aggressive - Veterinary Public Health Institute, Animal Welfare Division

ABSTRACT
Aggression is the most common reason for referrals of dogs to behavior specialists (Overall
1997, Landsberg et al. 1997, Askew 1996).
The theoretical section of this dissertation gives an overview of different aspects of aggression,
its classifications and definitions and describes several approaches reported in the literature. The
practical section reports on a study comparing the results of five methods currently being used in
Switzerland for evaluating canine aggressive behavior.
Part I. Behavior and Aggression: A Theoretical Overview
Many classifications and definitions of different types of aggression can be found in the
literature. This section gives an overview and analysis of a selection of existing classifications of
aggression from the fields of veterinary behavioral medicine, ethology, psychology and
cynology.
There is also a wide range of possible approaches to aggression as a problem behavior. The
theoretical overview presented here consequently considers several of these approaches rather
than concentrate on one direction or school of thought.
Part II. The Comparative Study
Three behavioral tests currently being used in Switzerland were selected for this study. Eight
criteria related to aggression in dogs (including intraspecific aggression, interspecific aggression
and attachment to owner) were defined. Sixty dogs were tested in three behavioral tests.
Additionally, a veterinary behaviorist and an ethologist evaluated 12 of these 60 dogs.
Results: The results were compared descriptively and with a Kappa test in the categories
‘aggressive behavior’, ‘relationship dog – owner’, ‘test passed’ and ‘measures recommended (to
the dog/owner)’. Using the Fisher’s exact test in the program NCSS, a possible correlation of
certain criteria within the tests (e.g. the influence of the expert on the results) was analyzed.
Within the three tests, a significant agreement was found for the criteria ‘intraspecific aggression’
(Kappa = 0.1334, p = 0.0143) and ‘interspecific aggression’ (Kappa = 0.1351, p = 0. 0140). The
veterinary behaviorist consistently evaluated the dogs higher in the aggression scale than did the
tests. (No statistical test was possible for this comparison). Several significant intra-test
correlations were found with the Fisher’s exact test, e.g. of the categories ‘expert’ and ‘hierarchy’
(p = 0.0037), ‘attachment’ (p = 0.0114) and ‘measures’ (p = 0.0003) in one test, or of the
categories ‘interspecific aggression’ and ‘measures’ (p = 0.0351) in another test.
When comparing the results of the tests to those of the behavioral specialists (veterinarian and
ethologist), a tendency to a low sensitivity is found.
Discussion and Conclusion: The agreement of the three tests in the criteria ‘intraspecific
aggression’ and ‘interspecific aggression’ is an important fact for public security. The difference
between the results of the tests and of the evaluations of the veterinary behaviorist and ethologist
suggest a closer look be made at possible ‘false negatives’ which might pose a latent danger for
the public.
1
INTRODUCTION
1. MOTIVATION FOR THIS STUDY
Aggression, aggressive behavior, dangerousness, vicious dogs, dangerous dogs, breed specific
legislation, behavioral test, temperament test, animal therapists, animal psychologists, veterinary
behavioral medicine, … – these are some of the words and phrases that have been appearing in
the media more and more frequently in the recent years. The German speaking media speak about
‘Kampfhunde’(‘fighting dogs’), a word many people use without really knowing what it means.
Exactly which breeds are included in this category? No scientific studies exist that define a
specific breed as a ‘Kampfhund’, although certain countries like France are trying to find
common morphological traits (such as size of the head, size of the dog, weight, etc.). The public
is led to believe that some breeds of dogs are a priori more aggressive and more dangerous than
are other breeds.
These increasingly frequent reports and articles in the popular press about dangerous dogs and
dog attacks – especially those that led to the killing of human beings – have contributed to the
buildup of a kind of hysteria in the population. Shocked by these media reports, the public in
many countries has urged their politicians to ‘do something’. In Germany, not undisputed
political measures have been taken. Several lists of ‘dangerous’ dogs have been set up, several
behavioral tests have been developed, dog taxes raised and new laws passed. A similar situation
exists in France. In Great Britain, the ‘Dangerous dog act’ had already been enacted in 1997.
Switzerland has been – and still is – being forced to consider how to handle this problem, how to
‘soothe the public’. In Switzerland, the canton of Basel-Stadt has introduced a breed-list and has
made a test obligatory for eight breeds. The canton of Basel-Land is working on a similar
proposal. Geneva has a breed-list, but no test. Other cantons are developing other solutions,
which are not based on breed-lists. In short, several suggestions to solve this problem are being
proposed, coming from many different directions and groups.
The same basic question must be faced, however, by all these countries and cantons - who is to
decide which dogs are dangerous and which ones are not? A collateral question is - what
education should the people have, who decide on the fate of the dogs?
The current rather confused situation, with a plethora of approaches and possible tests
administered by a wide variety of different kinds of ‘experts’, inspired us to look more closely at
the concepts and methods of testing of ‘aggressive behavior’ and ‘dangerousness’. We selected
five of the methods currently being used to evaluate canine behavior in Switzerland and
compared how they evaluate the same 60 dogs. We wanted to know if these methods all come to
the same evaluation of these dogs. If not, what might be the factors contributing to differing
evaluations?
2. FOCUS OF THE DATA ANALYSIS
The large amount of work necessary for the preparation and running of this comparative study led
to the decision to split up the resulting data among two doctoral students responsible for the
project. In a separate dissertation, Doris Lehmann (2003) focuses on the factor of dangerousness
and reviews a selection of existing behavioral tests for dogs. This study will be focussing on
aggression and different approaches to aggressive canine behavior.
2
3. STRUCTURE OF THIS REPORT
The first section of this dissertation gives an overview of different aspects of aggression, its
classifications and definitions and describes several approaches reported in the literature. The
second section reports on a study comparing the results of five methods currently being used in
Switzerland for evaluating canine aggressive behavior
Part I. Behavior and Aggression: A Theoretical Overview
‘Aggression’ is a word, which is difficult to define and to classify.
There is a wide range of possible approaches to aggression as a problem behavior – as
well as to problem behavior in general. The theoretical overview presented here will
therefore consider several possible approaches rather than concentrate on one direction or
school of thought.
Part II. The Comparative Study
Three behavioral tests currently being used in Switzerland were selected for this study.
Sixty dogs were run through all three of these tests. Additionally, a veterinary behaviorist
and an ethologist, both working in the area of behavioral therapy in dogs, were each asked
to evaluate 12 of these dogs according to their own method.
This project targets two aspects - ‘aggressive behavior’ and ‘dangerousness’, as these are
the two topics currently most discussed concerning dogs. They are also the most common
reasons for referrals of dogs to behavior specialists (Overall 1997, Landsberg et al. 1997,
Askew 1996). We are aware of the fact that a variety of other behavioral problems exists
in dogs and that some of the methods included in this study give information on other
factors than just aggression and dangerousness. However, the consideration of all these
other factors would surpass the scope of this two-year project.
4. UNDERLYING QUESTIONS
The underlying questions about the targeted topics are as follows:
•
•
•
•
•
•
Is it possible to test a dog’s behavior?
As all these methods claim to be able to evaluate a dog’s behavior correctly, the assumption
can be made that they should agree in their evaluations of the same dogs. Is this the case?
Specifically, do the three selected tests judge the same dogs to be similar in their behavior
related to ‘aggression’ and ‘dangerousness’? Are the evaluations of ‘aggressive’ or
‘dangerous’ by the two behavioral specialists similar to those of the three tests (based on the
same criteria)? If no, what are the possible reasons?
Do the test results correspond to what the owners say about their dogs in a questionnaire sent
to them in advance?
Is there a correlation of ‘aggressiveness’ and ‘dangerousness’ within the methods?
When talking about aggressive and ‘dangerous’ dogs, mainly the public security is of interest.
In this respect, it is interesting to know how great the specificity and the sensitivity of the
methods are and if there are many false negatives, i.e. how big is the incidence of dogs being
evaluated ‘not dangerous’ when in fact they are ‘really’ ‘dangerous’?
What is the importance of validity when evaluating canine behavior?
3
The null hypothesis of this comparison is as follows: As all five methods are currently being used
in Switzerland, it is to be assumed that they all agree on the same dogs’ behavior to a certain
extent.
Due to the different approaches compared in this project, and the fact that none of the tests
involved have been validated, several areas of this study may be considered a heuristic approach
to the comparison of some of the existing methods whose aim is to recognize potentially
dangerous dogs. 1
1
Heuristic comes from the Greek word ‘heuriskein”, meaning, ‘to discover”. A ‘heuristic’ method is an
approach for gaining knowledge about an area also through experience rather than solely through the
proving or disproving of hypotheses based on a scientific model. The heuristic paradigm has been used in
many sciences which involve the study of behaviors and includes concepts and tools for recognizing and
regulating the biases researchers set up as part of the research process.
4
PART I: BEHAVIOR AND AGGRESSION IN DOGS: A
THEORETICAL OVERVIEW
1. INTRODUCTION AND AIMS OF THE THEORETICAL PART
Problem behaviors in dogs have been discussed in many fields resulting an abundance of
information in the literature on how to approach these problems. However, a general overview of
different approaches does not yet exist. The literature mostly concentrates on the individual
author’s specific field, e.g. veterinary behavioral medicine, ethology, cynology or psychology.
The authors often do not define the model upon which their approach is based.
A comparison of the effectiveness of the approaches is not possible at the moment due to the
following factors:
• Not all authors describe, define or classify the behavioral problems they speak about, hence,
there is a general lack of descriptions, definitions and classifications;
• Many authors use the same terminology, but it refers to different things or they use different
words for the same problems. This leads to a lack of common definitions, agreed upon by
everybody independent of their specialized field (veterinary medicine, ethology, cynology,
psychology, etc.). A first task clearly would be to make sure that everybody is talking about
the same thing;
• The population of dogs seen and treated by the different experts is often not reported and
hence might differ considerably;
• The type and duration of the therapies, the frequency and duration of the consultations and
follow-ups, and the success rates of the treatments for the experts are not known. Hence, it is
not possible to compare the results of different experts. However, it is possible to give a
descriptive overview of several approaches and to show where they agree and where they
disagree.
The theoretical first part of this study thus consists of two sections. The first describes the
important aspects of aggression. The second is an overview of different approaches,
concentrating mainly on their handling of aggression.
5
2. LITERATURE SOURCES
For the overview, a selection was made of several textbooks, which deal with problem behaviors
in dogs in the areas of veterinary behavioral medicine, psychology, ethology and cynology. The
choice of literature was based on the recommendations of experts in the fields of veterinary
behavioral medicine and ethology.
Two kinds of analyses were made:
(A) The proportion of pages dealing with specific themes (such as normal canine behavior,
consultation, medication, therapies, and classification of disorders/symptoms/syndromes) per
total number of pages was calculated.
(B) An overview was made of factors related to approaching and treating aggressive behaviors,
including the following:
- The consultation (e.g. interview with the owner, observation of the dog, location of
consultation, physical examination),
- The treatment (e.g. medication, behavioral, systemic, cognitive),
- The subjects mentioned to be of interest during the interview (e.g. behaviors of the
dog, history of the dog, health of the dog)
The classification of problems (nosography) and the definitions of terminology are shown for
every approach.
The books included in this review are the following:
• Overall, K. 1997
•
•
•
•
•
Clinical behavioral medicine for small animals, Mosby Year Book,
Inc.
Pageat, P. 1998
Pathologie du comportement du chien, Editions du Point Vétérinaire
Landsberg G., Hunthausen W., Ackerman L. 1997
Handbook of behavior problems of the dog and cat, Oxford:
Butterworth-Heinemann
O’Farrell V. 1991
Verhaltensstörungen beim Hund, Ein Leitfaden für Tierärzte, Verlag
M. & H. Schaper GmbH & Co., KG, Alfeld, Germany
Askew H.R. 1996
Treatment of behavioural problems in dogs and cats: a guide for the
small animal veterinarian, Blackwell scientific publications, Oxford
Lindsay S.R. 2001
Handbook of applied dog behavior and training, Iowa State University
Press, two volumes
Books reviewed only for the topic ‘aggression’:
• Dehasse, J. 2002
Le chien agressif, Publibook
• Campbell, W.E. 1999 Behaviour Problems in Dogs, BehavioRx® Systems, Grants Pass,
Oregon
• Feddersen-Petersen, D. 2000
"Hundepsychologie", Kosmos-Verlag
• Houpt, K.A., Wolski, T.R. 1998
Domestic Animal Behavior for Veterinarians and Animal Scientists.,
The Iowa State University Press, Ames, IO
6
3. DEFINING BEHAVIOR
As aggression is a part of normal behavior, it is appropriate to begin with a brief consideration of
‘behavior’.
Surprisingly, it is quite difficult to find a satisfying definition or even a simple definition of the
word ‘behavior’ in dictionaries of ethology. Barrows (2001) refers to the Oxford English
dictionary. There is no definition in Heymer (1977), or in Mac Farland (1990). Immelmann
(1983: 13) says “behavior relies on movements, noises and body postures of an animal,
furthermore on those externally recognizable changes that are used for mutual communication
and can therefore elicit behaviors in the according partner” ∗ . Eibl-Eibesfeld (1984) refers to
behavior as a functional unit of (a sequence of) acts, which is externally observable. Piaget
defines behavior as “the whole of the actions that organisms shows towards their surroundings to
modify the state of it or to change their own situation in reference to it” (Piaget, 1976 cited in
Doré 1983: 7) 2 . Immelmann & Beer (1989) make a subtle differentiation based on the behaviorist
model: molecular behavior corresponds to movements, and molar behavior to actions.
A behavior is a functional unit constituted of an observable sequence of different acts
(movements).
Based on the notion that a behavior cannot be separated from the individual showing it, Dehasse
(2002b) developed a model of ‘psychels’ or ‘psycho-biological elements’. In this model, these
elements are considered the ‘basic atoms’ that influence the behavior of an individual. All the
psychels influence one another as they are organized in a hierarchical structure composed of four
levels. The ones higher in the hierarchy have an influence on all the ones below, e.g. a change in
the higher level of mood will have effects on the lower level structures of emotion, cognition,
perception, motor acts and neuro-vegetative acts (see Fig.1.1).
Figure 1.1 Organization of psychels in the model of Dehasse (2002b)
1.
Organism
2.
Mood
3.
4.
Emotions
Motor acts
Cognitions
S
T
R
O
N
G
Perceptions
Neuro-vegetative acts
∗
I
N
F
L
U
E
N
C
E
Translation by M. Braem (2003)
“l’ensemble des actions que les organismes exercent sur leur milieu extérieur pour en modifier des états ou pour
changer leur propre situation par rapport à lui”
2
7
W
E
A
K
I
N
F
L
U
E
N
C
E
Behavior, according to Mertens (2001: 14) is “the organic control and regulation of
environmental relations based on the exchange of information including the use of experience“.
Abrantes (1997) simply defines behavior as “a way of acting”. Everything in the way the
individual appears to others is counted as behaviour”. Pageat (1998: 368) defines behavior as “
"The whole of motor and communicational productions" 3 .
These are just a few examples, which illustrate the diversity of opinions, and definitions found in
the literature on behavior. The question arises: Is there a need to define this term precisely? One
could consider the study of behavior as part of the ‘science of the imprecise’, a ‘fuzzy concept’.
According to Moles (1995: 47) there is no necessity for too precise definitions in the ‘science of
the imprecise’, as the greater the precision of a definition, the smaller the operational and
heuristic value, the smaller the flexibility.
3.1
NORMAL BEHAVIOR
Odendaal says (1997) that as behavior is a complex phenomenon, the word itself cannot be
unanimously defined, it is very difficult to determine and define what ‘normal’ behavior - and
therefore what ‘abnormal’ - behavior is.
In semantics, the word ‘normal’ refers to a norm. Questions we will encounter throughout this
study are: Who establishes the norm? What is the norm, the standard, the reference? Is it actually
possible to define a ‘norm’, a ‘standard’?
There are several possibilities to define ‘the norm’. It may be defined by the observer leading to
the possibility of this definition becoming an anthropocentric model based on a statistical
average. As no actual ethogram of the domestic dog exists, there is no standard of reference to
establish what a normal or average behavior could or should be.
The norm may be centered on the animal itself, and called ‘physiological’ or ‘adaptive’ or
‘normal’ according to the average of the individuals of a species or according to the fitness of
survival of the species. Depending on the view, the norm for a species as a whole may be quite
different from the adaptive norm of the individual. For example, intraspecific aggression leading
to the killing of a conspecific may well be adaptive (‘normal’, ‘physiological’) for an individual,
but maladaptive (‘abnormal’, ‘pathological’) for the species.
The owner often decides for him- or herself if the behavior is acceptable for the system he or she
lives in or not, before consulting a specialist. A behavior may be ‘normal’ for the species,
‘adaptive’ for the individual, but it may be considered ‘abnormal’ by the owner, simply because
it is a nuisance. For example, a dog barking at a postman passing by may be acting according to
its species’ behavior, this behavior being adaptive and ‘normal’ for a dog living in a pack. This
barking may, however, be unacceptable in our society, annoying to the postman and
embarrassing to the owner. The opposite can be true as well, a behavior which is completely
‘abnormal’, maladaptive for an individual, such as lying around eight hours a day waiting for the
owner to come home, is considered as a desirable trait in many Western societies. The boundaries
between ‘abnormal’ and ‘unacceptable’, ‘normal’ and ‘acceptable’ are not strict but are
dependent on the context.
3
Translated by M.Bräm
8
3.2
BEHAVIOR OF WOLVES AND DOGS
The wolf is often considered the ‘natural’ example and primary reference point for canine
behavior. By extrapolating wolf behaviors to dog behaviors, the wolf often becomes the ‘norm’
for the behavior of our pet dogs. However, one must consider this extrapolation critically, as one
tends to forget that dogs are not simply domesticated wolves, but that they have different
ecological, social and physiological needs than wolves (Feddersen-Petersen 2001a, Askew 1996).
Askew (1996: 5) notes that “In addition to understanding how the domestic animals’ wild-living
ancestors lived and behaved, one must also strive to understand how (and why) this behavior has
been modified during the domestication process.”
Most dogs are surrounded primarily by human beings, they spend most of their lives in buildings,
they do not have to chase prey to survive, they are expected to live according to human rules,
they are expected to be inactive for the major part of the day. Some of them might have specific
tasks, such as assisting a blind person or guarding a house. As Brunner (1988, cited in FeddersenPetersen 2001a) says, there are behaviors that are considered normal in the wild form, but that
when elicited in a pet dog are regarded as abnormal. An example of this is shyness or fear of
humans, a trait that can be connected with a certain degree of aggression. As Schöning (2001: 97)
suggests “ ... one should not simply provide an ethological profile to the dog in comparison to the
wolf, but one should also compare different breeds of dogs ...“. Feddersen-Petersen has done
many studies on the comparison of the behaviors and the behavioral development of wolves with
different dog breeds (Feddersen-Petersen 2000b, 2001a) and has found that there are quite a bit of
differences, even according to the dog breed. However, here also, one must be careful with
extrapolating these results to pet dogs. The dogs studied in these projects all live in packs made
up of dogs or wolves and dogs living in kennels and not in a home situation. Our pets mainly live
in human families. What we need is an ethogram of the ‘pet dog’, living in a household with
humans. Here also, Feddersen-Petersen (2001: 175) limits the possibilities of generalization when
she states that “… the pet dog exists as little as the normal pet dog behavior”*. What may be
adequate and necessary for one breed might be not enough or even a torture for another
(Feddersen-Petersen 2001). In order to be sure to give every dog what it needs, she recommends
making ethograms according to dog breeds.
3.3
NOT NORMAL BEHAVIOR
As it is not easy to define ‘normal behavior’ definitely, it is consequently difficult to define
abnormal behavior. Beaver (1994: 3) defines ‘abnormal behavior’ as “any behavior that varies
from the norm expected for a species can be called abnormal. This variation can relate to the style
or timing of the act, represent a normal behavior used in an inappropriate place or time, or be a
behavior not typically used by a species.” She defines the norm as “what is expected for a
species”, however, she does not define who decides what is normal for a species. Her definition
four factors influencing whether a behavior is considered normal or not: style, time, place, and
species.
To complicate matters, not every school, not every model, not every scientist categorizes
behavior problems in the sense of “abnormal” in the same way.
9
Overall (1997), an exponent of the Anglo-Saxon school of thought, differentiates between
‘appropriate’ and ‘inappropriate’ behavior, whereas Pageat (1998), as a leader of the French
school of thought, considers a behavior either as ‘physiological’ or ‘pathological’, as ‘adaptive’
or ‘non-adaptive’. Only the French model recognizes the aspect of ‘pathology’ in psychological
disorders, as a parallel to the same notion in somatic illnesses. This means that an animal is not
adaptive to the current situation, it does not return to the state of physiological homeostasis. Are
these two authors who represent two different schools of thought talking about the same things?
Coming back to aggression – as this is a behavior, this word cannot definitely be classified as
‘normal’ or ‘abnormal’, ‘physiological’ or ‘pathological’, as ’appropriate” or ‘inappropriate’.
Unless there is an underlying disease that can be diagnosed objectively as being ‘pathological’, as
deviating from the ‘physiological norm’ (which is the advantage of the veterinary behavioral
specialists). Depending on the school of thought (French, Anglo-Saxon, ethology, psychology or
cynology) and even on the author, aggression is defined differently on the basis of possible
causes, functions or simple descriptions. Although all these different approaches do agree in
certain aspects, there are larger and subtler differences worth paying attention to. Some of these
aspects will be discussed in the chapter ‘overview of different approaches’ (chapter 3.9).
The boarder between normal and non normal aggression is fluid and experts communicating with
each other must first make sure they are really speaking about the same problem or using the
same model or paradigm. The goal – perhaps an utopian idea - would be to develop a ‘common
language’ amongst experts from different areas.
4. AGGRESSION
4.1
DEFINITIONS OF AGGRESSION
The attempt to frame a simple, precise and universally applicable definition of aggression has
historically not met with much success.
Voith and Borchelt (1985: 949) describe the general situation as follows:
“Despite thousands of research reports investigating aggression, no uniformly acceptable
definition has ever been agreed on. By one definition, aggression is behavior that leads to,
or appears to an observer to lead to, the damage or destruction of some target. This is a
fairly objective, observational definition of aggression and would include predation.“
“If aggression in animals is defined solely as inflicting injury on another individual, then
such threats as growling, baring the teeth, or snapping at an individual do not fit the
definition. If aggression is defined solely in terms of certain acts, such as growling or
biting, the definition is also unsatisfactory ... If aggression is defined solely as acts that
occur when an individual is in an emotional state, e.g., “anger”, the formidable task arises
of defining and identifying the emotional state in order to assess whether or not an animal
is aggressive. (950)”
10
The following definitions of ‘aggression’, culled from different sources, further illustrate the
difficulty of the task:
Heymer (1977: 24ff): “A physical act or threat of action by one individual that reduces the
freedom or genetic fitness of another.”*
Immelmann & Beer (1989: 8-10):
“A general term for all elements of attack, defense, and threat
behavior.”
Eibl-Eibesfeld (1984) cited in Dehasse (2002b: 13):
“Any behavior resulting in forcing another individual to stay at a
spatial or social (hierarchical) distance can be called aggressive, even
if it does not result in any physical damage.” *
Pageat (1998: 365) "A physical act or threat that allows an individual to reduce the
liberty and the genetical potency (reproduction) of another
individual".*
Dehasse (2002b: 14) "Aggression is defined as a behavior that leads to - or of which the
apparent aim is to - to do harm to the physical (and/or psychological)
integrity or freedom of another individual." *
Feddersen-Petersen (2001a: 27)
"Aggressive behavior [is when] motor patterns are shown that are
directed towards the bodily intactness of social partners.” *
Landsberg (1996: 129)
"In general [the term aggression] refers to threatening or harmful
behaviour directed towards another individual or group.”
Lindsay (2001: 175) “aggressive behavior, not stemming from idiopathic or pathological
causes, can be viewed as an adaptive effort to establish control over
some vital resource or situation that cannot be effectively controlled
through other means.”
Overall (1997: 88) Aggression "is defined within a given context as an appropriate or
inappropriate threat or challenge that is ultimately resolved by
combat or deference."
4.2
WHAT IS AGGRESSION?
The word ‘aggression’ originates from the Latin word ‘aggredi’ which means ‘to approach, to try
to win somebody for oneself’, but also ‘to attack’. The word has a negative connotation for many
people; it triggers negative feelings. However, aggression is a normal and necessary part of
canine (and in general, mammalian) behavior (AGGH 2000, Dehasse 1999, Feddersen-Petersen
2000a, 2001b, Pillonel 2001). Like the original Latin word, aggression does not have to be
understood only negatively.
Aggression is a ‘meta-class’ of behaviors, composed of several different sub-classes of
‘aggressions’, the specification of which differs according to author and theoretical model.
Aggressive behavioral sequences are found in agonistic behavior. ‘Agon’ in Greek means ‘to
struggle, to fight’. ‘Agonistic behavior’ is defined by Borchelt and Voith (1985: 951) as “ … used
in connection with any sort of behavior that involves a contest or conflict between two animals,
usually of the same species ... .Agonistic behaviors include threats ( ... ), offensive aggression
(such as chasing and biting) and defensive behaviors (including aggression, fleeing, submissive
signals, threats, and biting)“. Dehasse (2002b: 14) defines it as „any behavior that can contributes
11
to the solution of a conflict, meaning aggression itself, but also threatening behavior, flight,
appeasing and submissive behaviors, etc“ *.
There seem to be three main points common to these definitions:
• At least two individuals are necessary for agonistic behavior to take place;
• There is the component of threat or conflict;
• In nature, agonistic behavior, including aggressive behavior towards conspecifics, is aimed
at solving this conflict before it evolves into a serious fight, at avoiding injuries and death.
Aggression a priori is not dangerous or bad, but a necessary component of normal
communication between individuals socialized to each other. It is a communication between (at
least) two individuals about their relationship to each other, a mutual cost-benefit analysis of their
strengths and weaknesses (Feddersen-Petersen, 2001b). Amongst wolves, the main means of
communication is aggression (Feddersen-Petersen, 2001b) consisting of mutual threats – this is
necessary for the social system to function, for the hierarchy within the pack, for the relationships
between the individuals of the pack to stay stable and to avoid injuries within the group. The fact
that wolf behavior may not simply be projected to dog behavior will be discussed in a later
section. There are controversies about the fact whether predatory aggression actually is part of
aggressive behavior or not. Zimen (1992) states that predatory behavior is not primarily a
behavior motivated by aggression, as no signs of aggression (e.g piloerection, showing its teeth)
are shown during or after hunting.
The act of threatening is the first phase of the social (intragroup) aggressive sequence. Like many
behaviors, aggressive behavior consists of four phases (Dehasse, 2002b) elicited by a stimulus:
1) An appetent or beginning phase with signals of threat, such as showing teeth, growling
or piloerection;
2) A consumption phase or phase of action consisting of attack with or without biting;
3) A quieting or ending phase which allows the animal to ‘cool down’; and
4) A refractory phase during which no new aggressive behavior can be elicited.
This sequence is an important point to keep in mind when deciding on the diagnosis and choosing
the therapy for a specific problem. The integrity of the sequence helps in defining whether a
behavior is adaptive or maladaptive/pathological, predictable or not predictable. Dehasse (2002b)
shows a few examples of how the phases can differ according to the type of aggression (see
Table 1.1):
Table 1.1 Types of aggression according to Dehasse (2002b)
Type of aggression
Atypical
Competitive
Distancing
Hyperaggression
Irritation
Fear
Predatory
Etc.
Intimidation
Absent
Structured
Structured
Absent
Short
Signs of fear
Absent
12
Attack
Direct
Controlled
Controlled
Direct
Controlled
Almost direct
Typical sequence
4.3
AGGRESSION AND DANGEROUSNESS
For many people, an aggressive dog is automatically also a dangerous one and vice versa. This,
however, is not always the case.
Firstly, it depends on the definition of dangerousness, of which, as for aggression, there are
numerous. Pillonel (2001) defines a dangerous dog as “a dog that impairs or threatens to impair
another individual’s physical intactness or freedom of movement by its behavior”*. Dehasse
(2002b: 14) defines a dog as being ‘dangerous’ when “it endangers the physical and/or
psychological integrity of a human or other living being”*. Stur (1998) concludes from the
definition of the ‘dangerous dog’ by the VDH (‘Verein für das deutsche Hundewesen’):
“According to this definition, dogs are considered as being dangerous if they are socially
unacceptable and for which one can assume with high probability that they will bite a human or
animal in a conflict situation” 4 .*
Secondly, aggression, as discussed above, can be a part of the normal ethogram of a species. The
word ‘aggression’ refers a behavior sequence in a certain situation and at a certain moment.
‘Dangerousness’, on the other hand, is a concept that cannot be described as a ‘part of the normal
ethogram’, since behavioral elements reflect only some of the many factors involved in
‘dangerousness’.
Aggressive behavior is an important point to consider when evaluating the potential
‘dangerousness’ of a dog, but several other factors have to be taken into account as well. These
factors include: the behavior of the victim and the’attacker’ before, during and after the specific
incident; the situation before, during and after the incident, the size, body mass and health of the
victim and the attacker; the capability of the victim to understand canine behavior, the upbringing and keeping of the dog(s) etc. (Pillonel, 2001, Stur 1998). This interplay of multiple
factors underlines the fact that every dog is potentially dangerous, simply because it has the
means (size, weight, teeth, speed, etc.) to injure another individual. It is impossible to predict a
living being’s reactions in every situation and moment, and consequently, it is impossible to give
a one hundred percent guarantee that a dog is ‘not dangerous’. This complexity also implies that
only an experienced professional is capable of assessing the dangerousness of a dog.
Finally, a dog does not have to be aggressive to be dangerous. A large dog, for example, that
likes kids and jumps up at every kid it sees, is dangerous simply because it can push the child
over and injure it by doing so. On the other hand, a dog showing aggressive behavior does not
necessarily have to be more dangerous than one not showing this behavior, e.g. a dog growling at
kids out of fear and then escaping this fear-eliciting situation, is not dangerous, as there is no
confrontation. As discussed above, aggression is a way of communicating and, most often, this
communication stays on the level of threats without attack and without any injuries on either side.
For a more detailed discussion about ‘dangerousness’, see Doris Lehmann (2003).
4
“Nach dieser Definition gelten als gefährlich sozial unverträgliche Hunde, bei denen mit hoher Wahrscheinlichkeit
davon auszugehen ist, daß sie bei Auseinandersetzungen mit Menschen oder Tieren, auch Artgenossen, beißen”.
13
4.4
INFLUENCING FACTORS
Almost all authors agree that aggression can be influenced by many factors. This section gives a
brief overview of the most frequently found influencing factors in the literature.
4.4.1
INDIVIDUAL FACTORS
4.4.1.1
Genetics
Triggered by the media and the laws being introduced in several countries, including certain
cantons of Switzerland, there is a widespread popular belief that specific dog breeds are a priori
more aggressive (and by this dangerous!) than others. However, up to now, there is no scientific
evidence for breed-specific innate increased aggression or dangerousness (Bundesamt für
Veterinärwesen 2000, Arbeitsgruppe „Gesetzgebung betr. Gefährliche Hunde“ 2001, FeddersenPetersen 2001b, Stur 1998).
As Stur (1987: 957) points out, two major aspects must be considered when differentiating canine
temperament genetically: (1) “the genetically fixed species-specific temperament” and (2)
“behaviour patterns, that show genetic variance within the species”. Aggression belongs to both
of these groups. Aggressive behavior belongs to the first group in the sense that “the manner of
aggression is therefore genetically fixed in the species dog (Stur 1987: 959)”. It is also part of the
second group, when considering the “quantitative differences in the degree of aggression between
breeds of dogs and also within the breeds. (Stur 1987: 959)” Clinicians have observed that there
are certain lines within certain breeds that have a lower threshold to elicit aggression and
therefore a higher tendency to show (dangerous) aggression than do others (Feddersen-Petersen
2000a). Examples of this are certain lines of bull terriers (Feddersen-Petersen 2001a, 2000b) or
English cocker spaniel (Podberscek 1996). “ … in the course of domestication, man has modified
the aggression pattern as far as intensity is concerned but not its basic nature” (Stur 1987: 959).
The breeds shown to be most often involved in dog bites vary according to study, author and
country (Horisberger 2002, Beaver 1993)
Stur (1987: 959) states that “aggression is a characteristic of high heritability […] thereby
increasing aggression by selection can be done very easily”. However, she does not specify the
heritability lever or whether this increased aggression is specific to certain breeds. In a following
sentence, she adds “furthermore dogs with high aggressiveness are not always owned by people
who can guide this aggression into the right channels (959)”. The danger a dog presents is, as
mentioned above, always influenced by several factors. As Eichelberg (2000) says, the capacities
of a dog are usually much greater than those that are actually used. This leads to the hypothesis
that a good dog trainer could teach almost any dog to show certain behaviors, e.g. to guard a
house or to retrieve or…to attack, i.e. become dangerous. The breed of dog influences mainly the
time it takes to teach the dog the desired behavior. The human being has a great influence on the
development of a dog – “with the wrong human partner, even ‘peacefully’ predisposed, large
dogs can become a great danger … 5 ”* (Feddersen-Petersen 1990: 232).
5
Translated by M. Bräm
14
The main aim of breed specific legislation seems to be to soothe the public. However, this is a
critical point, as often the public will receive a false sense of accomplishment (American
Veterinary Medical Association (AVMA) 2001). Breed specific legislation may increase the
feeling that the ‘dangerous dogs’ are taken care of, and lead to a false security of the public
towards other dog breeds or to other factors which may in fact lead to dangerous situations.
In the United States of America, a task force was set up to analyze the aspect of breed specific
legislation (AVMA 2001). The conclusion of this task force is “Members of the Task Force
believe such ordinances are inappropriate and ineffective“, based on the following arguments
(AVMA 2001: 1735):
“1. First, a dog’s tendency to bite depends on at least 5 interacting factors: heredity, early
experience, later socialization and training, health (medical and behavioral), and victim
behavior
2. There is no reliable way to identify the number of dogs of a particular breed in the canine
population at any given time
3. Statistics may be skewed, because often they do not consider multiple incidents caused
by a single animal
4. The breed is often identified by individuals who are not familiar with breed
characteristics and who commonly identify dogs of mixed ancestry as if they were
purebreds
5. The popularity of breeds changes over time, making comparison of breed-specific bite
rates unreliable ... Law enforcement personnel typically have no scientific means for
determining a dog’s breed ... ”
In German, the word ‘Kampfhund’ (‘fighting dog’) has become very popular. The following
breeds are generally considered to belong to this category: American Staffordshire Terrier,
Bandog, Bullmastiff, Bullterrier, Dogo Argentino, Dogue de Bordeaux, Fila Brasileiro, Mastiff,
Mastin Espanol, Pit-Bull, Rhodesian Ridgeback, Staffordshire Bullterrier, Tosa-Inu, Doberman
and Rottweiler (Eichelberg 2000). But the question remains, what makes a ‘fighting dog’? What
do these breeds have in common? According to Eichelberg (2000: 93), there is no common
background (morphological, historical, genetical, etc.) for these breeds. “The only thing they
have in common is the fact that they are dogs, and that they can all bite without exception”*.
Stur (1998) states that elements of temperament (‘Wesensmerkmale’) are characteristics with a
low heritability 6 . Ruefenacht et al (2002) studied the heritability of seven different behavior traits
(self-confidence, nerve stability, temperament, hardness, sharpness, defense drive and fighting
drive) based on the results of German Shepherds (N = 3497) in the test for this breed in
Switzerland. The heritability found in this study for the above-mentioned behavioral traits ranged
from 0.09 to 0.24. She concludes that “The modest genetic improvement over the last 25 years in
the studbook population of the German Shepherd dog (GSD) was due to the low heritabilities of
the behaviour traits, but mainly because of the low selection intensities after the test (only 8%
failed)” (Rüfenacht 2002: 131).
6
Heritability = „Mass für den Anteil, den der Genotyp an der phänotypischen Merkmalsprägung ausmacht“ (Stur,
1998)
15
4.4.1.2
Hormones
The three main categories of hormones discussed in the literature to have an influence on
behavior, including aggression, are thyroid hormones, glucocorticoids (Cortisol) and sex
hormones (testosterone, estrogen, progesterone).
Several authors mention a possible correlation of thyroid hormone levels and behavioral changes,
including aggression, in dogs (Aronson & Dodman 1997, Pageat 1998, Hamilton, McBride &
Brown 1999, Morgan, Bradshaw, Casey & Thomas 1999, Barlow, Bradshaw & Casey 2001).
While there is no certainty about the influence, which thyroid hormones have on behavior, there
are, however, several hypotheses. Aronson & Dodman (1997) recommend hypothyroidism as a
differential diagnosis for dogs and horses showing ‘inappropriate aggression’.
One of the problems arising when correlating aggressive behaviors with hormones is that the type
of aggression is often not defined, such as in this study. Dodds & Aronson (1999) found thyroid
dysfunction in 62% of the aggressive dogs evaluated in a study run at the Tufts University on 634
canine cases of aberrant behaviors. These authors do not specify the type of aggression in this
study. It is not possible to correlate a specific hormone change with aggression in general,
because aggression in general does not exist. Aggression is shown in a specific context towards
one or several individuals. Other authors specify which aggression types may be influenced.
Hamilton, McBride & Brown (1999: 136) found results that “suggest that aberrant behaviours
[including dog to dog aggression, dog to human aggression and territorial aggression] may be one
of the earliest signs of thyroid deficiency”. These authors consider the behavioral changes to be a
result of the medical condition.
Several authors hypothesize that there is a correlation of thyroid hormones and cortisol (Barlow,
Bradshaw & Casey 2001, Dodds & Aronson 1999, Morgan, Bradshaw, Casey, & Thomas 1999,
Aronson & Dodman 1997). A decreased level of thyroid hormones inhibits cortisol clearance,
thereby increasing cortisol levels, which mimics a constant state of stress and has a depressing
effect on TSH, reducing thyroid hormones further. Cortisol may have effects on behavior as well.
The question remains, is the decreased thyroid hormone level a consequence of the increased
cortisol levels (resulting, e.g. from chronic stress) and behavioral changes, or is it the prerequisite
for the behavioral changes, or are the two aspects not correlated at all? Dodds & Aronson (1999)
state that depression, which is found in hypothyroid dogs, has been shown to change neural
activity (including that of serotonin and dopamine receptors) or volume in brain regions that
regulate behavior, including aggression. Serotonine decrease and dopamine increase have been
correlated with aggressive behavior. However, as we are lacking information on predictability of
hypothyroidism in aggression, these correlations are merely retrospective and not necessarily
causal (Van Erp & Miczek 2000, Soderstrom, Blennow, Manhem & Forsman 2001).
Practitioners often report informally that drugs thought to increase serotonin levels (SSRI) and
decrease dopamine levels (neuroleptics) are effective in decreasing certain aggressive behaviors.
Sex hormones are usually mentioned in combination with behavioral problems, including some
types of aggression. However, “few data exist on the decrease in aggression seen within each
type of aggression” (Overall, 1997: 97). The influence of sex hormones has been studied by the
change of behavior when the hormone is lacking or added to the system, i.e. the behavioral
changes found after castration and by behavioral changes after application of a sex hormone as a
therapy. Several studies have been done analyzing these effects: Neilsen, Eckstein & Hart
(1997a: 180) found that “with various types of aggressive behaviour, including aggression
towards human family members, castration may be effective in decreasing the aggression in some
16
[male] dogs, but fewer than one third can be expected to show marked improvement.” Wright &
Nesselrote (1987) found that intact males are more often involved in aggressive behavior than are
castrated ones. The hypothesis exists that testosterone has an effect on aggressive behavior in
dogs. As Overall (1997: 96) points out, “testosterone acts as a behavior modulator that makes
dogs react more intensely”.
Intact females seem to show (unspecific) aggression less often than neutered females (Gebers,
Fuchs & Barnabe 2002, O’Farrell & Peachey 1990; Wright & Nesselrote 1987). Here, the main
hormones considered are estrogen and progesterone. Estrogen is said to have “threshold-lowering
effects on the female” (Lindsay 2001: 182) and progesterone to “elevate the behavioral
thresholds” and to “assert a calming effect on dogs” (Lindsay, 2001: 182). Lindsay points out that
there is a controversy about the effects of progesterone on aggressive behavior with not all
authors agreeing on the anti-aggressive effect of this hormone. Overall (1997), for example,
states that progestines increase certain types of aggression, such as intrasexual interdog
aggression amongst females.
Another fact suggesting that hormones may have an influence on aggressive behavior is “the
increase of aggression and competitiveness are often associated with hormonal changes occurring
around puberty, a biological change that may lower the threshold for several significant sexrelated behavior patterns, including intermale and interfemale aggression“ (Lindsay, 2001: 181).
Pageat (1998) interprets this increase in aggression not as a consequence of a hormonal effect,
but of an increased dopaminergic effect in the adolescent dog’s brain, which leads to a decrease
in the regulation of self-control and aggression.
4.4.1.3
Neurobiochemistry
Many behaviors have been correlated in the literature to neurotransmitters. However, these
studies must be read critically, as correlation does not automatically imply causality. Several
dozen neurotransmitters and neuromodulators are known, but only a few of them are being used
as a model to explain certain changes in behavior (such as aggression). Serotonin, dopamine,
noradrenalin and GABA are the neurotransmitters most often found studied in correlation with
behavioral problems. In the history of psychopharmacology, there are phases of popularity of
specific neurotransmitters associated with certain behaviors. At the moment we are a ‘serotonin
phase’. Serotonin is currently thought to have an influence on aggressive behavior, as many drugs
used to treat aggressive behavior problems are thought to have an effect on serotonin activity
(e.g. serotonin-reuptake-inhibitors). However, the aggression decreasing effect has not been
proven to be linked to the supposed increase in serotonin levels. Other drugs classified within the
group of selective-serotonin-reuptake-inhibitors, such as sertralin or citalopram, may even
increase aggression. The conclusion is that each drug has to be studied individually and that it is
not possible to generalize the effects of one substance to a group of drugs.
4.4.1.4
Age
Problems in aging dogs (> 10 years) are mainly consequences of deficiencies in memory and
learning (Hart, Neilson & Ruehl 1997). Only „few behavioral problems could be directly
attributed to an underlying medical problem“ in a study run by Chapman % Voith (1990: 945).
Aggressive behavior problems, as well as other behavior problems, seem to appear for the first
time in old age rather than being a continuation or exacerbation of an already existing problem
(Chapman &Voith 1990). Hart, Neilson & Ruehl (1997: 31) believe that late-onset
aggressiveness is mostly irritable. They studied the correlation of this aggression with cognitive
dysfunction and define cognition as: “mental processes that are occurring within animals and
which cannot be directly observed but would include memory, learning, awareness and
17
perception”. They found that late-onset aggression was four times more likely to be found in dogs
diagnosed (by them) to have a cognitive dysfunction as in those that were not. According to the
results of this study, old dogs are not more or less likely to show aggression than younger dogs.
Chapman and Voith (1990: 946) have even reported that “few old dogs were referred because of
aggressive behavior”. Possible explanations for this are: (1) aggressive dogs are not kept, or (2)
other types of behavior problems are actually more frequent in older dogs. However, old females
were represented showing aggression twice as often than old males in the 146 dogs studied by
Hart, Neilson & Ruehl (1997).
Wright & Nesselrote (1987) studied the distribution of behavior problems in 105 dogs concerning
age, breed, sex and reproductive status. These authors found that not many dogs less than one
year of age were presented for behavioral problems. They formulate the hypotheses that (1)
young dogs might be less likely to show serious patterns of aggressive behavior, (2) the owners
of young dogs showing (aggressive) problem behaviors might try to solve the problem first on
their own, or (3) many owners hope that their young dog will ‘grow out of it’.
As mentioned above, an increase in aggressive behavior is often found around puberty which
may be explained by the changes taking place in the hormonal and neurotransmitter systems in
this phase of life as explained in the chapter “hormones”.
To summarize these findings: Aggression does not seem to be found more often in old dogs than
in young, however, a difference in the distribution among the sexes was noted. Age is only one
factor to consider in aggression, with decreased cognitive capacities being one major influencing
factor for behavioral problems to occur. However, all the other influencing factors mentioned in
this chapter are of course just as valid for old dogs as for young. Old dogs are naturally more
prone to showing bodily deficits, such as diminished eyesight or hearing or pain caused by
arthrosis, than are younger dogs, which may increase the probability of irritation aggression to be
shown.
4.4.2
ILLNESSES / PHYSICAL AILMENTS
4.4.2.1
Pain
Pain is defined as “the awareness or perception of a noxious stimulus that is potentially damaging
to tissue” by Kelly (in Ettinger, 1995: 21). Moreover, Kelly finds it is a mainly subjective
phenomenon, meaning that the same stimulus does not have to have the same behavioral effects
on every dog. Pain is known to be a possible cause of aggression. Most authors even give this
specific cause its own category as ‘pain-induced aggression’ (such as Houpt & Wolski 1998,
Overall 1997, Landsberg, Hunthausen & Ackerman 1997, Askew 1996, O’Farrell 1991).
However, pain is not a diagnosis, but a symptom (Kelly in Ettinger 1995) with aggression being
one of the possible behavioral changes seen in dogs suffering from pain. Every veterinarian or
dog owner knows the situation of a dog being hurt badly (a fracture, an open wound) and not
allowing anybody to touch it in this situation. Scott (2000: 37) claims that behavioral problems,
including “the three main categories of disorders that we as behaviourists recognize: aggression,
anxiety and compulsive type behaviours”, “can be caused by pain as the sole trigger”. She
describes mainly musculoskeletal pain, but also dental and visceral pain to be the main factors
involved in behavioral changes. Pain may increase aggression in several ways: It may be elicited
directly when a painful manipulation is done to the dog, which could be considered a type of
irritable aggression or it may be shown as a consequence of the general change in mood found
with chronic pain. The dog may be more irritable as a consequence of lack of sleep and rest,
uneasiness through limited range of movement, or constant stress.
18
Therapies of this type of aggression are primarily aimed at the causing element and not the
aggression itself.
4.4.2.2
Sensory Deficiencies
A dog that cannot see or hear is more likely to be startled by a sudden touch than a dog with good
eyesight or hearing. Sudden blindness or deafness can lead to confusion and anxiety in dogs, with
the possibility of increased irritation or fear aggression, which may evolve into secondary
hyperaggression as defined by Dehasse (2002) or Pageat (1998). These causes of aggression
should be checked before starting with a behavioral treatment, especially if dogs tend to be more
aggressive when there is less light or in noisy surroundings or when startled by somebody
approaching from behind.
4.4.2.3
Infection
Several viral (e.g. rabies, Aujetzky), bacterial e.g. Lyme borreliosis (Straubniger 2000) and
parasitic (encephalitozoon (Braund 2003e) infections are known to possibly cause aggression in
dogs. The most obvious is rabies, but any infection of the brain tissue, leading to encephalitis or
meningitis can lead to aggressive behavior.
