Hendriks, Anja Alide Jolijn

University of Groningen
The construction of the five-factor personality inventory (FFPI)
Hendriks, Anja Alide Jolijn
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to
cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2009
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Hendriks, A. A. J. (2009). The construction of the five-factor personality inventory (FFPI) Groningen: s.n.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Download date: 17-06-2017
8. Summary and Discussion
Thisstudydealswith the constructionand a first psychometricevaluationof the Five-Factor
Inventory(FFPI; Hendriks,Hofstee,& De Raad,in preparation;Hendriks,Hofstee,
Personality
five broad dimensions
De Raad,& Angleitner. 1995),which instrumentefhciently assesses
(factors)of individual differencesin behavior:Extraversion,Agreeableness,
Conscientiousness!
in
ErnotionalStability,and Autonomy. The FFPI consistsof 100 brief and concretestatements
thethird personsingular,and can be administeredin 10-15minutes.In additionto the five factor
scores,the FFPI may be usedto assess40 bipolar facet scoresthat ariseas blendsof the five
factors.
The FFPI was developedwithin the psycholexicalparadigmin which it is assumedthat
naturallanguageprovidesfor. or elsewill incorporate,all elementsthat peoplefind importantin
about individual differencesin behavior.In line with this assurnption,
theircommunications
r.vhichis also known as the "sedimentation
hypothesis"(Brokken, 1978)or "lexical hypothesis"
(Goldberg,l990;McCrae, 1990),adherentsof the paradigmtake the (unabridged)dictionaryas
theirmain sourcefor delineatingthe "personalitytrait domain".Presumablythe Englishscientist
andwriter Sir FrancisGalton (1884) was the first personto scana dictionaryfor personalitytrait
(John.Angleitner,& Ostendorf,1988).For the lexical approachto personality.
descriptors
however,Allport and Odbert's(1936) dictionarysaanhas beenof more significance.These
authorswere the first to systematicallylist and categorizeall terms in the English languagethat
can"... distinguishthe behaviorof one humanbeing from that of another"(p.24). Following
Allport and Odbert,dictionaryscanshave beenundertakenby researchers
in many different
countries.
These"(multi-)traitpsychologists"especiallyselectedand subsequently
empiricallyexamined
primarily by trait
the interrelationships
of the "stabletraits", which were taken to be represented
adjectives(talkative,docile, and so on), rvith the ultimategoal of revealingthe structureof
personality.Nowadays.more and more consensus
is emergingthat four or five replicablefactors
adequately
summarizepeople'svarianceon most of thesetraits.Moreover,while researchers
haveusedpartly different labelsin the past,it is now becomingcommonpracticeto refer to the
first four personalityfactorsas Extraversion,Agreeableness,
Conscientiousness,
and Emotional
Stability.Concerningthe fifth factor. debateis still going on, not only with respectto its label,
but also on whetherthis factor provesto be replicableat all. While it was originally refered to
as Culture(Tupes& Christal,1961),and subsequently
as Intellect(Goldberg,1990;Ostendorf,
The constructionof the Five-FactorPersonalitvInventorv
1990),recentlysuggestedlabelsfor this fifth factor are Creativity(Johnson,1994;Saucier,
1992), Originality (Saucier, 1992), or Imagination (Saucier, 1992, 1994).Another widely used
label is Openness
to Experience(Costa& McCrae, 1985, 1992);this label, however,doesnor
follow from studiesin the psycholexicaltradition.
From the momentit becameavailable,PrincipalComponentsAnalysis(PCA) followed by
varimax rotation has been rfte method to reveal the structureof personality traits. The ciassical
"Big-Five" simple-structuremodel, in which trait terms are assignedto the one factor in the
five-dimensional
trait spaceon which they have their largestprojection(loading),has been
refinedto the AbridgedBig-Five DimensionalCircumplex(AB5C) Model (Hofstee& De Raad,
1991;Hofstee,De Raad,& Goldberg,1992)someyearsago. Most trait terms appearedto be
blendsof two of the Big-Five factors:apartfrom having a high loadingon one factor.they have
a quite substantialloading on a second,which meansthat thesetrait terms are close to a vector
in the five-spacewhich is positionedsomewherein betweenthe pertainingtwo Big-Five factors.
