Association for Information Systems
AIS Electronic Library (AISeL)
ICIS 1980 Proceedings
International Conference on Information Systems
(ICIS)
1980
A CRITIQUE OF COGNITIVE STYLES
THEORY AND RESEARCH
Ronald N. Taylor
University of British Columbia
Izak Benbasat
University of British Columbia, [email protected]
Follow this and additional works at: http://aisel.aisnet.org/icis1980
Recommended Citation
Taylor, Ronald N. and Benbasat, Izak, "A CRITIQUE OF COGNITIVE STYLES THEORY AND RESEARCH" (1980). ICIS 1980
Proceedings. 25.
http://aisel.aisnet.org/icis1980/25
This material is brought to you by the International Conference on Information Systems (ICIS) at AIS Electronic Library (AISeL). It has been accepted
for inclusion in ICIS 1980 Proceedings by an authorized administrator of AIS Electronic Library (AISeL). For more information, please contact
[email protected].
11111111111 1 011011111 lili 111 lili
A CRITIQUE OF
COGNITIVE STYLES THEORY AND RESEARCH
RONALD N. TAYLOR
Faculty of Commerce and Business Administration
University of British Columbia
IZAK BENBASAT
Faculty of Commerce and Business Administration
University of British Columbia
ABSTRACT
Within the past decade many writers and researchers have
recognized the importance of including psychological
characteristics in managerial decision models. Cognitive
styles have been singled out for research attention to
determine their importance in information use and the
acceptance of the recommendations made by management
Despite the evidence that cognitive styles
scientists.
influence information processing in these contexts, conceptual
and methodological weaknesses have seriously hampered the
Among the
conclusiveness of cognitive syles research.
weaknesses evident
in cognitive styles research studies are:
1) inadequately formulated theory of cognitive styles, 2) the
use of a great many inadequately validated measuring
instruments for assessing cognitive styles, and 3) faulty
research designs.
A review of the cognitive styles literature
is presented; then these three problems are discussed and
suggestions are made for correcting these limitations in
cognitive styles research.
1.
A
INTRODUCTION
Within the past decade, the accounting
and
mangement
information
(MIS) ·design
literature has
reflected an increased
attention
the
to
recent
report
of
the
American
Accounting Association Committee on Human
Information Processing (2) states that an
understanding of the variables and
processes involved in human information
processing and decision making
psychological
is
a
characteristics of the decision maker.
Research in this area has been stimulated
prerequisite to improving human decisions.
by hypotheses such as "the utility of a
discussed in the above report and one
One
particular type of information can not be
(4)
attuned and
tested
the
impact
of
literature
Witkin
(44)
ib
defines
have discussed three types of
managerial
empirically
cognitive
information systems design:
complexity,
2)
independence-dependence,
psychological
systematic-heuristic.
on
information system
and
use.
Among the
acceptance
characteristics examined are tolerance
MIS
cognitive style which have relevance for
will use most effectively' (30, p. 478).
charactistics
categories
cognitive styles as "the characteristic,
self-consistent modes of functioning which
individuals show in their perceptual and
intellectual activities." Bariff and Lusk
should be given the kind of information to
have
and
"cognitive styles."
to conform to one type of information
system, rather each psychological type
number of studies
psychological
accounting
that "the designers of information systems
should not force all psychological types
A
the
which has received attention in the
effectively evaluated apart fromthe users
of that information," (15, p. 518), and
which he is psychologically
of
These
for
ambiguity (15), dogmatism (25,37), risktaking propensity (37), and cognitive
roles
styles
in
cognitive
styles
1)
field
and
and
3)
the
research which has investigated their
(9).
82
in accounting and MIS are examined
paper.
The research evidence
this
suggests that both problem characteristics
and
user
Two additional studies. relating to
influence
characteristics
cognitive complexity are discussed below.
decision-making strategies and dictate
information system design.
Simon and
Newell have argued that "a few, and only a
The Decision Style model of Driver and
Mock (17) measures cognitive complexity
and segregates decision makers in terms
of the amount of information they tend to
few, gross characteristics of the human
information-processing
system
use
are
(36,
p.
148).