4.4.2.4
Metabolism
The influence of thyroid hormones and cortisol, and therewith of hypothyroidism, hyper- and
hypoadrenocorticism, on aggressive behavior is discussed in the section ‘hormones’ and will not
further be discussed here.
Aggression can be one of the behavioral changes found in several rare storage disorders, such as
ceroid lipofuscinosis and gangliosidosis. (Braund 2003a)
4.4.2.5
Neurological
It seems obvious that aggression can be shown as a symptom of almost any process involving the
brain. Ranging from infections, which have been mentioned above, through brain tumors and
developmental disorders, epilepsy (Berendt 2002), hepatic encephalopathy and neurological
syndromes, such as diencephalic syndrome or the cerebral syndrome (Braund 2003d). Active
agents found in toxins (Braund 2003c) or medications may lead to increased aggressive behaviors
as well.
Learning
4.4.2.6
Learning will be discussed in terms of two categories, which focus not on the manner of learning,
but rather its importance to humans:
Taught behaviors, i.e. behaviors that a human being willingly teaches a dog, behaviors the human
being wants the dog to learn, and
Learned behaviors. As Feddersen-Petersen (2002) says, „every stimulus can become a trigger of
fear and aggressive behavior through learning“ *.
19
Taught Behaviors:
Dogs can be taught to be aggressive on command. In certain ‘dog jobs’ this is desired, such as
with police dogs or military dogs. However, more and more private persons are starting to
educate their dogs as ‘Schutzhunde’. While working with dogs is valuable, for both the dog
and the owner, care must be taken with dogs being trained to show learned aggression.
Feddersen-Petersen (2000b) urges caution when buying an already trained dog, as the lack of
attachment to the new owner can easily lead to the dog’s aggression slipping out of control.
She also points out the danger of dogs that have not finished the education because they were
not ‘suited’ or were getting too ‘fierce’. If these dogs end up with new owners who have no
idea of the history of the dog, there is a great danger of incidents occurring. (FeddersenPetersen 1991). Under the guidance and supervision of a competent and educated owner, such
training should not increase the potential dangerousness of a dog.
Learned Behaviors
Dogs, as humans, learn all the time, no matter what is happening around them. Almost any
type of aggression can become instrumentalized, meaning that the dog learns through
instrumental and/or classical conditioning that its behavior has certain effects. If these affects
are positive, it will more likely repeat the behavior. Dehasse (2002b: 153) describes this
phenomenon as ‘secondary hyperaggression’ being “the instrumentalized consequence of
another type of aggression. There is a loss of adaptive capacity, a regression of intelligence
level to the state of reflexes”*.
One well-known example is the dog barking and growling at people passing by its fenced-in
garden. The passerby approaches the dog’s territory, the dog barks and growls, the passerby
continues on his way, as he has no intent of entering the property. In the eyes of the dog, its
behavior was successful, because it attained what it wanted to, namely, that the intruder
leaves. So the dog will repeat this behavior with the next passerby. With time, the passerby
has become a conditioned stimulus that automatically triggers the aggressive behavior in the
dog.
Another example is the fearful dog that does not want to be touched. It growls at the person
stretching out his hand to pet it, the person pulls back his hand. The dog was successful and
will repeat the behavior the next time this person tries to pet it.
Within this category of aggression, there is always the danger of aggravation and
generalization. In the example of the fearful dog, it might start out with a growl at one specific
person, with time it might generalize to biting any human being that tries to approach the dog.
4.4.2.7
Emotional State - Mood
As humans all know from themselves, the general emotional state we are in has a great influence
on how we react to outside stimuli. If we are in pain, tired, ill, afraid or irritated, we might be
more likely to react aggressively towards an otherwise ‘neutral’ stimulus than if we are healthy
and feeling good. The same can be found with dogs. It is known that anxious dogs can show fear
aggression in situations they perceive to be threatening. Part of the definition of Pageat’s
‘hypersensitivity-hyperactivity-syndrome’ (Pageat 1998) is the lack of self-control and bite
inhibition – these ‘hyper’ dogs tend to bite because they are not capable of controlling
themselves, of stopping. These incidents often happen during playing episodes.
Dogs suffering from chronic depression or hyperattachment may be more prone to show irritation
aggression (Dehasse 2002b). Pageat (1998) and Dehasse (2002b) describe uni- and bipolar
disorders (parallel to similar disorders found in humans), within this disorders, in which
distancing, irritation and fear aggression can be shown (Dehasse 2002b).
20
4.4.3 ENVIRONMENTAL FACTORS
We have spoken about the endogenous factors that can influence aggressive behavior. Of course,
the environment plays a role as well. Dehasse (2002b) thinks that genetics have about an average
influence of 30% (varying from 0 to more than 50%) on the development of behavior, leaving
about 70% to the environment. However, one must recognize, that raising a puppy in a perfect
environment, socializing it to as many stimuli as possible and giving it a good education, does not
guarantee a dog without problems. There still remains that 30% and all the many unknown and
uncontrollable factors involved in behavioral development.
4.4.3.1
The Mother Dog
The education by the mother is very important during the first few weeks of life. A good dog
mother teaches her puppies the submitting postures and self-control. If this is not the case, the
dog is likely to develop for example a hyperactivity disorder, where it is not capable of
controlling itself, or a disocialization disorder (Pageat 1998) where the dog simply did not learn
how to communicate with other dogs (Dehasse 2002b). Dehasse (2003b) mentions that puppies in
large litters exceeding 5-6 puppies are more likely to become hyperactive or to suffer of
disocialization, and consequently show certain types of aggression. He hypothesizes that with too
many puppies in her litter, the mother dog is overtaxed with the task to teach all of them the right
communication.
4.4.3.2
The Siblings
Puppies play. Play is an important factor in the acquisition of self-control and bite-inhibition.
Dogs are not born with an innate knowledge of how strongly to bite – they need to learn.
Through the reactions of their siblings to their bites, they learn to regulate the intensity, they learn
what is called ‘bite-inhibition’. It is known that in litters with few puppies, the likelihood is
greater than in medium-sized litters for the puppies to develop hyperactivity and disocialization,
which may include different types of aggressive behavior (Dehasse 2003b).
4.4.3.3
Socialization
During the sensitive period (about 3 weeks to 3 months of age, varying with breed, size and
individual), the puppy must be socialized with as many different stimuli as possible, mainly those
which it will encounter in its later life. This socialization allows the puppy to get habituated to
these stimuli and realize that they are no threat, that they are either harmless or even fun. Dogs
that were not in contact with a certain stimulus (such as other dogs or humans or children) during
this phase, are more likely to develop fear of these stimuli later in life and may react with selfdefense aggression, mainly distancing, irritation and fear aggressions (Dehasse 2002b). Pageat
(1998) and Dehasse (2002b) call this syndrome ‘primary deprivation disorder’.
Communication and Social Structure
4.4.3.4
It is important that the owners are familiar with canine communication, with the dog’s natural
needs and its normal social structure. After an incident, owners often claim “he attacked
suddenly, without provocation”. Donaldson (1996: 59) explains that “a major element of the
culture clash between dogs and humans is the differing perceptions of what constitutes a threat”.
If dangerous aggression should be avoided as much as possible, humans have to learn to
understand the canine language and to think ‘canine’. How we humans perceive things is not
always how dogs perceive them. Being able to understand canine communication helps in
21
predicting and recognizing aggressive behavior and decreases the likelihood of an injury
occurring.
Hierarchical problems are most probably overdiagnosed, often because the two terms
‘dominance’ and ‘obedience’ are confused. These words are not synonyms. A dominant dog may
obey perfectly, but show aggressive behavior towards its owners at home anyway. (This aspect
will be further discussed in the chapter ‘dominance, hierarchy and aggression’.) However, there
are cases where the problem lies in the hierarchical structure of the system, for example with
owners who want to keep their dog ‘anti-authoritarian’ and thereby allowing their dog to decide
when and where which interaction or action shall take place. This is no problem, until one day,
the owners do not agree with what the dog wants - e.g. the dog does not allow the owner to get
into bed anymore - and then there is a clash. Pageat (1998) gives the label ‘sociopathy’ to
hierarchical problems within the same household and couples this disorder with three types of
aggression: hierarchical, irritation and territorial and/or maternal. The therapy here, however, is
not obedience, but a change in the social structure. Elements of obedience training may be used,
but should not be confounded with actual obedience.
4.4.3.5
Obedience and Education
Obedience and education of the dog are naturally very important to ‘keeping the dog under
control’. A dog under control, a dog that obeys is, in the presence of its owner, a smaller risk to
the surroundings. Even an obedient and well educated dog, however, can show aggressive
behaviors that may become dangerous, especially if the human’s aim is to have an aggressive,
dangerous dog.
4.4.3.6
Medications
Several medications can influence behavior. Dissociative anesthetics, such as ketamin, are known
to elicit hallucinatory eidolies in dogs, during which predatory or irritation aggression may be
shown (Dehasse 2002b). Certain psychopharmaca, also used to treat behavioral problems, may
act in a disinhibiting way (such as phenothiazines, butyrophenones, certain SSRIs). If an
aggressive tendency is present and the dog is disinhibited, the likelihood of this individual to
show aggression increases. Therefore, care must be taken in the choice of the medication used to
treat behavioral problems.
Ecology and Ethology
4.4.3.7
The natural needs of the canine species have to be taken into account and have to be fulfilled. As
mentioned in the section on ‘socialization’, a dog that is deprived of certain experiences (such as
contact with the outside world, because it was constantly kept in a closed kennel during
puppyhood) is likely to develop phobias or even general anxiety. This, in turn, can lead to certain
types of aggression, such as fear aggression, distancing aggression and irritation aggression. A
dog has certain biological needs such as a certain degree of exercise every day, a certain degree
of attention and play etc. If these needs are not fulfilled, the likelihood of behavioral problems,
including aggression, increases.
Nutrition
4.4.3.8
Several studies have been done looking at the influence of specific nutrients on the behavior of
dogs. Hughes & Duncan (1988, cited in Anderson & Marinier, 1997: 183) claim that
„inappropriate or insufficient feeding will result in physiological ‘hunger’ which often manifests
itself in behavior such as aggression, in-pack fighting…“.
22
Anderson and Mariner (1997) found a correlation between diet, restricted exercise and behavior
by treating dogs with behavioral problems with a specific diet and restricted exercise. Concerning
the diet, they found that increased digestible protein in the diet stabilized the behavior of the dogs
treated by them.
Dodman et al. (1996: 376) found that “a reduction in dietary protein content is not generally
useful in the treatment of behavior problems in dogs, but may be appropriate in dogs with
territorial aggression that is a result of fear”. They set up the hypothesis that low dietary protein
content might facilitate the amount of tryptophan passing from the blood into the brain.
Tryptophan is a precursor of serotonin, a neurotransmitter that has been linked to aggressive
behaviors. Low serotonin activity in the brain has been associated with increased aggression.
Additionally, tryptophan competes with other amino acids at the same transport mechanism from
the blood to the brain. If a diet has a high content of other amino acids, the competition for the
transportation of tryptophan into the brain is greater, proportionally less tryptophan will be
transferred into the brain, leading to a reduction in the content of serotonin and a possible
increase of aggression. However, every model linked to neurotransmitters must be regarded with
care, as the effect of only one of several dozen known neurotransmitters or neuromodulators is
considered, ignoring possible influencing or intervening effects of these. The behavioral effect
seen might not be the result of the action of only this one neurotransmitter that is being studied.
These authors searched the literature for publications showing the influence of fat content and
carbohydrate content on behavior. They did not find any evidence that fat has an effect on
behavior, however, carbohydrates seem to have one.
In a later study, DeNapoli et al. (2000) studied the effects of dietary protein content and
tryptophan supplementation on two types of aggression, namely, dominance aggression and
territorial aggression, and on hyperactivity in dogs. They found that the addition of tryptophan to
diets high in protein or the change to a diet low in protein may decrease aggression in dominant
aggressive dogs. As found in the earlier study, for dogs showing territorial aggression, a low
protein diet may be recommended, however, they do add that the supplementation of this diet
with tryptophan may help in reducing this type of aggression in dogs.
4.4.3.9
Context of a Situation
The context in which aggressive behavior must be considered when diagnosing the type of
aggression and the prognosis, when deciding whether the behavior is adaptive/normal or not
adaptive/pathological/not normal. In every case of aggression, one can ask oneself: is this a
normal dog in normal surroundings? A normal dog in an abnormal surroundings? An abnormal
dog in a normal surrounding? Or an abnormal dog in abnormal surroundings? All the influencing
factors mentioned above must be taken into account before definitely diagnosing the type of
aggression and formulating a prognosis.
23
4.5
DOMINANCE, HIERARCHY, OBEDIENCE AND AGGRESSION
Dominance and hierarchy are two words often associated with aggressive behavior. It is
interesting to note that many authors talk about dominance, hierarchy and dominance aggression
without defining the terms.
4.5.1 DOMINANCE AND AGGRESSION
In the general population it is often assumed that the dominant dog is automatically the more
aggressive and the more dangerous. The difference between aggression and dangerousness has
been discussed above. Here are explanations of why dominance is not equal to aggression.
Below is a selection of the many existing definitions of the term ‘dominance’:
• Dehasse (2002b: 14) defines dominance as “the capacity to attain privileges and to defend
them"*.
• Drews (1993) (cited in Lindsay 2001: 234): “dominance is an attribute of the pattern of
repeated, agonistic interactions between two individuals, characterized by a consistent
outcome in favor of the same dyad member and a default yielding response of its opponent
rather than escalation. The status of the consistent winner is dominant and that of the loser
subordinate.”
• Overall (1997: 115): "Dominance is a concept found in traditional ethology that pertains to an
individual's ability, generally under controlled conditions, to maintain or regulate access to
some resource."
• Lindsay (2001: 238) himself notes that “dominance typically denotes a social relationship
based on a regular exchange of species-typical threat and appeasement signals between at least
two individuals”.
All these definitions tell us that dominance
• has to do with privileges / resources
• has to do with defending, maintaining or regulating access to these privileges / resources
• is a characteristic of a relationship, not of an individual.
• is coupled with submission/appeasement of the other partner in the relationship.
What is striking though is that none of these definitions clearly includes all these points. Dehasse
uses the word ‘privilege’, Overall ‘resource’ to describe the defense or regulation and
maintaining of privilege resp. resource, but it neither explicitly states that this is a characteristic
of a relationship, not of an individual. Only through implication is it clear that one can only
defend and control access to something if another individual is present, implying the necessity of
a relationship between two individuals.
In Drews’ definition as well as in Lindsay’s, the aspect of ‘a third party’, namely a privilege or a
resource, is not considered, dominance is an attribute of an interaction. The fact that the
dominance-subordinate relationship is dependent on the context and that it may change over time
does not become clear in any of the definitions. Drews limits the communication of the two
individuals to agonistic behavior and claims the outcome to be consistently in favor of the
dominant, emphasizing the necessity of repetition for a relationship to be stabilized. This is
certainly necessary for a stable hierarchical structure to be built up in a social system. However, a
dominance-subordinate relationship can be established in a specific context between two
24
individuals in different manners, such as a dominant representatio of a trainer, by applying force,
by the dog coupling the present situation with experiences (Dehasse 2002a). It seems that the two
words ‘dominance’ and ‘hierarchy’ are being slightly confused in Drews’s definition. It is not an
attribute of dominance that escalation is avoided, but an attribute of the communication between
the two individuals involved, leading to a clear relationship between the two individuals, i.e. one
being dominant over the other. If both sides accept this, the danger of escalation is diminished.
As Overall (1997: 115) puts it “…it is important to realize that it [dominance] is not defined as
aggression on the part of the ‘dominant’ individual, but rather as the withdrawal of the
‘subordinate’”. She continues “the behavior of the lower status individuals, not the higher
ranking one, is what determines the relative hierarchical rank”. Dominance and hierarchy are not
the same thing. Dominance alone does not lead to avoidance of serious fights. Avoidance of
escalation is one of the advantages of a stable hierarchy, every member knows its place and its
tasks. As Dehasse (2002a: 87) states “a hierarchy allows to precise limits and to appease the
members of the group”*.
One animal alone cannot be described as being dominant or submissive, there has to be an
interaction of at least two individuals within a certain context. Aggression refers to a certain
behavior of an individual, dominance to the relationship two individuals have to each other. They
are two different concepts describing two different things and cannot be used as synonyms.
The aspect of context is important to note as well. Within a relationship of two dogs, dog A may
be dominant over dog B in a specific context, such as in the presence of food, whereas dog B
may be dominant over dog A in the context of a privileged place to rest. Dog A may be dominant
over dog B in the presence of food in normal circumstances, if however, dog B is starving, this
dog will fight for the food and might win access to it. Overall (1997: 115) emphasizes the fact
that dominance may not be confused with status and that it is not necessarily coupled with the
“priority of access to resources”.
The term ‘dominance aggression is a widely used in the world of behavioral medicine in the
Anglo-Saxon school; the French school does not recognize this notion. This term is confusing, as
it suggests that this type of aggression is based on a dominance status of the dog in a hierarchy.
Overall (1997: 512) defines dominance aggression as “abnormal, inappropriate, out-of-context
aggression (…) consistently exhibited by dogs towards people under any circumstance involving
passive or active control of the dog's behavior or the dog's access to the behavior”. Describing
this type of aggression in detail passes the scope of this study, but it seems important to point out
the fact that dominance aggression is not the same as dominance. Overall (1997: 116) hypotheses
that it is a manifestation of an underlying anxiety disorder and that it “is about the concept of
control, not about overt challenges…or ambiguous challenges…”.
4.5.2 HIERARCHY AND AGGRESSION
Aggressive problems are often assumed to stem from management problems, including problems
of the hierarchy, i.e. the dog being dominant over its owners. The notion of hierarchical problems
is very well recognized in the field of behavioral veterinary medicine (Pageat: hierarchical
aggression, problems of hierarchy (sociopathy), Dehasse: hierarchical aggression), but do not
seem to be diagnosed by veterinarians as often as by dog trainers or animal psychologists.
Overall (1997: 1) states that “it is inexcusable and irresponsible to advance poor management as
the primary cause of behavioral disorders”.
25
Hierarchy is the organization of several individuals into social ranks, “to minimize social contests
and prevent the outbreak of overt and potentially damaging fighting between competitors,
thereby laying a foundation for social order and harmoniously organized group activity” (EiblEibesfeldt 1979 cited in Lindsay 2001: 235). Hierarchy is the whole of the dominance –
subordinate relationships within a pack. This hierarchy is not maintained by constant battles
between all the members of the (family-) pack, but by a functioning communication among the
individuals of the pack that allows every member to know where it stands relative to the other
members of the pack, what its privileges are and what ‘language’ (i.e. ‘dominant’ or ‘submissive,
appeasing’) it is supposed to use when communicating with which individual of the pack
(Dehasse 2002a).
In the Oxford Universal Dictionary (1955), hierarchy is defined as “a body of persons or things
ranked in grades, orders, or classes, one above another.” This definition suggests a linear
organization of a hierarchy. For canines, it is not as simple as that. Several theories exist on how
the social hierarchy is organized in dogs and wolves. Dehasse (2002a) represents the model of a
separate, linear hierarchy among the adult males and females of the pack. There is an alphaindividual in both these categories, followed by a linear hierarchy with an omega animal ‘at the
bottom’. Below this ‘adult hierarchy’, the whole lot of adolescent dogs are found, regardless of
their sex. Puppies seem to be ‘excluded’ from the hierarchy as to that they have many privileges
normally dominant dogs maintain and defend. (Dehasse, 2000: 43, summarized in Fig. 1.2)
Figure 1.2 Social organization within a dog pack according to Dehasse (2000)
α
α
β
β
ω
ω
Adolescents
Puppies
Lindsay (2001: 236) states that “Among wolves, social ranking is mostly non-linear in
structure…Instead of a neatly defined hierarchy of status relations based on individual
competitive success, wolf-pack social organization is affected by other less obvious political
factors, such as kinship relations (dependent ranking) and various coalitions exerting pressures on
the dominance structure.”
26
Hierarchy is often explained referring to the wolf living in packs in the wild or in wild life
reserves. As put into question above: can the social structure of wolves simply be extrapolated to
pet dogs? It is often claimed that pet dogs see their human family members as their pack,
implying that they do not make a difference between dogs and humans (and other animals).
Feddersen-Petersen (2001a) found that poodles showed a high degree of agonistic behaviors
towards conspecifics, however less towards humans. She concludes that dogs are capable of
differentiating between a conspecific and a human as a social partner and thereby deducts that the
above mentioned hypothesis is not quite as simple as it sounds. Dehasse (2002a: 88) refers to the
cohabitation of dogs and humans as ‘la famille-meute’, the “family-pack” suggesting the
functional hypothesis that the family may be looked at as sort of a pack, but not as a pack of
dogs. Here again, there is a need of an ethogram of the pet dog living in the ‘family-pack’. The
relationship within the social structure of more than one species living together becomes even
more complex. Dehasse (2002a) and Lindsay (2001) describe several possible hierarchical
structures within a ‘family-pack’. It is an important fact in order to avoid problems that the
human members of the family-pack understand the social structure and communication of their
canine family members.
4.5.3 OBEDIENCE AND AGGRESSION
Dunbar (cited in Overall, 1997: 1) classifies problems with behavior into three categories:
“temperament problems, behavior problems, and obedience problems”. Disobedience and
dominance are often confused. It is true that it might be easier for the owner to ‘control’ a dog
that takes a submissive role towards him, because it is clear to the dog that it is its ‘duty’ to do
what the dominant says. If the hierarchical situation is not clear within the family pack, it will be
the dog with the tendency to more self-secure behavior that will try to ‘climb up’ the hierarchical
ladder and dominate one or several of the human family members. This is not a problem until the
day comes when the human wants, for example, to get a snack from the refrigerator and is
confronted with a growling dog. This dog has a privilege (the food in the refrigerator) and is
defending it and controlling the access to it by its owner. According to the definition of
dominance of Dehasse and Overall, this dog is dominant over the owner in this particular
situation. It may, however, otherwise be a completely obedient dog. For example, it might even
react to commands such as ‘sit’, but continue the growling and threatening behavior at the same
time. This is one example of why obedience is not necessarily connected to less aggression and
less dominance. Overall (1997: 89) underlines this by saying “It is important to realize that dogs
exhibiting inappropriate, out-of-context aggressions are not misbehaved or poorly behaved – they
are clinically abnormal and must be regarded as such”.
Obedience does not actually decrease aggressive behavior, but it may decrease the danger the dog
presents by showing aggressive behavior, as in the following situations:
• Having the dog under control diminishes the likelihood of an incidence occurring;
• Challenging the dog with something in a particular moment redirects its attention from a
possible dangerous situation to its owner;
• Working with the dog in any way, if on the basis of reinforcement, increases the interaction
time with the dog and therefore is a good basis for a healthy attachment to develop. An intact
hierarchy is only possible, if there is an attachment among the members of the social system.
(Dehasse 2002a)
Viewed from another angle, an aggressive dog might become more controllable with obedience
training, but it will not become less aggressive. Although the therapies recommended by
therapists often seem like obedience tasks, their aim is a different one. The goal of obedience
27
training is that the dog ends up doing what we want from it. Overall (1997: 408) explains that the
goal of behavioral therapy is to allow the dog to relax and thereby render it receptive to learning
new, more appropriate behaviors.
4.6
FEAR, SUBMISSION AND AGGRESSION
A dog showing submissive behavior is often misinterpreted to show fear. These two aspects
cannot be equated, as submission results from an interaction between at least two individuals, i.e.
is an attribute of a relationship, whereas fear is an emotion shown in a specific context. What
these two behaviors might have in common is the low posture the dog tends to show in both
situations.
Fear and aggression may be linked leading to the notion of ‘fear aggression’ which is recognized
by most authors as a category of aggression (Dehasse 2002b, Lindsay 2001, Pageat 1998, Houpt
& Wolski 1998, Overall 1997, Landsberg, Hunthausen & Ackerman 1997, Askew 1996). The
characterstics of this type of aggression in almost all definitions are:
• Signs of fear (Dehasse 2002b, Houpt & Wolski 1998, Lindsay 2001, Landsberg, Hunthausen
& Ackerman 1997)
• Aggression shown in situations where flight is not possible (Dehasse 2002b, Lindsay 2001,
Pageat 1998)
• Aggression shown in hopeless situations, i.e. situations where every other behavior is
ineffective (Dehasse 2002b, Lindsay 2001, Pageat 1998, ), Landsberg, Hunthausen,
Ackerman 1997)
Dogs showing aggression out of fear can be very dangerous, as the aggressive behavior is their
last resort and their behavior is not controlled which can lead to very serious injuries (Dehasse
2002b).
4.7
PLAY AND AGGRESSION
Here also, the boundaries are very fluid – where does play become aggression? What is play?
Lindsay (2001: 250) says, “play offers a powerful nonintrusive means to control the direction of
social polarity and attention, to balance affection and leadership, and to increase feelings of
affiliation and cooperation between people and dogs. Play is relatively incompatible with
aggression and fear, although, under the influence of escalating frustration or threat, play may
slip over into overt aggression“. The two behavioral elements encompass similar body postures
and behavioral sequences. As already mentioned, it is thought that puppies learn aggressive
communication and regulation through playing with their siblings and their mother (Dehasse
2002a), however, the dogs have ways to signal ‘don’t take this seriously, it is only play’. Lindsay
(2001: 251) refers to these as “cutoff or compromise signal” that “…is not a submissive gesture
but an opportunity for the contestants to call a draw and disengage without loss or gain”. A
typical example for such a cutoff signal is the play bow.
Play is more likely to escalate into aggression if two unfamiliar dogs play together for the first
time or when one of the play-mates decides he or she had enough before the other is ready to quit
(Lindsay 2001).
28
4.8
CLASSIFICATIONS – TYPES OF AGGRESSION
There are several existing options to define and classify behaviors, including types of aggressive
behavior. The following definitions of the different types of classifications are taken from
Dehasse et al. (2003a):
Descriptive
Functional
Causal-motivational
Contextual
Neural
Learning
Complex and/or diagnosis
Therapeutic
The definitions of the several authors will be analyzed as to which of Dehasse’s definition
categories they include. The aim is to show tendencies, showing which authors use which types
of definitions. The question remains, of whether it is possible to look at a definition independent
of its author and attempt to determine the categories of definition used. Some authors specify
what type of classification they use, e.g. Dehasse et al. (2003a) claim to use a descriptivecontextual classification, whereas Landsberg et al. (1997) say they use a functional one. The
results of the analysis done in this chapter indicate that, according to the definitions used here, the
definitions of Dehasse et al. (2003a) also include quite a bit of causal-motivational aspects, and
Landberg et al.’s (1997) definitions would more likely be classified as contextual than functional.
One is confronted with the choice of relying on what the author says he or she is using and
ignoring the other aspects that possibly can be found or including all the aspects one sees in the
definitions, but concentrating on what the authors emphasizes as being important?
This chapter analyses the definitions of the authors independent of what they claim to be using,
and then compares the results with what the author says he or she uses.
4.8.1 DESCRIPTIVE
Here the behavior unit shown by the animal is described. This is the most widely used way to
define behaviors.
4.8.2 FUNCTIONAL
“The functional definition adds the observer’s interpretation of the function of the animal’s socalled intent and is shaped by sociobiology models…. The functional classification postulates the
hypothesis of biological function, i.e. of adaptive fitness.” (Dehasse at al. 2003a: 3)
A functional definition is an interpretation of the hypothesized function of the behavior.
However, a behavior unit may seem to have several different functions, in these cases, the most
apparent or most logical hypothesis of the observer may be chosen. (Dehasse et al. 2003a)
Example:
• Distancing aggression (Dehasse 2002b): “essentially proactive aggression against
conspecifics or other animals (and humans) (particularly unknown or from other groups)
having the apparent function to keep them out of the security distance of the aggressor dog.” *
29
4.8.3 CAUSAL-MOTIVATIONAL CLASSIFICATION
“This classification sorts the aggressive behaviors by their hypothesized trigger (or eliciting)
factor, motivation or cause. This is a modeled categorization.” (Dehasse et al. 2003a). Dehasse
(2003a) classifies this type of definition into several sub-categories:
affective-emotional
• Fear aggression (Overall, 1997: 512): “Aggression that consistently occurs concomitant with
behavioral and physiological signs of fear as identified by withdrawal, passive, and avoidance
behaviors associated with the sympathetic branch of the autonomic nervous system.”
Cognitive
• Possessive aggression (Overall 1997: 516): “Aggression that is consistently directed towards
another individual who approaches or attempts to obtain a nonfood object or toy that the
aggressor possesses or to which the aggressor controls access.”
Psychological
• Psychological disturbance-related aggression (Overall 1997): “aggressive behaviors that do
not meet the criteria for specific aggressions.” [Causal, but very imprecise]
Sensorial
• Pain aggression (Overall, 1997: 516): “Consistent aggressive behavior, in excess of that
required to indicate concern and to effect restraint, demonstrated only in a context known or
potentially associated with pain, but that may not be painful itself.”
•
•
Space control
Distancing aggression (Dehasse 2002b): “Aggression shown towards humans or dogs that
do not belong to the family group (or individuals that have been seen regularly since
puppyhood) (85)“. The effect of the aggression is to keep the individuals at the security
distance (78)”.
Social
Hierarchical aggression (Pageat 1995: 146): “Behaviors that are triggered in situations of
hierarchical competition…The sequence of this aggression is typically composed of three
phases: phase of intimidation, phase of the attack, phase of appeasement”.*
Somatic
• Atypical aggression: unspecific aggressive behavioral sequence not classified elsewhere, and
due to a somatic disorder, such as a metabolic (hypoglycemia, porto-systemic shunt), an
endocrine, an inflammatory/infectious, a degenerative, a tumoral disorder, painful disorder,
etc.
30
4.8.4 CONTEXTUAL
“This category groups aggressive behaviors correlated to a specific repetitive context. It does not
analyze the aggression in detail, but only the circumstances surrounding it.” (Dehasse et al.
2003a).
Example:
• Play aggression (Overall 1997: 516): "consistent aggression that occurs in contexts in which
play behaviors (…) would normally occur.”
4.8.5 NEURAL
“This categorization is based on the experimental discoveries of genetically determined “hardwired” neural circuitry organizations.” (Dehasse et al. 2003a)
Example:
• Fear aggression (Pageat 1995: 149): behavior under the control of the anterior hypothalamus
and ventromedian nucleus. *
4.8.6 LEARNING
This is a category, which is difficult to circumscribe, as almost all types of aggression are
influenced by learning processes (classical conditioning, instrumental learning …)
Example:
• Hyperaggression (Dehasse, 2002b: 153): “Secondary hyperaggression is the
instrumentalized consequence of another type of aggression. There is a loss of adaptive
capacity, a regression of intelligence level to the state of reflexes.” *
“Tertiary hyperaggression is the aggression taught to a dog by a trainer.”
4.8.7 COMPLEX AND/OR DIAGNOSIS
“Diagnoses of problem behaviors are not descriptions of behavioral events, but are hypotheses of
useful units based on several signs. They do not imply an underlying mechanistic
phenomenon”(Overall, 1997: 3-4). “The diagnoses classification pretends to be functional. In
fact, it is essentially a contextual, non-causal, classification.” (Dehasse 2003a).
Example:
• Dominance aggression (Overall 1997: 512): “Abnormal, inappropriate, out-of-context
aggression consistently exhibited by dogs towards people under any circumstance involving
passive or active control of the dog’s behavior or the dog’s access to the behavior.”
4.8.8 THERAPEUTIC
Definitions based on the effects of therapies and drugs on the problem behavior. As of this time,
no such classification exists.
31
4.8.9 OPERATIONAL CLASSIFICATIONS
As opposed to the preceding possibilities of defining and classifying types of
behaviors/aggression according to the observed behavior, or to the individual and/or context
itself, the operational definition depends on the author. The author defines what he or she means
with a particular word, which makes the definition operational for his or her model. However, the
same word may have a completely different meaning for another author, but, with the definition
of this author, be operational for his or her model as well. The lack of universal definitions is a
big problem in the world of behavior.
(Immelmann & Beer 1989 cited in Barrows 2001 cited in Dehasse et al. 2003a).
Examples:
• Not operational: Anorexia (“absence of appetite for food” (Tilley & Smith 1997)). If this
word is used, it will mean the same thing to everybody in the medical field.
• Operational: Parental aggression may mean aggression towards offspring (Wittenberger
1981) or aggression to defend offspring (Archer 1988).
4.8.10 VETERINARY MEDICINE 7 8
FRENCH
*
4.8.10.1 Pageat (1998)
Pageat recognizes six groups of aggression:
• Predatory aggression
• Hierarchical aggression
• Irritation aggression
• Territorial aggression
• Maternal aggression
• Fear aggression
Pageat uses mainly the complex and/or diagnosis way of defining the types of aggression (36%)
and the causal-motivational type (29%). These characteristics of a definition are combined in
three of his definitions of types of aggression.
*
4.8.10.2 Dehasse (2002b)
In his book “Le chien aggressif”, Dehasse (2002b) suggests two ways of classifying aggression:
1) Defensive and offensive aggression depending on the relative direction of movement of the
individuals involved in the aggressive sequence (personal communication: preference of
‘proactive aggression’ instead of ‘offensive aggression’)
7
8
For the definitions of the types of aggression mentioned in this section see Appendix A
For the results of the analysis of the types of definitions see Appendix B
32
•
•
•
Atypical aggression
Defensive aggression
o Irritation, frustration-induced and pain-induced aggression
o Fear aggression
o Territorial defense aggression
o Litter defense aggression (maternal aggression)
o Distancing aggression
Offensive aggression
o Competitive aggression
o Hierarchical aggression
o Possessive aggression
o Redirected aggression
o Predatory aggression
o Aggression of pursuit
o Instrumental aggression, hyperaggression
o Aggression on command
o Extragroup aggression
2) Aggressed subject: towards humans, towards dogs, towards other animals, etc.
In his 2002 book, Dehasse relies mainly on causal-motivational (33%), descriptive (30%), and
contextual (27%) definitions whereby two or all of these aspects are found in combination in 9 of
the 17 definitions of aggression.
In Dehasse et al. (2003a), a new classification of aggressions is proposed based on descriptivecontextual definitions. There is a division of the types of aggression into two main groups relative
to the threatening motion: proactive versus reactive. The types of aggression are listed in a
sequence from mild to lethal consequences.
•
•
•
•
•
Education of the young/parental disciplinary and weaning aggression
o Parental weaning aggression
o Parental disciplinary aggression
o Parental educative aggression
Play aggression
o Play fighting
Competitive-social aggression
o Food-elicited aggression
o Non-food object elicited/possession aggression
o Resting area-elicited aggression
o Social interaction/alliance-control elicited aggression
o Space-control-elicited aggression
o Sexual-control-elicited aggression
o Other resource-elicited aggression, non specified
o Dueling fights:
dominance aggression
Intrasexual (intraspecific intergroup) aggression
Sexual aggression
33
•
•
•
•
•
•
•
•
•
•
•
Irritation (irritable) aggression
o Frustration-related aggression
o Pain-related aggression
Defense of young
o Maternal aggression
Defense of space/space-eliciting/managing aggression
o Distancing/deterring aggression
o Group-defense aggression
o Territorial aggression
Redirected aggression
Pursuit aggression
Critical aggression
o Fear-related/elicited aggression
o Antipredatory aggression
Learned aggression
o Instrumental aggression
o Trained aggression
Mobbing/ganging
Infanticide
Predatory aggression
o Group hunting
Atypical aggression
o Overactivity-related aggression
o Somatic disorder-induced hyperaggression
o Idiopathic aggression
o Disocialisation-related aggression
Here, a contextual aspect of the definition can be found in 44% of the cases, causal-motivational
in 30% and descriptive in 23%. This is quite surprising, as the author’s aim was to create a
descriptive-contextual classification. When analyzing the results, we find that causalmotivational and contextual aspects are combined in 16 of the 38 definitions and a combination
of descriptive and causal-motivational aspects is found in 3 definitions, the causal motivational
aspect alone is only found in 2 cases. As the causal-motivational aspect appears in combination
with either descriptive or contextual aspects in all but two cases and the authors emphasize that
the definitions are to be understood in a descriptive-contextual way, not too much emphasis
should be put on the possible causal-motivational aspect. However, the question remains - can it
be ignored?
ANGLO-SAXON
4.8.10.3 Overall (1997)
Overall (1997: 102) categorizes the types of aggression according to Young (1988), after
Borchelt & Voith (1982) and based on Moyer (1968). She views these definitions all as
“diagnoses: none of the aggressions represent normal, appropriate, in-context behaviors” (Overall
1997: 102).
34
•
•
•
•
•
•
•
•
•
•
•
•
Maternal aggression
Play aggression
Fear aggression
Pain aggression
Territorial and protective aggression
Interdog aggression
Redirected aggression
Food-related aggression
Possessive aggression
Predatory aggression
Dominance aggression
Idiopathic aggression
Overall (1997: 102)mentions in her book: “ ... aggression can be viewed as a range of solutions to
certain problems (Archer 1988); hence, functional and causal groupings of aggressions can be
based on contextual behaviors (Brain 1981)". She uses a functional classification which “enables
the clinician to study constellations of signs that are highly correlated with specific behavioral
traits, allowing the clinician to formulate diagnoses (hypotheses) that can be tested in a rigorous
and paradigmatic manner through treatment.
Correlation is not analogous to underlying causality, ... ” (3) As she mentions herself, mainly
complex or diagnosis and contextual definitions (each 32%) are used, which are defined by her as
functional, whereas in the classification of Dehasse et al (2003a) they would be classified in the
categories of complex or diagnosis and contextual definitions. A combination of the two aspects
is found in 11 of the 13 definitions of aggression.
*
4.8.10.4 Feddersen-Petersen (2000b)
Feddersen-Petersen cites Hassenstein (1980) who says that “aggression can be the expression of
very different internal conditions or a varying readiness to show a behavior (72)“ and that it can
be the result of multiple motivations. This can be interpreted in the way that she mainly relies
causal-motivational characteristics to define a type of aggression.
She differentiates between intraspecific and interspecific aggression. And recognizes the
following categories:
• Aggression as general defense
• Readiness to aggress prey animals
• Aggression as a counterattack
• Aggression towards sexual rivals
• Aggression when defending a territory
• Aggression as a test of power
• Group aggression
• Aggression because of frustration
• Social aggression
According to the definitions used in this analysis, Feddersen-Petersen mainly relies on contextual
(46%) and causal-motivational (39%) aspects when defining types of aggression. In three of the
five definitions she uses both aspects in combination, in the remaining two, it is one or the other
in combination with a functional aspect.
35
4.8.10.5 Landsberg, Hunthausen & Ackerman (1997)
According to Landsberg, Hunthausen & Ackerman themselves, the definitions are mainly
functional. However, 75% of the characteristics are contextual and 25% causal-motivational and
none are functional according to the definitions of categories used in this analysis. Landsberg et
al. define a functional classification as one that “takes into account the circumstances in which
the aggression occurs as well as the organisation of behaviour patterns involved. This allows for
inferences concerning the motivation for the behaviour as well as aiding in uncovering the factors
leading to the behaviour, thus facilitating the fabrication of a solid prognosis and treatment plan.”
(129). This definition does not agree with the definition for ‘functional’ Dehasse et al. (2003a)
use in their article. However, when analyzing these definitions further, what Landsberg et al.
define as ‘functional’, is defined as ‘contextual’ and ‘causal-motivational’ by Dehasse et al.
(2003a). This shows how important it is to clearly define what is meant with the terms used.
•
•
•
•
•
•
•
•
•
•
•
•
•
Dominance aggression
Possessive aggression
Fear aggression
Territorial or protective aggression
Predatory aggression
Pain-induced aggression
Play aggression
Maternal aggression
Redirected aggression
Intraspecies aggression: intermale, interfemale, social status aggression
Pathophysiological aggression
Idiopathic aggression
Learned aggression
4.8.10.6 Houpt & Wolski (1998)
Houpt & Wolski use the following classification of aggression:
•
•
•
•
•
•
Social or dominance-related aggression
Territorial aggression
Pain-induced aggression
Fear-induced aggression
Maternal aggression
Predatory aggression
These definitions are mainly functional (46%) and contextual (31%), using either one or the other
or both in combination.
36
4.8.11 PSYCHOLOGY
Several psychologists have published books on the behavior of dogs. Two examples, Askew and
O’Farrell were chosen for this analysis.
4.8.11.1 Askew (1996)
Askew differentiates the two main categories ‘intraspecific aggression’ and ‘interspecific
aggression’. Within the group of intraspecific aggression, he defines two sub-categories:
‘intragroup’ and ‘extragroup’ aggression.
Askew says he uses proximate, function-related definitions and places some forms of aggression
into more than one major category.
•
•
Intraspecific aggression
o Intragroup aggression
Competitive aggression (dominance-related or possessive)
Self-protective aggression
Defense of young
Playful aggression
o Extragroup aggression
Competitive aggression (dominance-related or possessive)
Self-protective aggression
Group-defensive aggression
Interspecific aggression
o Competitive aggression (interspecific competition)
o Self-protective aggression
o Group-defensive aggression
o Predatory behavior
As does Overall, Askew adopts the classification of Borchelt & Voith (1982). He claims his
definitions are based on the eliciting situations which would correspond to the contextual
category used in this study, and on the etiology of the behavior shown, which is part of the
causal-motivational category. He mentions however, that his system of classification is a
functional one.
Analyzing the definitions of these types of aggression leads to an agreement with the claims the
author makes. Fifty-five percent of the definitions include a contextual characteristic, 21% a
causal-motivational and 18% a functional. He also often combines several aspects in one
definition. Functional and contextual are combined in 4 of the 21 definitions, most frequently,
however, the definitions are purely contextual (11 of 21).
37
4.8.11.2 O’Farrell (1991)
O’Farrell distinguishes two main categories of aggression: ‘dominance aggression’ and ‘prey
aggression’.
•
Dominance aggression: towards dogs (in the same household (males and females) vs.
different households (mostly males))
o Territorial aggression
o Protective aggression
o Pain-induced aggression
o Maternal aggression
• Prey aggression:
o Towards dogs
o Towards humans
o Towards other living beings
These definitions consist mainly of contextual aspects (40%), however descriptive, functional
and causal-motivational aspects are found in each 20% of the definitions as well. In three of the
six definitions, combinations of two or three of the above mentioned categories are found.
4.8.12 CYNOLOGY
4.8.12.1 Lindsay (2001)
Lindsay makes the differentiation of intraspecific versus interspecific aggression, however, he
does not use this notion to categorize the types of aggression. The listing is in an alphabetical
order.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Avoidance-motivated
Control (dominance)-related
Dysfunctional
Fear related
Idiopathic
Instrumental
Intermale/Interfemale
Irritable
Low threshold
Maternal
Pathophysiological
Playful
Possessive
Predatory
Protective
Redirected
Territorial defense
Trained
Xenopic (fear related)
38
Lindsay claims to use descriptive and functional characteristics to define aggression. Analysis of
his definitions leads to the conclusion that, according to the definitions used in this analysis, he
uses mainly contextual (56%) and causal-motivational (22%) characteristics to define the types of
aggression in his book.
4.8.12.2 Campbell (1999)
Campbell distinguishes the following types of aggression:
•
•
•
•
•
•
•
Aggression towards owner
Aggression towards outsiders
Overprotectiveness
Aggression towards other dogs
Predatory behavior (Killing other animals)
Chasing Vehicles and other moving objects
Sexual mounting and jumping up
Eighty-three % of these definitions include causal-motivational characteristics and 17%
contextual ones.
4.8.13 ETHOLOGY
*
4.8.13.1 Zimen (1992)
In ethology, several ‘Funktionskreise’ exist. This is a model to classify (normal) behavior in
order to better understand the function of every behavior. The word ‘Funktionskreis’ includes the
aspect of function leading to the assumption that we might expect a functional classification of
aggressive behavior of ethologists. Zimen says: “A dog's attacks towards conspecifics, towards
other animals or humans can be attributed to different ‘Funktionskreise’: Hunting for prey,
territorial defense, social behavior within a group and sometimes also play behavior (384)”.Not
only the function, but also the motivation is an important aspect in ethology, however, Zimen
continues that “often, it is not impossible to distinguish the separate motivations ... ” (384) and
that "a variety of factors are involved in every dog attack” (285).
•
•
•
predatory aggression: towards „not humans“ versus „humans“
socially motivated aggression towards humans
territorial aggression: towards „not humans“ versus „humans“
It is interesting to note that 50% of the definitions include a contextual aspect and 25% a
descriptive one. Only 13% each involve functional or causal-motivational characteristics
according to the definitions used in this analysis.
39
4.8.14 DISCUSSION OF THE CLASSIFICATION OF AGGRESSION
It is clear that there is a lack of universally accepted definitions in the field of behavior in
general. The diversity of existing definitions and classes of definitions inevitably leads to
misunderstandings amongst professionals and nonprofessionals. The aim should be definitions
with as little subjective value, as little imprecision as possible. In their article, Dehasse et al.
(2003a) propose a descriptive-contextual classification as fulfilling this requirement.
A first point to consider when agreeing on definitions is: What is the aim of defining the
particular word? For whom is it being defined?
Owners of dogs showing behavioral problems (such as aggression) are often interested in the
cause, the motivation and function of the behavior shown. They want to understand what is going
on with their pet; It is human nature to want to understand, to look for causes in everything. In
this respect, a functional or causal-motivational definition might be appropriate. However, a great
disadvantage of the use of this type of defining is the subjectivity, the interpretation and inference
inevitably associated to it.
Take the following example: A dog is diagnosed to show ‘dominance aggression’ and let us first
choose O’Farrell’s (1991: 97) causal-motivational definition of this type of aggression: “This
aggression is normally triggered by actions of the attacked individual, that are considered as a
threat to the dog's dominance…”*. She interprets the emotions the dog might feel in a certain
situation, the dog feels threatened. Do we know for sure? We might be able to extrapolate from
earlier experiences with dogs, from our own experiences, from research done in the literature,
but, we will never be sure if what we consider a threat is actually also one for the dog and vice
versa. We do not know if what one person considers a threat to the dog would also be considered
a threat by another person.
If a human infant cries, we might interpret its behavior to mean it is hungry, or has a
stomachache, or has wet diaper - however, we will never know for sure what the actual reason or
the ‘aim’ of the infant’s crying was, we can only guess. This type of definition might satisfy the
client, as he or she has found a cause and can understand his or her pet’s behavior, it should not
be satisfactory for a professional, as he or she should rely on objective information.