The AB5C model takesthis finding into account.The model distinguishesbetweenfactor-pure
variables,which have no secondaryloadingof any significance,and two blends:one in which a
specificBig-Five factor plays a primary role and one in which that samefactor plays a
secondary
role. Examplesare the two blendsI+II+ ("cheerfulness")
and II+I+ ("kindness',),as
distinguished
from factor-pureI+ (Extraversion)and factor-pureII+ (Agreeableness)
variables.
The resultis a partitioningof the five-dimensionaltrait spaceinto 90 "facets"containingclusters
of trait termsthat are much more homogeneous
than the five large clustersin the Big-Five
simple-structure
model. The AB5C model thus offers a much betteranchorpoint for factor
interpretation,
by virtue of its representation
of nuancesin trait meaning.
We took the AB5C model as a point of departurefor the constructionof an inventorythat
coversthe five-dimensionaltrait space.Basedon the contentof all well-filled (65) facets.brief
concretestatements
(sentenceitems) were written (seesection2.1). We considereda facet to be
well-filledif it containedat leastthreetrait adjectiveshaving a projectionof .40 or more on the
pertainingfacet vector. The instructionsto team memberswere to define the meaning of each
AB5C facetin a recursiveway, that is, by taking the sharedmeaningof its composingclusterof
trait terms,while contrastingit to the meaningof its oppositeclusterand centeringit betweenits
two adjacentclusters.Each team memberindependentlywrote as many items as he or she could
think of --typicallysome sevenper facet-- given the contentof the AB5C facetsthat were taken
as a point of departurefor a particularteam session.Per team session,which were held once or
twice a week,two to three facetswere thus addressed.
All independentlyproducediternswere
examinedone by one for their fit to the facet for which they were written, and on whether they
fulfilled explicit guidelinesfor item production(Hofstee,1991).Theseguidelinesservedthe
puryoseof creatingan instrumentthat can be usedfor a broad rangeof educationalleveis,
avoidsdiscriminationof certainpeopleor groupsof people,and elicits ratingsas objectiveas
possible.This first stageof the project resultedin a preliminaryDutch item pool of 909 sentence
items.Examplesare: Has a good word for everyoneand Makesfriends easily.
Additional sentenceitems were producedon the basisof personalitydescriptiveverbs(see
section2.2). This word classis fit par excellencefor use in brief behaviorallyconcrete
statements
because,in contrastwith trait adjectivesand nounswhich may be usedin simple
ratinglists,verbs needfurther specificationin order to make clear what is meant.For instance,
Listensto others is somethingtotally different from Makes others listen to him/her. Furthermore,
itemsbasedon personalitydescriptiveverbs may add meaningthat is not containedin trait
(De Raad,Mulder, Kloosterman,& Hofstee,1988).In total 136 such sentenceitems
adjectives
wereaddedto the Dutch item pool. Examplesarel.Insultspeople and Knows how lo convince
others.
At this stageof the project,the items were translatedinto American-English,in cooperation
with Lewis R. Goldbergfrom the OregonResearchInstitute,and into German,by Alois
Angleiterand his team from the University of Bielefeld.Translationdifficulties are a notorious
problemwhen testswhich are developedin one languageare translatedinto other languages.
For
this reason,we consideredthe translatabilityof the items to be a prerequisitefor their inclusion
in the final item pool. This strategyhas the advantagethat any difficulties in this respectare
encountered
before,and not after, the final item selectionhas taken place.Becauseof the
time-consuming
natureof sucha procedure,we had to limit ourselvesto American-Englishand
German.But, if items provedto be translatableinto thesetwo languages,their chanceof being
translatable
into still other languageswas expectedto be greatlyenhanced.
Againstthis background,the initial Dutch item pool was extendedfrom a third source(see
section2.3). As the Dutch FactorV is best interpretedas Rebelliousness
or Spirit (Hofsteeet al..
1992).whereasthe Americanand GermanFactorsV are much more Intellectual,additional
sentence
items referringto Intellectwere written in order to link up the initial Dutch item pool
to the Americanand GermanFactorsV. In total 266 Intellectsentenceitems were addedto the
pool. Examples arc',Wantsto understandthings and Wantsto form his/her own opinions.