Although
there
is
and
the
number
of solutions
they
However, two experimental
generate.
studies (28,34) have failed to support the
invariant over task and problem solver"
some
validity of the model.
disagreement regarding the relative impact
of problem characteristics and problem
solver
the
characteristics,
McKenney and Keen (29)·have developed
evidence
a model of how decision makers differ in
emerging from experimental studies
suggests that both problem and decision-
their data extraction modes.
Receptive
types display a desire to analyze all the
maker characteristics influence
problem-solving strategies and, moreover,
detailed
raw
decision.
these two sets of characteristics may
data
before
making
a
the apparent importance of
Preceptive types look for
certain cues, deviations from or
conformities with their expectations. The
intake of information of a preceptive is
cognitive styles for managerial decision
dictated by his heuristics for cataloguing
interact in exerting such influence.
Despite
what he finds.
Tests used to measure
these styles are discussed in (21).
making and information use, payoff from
this line of research has been limited by
research which fails to meet adequate
The results are
standards of quality.
2.2 STUDIES OF FIELD INDEPENDENCE-
evident in the discouragement expressed by
DEPENDENCE
Chervany and Dickson (13) regarding
cognitive styles research in MIS.
Standards
Witkin
a1-1- (44) define field
independence-dependence in the following
for evaluating studies of
1) appropriate
fashion: "In a field dependent mode of
perceiving, perception is dominated by the
and valid measures of cognitive style, and
3) use of research designs capable of
overall organization of the field; there
cognitive
styles include:
theory developments,
2)
use of reliable
is relative inability to perceive parts of
This global quality
is indicative of limited differentiation.
Conversely, a field independent style of
perceiving, in which parts of a field are
experienced as discrete from organized
background, rather than fused with it, is
a field as discrete.
In the next
yielding conclusive results.
sections of this paper cognitive styles
research related to accounting and MIS is
first reviewed and then evaluated in the
light of these standards.
.
a relatively differentiated way of
2. REVIEW OF COGNITIVE STYLES RESEARCH
functioning."
RELATED TO ACCOUNTING AND MIS
Studies by Witkin et al. have shown
that
2.1 STUDIES OF COGNITIVE COMPLEXITY
Cognitive
differentiation
extracted
complexity
(number
involves
of dimensions
Central
and
integration
(number
independents
have
more
to Witkin's
theory
is
the
contention that the ability to "break-up"
from the data), articulation
(the fineness of the discrimination
process),
field
analytical and structuring abilities
compared to field dependent types (45).
a configuration reflects not only
of
perception,
interconnections among rules for combining
but also
indicates
approach to problem solving.
data).
a basic
Although
critics have challenged the validity and
value of Witkin's conceptualization of
cognitive styles
Schroeder, Driver, and Streufert (35)
have done extensive theoretical and
experimental
work
in
cognitive complexity.
Paragraph
Completion
the
of
of Witkin and his associates has uncovered
They developed the
an important pyschological construct to
Test
field
(16,47), there exists
strong experimental evidence that the work
(19,35)
to
explain managerial information use.
measure complex, integrative thinking.
Empirical studies (24,35) have shown that
integratively more complex individuals can
process more concpptual data, derive more
clues and search for more information than
integratively simple individuals.
Such
findings
Of the three tests developed by Witkin
et al. to measure this cognitive style,
the Group Embedded Figures Test (45) was
selected for use in experiments on the
acceptance of managerial reports and
are relevant to MIS since the
information use.
information load appropriate for effective
information processing by individuals of
different integrative complexity is an
In summary, the experimental evidence·
published so far on individual differences
system design (31).
high analyics (field independents)
important determinant of
information
as
83
measured by the
GEFT
indicates
that
out
perform low analytics ( field dependents)
measures 1) perception of objects along a
in certain problem solving and decision
Low analytics prefer
making tasks (8,27).
disaggregated data (4,27) and perform
evaluation of objects along a thinking vs.
sensing
(27)
logical process,
or equally
STUDIES
OF
the
at an
impersonal
The feeling style bestows on
a personal, subjective value.
things
SYSTEMATIC-
feeling
HEURISTIC
types
tend
to
take more
stands
(30).
style
measuring systematic-heuristic styles were
statistically significant but relatively
different problem-solving approaches.
Systematic reasoning refers to a tendency
problems
with
rather
than
a
system
are
that
(as
systematic
or heuristic)
was
more
highly correlated to all other tests than
solved
any other single test.
of
Common
underlying causal relationships.
"the
operationally defined by the test itself."