For the professional involved with canine behavior, a descriptive, contextual definition is of
greater use, as it is objective and describes only what is seen in which circumstances. These types
of definitions allow communication amongst professionals working throughout the different
fields of (canine) behavior, such as the veterinarian, the ethologist, the dog trainer, etc. Of course,
it is also important to define the triggers of the problem behavior in order to be able to
successfully treat an animal. However, this aspect need not necessarily be included in the
definition of the behavior, as it is specific to each individual.
4.9
DRUGS FOR TREATING PROBLEMS OF AGGRESSION
There is an abundance of the information on neurotransmitters, which possibly have an influence
on aggressive behaviors. In this review, I will focus on the few neurotransmitters, which are
thought to be affected by the drugs most commonly used to treat (aggressive) behavior problems
in dogs, serotonine, dopamine, noradrenaline and GABA (gamma-amino-butyric-acid). The
effects of these drugs have not been proven to be linked to the influence of the substance on a
specific neurotransmitter, as one specific substance is quite likely to modify the activity and
availability of several neurotransmitter systems at once. It must also be emphasized that the
classifications of drugs into groups is an attempt to create a simplified model for the human brain
40
to easier understand the effects of drugs, however, substances grouped in the same category do
not necessarily have the same effects when actually applied.
If one looks at the drugs available to the behaviorist veterinarians, one sees that only a limited
nubmer of these drugs are in fact used. The following drugs seem to be used in 80% of the cases:
fluoxetine, fluvoxamine, clomipramine, carbamazepine, selegiline and cyproterone. There are a
few variations from one prescribor to the other with some of them adding alprazolam,
phenobarbital, amitryptiline, pipamperone and other drugs.
Among the major drugs all belong to the group of serotonine reuptake inhibitors (SRI), specific
or unspecific. This does not mean that all SSRI drugs are useful to treat aggressive problems,
since sertraline, paroxetine, citalopram and others may even increase aggressive behaviors. It is
possible that the way these molecules act may be different from their SSRI effect. From this, one
may conclude that the effect on neurotransmission is not of much practical use in the decision of
what drugs to prescribe (Dehasse 2002b).
4.10
CONCLUDING REMARKS
Several alternatives exist to address aggressive behavioral problems in dogs. It is reasonable to
assume that no one approach suits every owner, every dog, every problem or every situation. This
study has given an overview of overlapping disciplines, pointing out the differences between the
methods and providing information which allows the clinician and the client to select which
approach, or combination of approaches would be most suitable for the individual case.
41
5. OVERVIEW TO EVALUATING (AGGRESSIVE) PROBLEM
BEHAVIORS OF DOGS ACCORDING TO A FEW
SELECTED MODELS
5.1
OVERVIEW OF CHAPTERS
Problem behaviors in dogs have been discussed in the literature of many fields (e.g. veterinary
behavioral medicine, cynology and psychology); However, a general overview of different
approaches does not yet exist. A brief overview of the issues discussed in several of these
different approaches will be given here, including notes of which areas seem to be given
particular weight by the different authors.
For this overview, a selection was made of several textbooks on problem behaviors in dogs in the
areas of veterinary behavioral medicine, psychology and cynology. No books could be found on
this subject written by an ethologist.
The eighteen criteria looked at in theses sources were the following:
Epidemiology
Origins, History, Domestication
Normal canine behavior
Neurobiology
Psychopathology
Consultation / approaching problem behaviors
Learning Theories
Biological therapies/medication
Therapies
Categorization of Problems
Prevention of behavioral problems
Involvement of owner
Legal Aspects
Social work and behavioral problems
Future of treatment of behavioral problems
Glossary
Not canine behavior (i.e. feline behavior)
Miscellaneous other subjects
In order to calculate the weight, which is given by individual authors to these criteria, the
proportion of pages dealing with these each criterion was calculated per total number of pages of
the publication.
An overview of the results is presented in Table 1.2. This overview is meant to offer a quick
reference tool to determine the centers of interest of the selected authors, representing certain
schools of thought. As the reporting of the detailed descriptions in each of the categories of this
table would be beyond the bounds of the present report, they are being prepared for presentation
on a website and CD-ROM.
Table 1.2 Themes treated in books of 6 authors
42
Approach/
theme
Education
Total number
of pages
(minus index)
not canine
%
Epidemiology
%
Origins,
History,
Domesticatio
n
%
normal canine
behavior
%
Neurobiology
%
Psychopathol
ogy
%
Consultation /
approach
%
Learning
Theories
%
biological
therapies/med
ication
%
Therapies
%
Categorizatio
n of Problems
%
Prevention of
behavioral
problems
%
Owner
%
Legal aspects
%
Social work
and
behavioral
problems
%
Future of the
treatment of
behavioral
problems
%
Glossary
%
Appendixes
%
Other
%
Askew (1996)
Lindsay (2000
/ 2001)
Dog trainer
O'Farrell
(1991)
Psychologist
Overall (1997)
Pageat (1998)
Psychologist
Landsberg et
al (1997)
Veterinarian
Veterinarian
Veterinarian
323
198
706
129
530
366
61
17.78%
17
8.59%
89
16.79%
4
0.75%
2
50
4
0.58%
7.08%
0.75%
8
80
15
36
31
2.33%
11.33%
54
7.65%
11.63%
6.79%
8.47%
68
18.58%
62
12
59
7
11
53
18.08%
6.06%
8.36%
5.43%
2.08%
14.48%
192
18
27.20%
13.95%
2
18
1
30
67
0.58%
17
4.96%
9.09%
15
7.58%
0.78%
5
3.88%
5.66%
19
3.58%
18.31%
25
6.83%
179
110
229
49
129
97
52.19%
55.56%
32.44%
37.98%
24.34%
26.50%
7
21
5.43%
16
12.40%
3.96%
17
8.59%
35
4.96%
10
1.89%
11
2.08%
8
2.33%
2
1.55%
4
2.02%
12
3.28%
138
26.04%
28
5.28%
43
Approach/
theme
Empty pages
%
Total over all
absolute
Total over all
%
Askew (1996)
Lindsay (2000
/ 2001)
7
0.99%
O'Farrell
(1991)
9
6.98%
Overall (1997)
4
1.17%
Landsberg et
al (1997)
5
2.53%
343
198
706
129
530
366
100.00%
100.00%
100.00%
100.00%
100.00%
100.00%
44
Pageat (1998)
13
3.55%
5.2
APPROACHING PROBLEMS OF AGGRESSION
This section gives a short overview of how a few selected authors approach problem behaviors, e.g. aggressive behavior (see Table 1.3). For
a more detailed overview, see Appendix C.
Table 1.3 Approaching Problems of Aggression
Approach
How to conduct a
consultation
Overview of how
to approach
problems of
aggression
Aim of
consultation
Askew (1996)
Dehasse (2002b)
Landsberg et al.
(1997)
Lindsay (2000;
2001)
O'Farrell (1991)
Overall (1997)
Pageat (1998)
x
x
x
x
x
x
x
x
x
x
x
x
2-3h for a dog
Maximum of
information in
minimum of time
Relationship with
owner
Contract with the
client
Demand of the
client
Motivation for the
consultation
Expectations of
normal / acceptable
vs. abnormal /
unacceptable
30 min. -2 hours for
initial interview
x
Follow up
Location of the
consultation
x
adaptive,
physiological vs.
pathological
Categorization of
behaviors
Duration of
Consultation
x
client's home or
counselor's office
consultation room
x
predatory
aggression vs.
dominance
aggression
inappropriate, in
context vs.
appropriate
behavior
clinically normal vs.
abnormal animals
longer than a
normal veterinary
consultation
x
x
telephone
interview, home
interview
consultation room
x
x
x
adaptive vs.
pathological
x
x
x
x
consultation room
x
x
x
x
x
x
x
45
x
x
x
x
x
Approach
the client
The Problem
Signalement of
the dog
Questionnaire
Behavioral history
/ speaking with
the client
Observation
(dog/owner)
Health / physical
examination
Analysis of the
system
Multifactoriality in
behavioral
problems
Influence of
learning
Influence of
genetics
Influence of
hormones
Determination of
type of
aggression
Behavioral
sequence
Characteristics of
the dog
Characteristics of
the owner
Circumstances/Sti
muli / context
social contacts
Body language /
communication
What happens
after the event?
Quantification of
the problem
Predictability of
Askew (1996)
Dehasse (2002b)
Landsberg et al.
(1997)
Lindsay (2000;
2001)
O'Farrell (1991)
Overall (1997)
Pageat (1998)
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
46
x
Approach
Askew (1996)
the aggression
Patterns
Age of onset
Duration of
problem
Intensity/control
of bite
Victim
Dehasse (2002b)
Landsberg et al.
(1997)
Lindsay (2000;
2001)
x
x
x
x
x
O'Farrell (1991)
Overall (1997)
x
x
x
x
x
x
x
x
x
Progression of
the problem
Neurological
centers involved
Provocation of
the behavior
danger
General medicine
/ general health
Diagnosis of the
problem
Hypothesis
Previous
therapeutical
attempts of the
owner
Prevention of
problems of
aggression
Consequences of
the aggression
Legal issues
Ethics
Pageat (1998)
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
47
5.3
TREATING PROBLEMS OF AGGRESSION
Table 1.4 gives an overview of what the selected authors take into consideration to treat problems
of aggressive behavior. For a more detailed overview, see Appendix D.
Table 1.4 Treating Problems of aggression
Approach
Approach to
treatment of
aggressive
problems
Goal of
treatment
Safety
Duration of
therapy
Urgency
Prognosis
Guarantee
Medication
Homeopathy
Pheromones
Surgery
Behavioral
and/or
cognitive
therapy
Ethological
"therapies"
strategic or
systemic
therapies
Askew
(1996)
Dehasse
(2002b)
Landsberg
et al. (1997)
x
x
x
x
x
x
x
Lindsay
(2000;2001)
O'Farrell
(1991)
Overall
(1997)
Pageat
(1998)
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Environmental
modification
x
x
x
x
Mechanical
Contact /
Handling
Autocontrol
Avoidance
Cognitive
appraisal
Countercondit
ioning
Desensitizatio
n
Diet
Diffusing
aggression
Euthanasia
Exercise /
activity
Flooding
Hierarchy
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
48
x
x
x
x
x
Askew
(1996)
Approach
Hormones
Ignoring
Obedience
training/contr
ol
Owners
Punishment
Replacement/r
ehoming
Reinforcement
Reward
Relaxation
Self
affirmation
Success
Therapies not
to be
recommended
Validation
5.4
Dehasse
(2002b)
Landsberg
et al. (1997)
Lindsay
(2000;2001)
O'Farrell
(1991)
Overall
(1997)
x
Pageat
(1998)
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
SUMMARY
The overview of these sources reflects the fact that the authors are using different scientific
models. These underlying models are, however, rarely explicitly mentioned or described. The
reader may be able to guess the model by analyzing the different themes such as learning
theories, therapies or medication.
These books are written to help understand and treat animal behavior problems, all the authors
put their main emphasis on the categorization, classification and description of behavioral
problems (25% to 56%). This makes sense, as one has to know what symptoms and signs to look
for in order to be able to diagnose the problem.
Not surprisingly, more emphasis is put on medical therapies by the veterinarians: Pageat (18%),
Landsberg (9%) and Overall (6%). The other professional groups mention this means of treating,
but do not devote many pages to it. This also is not surprising, as prescribing medication is not
within the range of treatments these other professionals use. What is very striking, however, is
how little information can be found on learning theories. Only two authors talk about learning in
depth: Lindsay (27%), a dog trainer, and O’Farrell (14%), a psychologist.
When considering that therapies of behavioral problems consist mostly of a combination of
different types of treatment (behavioral therapy, medication, environmental therapy, systemic
therapy etc.), it is interesting that the theoretical bases of these different treatments is not
discussed in greater detail by many authors.
Communication is an important part of behavioral therapy. Many different kinds of
communication are involved: communication of the dog with the owner, the owner with the dog,
the owner with the behavioral specialists, the behavioral specialist with the owner, the dog with
the behavioral specialists and the behavioral specialist with the dog. The authors reviewed here
dedicate between 5% and 18% of their books to this very important theme. However, when they
do address this subject, all of the authors consider more or less the same points, i.e. they receive
information mainly through the owner and partially through observing the animal. However,
these sources of information are emphasized in differing degrees according to the author. This
becomes clear in the more detailed overviews in the Appendices C and D.
49
Only the veterinarians seem to take into account the sequence of acts shown in the context of the
problem behavior, as this is a very important part of the diagnostic procedure.
In order to be able to guess the model used by the author, the reader will have to analyze the
single criteria more closely, such as ‘psychopathology’, which is central to the medical model
and is only mentioned in one book (Pageat 1998). The notions of ‘pathology’ and ‘pathological
behavior’ are only used by the French authors of veterinary behavior medicine, represented in
this study by Dehasse (2002b) and Pageat (1998). The other schools of thought use terms like
‘abnormal’ or ‘inappropriate’ behaviors.
Summary: It is reasonable to assume that no one approach suits every owner, every dog, every
problem or every situation. This overview may provide information, which would allow the
clinician and the client to select the approach or combination of approaches most suitable for the
individual case. In order to make it easier for the reader to understand the material and to apply
the information in practice, one would wish in future publications for a short section introducing
and defining the model used by the author.
50
PART II
COMPARISON OF FIVE SELECTED METHODS TO
EVALUATE THE AGGRESSIVE BEHAVIOR OF DOGS
1. INTRODUCTION
1.1
AIMS OF THE COMPARATIVE STUDY
Three behavioral tests currently being used in Switzerland were selected for this study. Sixty
dogs were run through all three of these tests. Additionally, a veterinary behaviorist and an
ethologist, both working in the area of behavioral therapy in dogs, were each asked to evaluate 12
of these dogs according to their own method.
This project targets two aspects - ‘aggressive behavior’ and ‘dangerousness’, as these are the two
topics currently most discussed concerning dogs. They are also the most common reasons for
referrals of dogs to behavior specialists (Overall 1997, Landsberg et al. 1997, Askew 1996). We
are aware of the fact that a variety of other behavioral problems exists in dogs and that some of
the methods included in this study give information on other factors than just aggression and
dangerousness. However, the consideration of all these other factors would surpass the scope of
this two-year project.
Underlying Questions
The underlying questions about the targeted topics are as follows:
• Is it possible to test a dog’s behavior?
• As all these methods claim to be able to evaluate a dog’s behavior correctly, the
assumption can be made that they should agree in their evaluations of the same dogs. Is
this the case? Specifically, do the three selected tests judge the same dogs to be similar
in their behavior related to ‘aggression’ and ‘dangerousness’? Are the evaluations of
‘aggressive’ or ‘dangerous’ by the two behavioral similar to those of the three tests
(based on the same criteria)? If no, what are the possible reasons?
• Do the test results correspond to what the owners say about their dogs in a questionnaire
sent to them in advance?
• Is there a correlation of ‘aggressiveness’ and ‘dangerousness’ within the methods?
• When talking about aggressive and ‘dangerous’ dogs, mainly the public security is of
interest. In this respect, it is interesting to know how great the specificity and the
sensitivity of the methods are and if there are many false negatives, i.e. how big is the
incidence of dogs being evaluated ‘not dangerous’ when in fact they are ‘really’
‘dangerous’?
• What is the importance of validity when evaluating canine behavior?
The null hypothesis of this comparison is as follows: As all five methods are currently being used
in Switzerland, it is to be assumed that they all agree on the same dogs’ behavior to a certain
extent.
51
Due to the different approaches compared in this project, and the fact that none of the tests
involved have been validated, several areas of this study may be considered a heuristic approach
to the comparison of some of the existing methods whose aim is to recognize potentially
dangerous dogs. 9
1.2
WHAT IS A TEST?
Before beginning a comparison of several tests, it should be made clear what a test is. Hirsig
(1993: 4.1) defines a test as follows: “A test is a standardized procedure to assess a sample of
behaviors with the aim to uncover the development of a person’s behavior over time or to make
possible a comparison of one person’s behavior with that of other persons (intra-individual
comparison vs. inter-individual comparison) 10 .
Tests are evaluated according to certain criteria, among which the most important are the
following:
• Reliability
• Validity:
• Sensitivity
• Specificity
“Reliability refers to the degree to which test scores are free from errors of
measurement…” (The American Psychological Association, APA, 1985:
19).
“The concept [of validity] refers to the appropriateness, meaningfulness,
and usefulness of the specific inferences made from test scores. Test
validation is the process of accumulating evidence to support such
inferences” (APA 1985: 9).
Sensitivity is defined as the ability of the test to identify correctly those
individuals showing the behavior tested for.
Specificity is defined as the ability of the test to identify correctly those
individuals not showing the behavior tested for.
These criteria will be discussed more extensively in a later section (chapter 6.1 “Discussion of
methods”).
9
Heuristic comes from the Greek word ‘heuriskein”, meaning, ‘to discover”. A ‘heuristic’ method is an approach for
gaining knowledge about an area also through experience rather than solely through the proving or disproving of
hypotheses based on a scientific model. The heuristic paradigm has been used in many sciences which involve the
study of behaviors and includes concepts and tools for recognizing and regulating the biases researchers set up as
part of the research process.
10
All English translations from the German and French original texsts by M. Bräm (2003)
52
2. ANIMALS, PERSONS, MATERIALS AND METHODS
2.1
DOGS AND OWNERS
2.1.1 FINDING THE DOGS AND THEIR OWNERS
Information about this project together with a registration sheet was distributed to several
veterinary clinics and dog clubs in the cantons of Bern and Fribourg, Switzerland. In addition,
several veterinarians practicing behavioral medicine were asked if they had any clients interested
in participating in this project.
2.1.2 NUMBER OF DOGS AND OWNERS
A total of 60 dogs were evaluated by the selected tests, and a subset of 12 were also evaluated by
the veterinary behaviorist and by an ethologist:
Tests:
Total of 60 dogs
51 Owners, 42 participating with one dog, 9 with two dogs
Veterinary behavior therapists: 12 dogs belonging to 9 owners (6 owners with one dog, 3
owners with two dogs)
Ethologist:
12 dogs belonging to 9 owners (6 owners with one dog, 3
owners with two dogs)
Three Dogs were evaluated by all five methods.
The dogs were numbered from 1 to 69 according to the sequence of application for participation.
We are aware of the fact that a larger number of individuals would have had to be considered to
obtain statistically significant information. However, because of time and financial constraints, it
was not possible to work with a larger number of dogs. The organization of the project with just
60 dogs involved constructing six possible test sequences and time schedules for 51 owners who
had to show up three to five times within a period of two to four weeks. Additionally, it was not
the aim of this project provide statistical information on the dog population of Switzerland, but to
compare different methods of evaluating dogs.
The tests were prepared and run jointly by Maya Bräm and Doris Lehmann (2003), who then
evaluated the resulting data separately for their individual doctoral dissertations.
Originally, the plan was to have all 60 dogs evaluated by the behavior therapists. This was not
possible for three reasons:
• One session with a behavioral therapist lasts between 1 and 2 1/2 hours. Therefore, the
number of dogs a therapist can evaluate a day is limited. It was not possible for the two
participating behavioral specialists to ‘sacrifice’ so much time for our project, as they
also had to keep running their own practices.
• The financial capabilities of our project were limited. Although the two therapists
reduced their usual salaries significantly for us, it was not possible to pay them for more
than the 12 dogs evaluated.
• The three tests alone were already time consuming for the dog owners, some of whom
had over an hour’s drive to the testing area – we could not expect all 51 owners to come
an additional two times (for a total of five times) for more consultations.
53
2.1.3 BREEDS OF DOGS
Because of the small number of dogs within each breed, the participating breeds were pooled into
4 breed groups encompassing 9 breed groups of the ‘Fédération Cynologique International’
(FCI), which are labeled in Tables 2.1 and 2.2 with Arabic letters.
Table 2.1 Breed groups according to the FCI
Breed groups
A
D
G
H
M
S
T
U
V
Total
Breeds
Retrievers - Flushing Dogs - Water Dogs
Dachshunds
Companion and Toy Dogs
Sheepdogs and Cattledogs (except for Swiss Cattledogs)
mixed breeds
Pinscher and Schnauzer - Molossians and Swiss
Mountain- and Cattledogs
Terrier
Spitz and primitive types
Pointing Dogs
number
%
13
21.7%
1
1.7%
2
3.3%
19
31.7%
11
18.3%
4
6.7%
4
3
3
60
6.7%
5.0%
5.0%
100%
Table 2.2 Breed groups used in this study
Breed groups
A
H
M
R
Total
Breeds
Retrievers
Shepherds
mixed breeds
other pure breeds (D, G, S, T, U, V of FCI group)
54
number
13
19
11
17
60
%
21.7%
31.7%
18.3%
28.3%
100%
2.1.4 SEX OF THE DOGS
In this study, twenty-six (43%) of the dogs were males, 34 (57%) females. Eleven (42%) of the
males and 24 (71%) of the females were neutered (see Fig. 2.1).
Figure 2.1 Sex of dogs
fc = female spayed, fnc = female intact, mc = male castrated, mnc = male intact
2.1.5 AGE OF DOGS
The ages of the dogs participating in this study ranged from 1.5 years to 13 years (see Fig. 2.3).
The mean age was 4.6 years, the median age 3.5 years
Table 2.3 Age of dogs
Age
< 2 years
2 years
3 years
4-6 years
7-9 years
> 9 years
Number of dogs
14
8
12
13
9
4
60
Total
2.1.6 WEIGHT OF DOGS
The weights of the dogs participating in this study were pooled into two groups (see Table 2.4).
Dogs weighing 25 kg or less and dogs weighing more than 25kg. The weights of two dogs are
missing.
Table 2.4 Weight of dogs
Category
1
2
Missing
values
Total
Weight range
≤ 25 kg
> 25 kg
55
Nr of dogs
33
25
2
%
55%
42%
3%
60
100%
2.1.7 STIMULUS DOGS
All three tests included situations where the tested dog is confronted with another dog (‚stimulus
dog‘). As can be seen in Table 2.5, a total of 11 stimulus dogs participated in this project. Nine
dogs were used alternatingly in Test A, 3 dogs in Test C. The stimulus dogs for Test B were
organized by the test experts themselves. We do not have any information on these dogs.
Table 2.5 Stimulus dogs
Stimulus
dog
S1
S2
S10
S3
S4
S5
S6
S7
S8
S11
S9
Breed
Sex
Age
Weight
category
40kg
White Shepherd
Golden Retriever
German Shepherd
mix
Appenzeller
Sennenhund mix
White Shepherd
Border Collie mix
Belgian Shepherd
Husky
German Shepherd
mix
German Shepherd
mix
American
Staffordshire
Terrier
m
mc
m
3 years
5 years
9 months
fc
7 years
36kg
1
f
m
fc
m
mc
6 years
3 years
9 years
3 years
2 years
32kg
25kg
22 kg
20.5kg
28kg
3
1
3
13
23
m
8 years
39kg
mc
4 years
30kg
25kg
N° of times used in
Test C
19
21
20
1
7
58
S3, S8
total
2.2
N° of times used in
Test A
4
2
60
QUESTIONNAIRE
The questionnaire filled out by the owners is based on questionnaires found in tests described in
the literature, especially the test from Niedersachsen, Germany (which originally was planned to
be included in this comparison). The questionnaire contains information on the living
environment and routines, the education, health history and behavior of the dogs.
The following kinds of information is collected by this questionnaire:
• General information about the dog: name, breed, age, weight, sex, age and reason of
castration
• Owner’s reason for choice of the dog and information about the dog’s origin: experience
of the owner with dogs, reason for getting a dog, origin of dog, reason for exactly why
this individual dog was chosen, age when adopted, raising conditions of the puppies.
• Education: participation in puppy-courses, further education of the dog, person mainly
responsible for the dog’s education, time spent with working with the dog, special
education.
• Behavior: house training, temperament, greeting rituals, behavior around visitors,
behavior in several specific situations, such as contact with other male or female dogs,
56
contact with unknown grown-ups or children, etc., behavior on the leash, behavior
towards other dogs, chasing behavior, obedience, agonistic behavior, biting history,
• Surroundings, daily routines: place where dog stays ‘at home’, ‘house rules’, persons
who mainly take care of the dog, feeding routines, walks, time spent alone by the dog,
places where dog is left alone, contact-persons, animals and objects.
• Health: history of illnesses or accidents, medication
Most of the answers were set up in a multiple-choice style. For more detailed information on the
questionnaire, its analysis and description, please refer to the dissertation of Doris Lehmann
(2003). An example of the questionnaire (in German) can be found in the Appendix E.
2.3
THE TESTS THAT WERE COMPARED
Three tests were chosen that include similar aims, but were somewhat different in structure.
Originally, we had intended to include one of the longer tests, i.e. the behavioral test of
Niedersachsen, Germany. Because of organizational reasons, it was not possible to include this
test. Since this is a project of a Swiss University, the following three tests actually run in
Switzerland at the beginning of this project were selected for this comparative study.
A Test of the American Staffordshire Terrier Club of Switzerland
B ‚Halterprüfung‘ by Hans Schlegel, Switzerland
C Test of the Canton of Basel-Stadt, Switzerland
Differences between the three tests:
• They are based on different ideas and theoretical schools.
• They are structured differently, e.g. concerning location, duration and degree of security for
the participants.
• They are designed for different breeds
• They were developed on the basis of different motivational backgrounds
Similarities of the three tests:
• They all include the same aim of uncovering potentially dangerous dogs
• They all consider ‘aggressive behavior’ as an undesired behavior.
• They were all developed by people familiar and experienced with canine behavior.
• They are currently (2002/2003) run by experts in Switzerland.
2.3.1
THE TEST OF THE AMERICAN STAFFORDSHIRE TERRIER CLUB (ASTC)
OF SWITZERLAND, TEST A
2.3.1.1
History
The American Staffordshire Terrier is one of the breeds included in almost all the breed-lists of
countries having breed-specific legislation. The American Staffordshire Terrier Club (ASTC) of
Switzerland developed this test as a reaction to the increased insecurity of the public towards socalled ‘fighting dogs’ and the tendency of governments in several countries to introduce
behavioral tests for certain breeds. This test is an adapted version of the test of the Staffordshire
Bullterrier Club of Germany.
57
2.3.1.2
Aim of the Test
The American Staffordshire Terrier Club of Switzerland is interested in healthy dogs, physically
as well as mentally. This test aims at recognizing potentially dangerous dogs, dogs showing
‘undesired aggressive behavior’ and recommend further measures to guarantee, as far as possible,
the public the safest dogs possible.
2.3.1.3
Breeds
This test is obligatory for all members of the ASTC of Switzerland who own an American
Staffordshire Terrier or American Staffordshire Terrier mixes. It is recommended for all other
members of the ASTC who own other breeds. In this project, we did not have any breed
regulations – all breeds were welcome. No American Staffordshire Terrier participated in this
study.
2.3.1.4
Prerequisites for the Normal Test
In the course of the test normally being run for the club, all participating dogs must be
microchipped (which was not the case for our tests) and registered.
Every dog must pass the test three times, the first time with 10-15 months, the second time with
24-36 months, and the third with 4-6 years.
2.3.1.5
Measures if the Test is Not Passed
Dogs that do not pass the test must be reported to the corresponding department of the SKG
(Schweizerische Kynologische Gesellschaft, Swiss Cynologic Society) who then decides,
together with the managing committee of the ASTC what further measures should be imposed.
These dogs must repeat the test.
Failing the test in the context of this project had no official consequences for the dog and its
owner.
58
2.3.1.6
The Evaluation
The following behaviors (see Table 2.6) are marked as ‘desired’ or ‘not desired’ in the American
Staffordshire Terriers being run through this test under normal circumstances (Reglement für die
obligatorische Verhaltensüberprüfung des ASTC – Schweiz 2000)..
Table 2.6 Desired and undesired behaviors in the test of the American Staffordshire Terrier Club,
Switzerland
Dog
Owner
Desired
• Balanced, interested, attentive
behavior adapted to the specific
situation, with a medium to high
threshold.
• Quick recovery after a situation.
• Open, people-friendly character
• Must be capable of influencing his/her
dog correctly
• Must have his/her dog under control in
any situation.
• Must influence his/her dog
consequently, but must not trespass the
limits of the law of animal protection.
Undesired
• Aggression, disobedience, fearfulness
• Low or too low threshold leading to
slow or no recovery at all after stress
•
Uncontrolled and insecure behavior
towards his/her dog
• Violation of the animal protection law
2.3.1.7
Why This Test Was Chosen
• The American Staffordshire Terrier is one of the breeds affected by breed-specific-legislation
in several countries, including Switzerland.
• The test is aimed at uncovering aggressive behavior and potentially dangerous dogs within this
specific breed.
• It is structured differently from the other two participating tests, for example in that it is
located in a fenced in area and on a road. It is the only one of the three selected tests that
includes a situation where the behavior of the dog is evaluated in the absence of its owner. The
dogs are judged by experts experienced in evaluating canine behavior (based on a cynological
background) who were willing to participate in our project.
The Experts
2.3.1.8
It was important to us in all three tests that the experts who evaluated the dogs for our project are
also familiar with the test. We were fortunate that two experts who normally judge the dogs in
this test were willing to collaborate. Unfortunately, one of the experts had to cancel his
participation in the last moment, which obliged us to ask someone who was not very familiar
with the test, (‘expert A2’), to help out. The ordinary expert, ‘expert A1’, has a long experience
of working with dogs and judging dogs (of several breeds). Expert A2 has experience with
working with dogs and dog training. The effects of this fact will be discussed in the discussion
part of this study.
59
2.3.1.9
Location
This test was run on the training grounds of the ‘Kynologische Verein Düdingen’, in Düdingen,
Fribourg, Switzerland. For our project, the area was enclosed with a fence (see photo of location).
Photos$
2.3.1.10 Time Schedule
The test was scheduled to run on 4 days, each expert testing 15 dogs on two days. The
cancellation of the one expert forced us to reschedule the 15 dogs of that specific day – the other
expert, A1, was able to evaluate an additional 4 dogs on one day, the other 11 dogs were
rescheduled on another day and evaluated by expert A2.
2.3.1.11 Duration
The test normally lasts 20 to 30 minutes, the difference in duration stemming from the possibility
of repeating some parts of certain situations or the whole situations or omitting a situation
because of various reasons (such as the dog being too dangerous or too anxious). As the experts
were asked not to give any feedback to the owners, the test only lasted an average of 15 minutes
for this project.
60
2.3.1.12 People, Animals and Material Necessary
Location:
fenced in area, a paved road
People:
minimum of six persons to form the ‘group of people’ situation, one bicycle rider,
one jogger. A total of 28 people helped us during the four days of this test, the
roles were shared and switched, so the same person did not have the same task
during the whole day.
Animals:
One dog for the situation of the ‘confrontation’ of the test dog with another dog
(Stimulusdog). A total of nine stimulus dogs shared this job during the four days
of testing.
Material:
dog toys, flags to signalize distances, a bicycle, a hook to which the dog could be
tied.
2.3.1.13 The Actual Test
There are two parts to this test, the first is located within a fenced-in area where the dog is mostly
off-leash, the second on a quiet road bordering the fenced in meadow. The owners were allowed
to motivate their dogs as they usually did.
On the Fenced-in Area (dog mostly off leash) (See Fig. 2.2)
1. The dog is let free to wander around and investigate anything it likes within in the fenced-in
area (for one to two minutes). Then, the owner calls the dog back
2. The owner motivates the dog to play – dog and owner play for about half a minute, then the
owner gives the command to stop playing, then the owner resumes playing with the dog for
another 40 seconds.
3. The owner and the dog (off leash) walk through a group of people moving around (duration
about half a minute).
4. The group of people form a circle around the owner and dog – and close and then open the
circle around the team, the first time walking slowly, the second time running (duration of the
whole situation about 45 seconds).
5. The owner walks the dog back and forth on the leash. On the way down the owner puts the dog
into a sit, then continues to walk with the dog. The owner walks the dog (on the leash) back
and forth a second time. On the way back the expert walks towards the team and first greets
the owner by handshake, then greets the dog (the whole situation lasts about one minute).
6. The owner walks the dog to the end of the paddock and puts the dog into a down or sit and
gives it the command to stay, the owner walks about 20 steps away from the dog. The expert
walks past and around the waiting dog several times. The owner goes and gets the dog.
Duration of the whole situation about one and a half minutes.
61
Figure 2.2 American Staffordshire Terrier Test on the enclosed area
On the Road (dog on leash all the time) (See Fig. 2.3)
7. The owner and dog are passed by a biker, then a jogger from the back and then from the front.
8. The owner and dog pass a dog and its owner standing on the left hand side of the road. The
team turns around and walks back the way it came, thereby passing the dog and its owner on
the right hand side of the road.
(Duration of points 7 and 8 together about one and a half minutes)
9. The dog is tied to a pole and the owner disappears out of sight of the dog. The expert walks
past the dog noisily several times. (Duration of the situation about one to one and a half
minutes)
Figure 2.3 American Staffordshire Terrier Test on the road
62
2.3.2
THE ‘HALTERPRÜFUNG’, SWITZERLAND, TEST B
2.3.2.1
History
The ‘Halterprüfung’ (‘test of the dog owner’) was developed by Hans Schlegel on the basis of his
own many years of experience with dogs and wolves. The first voluntary ‘Halterprüfung’ in
Switzerland was held in March 2001. It is not a compulsive test for any dog breed or population
category; it is voluntary.
2.3.2.2
Aim of the Test
The aim of this test is primarily to find out if the owner has his or her dog under control in every
day situations. The second goal is to be able to assess the basic character of the dog and
consequently be able to decide whether the animal is a risk factor for its surroundings. As H.
Schlegel is the owner of a dog school in Switzerland, an additional objective is also to pass on
knowledge to dog owners, counsel them and correct them in the form of courses.
2.3.2.3
Breeds Normally Accepted to the Test
All breeds are accepted for the Halterprüfung in the ‘normal’ circumstances as well as in this
project.
2.3.2.4
Prerequisites for the Normal Test
No aids are allowed (food, toys, clicker, etc.) except for the voice and petting. The dog is offleash during most of the test, so a certain degree of obedience is necessary to be able to
participate.
2.3.2.5
Measures if the Test is Not Passed
Since this is a test issued by a single person and not an organization or the state, no measures
have to be taken, if a dog fails the test. The owner has the possibility to participate in dog training
courses offered on the spot.
2.3.2.6
The Evaluation
This test considers two main issues. The first is the ‘Führigkeit’ (the ‘leadability’ of the dog), the
second is the ‘das Wesen’ (the ‘character/nature/nature’ of the dog).
Within the aspect of ‘leadability’, the following points are evaluated:
• ‘leadability’ in the ‘Ist-Stand’ 11 (‘present state’)
• ‘leadability’ and behavior under distraction
• ‘leadability’and behavior in every day situations
• ‘leadability’and behavior in the presence of conspecifics
11
The ‘Ist-Stand‘ is the term Schlegel has coined to determine the present state of the dog before it is put into the
various test conditions.
63
Within the aspect of ‘character / nature’, the following points are evaluated:
• Influences of the environment (noise, fast movements, sudden influences, objects)
• Fearfulness
• Threshold to withstand pressure (‘Belastbarkeitsgrenze’ in every day situations)
• Risk group for the dog’s environment and humans surrounding it
• Measures and recommendations
The test puts a great deal of emphasis on obedience and how well the owner has his or her dog
under control, in normal situations as well as in situations of a higher degree of stress. This
second point has more weight in this test than it does in the other two.
2.3.2.7
Why This Test Was Chosen
When the whole problem of ‘dangerous’ dogs was re-ignited after an incident in Germany in the
year 2000 where a boy was killed by two dogs, the Swiss Federal Veterinary Office (FVO)
organized a meeting of known dog professionals to discuss possible measures. In this context,
Hans Schlegel suggested his ‘Halterprüfung’ as a solution. According to Schlegel, if all owners
had their dogs under control all the time, the risk of incidents would probably decrease (AGGH,
2000).
One of the aims of this test is to evaluate the risk incorporated by dogs, including aggressive
behavior, which consequently also leads to discovering potentially dangerous dogs
This test is structured differently from the other two tests in that it emphasizes obedience and the
role of the owner much more than do the other two tests, which mainly take the dog into account.
As can be seen from the web-page (www.wolfsprung-kennels.ch), the demands that the dogs
must fulfill in order to pass this test are (1) Walking on the leash, (2) Walking without leash, (3)
‘Down’ out of a movement and ‘down’ in a situation with distraction, and (4) Being called back
out of a game. These are more strict prerequisites than are demanded by any of the other tests.
It is the only test to look at the dog’s reactions to other animals (chicken and goats) and the only
test where the dog is off leash the major part of the time.
2.3.2.8
The Experts
As in test A, the experts familiar with this test evaluated the dogs for this project. Experts B1 and
B2 shared this job. They are both experienced in working with dogs and in evaluating canine
behavior, and they are both professional dog trainers.
64
2.3.2.9
Location
This test was run in Düdingen, Fribourg, Switzerland. It started on an open meadow and ended in
a wooded area (See Photo below).
Photos
2.3.2.10 Time Schedule
The test was run on four consecutive days, from 8am until about 1pm.
2.3.2.11 Duration
The entire test normally lasts about 30 minutes. As the experts were asked not to give any
feedback to the owners, the ‘Ist-Stand’ evaluation of the present state of the dog, lasted an
average of 5 minutes and the test itself lasted ca. 10 minutes.
2.3.2.12
People:
Animals:
Material:
People, Animals and Material Necessary
10 to 15 persons for the group of people
1 jogger
1 bicycle rider
2 to 3 medium to large sized dogs playing
Several chickens in a fenced in area
2 medium-sized goats
One bicycle
Objects to make noise with (bottles filled with cobble stones, tin cans, etc.)
A white plastic band as limitation line
Dog toys
65
2.3.2.13 The Actual Test
This test is divided into two parts, the evaluation of the present state (‘Ist-Stand’) and the
‘Halterprüfung’, the actual test.
Evaluation of the present state (‘Ist-Stand’)
Passing this part is a prerequisite to participating in the ‘owner’s test’ (‘Halterprüfung’). As
mentioned above, certain demands have to be fulfilled in order to be able to pass the
Halterprüfung. The following four main points are considered: (1) Walking on the leash, (2)
Walking without leash, (3) Down out of a movement and down with distraction, and (4) Being
called back out of a game. These points are tested in the Ist-Stand.
Walking on the leash (See Figure 2.4)
1. The dog walks next to its owner on the leash. The dog must not pull or avoid or bother its
owner in any way.
2. When the owner stops at B, the dog must sit down next to him.
Walking off the leash:
3. Still at B, the dog is let off the leash. The dog must walk off the leash at heal of his owner from
A->C->D->C->A, the dog must not run away or bother his owner in any way
4. When the owner stops at A, the dog must sit down immediately
Lying down out of a movement and lying down with distraction: during this whole situation, a
group of people keeps moving around the whole field.
5. The dog (off leash) follows the owner at heel from A to B (about 10 steps)
6. At B, the dog is put into ‘down’ and the owner keeps on walking with the same speed without
turning around until he/she reaches point C
7. Two persons holding hands pass beside each side of the dog (with the hands passing over the
dog), first walking normally from the front and then back, then bent over with slow steps and
staring at the dog, again from the front and then back.
8. The owner goes and gets the dog at B and walks to A
Being called back from a game
9. At A, the owner receives a toy and starts playing with the dog
10. As soon as the dog has started to play, the owner hands the toy over to the expert who
continues animating the dog to play
11. The expert throws the toy away past the white limitation line (ca. 20m)
12. The owner sends his dog after the toy. At 10 m distance from the toy, the owner calls the dog
back or gives it the command to lie down.
66
Figure 2.4 ‘Ist Stand’
White limitation line
Grass
If the dog-owner-team passes this ‘Ist-Stand’, then it can go on to the actual ‘Halterprüfung’. The
teams who did not pass the ‘Ist-Stand’ were additionally confronted with two situations of the
actual test to allow the evaluation of their behavior towards other animals and people, which is
not part of the normal Ist-Stand. These two situations were the confrontation with goats and
chicken and the testing of the dog’s reaction to people in a circle (see below).
67
The Actual Test
The ‘Halterprüfung’ took place on an open field and on a path through woods (see Fig. 2.5)
The dog is off-leash during the whole test.
• The dog is confronted with unknown animals:
1. Chickens in a pen
2. Two goats tied to a tree trunk
• Single active person
3. A person riding a bike
4. A person jogging
• Active group of people, in a stress situation
5. Several people form a circle in the center of which the owner and the dog are standing.
The persons in the circle all jump up into the air at the same time and shout aloud when
landing. To relax the situation, they crouch down afterwards and make contact with the
dog.
• Every day situations
6. A group of people making a lot of noise walks towards the team. The dog must follow its
owner through this group. The group lines up to form a passage for the team. When dog
and owner pass through this passage, the people drop their ‘noisy’ objects (bottles filled
with pebbles, tin cans…)
• Other conspecifics
7. Dog and owner must walk through a group of people playing with their dogs.
Figure 2.5 Physical situation for the ‚Halterprüfung‘
1
3
2
4
5
7
6
All the teams first went through the ‘Ist-Stand’, the ones who passed continued on to the
‘Halterprüfung’, the ones that did not pass were tested in the above mentioned two additional
situations.
68
2.3.3
THE TEST OF THE CANTON OF BASEL-STADT, SWITZERLAND, TEST C
2.3.3.1
History
The Canton of Basel-Stadt developed this test in 2001 as a reaction to several incidents. One of
these was the killing of a little boy by two dogs of the so-called ‘fighting breeds’ in the summer
of 2000 in Hamburg, Germany. Additionally, the Canton has statistical information on the
increase of incidents caused by certain breeds within the Canton. The increasing number of the
Rottweiler and Pitbull Terriers breeds in the Canton and the ownership of problematic dogs by
individuals who are not able to handle such types of dogs were pushing factors contributing to the
introduction of a test.
It must be emphasized that the test is not the only measure taken concerning this issue of
‘potentially dangerous’ dogs in Basel. There are several other legal requirements for persons who
would like to own such a dog (see listing of breeds below). The officials also collaborate very
closely with dog trainers and animal therapists.
2.3.3.2
Aim of the Test
The aims of this test (plus the additional measures mentioned below) are (Kantonales
Veterinäramt, Basel-Stadt 2001):
• To recognize conspicuous dogs with a short behavioral test in the course of giving the
approval to own a dog belonging to the so-called ‘fighting-dog’ breeds.
• To protect humans and animals from canine aggression
• To protect the dog itself of being kept in conditions not appropriate for its species.
2.3.3.3
Breeds Normally Accepted to the Test
This test is obligatory for all the so-called ‘potentially dangerous’ dog breeds in the Canton of
Basel-Stadt. These are: the Pitbull Terrier, the American Staffordshire Terrier, the Staffordshire
Bullterrier, the Rottweiler, the Doberman, the Fila Brasileiro and the Dogo Argentino. Dogs of
other breeds who have been involved in incidents are also evaluated in this test.
In our project, there was no breed limitation. It was interesting to see how breeds that normally
do not participate in this test handle the situations. It was also interesting to observe the reactions
of the experts to breeds normally not tested by them in this test.
Prerequisites for the Normal Test
2.3.3.4
Persons who would like to own a ‘potentially dangerous dog’ (see breed list above) must get an
approval from the Canton before acquiring the animal. The person must be over twenty years of
age and must not have previous convictions of supporting prostitution. He or she must prove
enough cynologic knowledge to own a dog of this breed.
The dogs must stem from a breeder who socializes his puppies sufficiently. The dog must be
microchipped. It is obligatory for the owners of these breeds to participate in puppy play courses
and obedience classes with their dogs. The owner must take out liability insurance covering 3
Million Swiss Francs and pay an approval charge of 250 Swiss Francs.
The test is only part of a whole procedure of prevention and measures taken in order to increase
public security. (Kantonales Veterinäramt, Basel-Stadt 2001)
69
2.3.3.5
Measures if the Test is Not Passed
The Canton of Basel-Stadt sets great store on the fact that it collaborates with dog trainers, dog
schools, dog therapists, etc. Puppy play groups and obedience classes are obligatory prerequisites
to own the targeted breeds, but obedience classes, training lessons and consultations with a
therapist can also be measures used after not having passed the test. The dog is then observed
during this time and must reappear for another test. Further measures that can be imposed are
constraint by muzzle or leash, prohibition of a certain forms of training (e.g. ‘Schutzdienst’),
rehoming of the dog or euthanasia.
As with the other two tests included in this project, the failing of this test in the frame of our
project had no official consequences whatsoever for the dogs or its owners.
2.3.3.6
The Evaluation
The following aspects are the normal features of the evaluation of the test run in the Canton of
Basel-Stadt, although not all were utilized for this project:
Both the behaviors of the dog and of the owner are taken into account in all situations.
Concerning the dog’s behavior: Normal, aggressive, threatening, fearful, biting or biting
without threatening; the dog is leading its owner; aggressive behavior towards the owner;
good, acceptable or bad obedience.
Concerning the owner: no reaction to the dog’s behavior; correction of the dog (with the
voice or physically); fearful, relaxed, insecure or dominant behavior; the owner has to get
the dog (in the situation of calling it to come); no or fearful reaction to or correction of the
dog showing aggressive behavior towards its owner.
2.3.3.7
Why This Test Was Chosen
This is the only test for dogs based of certain breeds made obligatory based on Cantonal laws in
Switzerland. Probably partly due to this fact, much attention is given to establishing as much
security as possible for the persons and dogs involved in running this test. This leads to an
interesting point in this test - its structure. As opposed to the other two tests that are located on a
field (fenced in or not) and a wood, this test is run on a quite small area. To guarantee the security
of the experts and the stimulus dogs, these are situated within a kennel during most of the test,
while the owner-dog-team passes by outside this kennel. The stimulus dog used for the
confrontation situation with another dog is in another kennel to guarantee its safety as well. Since
this is an obligatory test, which should be run on as many dogs as possible, the test has to be of a
short duration. This test only lasted an average of 5 minutes in this study, which is another point
in which this test differs from the others in this comparison.
2.3.3.8
The Experts
Experts C1 and C2, who are both experienced judges of canine behavior and also normally
evaluate the dogs in Basel-Stadt, came to Bern one day each to assess the behaviors of all the 60
dogs for our project.
70
2.3.3.9
Location
This test was run on the premises (Forstzentrum) close to the School of Veterinary Medicine of
Bern, Switzerland. The most important characteristics of the original settings in Basel-Stadt were
reconstructed as closely as possible in Bern. These were the sensation of narrowness in the whole
area, kennels for the stimulus dog and the expert, and barriers (concrete walls in Basel, kennels in
Bern) surrounding the whole area. (See Photo 2.3)
Photo 2.3 Premises of Test C
2.3.3.10 Time Schedule
This test was run on two days with both experts evaluating 30 dogs each on one day.
2.3.3.11 Duration
The normal duration of this test in Basel-Stadt is 15 minutes. As in the other tests, the experts
were asked not to give any feedback to the owners, which reduced the duration of the test to an
average of about 5 minutes.