The initial Dutch item pool eventuallycontaineda total of 1,311sentenceitems.The
translationof this item pool into American-Englishand Germanappearedto be an intricate
75
l
5urrdl[y lnvenlory
proces,and is best referredto as an "internationallyinteractive"way of constructingthe trnal
itenrpool (seeChapter4). First of a|l, 397 items were discardedin the processbecauseof oneof
the following reasons:(1) no good translationcould be found in one or either ofthe other
languages,
or (2) the item wasjudged to be inferential,that is, requiringan inferenceby the
rater.or (3) the item was judged to be a social effect, or (4) its translationin one or both of the
revolvedarounda trait adjective,or (5) the item wasjudged to be arnbiguous.
otherlanguages
or
(6) the item was judged to be too specific.Severalother items neededadaptations,
which could
meanreplacingthem by their closestback-translation.
After a final round of cleansingwith
regardto the guidelinesfor item production,a total of 914 sentenceitems remainedto constitute
the trilingualitern pool.
In The Netherlands.self- and others'ratingswere collectedon the 914 sentenceitems,
togetherwith ratingsof this sampleon a 225-itemtrait-adjectiverating list that coversthe
five-space.
The aim of the study was to determinethe relationshipsof the sentenceitems with
the adjective-based
Big-Five dimensions,and to evaluatethe items' psychometricquality (see
Chapter5). Target subjectswere 167 first-yearstudentsand staff membersof psychology.With
the exceptionof eight of them, eachtargetprovideda self-rating,and was independentlyrated
by two to four otherswho knew the targetpersonwell, giving a lotal of 790 raters.
A principalcomponentsanalysis(PCA) was performedon the pooled (1/: 790) sampleof
self-and others'ratingson the 225 trait adjectives;eachsubject'sraw scoreswere corrected
beforehand
for acquiescence
responseset (Hofstee,1994b).The appropriateness
of pooling the
samplewas checkedin a pre-analysis(seesection5.1.4).The screetest clearly indicatedfive
l-actors,
which were subsequentlyvarimax rotated.Next, Pearsoncorrelationswere calculated
betweenthe 914 sentenceitems and the five varimax-rotatedfactors.which were identifiedas
Extraversion,
Agreeableness,
Conscientiousness.
EmotionalStability.and Autonomy. The AB5C
lacetprojectionsof the sentenceitems appearedto rangefrom -.66 to.71; their medianabsolute
valuewas found to be .47. In other words, we found a clear demonstrationof the relationship
betweenthe sentenceitems on the one hand and the adjective-based
Big-Five dimensionson the
rtherhand.In addition,Pearsoncorrelationswere calculatedbetweenself- and averagedothers'
'atings,indicatingthe self-peervalidity of the sentenceitems.
Thesevalueswere found to be
rerypromising(medianvalue: .35)^
We furtherinvestigatedthe effectsof social desirabilityand observabilityon interjudge
(self-[averaged]others
.greement
correlations),following John and Robins (1993). The sentence
lemshad beenratedon thesetwo variables(seesection3.2), as well as on comprehensibility
(seesection3.1), by independentsanplesat an earlier stageofthe project.Theseratingswere
collectedin order to be usedas secondarycriteria for item selection.We only partly replicated
Johnand Robins' findings: a moderatelystrongpositiverelationship(.30) betweenobservabilitl'
andinterjudgeagreement,but no relationshipof any significancebetweensocial desirabilityand
interjudgeagreementwas found.
equivalencein the three languagesas one of the
Our intentionwas to use nleasurement
plimary criteria for item selection(seesection6.1), next to the items' f'actorloadings,self'-peer
validity,and comprehensibility.Data (N = 766, self-ratings)on the English items were made
availableby Lewis R. Goldberg,and data(1/ = I 18, peer ratings)on the Germanitemswere
madeavailableby Alois Angleitner.Accordingto a pilot study (Houtman, 1994),a Moklien
scaleanalysis(Molenaar,Debets,Sijtsma.& Hemker, 1994)followed by an analysison
DifferentialItem Functioningand DifferentialTest Functioning(Stout & Roussos.1992,1994),
bothin their versionsfor polytomousitems, seemeda promisingcombinationof methodsfor our
goal.We had to conclude,however,that the undertakingmet a dead end. The main reasonto
abandonmeasurement
equivalenceas a selectioncriterionwas a substantialloss of items,and
item variety,due to large differencesin the scalabilityof the items betweenthe Dutch and
Germandata on the one hand and the American data on the other hand, for which differenceswe
hadno good explanation.A secondthoughtin this respectwas that one is not necessarily
interested
in item equivalenceon a one-to-onebasis.but in equivalenceof scalescores.