He observed that a self-evaluation rating
The search is
familiar
interrelated concepts
measure
to an emphasis on workable solutions to
problem situations.
that
stated
measurements of cognitive style do not
measure exactly the same feature, but
A
more or less explicit model, often stated
in quantitative terms, forms the basis for
Heuristic reasoning refers
each decision.
analogies
he
Therefore,
low.
to reduce problem situations to a core set
of underlying causal relationships.
found that the
correlations among five different tests
characterizes
which
(40)
Vasarhelyi
Systematic-heuristic describes a
cognitive
for
2)
While thinking types tend to systematize,
STYLES
total
aimed
finding.
(8) with disaggregated data compared
to aggregated data.
2.3
and
thinking style tends to be based on a
High analytics prefer
(26), and perform better
scale
On the evaluation dimension,
(26) observed that low analytics pre ferred
well
intuition
feeling scale.
better with them (8,27), although Lusk
summary reports.
vs.
important role to the extent that they are
In a laboratory experiment Vasarhelyi
(41) found that heuristics used less
information (presented in aggregrate form)
applied to the total situation
than
sense, intuition, and unqualified feelings
about
future
developments
play
an
(9).
Analytics
analytics.
tended
to
utilize computers in planning more than
heuristics did; whereas heuristics
expressed more concern with the lack of
flexibility of man/machine systems than
There is a great diversity in the
psychological instruments uaed to measure
The findings
systematic-heuristic styles.
did analytics.
of the studies on cognitive complexity and
An analytic-heuristic questionnaire
developed at the University of Minnesota
field independence- dependence styles
could be compared since they used the same
psychological instruments: however, the
same is not true for studies based on the
was used in six studies to measure
use
systematic- heuristic styles.
measure
Most
of
these
studies were
inventory decision making
The pioneering research into the
to
(6).
conducted under similar production/
the
impact
settings.
(6) ·reported that the analyticdifferences were not
heuristic
Barrett
systematic-heuristic style was conducted
He conducted an
by Huysmans (20).
experiment
the
impact of cognitive style on information
significantly correlated with performance
of
cognitive style on acceptance of operation
He
in the experimental task.
partly
Huysman
explained this finding by indicating that
concluded that implementation was more
the instrument used to measure styles is a
preference inventory, that is, it does not
research
recommendations.
effective when the manager's style matched
measure performance abilities.
the analytic approach used to arrive at
the operations research recommendation
report.
McKenny
and
Keen
(29)
3. EVALUATION OF COGNITIVE STYLES RESEARCH
RELATED TO MIS
developed a
model which distinguishes between
EVALUATING THEORIES OF COGNITIVE STYLE
systematic and intuitive (heuristic)
USED IN MIS RESEARCH
In a problem solving study Keen
modes·
(21) found that analytics are more likely
Most cognitive styles are loosely
defined and based upon only rudimentary
to pick planning type problems, and
heuristics the kind of problems which
theories.
involve hypothesis testing.
Mason and Mitroff (30) have proposed
the use of the Myers-Briggs Type Indicator
to adequately describe the individual.
None of the theories of cognitive style
to measure psychological types of decision
makers.
For example,' there appears to
be no simple answer to the question of how
many cognitive styles must be postulated
which
The Myers-Briggs Type Indicator
84
have
been
applied
to
MIS
have
generated
as
yet
a
systematic
set
administered
of
to the
same sample yields
related constructs for incorporating
cognitive styles within the framework of
little information about the validity of
The Zmud study (48) is an
the tests.
personality theary.
example of this fallacy applied
cognitive styles research.
An isolated measure
of a personality construct, not placed
within the network of other personality
constructs, is of limited value in
understanding behavior. The extent to
which cognitive styles overlap with other
accounting and MIS research.
constructs
(40)
(e.g., general
to
MIS
There is little empirical work Which
relates various cognitive styles used in
intelligence or
study described
in
Vasarhelyi's
the
previous
spatial perception) is not clear, yet this
section has shown that the correlations
is important in conceptualizing cognitive
among various cognitive style measuring
styles as a personality construct. While
there is little harm in using a construct
which overlaps with other constructs in
instruments were statistically significant
Keen (21) has
but relatively low.
correlated his systematic-intuitive
instrument with the Myers-Briggs Test and
reported that, although many of the
the literature, it does limit the
contribution the new construct can make to
understanding information processing and
"thinking"- types
decision making.
it is apparent that a "feeling" type is
unlikely to be systematic.