2.3.3.12
People:
Animals:
Material:
People, Animals and Material Necessary
None were necessary besides the expert
An intact male dog. This job was shared by three dogs during the two testing days.
Three dog kennels (one for the expert, one for the stimulus dog, one to construct a
narrow passageway). Gates to close off the test area to avoid dogs escaping.
Canvasses to cover the area where the stimulus dog stayed when not confronted
with the test dog.
71
2.3.3.13 The Actual Test
The structure and sequence of the actual test is shown in Fig. 2.6.
Figure 2.6 Test of the Canton of Basel-Stadt
2
6
Narrow
passageway
1
3
4
5
kenne ls
Stimulus
dog
Expe rt
Fenced-in
area
1. First contact of expert and dog-owner-team, separated by a gate. The Expert goes into a kennel
and asks the team to enter into the fenced-in area.
2. Owner and dog on the leash walk past the kennels as if on a ‘normal’ walk.
3. Owner lets his dog off the leash and both walk back and forth past the kennels again.
4. A stimulus dog (intact male) is let out into the second kennel; owner and dog walk past the
kennels again
5. Expert leaves the kennel and reads the microchip on the dogs left shoulder
6. The owner takes his dog on the leash again and the team walks through a gate and a narrow
passageway.
2.3.4 THE EXPERTS
It was very important that the people evaluating the dogs in the tests were familiar with the test
and generally had had experience in the evaluation of canine behavior. For every test, the two
experts who normally are the judges in the tests were asked to participate. All six of them were
willing to collaborate. Unfortunately, one expert (A3) of test A had to cancel his participation at
the last moment and it was not possible for the other expert (A1) of this test to jump in for him on
such short notice. This made it necessary for us to call in an external expert (A2), who had
experience in evaluating canine behavior and working with dogs, but was not familiar with the
test. Expert A2 evaluated 11 of the dogs.
72
2.4
THE BEHAVIORAL SPECIALISTS
It is difficult to find a term that encompasses the whole of the professions concerned with
evaluating dog behavior and therapy of behavioral problems. We decided on the term ‘behavioral
specialists’, referring, in our case, to a veterinarian and an ethologist, specialized in animal, and
especially canine, behavior problems and schooled in treating them. What is interesting to us here
are the different approaches to the same problem.
For more discussion of this point, please refer to the section 5 of the Part I: ‘Overview to
Evaluating (Aggressive) Problem Behaviors of Dogs According to a few Selected Models.
2.4.1
THE VETERINARY BEHAVIORAL THERAPIST
2.4.1.1
Education – Approach
The veterinary behaviorist (VB) in this study is a veterinarian specialized in animal behavioral
medicine alongside of his work in his general veterinary practice for small animals in the French
speaking part of Switzerland. In September of 2002, he successfully passed the exams of the
course ‘Zoopsychiatry’ 12 held in Switzerland.
The VB combines his knowledge from several different approaches in his own individual
approach. He generally receives his clients and patients in his practice. His diagnosis is based
mainly on a discussion with the clients, i.e. he relies greatly on what the clients tell him and what
he sees of the animal in the consultation room. However, depending on the problem the animal
shows, he might visit the pet’s home, evaluate the animal in the specific problem situation or ask
a third party about their opinion (such as the victim or the dog’s regular veterinarian). A
somatically healthy animal is the basis for this work, so if he thinks certain medical diagnostic
procedures are essential, he will ask for them or run them himself.
12
The course was organized by the STVV (Schweizerische Tierärztliche Vereinigung für Verhaltensmedizin, ‘Swiss
veterinary association for behavioral medicine’) and is based on a medical model, i.e. it recognizes the notions of
pathology and adaptation and the structure of knowledge in pathogeny, semiotics, diagnosis and therapy.
73
2.4.1.2
The Dogs
The VB evaluated a total of 12 dogs belonging to 9 owners (see Table 2.7). The dogs were
selected from the 60 dogs that participated in the tests, based on the readiness and interest of the
owners to have their pets assessed by a VB in addition to the evaluation of the three tests.
Table 2.7 Description of dog population
Sex
Number of dogs
Breed
Number of dogs
Female spayed
7
Retrievers
3
Female intact
2
Shepherds
1
Male castrated
1
Mixed Breeds
4
Male intact
2
Other pure breeds
4
Total
12
total
12
Age
Number of dogs
< 2 years
4
2 years
1
3 years
4
≤ 25 kg
4-6 years
2
> 25 kg
5
7-9 years
1
total
12
total
12
Weight group
Number of dogs
7
2.4.1.3
Location and Time Schedule
The veterinary behavioral therapist evaluated the dogs in a room on the premises of the animal
hospital in Bern, Switzerland, on four days during April 2002. Since the VB was the last expert to
evaluate the dogs, he was the only one allowed to discuss the problems with the owners and to
give them feedback.
2.4.1.4
Duration
The average duration of a consultation was one hour. Some of the owners brought both their dogs
to be evaluated, which increased the duration somewhat.
2.4.1.5
People, Animals and Material Necessary
No additional help or material was necessary besides a room, a table, chairs, pen and paper.
The Evaluation
2.4.1.6
For our project, the VB evaluated the dog in a consultation with the owner as he usually does. He
got information about the dog’s behavior directly by watching it during this time and indirectly
by the interview of the owners.
Although he did not agree completely with the criteria, he did agree to answer the same criteria as
the test experts did in order to allow us to compare his evaluations with the test results.
Additional remarks were added to allow interpretation or understanding of his line of thought.
74
2.4.2
THE ETHOLOGIST
2.4.2.1
Education and Approach
The Ethologist (ET) participating in this project studied ethology at the University in Zürich, with
the aim to treat dogs‘ behavioral problems. She practices behavioral therapy in the Canton of
Zürich and is enrolled as an instructor in the education of the VIETA 13 offered by D. Turner in
Zürich.
She has a questionnaire, which she uses to ask the clients about their dogs’ behaviors. She usually
goes to the problem-pet’s home in order to be able to evaluate the behavior of the animal in its
normal surroundings and interviews the owners in their home.
2.4.2.2
The Dogs
The Ethologist evaluated a total of twelve dogs (see Table 2.8). These twelve dogs belong to nine
owners (six owners with one dog, three owners with two dogs).
As with the VB, the dogs were selected from the 60 dogs that participated in the tests, based on
the readiness and interest of the owners to have their pets assessed by a VB in addition to the
evaluation in the three tests.
Table 2.8 Description of dog population
Sex
Number of
dogs
Female spayed
Female intact
Male castrated
Male intact
5
1
3
3
Total
12
Age
1 < 2 years
2 = 2 years
3 = 3 years
4 = 4-6 years
7 = 7-9 years
9 > 9 years
total
Breed
A = Retrievers
H = Shepherds
M = Mixes Breeds
R = Other pure
breeds
Total
Number of
dogs
2
1
Weight group
2
5
1
1
12
≤ 25 kg
> 25 kg
total
Number of
dogs
3
6
1
2
12
Number of
dogs
6
6
12
2.4.2.3
Location and Time Schedule
The Ethologist visited the owners and their dogs at their homes during one week in February,
2002, i.e. preceding the tests. She was asked not to reveal any information on the outcome of her
evaluation to the pet owners.
13
Verband der I.E.T. - Institut für angewandte Ethologie und Tierpsychologie - Absolventen, ‘Association of alumni
of the I.E.T (institute for practical ethology and animal psychology)
75
2.4.2.4
Duration
Considering the fact that the Ethologist went to the homes of the pets, more time had to be
allowed. The actual evaluation lasted between two and three hours per owner. Here also, some of
the owners had two pets to be evaluated.
2.4.2.5
People, Animals and Material Necessary
No additional material or persons were necessary.
2.4.2.6
The Evaluation
The ET also used the same evaluation sheet as the tests to assess the behaviors of the dogs. This
allows us to make a descriptive comparison of the evaluations of the ET with the test results.
2.5
THE TIME SCHEDULE
After having consulted several people experienced with canine behavior, we decided that one
evaluation, i.e. test or consultation by a behavioral specialist, would be run per dog per day. The
aim of this was to avoid tiring effects of the dogs and owners. In the original time schedule
(Table 2.9), the plan was to run every dog through one test a day and all three tests within one
week. The ethologist evaluated 12 dogs two weeks before the tests started; the veterinary
behaviorist 12 dogs within a month after completion of the tests. As a consequence to the
cancellation on short notice of the participation of expert A3 in the test A on Monday, March 11th
2002, the decision had to be made of either doing without the 15 dogs scheduled for that
particular day or of rescheduling them to another day. Since the total number of dogs (N = 60) is
the lower limit needed for statistical information, we decided on the second solution. Eleven of
the 15 dogs were rescheduled for Wednesday, March 13th 2002, and evaluated by the independent
expert A2, and four were additionally evaluated by expert A1 on Friday, March 15th 2002. This
led to some changes in the other requirements as well – four of these eleven dogs had two tests
on the same day (Test A and C) and the test sequence of five dogs was altered (see Table 2.9).
76
Table 2.9 The original time schedule
Date
6.3. 7.3. 8.3. 9.3. 10.3. 11.3. 12.3. 13.3. 14.3. 15.3.
Test
A
Number of dogs 15
A
15
B
15
A
15
B
15
B
15
A
15
B
15
C
30
C
30
A = Test A; B = Test B, C = Test C, Numbers = number of dogs per day
Table 2.10 The altered time schedule
Date
6.3. 7.3. 8.3. 9.3. 10.3. 11.3. 12.3. 13.3. 14.3. 15.3.
Test
A
Number of dogs 15
A
15
B
15
A
11
B
15
B
15
A
19
B
15
C
30
C
30
A = Test A; B = Test B, C = Test C, Numbers = number of dogs per day
2.6
STATISTICAL TESTS
The following sections describe the statistical tests used in this study.
2.6.1 DESCRIPTIVE RESULTS
No statistical tests were used for this part.
2.6.2 COMPARISON OF THE METHODS
Kappa is a measure of association (correlation or reliability beyond chance) between two
measurements of the same individual when the measurements are categorical, and when a gold
standard is not available. Significant agreement thus is interpreted as an indication of validity
while disagreement suggests that the tests are untrustworthy. It in principle tests if the counts
along the diagonal of table are significantly large. Because Kappa is used when the same variable
is measured twice, its standard form is only appropriate for square (2x2) tables. Kappa ranges
from 1 (complete agreement) to 0 (agreement that is equal to that expected by chance), whereas
negative values indicate agreement less than is expected by chance (i.e. disagreement). Rules of
thumb for the interpretation of kappa results from 2x2 tables: less than 0.2 indicate slight
agreement, values between 0.21 and 0.40 indicate fair agreement, values between 0.41 and 0.60
77
indicate moderate agreement, values between 0.61 and 0.80 indicate substantial (high)
agreement, and values over 0.81 indicate very high agreement between the two raters.
An extension of Kappa within the statistics software STATA 7 can be used to study the
agreement of two or more raters such as judges or doctors. There, each rater classifies each
individual into one of k categories. Again the kappa value – indicating the percent agreement of 2
or more raters beyond chance – is calculated together with a test statistic and it’s p-value.
Positive Kappa values with p-values < 0.05 indicate that the agreement beyond chance alone is
significantly greater than zero (0) (and nothing more)! (Thursfield 1995: 280-282)
2.6.3
CORRELATION WITHIN THE TESTS, CORRELATION OF AGGRESSION
AND DANGEROUSNESS
Here, the correlation between the individual criteria within the tests are examined. This was done
with the program NCSS using cross tabulations. Where the conditions of a Chi-square test were
fulfilled, this test was run. The Chi-square test is an analysis of numbers of subjects in categories.
The minimal conditions for this test are:
• With two groups and two variables, the expected values of every frequency should be at
least five
• With more than two groups or variables, the expected values should not be less than 5 in
more than 20% of the total expected categories and no expected values should be less
than one.
In the cases where these conditions were not fulfilled, the groups were pooled to categories
consisting of a larger number of individuals where it made sense. Where it was possible the
categories were pooled to 2 x 2 tables, making a Fisher’s exact test possible, with the others the
Chi-square test was run with the pooled results.
The null-hypothesis in this section is that there is no correlation between the points looked at.
The alternative hypothesis is that there is a correlation.
2.6.4 COMPARISON OF APPROACHES
The aim of this part was to define the sensitivity and specificity of the test results when compared
to the evaluations veterinary behaviorist and the ethologist. The specificity and sensitivity were
calculated with the Win Episcope program.
78
3. EVALUATION
3.1
QUESTIONNAIRE
For the discussion of the evaluation of the questionnaire, please see the dissertation of Doris
Lehmann (2003).
3.2
THE METHODS
The aim of this project is to compare how three different tests and two behavioral specialists
evaluate the aggressive behavior and potential dangerousness of the same dogs (60 for the tests,
12 each for the comparison of the results of the veterinary behaviorist and the ethologist with the
tests). In order to be able to compare the results, a common basis had to be created. Since the
three participating tests all have different ways of evaluating the dogs‘ behaviors, we set up our
own 15 criteria. These points are based on the analysis of the situations of several tests (amongst
others, the test of Niedersachsen, Germany (Niedersächsisches Ministerium für Ernährung,
Landwirtschaft und Forsten 2001), the test for aggressive behavior by Netto and Planta, 1999, the
Netherlands and the three participating tests). The terms used are based on popular definitions,
are mainly contextual and are always based on the evaluation sheet used. It is impossible to make
one definition that will suit everybody; the terms used here are always framed by quotation
marks.
We concluded that the following fifteen points are common to most tests evaluating canine
behavior:
•
Other dogs
1. ‘Intraspecific aggression’ of the dog towards other dogs
2. ‘Potential dangerousness’ of the dog-owner team for other dogs
Originally, we had considered splitting this question up into two separate aspects according
to the gender of the other dog. Since no test required both sexes of stimulus dogs to be
present, we decided not to include this separation. However, the gender, breed and weight
group of the stimulus dogs were noted for every test dog and were run through a cross
tabulation test to find possible correlations with the test results.
•
Human beings
3. ‘Interspecific aggression’ of the dog towards humans
4. ‘Potential dangerousness’ of the dog-owner team for humans
Here also, we had first considered differentiating between the behavior of the test dog
towards children and adults, men and women. This was based on the longer tests we had
considered including in this study. We decided against this differentiation because the three
participating tests do not specifically test for this. Understandingly, it is ethically not possible
to include a test situations where children are exposed to dogs.
79
Since the situations where the dog’s reaction to people is tested usually consist of a group of
people made up of men and women walking around, we did not specify the gender to which
the dog reacted.
14. Behavior of the dog (alone) towards humans
15. ‘Potential dangerousness’ of the dog (alone (i.e. in the absence of the owner)) for humans
This point was only tested in one test, test A, and therefore does not appear in the
comparison of the three tests. However, it is a plus point for this test in that it considers a
plausible situation the other two tests do not (e.g. the dog is tied in front of a shop)
•
Moving objects: jogger, bicycle
5. The behavior of the dog towards moving objects (jogger, bicycle)
6. ‘Potential dangerousness’ of the dog-owner team for moving objects (jogger, bicycle)
This point was tested by tests A (jogger and cyclist) and B (cyclist).
•
Other animals
7. ‘Aggressive behavior’ of the dog towards other animals
8. ‘Potential dangerousness’ of the dog-owner-team for other animals
Only one test, Test B, considered this aspect. Consequently, this point cannot be taken into
account when comparing the results of the three tests. We decided on considering it anyway,
because it is often the cause of many problems for owners. It is a plus for this one test that it
can give more information on this subject than can the others.
•
Indirect factors
Three additional factors were considered that are not directly involved in aggressive behavior
or dangerousness, but that might have an indirect influence on either the behavior itself or on
the way the behavior is evaluated by the experts:
9. Walking on the leash
This is indirectly connected to the aggressive behavior and potential dangerousness of dogs.
It is often said that being held on a leash makes a dog even more aggressive than if it were
free. However, this point could not be evaluated in this project, because the dogs were not
only presented to the according situations when on the leash, but also when free – and it is
not known upon which situation the expert based his decision.
It is an important fact though, when evaluating the dangerousness of the dog. If the dog is on
the leash all the time and the owner therefore has it under control, the potential
dangerousness of the team is reduced. Doris Lehmann (2003) evaluates this aspect in her
dissertation.
80
10. Obedience
Again, this is an issue indirectly related to dangerousness for the same reason as walking on
the leash. If the dog is under control in every situation, the potential dangerousness of the
team is reduced. If the dog is not under control, it depends on the dog’s behavior, how
dangerous it is. It may be dangerous because it shows aggressive behavior (as defined in this
project), or it can be hazardous for its surroundings simply by running around out of control,
increasing the risk of, for example, a traffic accident or running into people. This aspect of
dangerousness, however, was not considered in this project.
11. The relationship of the dog and its owner
Another indirectly related issue of aggressive behavior and potential dangerousness is the
relationship of the team. Here, the ‘hierarchy within the team’ and the fact if there was an
‘attachment’ or not was of interest. The use of these terms is based on the popular definition
of ‘hierarchy’ and ‘dominance’ and not on the ethological definitions, meaning that if the
owner is ‘dominant’ over the dog, the hierarchy is intact, and if the dog is ‘dominant’ over
the owner, hierarchical problems are more likely to occur. The simplified use of these terms
is indicated by the single quotes around the term.
•
Passing the test
12. Test passed or not
Since the three tests value specific aspects differently, it is interesting to compare how many
and which dogs pass the test and what the reasons might be for not passing the test.
This point is not an overall summary of all the other points – it is considered as a separate
subject.
•
Measures
13. Measures recommended
This question asks the experts what measures they would recommend for the team being
tested, if it were a ‘real’ test. Again, the different structures, background and aims of the tests
highlight some differences, which will be shown in the result section and discussed in the
discussion section.
The behavioral specialists were able to answer all the points, however, they cannot be included in
the statistical evaluation, as the number of 12 dogs does not allow a reliable interpretation of the
results.
An example of the criteria (in German) can be found in the Appendix F. As mentioned above, the
evaluation and discussion of the results is split up between the dissertation of Lehmann (2003)
and this report. The results and discussions of points ‘potential dangerousness for other dogs, for
humans, for moving objects and for other animals’, ‘walking on the leash’ and ‘obedience’ can
be found in the dissertation Lehmann (2003) and of the points ‘intraspecific and intraspecific
aggression’, ‘behavior towards moving objects and other animals’, ‘relationship of the dog-owner
team’, ‘test passe’ and ‘measures recommended’ in this report, together with the correlation of
aggression with dangerousness.
Not every test was able to evaluate all the criteria:
81
Test A is the only test to evaluate the behavior of the dog alone towards people and the potential
dangerousness of the dog alone for other people (points 14 and 15)
Test B is the only test to evaluate the aggressive behavior of the dog towards other animals and
the potential dangerousness of the dog for other animals (points 7 and 8)
Test C is the shortest test and is not able to evaluate the aggressive behavior of dogs towards
moving objects (point 5), towards other animals (point 7) and of the dog alone towards
people (point 15) , and the according potential dangerousness (points 6, 8, and 14)
The behavioral specialists were able to evaluate all the criteria.
Additionally, the experts could add comments to every point. These additional comments were
not considered in the evaluation, because they are not categorizable, but they were used to give
the owners feedback and to explain the evaluations by the experts.
3.3
THE EVALUATION SCALE
3.3.1 THE ORIGINAL SCALE
The evaluation sheets were organized in a multiple-choice manner. For some points, only one
answer was to be ticked (criteria 1, 3, 5, 7 and 14), for the other points the experts had the choice
to select several possible answers.
Table 2.11 shows an example-question from the original evaluation sheet 14 , which the experts
were asked to fill in for every dog. Every test and both behavior specialists received the same
questions in the same format. Because of the differences in structure of the tests, not all points
were answered by all the tests. Criterion 7 (behavior towards other animals) was only tested by
test B, whereas criterion 14 (interspecific aggression of the dog alone) was only tested in test A.
Test C is the shortest and least extensive test, it did not test any situations the other two tests did
not also consider.
Table 2.11 – Example of criteria
1. Intraspecific aggression
Behavior of the dog towards other dogs
1
2
3
4
5
6
open, friendly, neutral
dominant behavior
mistrustful, fearful, without aggression
Threatening / menacing
overt aggression / attack / biting with previous threatening
overt aggression / attack / biting without previous threatening
Comment:
A summary of the original scale the experts used is shown in Table 2.12.
14
translated from German into English by M. Bräm,
82
Table 2.12 – The original scale
Criteria
1. Intraspecific aggression
3.
Interspecific aggression
14. Behavior of dog alone
5.
7.
Behavior towards moving
objects
Aggressive behavior
towards other animals
11. Relationship dog – owner
12. Test passed or not
13. Measures suggested
Βold:
Possible answers
1 open, friendly, neutral
2 dominant behavior
3 mistrustful, fearful, without aggression
4 threatening, warning
5 overt aggression / attack / bite with threatening /
warning
6 overt aggression / attack / bite without threatening /
warning
1 open, friendly, neutral
2 mistrustful, fearful, without aggression
3 threatening, warning
4 overt aggression / attack / bite with threatening /
warning
5 overt aggression / attack / bite without threatening /
warning
1 hierarchy intact
2 dog dominant over owner
3 attachment existent
4 no attachment
5. dog shows aggressive behavior towards its owner
5.1 threatening / warning towards owner
5.2 attack / bite without threatening / warning
5.3 attack / bite without threatening / warning
5.4 owner corrects the dog
5.5 owner does not correct the dog when it is showing
aggressive behavior towards the owner
6 dog shows fearful behavior towards the owner
1 test passed
2 test not passed
3 ‘Ist-Stand’ passed
4 ‘Ist-Stand’ not passed
1 none
2 muzzle
3 leash
4 training course, education class
5 therapy
6 rehoming
7 euthanasia
only test B
83
3.3.2 THE DEFINITE SCALE
The scale for evaluating the aggressive behavior had to be altered and adapted for the following
reasons.
• The criteria that were not answered by any test were left out
• The scale was formed into an ordinal scale for the points 1, 3, 5, 7, 8 and 14. This meant
combining the second and the third points of the aggressive behavior, i.e.’ dominant
behavior’ and ‘mistrustful, fearful, without aggression’ to one second point ‘dominant
behavior or mistrustful fearful behavior without aggression’. The end scale for these six
criteria has five ordinal points for the aggressive behavior.
• If two boxes were ticked where only one was supposed to be within the answers of
aggressive behavior (1, 3, 5, 7, 14), the higher score was chosen.
• Within the points numbers 11 and 13 (relationship of dog and owner and measures),
every answer was considered separately, as they do not present an ordinal scale.
An overview of the definite scale is shown in Table 2.13.
Table 2.13 – The definite scale
Criteria
1. Intraspecific aggression
3.
Interspecific aggression
14. BEHAVIOR OF DOG ALONE
5.
Behavior towards moving objects
7.
Aggressive behavior towards
other animals
Answers
1 open, friendly, neutral
2 dominant behavior or mistrustful, fearful, without
aggression
3 threatening, warning
4 overt aggression / attack / bite with threatening /
warning
5 overt aggression / attack / bite without
threatening / warning
1 open, friendly, neutral
2 mistrustful, fearful, without aggression
3 threatening, warning
4 overt aggression / attack / bite with threatening /
warning
5 overt aggression / attack / bite without
threatening / warning
1 open, friendly, neutral
2 mistrustful, fearful, without aggression
3 threatening, warning
4 overt aggression / attack / bite with
threatening / warning
5 overt aggression / attack / bite without
threatening / warning
84
Criteria
11. Relationship dog – owner
12. Test passed or not
13. Measures suggested
Βold
CAPITALS
Answers
1 hierarchy intact
2 dog dominant over owner
3 attachment existent
4 no attachment
1 test passed
2 test not passed
3 ‘Ist-Stand’ passed
4 ‘Ist-Stand’ not passed
1 none
2 muzzle
3 leash
4 training course, education class
5 therapy
6 rehoming
-> only test B
-> ONLY TEST A
85
4. PROBLEMS DURING TESTING
As to be expected in any research project, some problems turned up during the implementation.
The problems that cropped up on this project were the following:
• Dogs missing:
• Two owners forgot their appointment and missed the test B
• One dog was injured on her paws and could not participate in test A
• Accidents: Unfortunately, there was one biting incident during a break. A female dog, just
having passed test C, attacked another female spayed dog during the coffee break. This is a
reminder that tests are only momentary pictures of a momentary situation of a dog’s behavior
and furthermore cannot include every possible situation that might occur in real life. It turned
out that the attacking dog had done this previously already - only towards female spayed dogs
– the stimulus dog in the test was an intact male. Consequently, she did not show any
aggressive behavior towards the stimulus dog in the test and passed the test successfully.
• Experts: One expert of Test A had to cancel his participation on short term, which forced us
to reschedule all 15 dogs for that day to other days. Four of these 15 dogs could be tested
another day by the other expert (Expert A1) and 11 were evaluated by an independent expert
(Expert A2) who was organized on short term. The consequence of this was that eight of the
dogs had two tests on one day and that the sequence of the tests could not be held for ten
dogs. We decided on this solution, because having all 15 dogs not run through this one test
would have forced us to not consider them at all in the comparison – which would have left us
with 45 dogs and would not be enough for statistical evaluation.
• Bias: While listening to the experts’ discussions among themselves and their remarks to us,
we noticed that a few of them were quite biased regarding some breeds. Looking at the
function and aim of their test explains why. It seemed to us that in the test where normally
only specific breeds are evaluated, these would have been judged more strictly. However,
none of the ‘fighting dog’ breeds participated in our project. In one test, it seemed that small
dogs were not taken as seriously as the large ones.
• Evaluation: The testers varied somewhat in their answers in the following ways: Some
questions were forgotten, some points not filled in and some boxes ticked twice where only
one answer would have been possible.
86
5. RESULTS
5.1
STATISTICAL PROGRAMS USED
The statistical evaluation was done with Microsoft Excel and NCSS (Number Cruncher
Statistical System).
Percentages, distributions and Chi-Square-Test were run in Microsoft Excel. Cross-Tabulations
were used to find the correlation between several points in NCSS. The Kappa-Test was run with
the SPSS-Program.
5.2
DESCRIPTIVE EVALUATION OF EVERY METHOD
The percentages of the criteria are given with 100% considered as the total number of dogs (N =
60) minus the missing answers. The missing values vary with each test and each criterion, as this
number is made up of:
• dogs that did not participate in the test
• dogs for which no answer were marked
• dogs receiving contradictory answers (e.g. ‘no attachment’ and ‘attachment’ both
marked)
For an overview of the results, see Appendix G.
5.2.1 DOGS
Tests
Fifty-nine of the planned 60 dogs were evaluated in Test A. One dog could not participate
because it was wounded. In Test B, 58 of the planned 60 dogs were evaluated. Two owners did
not make it to this test and hence, did not participate. All 60 dogs were evaluated in Test C.
Behavioral Specialists
The veterinary behaviorist and the ethologist each evaluated 12 dogs belonging to 9 owners.
Three of these dogs were seen both by the veterinarian and by the ethologist.
5.2.2 DATES
Tests
All tests were run on the actual dates planned, with the one above mentioned exception of Test A,
where 15 dogs had to be rescheduled to another day (see Fig. 2.7 and 2.8). This explains the
uneven distribution of the dogs among the four testing days.
In Tests B and C, all dogs were tested on the planned days. Two owners did not show up for Test
B on March 12th 2002, which explains the total number of only 58 dogs.
87
Figure 2.7 Dates Test A
Figure 2.8 Dates Test B
Behavioral Specialists
The ethologist evaluated 12 dogs in the month of February 2002, before the dogs ran through the
tests.
The veterinary behaviorist evaluated the 12 dogs on 4 days in the month of April 2002. The dogs
had all run through the tests already.
5.2.3 EXPERTS
Test A
Forty-nine of the dogs were tested on three days by Expert A1 who was familiar with the test, 11
dogs were evaluated on one day by Expert A2 who was not familiar with the test, but familiar in
evaluating canine behavior. The one missing dog is the one mentioned above that could not
participate because of a medical problem.
Figure 2.9 Experts Test A
Test B
Expert B2 evaluated all the dogs in the ‘Ist-Stand’, whereas Expert B2 and Expert B1 split the job
of judging the dogs that passed the ‘Ist-Stand’ in the actual test. Here, the judgements of the
person who had evaluated the dog in the actual test were considered, as that was the Expert who
answered the criteria of evaluation. The teams who did not pass the ‘Ist-Stand’ were additionally
tested in certain situations of the actual test in order to be able to answer our questions. In these
cases, it is the only expert who saw the dog who was taken into account.
There are three dogs missing, two of them had not shown up – and with one it was not noted
which expert did the evaluation.
Figure 2.10 gives an overview of the number of dogs evaluated by each expert and the dogs
missing.
88
Figure 2.10 Experts Test B
Test C
Figure 2.11 shows that both experts evaluated 30 dogs each on one day each. All dogs went
through this test.
Figure 2.11 Experts Test C
5.2.4 STIMULUS DOGS
Test A
A total of nine stimulus dogs of both sexes participated in this test (see Fig. 2.12) – a closer
description of them can be found in the section 2.1.7 ‘Stimulus dogs’.
Stimulus dogs S7 and S8 were used the most (13 and 23 times), dogs S8 and S3 were used
together 7 times, because they belonged to the same owner and it was not possible to put one of
these dogs into the car all the time. In two cases there is no information on which stimulus dog is
used. One of these dogs did not participate in the test, with the other the information is missing.
89
Figure 2.12 Stimulus dogs
Test B
The stimulus dogs used in this test were organized by the experts themselves and the
corresponding data for the stimulus dogs in this test were not available to us.
Test C
The dogs used in this test and the number of times they participated are shown in Figure 2.13.
Three intact male stimulus dogs split the job during the two testing days. They were all used on
an average of 20 times.
Figure 2.13 Stimulus dog Test C
90
5.2.5 SEQUENCE
Tests
Originally, 10 dogs were planned for every test sequence. Because of the rescheduling, there are
only 5 dogs left in the group BAC and 15 in the group BCA. One of the consequences of this reorganization was that 8 dogs had 2 tests, Test A and C, on the same day. (See Fig. 2.14)
Figure 2.14 Test sequence
5.2.6
EXACTNESS OF ANSWERS
Tests
It is quite striking to note that in Test B, 50 points are missing answers in as opposed to the 8
missing in of Test A and 2 in Test C. Twenty-three of these 50 missing answers are in the
criterion ‘behavior towards moving objects’; 12 for the criterion ‘behavior towards other
animals’. (See Fig. 2.15)
Figure 2.15 Exactness of answers
91
Behavioral Specialists
The veterinary behaviorist’s (VB) evaluation had a total of 6 missing answers, 4 of which were
within the criterion ‘hierarchy / attachment’. The evaluation of the ethologist (ET) showed two
missing values. (See Figure 2.16)
Figure 2.16 Exactness of answers
5.2.7 INTRASPECIFIC AGGRESSION (CRITERION 1)
Tests
All three tests evaluated the majority of the dogs to show ‘open, friendly, neutral’ behavior
towards other dogs, Test C found 12% of the dogs to show ‘threatening behavior’ as opposed to
only 3% and 4% in the other two tests. No dog attacked without threatening in any of the three
tests. (See Table 2.14)
Table 2.14 Intraspecific aggression, Tests A, B and C
Absolute
number of
dogs
Test score / Tests
A
B
Percentage of total (N =
60) minus missing values
C
A
B
C
Open, friendly, neutral behavior
45
43
38
76%
78%
64%
Dominant behavior or mistrustful, fearful
behavior without aggression
12
8
13
20%
15%
22%
Threatening behavior
2
2
7
3%
4%
12%
Attacking with threatening
0
2
1
0%
4%
2%
Attacking without threatening
0
0
0
0%
0%
0%
Missing values
1
5
1
Total
60
Behavioral Specialists
92
60
60
100%
100%
100%
The veterinary behaviorist evaluated each 25% (N = 3) of the dogs to show ‘open, friendly,
neutral behavior’, ‘dominant behavior or mistrustful, fearful behavior without aggression’ and
‘threatening behavior’ towards other dogs, two dogs were evaluated to attack with threatening
and one dog to show ‘attacking behavior without threatening’
The ethologist evaluated 50% (N = 6) of the dogs to show ‘threatening behavior’ towards other
dogs, 25% (N = 3) to show ‘dominant behavior or mistrustful, fearful behavior without
aggression’, 17% (N = 2) to show ‘open, friendly, neutral behavior’ and 8% (N = 1) to ‘attack
with threatening’. No dogs were considered to ‘attack without threatening by the ethologist. (See
Table 2.15)
Table 2.15 Intraspecific aggression, behavioral specialists
Absolute
number of
dogs
Test score
Percentage of
total
(N =
12)
VB
ET
VB
ET
Open, friendly, neutral behavior
3
2
25%
17%
Dominant behavior or mistrustful, fearful
behavior without aggression
3
3
25%
25%
Threatening behavior
3
6
25%
50%
Attacking with threatening
2
1
17%
8%
Attacking without threatening
1
0
8%
0%
Total
12
12
100%
100%
5.2.8 INTERSPECIFIC AGGRESSION (CRITERION 3)
Tests
Most of the dogs were evaluated to show ‘open, friendly, neutral’ behavior towards humans in all
three tests, with Test B giving this answer in only 61% of the cases as opposed to the other two
(Test A 83% and Test B 88%). 30% of the dogs in Test B were evaluated to show ‘dominant
behavior or mistrustful, fearful behavior without aggression’ which is almost three times more
than this was the case in the other two tests. Test A found three dogs to attack with threatening,
Test B one, Test C none. (See Table 2.16)
93
Table 2.16 Interspecific aggression, Tests A, B and C
Percentage of total (N = 60)
minus missing values
Absolute number of dogs
Test score
A
B
C
A
B
C
Open, friendly, neutral behavior
48
34
52
83%
61%
88%
Dominant behavior or mistrustful,
fearful behavior without aggression
5
17
6
9%
30%
10%
Threatening behavior
2
4
1
3%
7%
2%
Attacking with threatening
3
1
0
5%
2%
0%
Attacking without threatening
0
0
0
0%
0%
0%
Missing values
2
4
1
Total
60
60
60
100%
100%
100%
Behavioral Specialists
Three dogs each were evaluated by the veterinary behaviorist to show ‘threatening behavior’ and
‘attacking without threatening’ and two dogs ‘dominant behavior or mistrustful, fearful behavior
without aggression’.
The majority of the dogs (75%, N = 9) evaluated by the ethologist was considered to be ‘open,
friendly, neutral’ towards humans, the remaining 3 dogs being evenly distributed amongst the
categories ‘dominant behavior or mistrustful, fearful behavior without aggression’, ‘threatening
behavior’ and ‘attacking with threatening’. No dogs were evaluated to ‘attack without
threatening’ by the ethologist. (See Table 2.17)
Table 2.17 Interspecific aggression, behavioral specialists
Absolute
number of
dogs
Test score
Percentage of
total (N = 12)
VB
ET
VB
ET
Open, friendly, neutral behavior
4
9
33%
75%
Dominant behavior or mistrustful, fearful
behavior without aggression
2
1
17%
8%
Threatening behavior
3
1
25%
8%
Attacking with threatening
0
1
0%
8%
Attacking without threatening
3
0
25%
0%
Total
12
12
100%
100%
94
5.2.9 INTERSPECIFIC AGGRESSION, DOG ALONE (CRITERION 14)
Tests
Test A was the only one to evaluate this situation. Eighty-eight% of the dogs were considered to
show ‘open, friendly, neutral’ behavior towards humans, 8% showed ‘mistrustful, fearful
behavior without aggression’ and only 1 dog threatened and 1 dog attacked with threatening. (See
Table, 2.18)
Table 2.18 Interspecific aggression, dog alone, Test A
Percent
Absolu age of
te
total (N
numbe
= 60)
r of
minus
dogs missing
values
Test score
A
A
Open, friendly, neutral behavior
52
88%
Dominant behavior or mistrustful, fearful behavior
without aggression
5
8%
Threatening behavior
1
2%
Attacking with threatening
1
2%
Attacking without threatening
0
0%
Missing values
1
Total
60
100%
Behavioral Specialists
The veterinary behaviorist evaluated 42% (N = 5) of the dogs to show ‘open, friendly, neutral
behavior’ towards humans when left alone, 2 each to show ‘dominant behavior or mistrustful,
fearful behavior without aggression’ and ‘threatening behavior’, 3 dogs were considered to
‘attack without threatening’. (See Table 2.19)
The ethologist did not evaluate this criterion.
95
Table 2.19 Interspecific aggression, dog alone, veterinary behaviorist
Absolute number of
dogs
Percentage of total
(N = 12)
VB
VB
Open, friendly, neutral behavior
5
42%
Dominant behavior or mistrustful, fearful behavior
without aggression
2
17%
Threatening behavior
2
17%
Attacking with threatening
0
0%
Attacking without threatening
3
25%
Total
12
100%
Test score
5.2.10 BEHAVIOR TOWARDS MOVING OBJECTS (CRITERION 5)
Tests
Test A and B are the two tests that evaluated this point. Twenty-three dogs did not receive an
answer to this question in Test B. Results of Test A evaluated a higher percentage of dogs to
show ‘dominant behavior or mistrustful, fearful behavior without aggression’ than did Test B.
This difference may have to do with the fact that no answer was given for 38% of the dogs in
Test B, which makes a comparison more difficult, as only 37 dogs are left to be considered,
which were tested in both tests. (See Table 2.20)
This situation was not tested in Test C.
Table 2.20 Behavior towards moving objects, Tests A and B
Absolute number of
dogs
Test
Percentage of total (N = 60)
minus missing values
A
B
A
B
Open, friendly, neutral behavior
49
35
83%
95%
Dominant behavior or mistrustful, fearful
behavior without aggression
10
2
17%
5%
Threatening behavior
0
0
0%
0%
Attacking with threatening
0
0
0%
0%
Attacking without threatening
0
0
0%
0%
Missing values
1
23
Total
60
60
100%
100%
score
Behavioral Specialists
96
The veterinary behaviorist considered 45% (N = 5) of the dogs to show ‘open, friendly, neutral’
behavior towards moving objects, whereas 55% (N = 6) were evaluated to ‘attack without
threatening’.
The ethologist evaluated 82% (N = 9) of the dogs to show ‘open, friendly, neutral behavior’
towards moving objects and only 18% (N = 2) to show ‘dominant behavior or mistrustful, fearful
behavior without aggression’. (See Table 2.21)
Table 2.21 Behavior towards moving objects, behavioral specialists
Absolute number of dogs
Test score
Percentage of total (N = 12)
minus missing values
VB
ET
VB
ET
Open, friendly, neutral behavior
5
9
45%
82%
Dominant behavior or mistrustful, fearful
behavior without aggression
0
2
0%
18%
Threatening behavior
0
0
0%
0%
Attacking with threatening
0
0
0%
0%
Attacking without threatening
6
0
55%
0%
Missing values
1
1
Total
12
12
100%
100%
5.2.11 AGGRESSIVE BEHAVIOR TOWARDS OTHER ANIMALS (CRITERION 7)
Tests
This situation was only evaluated in Test B. Unfortunately, 20% of the dogs did not receive an
answer to this point (taking into account that two dogs did not run through this test). 79% of the
animals actually tested were evaluated to show ‘open, friendly, neutral’ behavior towards other
animals. (See Table 2.22)
97
Table 2.22 Behavior towards other animals, Test B
Absolute number of dogs
Percentage of total (N =
60) minus missing values
Test score
B
B
Open, friendly, neutral behavior
38
79%
Dominant behavior or mistrustful, fearful behavior without
aggression
7
15%
Threatening behavior
2
4%
Attacking with threatening
1
2%
Attacking without threatening
0
0%
Missing values
12
Total
60
100%
Behavioral Specialists
The veterinary behaviorist considered nearly all the dogs (N = 9) to ‘attack without threatening’,
whereas 2 ‘attack with threatening’.
The ethologist evaluated 45% (N = 5) of the dogs to show ‘open, friendly, neutral behavior’
towards other animals, 36% (N = 4) ‘dominant behavior or mistrustful, fearful behavior’ and 17%
(N = 2) ‘threatening behavior’. (See Table 2.23)
Table 2.23 Behavior towards other animals, behavioral specialists
Absolute number of dogs
Test score
Percentage of total (N = 12)
minus missing values
VB
ET
VB
ET
Open, friendly, neutral behavior
0
5
0%
45%
Dominant behavior or mistrustful, fearful
behavior without aggression
0
4
0%
36%
Threatening behavior
0
2
0%
18%
Attacking with threatening
2
0
18%
0%
Attacking without threatening
9
0
82%
0%
Missing values
1
1
Total
12
12
100%
100%
98
5.2.12 RELATIONSHIP DOG AND OWNER (CRITERION 11)
5.2.12.1 ‘Hierarchy’
Tests
In the majority of the teams, the owner was evaluated to be ‘dominant’ in all three tests (Test
A: 81%, Test B: 93%, Test C: 91%). Test A considered 10 dogs to be ‘dominant over their
owners’, whereas this option was chosen less often in tests B (N = 1) and C (N = 5). The
number of missing answers to this question is high for Test B (N = 46). (See Table 2.24)
Table 2.24 ‘Hierarchy’, Tests A, B and C
Absolute
Percentage of total (N =
number of dogs 60) minus missing values
Answer
A
B
C
A
B
C
Owner dominant
43
13
50
81%
93%
91%
Dog dominant
10
1
5
19%
7%
9%
Missing values
7
46
5
Total
60
60
60
100%
100%
100%
Behavioral specialists
For the 8 teams for whom the veterinary behaviorist answered this criterion, the owner was
considered to be ‘dominant’.
The ethologist considered the owners to be ‘dominant’ in half the cases (N = 6) and the dogs
to be ‘dominant’ in the other half. (See Table 2.25)
Table 2. 25 ‘Hierarchy’, behavioral specialists
Absolute
number of
dogs
Answer
Percentage of
total (N = 12)
minus missing
values
VB
ET
VB
ET
Owner dominant
8
6
100%
50%
Dog dominant
0
6
0%
50%
Missing values
4
0
Total
12
12
100%
100%
99
5.2.12.2 ‘Attachment’
Tests
This criterion was not answered consequently, mainly in Test C where 50% of the dogs did
not receive an answer. Of the dogs evaluated, the majority was considered to ‘have an
attachment’ to their owners in all three tests. 77% (N = 40) of the dogs receiving an answer
of this question in Test A, 83% (N = 35) in Test B and 97% (N = 29) in Test C had the point
‘attachment existent’ marked. (See Table 2.26)
Behavioral Specialists
The veterinary behaviorist and the ethologist considered all 12 dogs to have ‘an attachment’
to their owner. (See Table 2.27)
Table 2.26 ‘Attachment’, Tests A, B and C
Absolute
number of dogs
Percentage of total (N =
60) minus missing
values
Testscore
A
B
C
A
B
C
Attachment existent
40
35
29
77%
83%
97%
No attachment
12
7
1
23%
17%
3%
Missing values
8
18
30
Total
60
60
60
100%
100%
100%
Table 2.27 ‘Attachment’ behavioral specialists
Absolute Percentage of
number of total
(N =
dogs
12)
Test score
VB
ET
VB
ET
Attachment existent
12
12
100%
100%
No attachment
0
0
0%
0%
Total
12
12
100%
100%
100
5.2.13 TEST PASSED / NOT PASSED (CRITERION 12)
All dogs passed Test C and 97% of the dogs passed Test A.
In Test B, passing the ‘Ist-Stand’ and the actual test was considered separately. 36% of the dogs
failed the ‘Ist-Stand’, and therefore were also considered to have failed the test, 68% passed the
‘Ist-Stand’. Of these 38 dogs that passed the ‘Ist-Stand’, 29 also passed the test, whereas 9 failed.
(See Table 2.28)
Table 2.28 Test A, B and C, passed / not passed
100% = Total minus missing
Absolute
Percentage of total (N =
number of dogs 60) minus missing values
Test score
A
B
C
A
B
C
Passed
57
29
60
97%
50%
100%
Failed
2
29
3%
50%
0%
Ist-Stand passed
38
0%
66%
0%
Ist-Stand failed
20
0%
34%
0%
100%
200%
100%
Missing values
1
2
0
Total
60
118
60
5.2.14 MEASURES RECOMMENDED (CRITERION 13)
OVERVIEW
Test
Tests A and C did not recommend any measures in the majority of the cases (85% (N = 50)
and 82% (N = 49), resp.), with the remaining 15% (N = 9) and 18% (N = 11) of the teams
receiving one measure. Test B however, recommended one measure in 64% (N = 32) of the
cases, and two measures in 18% (N = 9), leaving only 18% (N = 9) of the teams without any
measures. (See Table 2.29)
Table 2.29 Measures recommended by Tests A, B and C
Absolute
number of
dogs
Percentage of total (N =
60) minus missing values
Test score
A
B
C
A
B
C
No measures
50
9
49
85%
18%
82%
One measure
9
32
11
15%
64%
18%
> 1 measure
0
9
0
0%
18%
0%
No answer
0
8
0
Missing values
1
2
0
Total
60
60
60
100%
100%
100%
101
Behavioral Specialists
The veterinary behaviorist gave an answer to this question for all dogs. For 50% (N = 6) no
measures were recommended, 2 received one measure and 4 received more than one
measure.