Instead.we usedthe resultsin the Dutch data set to make a preselectionof 284 sentence
items.on which self- and others' ratingswere collectedin a fresh Dutch sample.This strategy
enabledus to enlargethe sampleof subjectson which the final item selectionwould take place,
while having reducedthe numberof items to a more manageable
(for subjects)proportionlr'hich
only containedthe most promisingones,and to collect dataon convergentmeasuresin orderto
be ableto pertbrm a preliminaryvalidity study.Targetsubjectswere 125 first-yearstudentsof
psychology.With the exceptionof eight of them, eachtargetgave a self-rating.and eachtarget
wasratedby two to ibur otherswho knew the targetpersonwell, giving 606 ratersin total.
Self-ratingswere collectedon the 284 sentenceitems,the 225-itemtrait-adjectiverating list, and
the NEO-PI-R (Costa& McCrae, 1985, 19921.
Dutch translation:Hoekstra,Ormel, & De Fruyt,
1996).Others'ratingswere collectedon the 284 sentenceitems only. This samplewas pooled
with the available(1/ = 790) data set,giving a total ol 1,3I I raters(N : 292 targets),after
deletionof subjectswith suspectresponseprofiles.Becausethey had had different instructions
for their ratings.a checkon the appropriateness
of pooling the two sampleswas performed.An
The constructionof the Five-FactorPersonalitvlnventorv
analysisof variancein intra-individualspreadsacrossthe 284 sentenceitems revealedsignificant
in all analysesusing
main effectsfor Instructionand Rater (self- or other-rating).Consequently,
this pooledsample,eachsubjects'raw scoreswere correctedbeforehandnot only for
acquiescence,
but also for differencesin intra-individualspreads.
In order to determinetheir AB5C-facetpositions,a PCA was performedon the 284 sentence
items,followed by a varimax rotationof the first five factors(in accordancewith the screetest).
The llnal positionsof the axes,however,were subsequently
basedon the joint resultsin the
Dutchand Americandata sets;not on the Germandata set. by reasonof its much smaller
nurnberof raters.We decidedto take the American-Englishstructureinto accountfor the final
item selectionbecause,in spiteof the disappointingresultsin the analysisof measurement
equivalence.
the two structuresappearedto be remarkablysimilar: the congruencecoefficients
for the varimax-rotatedprincipal componentswere all found to be (well) above .87. In orderto
establishthe final positionsof the axes,a Procrustesrotationto optimal agreementwas
performedon the Dutch and Americanmatricesof loadings,whereuponthe consensusmatrix
wasvarimaxrotatedonce more (Evans,1971,p.43; Hakstian,1973,p. 226; seealso Kiers,
1995,for a comparisonof methods).Next, the items' AB5C facet positionswere determinedby
takingtheir two highestloadings.
Severalitem samplingplans were considered(seesection6.4.1).We finally decidedto select
20 good sentenceitems per factor having their primary loadingon that factor, to be spread
acrossthe differentfacetsof the factor so as to avoid redundancy.Consequently,the unweighted
sum scoresof the items for the five scalesare mostly positivelycorrelated;an orthogonalization
procedure(cf. Costa& McCrae, 1992)is neededto obtain orthogonalfactor scores.
The 100 items that constitutethe FFPI were primarily selectedon the basisof having a large
projection,a substantialself-(averaged)others
correlation(self-peervalidity), and a low Difficulty
score(high comprehensibility).
Observabilityand non-extremesocial desirabilityservedas
marginalcriteria.The items meet the standardsfor item quality to a considerableextent,as
Table 10 (Chapter6) shows.
A first psychometricevaluation(seeChapter7) revealedthat the FFPI scaleand factor
scoresshowhigh internalconsistencies,
substantial(six months)stabilities,and very promising
self-peervalidities.Theseresultscan only be taken to be tentative,however.as this first
psychometric
evaluationwas conductedwithin the availabledata set in which also the item
selectiontook place.Also, thesevaluesmay be expectedto shrink somewhatin other-language
versions.