Zmud (48)
At
relative generality of these styles.
one extreme, theorists have sought a
master cognitive style variable which has
control
over
intuitive
style,
concluded that the Minnesota instrument
primarily measures whether an individual
uses a judging or perceptive approach to
information
acquistion and processing activities
problem solving; that is,
(e.g.,
planned
Witkin).
in
correlated the analytic-heuristic
instrument used in the Minnesota studies
He
(6) with the Myers-Briggs Test.
Crucial to the issue of how many
cognitive styles to postulate is the
primary
are
At the other extreme,
or
spontaneous
whether a
strategy
is
researchers have attempted to specify a
employed.
number of discrete stylistic variables.
While a single master style variable
factor analytic study which revealed that
the Minnesota analytic-heuristic
instrument and the Paragraph Completion
Larreche (24) conducted a
permits simplicity in theories, this
parsimony may be gained at the expense of
test were part of the same factor,
more Precise explanation of behavior in
the Witkin Embedded Figures
specific situations.
whereas
test was
loaded more heavily under a different
Yet, extreme
specificity can lead to a chaotic and
factor.
conceptually sterile position of positing
dimensionality of the cognitive styles
a unique cognitive style for each task.
The solution to this problem may lie in
the organization of cognitive style
construct, the factor-analytic studies add
variables.
It is possible that,
Apart from indicating the multi-
little
insight
into
the
nature
cognitive styles measures (3).
these factor-analytic studies
just as a
primary motive or need may subsume a
variety of subordinate motives or needs, a
relatively independent
more basis cognitive style variable such
be high on one measure of cognitive style
organize
a
number
of
less
tapped by
the various tests; thus it is possible to
and low on another measure of cognitive
as field independence-dependence may serve
to
factors
of
All of
showed
general
style.
cognitive style variables. Nevertheless,
this variety of cognitive styles used in
It is apparent that the various
MIS research has led to confusion about
measures of cognitive styles used in the
literature are not interchangeable and,
presumably, will differ considerably in
the nature of the construct
their implications for MIS design.
and has
limited the generalizability of findings
from
studies
using
the
different
The inadequacy of theory development
measures.
in cognitive styles is evident in view of
the great many specific cognitive styles
reported in the literature, each allegedly
Clarifying the nature of the cognitive
styles constructs can be approached by
indentifying the dimensionality of the
measured
constructs.
generally attempt
limited
yet, despite several attempts
to determine the dimensionality of
by
a number of
The
tests.
theories that have been developed
to
predict
specific
information- processing or decision-
cognitive measures, the results are far
The great number of measures
from clear.
making
behaviors
and
have
not
been
concerned with a systematic mapping of the
proposed to reflect cognitive styles
cannot all be included in a single factoranalytic study, and the studies performed
one
(10,21,39,40,42)
included
prefer disaggregated reports, it fails to
relatively few measures of cognitive
style.
Moreover, as Fiske (18) has
demonstrated observing patterns of
how the analytic- heuristic cognitve
correlation
for
have
scores
each
on
two
psychological domain of cognitive styles.
For example, while this approach permits
to
assert
that
heuristics
tend
to
explain why this relationship holds and
sytles related either to other cognitive
tests
styles or to other personality constructs.
85
The confusion regarding cognitive styles
can be somewhat alleviated by specifying
the cognitive styles in a generic sense.
EVALUATING
3.2
AND
RELIABILITY
THE
VALIDITY OF COGNITIVE STYLES MEASURES USED
IN MIS RESEARCH
To
be
treatment, for example, analytics can be
compared with heuristics to determine the
effects due to cognitive style.
artifactual
age ,'sex ,
(e.g.,
Covariates
experience)
however,
may produce
results.
It
is
always
possible that other characteristics of
interpreted
meaningfully,
subjects which are systematically related
must demonstrate adequate reliability and
to cognitive styles may remain an
alternative explanation for the effects
validity, yet little concern for these
essential features of tests is evident in
control over artifactual results can be
measures of psychological characteristics
Some
attributed tp cognitive styles.
attained either in f priori assignments of
subjects to treatments (e.g., matching
research into cognitive styles.