67% (N = 8) of the dogs evaluated by the ethologist, received no measures and 33% (N = 4)
one measure. (See Table 2.30)
Table 2.30 ‘Measures’ recommended by the behavioral specialists
Absolute
number of
dogs
Test score
Percentage of
total (N = 12)
VB
ET
VB
ET
No measures
6
8
50%
67%
One measure
2
4
17%
33%
> 1 measure
4
0
33%
0%
No answer
0
0
0%
0%
Total
12
12
100%
100%
MUZZLE
This measure was recommended only once by Test C and three times by the veterinary
behaviorist. (See Table 2.31)
Table 2.31 Muzzle, recommended by Tests A, B and C
Absolute
number of dogs
Percentage of total
(N = 60) minus
missing values
Test score
A
B
C
A
B
C
Marked
0
0
1
0%
0%
2%
Not marked
59
58
59
100%
100%
98%
Missing values
1
2
0
Total
60
60
60
100%
100%
100%
OBLIGATORY LEASH
Tests
Test B is the only test to have recommended this measure (18% (N = 11) of the cases). These
18% are the same 18% receiving more than one measure. This measure was always
combined with the measure ‘education class’. (See Table 2.32)
102
Table 2.32 Obligatory leash, recommended by Tests A, B and C
Absolute
number of
dogs
Percentage of total (N =
60) minus missing values
Test score
A
B
C
A
B
C
Marked
0
11
0
0%
19%
0%
Not marked
59
47
60
100%
81%
100%
Missing values
1
2
0
Total
60
60
60
100%
100%
100%
Behavioral Specialists
The Veterinary behaviorist recommended the measure of ‘obligatory leash) to 3 teams, the
ethologist in no case. (See Table 2.33)
Table 2.33 Obligatory leash, recommended by the behavioral specialists
Absolute
number of
dogs
Test score
Percentage of
total (N = 12)
VB
ET
VB
ET
Marked
3
0
25%
0%
Not marked
9
0
75%
0%
Total
12
0
100%
0%
EDUCATION CLASS
Tests
‘Education class’ is the answer most often given by Test B, 66% (N = 38) of the dogs
received this measure, as opposed to only 12% (N = 7) and 17% (N = 10) in tests A and C,
resp. (See Table 2.34)
Table 2.34 Education class, recommended by Tests A, B and C
Absolute number
of dogs
Percentage of total (N = 60)
minus missing values
Test score
A
B
C
A
B
C
Marked
7
38
10
12%
66%
17%
Not marked
52
20
50
88%
34%
83%
Missing values
1
2
0
Total
60
60
60
100%
100%
100%
103
Behavioral specialists
The veterinary behaviorist recommended this measure for 2 teams (17%). The ethologist
Imposed this measure in one case (8%). (See Table 2.35)
Table 2.35 Education class, recommended by the behavioral specialists
Absolute number
of dogs
Test score
Percentage of
total (N = 12)
VB
ET
VB
ET
Marked
2
1
17%
8%
Not marked
10
11
83%
92%
Total
12
12
100%
100%
THERAPY
Therapy was recommended in only in 2 cases in Test A (3%) and 1 case in Test B (2%) and
in 5 cases (42%) by the veterinary behaviorist. The ethologist did recommended this measure
in 25% (N = 3) of the cases. (See Tables 2.36 and 2.37)
Table 2.36 Therapy, recommended by Tests A, B and C
Absolute
number of dogs
Percentage of total (N
= 60) minus missing
values
Test score
A
B
C
A
B
C
Marked
2
1
0
3%
2%
0%
Not marked
57
57
60
97%
98%
100%
Missing values
1
2
0
Total
60
60
60
100%
100%
100%
Table 2.37 Therapy, recommended by the behavioral specialists
Absolute
number of
dogs
Percentage of
total (N = 12)
VB
ET
VB
ET
Marked
5
3
42%
25%
Not marked
7
9
58%
75%
Total
12
12
100%
100%
Test score
104
REHOMING
Test B was the only test to recommend rehoming of two dogs. The veterinary behaviorist did
not consider this measure in any of the cases. The veterinary behaviorist and ethologist did
not recommend this measure in any case. (See Table 2.38)
Table 2.38 Rehoming, recommended by Tests A, B, and C
Absolute
number of
dogs
Percentage of total (N =
60) minus missing values
Test score
A
B
C
A
B
C
Marked
0
2
0
0%
3%
0%
Not marked
59
56
60
100%
97%
100%
Missing values
1
2
0
Total
60
60
60
100%
100%
100%
EUTHANASIA
Euthanasia was not recommended by any test for any dog.
105
5.3
RESULTS OF THE COMPARISON OF THE FIVE METHODS
This chapter is organized into subchapters according to the criterion considered. Within these
subchapters a distinction is made between ‘descriptive comparison’, where the results are
compared in absolute numbers and ‘statistical comparison’, where a Kappa Test was run.
With the Kappa Test, the results of the criteria ‘intraspecific aggression towards other dogs’,
‘interspecific aggression towards humans’, ‘hierarchy’, ‘attachment’ and ‘measures’ of the three
tests were compared using the kappa test. The criterion ‘behavior towards moving objects’ of
Test A and B was compared (Test C did not evaluate this point).
For an overview of the results of the Kappa-statistics of this section see Appendix H, for an
overview of the descriptive comparison of the three tests see Appendix I
Explanation of the Scales
The aggression scale on the y-axis:
1 open, friendly, neutral behavior
2 dominant behavior or mistrustful, fearful behavior
3 threatening behavior
4 attack with threatening
5 attack without threatening
‚Hierarchy‘
1 owner dominant over dog
2 dog dominant over owner
‚Attachment‘ to the owner
1 attachment existent
2 no attachment
‚Measures‘
1 no measures recommended
2 measures recommended
106
5.3.1
INTRASPECIFIC AGGRESSION OF THE DOG TOWARDS OTHER DOGS
(CRITERION 1)
5.3.1.1
Descriptive Results
All three tests agree in 25 and disagree in 29 of the 54 cases that were evaluated by all the tests.
Tests A and B have the highest percentage of agreement (76%) and Tests A and C, and Tests B
and C, both show an agreement in 59% of the cases. (See Fig. 2.17a and 2.17b)
Figure 2.17a Intraspecific aggression in Tests A, B and C, dogs 1 to 29
Figure 2.17b Intraspecific aggression in Tests A, B and C, dogs 30 to 60
107
The highest agreement was found between the veterinary behaviorist and Test B with 33% (N =
4), the lowest with Test C (17%). (See Fig. 2.18a)
Figure 2.18a Intraspecific aggression, veterinarian
The highest agreement was found between the evaluation of the ethologist and Test A (25%), the
ethologist agreed with the other two tests in two cases each (17%). All four experts agreed in one
case. (See Fig. 2.18b)
Figure 2.18b Intraspecific aggression, ethologist
108
5.3.1.2
Statistical Results
A significant agreement of the results was found when comparing the three tests (Kappa = 0.133,
p = 0.014), meaning that the agreement was higher than what would be expected by chance. For
the answers ‘open, friendly, neutral’ and ‘threatening behavior’, the agreement was significantly
different from zero.
5.3.2
INTERSPECIFIC AGGRESSION OF THE DOG TOWARDS HUMANS
(CRITERION 3)
5.3.2.1
Descriptive Results
Tests A and C showed an agreement of 77% (N = 44), whereas tests A and B, and Tests B and C
agreed in 65% of the cases each. An agreement of 55% was found of all three tests. (See Fig.
2.19a and 2.19b)
Figure 2.19a Interspecific aggression in Tests A, B and C, dogs 1 to 29
Figure 2.19b Interspecific aggression in Tests A, B and C, dogs 30 to 60
109
The veterinary behaviorist and the expert of Test A agreed on their evaluation of the dogs in 45%
of the cases (N = 5). The agreement with tests B (27%) and C (33%) was lower. All four of the
experts agreed only in 20% (N = 2) of the cases evaluated by all of them. (See Fig. 2.20a)
Figure 2.20a Interspecific aggression, veterinarian
There is a high agreement of the ethologist with all three tests (73% to 82%), in 73% of the cases,
all four experts agree on the evaluation of the dog. (See Fig. 2.20b)
Figure 2.20b Interspecific aggression, ethologist
5.3.2.2
Statistical Results
A significant agreement of the results was found when comparing the three tests (Kappa = 0.135,
p = 0.014). For the answers ‘open, friendly, neutral’ ‘threatening behavior’ and ‘attacking with
threatening’, the agreement was significantly different from zero.
110
5.3.3 BEHAVIOR OF THE DOG (ALONE) TOWARDS HUMANS
Only Test A evaluated this situation, consequently, it cannot be compared to the results of the
other two tests. The veterinary behaviorist and Test A agree on the behavior of the dogs in 50%
(N = 6) of the cases (See Fig. 2.21). The ethologist did not answer this point for any dog.
Figure 2.21 Interspecific aggression (dog alone), veterinarian
5.3.4
THE BEHAVIOR OF THE DOG TOWARDS MOVING OBJECTS (JOGGER,
BICYCLE)
Only Tests A and B included a situation where this criterion could to be answered. Both the
veterinary behaviorist and the ethologist assessed the behavior of the dogs in these situations.
5.3.4.1
Descriptive Results
Tests A and B agreed in 84% (N = 31) of the cases evaluated by both of them. (See Fig. 2.22a
and Fig. 2.22b)
Figure 2.22a Behavior towards moving objects in Tests A, B and C, dogs 1 to 29
111
Figure 2.22b Behavior towards moving objects in Tests A, B and C, dogs 30 to 60
The veterinary behaviorist agreed with the experts of Tests A and B in 36% (N = 4) and 40% (N
= 2) respectively. All three experts agreed in 40% (N = 2) of the cases evaluated by all of them.
(See Fig. 2.23a)
Figure 2.23a Behavior towards moving objects, veterinarian
112
The ethologist agreed in 82% (N = 9) of the cases with Test A and in 78% (N = 7) with Test B.
All three methods agreed in 73% (N = 7) of the cases evaluated by all three of them. (See Fig.
2.23b)
Figure 2.23b Behavior towards moving objects, ethologist
5.3.4.2
Statistical Results
Tests A and B evaluated this situation, whereas Test C did not. Consequently, the only
comparison that can be made is between the first two tests.
There is no agreement surpassing the randomly expected results when comparing this criterion of
Test A with Test B (p 0.091).
113
5.3.5 AGGRESSIVE BEHAVIOR OF THE DOG TOWARDS OTHER ANIMALS
This situation was only evaluated in Test B. The veterinary behaviorist does not agree with Test
B in any of the 9 cases evaluated by both methods and consistently evaluates the dogs higher in
the scale than does the expert of Test B. (See Fig. 2.24a)
Figure 2.24a Behavior towards other animals, veterinarian
The ethologist agrees with Test B in 50% (N = 5) of the cases, evaluating the dogs higher in the
aggression scale in the cases where they differ. (See Fig. 2.24b)
Figure 2.24b Behavior towards other animals, ethologist
114
5.3.6
5.3.6.1
THE RELATIONSHIP OF THE DOG AND ITS OWNER
Descriptive Results
‘Hierarchy’
Tests B and C agreed on the behavior of 84% (N = 21) of the dogs evaluated by both tests, Tests
A and B in 72% (N = 26), and A and C in 76% (N = 37) of the cases. All three methods agreed in
36% (N = 16) of all the cases that received an answer to this question. (See Fig. 2.25a and 2.25b)
Figure 2.25a, ‘Hierarchy’ in Tests A, B and C, dogs 1 to 29
Figure 2.25b ‘Hierarchy’ in Tests A, B and C, dogs 30 to 60
115
An agreement of 75% (N = 6) was found between the veterinarian and Test A, and of 83% (N =
5) between the veterinarian and Test C. Test B did not answer this point for any of the dogs
evaluated by the veterinary behaviorist. (See Fig. 2.26a)
Figure 2.26a ‘Hierarchy’, veterinarian
The ethologist agreed in 100% of the cases with Test B, however, this test only evaluated 4 dogs
in this point. 80% (N = 8) agreement was found with Test A, and 55% (N = 6) with Test B. All
three tests agree in 3 cases. This number is limited by Test B, which only answered this question
in 4 cases. (See Fig. 2.26b)
Figure 2.26b ‘Hierarchy’, ethologist
116
‘Attachment’
Test C has an agreement of 81% (N = 21) and 84% (N = 21) with tests A and B, respectively.
Tests B and C agree in 72% (N = 26) of the cases evaluated by both of them. All three tests agree
in 36% (N = 16) of the cases. The total number of dogs evaluated by all three tests is mainly
limited by Test C, where only one expert answered this point, and by Test B where not all the
dogs received an evaluation either. (See Fig. 2.27a and 2.27b)
Figure 2.27a ‘Attachment’ in Tests A, B and C, dogs 1 to 29
Figure 2.27b ‘Attachment’ in Tests A, B and C, dogs 30 to 60
117
Behavioral Specialists
The veterinarian agreed with Test C in 86% (N = 6) of the cases, with Test B in 67% (N = 8) and
with Test A in 55% (N = 6) of the cases evaluated by the both test considered. (See Fig. 2.28a)
Figure 2.28a ‘Attachment’, veterinarian
In those cases where both the ethologist and the test gave an answer to this criterion, they both
agreed in 100% of the cases that there was an attachment of the dog to the owner. (See Fig.
2.28b)
Figure 2.28b ‘Attachment’, ethologist
5.3.6.2
Statistical Results
There is no agreement of the results surpassing the randomly expected when comparing the
criteria ‘hierarchy’ of the three tests. A significant agreement was found for the point
‘attachment’ (Kappa = 0.352, p = 0.003).
118
5.3.7 TEST PASSED OR NOT
This point was not run through the Kappa test, as almost all the dogs passed all three tests.
5.3.8
MEASURES RECOMMENDED
5.3.8.1
Descriptive Results
Tests A and C agree in 75% (N = 44) of the cases, tests A and B in 54% (N = 31) and Tests B and
C in 50% (N = 29). All three tests give the same answer in 50% (N = 29) of the cases. (See Fig.
2.29a and 2.29b)
Figure 2.29a Measures recommended by Tests A, B and C, dogs 1 to 29
Figure 2.29b Measures recommended by Tests A, B and C, dogs 30 to 60
119
The veterinary behaviorist agrees with Test A in 58% (N = 7) of the cases, with Test B in 45% (N
= 5) and Test C 42% (N = 5). There is no case where all four methods agree on the same
measures. (See Fig. 2.30a)
Figure 2.30a Measures recommended by the veterinarian
The ethologist and Test B agree on whether to recommend measures or not in 70% (N = 7) of the
cases. The agreement with both Tests A and C is 67% (N = 8). All four methods agree in 30% (N
= 3) of the cases evaluated by all of them. (See Fig. 2.30b)
Figure 2.30b Measures recommended by the ethologist
5.3.8.2
Statistical Results
There is no agreement surpassing the randomly expected results when comparing the criteria of
measures of the three tests.
120
5.4
RESULTS OF CORRELATION WITHIN THE TESTS / METHODS
For a detailed overview of the results, see Appendix J.
The following groups were pooled in order to create larger groups consisting of a greater number
of individual subjects:
1. The sex of the dogs was pooled into two groups: only males and females are distinguished,
the neutering status is not taken into account.
2. The weight of the dogs was pooled into two groups: 25 kg or less and more than 25 kg.
3. The weights and sex of the stimulus dogs were grouped in the same way.
4. The age was pooled into two groups: three years or younger and older than three years
5. Relationship dog-owner: The criterion ‘relationship dog-owner (Criterion 11) was divided
into two categories, allowing the consideration of the missing values. These two categories
are ‘hierarchy’ and ‘attachment’ with the according answers being ‘owner dominant over
dog’ and ‘dog dominant over owner’ and ‘attachment existent’ and ‘no attachment’,
respectively.
6. Measures: a new criterion ‘measures’ was created to give an overview of the measures. The
two possible answers are: ‘no measures recommended’ and ‘measures recommended’
7. The aggression scale was pooled to a two point scale. For the tests, the two values ‘open,
friendly, neutral behavior’ (which will be referred to as ‘neutral behavior’) and ‘dominant
behavior or mistrustful-fearful or aggressive behavior’ (which will be referred to as ‘not
neutral behavior’) were created. As the veterinary behaviorist and the ethologist generally
evaluated the animals higher in the aggression scale, it was possible to pool the answers
into the two groups ‘not aggressive’ and ‘aggressive’ behavior. This does not make much
sense for the tests, as almost all the dogs would be found in the group ‘not aggressive’
behavior.
The number of dogs being evaluated varies from criterion to criterion, hence, the total number of
dogs showing, for example, ‘neutral behavior’ towards other dogs does not always correspond
with the total number of dogs showing this behavior towards humans or moving objects, etc.
Caution must therefore be taken when comparing the percentages of the results. The results will
be given as follows: x% (N, T), with x% standing for the percentage of the Total (T) of the dogs
that were evaluated in this category and ‘N’ being the absolute number.
The null hypothesis of all the tests of correlation within this chapter is that there is no correlation
between the two criteria being looked at. The alternate hypothesis is that there is a correlation.
Within some of the chapters, it was predicted in advance in what direction the correlation was
expected to go (one-tailed p-value) in others not (two-tailed p-value).
5.4.1 BREED
All the groups in the original scale of all three tests showed at least one group with expected
values less than five individuals, not allowing a reliable interpretation of the Chi-square test. The
pooling of the answers did not increase the number of individuals within every group enough to
receive reliable results in a Fisher’s exact test. Hence, no statement can be made whether this
point is correlated with any of the others or not.
121
5.4.2 SEX OF DOG
None of the criteria of any of the three tests fulfilled the condition for a Chi-square test. Pooling
the answers into ‘neutral behavior’ and ‘not neutral behavior’ on the one hand for the tests and
into ‘aggressive behavior’ and ‘not aggressive behavior’ for the veterinary behaviorist and the
ethologist, and female and male on the other hand, allows the Fisher’s exact test to be run. No
correlation was found for Tests A and B in any point.
Test C
In Test C, a significant correlation (p = 0.007, two-tailed) was found with the variable
‘intraspecific aggression’. 79% (N = 27, T = 34) of the female dogs were evaluated to show
‘neutral behavior’ towards other dogs, whereas 44% (N = 11, T = 25) of the male dogs were
considered to show this behavior. (See Table 2.39)
Table 2.39 Correlation of sex and intraspecific aggression, Test C
Sex
Absolute number of dogs
Test C
female
male
Total
Neutral behavior
27
11
38
46%
19%
64%
Not neutral behavior
7
14
21
12%
24%
36%
Total
34
25
59
58%
42%
100%
Intraspecific aggression
p-value
Percentage of Total (N =
59)
female male
Total
0.007
two-tailed
No correlation was found within the cases evaluated by the veterinary behaviorist and the
ethologist.
5.4.3 WEIGHT OF DOG
None of the criteria fulfilled the condition for the Chi-square test to be run with the original
results.
Tests A and B
The Fisher’s Exact Test was run on the pooled answers of the two-point aggression scale and on
dogs weighing 25kg or less and more than 25kg. This led to a significant correlation with the
variable ‘intraspecific aggression’ in Test A (p = 0.015, two-tailed) and C (p = 0.013, two-tailed).
No correlation was found in Test B.
In Test A, 89% (N = 29, T = 33) of the dogs weighing 25kg or less were evaluated to show
‘neutral behavior’ towards other dogs, whereas 59% (N = 14, T = 24) of the dogs weighing more
than 25 kg were considered to show this behavior. (See Table 2.40)
In Test C, 78% N = 25, T = 32) of the dogs weighing 25kg or less were considered to show
‘neutral behavior’ towards other dogs, whereas only 44% (N = 11, T = 25) of the dogs weighing
more than 25 kg showed this behavior. (See Table 2.40)
122
Table 2.40 Correlation of intraspecific aggression with weight, Tests A and C
Absolute number of dogs
Test A
Intraspecific
aggression
≤ 25 kg
> 25 kg
Total
≤ 25 kg
> 25 kg
Total
Neutral behavior
29
14
43
51%
25%
75%
Not neutral behavior
4
10
14
7%
18%
25%
Total
33
24
57
58%
42%
100%
Absolute number of dogs
Test C
Weight
Percentage of Total
(N = 57)
Intraspecific
aggression
p-value
0.014
two-tailed
Weight
Percentage of Total
(N = 57)
≤ 25 kg
> 25 kg
Total
≤ 25 kg
> 25 kg
Total
p-value
Neutral behavior
25
11
36
44%
19%
63%
Not neutral behavior
7
14
21
12%
25%
37%
Total
32
25
57
56%
44%
100%
0.013
two-tailed
No correlation was found with the weight in any point for the dogs evaluated by the veterinary
behaviorist or the ethologist.
5.4.4 AGE
Test C
A correlation was only found in Test C with the variable ‘interspecific aggression (p = 0.048,
two-tailed). 93% (N = 42, T = 45) of the dogs older than three years of age were evaluated to
show ‘neutral behavior’ towards humans. 71% (N = 10, T = 14) of the dogs three years or
younger showed this behavior. (See Table 2.41)
Table 2.41 Correlation of interspecific aggression with age, Test C
Age
Absolute number of dogs
Test C
Interspecific
aggression
≤ 3 years
> 3 years Total p-value
Percentage of Total
(N = 59)
≤ 3 years
> 3 years
Total
Neutral behavior
10
42
52
17%
71%
88%
Not neutral behavior
4
3
7
7%
5%
12%
Total
14
45
59
24%
76%
100%
123
0.048
twotailed
5.4.5 EXPERT
Test A
In Test A, a correlation was found in the Fisher’s Exact Test between the group ‘expert’ and the
variables ‘hierarchy’ (p = 0.004, two-tailed), ‘attachment’ (p = 0.011, two-tailed), ‘measures’ (p
< 0.001, two-tailed) and ‘education class’ (p < 0.001, two-tailed). A higher percentage of owners
were found with expert A1 than with expert A2 to be ‘dominant over their dogs’ (A1: 89% (T =
45), A2: 38% (T = 8). A higher percentage of dogs were evaluated by expert A1 to have ‘an
attachment’ (A1: 84% (T = 44), A2: 38% (T = 8)), not to receive measures (A1: 94% (T = 49),
A2: 40% (T = 10)) and to not receive the measure ‘education class’ (A1: 98% (T = 49), A2: 40%,
p < 0.001 (T = 10)). (See Table 2.42)
Test B
In Test B, a correlation was found of the group ‘expert’ and the variable ‘measures’ (p = 0.022,
two-tailed). After pooling the answers of the aggression scale into two answers, a correlation was
found with the variable ‘intraspecific aggression’ (p = 0.012, two-tailed). 89% (N = 32, T = 36)
of the dogs evaluated by expert B1 were considered to show ‘neutral behavior’ towards other
dogs, 56% (N = 10, T = 18) by expert B2. 27% (N = 9, T = 33) of the dogs evaluated by expert
B1 did not receive any measures, whereas all the dogs evaluated by expert B2 (T = 16) received
measures. (See Table 2.42)
No correlation was found with the experts in Test C.
Table 2.42 Correlation of 6 criteria with experts, Tests A and B
Expert
Absolute number
Test A
p-value
Percentage of total
(N = 53)
A1
A2
Total
‘Hierarchy’
A1
A2
Total
Owner dominant over dog
40
3
43
75%
6%
81%
Dog dominant over owner
5
5
10
9%
9%
19%
Total
45
8
53
85%
15%
100%
0.004
two-tailed
Expert
Absolute number
Test A
‘Attachment’
A1
A2
Total
Attachment existent
37
3
40
71%
6%
77%
No attachment
7
5
12
13%
10%
23%
Total
44
8
52
85%
15%
100%
124
p-value
Percentage of total
(N = 52)
A1
A2
Total
0.011
two-tailed
Expert
Absolute number
Test A
p-value
Percentage of total
(N = 59)
A1
A2
Total
Measures
A1
A2
Total
No measures
46
4
50
78%
7%
85%
Measures
3
6
9
5%
10%
15%
Total
49
10
59
83%
17%
100%
< 0.001
two-tailed
Expert
Percentage of total
(N = 59)
Absolute number
Test A
Education class
A1
A2
Total
Not recommended
48
4
Recommended
1
Total
49
p-value
A1
A2
Total
52
81%
7%
88%
6
7
2%
10%
12%
10
59
83%
17%
100%
< 0.001
two-tailed
Expert
Absolute number
Test B
p-value
Percentage of total
(N = 54)
B1
B2
Total
Intraspecific aggression
B1
B2
Total
Neutral behavior
32
10
42
59%
19%
78%
Not neutral behavior
4
8
12
7%
15%
22%
Total
36
18
54
67%
33%
100%
0.012
two-tailed
Expert
Absolute number
Test B
B1
B2
Total
Not recommended
9
0
9
18%
0%
18%
Recommended
24
16
40
49%
33%
82%
Total
33
16
49
67%
33%
100%
Measures
p-value
Percentage of total
(N = 49)
B1
B2
Total
0.022
two-tailed
5.4.6 TEST SEQUENCE
Pooling the answers did not result in large enough groups to reliably run the Chi-square test or
the Fisher’s exact test.
5.4.7 STIMULUS DOG
Test A
No correlation was found with the weight or sex of the stimulus dogs (n=9) within Test A. Seven
of the dogs were confronted with two dogs at the same time (one male, one female). The three
groups ‘male’, ‘female’ ‘two dogs’ did not fulfill the conditions for a chi-square test to be run.
Considering the results of the dogs confronted with the two stimulus dogs as missing values,
leaving the two groups ‘male’ and ‘female’, did not show any correlation.
125
Test B
The dogs used in Test B are not known.
No Chi-square test was run for this test.
Test C
Test C only used male dogs (n=3) that were more or less the same weight.
No Chi-square test was run for this test.
5.4.8
‘HIERARCHY’
For a hierarchy to be able to exist, at least two individuals have to be in a relationship with each
other. Most relationships involve a certain degree of attachment (and/or dependency). An
existing, stable ‘hierarchy’ implies a certain degree of attachment (and/or dependence) in the
human – and/or the dog’s - social system. Based on this assumption, the alternate hypothesis was
set up that dogs being evaluated to be in a hierarchy where the owner is ‘dominant’, are more
likely to also be considered to have ‘an attachment’ to their owner.
In dog training, the aspect of ‘hierarchy’ is often taken to be a potential cause for behavioral
problems, such as disobedience or lack of control. This hypothesis leads us to the alternate
hypothesis that dogs that are ‘dominated by their owners’ are less likely to receive measures than
dogs that ‘dominate their owners’. The same reasoning can be made with the types of measures,
such as ‘obligatory leash’ and ‘education class’.
Hence, the one-tailed p-value was chosen in the Fisher’s Exact Test for the correlation
‘hierarchy’ and ‘attachment’ and the correlation ‘hierarchy’ and ‘measures’, ‘obligatory leash’
and ‘education class’.
Test A
For Test A, the alternate hypothesis was accepted in the following cases: ‘attachment’ (p < 0.001,
one-tailed), ‘measures’ (p = 0.002, one-tailed) and ‘education class’ (p < 0.001, one-tailed). After
pooling the answers, a correlation was also found in the case of ‘interspecific aggression’ (p =
0.008, two-tailed).
92% (N = 35, T = 38) of the dogs that were evaluated to be ‘dominated by their owners’ were
also considered to have ‘an attachment’ to their owners. 95% (N = 41, T = 43) of the dogs
evaluated to be in a hierarchy where the owner is dominant over the dog did not receive any
measures, opposed to 50% (N = 5, T = 10) of the dogs that were considered to be ‘dominant over
their owners’. All 42 teams where the owner was evaluated to be ‘dominant over the dog’ did not
receive the measure ‘education class’, 50% (N = 5, T = 10) of the dogs considered to be
‘dominant over their owners’ did receive this measure. 91 % (N = 38, N = 42) of the dogs
evaluated to be in a hierarchy where they are ‘dominated by their owners’ showed ‘neutral
behavior’ towards humans, whereas only 50% (N = 5, T = 10) of the dogs evaluated to be
‘dominant over their owners’ showed this same behavior. (See Table 2.43)
Test B
In Test B, the null hypothesis was accepted in all cases, i.e. there is no correlation between any
two variables.
126
Test C
For Test C, a correlation was found between the group ‘hierarchy’ with the variables ‘measures’
(p = 0.018, one-tailed) and ‘education class’ (p = 0.012, one-tailed). 90% (N = 45, T = 50) of the
teams where the owners were evaluated to be ‘dominant over their dog’ did not receive any
measures, whereas only 40% (N = 2, T = 5) of the dogs considered to be ‘dominant over their
owners’ did not receive any. 92% (N = 46, T =50) of the teams with the owner considered to be
‘dominant’ over his or her dog did not receive the measure ‘education class’, whereas 40% (N =
2, T = 5) of the dogs ‘dominant over their owners’ did not receive these measures. (See Table
2.43)
Behavioral Specialists
The null hypothesis was accepted in all the cases evaluated by the veterinary behaviorist and the
ethologist, i.e. no correlations were found within these approaches.
Table 2.43 Correlation of 6 criteria with ‘hierarchy’, Tests A and C
‘Hierarchy’
Absolute numbers
Test A
Interspecific
aggression
Percentage of Total
(N = 52)
owner
domina
nt
dog dominant
Total
43
73%
10%
83%
5
9
8%
10%
17%
10
52
81%
19%
100%
owner dominant
dog dominant
Total
Neutral behavior
38
5
Not neutral behavior
4
Total
42
pvalue
0.008
twotailed
‘Hierarchy’
Absolute numbers
Test A
Percentage of Total
owner
domina
nt
dog dominant
Total
37
76%
4%
80%
9
7%
13%
20%
83%
17%
100%
owner dominant
dog dominant
Total
Attachment existent
35
2
No attachment
3
6
‘Attachment’
Total
38
8
(N = 46)
46
pvalue
<
0.001
onetailed
‘Hierarchy’
Absolute numbers
Test A
Percentage of Total
owner
domina
nt
dog dominant
Total
46
77%
9%
87%
5
7
4%
9%
13%
10
53
81%
19%
100%
owner dominant
dog dominant
Total
No measures
41
5
Measures
2
Total
43
Measures
(N = 53)
127
pvalue
0.002
onetailed
‘Hierarchy’
Absolute numbers
Test A
Percentage of Total
owner
domina
nt
dog dominant
Total
48
81%
9%
91%
5
5
0%
9%
9%
10
53
81%
19%
100%
owner dominant
dog dominant
Total
Not recommended
43
5
Recommended
0
Total
43
Education class
(N = 53)
pvalue
<
0.001
onetailed
‘Hierarchy’
Absolute numbers
Test C
Percentage of Total
owner
domina
nt
dog dominant
Total
47
82%
4%
85%
3
8
9%
5%
15%
5
55
91%
9%
100%
owner dominant
dog dominant
Total
No measures
45
2
Measures
5
Total
50
Measures
(N = 55)
pvalue
0.018
onetailed
‘Hierarchy’
Absolute numbers
Test C
Percentage of Total
owner
domina
nt
dog dominant
Total
48
84%
4%
87%
3
7
7%
5%
13%
5
55
91%
9%
100%
owner dominant
dog dominant
Total
Not recommended
46
2
Recommended
4
Total
50
Education class
(N = 55)
128
pvalue
0.012
onetailed
5.4.9
‘ATTACHMENT’
Attachment to an attachment figure establishes confidence and a sense of security in the presence
of the attachment figure. As Dehasse (2002a: 37) mentions “thanks to this privileged contact, the
attachment figure / individual has the best position to become the teacher, the master. The student
is affiliated with, focuses its attention and concentration on this attachment figure…Without
attachment neither learning, nor impregnation, nor socialization, nor habituation nor social
communication is possible.” 15 * All the test situations but one in Test A were run in the presence
of the owner (hypothesized to be the attachment individual).
The alternate hypothesis was that dogs having an attachment to their owners are more likely to
show neutral behavior towards the stimuli they are confronted with in the tests than are dogs that
have no attachment to the owner. The variable ‘behavior towards moving objects’ was excluded
from this alternate hypothesis, as there is the aspect of ‘chasing’ involved there, which has a high
genetic influence and is difficult to handle even with an existing attachment. It is assumed that
animals having an attachment to their owners are more likely to stay close to them and are more
likely to obey. Because of the higher sense of security through the presence of the attachment
figure and the higher likelihood of the dog obeying an owner to whom it is attached, there is less
likelihood that the dog will show signs of fear or mistrust towards other animals or humans. The
alternate hypothesis was that dogs having an attachment to their owner are less likely to receive
measures than dogs without attachment. Hence, the one-tailed p-value (P1 > P2) of the Fisher’s
Exact Test was chosen.
Test A
For Test A, a correlation was found of the group ‘attachment’ with the variables ‘measures’ (p =
0.001, one-tailed) and ‘education class’ (p < 0.001, one-tailed). When pooling the answers for
aggression, a correlation was also found with the criterion ‘interspecific aggression towards
humans’ (p = 0.042, one-tailed). 87% (N = 34, T = 39) of the dogs evaluated to have ‘an
attachment’ to their owners showed ‘neutral behavior’ towards humans, 95% (N = 38, T = 40)
did not receive any measures and none of them received the measure ‘education class’. Whereas
50% (N = 6, T = 12) of the dogs considered to have ‘no attachment’ to their owners did not
receive any measures, 50% (N = 6, T = 12) did not receive the measure ‘education class’ and
58% (N = 7, T = 12) showed ‘neutral behavior’ towards humans. (See Table 2.44)
Test B
For Test B, a correlation was found in the Chi-square test of the group ‘attachment’ with the
variables ‘obligatory leash’ (p < 0.001, one-tailed). When pooling the answers, a correlation was
also found with the variable ‘intraspecific aggression’ (p = 0.012, one-tailed). 91% (N = 32, T =
35) of the dogs evaluated to ‘have an attachment’ to their owners did not receive the measure
‘obligatory leash’ and 82% (N = 27, T = 33) were considered to show ‘neutral behavior’ towards
other dogs. Whereas 14% (N = 1, T = 7) of the dogs evaluated to have ‘no attachment’ to their
owners, did not receive the measure ‘obligatory leash’ and 29% (N = 2, T = 7) showed ‘neutral
behavior’ towards other dogs. (See Table 2.44)
No correlation was found in Test C or the cases evaluated by the veterinary behaviorist and the
ethologist.
15
Translated by M. Bräm
129
Table 2.44 Correlation of 5 criteria with ‘attachment’, Tests A and B
‘Attachment’
Absolute numbers
Test
A
Percentage of Total
Existent
No
attachment
Total
Neutral behavior
34
7
Not neutral
behavior
5
Total
39
Interspecific
aggression
Existent
No
attachment
Total
41
67%
14%
80%
5
10
10%
10%
20%
12
51
76%
24%
100%
Percentage of Total
(N = 52)
p-value
0.042
one-tailed
‘Attachment’
Absolute numbers
Test
A
Existent
No
attachment
Total
No measures
38
6
Measures
2
Total
40
Measures
(N = 51)
existent
No
attachment
Total
44
73%
12%
85%
6
8
4%
12%
15%
12
52
77%
23%
100%
Percentage of Total
(N = 52)
p-value
0.001
one tailed
‘Attachment’
Absolute numbers
Test
A
Existent
No
attachment
Total
Not
recommended
40
6
Recommended
0
Total
40
Education class
existent
No
attachment
Total
46
77%
12%
88%
6
6
0%
12%
12%
12
52
77%
23%
100%
Percentage of Total
(N = 40)
p-value
< 0.001
one-tailed
‘Attachment’
Absolute numbers
Test
B
Intraspecific
aggression
Existent
No
attachment
Total
p-value
existent
no
attachment
Total
Neutral behavior
27
2
29
68%
5%
73%
Not neutral
behavior
6
5
11
15%
13%
28%
Total
33
7
40
83%
18%
100%
130
0.011
one-tailed
‘Attachment’
Absolute numbers
Test
B
Percentage of Total
Existent
no
attachment
Total
Not
recommended
32
1
Recommended
3
Total
35
Obligatory leash
(N = 42)
existent
no
attachment
Total
33
76%
2%
79%
6
9
7%
14%
21%
7
42
83%
17%
100%
p-value
< 0.001
one-tailed
5.4.10 MEASURES
The significant results of a correlation of recommended measures (such as muzzle, leash or
education class) with the variables ‘breed’, ‘weight’, ‘sex’, ‘stimulus dog’, ‘expert’, ‘age’,
‘hierarchy’ and ‘attachment’ are listed in the previous sections. The results of possible
correlations with the aggression categories are the focus of this section.
The alternate hypothesis was that dogs showing ‘neutral behavior’ are less likely to receive
measures than dogs showing ‘not neutral behavior’. Dogs that do not show aggression, fearful or
mistrustful behavior towards conspecifics, other animals and humans, or chase moving objects,
are not considered a problem for their owners or for public security. Hence, there is no need to
change anything by imposing measures. Additionally, the alternate hypothesis is made that dogs
that did not receive any measures are more likely to have passed the test.
Test A
In Test A, a correlation was found between the points ‘measures’ and ‘interspecific aggression
towards humans’ (p = 0.005, one-tailed), and between the group ‘education class’ and the
variable ‘interspecific aggression’ (p = 0.013, one-tailed) when pooling the answers. 90% (N =
44, T = 49) of the dogs that did not receive any measures showed ‘neutral behavior’ towards
humans, 88% (N = 45, T = 51) of the dogs that did not receive the measure ‘education class’
showed this behavior towards humans. Whereas, only 44% (N = 4, T = 9) of the dogs that did
receive measures were considered to show ‘neutral behavior’ towards humans and 43% (N = 3, T
= 7) of the dogs that did receive the measure ‘education class’ showed ‘neutral behavior’ towards
humans. (See Table 2.45)
Test B
In Test B, the zero hypothesis was rejected when correlating the group ‘measures’ with the
variables ‘test passed’(p = 0.0002, one-tailed) and, after pooling the answers for aggression, with
the variable ‘interspecific aggression’ (p = 0.0351, one-tailed). All the dogs (N = 9) that did not
receive measures passed the test and 89% (N = 8, T = 9) were evaluated to show ‘neutral
behavior towards humans’. Whereas 32% (N = 13, T = 41) of the dogs that did received measures
passed the test and 50% (N = 20, T = 40) showed ‘neutral behavior’ towards humans. A
correlation was also found between the group ‘obligatory leash’ and the variable ‘test passed’ (p
= 0.0002, one-tailed), and after pooling of the results of the aggression categories, with the
variables ‘intraspecific aggression’ (p = 2 x 10-5, one-tailed) and ‘behavior towards other
animals’ (p = 0.0471, one-tailed). 62% (N = 29, T = 47) of the dogs that did not receive the
measure ‘obligatory leash’ passed the test, 91% (N = 41, T = 45) showed ‘neutral behavior
towards other dogs’ and 85% (N = 34, T = 40) showed ‘neutral behavior’ towards other animals.
However, none of the dogs that received the measures ‘obligatory leash’ (N = 11) passed the test,
131
20% (N = 2, T = 20) were evaluated to show ‘neutral behavior’ towards other dogs and 50% (N =
4, T = 8) to show ‘neutral behavior’ towards other animals. (See Table 2.45)
Test C
After pooling the results of the aggression scale in Test C, a correlation was found of the group
‘measures’ with the variable ‘intraspecific aggression’ (p = 0.0376, one-tailed). 71% (N = 34, T =
48) of the dogs that did not receive any measures were evaluated to show ‘neutral behavior’
towards other animals. Whereas, 36% (N = 4, T = 11) of the dogs that did receive measures were
evaluated to show this behavior. (See Table 2.45)
Veterinary behaviorist
Within the group of dogs evaluated by the veterinary behaviorist, a correlation was found of the
measure ‘muzzle’ with the variable ‘interspecific aggression (dog alone)’ (p = 0.0455, one-tailed)
and between the measure ‘obligatory leash’ and the variable ‘interspecific aggression (dog alone)
(p = 0.046, two-tailed. 78% (N = 7, T = 9) of the dogs that did not receive the measure ‘muzzle’
and ‘obligatory leash’ were evaluated to show ‘neutral behavior’ towards humans when separated
from their owners. None of the dogs that did received one of these measures was evaluated to
show this behavior. (See Table 2.45)
Ethologist
No correlation was found for the within the group of dogs evaluated by the ethologist.
Table 2.45 Correlation of 4 criteria with measures recommended by Tests A, B and C and by the
veterinary behaviorist
Measures
Absolute numbers
Test Interspecific
A aggression
Percentage of Total
(N = 58)
No
measures
Measures
Total
48
76%
7%
83%
5
10
9%
9%
17%
9
58
84%
16%
100%
No measures
Measures
Total
Neutral
behavior
44
4
Not neutral
behavior
5
Total
49
132
p-value
0.005
onetailed
Education class
Absolute numbers
Percentage of Total
(N = 58)
Not
recommende
d
Recommen
ded
Total
Neutral
behavior
45
3
48
78%
5%
83%
Not neutral
behavior
6
4
10
10%
7%
17%
Total
51
7
58
88%
12%
100%
Test Interspecific
A aggression
p-value
0.013
onetailed
Not
Recommende
Total
recommend
d
ed
Measures
Absolute numbers
Test Interspecific
B aggression
Percentage of Total
(N = 49)
No
measures
Measures
Total
28
16%
41%
57%
20
21
2%
41%
43%
40
49
18%
82%
100%
No measures
Measures
Total
Neutral
behavior
8
20
Not neutral
behavior
1
Total
9
p-value
0.035
onetailed
Measures
Absolute numbers
Test
Test passed
B
Percentage of Total
(N = 50)
No
measures
Measures
Total
22
18%
26%
44%
28
28
0%
56%
56%
41
50
18%
82%
100%
No measures
Measures
Total
Test passed
9
13
Test failed
0
Total
9
133
p-value
< 0.001
twotailed
Obligatory leash
Absolute numbers
Percentage of Total
(N = 55)
Not
recommende
d
Recommen
ded
Total
Neutral
behavior
41
2
43
75%
4%
78%
Not neutral
behavior
4
8
12
7%
15%
22%
Total
45
10
55
82%
18%
100%
Test Intraspecific
B aggression
p-value
Not
Recommende
Total
recommend
d
ed
< 0.001
one-tailed
Obligatory leash
Absolute numbers
Percentage of Total
(N = 48)
Not
Recommende
Total
recommend
d
ed
Test Beh. towards
B other animals
Not
recommende
d
Recommen
ded
Total
Neutral
behavior
34
4
38
71%
8%
79%
Not neutral
behavior
6
4
10
13%
8%
21%
Total
40
8
48
83%
17%
100%
p-value
0.047
one-tailed
Obligatory leash
Absolute numbers
Percentage of Total
(N = 58)
Not
recommende
d
Recommen
ded
Total
Test passed
29
0
29
50%
0%
50%
Test failed
18
11
29
31%
19%
50%
Total
47
11
58
81%
19%
100%
Test
Test passed
B
134
p-value
< 0.001
two-tailed
Not
Recommende
Total
recommend
d
ed
Education class
Absolute numbers
Percentage of Total
(N = 55)
Not
recommende
d
Recommen
ded
Total
Neutral behavior
17
26
43
31%
47%
78%
Not neutral
behavior
1
11
12
2%
20%
22%
Total
18
37
55
33%
67%
100%
Test Intraspecific
B aggression
Not
Recommende
Total
recommend
d
ed
0.039
onetailed
Education class
Absolute numbers
Percentage of Total
(N = 58)
Not
recommende
d
Recommen
ded
Total
Test passed
16
13
29
28%
22%
50%
Test failed
4
25
29
7%
43%
50%
Total
20
38
58
34%
66%
100%
Test
Test passed
B
Not
Recommende
Total
recommend
d
ed
0.002
twotailed
Measures
Absolute numbers
Test Intraspecific
C aggression
Percentage of Total
(N = 59)
No measures
Measures
Total
No
measures
Measures
Total
Neutral behavior
34
4
38
58%
7%
64%
Not neutral
behavior
14
7
21
24%
12%
36%
Total
48
11
59
81%
19%
100%
135
0.038
onetailed
Muzzle
Absolute numbers
VB
Percentage of Total
Not
recommende
d
Recommen
ded
Total
Aggressive
behavior
8
3
11
Total
8
3
11
Beh. towards
other animals
p-value
(N = 11)
Not
Recommende
Total
recommend
d
ed
0.045
onetailed
73%
27%
100%
73%
27%
100%
Obligatory leash
Absolute numbers
VB
Percentage of Total
(N = 12)
Not
recommende
d
Recommen
ded
Total
Not aggressive
behavior
7
0
7
58%
0%
58%
Aggressive
behavior
2
3
5
17%
25%
42%
Total
9
3
12
75%
25%
100%
Interspecific
aggression
(dog alone)
136
p-value
0.046
twotailed
Not
Recommende
Total
recommend
d
ed
5.4.11 CORRELATION OF TYPES OF AGGRESSION
All the following results were found after pooling the data into larger categories. The null
hypothesis is that there is no correlation of the two points studied. The alternate hypothesis is that
there is a correlation between the two points looked at, but no assumption was made in advance
in which direction this correlation would go. Hence, the two-tailed p-value was chosen and a
value below 0.05 was considered significant.
An overview of the correlation of the types of aggression found within each method is given in
Figure 2.31.
For Test A, a correlation was found between ‘intraspecific aggression’ and ‘interspecific
aggression towards humans’ (p = 0.009, two-tailed), ‘behavior towards moving objects’ (p =
0.045, two-tailed) and ‘interspecific aggression towards humans (dog alone)’ (p = 0.048, twotailed); between ‘interspecific aggression towards humans’ and ‘behavior towards moving
objects’ (p = 0.039, two-tailed) and ‘interspecific aggression towards humans (dog alone)’ (p =
0.006, two-tailed); and between ‘behavior towards moving objects’ and ‘interspecific aggression
towards humans (dog alone)’ (p < 0.001, two-tailed).
91% (N = 40, T = 44) of the dogs evaluated to show ‘neutral behavior’ towards other dogs also
were considered to show this behavior towards humans, 89% (N = 40, T = 45) towards moving
objects, and 93% (N = 42, T = 45) towards humans when separated from their owners. Of the
dogs being considered to show ‘neutral behavior’ towards humans, 90% (N = 43, T = 48) were
evaluated to show this same behavior towards moving objects and 96% (N = 46, T = 48) towards
humans when separated from their owners. 92% (N = 48, T = 52) of the dogs showing ‘neutral
behavior’ towards humans when separated from their owners also showed this behavior towards
moving objects. (See Table 2.46)
For Test B, a correlation was found between ‘intraspecific aggression towards other dogs’ and
‘interspecific aggression towards humans (p = 0.022, two-tailed), between ‘aggressive behavior
towards other animals’ and ‘intraspecific aggression towards other dogs’ (p = 0.007, two-tailed),
‘interspecific aggression towards humans’ (p = 0.003, two-tailed) and ‘behavior towards moving
objects’ (p = 0.009, two-tailed).
Of the dogs evaluated to show ‘neutral behavior’ towards other dogs, 71% (N = 30, T = 42) were
considered to show the same behavior towards humans and 92% (N = 33, T = 36) towards other
animals. 93% (N = 28, T = 30) of the dogs showing ‘neutral behavior’ towards humans were
evaluated to show the same behavior towards other animals. 94% (N = 33, T = 35) of the dogs
showing ‘neutral behavior’ towards moving objects also show this behavior towards other
animals. (See Table 2.47)
No correlation was found for Test C, the veterinary behaviorist or the ethologist.