S u m m a r ya n d D i s c u s s i o n
Additional indicationsof constructvalidity were obtainedfrom correlationsbetweenFFPI
factorscoresand scoreson two other five-factorpersonalityinventories.We found a clear-cut
convergentvalidity betweenthe FFPI and the 225-itemtrait-adjectiverating list. Althoughfar
lrom being trivial, theseresultsservemainly to confirm an expectation,sinceboth instruments
stemlargely from the samesource,namely the AB5C model of personalitytraits. With respectto
theNEO-PI-R.clear-cutconvergentvalidity was found for Extraversion,Agreeableness,
to
Conscientiousness,
and Emotional Stability (reversed).FFPI-Autonomyand NEO-Openness
Experienceappearedto have almostnothing (r -.10) in common.Although this correlationof
.10 will almostcertainlyprove to be an underestimate
due to a samplepeculiarity,we do expect
a lack of commonvariancebetweenthesetwo variablesto be a robust finding (see,for instance:
McCrae,1990).This is not necessarilya problem in itself; the questionis more generallywhat to
thini<oi this lack ol commonvariance.
Reflectionson Factor V
McCraeand Costa(McCrae, 1990)believethat the fifth factor identifiedin psycholexical
studiesis best interpretedas a variant of Opennessto Experience,which they considerto be a
psychologically
fundamentaldimension;other interpretations
are taken to be conlbundingsof
intelligence,
educationand sophisticationwith this more basicfactor (.p. 122). Accordingto
McCraeand Costa,the reasonwhy the lexical ttadition has failed to portray adequatelyFactorV
asOpenness
to Experience
is the customaryrigid ("single-word")
applicationo1'thelexical
hypothesis,
which, to their opinion, resultsin an underrepresentation
of certainaspectsof the
personalitydomain.As McCrae (1990) states:"When examinedby facet,it appearsthat English
hasntanywords that expressOpemessto Ideas.but fewer that captureother facetsof Openness"
(p. 12a).Thereis no simple answerto this remark as yet, at leastwith respectto the FFPL If
oneadheresto the lexical tradition in its strongest(single-word)formulation,the obviousreply
would be that peoplemost probablyfind theseother facetsof Openness(to Fantasy,Aesthetics,
Feelings.Actions.and Values)of lessimportance.But if one, like the Groningenteam, adheres
to the essenceof the lexical hypothesis,which merely statesthat individual differencesin
behaviorthat peoplefind importantto communicateaboutwill eventuallybecomeencodedinto
the naturallanguage,the situationis lessclear-cut.Did we fail to portray adequatelyFact<lrV as
Openness
to Experience?When it doestake sentences
to conveyindividual differencesin (facets
of) Openness
to Experienceas it is suggested
by McCraeand Costa,indeedthere may be an
T h e c o n s t r u c t i o no f t h e F i v e - F a c t o rP e r s o n a l i t vI n v e n t o r v
omissionon the part of the FFPI: its basishas been(clustersof) singletrait terms,Intellect
terms.and personalitydescriptiveverbs.De Raad(1994),however,points at the inconsistent
findingson which the postulationof Opennessto Experienceas one of the basicpersonality
dimensionswas based.So, for the time being. Opennessto Experienceis but one of the
candidates
for a universalFactorV, no more so than Intellect,or Creativity,or Imagination,or.
for instance,Autonomy. That is to say, if a fifth factor provesto be replicableat all.
Althoughwe did our bestto come up with Intellectas a fifth dimension(seeChapter2), we
endedup with Autoltomy instead,as factor-pureitems like Linksfacts together.Ilctntsto form
hi.s/herown opinions,andAnol),ses
problemsloadingon the positivepole. versusFollows the
crowd,Copiesothers, and Does what othersdo loadingon the negativepole. reveal.The large
majorityof the extra Intellectsentenceitems that were includedin the final item pool appeared
to be blendsof FactorsV and III (Conscientiousness).
Also trait adjectiveshkeintelligent,u-ise,
quick,cleveras opposedto unintelligent,unwise,gultible, stupid werefound to be associated
with the V+lll+/V-lll- blend in the final structure.Interestinglyenough,it can be noted that of
the threeGermaniclanguagesonly the Germantrait structure(Ostendorf,1990.Table 50) shows
a clearInlellectfactorwhen one appliesan AB5C-modellingto the data.The coreof the
American-English
FactorV (seeSaucier& Goldberg, 1996)appearsto be represented
by traits
that ratherrefer to Autonomy Qthilosophical.inquisitive,insightfut),whereastraits referringto
Intellect(intelligent,intellectual,smart)have their largestprojectionson the v+lII+ blend, like
in the presentfindings.An exciting thought in this respectis that it also is more easyto conceive
of Creativity,Imaginativeness,
and Opennessto Experienceas facetsof Autonomy than it is vice
versa(Autonomyas a facet of one of the others).