Frequently this information, which is
essential for determining the value of the
subjects
measuring instruments, is not reported in
published studies. Exceptions are Vannoy
serious problem remains in that subject
and
(39),
(6)
Barrett
who
on
the covariates)
or
by
Yet, a particularly
covariate analysis.
characteristics may interact with task
reported
characteristics in determining performance
-- thus limiting the extent to which the
findings can be generalized. For example,
reliability figures.
Validity of cognitive styles measures
cognitive style may interact with task
requirements for special knowledge (e.g.,
can best be approached by construct
validation demonstrating a network of
as in a warehouse expansion problem) in
theory-supporting interrelationships among
The result is
producing task performance.
Vannoy (39) provided one
variables.
example of using the.multi-trait multi-
a theory of cognitive style which is
method matrix to test for discriminant and
complex and difficult to interpret since
convergent validation
cognitive
(11),
but he found
theory
to
cognitive
specify
styles
the
into
may
influence
task
various levels of experience in the task.
The population of primary interest in
managerial information processing and
placement of
variables
style
performance differently for subjects at
that the different measures of cognitive
styles that he used failed to converge on
the construct.
Unfortunately, construct
validation requires a well-developed
decision making
a
is
the
organizational
theoretical network of relationships (11,
manager and it is important to generalize
32).
research conclusion with confidence
Such
a
theory
has
not
formulated.
yet
been
EVALUATION RESEARCH DESIGNS USED
COGNITIVE STYLES RESEARCH FOR MIS
this population.
sufficiently similar to business managers
in their information use to justify
IN
generalizing findings
from research
conducted on them to managers. Caution
must be exercised when managers, due to
Research designs typically used to
investigate the impact of cognitive styles
on information processing are deficient in
two important methodological features:
to
If other subjects are
used (e.g., business students) a central
issue is whether these subjects are
·
different'experience,
status,
or
1)
motivations
tend to deal with information
in ways which differ from processes
Adequate provisions are seldom made to
typical of subjects who have not acquired
these skills, status, or motives. An even
more serious problem is the possibility
generalizability and 2) control.
insure
generalizability of
findings
to
that
populations of interest. The inadequacies
include small sample sizes, failure to
randomly assign subjects to treatments,
the
differences in subject
may interact with
characteristics
experimental treatments, thus threatening
both internal and external validity (12).
unequal and nonproportionate cell sizes
As
for F-tests, and experimental research
a
consequence
experimental designs
limited to laboratory settings.
of
these
flaws,
for cognitive styles
research fail to fully captialize upon the
Since cognitive styles are attribute
ability of experiments to cancel out or
variables, subjects are assigned to
treatments on the basis of attributes they
control for alternative explanations.
Similar methodological
Generally it is assumed that limited
difficulties are presented in attempting
availability of subjects of interest and
cost of data collection make it advisable
already possess.
to include other attribute variables (e.g.
male vs. female) in experimental designs.
One way to partially resolve this
to use as few subjects as the requirements
difficulty is to measure cognitive styles
design permit.
separately and, then, to randomly assign
subjects to treatdents.
Hence, for each
and the accompanying statistical tests
specify both the requirements for cell
of statistical inference and experimental
86
The experimental design
frequencies and any additonal requirements
using dummy variables in which l's and 0's
are assigned to subjects depending on
to meet the assumptions of the statistical
tests used;
whether they possess or do not possess a
for example, the existence of
equal or proportionate cell frequencies.
characteristic.
In this sense, the sample size depends on
the statistic used, the number of
independent variables to be studied
partitioned.
simultaneously,
and
be
included
as
Continuous variables can
such
rather than being
Similarly, experimental
treatments can be handled as variables.
Unequal n's are still a problem with
the number of levels
multiple regression analysis, but much
to be assigned to each independent
Ackoff (1) has presented a
variable.
useful discussion of this issue and
less so than is the case with analysis of
variance.
Also, it is still possible (as
in the use of analysis of covariance with
provided a table to generate sample sizes
experimental designs)
from these considerations.
to statistically
control the effects of certain variables
Sample
size
researcher
information
can
be
reduced
is
willing
about
certain
to
if
on relations of interest.
the
give
(14)
up
provide insightful discussions of the use
of multiple regression in this type of
interactions
among the experimental variables.