137
Figure 2.31 Correlations of types of aggression
Intraspecific
aggression
Interspecific
aggression towards
Test A
Behavior
towards
moving
Interspecific aggression
towards humans (dog
alone)
Interspecific
aggression towards
humans
Intraspecific
aggression
Test B
Behavior towards other
animals
Behavior
towards
moving
Test C
Intraspecific
aggression
Ethologist
Intraspecific
aggression
No
correlatio
Behavior
towards
moving
No
correlatio
Veterinarian
No
correlatio
Interspecific aggression
towards humans
Interspecific aggression
towards humans
Behavior towards other
animals
Interspecific aggression
towards humans (dog
alone)
138
Table 2.46 Correlation of types of aggression, Test A
Absolute numbers
Test A
Interspecific
aggression
Neutral behavior
Not neutral behavior
Total
Neutral
behavior
40
4
44
Not neutral
behavior
Intraspecific aggression
Percentage of Total
58)
(N =
Total
Total
8
6
14
48
10
58
p-value
0.009 twotailed
Neutral
behavior
69%
7%
76%
Not neutral
behavior
14%
10%
24%
83%
17%
100%
Interspecific aggression
Absolute numbers
Test A
Moving objects
Neutral
behavior
Not neutral
behavior
Neutral behavior
43
6
Not neutral behavior
5
Total
48
Percentage of Total
(N = 58)
Neutral
behavior
Not neutral
behavior
Total
49
74%
10%
84%
4
9
9%
7%
16%
10
58
83%
17%
100%
Total p-value
0.039
Interspecific aggression
Absolute numbers
Test A
Neutral
behavior
Not neutral
behavior
Neutral behavior
46
6
Not neutral behavior
2
Total
48
Interspecific aggr.
(dog alone)
Percentage of Total
(N = 58)
Neutral
behavior
Not neutral
behavior
Total
52
79%
10%
90%
4
6
3%
7%
10%
10
58
83%
17%
100%
Total p-value
0.006
Interspecific aggression (dog alone)
Absolute numbers
Test A
Moving objects
Neutral
behavior
Not neutral
behavior
Neutral behavior
48
1
Not neutral behavior
4
Total
52
Percentage of Total
(N = 59)
Neutral
behavior
Not neutral
behavior
Total
49
81%
2%
83%
6
10
7%
10%
17%
7
59
88%
12%
100%
139
Total p-value
< 0.001
Table 2.47 Correlation of types of aggression, Test B
Absolute numbers
Test B
Neutral
behavior
Not neutral
behavior
Neutral behavior
30
4
Not neutral behavior
12
Total
42
Interspecific
aggression
Intraspecific aggression
Percentage of Total
Neutral
behavior
Not neutral
behavior
Total
34
56%
7%
63%
8
20
22%
15%
37%
12
54
78%
22%
100%
Total p-value
0.022
Intraspecific aggression
Absolute numbers
Percentage of Total
Test B
Neutral
behavior
Not neutral
behavior
Neutral behavior
33
5
Not neutral behavior
3
Total
36
Other animals
Not neutral
behavior
Total
38
72%
11%
83%
5
8
7%
11%
17%
10
46
78%
22%
100%
Neutral
behavior
Not neutral
behavior
Neutral behavior
28
10
Not neutral behavior
2
Total
30
Other animals
0.007
Interspecific aggression
Percentage of Total
(N = 48)
Not neutral
behavior
Total
38
58%
21%
79%
8
10
4%
17%
21%
18
48
63%
38%
100%
Neutral
behavior
Not neutral
behavior
Neutral behavior
33
0
Not neutral behavior
2
Total
35
Other animals
Total p-value
Neutral
behavior
Absolute numbers
Test B
(N = 46)
Neutral
behavior
Absolute numbers
Test B
(N = 54)
Total p-value
0.003
Moving objects
Percentage of Total
(N = 37)
Neutral
behavior
Not neutral
behavior
Total
33
89%
0%
89%
2
4
5%
5%
11%
2
37
95%
5%
100%
140
Total p-value
0.009
5.5
RESULTS OF THE TESTS CONCERNING THE CORRELATION
OF AGGRESSIVE BEHAVIOR AND DANGEROUSNESS
A Chi-square test or a Fisher’s Exact Test was run in order to find out if the experts had a
consistent pattern in evaluating the aggressive behavior and the dangerousness of the dogs, i.e. if
a certain behavior generally leads to the same level of dangerousness. It would be expected that
there would be some kind of correlation of the aggression scale and the dangerousness scale for
every test. The alternate hypothesis that there is a correlation between the dog’s behavior and the
potential dangerousness it incorporates for its surroundings. The two-tailed p-value was chosen,
with a value of p < 0.05 being significant. The percentages refer to 100% being the total of dogs
evaluated by the specific method in the targeted criterion.
However, the problem is again that the population is too homogenous and only few dogs actually
were evaluated to show aggressive behavior or to be ‘dangerous’. As the majority of the dogs
was considered to show ‘open, friendly, neutral behavior’ as well as to ‘not be dangerous and in
control of their owners’ in all three tests, the results were again pooled into larger groups to allow
a Fisher’s Exact Test to be run.
Original scales:
Aggression:
1 = open, friendly, neutral behavior
2 = dominant behavior or mistrustful, fearful behavior without aggression
3 = threatening behavior
4 = overt attack / biting with threatening
5 = overt attack / biting without threatening
Dangerousness:
1 = dog is no danger, dog is under the control of its owner
2 = dog is no danger, dog is not under the control of its owner
3 = dog is a potential danger, dog is under control of the owner
4 = dog is a potential danger, dog is not under control of the owner
Pooled scale:
Tests: A pooling to the two groups ‘aggressive behavior’ and ‘not aggressive behavior’ did not
make much sense, as almost all the dogs would have been in the category ‘not aggressive
behavior’. Consequently, we chose to pool the answers into the two categories of ‘neutral’ and
‘not neutral’ behavior.
Aggression:
open, friendly, neutral behavior, which will be referred to as
‘neutral behavior’ (answer 1 of the original scale)
not open, friendly, neutral, behavior, which will be referred to as
‘not neutral behavior (answers 2 to 5 of the original scale)
Dangerousness:
dog ‘not dangerous’ (answers 1 and 2 of the original scale)
dog ‘dangerous (answers 3 and 4 of the original scale)
141
Veterinary behaviorist and ethologist:
Aggression
aggressive behavior (answers 1 and 2 of the original scale)
Not aggressive behavior (answers 3 to 5 of the original scale)
Dangerousness:
dog ‘not dangerous’ (answers 1 and 2 of the original scale)
dog ‘dangerous (answers 3 and 4 of the original scale)
5.5.1
INTRASPECIFIC AGGRESSION TOWARDS OTHER DOGS AND
DANGEROUSNESS FOR OTHER DOGS
Tests B and C show a correlation of ‘intraspecific aggression’ and ‘dangerousness for other dogs’
(Test B: p < 0.001, Test C: 0.013, two-tailed). All the dogs (N = 43) evaluated to show ‘neutral
behavior’ and 42% (N = 5) of the dogs showing ‘not neutral behavior’ towards other dogs in Test
B were considered to be ‘not dangerous’. In Test C, all the dogs (N = 38) evaluated to show
‘neutral behavior’ and 81% (N = 17) of the dogs showing ‘not neutral behavior’ towards other
dogs were considered to be ‘not dangerous’. (See Table 2.48)
Table 2.48 Correlation of intraspecific aggression and dangerousness for dogs
Tests B and C
Intraspecifi c aggression
Absolute number of dogs
neutral
not neutral
Total
neutral
not neutral
Total
not dangerous
dangerous
43
0
5
7
48
7
78%
0%
9%
13%
87%
13%
Total
43
12
55
78%
22%
100%
Dangerousness for dogs
Test B
Dangerousness for dogs
Test C
Percentage of total (N = 55)
not dangerous
dangerous
Total
Absolute number of dogs
neutral
not neutral
Total
38
17
55
0
4
4
38
32
5.5.2
59
Percentage of total (N = 59)
neutral
not neutral
Total
64%
29%
93%
0%
7%
7%
64%
36%
100%
p-value
< 0.001
two-tailed
p-value
0.013
two-tailed
INTERSPECIFIC AGGRESSION TOWARDS HUMANS AND
DANGEROUSNESS FOR HUMANS
A correlation of the points ‘interspecific aggression towards humans’ and ‘potential
dangerousness for humans’ was found in all three tests (Test A: p = 0.004, Test B: p = 0.02, Test
C: p = 0.001, two-tailed). All of the dogs (N = 48) evaluated to show ‘neutral behavior’ and 70%
(N = 7) of the dogs showing ‘not neutral behavior’ towards humans in Test A were considered
‘not dangerous’. In Test B, all the dogs evaluated to show ‘neutral behavior’ (N = 34) and 82%
(N = 18) of the dogs showing ‘not neutral behavior’ towards humans were considered to be ‘not
dangerous’ for humans. In Test C, 100% (N = 52) of the dogs evaluated to show ‘neutral
behavior’ and 57% (N = 4) of the dogs showing ‘not neutral behavior’ were considered to be ‘not
dangerous’. (See Table 2.49)
142
Table 2.49 Correlation of interspecific aggression and dangerousness for humans,
Tests A, B and C
Dangerousness for humans
Test A
not dangerous
dangerous
48
0
7
3
55
3
83%
0%
12%
5%
95%
5%
Total
48
10
58
83%
17%
100%
Dangerousness for humans
Test B
not dangerous
dangerous
Total
Dangerousness for humans
Test C
Interspecific aggression
Absolute number of dogs
Percentage of total (N = 58)
Not
Neutral
Total
Neutral
Not neutral
Total
neutral
not dangerous
dangerous
Total
Interspecific aggression
Absolute number of dogs
Percentage of total (N = 56)
Not
Neutral
Total
Neutral
Not neutral
Total
neutral
34
18
52
61%
32%
93%
0
4
4
0%
7%
7%
34
22
56
61%
39%
100%
Interspecific aggression
Absolute number of dogs
Percentage of total (N = 59)
Not
Neutral
Total
Neutral
Not neutral
Total
neutral
52
4
56
88%
7%
95%
0
3
3
0%
5%
5%
52
7
5.5.3
59
88%
12%
100%
p-value
0.004
two-tailed
p-value
0.02
two-tailed
p-value
0.001
two-tailed
INTERSPECIFIC AGGRESSION (DOG ALONE) TOWARDS HUMANS AND
DANGEROUSNESS OF THE DOG ALONE FOR HUMANS
A correlation of these two points was found in the only test that evaluated these criteria (Test A,
p = 0.012, two-tailed) and within the group of dogs evaluated by the veterinary behaviorist (p =
0.045, two-tailed). All the dogs evaluated to show ‘neutral behavior’ (N = 52) and 71% (N = 5)
of the dogs considered to react ‘not neutrally’ towards humans in the absence of their owners
were considered to be ‘not dangerous’ for humans in Test A. The veterinary behaviorist found all
the dogs he evaluated to show ‘not aggressive behavior’ (N = 7) and 40% (N = 2) of the dogs he
considered to show ‘aggressive behavior’ to be ‘not dangerous’ for humans in the absence of
their owners. (See Table 2.50)
143
Table 2.50 Correlation of interspecific aggression (dog alone) and dangerousness for humans
(dog alone), Test A and veterinarian
Interspecific aggression / dog alone
Absolute number of dogs
Percentage of total
Dangerousness humans
(dog alone)
Neutral
Not neutral
Test A not dangerous
dangerous
52
0
5
2
Total
52
7
VB
not dangerous
dangerous
(N = 59)
Neutral
Not neutral
Total
57
2
88%
0%
9%
3%
97%
3%
59
88%
12%
100%
Interspecific aggression / dog alone
Absolute number of dogs
Percentage of total
(N = 12)
Not
Not aggressive Aggressive
Total
Aggressive
Total
aggressive
7
2
9
58%
17%
75%
0
3
3
0%
25%
25%
Total
7
5
12
58%
42%
100%
5.5.4
BEHAVIOR TOWARDS MOVING OBJECTS AND DANGEROUSNESS FOR
MOVING OBJECTS
A correlation was found within the group of dogs evaluated by the veterinary behaviorist (p =
0.015, two-tailed). 80% (N = 4) of the dogs evaluated by the veterinary behaviorist to show ‘not
aggressive behavior’, and none of the 6 dogs considered to show ‘aggressive behavior’ towards
moving objects were found to be ‘not dangerous’ for moving objects. (See Table 2.51)
Table 2.51 Correlation of behavior towards moving objects and dangerousness for moving
objects, veterinarian
Dangerousness for
moving objects
VB
Behavior towards moving objects
Absolute number of dogs
Percentage of total
(N = 11)
not
not
aggressive
Total
aggressive
Total
aggressive
aggressive
not dangerous
dangerous
4
1
0
6
4
7
36%
9%
0%
55%
36%
64%
Total
5
6
11
45%
55%
100%
144
p-value
0.015 twotailed
p-value
0.012
twotailed
p-value
0.045
twotailed
5.5.5
BEHAVIOR TOWARDS OTHER ANIMALS AND DANGEROUSNESS FOR
OTHER ANIMALS.
A correlation of the points ‘behavior towards other animals’ and ‘dangerousness for other
animals’ was found within Test B (p = 0.0399, two-tailed). All the dogs evaluated to show
‘neutral behavior’ (N = 38) and 80% (N = 8) of the dogs considered to show ‘not neutral
behavior’ towards other animals were found to be ‘not dangerous’ for other animals in this test.
(See Table 2.52)
Table 2.52 Correlation of behavior towards other animals and dangerousness for other animals,
Test B
Test B
Dangerousness for
other animals
not dangerous
dangerous
Total
Behavior towards other animals
Percentage of total
(N =
Absolute number of dogs
48)
Not
Not
Neutral
Total Neutral
Total
neutral
neutral
38
8
46
79%
17%
96%
0
2
2
0%
4%
4%
38
10
48
145
79%
21%
100%
p-value
0.04
twotailed
5.6
COMPARISON OF THE APPROACHES USING A REFERENCE
VALUE
The aim of this analysis is to compare the different approaches, the tests on the one hand and the
behavioral experts on the other. The two aspects of interest are the sensitivity and the specificity
of the tests. Sensitivity is the ability of the test to identify correctly those individuals showing the
behavior tested for (e.g. aggression). Specificity is the ability of the test to identify correctly those
individuals not showing the behavior tested for (e.g. not aggressive) (Planta, 2001). What is of
interest for public security is the degree of sensitivity of the test, i.e. one wants as few false
negatives as possible. What is of interest for animal protection is the specificity, i.e. the false
positives, as these dogs might receive measures or even be put down as a consequence of the test
result, without the dog actually being a particular danger to its environment. Planta (2001) relied
on the information the owners gave about the history of biting incidences in a questionnaire and
on the observations of behavioral counselors. In this study, the veterinary behaviorist and the
ethologist obtained the major part of the information of the dog’s behavior through interviews
with the owners; so, these two approaches were taken as reference values for the tests. The aspect
of public security and the consequent interest in the dogs falsely assessed to show ‘not
aggressive’ behavior is another reason to choose the behavioral specialists as reference.
Landrivon (2002) says to use a pragmatic approach, if no standard reference exists. This
pragmatic approach would be the public security, i.e. to choose the approach that evaluated the
highest number of dogs to be aggressive. These two experts generally evaluated the dogs to be
more ‘aggressive’ than did the tests.
In several situations, the specificity and sensitivity could not be assessed, as one of the categories,
most often the ‘aggressive’ category of the tests or the ‘not aggressive’ category of the
specialists, did not include any individuals.
The tests were run in the program Win Episcope 2.0, calculated with a 95% level of confidence.
5.6.1 INTRASPECIFIC AGGRESSION
Testing for sensitivity of Test A, B and C in reference to the ethologist, a percentage of 29% was
found. Specificity in this case was 100%. (See Table 2.53)
Sensitivity for Test C in reference to the veterinary behaviorist was found to be 14%. Specificity
was 100%. In reference to the ethologist sensitivity was higher, 29% and specificity lower with
60% and a range of 17 to 100%. (See Table 2.53)
146
Table 2.53 Sensitivity and specificity of intraspecific aggression, Tests A, B and C
Intraspecific aggression
ET
Test A
aggressiv
not aggressive
Total
aggressive
not aggressive
Total
2
0
10
5
5
2
5
7
12
%
lower limit
upper limit
Sensitivity
29
0
62
Specificity
100
100
100
%
lower limit
upper limit
Intraspecific aggression
ET
aggressive
not aggressive
Total
aggressive
2
0
2
Sensitivity
29
0
62
not aggressive
5
4
9
Specificity
100
100
100
Total
7
4
11
%
Lower limit
Upper limit
Test B
Intraspecific aggression
ET
Aggressive
Not aggressive
Total
Aggressive
2
2
4
Sensitivity
29
0
62
Not aggressive
5
3
8
Specificity
60
17.059
100
Total
7
5
12
%
Lower limit
Upper limit
Test C
Intraspecific aggression
VB
Aggressive
Not aggressive
Total
Aggressive
1
0
1
Sensitivity
14
0
40
Not aggressive
6
5
11
Specificity
100
100
100
Total
7
5
12
Test C
5.6.2 INTERSPECIFIC AGGRESSION
A sensitivity of 40% and a specificity of 100% was found for Test A in relation to the veterinary
behaviorist. Test B was shown to have a sensitivity of 17% and a specificity of 100% in reference
to the veterinary behaviorist, and a sensitivity of 50% and a specificity of 100% in reference to
the ethologist. Testing Test C referring to the ethologist showed a sensitivity of 50% and a
specificity of 100%. (See Table 2.54)
Table 2.54 Sensitivity and specificity of Tests A, B and C, interspecific aggression
Interspecific aggression
147
VB
Aggressive
Not aggressive
Total
Aggressive
2
0
2
Not aggressive
3
6
9
Total
5
6
11
Test A
%
Lower limit
Upper limit
Sensitivity
40
0
83
Specificity
100
100
100
%
Lower limit
Upper limit
Interspecific aggression
VB
Aggressive
Not aggressive
Total
Aggressive
1
0
1
Sensitivity
17
0
46
Not aggressive
5
5
10
Specificity
100
100
100
Total
6
5
11
%
Lower limit
Upper limit
Test B
Interspecific aggression
ET
Aggressive
Not aggressive
Total
Aggressive
1
0
1
Sensitivity
50
0
100
Not aggressive
1
10
11
Specificity
100
100
100
Total
2
10
12
%
Lower limit
Upper limit
Test C
Interspecific aggression
ET
Aggressive
Not aggressive
Total
Aggressive
1
0
1
Sensitivity
50
0
100
Not aggressive
1
9
10
Specificity
100
100
100
Total
2
9
11
Test B
5.6.3 INTERSPECIFIC AGGRESSION (DOG ALONE)
The comparison of Test A with the veterinary behaviorist as a reference showed a sensitivity of
20% and a specificity of 100%. (See Table 2.55)
Table 2.55 Sensitivity and specificity of Test A, interspecific aggression (dog alone)
Interspecific aggression (dog alone)
VB
Test A
Aggressive
Not aggressive
Total
148
%
Lower limit
Upper limit
Aggressive
1
0
1
Sensitivity
20
0
55.062
Not aggressive
4
7
11
Specificity
100
100
100
Total
5
7
12
5.6.4 BEHAVIOR TOWARDS MOVING OBJECTS
The test could not be run in any of the cases, as all the dogs were evaluated to show ‘not
aggressive’ behavior towards moving objects in all the three tests.
5.6.5 BEHAVIOR TOWARDS OTHER ANIMALS
The test could not be run in any of the cases, as all the dogs were evaluated to show ‘not
aggressive’ behavior towards other animals in both Tests A and B.
5.6.6 ‘HIERARCHY’
The dog being ‘dominant’ over its owner was considered to be the positive answer, whereas the
owner being ‘dominant’ over his or her dog was the negative answer, i.e. the situation strived for.
Sensitivity and specificity were only able to be defined for Tests A and B in reference to the
ethologist. Test A showed a sensitivity of 50% and a specificity of 100%, whereas Test B showed
both a sensitivity and a specificity of 100%. (See Table 2.56)
Table 2.56 Sensitivity and specificity of Tests A and B, ‘hierarchy’
‘Hierarchy’
ET
Dog dominant
Owner dominant
Total
Dog dominant
2
0
2
Owner dominant
2
6
8
Total
4
6
10
Dog dominant
Owner dominant
Total
Dog dominant
1
0
1
Owner dominant
0
3
3
Total
1
3
4
Test A
%
Lower limit
Upper limit
Sensitivity
50
1
99
Specificity
100
100
100
%
Lower limit
Upper limit
Sensitivity
100
100
100
Specificity
100
100
100
ET
Test B
149
5.6.7 ‘ATTACHMENT’
The test for sensitivity and specificity could not be run, as both the veterinary behaviorist and the
ethologist considered all the dogs evaluated by them in this point to have an ‘attachment’ to their
owners.
5.6.8 MEASURES
Sensitivity of Test A in reference to the veterinary behaviorist was 33.3%, specificity 83.3%. Test
C showed a sensitivity of 66.7% and a specificity of 16.7% in relation to the veterinary
behaviorist. For Test B, a sensitivity of 100% and a specificity of 50% were found in reference to
the ethologist. (See Table 2.57)
Table 2.57 Sensitivity and specificity of Tests A, B and C, measures recommended
VB
Measures
No measures
Total
Measures
2
1
3
No measures
4
5
9
Total
6
6
12
Measures
No measures
Total
Measures
1
2
3
No measures
5
4
9
Total
6
6
12
Measures
No measures
Total
Measures
4
3
7
No measures
0
3
3
Total
4
6
10
Test A
%
Lower limit
Upper limit
Sensitivity
33.3
0
71.054
Specificity
83.3
53.513
100
%
Lower limit
Upper limit
Sensitivity
66.667
28.946
100
Specificity
16.667
0
46.487
%
Lower limit
Upper limit
Sensitivity
100
100
100
Specificity
50
9.992
90.008
VB
Test C
ET
Test B
150
6. DISCUSSION
6.1
DISCUSSION OF METHODS
The discussion of the methods here is guided by the viewpoints expressed by Dobbert in the
following statements:
“The relativity of validity and reliability is often problematic to persons accustomed to
using quantitative research methods where both appear to be easily measurable, but actually
absolute validity and reliability cannot be obtained in any science. A researcher must
always interfere with the subject in order to study it and can never control all the
circumstances surrounding any study. This is true for both physical and social sciences and
for both experimental and observational methodologies (Dobbert, 1982, p 259).
“The object of all research is to produce information that is usable for some defined
purpose. Valid and reliable data, then, are data that are accurate for a given purpose. Since
all research interferes with and shapes data, the object of trying to achieve validity and
reliability is to design the methods so that the necessary interference is well understood and
is limited to effects that are of smaller significance for the end at hand“ (Dobbert, 1982, p.
260).
6.1.1 SCIENCE AND RESEARCH
‘Science’ is generally considered to consist of two main categories:
• “exact” sciences such as mathematics, chemistry and physics;
• The ‘sciences of the imprecise’ (Moles, 1995) such as psychology, biology, ethology,
and also behavioral medicine.
As is so often the case in research, here again there seems to be no universally accepted definition
of ‘Science’ in the literature. Following are two definitions of ‘Science’:
“Knowledge of facts, phenomena, laws, and proximate causes, gained and verified by exact
observation, organized experiment, and ordered thinking”. (Michaelis cited in Barrows
2001: 615)
“Any department of knowledge in which the results of investigation have been logically
arranged and systematized in the form of hypotheses and general laws subject to
verification. (Michaelis cited in Barrows 2001: 615)
The most important characteristics that make a study ‘scientific’ seem to be the following:
• Knowledge
• Observation, experiment
• Logic
• Hypotheses and verification of these
• Verification of hypotheses
151
Moles (1995) describes three possible approaches to research depending on the degree of
interaction between the observer and the observed,
• Strong interaction: experimentation;
• Slight interaction: observation;
• No interaction: modelization.
On the basis of Moles’ model, one can define Science is the knowledge gained through research.
Moles’ three categories of research cannot, however, be strictly separated. A research project can
involve one, two or all three of these approaches. The study reported upon here includes aspects
of all three categories; The theoretical Part I is an example of ‘modelization’, Part II, of
experimentation (the tests) as well as ‘observation’ (ethologist and veterinary behaviorist).
6.1.2 VALIDITY AND RELIABILITY
The two most important considerations in test evaluations are ‘validity’ and ‘reliability’.
Validity
Only one scientifically validated test to evaluate canine aggressive behavior exists in the
literature. This is the “test for aggression in the domestic dog” developed by Netto and Planta
(1999). This lack of validated tests indicates that such validations of a test are not as easy as they
might seem.
Validity is not a simple, unitary concept; many aspects must be considered when using this term.
The American Psychological Association makes the following distinctions:
“The concept [of validity] refers to the appropriateness, meaningfulness, and usefulness of
the specific inferences made from test scores. Test validation is the process of accumulating
evidence to support such inferences” (1985: 9). More simply expressed, measures of
validity as the question: “is the test measuring what it claims to be measuring?” It is
important to note that it is not the test itself that is validated, but that “validity always refers
to the degree to which that evidence supports the inferences that are made from the scores.”
(APA, 1985: 9).
There are three traditional types of evidence: context, criterion and construct evidence of validity
(APA 1985) all of which are described in Table 2.58.
Table 2.58 Evidence for validity
Content
Demonstrates the degree to
which the sample of items,
tasks or questions on a test
are representative of some
defined universe or domain
Definition of content.
Criterion
Demonstrates that the test
scores are systematically
related to one or more
outcome criteria. In this
context, the criterion is the
variable of primary interest.
How accurately can
criterion performance be
predicted from scores on the
test?
152
Construct
Focuses on the test score as
a measure of the
psychological characteristic
of interest
Content
1. Adequate specification
of the universe of content
that the test is intended to
represent (by expert
judgment)
2. If using an available test
for other than the original
purpose, the
appropriateness of the
Procedure
original domain
definition must be judged
for the new use.
3. Determination of the
degree to which format
and response properties
of the sample are
representative of the
universe.
Criterion
Construct
Predictive study:
1. Test development
obtains information about
2. Continuation of the test
the accuracy with which
until a pattern of
early test data can be used
empirical relationships
to estimate criterion scores
between test scores and
that will be obtained in the
other variables clearly
future
indicates the meaning of
Concurrent study:
the test scores
has the same purpose, but it 3. Define meaning of the
obtains prediction and
test scores
criterion information
simultaneously.
Sensitivity
Specificity
Evidence
Intercorrelation among
items -> does test measure a
single construct?
Analysis of individual
responses
These three categories cannot, however, be clearly separeated in practice. In fact, an “ideal
validation includes several types of evidence, which span all three of the traditional categories”
(APA, 1985, p 9). The decision on how to look at the judgement of validity depends on the
particular validation problem and upon which category it is primarily based.
Other authors differentiate among different types of validation. Hirsig (1993), for example, first
distinguishes between ‘external’ and ‘internal validity’. External validity signifies that “the
construct is also valid in a context surpassing the concrete survey situation” (Hirsig 1993: 6.22).
Internal validity, on the other hand, refers to the fact that “the meaning of the construct is limited
to the survey situation” (Hirsig 1993: 6.22), or, as Winter (2000) puts it “Internal validity relates
to whether the findings or results of the research relate to and are caused by the phenomena under
investigation and not other unaccounted for influences“.
Hirsig also categorizes types of validity according to what the test interpretation is based on:
• The test interpretation is based on a representative conclusion, including face validity
and content validity.
- Face validity: “A survey procedure is face valid if it is self-evident”* (Hewson,
Luescher & Ball (1999: 52) say that “face validity implies that the questions used
in the rating scales make sense to the person using them”.)
- Content validity: A test is content valid if, in the eyes of the expert/examiner, it
considers a representative sample of the traits looked for. Hewson, Luescher &
Ball (1999: 52) do not specify who decides if the relevant aspects are represented:
“Content validity implies that all relevant aspects of the behaviour are
represented”.
153
• The test interpretation is based on an empirical conclusion of correlation, meaning that
there is an empirically provable correlation between the results obtained in a specific test
situation and the behavior in reality. The factors involved here are criterion validity,
correspondence validity and prospective validity.
- Criterion validity: “… is measured when an established scale exists and a new
scale is compared to it in an assessment of relative cost and ease of use“ (Hewson,
Luescher & Ball 1999: 52)
- Correspondence validity is present if the survey of the criterion and the testing lie
very close together in time.
- Prospective validity means that the specific test can be used to predict a certain
behavior that will occur in the future.
• The test interpretation can also be based on the estimation of a psychological construct
variable. Hirsig (1993) defines a ‘psychological construct variable’ as a not directly
observable trait that is only defined according to a certain psychological theory, such as
‘fearfulness’, and also ‘aggression’. This type of validity is never definite, but always
changing, according to the model upon which it is based.
Summary of the discussion of validity:
As can be seen, the aspect of validity is far from simple. When talking about validation, one must
be careful to define the following:
(1) The type of validation: content-related, criterion-related or construct-related
(2) The methods used for validation, which depend on the type of evidence of validation
chosen.
(3) The aspect of the test that was validated, i.e. the criterion, the content and/or the construct.
In the case of the present study, this would be “aggressive behavior” in dogs.
Reliability
“Reliability refers to the degree to which test scores are free from errors of measurement…”
(APA 1985: 19). This involves “identifying the major sources of measurement error, the size of
the errors resulting from these sources, the indication of the degree of reliability to be expected
between pairs of scores under particular circumstances, and the generalizability of results across
items, forms, raters, administrations, and other measurement facets.” (APA 1985: 19). The
greater the measurement error, the smaller the reliability of the test.
Hirsig (1993: 6.3) defines reliability in the following way “…a psychometric test is reliable if it
measures what it measures, accurately” 16*. There are two additional aspects related to
‘reliability’: accuracy and replicability. The accuracy of a result can only be estimated by
repeating the procedure over and over again. This leads to the notion of replicability: Does the
same test evaluate the same individual equally in several repetitions? Dobbert (1982: 261) states
that “ the following considerations deal with validity and reliability simultaneously, but it should
be noted that fieldworkers employing the anthropological paradigm are generally more concerned
with validity than replicability. This stems from the aims of the discipline and from the nature of
the subject matter: First, there is always more than one valid view of any social situation ...
Second, human social situations are always in flux, thus making an emphasis on reliability
somewhat artificial. Third, ethnographers attempt to present material humanistically, so that it
represents a participant’s perspective, but the achievement of high degrees of reliability usually
requires abstraction and generalization, an ignoring of the particularistic, which indeed is what
16
“…ein psychometrischer Test, ist dann reliabel, wenn er das, was er misst, genau misst”
154
makes the situation real to its actors“. Although Dobbert is referring to human behavior in her
book, her definitions can, to a certain point, be generalized to animal behavior, including canine
behavior.
When working with living beings that are capable of learning, it is to be expected that there will
be a certain learning effect when repeating the same test, i.e. learning will bring better results the
second time or, conversely, worse results, if the dog lived through a negative experience the first
time. Additionally, the results depend on a variety of external and internal factors influencing the
animal’s, the owner’s and the examiner’s behavior. It is impossible here to attain one hundred
percent reliability.
To summarize the discussion of reliability: The aim of an evaluation method is to have as few
sources of measurement error as possible, i.e. as little variance as possible based on undesired
influences. In the perfect situation, the only variable influencing the result would be the studied
characteristic, for example, aggression, with all the other factors standardized to a norm and with
no measurement errors.
6.1.3 SUBJECTIVITY IN SCIENCE
In the field of behavior studies, there is no reference test of behavior that can be considered to
give a specification of the animal’s ‘real’ behavior. In any evaluative situation, what the expert
sees depends on his or her point of view (Dobbert 1982). This leads to the question: What is
reality? And from there to the question: Who defines what ‘reality’ is, or specifically for the
present study, what the dog’s behavior ‘really’ is?
Subjectivity can be found on two main levels of the scientific research process: in the phase of
observation and in the phase of interpretation.
Subjectivity at the Observation Level
The first research level, in which subjectivity interferes, is the observation, the collection of the
data. How good is the observation? Are data obtained valid and reliable? How objective are the
results? Do the results reflect the ‘real’ situation? What is reality? In his chapter on ‘Reality’, the
philosopher Grayling (Grayling 2002: 231) comments upon the general problem of the observer
influencing the observed object: “ The underlying dispute is an ancient one in philosophy, and it
stems from a simple obvious, but deeply puzzling fact: that one can by definition never observe
an unobserved thing, and so can at best only guess what any unobserved thing is like. One cannot
even claim with certainty that an unobserved thing exists. So any description one gives of a thing
is a description of it as an observed thing. But to stand in the relation of the observer to
something is to interact with it. Might not interaction modify its character, perhaps drastically? It
seems that we can never claim to know how anything is in itself, undisturbed by our relation to
it.”
This notion is very applicable to research: It is never possible to obtain entirely objective data, as
observation cannot be separated from the person who is evaluating and therewith relating to the
element being observed.
A second source of observational subjectivity is the fact that already in the act of observation,
every expert uses his or her experience, his or her reality as a reference, which may not agree
with the realities of others. In the end, all observation is subjective.
The aim on the level of observation is to minimize the subjectivity as far as possible, but to be
aware of the fact, that it is impossible to eliminate it.
155
Subjectivity at the Interpretation Level
The second level where subjectivity plays a role is in the interpretation of the results. All
knowledge is constructed on a model, on a certain idea of reality. In the process of interpreting
data, first a theory is constructed, upon which the research hypothesis is based. The hypothesis
leads to certain expectations. Terms are defined, the data is put into context and interpreted. For
example, in the context of the test situation of confronting a stimulus dog, the test dog is in a high
posture, ears and tail up high, growls, shows its teeth and shows piloerection – is this dog really
showing aggression? In order to be able to interpret the observed behavior, the term ‘aggression’
and the corresponding behaviors believed to be shown in this context, must be clearly defined. A
problem in the interpretation, e.g. not clearly defining the terms used, lead to problems in the
methodology.
The goal on this level is not to minimize subjectivity, but to be explicit about what one is talking.
6.1.4 STANDARDIZATION
In present study, one does not know if the comparison results obtained are reliable or whether the
methods involved deliver valid results in the targeted aspect, i.e. aggressive behavior. One can
know that the test situations are not standardized in that the situations are not the same for every
dog and for every time the test is run. The helpers change, the stimulus dogs change, the number
of people watching and making noise around the testing area differs from one dog to the other. A
field situation was deliberately chosen, as it is a more ‘natural situation’. Such ‘natural situations
are, however, quite difficult to standardize. Laboratory situations are easier to standardize, but
risk becoming ‘artificial’ in the sense that the situation is less comparable to a natural situation. If
one does choose a laboratory situation, with standardized conditions, one then has to make sure
that the results obtained in these ‘artificial’ surroundings correlate with what one wants to test in
natural surroundings.
6.1.5 CHOICE OF THE METHODS
As mentioned above, it would have been scientifically correct to choose scientifically validated
tests for this study. The actual situation at the time of testing was, however, that none of the tests
being used in Switzerland are validated yet. Apart from the fact that the tests were not validated,
there are several other factors that are important to be discussed: duration, structure and security.
Duration
The duration of a test ideally should be long enough to be able to assess the targeted behaviors,
but not so long as to tire the participants. The original plan was to include the Niedersachsen Test
(2001). This test, however, lasts an average of 2 hours per dog, which would have been quite
difficult to organize with the means at hand. The question also arises of whether a test of longer
duration actually does give a more accurate, reliable assessment than one of shorter duration.
This come back to the point of validity and reliability. This question can thus only be answered if
all the tests included have been validated to be testing what they claim to test, and if the results
obtained are reliable.
156
Structure and Security
The three participating tests were structured differently to the required space, the testing situation
and the security for the participating people and animals. Test B requires for the largest space,
using an open field and a path in the woods. This area was, however, not fenced and the dogs
were off-leash during the major part of this test, not involving any type of security precautions
for the participating people and animals (except for the hens that were kept in a fence). Test A
took place in a fenced in area and on a road where the dog was walked on the leash. The fence
hindered the dogs from easily running off and hence possibly putting nonparticipating persons in
danger. There were, however, no security precautions for the people helping with the test. Test C
shows the highest degree of security for all participants, with the expert and the stimulus dog
being in separate kennels, protected by the wire fence of the kennel walls, and the whole testing
area being closed off as well, hindering the dogs from escaping. Security is a very important
aspect when running such a project, as the highest priority should be to avoid anybody getting
hurt. The insurance and liability situation was clarified in advance.
6.1.6 DEFINITIONS
Something that strikes the reader of published literature is that authors often use words without
clearly defining what they mean by them. The researchers for the present study can,
unfortunately, be included in this group. The words ‘aggression’,’dangerousness’, ‘hierarchy’ and
‘attachment’ were not sufficiently clearly defined in advance. Additionally, the project’s material
used several words circumscribing behaviors vaguely, such as ‘friendly’, ‘open’, ‘neutral’,
‘fearful’, ‘mistrustful’, ‘threatening’ and ‘attacking’. As any lack of definition leads to
imprecision in the interpretation of the results, it is not clear that every expert was really talking
about the same thing when using the same word. The least number of misunderstandings would
occur, if a descriptive-contextual definition were given of the key words used, i.e. descriptions of
the behavioral unit and the circumstances in which it occurs (Dehasse, 2003). This is particularly
important when working with people from different “schools of thought” and coming from
different educational backgrounds. The same word must have the same meaning for all the
participants in order to allow a correct comparison of the evaluations. The lack of descriptivecontextual definitions from the beginning turned out to be a major point of imprecision in this
project.
6.1.7 STATISTICS: CHOICE OF SAMPLE POPULATION, EVALUATION
If the researchers do not have much experience in statistics (as was the case in this study) it is
wise to consult a statistician at the very beginning of the project. Such statistical advice was,
unfortunately, not available to the project from the beginning on. Based on the information given,
on the proposed methods and on the desired type of information, the statistician is able to advise
us about the type, structure, range and format of the scales and the statistical tests that should be
used to receive the best results. Certain statistical tests can be run without the help and support of
a statistician. A Chi-square test or a Fisher’s Exact Test are easily run in the appropriate
statistical programs. However, as there are conditions and limitations to which test can be used
under which circumstances, a person experienced in the field of statistics can best determine
these conditions. The choice of the appropriate statistical test depends on the size of the
population selected, on the number of results, on the distribution of the results, on the aim of the
project, on the questions asked, etc.
157
The Sample Population
For the present study, the consulting statistician recommended at least 60 dogs to ensure
statistically reliable results. It was only after the tests were run that all involved realized that
these 60 dogs were not sufficient when due to the small variability of behaviors that were shown
by the dogs in the test. The answers had to be pooled in order to make statistical evaluations.
For the data from the veterinary behaviorist and the ethologist, it was clear from the offset that
not enough dogs would be involved to allow statistical tests to be run. This was due to the fact
that the project did not have the financing, or the experts the time for evaluating more than 12
dogs each. These 12 comparisons were always intended to be descriptive and anecdotal.
The Evaluation Sheet and Questionnaire
An earlier involvement of the statistician would have no doubt resulted in a different format and
structure of the evaluation sheet. The possibility of marking ‘yes’ or ‘no’ for every answer and an
ordinal scale of answers would have been more appropriate for the information we wanted to
receive. The questionnaire could have been adapted to the evaluation sheet in order to allow a
statistical comparison of the two sources of information. Lehmann (2003) includes more detailed
discussion of this questionnaire.
The choice of the sample population
The manner of searching for the dog owners resulted in a pre-selection of persons interested in
their dogs’ behavior and working with their dogs. It was therefore clear from the beginning that,
given the group of people motivated to participate, not many real ‘problem cases’ would be
included. Several behavior therapists (veterinarians and non-veterinarians) were asked to send
patients, but not many replied. Consequently, the population participating in this project is not
representative (from the behavioral point of view) of the whole dog population of Switzerland.
However, it is not the aim of this project to generalize the behavioral profile of this sample
population to the whole canine population of Switzerland. The objective is to compare the results
of the tests, for which the ideal population would be one from a wide range of behaviors.
6.1.8 INFLUENCING FACTORS
When working with living beings, there are always multiple factors that influence their
behavior, both of the individuals being tested or observed (dogs and their owners), and of the
experts or observers evaluating the behavior. We have grouped the influencing factors into four
categories:
6.1.8.1
Identifiable, not controllable factors with unknown influence
Many factors can be identified, but their level of influence cannot be determined and the factors
themselves not be controlled. This category includes factors such as the mood, the health status,
the influence of hormones and sex cycle of all participating individuals (animals and humans), as
well as the learning effect that occurs with all the participants. Bias and subjectivity, on the side
of the owners and the experts, and the fact that dogs are always learning, have a great effect,
which is impossible to avoid.
158
Learning Effect on Dogs
As every animal, the dog is capable of learning. It is continuously learning, whether one wishes it
or not. The animal learns something (positive or negative for the surroundings) in every situation.
Considering this, the argumentation that the dogs might learn to react differently to a similar
situation in a following test and in this way falsify our results is a possibility to be taken into
account. While nothing can be done about the fact that dogs learn, the possible influences can be
identified and discussed. Several points were considered in this project:
• The time interval between two tests and the number of dogs tested per day were taken into
account by having each dog evaluated by one test a day within a period of one week.
• The sequence of the tests (See Table 2.59): To be able to uncover this potential learning
effect, the 60 dogs were divided into six groups differing in their test sequence (six
possible sequences). By correlating the results of the dogs with their test sequence in a
cross tabulation test, potential associations can be uncovered which may or may not be
based on a learning effect. However, the problem here was that the results of the dogs
were too homogenous to run a Chi-square test on this data, as there were too many
categories with less than 5 individuals in the expected count.
Table 2.59 Test sequence
Test sequence ABC ACB BAC BCA CAB CBA
Test 1
A
A
B
B
C
C
Test 2
B
C
A
C
A
B
Test 3
C
B
C
A
B
A
A = Test of the American Staffordshire Terrier Club of Switzerland
B = “Halterprüfung”
C = Test of the Canton of Basel-Stadt
The test sequence changed for 5 dogs, as Expert A1 of Test A was not able to evaluate one day
and the corresponding test had to be rescheduled to another day.
Bias of the Owner
Every owner is prone to bias. This has to be reduced or at least taken into account as far as
possible.
Not only the dog is capable of learning – so is the owner. Several measures were put into place to
avoid as much bias on the part of the owners as possible
• The tests included were not revealed to the owners in advance, in order to avoid their
practicing for the tests with their dogs.
• The experts were asked not to make any remarks, give any hints or feedback to the
owners to avoid these being motivated to practice to correct their mistakes with their
dogs on the next test. The aim was for the dog (and the owner) to be as much in the same
condition in every test as possible.
• There was no way, of course, to standardize the moods of the owners (or of the dogs, or of
the experts or the researchers). However, it should at least be recognized that differing
moods of all participants could also influence the dogs’ behaviors.
159
6.1.8.2
Identifiable, not controllable factors with identifiable influence
These factors can be determined and their influence may be identifiable retrospectively by
running statistical tests and looking for possible correlations. They cannot, however, be
controlled. One example for this type of factor is the ‘stimulus dog’, i.e. the dog with which the
test dog is confronted in certain test situations in order to evaluate the reaction of the test dog to
its conspecifics. The sex, status, size, breed and personal experience of these stimulus dogs may
influence the behavior and reaction of the test dog. However, one must take care when
interpreting any ensuing correlations: correlation does not automatically imply causality.
The results of the test also depend on the viewpoint of the examiner (Dobbert 1982), i.e. on his or
her bias. All the experts except one were familiar with the tests they were administrating. They
were the persons who usually judge these tests when they are run for their normal purpose.
However, when evaluating the data, the subjectivity of every expert must be taken into account as
well. Does one expert tend to judge more strictly than another does? Does one expert prefer a
certain breed to another? The fact that an external expert had to be found at short notice for
eleven dogs in Test A must be taken into consideration. This expert was not familiar with the test,
although she was familiar with canine behavior. In addition, this expert has a different
educational background (different school of thought in cynology than does the expert who ran the
same test.
For two of the tests, Test A and C, a stimulus dog was necessary.
As the tests were running all day, two to three dogs shared this job during the day, so as not to
tire the stimulus dogs. As any dog, these dogs are also capable of learning, have their individual
preferences, get tired, etc, which means that the stimulus dog was also not strictly standardized.
In one case, the stimulus dog Nr. 8 was not neutral towards the test dog, a German Shepherd.
Dog Nr. 8 had had a bad experience with a German Shepherd and consequently reacted
aggressively, which made it difficult to evaluate the behavior of the test dog.
The appointed stimulus dog was noted for every test dog, as to be able to uncover a possible
correlation between the stimulus dog and the test results. The correlation was run with a cross
tabulation test (see section 5.4 ‘Results of Correlation Within the Tests / Methods’).
6.1.8.3
Identifiable and controllable factors:
These are the factors that can be determined, identified and controlled to a certain point by
standardizing the test situation as far as possible. For instance, by making sure that the number of
helpers on the field, the distribution of genders among these helpers and the clothes they are
wearing is kept constant. These are all factors that could have been taken into account, but in the
present study, were not.
6.1.8.4
Identifiable factors that can be eliminated:
The influence of these factors can be eliminated. An example would be closing off the road
passing close-by while the tests are running to avoid traffic passing by, or forbidding the
participants to talk to avoid this type of noise influence. None of these kinds of factors was
explicitly dealt with in the field situations set up for this project.
160
6.1.9 CONCLUSION
The ideal any behavioral test strives for is to deliver valid, reliable and objective results in
standardized surroundings with the least possible varying influences. The big question remaining
is to what extent this is actually possible.
In the field of behavioral medicine, one works with unpredictable, constantly learning and
changing living beings whose behaviors are influenced by many “non-measurable”, partially noncontrollable or not identifiable factors. Again, the aim is to identify, control and eliminate these
factors as far as possible in any one study while realizing that complete control and elimination is
impossible.
161
6.2
DISCUSSION OF THE DESCRIPTIVE EVALUATION OF EVERY
METHOD
6.2.1 DOGS
The participation of the dog-owner teams in the project was greater than expected. Only three of
the originally scheduled teams were not able to participate. The loss of data through these dogs is
not very great. These three dogs (5%) were not considered when comparing the according test
with the other tests.
6.2.2 DATES
The rescheduling of 15 dogs of Test A to another day led to several consequences which must be
taken into account when evaluating the data.
• The test sequence was changed for five dogs from BAC to BCA. However, as the test
sequence could not be taken into account in any of the tests for correlation, because the
distribution of the dogs’ behaviors was too homogenous to run a Chi-square test, this
change does not have an identifiable effect on the reported results.
• Eight dogs ran through two tests on one day instead of the planned one test per day. It
cannot be determined whether this fact had an influence on the results of these eight
dogs or not. This belongs to the category of identifiable, not controllable factors with
unknown influence.
• In Test A, an independent expert unfamiliar with the test evaluated 10 dogs (one dog
could not participate on that day). This Expert A2 has a different educational
background than does Expert A1, which can partially be seen by the tendency of Expert
A2 to evaluate dogs to be ‘dominant’ over their owners, to have ‘no attachment’ and to
recommend measures, especially education classes.
These points have especially to be kept in mind in the consideration of correlations of the
affected data.
162
6.2.3 TEST EXPERTS
The only test with an even distribution of dogs tested by both experts is Test C with 30 dogs per
expert. The same had been planned for Test A, but had to be changed because of the last minute
substitution of one expert. Expert A2 only evaluated 10 dogs (one of the 11 dogs did not
participate), whereas Expert A1 evaluated the remaining 49. As the two experts have a different
educational background and work according to two different theories, a tendency of an internal
variation within the test can be seen, as is shown in the section 5.4 (‘Results of the Correlation
Within the Tests’). This internal imprecision should be kept in mind when comparing Test A to
the other two tests.