It would be interestingto clarify the relationshipbetweenFFPI-Autonomya1d. fbr instance.
measures
of selJ-eLficacy
(Bandura,1977) andlocus of control (Levenson.1914:Rotter, 1966).
An abundance
of publicationson the topic indicatesthat thesevariablesare viewed to be
importantconceptsin developmental,educational,clinical, and healthpsychology(e.g., Bekker,
1993;clark, Steer.Beck, & Ross,1995;cronbach,lglT; Mills, 1994;Ryff, t9g9). preliminary
findingswith respectto the Italian translationof the FFPI (Marino, Perugini,& Ercolani, 1996)
suggestthat the three conceptsare moderatelyto stronglyrelated.
Also in the realm of organizationalpsychologywould one be interestedin the relationship
betweenthe personalityvariablesautonomyand locus of control, as well as their relationship
with job characteristics
and job performance.Basedon a meta-analyticreview,of the literature
with respectto the relationshipbetweenthe Big-Five dimensionsand job performance.Barrick
to be about the only personalityvariablewith some
andMount (1991) report Conscientiousness
study.Barrick and Mount (1993) investigatedthe role ofjob
predictivevalidity. In a subsequent
uutonomyas a moderatingvariableinfluencingthe validity of personalitypredictors.Indeed.they'
founddegreeof autonomyon tlie job to slightly moderatethe validity of at leastthreeof the
Barrick and Mount
Extraversion,and Agreeableness).
Big-Fivedimensions(Conscientiousness,
concludethat future researchshouldconcentrateon the generalizabilityof their resultsin order
to lurther clarify the conditionsunderwhich personalityconstructsare likely to be relatedto
jobs. Hower,'er.personalityvariablesand situational(job) variables
perfbrrnance
in management
rnayhavebeenconfounded,therebyattenuatingthe relationshipsdue to restrictionof range.
Spectorand O'Connell (1994), for instance,found a correlationof -.31 betweenjob autonony
andlocusof control: "internals"are found in jobs rvith higher autonomy.Furthermore.resultsof
a smallstudy (,V: 50) of our own with respectto the ideal applicant'sBig-Five profile fbr
difltrent typesof jobs (Dotinga,Hoekendrjk,& Raa4makers.1995)suggestthat a person's
standingon Autonomy may be more importantfor high-leveljobs with managerialtasks,while
his/herstandingon Conscientiousness
may be more importantfor lower-leveljobs with no
rnanagerial
tasks.Note. however.that this latter study has no ernpiricalbasisother thanTudgctl
importanceof concreteBig-Five relatedbehaviorswith respectto job performance.Nevertheless.
thesefindingsseemto suggestthat studieswith respectto person-environment
fit, concentrating
on Big-Fiveprofiles of applicantsandjobs, may offer an alternativeroute to enhancethe
predictivevalidity of personalityvariableswith respectto job performance.Autonomy may
certainlybe an interestingvariablein this respect.
Conclusion
In conclusion,the FFPI efficiently assesses
five broaddimensions(factors)of individual
dilTerences
in behavior:Extraversion,Agreeableness.
Conscientiousness,
EmotionalStability"and
Autonomy.The first four factorsare well-established
and robust,as the literatureextensivell'
illustrates.
The fifth factor,Autonorly, is lesswell-established,
but appearsto be an interesting
conceptin differenttlelds of psychology.Accordingto a first psychornetricevaluation,the FFPI
scaleand t-actorscoresshow high internalconsistencies,
substantialstabilities,and good construcl
validity.For appliedpurposes,AB5C facet scorescan readily be obtainedfrom the fir'e factor
scores,with reliabilitiesthat are in the sameorder as thoseof the factor scores.The specialitem
format,consistingof brief concretebehavioralstatements
in the third personsingularwhich ,,vere
explicitly selectedfor their comprehensibility,makesthe FFPI applicablefor self- and other's
ratings,and for a broadrangeof educationallevels.Its shortness(administrationtime 10-15
minutes)makesthe FFPI an almostideal instrument,not only for appliedpurposes,but alsoto
be included as a standarddevice in studiesinvestigatingthe relationshipbetweenpersonalityand
othervariablesof interest.Naturally, further validity studiesare neededincluding oneswith
other-language
versionsof the instrument,in order to make clear whetherthe FFPI really merits
its placeamongthe availableinstrumentsin the domainof personality.Basedon the preliminary
resultsso far, however,we have every confidenceit does.