Among
research.
the options open to the researcher who has
no interest in certain interactions, and
thus is willing to have them confounded,
are
the
use
of
Control difficulties have also plagued
cognitive styles research.
fractional
blocks,
Both Darlington
and Kerlinger and Pedhazur (22)
Experimental
replication, and latin square designs
(23,43). Benbasat (7) provided an example
designs for cognitive styles research
seldom contain adequate control groups.
of the use of fractional factorial design
Especially questionable is the issue of
determining to what extent subject
in cognitive styles research.
characteristics interact with treatments
The class of experiments using subject
(mentioned above) without also determining
used in
base
rates
f o r the observed
information-processing behaviors.
Design
behavioral science research, yet the
of an experiment rests upon understanding
characteristics as
appears
problem
to be
of
potential
variables
(38).
independent variables
increasingly
maintaining
control
over
both the sources of error that may mislead
the
confounding when attribute
are used is a difficult one
Whereas subjects can be assigned to
experimental groups on the basis of type
of cognitive style possessed, when there
are two or more attribute variables in an
(e.g.,
control over potential sources of error -the control group is a central feature in
experimentation and permits any observed
differences in the experimental results to
eliminating some
subjects from the analysis) to have equal
or proportionate
cell
frequencies.
This
be
is because, as in the case of cognitive
the
influence of the
While the full experimental designs
A related problem is that partitioning
styles, as generally conceptualized, into
experimental variable and little control
over alternative explanations for research
high and low cognitive style (heuristic
vs. analytic, perceptive vs. receptive,
Dichotomizing
to
provide more conclusive research findings,
a class of quasi-experimental designs has
been developed for conditions where there
is limited possibility of manipulating the
a continuous variable such as cognitive
loses information.
attributed
treatment.
styles and general intelligence, such
variables are likely to be correlated, not
independent.
etc.)
the procedures
(12,33). A strength of the experimental
approach is that it does permit direct
experiment it is difficult without using
artificial means
researcher and
available for either controlling error or
compensating for uncontrolled error
findings through random assignment of
subjects
a
to
treatments.
experimental
between cell variance and nonsignificant
experiments (5), nonequivalent control
groups (12), cross-lagged panel
results
when,
relationships
significant.
in
may
fact,
be
the
tested
statistically
correlations
methods
(46),
such
Quasi-
continuous variable can mean reduction in
as
field
and time-series designs
A better method is to use
attempt to simulate manipulation, provide
interval or ordinal scales which represent
controls over confounding variables, and
For
probe for causal relationships.
elaboration of the features and relative
merits of these designs for the reader is
referred to
more accurately
style variables.
the continuous cognitive
One approach
for overcoming these
discussions by Campbell and
difficulties in cognitive styles research
is the use of multiple regression
analysis. Multiple regression analysis
assume an even more important role in
there are any) as independent variables
cognitive styles upon complex MIS
designs.
employs
the dichotimous variables
Stanley (12) and Taylor and Vertinsky
(38).
(if
87
Such quasi-experimental designs may
4. CONCLUSIONS
REFERENCES
1.
Based upon the critique of cognitive
Ackoff,
R.L.
Scientific
Method:
Optimizing Applied Research Decisions.
styles theory and methodology used in
New York:
empirical research of this variable, a
Wiley,
(1962).
number of conclusions seem appropriate.
First,
the generic term "cognitive styles"
2.
subsumes a great variety of specific
cognitve
styles
and,
while
useful
in
indicating the entire class of cognitive
styles, its use to refer interchangeably
to specific cognitive styles is confusing
and misleading.
The specific cognitve
style measured in a study should be
indicated; such as cognitve complexity,
and his electric factor analysis machine,
1967), 17-21.
4.
the same things.
styles
must
Management Science, 23, (1977), 820-829.
5.
Barnes, L. Organizational change and
In V. Vroom
field experiment methods.
(ed. ), Methods of Organizational Research,
University Pittsburg Press,
57-112.
be
adequately validated if they are to be
used to make decisions concerning either
the selection and training of decision
6.
The
proliferation
to the confusion
regarding
the
implications of cognitive styles for MIS
7.
design and use.
Benbasat,
I.
An
experimental
evaluation of the effects of information
research designs
Finally,
Cognitive style:
Barrett, Michael J.
a decision process research program.
Graduate School of Business, University of
Minnesota, (1978).
of
unvalidated tests of cognitive styles has
added
(1967),
an overview of the developmental phase of
makers or the design of information
systems.