The facts that the evaluation of the dogs was not restricted to one expert in Test B and that the
number of dogs evaluated by each expert is not the same, make this point difficult to consider in
this test as well. As Expert B1 started evaluating dogs that had passed the ‘Ist-Stand’ with Expert
B2 while Expert B2 was still going through the ‘Ist-Stand’ with the remaining dogs,
Consequently, Expert B1 ended up testing more teams (38) than did Expert B2 (19) and Expert
B2 evaluated more dogs that did not pass the test, as not passing the ‘Ist-Stand’ was considered as
not passing the test.
6.2.4 STIMULUS DOGS
Tests A and C required stimulus dogs to be organized by the project researchers; Test B experts
brought their own stimulus dogs for which no information was made available of the sex, breed
and age.
The requirement to use only intact male dogs in Test C proved to be misleading, as one of the
intact females that passed the test and was evaluated to show ‘open, friendly, neutral’ towards an
intact male stimulus dog in the test then attacked a spayed female during the break a few minutes
after having passed the test. This one test female was evaluated to show ‘dominant behavior or
mistrustful, fearful behavior without aggression’ towards the stimulus dog in Test A and
‘attacked with threatening’ in Test B, whereas she was evaluated to be ‘open, friendly, neutral’ in
Test C. Even though this is only one dog, it is exactly this kind of dog that should be identified in
a behavioral test with the aim to increase public security. The lesson to be learned from this is
that the choice of stimulus dogs must be carefully made.
The three dogs used for Test C were also partly used for Test A, meaning that some of the dogs
would have been familiar with the stimulus dog from the previous test. The consequence is that
the situation was not the same for all dogs, depending on which stimulus dog they were exposed
to in the previous test.
Retrospectively, it was clear that the history of all the stimulus dogs should be taken as well, in
order to avoid too great an influence of the stimulus dog on the test dog. One of the stimulus dogs
had a specific aversion towards breeds of the shepherd type, as a result (according to its owner)
of an incident he had had with a dog of this type. One of the dogs to be confronted with this
stimulus dog was a German Shepherd. The stimulus dog itself did not react neutrally to this dog;
it threatened the test dog, which makes the reaction of the test dog towards a ‘neutral’ dog
difficult to interpret. Luckily for us, this occurred in Test C where the stimulus dog was in the
kennel, so the degree of danger of injury was diminished by the wire fencing.
163
6.2.5 TEST SEQUENCE
Originally, the test sequence was considered in order to be able to recognize an eventual
influence of one test on the other. A definitive determination of this factor was difficult for
several reasons:
• As most dogs were considered to show ‘not aggressive behavior’, the distribution of
behaviors within the groups of test sequence was not large enough to allow a reliable
interpretation of the results of a Chi-square test. As a consequence, the potential
influence of the test sequence on the results could not be analyzed.
• The distribution of the dogs in every sequence-group is not equal. As at least five
individuals per group are necessary to obtain reliable results with the statistical tests, this
is not possible with one group consisting of only five individuals.
• As eight dogs ended up taking two tests on one day, the influence of their sequence of
tests is not easily interpretable. A tiring effect on the dogs and the additional stress
brought by the owners who had to travel from one test location to the other within a
short period of time, must be taken into account as well. These teams did not have the
same conditions as did the other teams.
6.2.6 EXACTNESS OF ANSWERS
An important point to consider when evaluating binary answers, such as ‘attachment existent’ or
‘no attachment’ and ‘owner dominant’ or ‘dog dominant over owner’ (which were considered as
two opposing answers in this study), is that the fact that one of these points is not marked does
not necessarily imply that the other is. Especially in Test B, where up to 23 dogs did not receive
an answer to a point, the possibility remains that no answer was chosen for this specific criterion.
This could have been avoided by using ‘yes/no’ answers for every point. However, since this had
not been done, it was decided for the evaluation of the results to combine the two answers
‘hierarchy intact (i.e. owner dominant over dog)’ and ‘dog dominant over owner’ into one
question ‘hierarchy’. The two original answers ‘hierarchy intact’ and ‘ dog dominant over owner’
are now the answers to the criterion ‘hierarchy’. This also allows the missing values (i.e. where
none of the two options was chosen) to be considered. The same was done with the two criteria
‘attachment existent’ and ‘no attachment’, which became the answers to the criterion
‘attachment’, and for ‘measures recommended’ and ‘no measures recommended’ for the criterion
‘measures’.
When describing the results, it is obvious that the experts for Test B were the most inconsistent
with marking the answers, whereas the experts of Test C were the most consistent in their
answers.
In the data for Test B, the two points with the most missing answers are ‘behavior towards
moving objects’ and ‘behavior towards other animals’. One possible explanation for this is that
these two points are not considered in the ‘Ist-Stand’ and that they were consequently not
additionally evaluated with the dogs that failed the ‘Ist-Stand’ and did not go on to the
‘Halterprüfung’. All the dogs for which these answers are missing failed the ‘Ist-Stand’.
For two of the missing answers within the criterion ‘hierarchy / attachment’, the veterinary
behaviorist did not think either of the two options given were suitable, and therefore chose not to
give any answer at all. An additional option should be given, such as ‘no evaluation possible’
164
6.2.7 INTRASPECIFIC AGGRESSION (CRITERION 1)
The Tests
All three tests evaluated most of the dogs to show ‘open, friendly, neutral’ behavior towards
other dogs.
Test C evaluated more dogs to show ‘threatening’ behavior than did the other two tests. 70% (N
= 5) of the 7 dogs evaluated to show ‘threatening behavior’ towards other dogs in Test C were
males, 3 of which were intact, only 2 were spayed females. A factor to take into account here is
that Test C experts’ a priori assumption that encounters between two males, especially intact
males, are often more aggressive than encounters between two females or a male and a female.
The two dogs with this score in Tests A and B are both males, one intact, one castrated.
Although not statistically significant, it is important to point out that two dogs were evaluated to
‘attack with threatening’ in Test B and one dog in Test C. One of these dogs in Test B, is the
above mentioned female (dog number 41, see Figure 2.17b) that attacked another female during
the break of Test C. These are the dogs that are actually ‘interesting’ to identify. It would be
interesting to compare these results with the history of the dogs. The veterinary behaviorist (VB)
also evaluated the dog that attacked in the break and considered it to ‘attack without threatening’.
However, the consultation with the VB took place after the tests, this allowed him to receive
some information on the effects of the tests on the dogs. This dog that attacked was reported by
its owner to have gotten even more aggressive towards other dogs after the tests.
In Test A, this female was evaluated to show ‘dominant behavior or mistrustful, fearful behavior
without aggression’ towards other dogs (See Figure 2.17b, dog number 41). In this test, the dogs
are on the leash when walking by the stimulus dogs, which was an intact male in her case, i.e. a
kind of dog with which she normally did not have has problems.
The group of stimulus dogs for Test B was made up of both sexes, which permitted observation
of the test dog’s reactions to different conspecifics of both sexes.
The Behavioral Specialists
The veterinary behaviorist and ethologist generally evaluated the dogs higher in the scale of
intraspecific aggression than did the tests (See Figures 2.18a and 2.18b). This could be due to the
additional information, which these experts receive from the owners. In the behavioral
consultation, the history, experiences and observations of the owner are the main source of
information. The ethologist additionally has the direct information from observation of the dog in
the actual problem situation, if possible, whereas the tests are momentary pictures in time, with
the only information coming from the behavior of the dog at the specific moment of the test.
6.2.8 INTERSPECIFIC AGGRESSION (CRITERION 2)
The Tests
For the criterion of ‘interspecific aggression’, most of the dogs were evaluated to show ‘open,
friendly, neutral’ behavior towards humans in every test. However, Test B did so in 61% of the
cases, as opposed to 83% and 88% in the Tests A and C (See Table 2.16). Test B considered a
higher percentage (30%) of dogs to show ‘dominant behavior or mistrustful, fearful behavior
without aggression’ than did the other two tests (9% and 10%) (See Table 2.16). This might have
to do with the fact that the people in the situations of Test B were the most ‘fear-inducing’, as
they were making a lot of noise when approaching the dog and also when forming a circle around
the dog. Here, there is an additional component, which is not present in the other two tests –
noise. It is likely that more dogs would react fearfully in such a situation, as opposed a more or
less silent moving group of people in Test A or the two persons standing quietly in a kennel in
165
Test B. Test A also has a stress situation consisting of the group of people standing in a circle and
closing in on the dog. This stress level might be considered similar to the stress level in the circle
situation of Test B, however with the missing stimulation of noise. Test A found 3 dogs to ‘attack
with threatening’, one of which was also found in Test B to show this behavior (See Figures
2.19a and 2.19b). Test C did not evaluate any dog to show this behavior. Again, the stress level
caused by humans for the dog in Test C seems to be of a lower degree, as the humans are either
in the kennel or only shortly in contact with the dog when reading the chip.
The Behavioral Specialists
The veterinary behaviorist evaluated the dogs higher in the scale of interspecific aggression than
did the three tests (See Figures 2.20a and 2.20b), with only 33% showing ‘open, friendly, neutral’
behavior, and 25% evaluated to show ‘threatening behavior’ and ‘attacking behavior without
threatening’ (See Table 2.17). Again, this information is based on the experience of the owner
with his or her dog.
The ethologist agreed more with the tests in the sense that she evaluated 75% (N = 9) of the 12
dogs to show ‘open, friendly, neutral’ behavior.
6.2.9 INTERSPECIFIC AGGRESSION, DOG ALONE (CRITERION 14)
The Tests
For interspecific aggression of the dog in absence of its owner, in Test A, the majority of the dogs
was evaluated to show ‘open, friendly, neutral behavior’ towards humans, only one dog was rated
as ‘threatening’ and one dog as ‘attacking with threatening’. These results reflect, once again, the
small variety within the dog population participating in this study. Most of the test dogs were
well trained, well educated and do not show much aggressive behavior in general during test
situations.
The Behavioral Specialists
Again, the veterinary behaviorist generally evaluated the dogs higher in this aggression scale than
did the test (See Figure 2.21). Of the three dogs evaluated by him to ‘attack without threatening’,
two were considered to show ‘mistrustful, fearful behavior without aggression’ and one to show
‘open, friendly, neutral’ behavior towards humans in the test. The one dog that ‘attacked with
threatening’ in the test was evaluated by the veterinary behaviorist to ‘threaten’. These
differences in evaluation may again be the consequence of the additional information the
veterinarian had about the history of the dog.
6.2.10 BEHAVIOR TOWARDS MOVING OBJECTS (CRITERION 5)
The Tests
More dogs in Test A were considered to show ‘open, friendly, neutral’ behavior towards moving
objects than in Test B (See Table 2.20). Care must be taken when comparing these numbers, as
38% of the dogs did not receive an evaluation by Test B in this point. When taking the total
number of dogs evaluated as the 100%, 95% of these were considered to show ‘open, friendly,
neutral’ behavior towards moving objects in Test B and 83% in Test A. Two of the 23 dogs
without an answer to this point did not participate in the test, all the rest are dogs that failed the
‘Ist-Stand’. The experts had been asked to run these dogs through the test situations to evaluate
the behavior towards other dogs, towards humans, towards other animals and towards moving
objects. This was not done with 21 of the dogs that failed the ‘Ist-Stand’.
166
The Behavioral Specialists
The question of whether predatory behavior and pursuit are components of aggressive behavior
or not, is a matter of dispute between the different schools of thought in this area. The two
answers given by the veterinary behaviorist ‘open, friendly, neutral’ and ‘attacking without
threatening’ make sense, when one considers the sequence of predatory behavior not to be part of
the aggressive behavior, as it does not make sense to threaten and forewarn your prey. This is an
explanation for the missing answers involving ‘threatening’. This behavior also does not seem to
be shown consequently by the dogs and does not occur in every situation. In Test A, the dog is on
a leash when the bicycle and jogger pass it, which decreases its possibility to attack and might
influence the evaluation of the dog. The information the veterinarian received, however, is based
on the owner’s experience in every day life. Even though not statistically significant, the
difference is quite striking, as the veterinarian evaluated 6 dogs to ‘attack without threatening’
and neither of the tests evaluated any dog to show this behavior (see Tables 2.20 and 2.21). Three
of the dogs evaluated to ‘attack without threatening’ by the veterinarian were considered to show
‘mistrustful, fearful behavior without aggression’ in the test: the other three were evaluated to
show ‘open, friendly, neutral’ behavior (see Figures 2.23a and 2.23b). One point that must be
considered is the capability of the owner to recognize and determine ‘threatening’ and
‘aggressive’ behavior. This is, therefore, a weak point in the information source of the
veterinarian who has to rely on the owners’ observations and memory.
The ethologist, again, agrees more with the tests than with the veterinary behaviorist in that 82%
(N = 9) of the dogs were considered by her to show ‘open, friendly, neutral behavior’, and none
‘attacked without threatening’ (see Table 2.21), as did 6 with the veterinary behaviorist (not the
same 12 dogs were evaluated by both). One possible explanation might be the type of moving
objects taken into account by the ethologist and the veterinary behaviorist. The ethologist relies
more on the behaviors she actually observes, whereas the veterinarian relies on what the owners
say. This difference in source of information might be seen in theoretical situations like the
following: A dog reacts to joggers, but no jogger was encountered on the walk with the
ethologist, so, she might not have considered this possibility; the veterinarian might have
received this information of the owner during the interview.
For his part, the veterinarian has to be very careful to define the words he uses, to make sure that
the words mean the same thing to him and the owner. For example, an owner might not consider
his or her dog aggressing, if the dog chases a bicycle. If the veterinarian is not very careful in
analyzing the sequence of and the body postures shown during the reported behavior, he will not
be able to reliably interpret what kind of behavior the dog is showing.
6.2.11 AGGRESSIVE BEHAVIOR TOWARDS OTHER ANIMALS (CRITERION 7)
The Tests
Aggressive behavior towards other animals was only evaluated in Test B. Here, 20% of the dogs
did not receive an answer, leaving 48 dogs that were evaluated on this point. Of these, 79%
showed ‘open, friendly, neutral’ behavior towards other animals (see Table 2.22), which is
surprisingly few, considering the number of dogs that show pursuit behavior in general, and the
fact that the dogs were off leash all the time. As mentioned, the dogs participating had a good
average obedience, i.e. they were in good control by their owners, which might mean that their
attention is more difficult to deviate from the owner to the distracting stimulus of the specific test
situation.
167
The Behavioral Specialists
Unlike the Test Experts, the veterinary behaviorist evaluated 75% (N = 9) of the dogs to ‘attack
other animals without threatening’ and two to ‘attack with threatening’. None of the dogs was
evaluated to show ‘open, friendly, neutral behavior’ towards other animals (see Table 2.23). Of
the 9 dogs considered to ‘attack without threatening’ by the veterinary behaviorist, 6 were
evaluated to show ‘open, friendly, neutral’ behavior towards other animals in the test, 2 to show
‘mistrustful, fearful behavior without aggression’ and 1 answer is missing (see Figure 2.24a).
Even though the dogs were off leash when confronted with the animals, they did not attack them
in the test.
The ethologist again considers the majority of the dogs evaluated by her (75%, N = 9) to show
‘not aggressive’ behavior and only 17% (N = 2) to threaten (see Table 2.23). However, there
must be a history of attacking in the three dogs that were evaluated by both behavioral specialists
(dogs number 9, 39 and 49), as this is the information received by the veterinarian. The behavior
of the dog might depend on the type of ‘other’ animals. The test situation included only goats and
chickens, animals mainly encountered on the countryside whereas the evaluation of the
veterinarian might have been based on the owner’s reports of situations with cats, rabbits, deer,
etc. In order to be able to compare the results of the test with those of the behavioral consultation
on this point, this aspect would have had to be defined more clearly in advance.
6.2.12 RELATIONSHIP DOG – OWNER (CRITERION 11)
6.2.12.1 ‘Hierarchy’
Tests
For the criterion ‘hierarchy’, both Tests A and C considered the owner to be ‘dominant’ over the
dog in the majority of the cases. In Test B, 77% (N = 43) of the teams did not receive an answer,
leading to the conclusion that the experts in Test B did not answer this question consistently.
When 100% represents the total number of dogs minus the missing answers, 93% of the 60
owners were evaluated to be ‘dominant’ in this test. Twelve of the 13 dogs passed the test and
one failed. Test A considered 10 dogs to be ‘dominant over their owners’ which is more than in
the other two tests (see Table 2.24).
In retrospect, it seems that this criterion might not have been appropriately labeled. It might have
been difficult for the expert to evaluate most accurately and comprehensively the hierarchical
relationship of a dog-owner team, as this relationship may change according to the circumstances
and surroundings. A dog may behave perfectly during training, but show dominant behavior
when at home. These results must therefore be interpreted with caution, as the terms ‘hierarchy’
and ‘dominance’ were used in the project’s evaluation sheet with their popular meanings and not
as ethological definitions.
Behavioral Specialists
The veterinary behaviorist considered the owner to be ‘dominant over the dog’ in all the cases for
which he answered this question (N = 8) (see Table 2.25). Two of these dogs were considered to
be ‘dominant over their owners’ by Test A, one by Test C (not the same dog as in Test A). This
answer cannot be compared to the results of Test B, as all 8 dogs that the veterinarian evaluated
in this point did not receive an answer in Test B (see Figure 2.26a).
The ethologist considered 6 of the 12 dogs evaluated by her to be ‘dominant’ over their owners
(see Table 2.25). In two cases, this agreed with the evaluation of Test A and in one of these cases,
Test B evaluated the hierarchy in the same way (see Figure 2.26b). The difference between the
ethologist and the tests may be explained by the different surroundings in which the dogs were
168
evaluated. It is possible that the dog will show a different behavior when at home than when
running through a test. As the aspect of ‘dominance’ is related to a social interaction in a specific
situation, the dog may be ‘dominant’ over the owner in a home situation, but not so in ‘field’
situations.
6.2.12.2 ‘Attachment’
The Tests
Test C did not answer the criterion ‘attachment’ consistently; 50% (N = 30) of the dogs were not
evaluated in this point (see Table 2.26). This inconsistency is linked to Expert C2 who only
answered this question for one of the 30 dogs he evaluated. Within Test A, it was Expert A2 who
mainly evaluated the dogs to have ‘no attachment’ to their owner. This may be a consequence of
the different educational background of the experts and the distinct emphasis given by them to
the points ‘attachment’ and ‘obedience’. As the word ‘attachment’ was not clearly defined, the
experts might be talking about something different when using the same word, resulting in
differing answers. Also, the terms ‘attachment’ and ‘obedience’ might not having been clearly
distinguished by all experts.
The Behavioral Specialists
The veterinary behaviorist considered all 12 dogs to have an ‘attachment’ to their owners (see
Table 2.27). Five of these dogs were considered to have ‘no attachment’ to their owners in Test A
and 4 in Test B, whereby 3 of these are the same as in Test A. Test C did not give an answer to
this point for 5 of these dogs. Of the remaining 7 dogs, one was considered to have no
attachment’, as in the other two tests (see Figure 2.28a). This difference between the veterinarian
and the test experts might be based on the above mentioned different definition of ‘attachment’
and the criteria on which the decision ‘existent’ or ‘not’ is based.
Both the ethologist and the tests considered the dogs of the 12 teams evaluated by all of them to
‘have an attachment to their owners’ (see Figure 2.28b).
6.2.13 TEST PASSED (CRITERION 12)
Test B is interesting, as only half the dogs passed this test, whereas nearly all dogs passed the
other two (see Table 2.28). Here, the stricter evaluation of Test B is relevant – 20 dogs failed the
‘Ist-Stand’ without even entering the actual ‘Halterprüfung’ and thereby were considered as
having ‘failed’ the test. Nine dogs passed the ‘Ist-Stand’, but then failed the ‘Halterprüfung’.
Nevertheless, half the dogs passed the ‘Halterprüfung’ in spite of its stricter conditions
concerning obedience and not being allowed to reward the dog as usual (only the voice and
petting are allowed as a reward). In addition, the aggressive behavior of the dog is not the only
basis for deciding whether a dog passes the test or not. Depending on the test, other criteria not
considered in this study have an important influence as well. This is the reason why this criterion
was taken into account on the same level as all the other points, and not as the ‘total score’ of all
the other criteria.
169
6.2.14 MEASURES RECOMMENDED (CRITERION 13)
The Tests
According to the more strict evaluation of Test B, this is the test where the fewest dogs were
given no recommended measures: 18% (N = 9) as opposed to 85% (N = 50) in Test A and 82%
(N = 49) in Test C. Correspondingly, measures were recommended more often in Test B than in
the other two tests (see Table 2.29). The measures ‘obligatory leash’ with 19% (N = 11) and
‘education class’ with 66% (N = 38) were the most often recommended. The recommendation of
‘education class’ is not too surprising here, as the Expert for Test B also runs his own dog school.
Classes to diminish the problems within a team are usually recommended when this test is
privately run by the expert. Test B also puts more emphasis on obedience, as is seen for example
in the fact that the ‘Ist-Stand’ contains many elements of obedience and that the dog is off-leash
during the whole ‘Halterprüfung’ – i.e. a high degree obedience is necessary for the dog to pass
through all the distracting situations.
A muzzle was only recommended once, by Test C. Therapies were recommended rarely, which is
not surprising either, as the population of dog owners participating in this study were mainly
already working with their dogs in one way or another. Rehoming was only taken into
consideration by Test B, in recommendations to two dogs. One of these was evaluated to show
‘attacking behavior with threatening towards humans’, ‘dominant over its owner’ and ‘to have no
attachment’. Several measures were recommended for this team: ‘Obligatory leash’, ‘education
class’ and ‘rehoming’. The second dog was evaluated to show ‘dominant behavior or mistrustful,
fearful behavior without aggression towards humans’, ‘was dominant over its owner’ and
considered to have ‘no attachment’ to its owner. This dog received the same three recommended
measures: ‘obligatory leash’, ‘education class’ and ‘rehoming’.
Test A and C both only recommended one measure per dog; Test B imposed one measure in 64%
(N = 32) of the cases receiving measures and more than one in 18% (N = 9). The measures
mostly combined were ‘education class’ and ‘obligatory leash’. Here, once again, the educational
aspect and the different background of the test development of Test B seem to be influencing
factors.
The Behavioral Specialists
The veterinary behaviorist and the ethologist recommended different measures than did the tests,
which may be explained by their different educational backgrounds and professions. A muzzle,
an obligatory leash and therapy were recommended for three dogs by the veterinary behaviorist.
Education classes were recommended less often, but a therapy was considered to be necessary for
5 of the 12 dogs evaluated by the veterinarian, and 3 of the 12 dogs evaluated by the ethologist.
The measure ‘education class’ was chosen for only 2 dogs in Test A and 1 dog in Test B. In Test
B, 9 of the 12 dogs also evaluated by the veterinary behaviorist received no measures. For one
dog for which the veterinarian had recommended the three measures, therapy was recommended;
in one dog the veterinarian and Test B agreed with the measure of ‘education class’.
In one case, Test A recommended education class where the veterinary behaviorist did not
recommend any measure. Test B imposed education class for 10 of the dogs also evaluated by the
veterinarian, agreeing with him in one case, although adding the measure of obligatory leash and
recommending rehoming as well. They agreed on a therapy for the same dog.
Test C agreed with the veterinary behaviorist in four cases of not recommending any measures
and imposed an education class in three cases where the veterinarian did not (see Figures 2.29a
through 2.30b and tables 2.29 through 2.38).
170
In general, there seems to be a tendency of each method to prefer particular measures: the
veterinary behaviorist tends to recommend therapy and muzzle, the ethologist tends to
recommend therapy, and Test B definitely tends to recommend education class. Tests A and C
are more restrictive in recommending measures. This may be a consequence of the underlying
aims and purposes of the methods. The veterinarian is very concerned with the safety of the
surroundings of the dog and with the wellbeing of the animal. He primarily works on the basis of
medication and behavioral therapies and thus recommends these methods, which are most
familiar to him. The aim of Test A is to recognize dangerous individuals; obedience is not as
important a factor. In this test, the emphasis is put on the dog-owner team and the cooperation
within the team. Test B is combined with an educational school of dog trainers leading to the fact
that the experts recommend their way of working and solving problems, which is usually
education classes. Test C is a test that is mainly concerned with the safety of the general public,
and also normally restricted to certain breeds. The measures recommended by this test also
depend on the history of the dog, whether its behavior has previously led to any incidents and/or
legal consequences. The education classes and therapies are not offered by the test experts
themselves, they refer the dogs to trainers and therapists.
171
6.3
DISCUSSION OF THE RESULTS OF THE COMPARISON OF THE
THREE TESTS
6.3.1 INTRASPECIFIC AGGRESSION OF THE DOG (TOWARDS OTHER DOGS)
The Kappa test shows that there is a significant agreement among the three tests. This agreement
however is quite low. When looking at the tables in Appendix I, it can clearly be seen that the
76% agreement of Tests A and B is higher than the correlation of either of these tests with Test C
(both 59%). The agreement between all three tests, consequently, is even lower (46%).
Considering the fact that these tests all evaluated the same sixty dogs, and all claim to be able to
evaluate correctly the behavior of the dog, a difference of 54% is large, hence the low correlation
in the Kappa test.
One possible explanation for the higher agreement of Tests A and B might be the way the dog‘s
behavior towards other dogs was evaluated in the tests. Tests A and B both had no restrictions
concerning the sex of the stimulus dog, allowing a wider range of dogs reactions. Test C requires
that only intact male dogs be the stimulus dogs, as, according to the developers of this test, this is
where most of the problems occur. It was in this test, however, that the four-year-old female
intact dog (dog number 41) passed the test without problems and attacked another female spayed
dog quite severely during the break (a deep wound that had to be sutured, including drainage, by
a veterinarian). As the stimulus dog in the test was an intact male and this bitch has no problems
with males, this behavior was not recognized. This is an anecdotal example, with no statistical,
but much informational value and might lead to the consideration of including a variety of dogs
of different sex in a behavioral test.
The surroundings of Test C differ from those of the other two tests, as the whole test is held
within an enclosed area, the stimulus dog being confronted within a person or dog in a kennel.
This, on the one hand, increases the security of all individuals involved as well as the possibility
of standardizing the test. On the other hand, however, it decreases the capacity to observe the
dog’s behavior in detail, as the bars obstruct the view, and the dog most probably does realize
that there is something that stops it from getting to the stimulus dog. Hence, one hypothesis for
the higher agreement of Tests A and B is that the test situations are similar in both tests.
Test C evaluated the dogs higher on the aggression scale in 63% of the differing cases when
comparing it to Test A and 68% of the differing cases when compared to Test B. This difference,
however, was not significant in either of the comparisons.
172
6.3.2 INTERSPECIFIC AGGRESSION OF THE DOG TOWARDS HUMANS
According to the Kappa test, there is a higher agreement in the criterion ‘interspecific aggression
towards humans’ among the three tests than would be expected with a random distribution of the
results within the categories. The three tests agreed in 55% of the cases that were tested by all
three tests. Here, the highest percentage of agreement (73%) is found of Test A and C. In 77% (N
= 10) of the 13 cases these two tests did not agree on (see Figures 2.19a and 2.19b), Test A
evaluated the dog higher in the scale as did Test C, whereas the opposite was only the case in
23% (N = 3). This leads to the conclusion that Test A evaluated more strictly in this point than
did Test C. The agreement of both Tests A and C with Test B are 65%. Test B evaluated the dogs
higher in the scale than Test C in 33% of the cases, the opposite is only found in 2% of the cases
(see Figures 2.19a and 2.19b). Test B also has the tendency to evaluate the dogs higher in the
scale that Test A, but the difference is not significant.
To summarize the results of interspecific aggression of the dog towards humans, the highest
agreement is found between Test A and C, Test B has the tendency to evaluate the dogs higher in
the scale than both Tests A and C. When taking into account the individual distribution of the
answers, the higher marking of Test B is due to evaluating more dogs to show ‘dominant
behavior or mistrustful, fearful behavior without aggression’ towards humans. This might be a
consequence of the way the test is set up, in the sense that there are several situations in this test
where the behavior of the dog towards humans can be evaluated. However, most of these
interactions in this rest are quite stressful for many dogs (involving loud noise, people forming a
circle around the dog, people jumping up and down). These stressful situations may be more
likely to elicit fearful, mistrustful behavior than for example the group of people walking around
more or less quietly in the fenced in area of Test A or the experts standing in a kennel in Test B.
6.3.3 BEHAVIOR OF THE DOG (ALONE) TOWARDS HUMANS
This point cannot be compared as only Test A evaluated this situation.
6.3.4
THE BEHAVIOR OF THE DOG TOWARDS MOVING OBJECTS (JOGGERS,
BICYCLE)
The Kappa test did not find an agreement higher than would be expected by random distribution.
However, when considering the percentage of agreement in relation to the total number of dogs
(N = 60) minus the 23 missing values (where the results were missing in either one of the tests or
in both), there is an agreement of 84%, and only a difference of 16%. When comparing this
percentage with for example the percentage of agreement in the criterion ‘interspecific
aggression’ of Tests A and B (59%) where a significant agreement was found, it is quite high.
Test A, however, did evaluate more dogs to show ‘dominant behavior or mistrustful, fearful
behavior without aggression’ than did Test B. The situation in which this behavior is tested is
quite different in both tests, the dog being off leash and confronted with a ‘moving object’
accompanied by one or two dogs in Test B, whereas on the leash and confronted once from the
back and once from the front by a jogger or cyclist.
6.3.5 BEHAVIOR OF THE DOG TOWARDS OTHER ANIMALS
This was only tested by one test and hence could not be compared.
173
6.3.6
THE RELATIONSHIP OF THE DOG AND ITS OWNER
6.3.6.1
‘Hierarchy’
Quite a large number of dogs did not receive an answer to the criterion ‘hierarchy’ in Test B,
which reduced the total number of dogs that can be compared with this test to 14. However,
almost all of these 14 dogs were evaluated to have an intact hierarchy, meaning that the owner
was considered to be ‘dominant’ over the dog. Care must be taken when interpreting these
results, as this is only a bit more than a sixth of the total number of dogs evaluated by the other
two tests.
The three tests seem to agree quite well on the ‘hierarchy’ of the dogs and none of the tests shows
any tendency in considering one answer significantly more than the other tests.
6.3.6.2
‘Attachment’
Test C did not answer this question in 30 cases, Test B in 18 cases. These missing values result in
a reduced number of dogs possible to compare among the three tests.
Within the differing results, Test B has a tendency to consider more dogs to ‘have an attachment’
when Test A does not evaluate them so and vice versa (see Figure 2.27a and 2.27b). One possible
explanation for this tendency might be the individual differing opinions of the experts in Test A.
In Test A, five (50%) of the ten dogs that were considered to have ‘no attachment’ were
evaluated by expert A2. Expert A2 only evaluated 10 dogs in total, which means that 50% of the
dogs evaluated by this expert were considered to have ‘no attachment’, whereas this was the case
in only 10% for Expert A1.
When interpreting the results of all comparisons with Test C, one must keep in mind that 30 dogs
did not receive an answer to this question, and that the thirty dogs, which were evaluated by the
experts of Test C in this point were considered to ‘have an attachment’. It seems quite logical
that, as a consequence, the only possible deviation of Test C from the other tests, is this test
considering certain dogs to ‘have an attachment’ that were considered to have ‘no attachment’ in
Tests A or B, as the opposite is not possible.
6.3.7 TEST PASSED
Tests A and C seem to have similar factors that are deemed necessary in order to pass the test,
whereas Test B seems to be stricter. Test B is more demanding in its conditions to allow teams to
participate in the actual ‘Halterprüfung’; in this project 20 dogs did not pass the ‘Ist-Stand’ and
never entered the actual test. These dogs were considered to have failed the actual test without
ever actually having taken it, but by virtue of having failed the ‘Ist-Stand’ pretest. Nine dogs that
entered the ‘Halterprüfung’ failed the actual test. These 9 dogs are still significantly more than
the 2 dogs that failed Test A. No dog failed Test C. As discussed in the following section, Test B
puts much emphasis on obedience and the lack of it seems to be an important factor in a dog
failing the test. 25 (86%) of the 29 dogs that failed received the measure ‘education class’,
whereas only 13 (45%) of the 29 dogs that passed did.
174
6.3.8 MEASURES RECOMMENDED
Test B recommends measures significantly more often (82%) than do either Tests A (15%) and C
(18%). The measure most often recommended is ‘education class’. This seems quite logical when
considering the fact that the developers of this test also run a dog training school. It also reflects
the fact that obedience seems to be an important factor to this test, which is also reflected in the
way Test B is structured. The preliminary ‘Ist-Stand’ evaluation focuses on several aspects of
obedience (such as ‘walking’ on the leash and ‘down out of movement’), no rewards are allowed
except for voice and touch, and the dog is off leash during the whole test.
Tests A and C agreed on the specific recommendation in the majority of the cases (75%).
Reflecting the higher proportion of recommended measures in Test B, the agreement of this test
with the other two tests is not very high.
The recommendation of an obligatory leash by Test B in 11 cases also reflects the importance of
obedience for this test, as a dog that is not obedient is not under control of its owner and putting it
on a leash should reduce the danger of the dog for its environment (and for itself). This measure
was always combined with the measure of education class.
It is interesting to note that the only voluntary test, Test B, is the one with the strictest evaluation
and the one recommending most measures. The other two tests are obligatory for a certain
population of dogs, but evaluate them more leniently.
175
6.4
DISCUSSION OF RESULTS OF CORRELATION WITHIN THE
TESTS / METHODS
6.4.1 BREED
As no results were obtained in this section, no interpretation was possible.
6.4.2 SEX OF DOG
Test C
In this study, females (79%, N = 27) were more likely than males (44%, N = 11) to show ‘neutral
behavior’ towards other dogs in Test C. Males (56%, N = 14) were found more likely to show
‘not neutral behavior’ towards conspecifics than were females (21%, N = 7). However, the group
‘not neutral behavior’ does not only include aggressive behavior, but also ‘fearful, mistrustful
behavior’ and ‘dominant behavior’. This sub-group is too small to be statistically analyzed,
however, it can be described descriptively. Of the 14 males evaluated to show ‘not neutral
behavior’ towards other dogs, 6 (43%) were assessed to react aggressively and 8 (57%)
‘dominantly’ or ‘fearfully, mistrustfully’.
In this test population, 23% (N = 6) of the male dogs versus 6% (N = 2) of the female dogs
reacted aggressively towards the stimulus dogs of Test C. One must keep in mind that upon
request of the experts of this test, all three stimulus dogs were intact males, based on their
hypothesis that aggression is most frequent between intact males.
Several findings concerning this point have been reported in the literature, including the
following: (1) Male dogs are more likely to show aggressive behavior towards other dogs than
are females (Lindsay 2001, Unshelm 1997, Sherman 1996), and (2) According to some authors
intrasex (male-male or female- female) is more common than intersex (male-female) aggression
(see for example, Overall 1997) (3) Other authors report intrasex-intermale aggression being
encountered more often than is either intrasex-interfemale aggression or intersex aggression
(Lindsay 2001, Unshelm 1997, Sherman 1996). The results of this test show a tendency of female
dogs to react more ‘neutrally’ and male dogs to be more likely to behave ‘not neutrally’ towards
other male dogs.
When considering any hypothesis concerning intraspecific aggression, however, care must be
taken to define (1) which sexes (and status) are involved in the aggressive behavior, and (2)
under what circumstances the aggression is shown, i.e. do the dogs belong to the same social
group or not? Do they live in the same household or not? In this study, the intraspecific
aggression towards extra-group, extra-household dogs was tested. The sex was chosen according
to the test’s requirements. Test C was the only one to request only male stimulus dogs. For the
other two both sexes were used. The veterinary behaviorist and the ethologist based their
evaluation on the information gained from the owner and from direct observation of the dog in
interacting with other dogs.
The imprecision of using only males as stimulus dogs in Test C can be underlined by the
anecdotal incidence with the intact female dog that passed the Test And attacked a female spayed
dog a few minutes later (open, bleeding wound requiring stitches and drainage by a veterinarian).
It was well known to the attacking dog’s owner that her dog shows aggressive behavior towards
other female dogs, but as the stimulus dog in Test C was a male, this behavior was not
recognized. This same dog was considered to show ‘dominant behavior or mistrustful, fearful
176
behavior’ in Test A (towards an intact male dog) and to ‘attack with threatening’ in Test B
(confrontation with several dogs of different sexes). The veterinary behaviorist evaluated this
same bitch to ‘attack other dogs without threatening’. The ethologist did not evaluate this dog.
This example shows how important it is for public security to keep the false negative evaluations
to a minimum and emphasizes the importance of the ‘whole picture’ of the dog and the danger of
relying on a momentary picture.
6.4.3 WEIGHT OF DOG
Tests A and C
Both in Tests A and C, dogs weighing 25kg or less were more likely to show ‘neutral behavior’
towards other dogs than dogs weighing more than 25kg. Conversely formulated, dogs weighing
more than 25 kg were evaluated to show ‘not neutral behavior’ more often than dogs weighing
less than 25kg. (See Tables 2.60 and 2.61)
As a correlation was found between the aspects ‘intraspecific aggression’ and sex for Test C (see
section 5.4.2 ‘sex of dog’ and tables 2.60 and 2.61) with more females showing ‘neutral
behavior’ towards other dogs than males, the distribution of the sexes of the dogs within the two
weight groups was looked at more closely. The question was raised whether there were more
females in the group weighing less than 25kg than the other groups. As the females had been
shown to act ‘neutrally’ more often towards other dogs and if more females were found in the
group of dogs weighing less than 25kg, this might be an explanation why more dogs showing
‘neutral behavior’ were found in this group.
A tendency supporting this hypothesis was found in Test C. Within the group of dogs showing
‘neutral behavior’ towards other dogs, twice as many females were found in the group weighing
less than 25kg than were females weighing more than 25kg or males weighing 25 kg or less. In
Test A, not as striking a difference in the distribution of sexes was found as in Test C, with 27%
of the females and 22% of the males weighing less than 25kg showing ‘neutral behavior’ towards
other dogs. However, no correlation between sex and ‘intraspecific aggression’ had been found
within Test A, so the hypothesis made for Test C cannot be given for Test A.
Looking at these results, females weighing less than 25kg are more likely to show ‘neutral
behavior’ in Test C, whereas dogs weighing less than 25kg independent of the sex reacted
‘neutrally’ to other dogs in Test A. Again, emphasis must be put on the fact that the three
stimulus dogs participating in Test C were intact males, whereas the stimulus dogs of Test A
were of both sexes. The same explanation as given in the section ‘sex of dog’ can be brought
here. Assuming the hypothesis that male dogs are more likely to react ‘not neutrally’ towards
other male dogs than towards females, i.e. female dogs are more likely to show ‘neutral behavior’
towards male dogs than males, is correct, then the fact that 3 intact males were used as stimulus
dogs in Test C is a possible explanation for the higher percentage of female dogs reacting
‘neutrally’. Unfortunately, no Test Can be run to statistically underpin this hypothesis.
In summary, the result from this study indicate that a dog that is more likely to show ‘neutral
behavior’ towards other dogs, if it weighs less than 25kg and is a female. However, further
studies in this area are needed in order to be able to give a more definite indication of the
correlation of weight, sex and aggression.
177
Table 2.60 Weight and sex of dogs, Test A
Sex/weight
Test A
neutral
not neutral
Total
neutral
not neutral
Total
f ≤ 25kg
f > 25kg
m ≤ 25 kg
m > 25 kg
missing
16
10
13
6
2
6
2
5
18
16
15
11
0
27%
17%
22%
10%
3%
10%
3%
8%
30%
27%
25%
18%
0%
Total
45
15
60
75%
25%
100%
Table 2.61 Weight and sex of dogs, Test C
Sex/weight
Test C
neutral
not neutral
Total
neutral
not neutral
Total
f ≤ 25kg
f > 25kg
m ≤ 25 kg
m > 25 kg
missing
17
8
8
3
1
6
6
8
18
14
14
11
3
28%
13%
13%
5%
2%
10%
10%
13%
30%
23%
23%
18%
5%
Total
36
21
60
60%
35%
100%
6.4.4 EXPERT
Test A
Expert A1 evaluated a higher percentage of dogs to be in an ‘intact hierarchy’ and to ‘have an
attachment’ than did the other expert. He recommended fewer measures in general and
specifically fewer education classes than Expert A2. Expert A2 recommended measures
(consisting mainly of education classes) to the absolute majority of dogs. Several possible
explanations for the divergences of these results can be given:
(1) A differing familiarity with the test of both experts,
(2) A different educational backgrounds of the two experts,
(3) A differing interpretation of the terminology involved,
(4) A differing degree of importance assigned to certain behavioral traits (such as obedience and
staying close to the owner) by both experts when evaluating canine behavior and choosing the
according measures.
Expert A1 is the expert who routinely evaluates the American Staffordshire Terriers in the
official test of the American Staffordshire Terrier club of Switzerland. This expert is familiar
with the test itself, with its values and with its desired and undesired behaviors. Expert A2 was
not familiar with the test, as she was asked to replace the second expert who usually runs the test
(with expert A1), on short term (one day in advance). The difference in evaluation by the two
experts suggests that the objectivity (independence of the results from the testing expert) of Test.
The experts for Test A have a different educational background, Expert A1 has a longer
experience in evaluating several dog breeds in different tests. Expert A2 has an experience in
agility and is in completing a private training program currently run privately in Switzerland. As
this school puts much emphasis on obedience, it is to be expected that Expert A2 would be more
strict in evaluating the dogs’ obedience leading to a higher percentage of recommendations of
measures, especially education classes, as Expert A1.
178
Test B
Expert B2 recommended measures to all the 16 dogs evaluated by him in this point. The experts
of Test B had split up the evaluating work into two parts, with Expert B2 evaluating all the dogs
in the ‘Ist-Stand’ and continuing with the dogs that did not pass this preliminary test, and Expert
B1 taking over for the actual test. Expert B2 mainly evaluated the dogs that did not pass the ‘IstStand’, meaning that, in the eyes of the expert, the team was not fit to participate in the actual
test. This implies that these owners do not have their dogs ‘under control’ sufficiently to
participate in this specific test which leads to the consequence of recommending measures. With
their background of running a dog training school, it is not surprising that the measure
recommended most frequently was an education class with both experts. Expert B1 evaluated 22
dogs that passed the actual test, increasing the likelihood of these teams not needing any
measures, specifically education class, as a quite high degree of obedience is a prerequisite to
participate in the actual test. Based on the fact that Expert B1 mainly evaluated dogs that passed
the Test And Expert B2 those that failed, it is more likely that a higher percentage of dogs
showing ‘neutral behavior’ towards other dogs would be amongst the dogs tested by Expert B1.
Hence, no comment can be made about the objectivity of the Test As the populations of dogs
evaluated by both experts differ in their degree of obedience.
6.4.5 TEST SEQUENCE
No statistical Test could be run, on the test sequence due to the following reasons: (1) The
rescheduling of 15 dogs resulted in a diminution of one group (test sequence B-A-C) to five dogs
and an increase of another (test sequence B-C-A) to 15 individuals. The category with only five
dogs decreases the possibility of having enough individuals in every group of expected values to
run a Chi-square test. (2) The homogeneity of the population, with few dogs showing ‘not neutral
behavior’, resulted in more than 20% of the groups consisting of less than five individuals, which
makes the Chi-test impossible. (3) The impossibility of reducing the test sequence groups into
groups that would allow a Fisher’s Exact Test to be run.
6.4.6 STIMULUS DOG
No statistically significant correlation was found with the stimulus dog.
6.4.7 ‘HIERARCHY’
As the term ‘hierarchy’ was not precisely defined in advance, the meaning of this word might
differ for each expert. A correlation was found between the aspect ‘hierarchy’ and the point
‘attachment’ in Test A. If ‘attachment’ is defined in the popular sense as ‘closeness to the owner’
and ‘hierarchy’ as ‘control/dominance over the other’, a correlation of the two points seems
logical. The dog can easier be controlled/dominated (i.e. ‘hierarchy intact’) by the owner if it
stays close to him/her (i.e. ‘attachment’), if it is the dog looking where the owner is and not vice
versa.
On the one hand, hierarchical problems are often considered to have an influence on behavioral
problems, including aggression, in dogs. On the other hand, hierarchy and obedience are often
not very well distinguished in the popular sense, hierarchical problems often being treated with
obedience training. However, a dog may obey perfectly, but may still show dominant behaviors –
one does not exclude the other. The chances that ‘hierarchy’ and ‘obedience’ are often confused,
is shown also by the results of this study where both in Tests A and C, significantly fewer dogs
179
that were considered to be in an ‘intact hierarchy’ with their owners received measures than dogs
that were judged not to be in an ‘intact hierarchy’. Recommendation of education classes was
made significantly less often to the same group of dogs than to dogs being evaluated to be
‘dominant over their owners’.
When considering the results of the veterinary behaviorist and the ethologist, one must, again,
keep in mind that the hierarchy is evaluated under completely different circumstances, i.e. within
the system the dog lives in, as opposed to the momentary picture of an interaction of two
individuals, which is the basis for the test evaluation. This makes a comparison of the two
approaches (tests versus ‘behavioral experts’) difficult. The fact that no correlation between the
aspect of ‘hierarchy’ and the points ‘attachment’ and ‘measures’ was found with the veterinary
behaviorist and the ethologist might be explained by their already discussed differing definitions
of the term. A system where a dog considers itself higher in the hierarchy than its owner may
lead to problems between the dog and the owner. Dehasse (2002) defines ‘dominance’ as the dog
having access to privileges and defending them towards one or several persons of the same
household. Depending on the circumstances, these dogs may show dominant behavior or
mistrustful behavior towards unknown humans or even threaten or attack.
This aspect cannot be interpreted for Test B, as only 14 dogs received an answer to this question
and only one of these was considered ‘dominant over its owner’
6.4.8 ‘ATTACHMENT’
As was the case with the term ‘hierarchy’, the term ‘attachment’ was not precisely defined in
advance.
The fact of ‘having an attachment’ was found to be correlated with showing ‘neutral behavior’
towards humans (Test A) and towards dogs (Test B).
Beata (2002 65) says that attachment ‘will allow important learning, such as self-control, social
communication and knowledge of different stimuli of the environment. Without attachment, there
is no possible exploration.’ Having an ‘attachment to the owner’ might be one important factor in
the sensation of security for a dog. If this attachment is weak or not existent, the insecurity of the
dog may increase, and the animal may be expected to either react more fearfully and mistrustfully
towards unknown humans or to show ‘dominant behaviors’.
The fact that having ‘an attachment’ reduces the likelihood of recommended measures can be
explained based on this hypothesis as well. A lack of attachment can lead to insecurity of the dog,
but also the owner, i.e. the team has to work on their relationship. This can be accomplished
either by doing something together, such as participating in an education class (the measure
chosen by Test A) or the dog can be controlled by physical means, such as being held on a leash
at all times (which is the recommendation Test B gave for these dogs). The obligatory leash is
more of a symptomatic treatment whereas the education class offers the possibility of actually
achieving a change in behavior. The option of going into a therapy exists as well, but was not
correlated with this category, as only few dogs received this measure.
180
6.4.9 MEASURES
Dogs showing ‘neutral behavior’ towards humans (Tests A and B) or other dogs (Test C) are less
likely to receive measures. A ‘neutral behavior’ is seldom a problem in our society, except for
dogs with certain tasks, such as police dogs or rescue dogs. However, even these dogs are
expected to stay neutral as long as they do not receive the command to act. Measures include the
aim to change the behavior of the dog, the owner, both, the situation and/or the surroundings.
Clearly, if there is no problem, no changes are necessary, i.e. no measures need to be taken. The
probability of a ‘mistrustful, fearful behavior’ of developing into an aggressive, potentially
dangerous behavior is more likely. It makes sense to address this problem as soon as possible and
to protect the dog’s surroundings from a behavior that might be potentially dangerous.
Depending on the severity of the behavior (slightly mistrustful or fearful to threatening behavior
to overt attack), different measures may be recommended, varying from physical constraint, such
as an obligatory leash (as recommended in Test B) or a muzzle (as suggested by the veterinary
behaviorist) or an education class (advised by Tests A and B). Therapies were recommended
rarely by the experts of the tests, perhaps due to a bias stemming from their educational
background. Most experts do primarily pragmatic work with the dog, such as obedience training
or special training of dogs (e.g. ‘Schutzdienst’). Recommending physical constraints to a dog
showing mistrustful, fearful behavior or threatening or attacking behavior towards other living
beings makes sense as a first measure of security, however, it does not influence the actual
problem behavior. In order to achieve a change in mood and behavior of the dog, any therapy is
best combined with other measures where the dog learns new strategies such as education class
(as recommended by Test B) or behavioral therapy).
Measures are mainly correlated with ‘interspecific aggression’ towards humans (Test A, Test B
and the veterinary behaviorist) pointing at the importance of public security.
The correlation with ‘intraspecific aggression’ found in Test B with a suggested obligatory leash
as a physical constraint and in Test C with a generally increased likelihood of receiving measures
indicates that at the problem of security of other dogs was also a concern.
6.4.10 DISCUSSION OF CORRELATION OF AGGRESSION
Correlations were only found in Tests A and B. In both, dogs showing ‘neutral behavior’ towards
other dogs were more likely also to show ‘neutral behavior’ towards humans and moving objects.
This implies that dogs showing a neutral behavior towards one ‘trigger’-category might be in a
more equilibrated general emotional state and thereby less likely to be ‘aroused’ by any stimulus.
Within Test A, all the categories are correlated in the sense that if a dog was evaluated to show
‘neutral behavior’ towards one of the ‘targets’, it was also more likely to show ‘neutral behavior’
towards the others. This leads to the hypothesis that it might be possible to give a dog an ‘overall
reaction score’.
Dogs showing ‘not neutral’ behavior seem to be in a higher ‘emotional/arousal state’, suggesting
that they do not react neutrally to a variety of stimuli. While Test A is the only test to support this
hypothesis, it would be worthwhile looking into this aspect more closely.
Dogs showing ‘neutral behavior’ towards other animals are more likely to also ‘behave neutrally’
towards conspecifics, humans and moving objects. This makes sense, as other dogs and people
fall under the category of ‘other animals’ and the moving objects consisted of a human riding a
bicycle or jogging past the dog.
181
6.5
DISCUSSION OF THE RESULTS OF THE COMPARISON OF
AGGRESSIVENESS AND DANGEROUSNESS WITHIN THE
TESTS
A dog showing aggressive behavior is obviously more likely to be dangerous than a dog showing
‘not aggressive behavior’. However, several other factors influence the notion of dangerousness
as well such as the context in which the behavior is shown, the mass of the dog, the control the
owner has over his/her dog, etc. (Pillonel, 2000, Lindsay, 2001, Dehasse, 2002, Lehmann, 2003).
A dog that does not threaten and does not show any other signs of aggression may be considered
less dangerous than a dog that threatens, simply by the level of emotional state the dog is in.
However, a dog that threatens before considering an attack is certainly less dangerous than a dog
that attacks without threatening, i.e. is unpredictable. Here, the notion of pathology has to be
considered, which frequently goes beyond the focus or possibilities of a test.
It is interesting to note that no correlation was found between any category of aggression with
dangerousness for the ethologist and only in two aspects for the veterinary behaviorist. This leads
to the hypothesis that these approaches, which are based more on evaluating the system and
history of the dog by questioning the owners, might allow access to and consider more
information than can be extracted by watching the dog in a test during a given period of time.
This allows the behavioral specialists to receive more information about additional factors
contributing to the dangerousness of the dogs (such as past incidences, or the circumstances of an
incident and the weight and age of the victim), which are not considered in the evaluation sheet
of this project. This underlines the assumption that aggression is not the only factor to influence
the degree of dangerousness of a dog, but that many other factors are involved as well.
Another explanation might be the differing pooling of the aggression scale for the ethologist and
veterinary behaviorist on the one hand (‘aggressive behavior’ vs. ‘not aggressive behavior’) and
the tests on the other (‘neutral behavior’ and ‘not neutral behavior’). Additionally, these two
specialists only evaluated 12 dogs each, which makes statistical tests more difficult.
The two categories considered in the behavior scale of the tests are ‘neutral behavior’ and ‘not
neutral behavior’, whereas the distinction ‘not aggressive behavior’ versus ‘aggressive behavior’
was only possible for the veterinary behaviorist and the ethologist. It does make sense though,
that ‘neutral behavior’ would be correlated with ‘not being dangerous’ more than ‘not neutral’
behavior would be, no matter what the trigger / target of the behavior is. The category ‘not
neutral behavior’ includes ‘dominant behavior’ and ‘mistrustful, fearful behavior’ as well as
aggressive behaviors. These are not directly related to the present degree ‘of dangerousness’, but
dogs showing these behaviors are more likely to develop dangerous aggressive tendencies than
dogs reacting ‘neutrally’, as the factor of insecurity and fear can lead to fear aggression and
distancing aggressions. If the owner does not have an uncertain, fearful dog under control, the
potential danger for the surrounding increases.
A correlation of the points concerning the safety of humans was found in all three tests and the
group of dogs evaluated by the veterinary behaviorist, no matter whether the dog was in the
presence or absence of the owner. Quite unsurprisingly, the likelihood of a dog being evaluated
as ‘not dangerous for humans’ is larger if it shows (or is said to show by the owner) ‘neutral
behavior’ towards humans, than if it shows signs of fearful, mistrustful, dominant or aggressive
behaviors. A consistency between the evaluation of the dogs’ behavior towards other dogs and
the judgement of dangerousness for other dogs is only given in two cases (Tests B and C). A
possible explanation might be that other factors influencing the potential dangerousness play a
more important role in this type of aggression than in others, such as the dogs’ sex and size. In
the only test (Test B) that evaluated the dogs’ reactions to other animals, a consistency in the
answers of the behavior and the evaluation of the dangerousness was found, whereas there was
182
no correlation within the groups of dogs evaluated by the veterinary behaviorist and ethologist.
This might again be caused by the additional information the latter have through owner
questioning and that the evaluation of the dangerousness not being only on the dog’s behavior in
the actual situation. For example, the veterinarian or ethologist might learn that the dog is always
held on the leash when in situations where other animals might be encountered, hence, there not
being much of a danger, whereas the dogs in Test B were required to be off the leash in these
situations, which increases the potential dangerousness for the other animals simply because the
dog is not physically constrained. Also, in the test situation, the dogs’ reactions were only tested
towards goats and chicken, whereas the veterinarian and the ethologist had information about a
larger variety of animals, including cats and birds. A dog might be likely to chase a cat, but will
not react to a goat or chicken. This behavior, as well as the behavior towards moving objects may
both be considered as predatory behavior, which then raises the question of whether this behavior
is actually ‘aggression’ or not. Several schools do not view predatory behavior as a type of
aggression, as the sequence of aggressive phases is not intact, the dog does not usually threaten
its prey before attacking it – other schools would see this as a type of aggressive behavior (Pageat
1998, Dehasse 2002). Again, the different pooling of the aggression scale has to be taken into
account. In Test B, the likelihood of a dog showing ‘neutral behavior’ towards other animals was
more likely to be evaluated ‘not dangerous’ for other animals. When looking at the results of the
veterinary behaviorist, it is evident that there are no dogs that reacted ‘not aggressively’ and that
all these dogs were considered to be ‘dangerous’ for other animal (but, again, a statistical test
cannot be run with only two groups).
6.6
DISCUSSION OF COMPARISON OF THE APPROACHES USING
A REFERENCE VALUE
For several of the results the test for sensitivity and specificity could be run. Often, a high
specificity up to 100% was found.
The sensitivity however, is quite low, meaning that there are many dogs that were evaluated to be
aggressive by behavioral specialists but not by the tests. Again the behavioral specialists rely on
the information that they receive from the owners, who, one must assume, have observed some
kind of troublesome behavior, which was not picked up by the test evaluations.
Caution must be taken when interpreting the results of sensitivity, as the upper and lower limits
lie quite far apart. The main problems are the small number of dogs (N = 12) that were included
in this test and the fact that the variety of behaviors shown by the population is quite small.
The results obtained in this small study, however, do show a tendency towards a less sensitivity
of the tests in reference to the evaluations of the behavioral specialists, which is of importance, as
the recommendations resulting from these tests may have consequences for public security.
183
7. CONCLUSIONS
Aggression is a fuzzy concept with often tangible consequences. Fuzzy, because it is not easily
defined, because there is no agreement in the literature on what aggression actually is and how it
should be classified. Its definition and classification varies depending on the school of thought
and on the individual author. The overview of the books of several authors in the first part of this
study shows the extent of differences and similarities of a representative sample of these
approaches.
On the other hand, aggression has tangible consequences in that it may become a problem for the
security of the general public, and clinicians are faced with it almost daily. Aggression is a way
of communication that should be taken seriously as it is an important factor in the evaluation of
the dangerousness of an animal.
What is important for the practitioner, however, is not so much the forest of definitions, but
rather the handling of aggressive problems in every day life. Trainers and behavioral therapists
use several educative and therapeutic techniques, including behavioral therapy, cognitive therapy,
systemic therapy, strategic therapy, mechanical devices and medication. Whatever the underlying
model and approach used, in the end, the therapies and the medications are quite similar, with
individual preferences depending on the educational background and convictions of the
individual.
Many tests have been developed with many different aims. The three tests included in this study
all include the aim of uncovering dogs that might be a potential danger to the general public and
the families with which they live. The great majority of tests in use is not scientifically validated,
which means that one does not know if they actually are testing what they claim to be testing.
The question remains to be discussed to which extent it is possible to validate a test assessing
behavior of living beings, as it is impossible to standardize or control every factor that has a
possible influence on the outcome or to completely objectify the evaluation. Our aim in this study
was not to validate the tests, but to run a preliminary study of comparison of the approaches. In
the process of the study, both the benefits and difficulties of validation became clear. We hope
this study might lead to the reconsideration of several existing tests in use and perhaps to the
validation of some of them.
The comparison of the tests in the Kappa-test show an agreement higher than randomly expected,
with Kappa-values varying between 0.133 and 0.352 in the aspects of ‘intraspecific aggression
towards dogs’, ‘interspecific aggression towards humans’ and ‘attachment’. These Kappa values
are far from the agreement of 1, which would be the ideal, but probably impossible to attain.
They are, however, significant, which is a starting point for a valuable evaluation of the three
tests. While it is not possible to completely standardize a test, there are possibilities of increasing
the standardization and decreasing the imprecision of some factors influencing the results, such
as standardizing the number of helpers participating, always using the same experts and stimulus
dogs, etc.
Clearly there is no absolute reference value that tells us how the dog really behaves. Given this, it
seems to this author, that one can come most closely to this value through approaches which
include the life-history of the dog and the experiences of the owner (such as are part of the
veterinary behaviorist and the ethologist approaches). Of course, these methods are not validated
either, but they do allow for the possibility of uncovering ‘false negative’ evaluations which
might be given by the tests, i.e. dogs that were evaluated to be ‘not aggressive’ in the tests, but,
according to the owner’s observations and experiences, do act aggressively in certain
circumstances. These ‘false negatives’ are of interest for public security. The ‘false positives’, i.e.
184
the dogs that were evaluated to be aggressive by the tests and not by the behavioral specialists are
also of interest for public security, as these dogs obviously did react aggressively in the tests, i.e.
they have a tendency to show aggressive behavior at least in the test situation. The tests have a
tendency to a relatively high specificity of 50 to 100%, and a generally low sensitivity of 14 to
100%. Further studies are recommended to investigate these tendencies, as the false positives
affect dog welfare and the false negatives have implications for public security. A low sensitivity
implies a larger number of false negatives, i.e. dogs that are not recognized as being aggressive in
the tests. A high specificity means that the tests and the behavior specialists mainly agree on the
dogs that are not considered not to be aggressive. The false positives in this case are of interest as
well, as these are dogs that obviously did react aggressively in a test situation – and were not
recognized as this by the behavioral specialists. The validation of the methods would be of help
here, allowing us to interpret the information we get from a dog showing aggressive behavior
during a test situation as being representative of the dog’s general behavior in non-test situations.
185
8. REFERENCES
American Staffordshire Terrier Club Schweiz (2000) Reglement für die obligatorische
Verhaltensüberprüfung des ASTC – Schweiz, 30.7. 2000
Abrantes, R.A.B. (1997) Dog Language: An Encyclopaedia of canine behaviour.Walkan Janka
Publisher, Illinois
Arbeitsgruppe “Gesetzgebung betr. gefährliche Hunde” (2001) Empfehlungen der Arbeitsgruppe
“Gesetzgebung betr. gefährliche Hunde” des Bundesamts für Veterinärwesen
Arbeitsgruppe “Gesetzgebung betr. Gefährliche Hunde” (2000) Bericht über das Hearing zum
Thema “Kampfhunde”, 31.8.2000
The American Psychological Association (Ed.) (1985) Standards for Educational and
Psychological Testing, Washington, D.C. The American Psychological Association
American Staffordshire Terrier Club, Schweiz Reglement des American Staffordshire Terrier
Clubs, Schweiz
American Veterinary Medical Association, Task Force on Canine Aggression and Human-Canine
Interactions (2001) A community approach to dog bite prevention, Journal of the
American Veterinary Medical Association, 218, 1732-1747
Anderson, G., Marinier, S. (1997) The effect of food and restricted exercise on behaviour
problems in dogs, Proceedings of the First International Conference on Veterinary
Behavioural Medicine, Birmingham, UK, 183-187
Archer, J. (1988) The behavioural biology of aggression, Cambridge University Press
Aronson, L.P., Dodman, N.H. (1997) Thyroid Dysfunction as a cause of aggression in dogs and
cats, 43. Jahrestagung der Deutschen Veterinärmedizinischen Gesellschaft Fachgruppe
Kleintierkrankheiten, Hannover
Askew, H.R. (1996) Treatment of behavioural problems in dogs and cats a guide for the small
animal veterinarian, Blackwell scientific publications, Oxford
Barlow, T.A., Bradshaw, J.W.S., Casey, R.A. (2001) Hypothyroidism and behavioural change in
the domestic dog Canis familiaris, In: Overall, K.L., Mills, D.S., Heath, S.E., Horwitz, D.
(eds) Proceedings of the Third International Congress on Veterinary Behavioural Medicine,
Vancouver, 130-132
Barrows, E.M. (2001) Animal Behavior Desk Reference, a Dictionary of Animal Behavior,
Ecology, and Evolution, Library of congress, CRC Press LLC, Boca-Raton, Florida
Beata, C. (2002) Attachment: New skin for an old theory, Proceedings of the 8th Meeting on
Veterinary Behavioural Medicine, Granada, Spain, 65-69
Beaver, B.V. (1994) The Veterinarian’s Encyclopedia of Animal Behavior, Iowa State University
Press, Ames, Iowa
Beaver, B.V. (1993) Profiles of dogs presented for aggression., Journal of the American Animal
Hospital Association, 29, 564-569
186
Berendt, M. (2002) Epilepsy, In: Clinical Neurology in Small ANimals - Localization, diagnosis
and treatment, K.G. Braund (Ed.), International Veterinary Information Service
(www.ivis.org), Ithaca, New York, USA
Borchelt, P.L., Voith, V.L. (1985) Aggressive behavior in dogs and cats, Compendium of
Continuing Education for the Practicing Veterinarian, 7 (11), 949-957
Borchelt, P.L., Voith, V.L. (1982) Classification of animal behavior problems, Veterinary Clinics
of North America and Small Animal Practice, 12 (4) 571-585
Brain, P.F. (2001) The Biology of Aggression-Dispelling the Myths, Proceedings of the 2001
CABTSG Annual Study Day, Birmingham “Aggression Dispelling the Myths”, 1-4
Brain, P.F. (1981) Differentiating types of attack and defense in rodents. In Multidisciplinary
Approaches to Aggression Research, eds. Brain PF, Benton D. Elsevier / North Holland,
Amsterdam, 53-78
Braund, K.G. (2003a) Storage Disorders, In: Clinical Neurology in Small Animals - localization,
diagnosis and treatment, Braund, K.G. (Ed.), International veterinary information service,
Ithaca, NY (www.ivis.org)
Braund, K.G. (2003b) Endogenous metabolic disorders, In: Clinical Neurology in Small Animals
- localization, diagnosis and treatment, Braund, K.G. (Ed.), International veterinary
information service, Ithaca, NY (www.ivis.org)
Braund, K.G. (2003c) Neurotoxic disorders, In: Clinical Neurology in Small Animals localization, diagnosis and treatment, Braund, K.G. (Ed.), International veterinary
information service, Ithaca, NY (www.ivis.org)
Braund, K.G. (2003d) Neurological Syndromes, In: Clinical Neurology in Small Animals localization, diagnosis and treatment, Braund, K.G. (Ed.), International veterinary
information service, Ithaca, NY (www.ivis.org)
Braund, K.G. (2003e) Inflammatory Diseases of the Central Nervous System, In: Clinical
Neurology in Small Animals - localization, diagnosis and treatment, Braund, K.G. (Ed.),
International veterinary information service, Ithaca, NY (www.ivis.org)
Campbell, W.E., 1999, Behaviour Problems in Dogs, BehavioRx® Systems, Grants Pass, Oregon
Chapman, B.L., Voith, V.L. (1990) Behavioral problem of old dogs: 26 cases (1984-1987),
Journal of the American Veterinary Medical Association, 196, 944-946
Dehasse, J, Bräm, M, Schroll, S. (2003a) Aggressive behaviors in dogs, a new descriptivecontextual classification, Proceedings of the 4th International Veterinary Behaviour
Meeting, Caloundra, Australia (unpublished)
Dehasse, J. (2003b) 2. Le développement comportemental des chiens et chats et leurs troubles.
Conférence donnée à La Louvière (Belgique), le 12 février 2003.
Dehasse, J. (2002a, 2nd edition) L’éducation du chien, Le Jour, éditeur
Dehasse, J. (2002b) Le chien agressif, Publibook, Paris
Dehasse, J. (2000a) Mon chien est-il dominant? Le Jour, éditeur
Dehasse, J. (2000b) Mon jeune chien a des problèmes, Le Jour, éditeur
187
Dehasse, J. (1999) Dog aggressive behaviour disorders and dominance aggression, I9th Annual
general Meeting, 1st ASEAN Small Animal Scientific Congress - Small Animal
Practitioners Association of Malaysia - Selangor Darul Ehsan (Kuala Lumpur)
DeNapoli, J.S., Dodman, N.H., Shuster, L., Rand, W.M., Gross, K.L (2000) Effect of dietary
protein content and tryptophan supplementation on dominance aggression, territorial
aggression, and hyperactivity in dogs, Journal of the American Veterinary Medicine
Association, 217 (4), 504-408
Dobbert, M.L. (1982) Ethnographic research: Theory and Application for Modern Schools and
Societies, Praeger Publishers, CBS Educational and Professional Publishing, a Division of
CBS Inc., New York
Dodds, W.J., Aronson, L.P. (1999) Behavioral changes associated with thyroid dysfunction in
dogs, Proceedings of the 1999 American Holistic Veterinary Medical Association Annual
Conference, 80-82
Dodman, N.H., Reisner, I., Shuster, L., Rand, W., Leuscher, U.A., Robinson, I., Houpt, K.A.
(1996) The effect of dietary protein content on behavior in dogs, Journal of the American
Veterinary Medical Association, 208, 376-379
Donaldson, J. (1996) The culture clash. A revolutionary new way of understanding the
relationship between humans and domestic dogs, James&Kenneth Publishers, USA,
Canada, UK
Doré François Y. (1983) L'Apprentissage, une Approche Psycho-Éthologique, Chenelière &
Stanké, Montréal & Maloine, Paris
Eibl-Eibesfeldt, I. (1984) Ethologie, biologie du comportement, Ophrys Ed. Scientif. Naturalia et
Biologica, Paris
Eichelberg, H. (2000) Kampfhunde-Gefährliche Hunde, Deutsche Tierärtzliche Wochenschrift,
107 (3), 91-93
Ettinger, S.J., Feldman, E.C. (Eds) (1995, 4th edition) Textbook of veterinary internal medicine,
W.B. Saunders company, Philadelphia
Feddersen-Petersen, D.U. (2002) Biologie der Aggression / Biologie des Lernens, Internationaler
Kongress, 100 Jahre Schweizerischer Collie-Club / Collie-Club Suisse, Wettingen, Schweiz
Feddersen-Petersen, D.U. (2001a) Hunde und ihre Menschen, Franckh-Kosmos Verlags-GmbH
Feddersen-Petersen, D.U. (2001b) Zur Biologie der Aggression beim Haushund, Deutsche
Tierärztliche Wochenschrift, 108, 94-101
Feddersen-Petersen, D.U. (2000a) Hunde in Berlin, Redebeitrag zur Anhörung der
Bundestagsfraktion Bündnis 99/Die Grünen vom 21.8.2000,
www.hundepruefung.de/texte/feddersen_petersen.htm,
Feddersen-Petersen, D.U. (2000b, 3. Auflage) Hundepsychologie. Wesen und Sozialverhalten,
Franck-Kosmos, Verlangs-GmbH & Co., Stuttgart
Feddersen-Petersen, D.U. (1991) Aggressive Hunde ein Tierschutzproblem, Schutz des Tieres
vor Missbrauch durch den Menschen bedeutet Menschenschutz, Tierärztliche Umschau, 46,
749-754
Feddersen-Petersen, D.U. (1990) Verhalten der Hunde, Deutsche tierärztliche Wochenschrift, 97,
231-236
188
Gebers, A.M., Fuchs, H., Barnabe, V.H. (2002) Partial results of ovariohysterectomy in female
dogs from the municipality of Sao Paulo, Brasil, Free Communications Book of the 27th
WSAVA Congress, Granada, Spain, 161
Grayling, A.C. (2002) The Reason of Things, Weidenfeld & Nicolson, The Orion Publishing
Group Ltd, London
Hamilton Andrews, S., McBride, E.A., Brown, I. (1999) Hypothyroidism and aberrant
behaviours in the Bearded Collie, The Second World Meeting on Ethology, Lyon, 133-138
Hart, B.L., Neilson, J.C., Ruehl, W.W. (1997) Behavioural Changes in Ageing Dogs A
Demographic Analysis, Proceedings of the First International Conference on Veterinary
Behavioural Medicine, Birmingham, UK, 31-33
Hewson, C.J., Luescher, U.A., Ball, R.O (1999) Epidemiological and psychometric issues in
clinical behavioural research: a review with examples, The Second World Meeting on
Ethology, Lyon , 44-54
Heymer, A. (1977) Vocabulaire éthologique, Ethological Dictionary, Ethologisches Wörterbuch,
Presses Universitaires de France, Paris.Presses Universitaires de France, Paris
Hirsig, R. (1993) Methodische Grundlagen der Testpsychologie, Skriptum zur Lehrveranstaltung
des Psychologischen Instituts der Universität Zürich
Horisberger, U. (2002) Medizinisch versorgte Hundebissverletzungen in der Schweiz Opfer –
Hunde – Unfallsituationen, Inaugural-Dissertation zur Erlangung der Doktorwürde der
Veterinär-medizinischen Fakultät der Universität Bern, Schweiz
Houpt, K.A., Wolski, T.R. (1998) Domestic Animal Behavior for Veterinarians and Animal
Scientists, The Iowa State University Press, Ames, Iowa
Immelmann, K. (1983) Einführung in die Verhaltensforschung, 3. Auflage, Verlag Paul Parey,
Berlin und Hamburg
Immelmann, K., Beer, C. (1989) A Dictionary of Ethology, Harvard University Press
Kantonales Veterinäramt Basel-Stadt (2001) Pressekonferenz “Potentiell gefährliche Hunde:
Erfahrungen mit der revidierten Hundegesetzgebung”, Montag 1. Oktober 2001
Landrivon, G. (2002) Méthode globale de lecture critique d’articles médicaux à l’usage de
l’étudiant et du practicien, Editions Frisons-Roche, Paris
Landsberg, G., Hunthausen, W., Ackerman, L. (1997) Handbook of behavior problems of the dog
and cat, Oxford, Butterworth-Heinemann
Lehmann, D. (2003) Comparaison de cinq méthodes d’évaluation de la dangerosité potentielle
chez le chien, Thèse inaugurale pour l’obtention du titre du Docteur de la Faculté de
médicine vétérinaire de l’Université de Berne, Suisse
Lindsay, S.R. (2000) Handbook of Applied Dog Behavior and Training: Volume One,
Adaptation and Learning, Iowa State University Press, Iowa
Lindsay, S.R. (2001) Handbook of Applied Dog Behavior and Training: Volume Two, Etiology
and Assessment of Behavior Problems, Iowa State University Press, Iowa
Mac Farland D (ed) (1990) Dictionnaire du comportement animal, Robert Laffont, Paris
Mertens, P. (2001) Ethogram, Weiterbildungszyklus Verhaltensmedizin Teil I Gefährliche
Hunde von der Fallaufnahme zum Bericht, 14-25
189
Moles, A.A. (1995) Les sciences de l’imprécis, Editions du Seuil
Morgan, T.A., Bradshaw, J.W.S., Casey, R.A., Thomas, A.L. (1999) Hypothyroidism and
behavioural change in the domestic dog canis familiaris, The Second World Meeting on
Ethology, Lyon, 130-132
Moyer, K.E. (1968) Kinds of aggression and their physiological basis, Communication Behav
Beiol, 2 (A) 65-87
Neilson, J.C., Eckstein, R.A., Hart, B.L. (1997a) Effects of castration on problem behaviors in
male dogs with reference to age and duration of behavior, Journal of the American
Veterinary Medical Association, 211 (2), 180-182
Niedersächsisches Ministerium für Ernährung, Landwirtschaft und Forsten (2001) Wesenstest für
Hunde, Zusammenfassung der Referate des Vets2001 Kontresses, Fribourg, Schweiz, I XXIX
Odendaal, J.S.J. (1997) A diagnostic classification of problem behavior in dogs and cats,
Veterinary Clinics of North America Small Animal Practice, 27 (3), 427-443
O’Farrell, V. (1991) Verhaltensstörungen beim Hund, Ein Leitfaden für Tierärzte, Verlag M. &
H. Schaper GmbH & Co., KG, Alfeld, Deutschland
O'Farrell, V., Peachey, E. (1990) Behavioural effects of ovariohysterectomy on bitches, Journal
of Small Animal Practice, 31, 595-598
Overall, K.L. (1997) Clinical Behavioral Medicine for Small Animals, Mosby Inc, Missouri
Oxford Universal Dictionary (1955, 3rd edition), Oxford University Press, Amen House, London
Pageat, P. (1998) Pathologie du comportement du chien, Editions du Point Vétérinaire, Maisons
Alfort
Pillonel, C. (2001) Aggression und Gefährlichkeit, Zusammenfassung der Referate des Vets2001
Kontresses, Fribourg, Schweiz, 105-109
Planta, D.J.U., Netto, W.J. (1999) Behavioural testing for aggression in the domestic dog, The
Second World Meetingon Ethology, Lyon, 18-27
Planta, D.J.U. (2001) Testing dogs for aggressive biting behaviour the MAG-test (sociable
acceptable behaviour test) as an alternative for the Aggression-test, In Overall, K.L.,
Mills, D.S., Heath, S.E., Horwitz, D. (eds) Proceedings of the Third International Congress
on Veterinary Behavioural Medicine, Vancouver, 142-144
Podberscek, A.L. (1996) The English Cocker Spaniels preliminary findings on aggressive
behaviour, Applied Animal Behaviour Science, 47, (1-2) 75-89
Roll, A, Unshelm, J. (1997) Aggressive conflicts amongst dogs and factors affecting them,
Applied Animal Behaviour Science, 52, 229-242
Ruefenacht, S., Gebhardt-Henrich, S., Miyake, T., Gaillard, C (2002) A behaviour test on
German Shepherd dogs heritability of seven different traits, Applied Animal Behaviour
Science, 79 (2), 113-132
Schöning, B. (2001) Dangerous dogs in Germany - And how to define “normal aggression”?, In
Overall, K.L., Mills, D.S., Heath, S.E., Horwitz, D. (eds) Proceedings of the Third
International Congress on Veterinary Behavioural Medicine, Vancouver, 96.99
190
Scott, S. (2000) The role of pain in canine behaviour problems, Proceedings of the 2000
CABTSG annual study day, Birmingham, 37-38
Shermann, C.K., Reisner, I.R., Taliafaro, L.A. et al, Characteristics, treatment and outcome of 99
cases of aggression between dogs, Applied Animal Behavioral Science, 47, 91-108
Soderstrom, H., Blennow, K, Manhem, A., Forsman, A. (2001) CSF studies in violent offenders
I. 5-HIAA as a negative and HVA as a postiive predictor of psychopathy, 108 (7), 869-879
Straubinger, R, K. (2000) Lyme Borreliosis in Dogs, In Recent Advances in Canine Infectious
Deseases, Carmichael L. (Ed.), International Veterinary Information Service, Ithaca, NY
(www.ivis.org)
Stur, I. (1998) Zur Frage der besonderen Gefährlichkeit von Hunden aufgrund der Zugehörigkeit
zu bestimmten Rassen, Auszug aus einem Gutachten vom Februar 1998 zuhanden der
steiermärkischen Landesregierung, www.hund-hundhalter.de/arbeitpapiere/material/seiten/arb-pap-003.html
Stur, I. (1987) Genetic aspects of temperament and behaviour in dogs, Journal of Small Animal
Practice, 28, 957-964
Thrusfield M. (1995) Veterinary Epidemiology, 2nd Edition, Blackwell Science
Tilley, L.P., Smith, F.W.K. (1997) The 5 Minute Veterinary Consult: Canine and Feline,
Williams & Wilkins, Baltimore
Van Erp, A.M., Miczek, K.A. (2000) Aggressive Behavior, Increased Accumbal Dopamine, and
Decreased Cortical Serotonin in Rats, J Neurosci, 20 (24), 9320-9325
Winter, G. (2000) A Comparative Discussion of the Notion of 'Validity' in Qualitative and
Quantitative Research, The Quality Report, 4, 3-4
Wright, J.C., Nesselrote, M.S. (1987) Classification of behavior problems in dogs distributions of
age, breed, sex and reproducitve status, Applied Animal Behaviour Science, 19, 169-178
www.wolfsprung-kennels.ch, May 17, 2003
www.veterinaeramt.bs.ch/Hunde/potentiell_gefaehrliche_hunde.htm
www.ml.niedersachsen.de/funcions/downloadObject/0,,c1107366_s20,00.pdf, January 8, 2003
Young, M.S. (392) Aggressive behavior. In Clinical Signs and Diagnosis in Small Animal
Practice, ed. Ford RB, Churchill Livingstone, New York, 135-150
Zimen, E. (1992) Der Hund, Abstammung - Verhalten - Mensch und Hund, Goldmann Verlag,
München
191
9. APPENDIX
192
10. TABLE OF CONTENTS
ABSTRACT
1
ZUSAMMENFASSUNG
2
INTRODUCTION
3
1.
MOTIVATION FOR THIS STUDY
2
2.
FOCUS OF THE DATA ANALYSIS
2
3.
STRUCTURE OF THIS REPORT
3
4.
UNDERLYING QUESTIONS
3
PART I: BEHAVIOR AND AGGRESSION IN DOGS: A THEORETICAL OVERVIEW 6
1.
INTRODUCTION AND AIMS OF THE THEORETICAL PART
5
2.
LITERATURE SOURCES
6
3.
DEFINING BEHAVIOR
7
3.1
Normal Behavior
8
3.2
Behavior of wolves and dogs
9
3.3
Not Normal Behavior
9
4.
AGGRESSION
10
4.1
Definitions of Aggression
10
4.2
What is aggression?
11
4.3
Aggression and Dangerousness
13
4.4
Influencing Factors
4.4.1 Individual factors
4.4.1.1 Genetics
4.4.1.2 Hormones
4.4.1.3 Neurobiochemistry
4.4.1.4 Age
14
14
14
16
17
17
193
4.4.2
Illnesses /physical ailments
Pain
Sensory deficiencies
Infection
Metabolism
Neurological
Learning
Emotional state - mood
4.4.3 Environmental factors
4.4.3.1 The Mother Dog
4.4.3.2 The siblings
4.4.3.3 Socialization
4.4.3.4 Communication and Social Structure
4.4.3.5 Obedience and education
4.4.3.6 Medications
4.4.3.7 Ecology and ethology
4.4.3.8 Nutrition
4.4.3.9 Context of a Situation
4.4.2.1
4.4.2.2
4.4.2.3
4.4.2.4
4.4.2.5
4.4.2.6
4.4.2.7
4.5
Dominance, hierarchy, obedience and aggression
4.5.1 Dominance and Aggression
4.5.2 Hierarchy and Aggression
4.5.3 Obedience and Aggression
18
18
19
19
19
19
19
20
21
21
21
21
21
22
22
22
22
23
24
24
25
27
4.6
Fear, Submission and Aggression
28
4.7
Play and Aggression
28
4.8
Classifications – Types of Aggression
4.8.1 Descriptive
4.8.2 Functional
4.8.3 Causal-motivational Classification
4.8.4 Contextual
4.8.5 Neural
4.8.6 Learning
4.8.7 Complex and/or diagnosis
4.8.8 Therapeutic
4.8.9 Operational Classifications
4.8.10 Veterinary medicine
4.8.10.1
Pageat (1998) *
4.8.10.2
Dehasse (2002b) *
4.8.10.3
Overall (1997)
4.8.10.4
Feddersen-Petersen (2000b) *
4.8.10.5
Landsberg, Hunthausen & Ackerman (1997)
4.8.10.6
Houpt & Wolski (1998)
4.8.11 Psychology
4.8.11.1
Askew (1996)
194
29
29
29
30
31
31
31
31
31
32
32
32
32
34
35
36
36
37
37
4.8.11.2
O’Farrell (1991)
4.8.12 Cynology
4.8.12.1
Lindsay (2001)
4.8.13 Campbell (1999)
4.8.14 Ethology
4.8.14.1
Zimen (1992) *
4.8.15 Discussion of the Classification of Aggression
38
38
38
39
39
39
40
4.9
Drugs For Treating Problems of Aggression
40
4.10
Discussion and Conclusion
41
5. OVERVIEW TO EVALUATING (AGGRESSIVE) PROBLEM BEHAVIORS OF
DOGS ACCORDING TO A FEW SELECTED MODELS
42
5.1
Overview of Chapters
42
5.2
Approaching Problems of Aggression
45
5.3
Treating Problems of Aggression
48
5.4
Summary
49
PART II: COMPARISON OF FIVE SELECTED METHODS TO EVALUATE THE
AGGRESSIVE BEHAVIOR OF DOGS
1.
INTRODUCTION
51
51
1.1
Aims of the Comparative Study
51
1.2
What is a Test?
52
2.
METHODS, ANIMALS AND MATERIALS
2.1
Dogs and Owners
2.1.1 Finding the Dogs and Their Owners
2.1.2 Number of dogs and owners
2.1.3 Breeds of Dogs
2.1.4 Sex of the Dogs
2.1.5 Age of dogs
2.1.6 Weight of dogs
2.2
The Questionnaire
53
53
53
53
54
55
55
55
56
2.3
The Tests Which Were Compared
57
2.3.1 The Test of the American Staffordshire Terrier Club (ASTC) of Switzerland, Test A
57
195
2.3.1.1 History
2.3.1.2 Aim of the Test
2.3.1.3 Breeds
2.3.1.4 Prerequisites for the Normal Test
2.3.1.5 Measures if the Test is not Passed
2.3.1.6 The Evaluation
2.3.1.7 Why This Test Was Chosen
2.3.1.8 The Experts
2.3.1.9 Location
2.3.1.10
Time Schedule
2.3.1.11
Duration
2.3.1.12
Material, People and Animals necessary
2.3.1.13
The Actual Test
2.3.2 The ‘Halterprüfung’, Switzerland, Test B
2.3.2.1 History
2.3.2.2 Aim of the Test
2.3.2.3 Prerequisites for the Normal Test
2.3.2.4 Measures if the Test is not Passed
2.3.2.5 The Evaluation
2.3.2.6 Why This Test Was Chosen
2.3.2.7 The Experts
2.3.2.8 Location
2.3.2.9 Time Schedule
2.3.2.10
Duration
2.3.2.11
Material, People and Animals necessary
2.3.2.12
The Actual Test
2.3.3 The Test of the Canton of Basel-Stadt, Switzerland, Test C
2.3.3.1 History
2.3.3.2 Aim of the Test
2.3.3.3 Breeds Normally Accepted to the Test
2.3.3.4 Prerequisites for the Normal Test
2.3.3.5 Measures if the Test is not Passed
2.3.3.6 The Evaluation
2.3.3.7 Why This Test Was Chosen
2.3.3.8 The Experts
2.3.3.9 Location
2.3.3.10
Time Schedule
2.3.3.11
Duration
2.3.3.12
Material, People and Animals necessary
2.3.3.13
The Actual Test
2.3.4 The Experts
2.4
The Behavioral Specialists
The Veterinary Behavioral Therapist
2.4.1.1 Education – Approach
2.4.1.2 The Dogs
2.4.1.3 Location and Time Schedule
2.4.1.4 Duration
2.4.1
196
57
58
58
58
58
59
59
59
60
60
60
61
61
63
63
63
63
63
63
64
64
65
65
65
65
66
69
69
69
69
69
70
70
70
70
71
71
71
71
72
72
73
73
73
74
74
74
2.4.1.5
2.4.1.6
Material, People and Animals Necessary
The Evaluation
2.4.2 The Ethologist
2.4.2.1 Education and Approach
2.4.2.2 The Dogs
2.4.2.3 Location and Time Schedule
2.4.2.4 Duration
2.4.2.5 Material, People and Animals necessary
2.4.2.6 The Evaluation
2.4.2.7 Location and Time Schedule
2.4.2.8 Duration
2.4.2.9 Material, People and Animals necessary
2.4.2.10
The Evaluation
2.5
74
74
75
75
75
75
76
76
76
Fehler! Textmarke nicht definiert.
Fehler! Textmarke nicht definiert.
Fehler! Textmarke nicht definiert.
Fehler! Textmarke nicht definiert.
The Time Schedule
76
2.6
Statistical Tests
2.6.1 Descriptive Results
2.6.2 Comparison of the Methods
2.6.3 Correlation Within the Tests, Correlation of Aggression and Dangerousness
2.6.4 Comparison of Approaches
3.
EVALUATION
77
77
77
78
78
51
3.1
Questionnaire
79
3.2
The Tests
79
3.3
The Evaluation Scale
3.3.1 The Original Scale
3.3.2 The Definite Scale
82
82
84
4.
PROBLEMS DURING TESTING
86
5.
RESULTS
79
5.1
Programs Used
87
5.2
Descriptive Evaluation of Every Method
Dogs
Dates
Experts
Stimulus dogs
Sequence
Exactness of answers
Intraspecific aggression (Criterion 1)
87
87
87
88
89
91
91
92
5.2.1
5.2.2
5.2.3
5.2.4
5.2.5
5.2.6
5.2.7
197
5.2.8
5.2.9
5.2.10
5.2.11
5.2.12
Interspecific aggression (Criterion 3)
Interspecific aggression, dog alone (Criterion 14)
Behavior toward moving objects (Criterion 5)
Aggressive Behavior Towards Other Animals (Criterion 7)
Relationship Dog And Owner (Criterion 11)
5.2.12.1
‘Hierarchy’
5.2.12.2
‘Attachment’
5.2.13 Test Passed / Not Passed (Criterion 12)
5.2.14 Measures recommended (Criterion 13)
5.3
Results of the Comparison of the Five Methods
Intraspecific Aggression of the Dog Toward Other Dogs (Criterion 1)
5.3.1.1 Descriptive Results
5.3.1.2 Statistical Results
5.3.2 Interspecific Aggression of the Dog Towards Humans (Criterion 3)
5.3.2.1 Descriptive Results
5.3.2.2 Statistical Results
5.3.3 Behavior of the Dog (Alone) Toward Humans
5.3.4 The Behavior of the Dog Toward Moving Objects (Jogger, Bicycle)
5.3.4.1 Descriptive Results
5.3.4.2 Statistical Results
5.3.5 Aggressive Behavior of the Dog Toward Other Animals
5.3.6 The Relationship of the Dog and its Owner
5.3.6.1 Descriptive Results
5.3.6.2 Statistical Results
5.3.7 Test Passed or Not
5.3.8 Measures Recommended
5.3.8.1 Descriptive Results
5.3.8.2 Statistical Results
5.3.1
5.4
Results of Correlation Within the Tests / Methods
Breed
Sex of Dog
Weight of dog
Age
Expert
Test Sequence
Stimulus Dog
‘Hierarchy’
‘Attachment’
Measures
Correlation of Aggression
5.11.1
5.11.2
5.11.3
5.11.4
5.11.5
5.11.6
5.11.7
5.11.8
5.11.9
5.11.10
5.11.11
5.5
Results of the Tests Concerning the Correlation of Aggressive Behavior and
Dangerousness
198
93
95
96
97
99
99
100
101
101
105
107
107
109
109
109
110
111
111
111
113
114
115
115
118
119
119
119
120
106
121
122
122
123
124
125
125
126
129
131
137
141
5.5.1
Intraspecific Aggression Towards Other Dogs and Dangerousness For Other Dogs
142
5.5.2 Interspecific Aggression Towards Humans <-> Dangerousness For Humans
142
5.5.3 Interspecific Aggression (Dog Alone) Towards Humans and Dangerousness of the
Dog Alone for Humans
143
5.5.4 Behavior Towards Moving Objects and Dangerousness for Moving Objects
144
5.5.5 Behavior Towards Other Animals and Dangerousness for Other Animals.
145
5.6
Comparison of the Approaches Using a Reference Value
5.6.1 Intraspecific Aggression
5.6.2 Interspecific Aggression
5.6.3 Interspecific Aggression (dog alone)
5.6.4 Behavior Towards Moving Objects
5.6.5 Behavior Towards Other Animals
5.6.6 ‘Hierarchy’
5.6.7 ‘Attachment’
5.6.8 Measures
6.
DISCUSSION
141
146
147
148
149
149
149
150
150
146
6.1
Discussion of Methods
6.1.1 Science and Research
6.1.2 Validity and Reliability
6.1.3 Subjectivity in Science
6.1.4 Standardization
6.1.5 Choice of the methods
6.1.6 Definitions
6.1.7 Statistics: Choice of Sample Population, Evaluation
6.1.8 Influencing factors
6.1.8.1 Identifiable, not controllable factors with unknown influence
6.1.8.2 Identifiable, not controllable factors with identifiable influence
6.1.8.3 Identifiable and controllable factors:
6.1.8.4 Identifiable factors that can be eliminated:
6.1.9 Conclusion
151
151
152
155
156
156
157
157
158
158
160
160
160
161
6.2
162
162
162
163
163
164
164
165
165
Discussion of the Descriptive Evaluation of Every Method
6.2.1 Dogs
6.2.2 Dates
6.2.3 Test Experts
6.2.4 Stimulus Dogs
6.2.5 Test Sequence
6.2.6 Exactness of Answers
6.2.7 Intraspecific Aggression (Criterion 1)
6.2.8 Interspecific Aggression (criterion 2)
199
6.2.9
6.2.10
6.2.11
6.2.12
Interspecific Aggression, Dog Alone (Criterion 14)
Behavior Toward Moving Objects (Criterion 5)
Aggressive Behavior Toward Other Animals (Criterion 7)
Relationship Dog – Owner (Criterion 11)
6.2.12.1
‘Hierarchy’
6.2.12.2
‘Attachment’
6.2.13 Test Passed (criterion 12)
6.2.14 Measures Recommended (Criterion 13)
Discussion of the Results of the Comparison of the Three Tests
6.3.1 Intraspecific Aggression of the Dog (Towards Other Dogs)
6.3.2 Interspecific Aggression of the Dog Towards Humans
6.3.3 Behavior of the Dog (Alone) Towards Humans
6.3.4 The Behavior of the Dog Towards Moving Objects (Joggers, Bicycle)
6.3.5 Behavior of the Dog Towards Other Animals
6.3.6 The Relationship of the Dog and Its Owner
6.3.6.1 ‘Hierarchy’
6.3.6.2 ‘Attachment’
6.3.7 Test Passed
6.3.8 Measures Recommended
166
166
167
168
168
169
169
170
6.3
172
172
173
173
173
173
174
174
174
174
175
6.4
176
176
176
177
178
179
179
179
180
181
181
Discussion of Results of Correlation within the Tests / Methods
6.4.1 Breed
6.4.2 Sex of Dog
6.4.3 Weight of Dog
6.4.4 Expert
6.4.5 Test Sequence
6.4.6 Stimulus Dog
6.4.7 ‘Hierarchy’
6.4.8 ‘Attachment’
6.4.9 Measures
6.4.10 Discussion of Correlation of Aggression
6.5
Discussion of the Results of the Comparison of Aggressiveness and Dangerousness
Within the Tests
182
6.6
Discussion of Comparison of the Approaches Using a Reference Value
Textmarke nicht definiert.
7.
CONCLUSIONS
FEHLER! TEXTMARKE NICHT DEFINIERT.
8.
REFERENCES
Fehler!
186
200
9.
APPENDIX
192
10. INDEX
FEHLER! TEXTMARKE NICHT DEFINIERT.
201