Cognitive
Bariff, M.L. and Lusk, E.J.
and personality tests in designing MIS.
Research should press forward to determine
just what the cognitive styles do measure
and the dimennsions represented in the
multitude of tests currently in use.
Tests of cognitive
Association.
The American Statistician, (December
systematic-heuristic.
It should be kept
in mind that tests included in each class
measuring
Accounting
Armstrong, J.S. Derivation of theory
3.
by means of factor analysis or Tom Swift
field independence-dependence, or
may not be
American
Report of the 1976-77 committee on human
information processing, (August 1977).
system and decision maker characteristics
used to
investigate the role of cognitive styles
on
decision effectiveness. Ph.D. Thesis,
University of Minnesota, Minneapolis,
in
Minnesota,
information
system design
must
be
The difficulties are due to 1)
improved.
the sampling and data analysis problems
introduced
by
using
an
8.
Value
Benbasat, I. and Dexter, A.S.
and events approaches to accounting:
an
attribute
(nonmanipulated) variable, such as
cognitive
styles,
in
experimental evaluation.
experimental
appropriate experimental or statistical
advocated
Multiple
regression
as a solution to many
difficulties
o f
9.
The
Benbasat, I. and Taylor, R.N.
impact of cognitive styles on information
is
Management
system design.
the
introduced by the use of
cognitve styles in experimental
Accountint
Review, LIV, (October 1979), 7 IS=-7-49:---
research, and 2) the failure to provide
controls.
(1974).
Information
Systems Quarterly, 2, (June 1978), 43-54.
research.
Provision of control groups will be even
10.
more important as research into the impact
cognitive style. Journal of Personality,
28, (1960), 167-185.
of cognitive
styles upon
information
Broverman,
D.M.
Dimensions
of
system design and use is performed in
field sites.
Far
11.
from
expressed
Campbell,
D.T.
and
Fiske,
D.W.
Convergent and discriminant validation by
sharing the pessimism
by Chervany and Dickson (13)
the multitrait - multimethod matrix.
Psychological
81-105.
regarding the future of cognitive styles
research in MIS, the authors believe that
the potential payoffs from sound research
into the psychological characteristics of
Bulletin,
56,
--
(1959),
12. Campbell, D.T. and Stanley, J.C.
Experimental and Quasi-Experimental
information system users are high.
Unless
the theoretical and methodological aspects
of such investigations are improved,
Designs
for
Research.
however, the
13. Chervany, N.L. and Dickson, G. W.
Chicago:
Rand-McNalley, (1966).
full benefits of this line of
research will not be realized.
On
the validity of the analytic-heuristic
88
dependence.
instrument utilized in the Minnesota
experiments - a reply.
Manfasment
191-202.
Science, 24, (1978), 1091-1092.
14.
27.
Multiple regression
Darlington, R.B.
Empirical Research in
Selected Studies, (1973),
Accounting:
----
Lusk,
E.
A
test
of differential
in psychological research and practice.
performance peaking for a disembedding
psychological
task.
Bulletin,
69,
(1968),
--
161-182.
15.
·
Cognitive characteristics
Denner, J.
and
the
perceived
48,(1973),
importance
Accounting
information.
J.
29. McKenny, J.L. and Keen, P.G.W.
How
Harvard Business
managers' minds work.
Review, (May-June 1974), 79-90.
Selected Studies,
30.
M.
and
Mock,
Human
T.
Review,
18.
personality
D.W.
Can
a
L,
Bulletin,
89-92.
80,
--
and
Mitroff,
I.I.
A
Management Science,
31. Miller, D. and Gordon, L.A.
Conceptual levels and the design of
accounting information systems.
Decision
3
Sciences,
construct be validated empirically?
Psychological
R.O.
for research on management
information systems.
12, (1973), 475-487.
theory, and accounting information
The Accounting
systems.
(1975), 490-508.
Mason,
program
information processing, decision style
Fiske,
effects of personality on a
--
203-209.
Driver,
The
subject's information processing.
The
Accounting Review, (July 1978), 681-697.-
Review,
511-519.
Research to Accounting:
17.
Research,
28. McGhee, W., Shields, M. and Birnberg,
of
A reply to cognitive
16. Dermer, J.
field
aspects of annual reports:
independence / dependence.
Empirical
(1973),
Journal of Accounting
(Spring 1979), 286-294.
(April 1975),
259-269.
Psychometric Theory,
32. Nunnally, J.C.
New York: McGraw-Hill, (1967).
(1973),
19. Gardiner, Gareth S. and Schroeder,
Harold M. Reliability and validity of the
33. Ross, J. and Smith, P.
Orthodox
experimental designs. In H. Blalock and
paragraph completion test:
theoretical
and empirical notes.
psych21291£21
Reports, 11, (1972), 959-962.
A. Blalock (eds.), Methodology in Social
The Implementation
20. Huysmans, J.H.B.M.
New York:
of Operations Research.
34.
Wiley-Interscience,
21. Keen, P.W.G.
35.
The Implications of
Thesis,
Research. New York:
in
Holt, Rinehard &
in 1970.
propensity
Brooks-Cole,
and
intelligence
of
decision-making strategies for a sample of
industrial managers.
Larreche,
managers
American Psychologist, 26, 2,
37. Taylor, R.M. and Dunnette, M.D.
Influence of dogmatism, risk-taking
(1968).
24.
Human
the state o f the theory
(1971), 145-159.
Experimental Design:
California:
Driver, M.J. and
Human Information
New York:
Holt, Rinehart &
problem solving:
Procedures for the Behavior Sciences.
Belmont,
H.M.,
S.
36. Simon, H.A. and Newell, A.
Behavioral
Winston, (1973).
23. Kirk, R.E.
decision making.
Processing.
Winston, Inc., (1967).
Ha rvard
22. Kerlinger, P.N. and Pedhazur, E.J.
Regression
The use of accounting
in
Schroeder,
Streufert,
University, (1973).
Multiple
R.
Press,
Accounting Review, (July 1977), 642-652.
(1970).
D.B.A.
Savich,
information
Cognitive Style for Individual Decision
Making,
Academic
]RI# rch, New york:
and
Psychology, 59,
Marketing
a search fo r a
Jean-Claude.
models:
Journal of Applied
(1974), 420-423.
better match.
Research Paper Series No.
157, The European Institute of Business
Administration (INSEAD).
38. Taylor, R. N. and Vertinsky, I.
Experimenting with organizational
behavior.
In W. Stabuck and P. Nystrom
25. Long, Barbara H. and Ziller, R.C.
Dogmatism and predecisional information
Oxford University Press, (1978).
(eds. ), Handbook o f Organizational Design.
39. Vannoy, J.S.
search.
Journal of Applied Psychology,
12, (1965), 376-378.
26. Lusk, E.
construct, Journal of Personality and
Social Psychology, 2, (1965), 385-396.
Cognitive style aspects of
annual reports:
fie·ld
independence
Generality of cognitive
complexity - simplicity as a personality
/
89
40. Vasarhelyi, M.A. Man-machine planning
systems:
a cognitive style examination
ofinteractive
decision making.
Ph.D.
Thesis, University of California at Los
Angeles, (June 1973).
Man-machine planning
45. Witkin, H.A., Oltman, P.K., Rusin, E.
and Karp, S.A. The Embedded Figures Test,
Consulting Psychologist Press, Inc., Palo
Alto, California, (1971).
46. Yee, A.H. and Gage, N.L. Techniques
for estimating the source and direction of
41. Vasarhelyi, M.A.
a cognitive style examination of
systems:
causal
interactive decision making.
115-126.
Accounting Research,
Journal of
(July
1977TT
influence
138-153.
47. Ziegler, E.F.
42. Vernon, P.E. The distinctiveness of
a theory.
field
independence.
Journal
panel
data.
(1970),
A measure in search o f
Comptemporary Psychology, f,
(1963), 133-135.
of
Personality, 40, (1972), 366-*17
48.
Zmud,
R.W.
On the validity of the
analytic-heuristic instrument utilized in
the Minnesota experiments.
Manatement
Statistical Principals in
43. Winer, B.J.
Experimental Design, 2nd Ed., New York:
McGraw-Hill,
in
Psychological Bulletin, 2,
----
Science, 14, (1978), 1088-1090.
(1971).
44. Witkin, H.A., Dyk, R.B., Patterson,
H.R., Goodenough, D.R., and Karp, S.A.
Psychological Differentiation, New York:
Wiley, (1962).
90
-----