COGNITIVE STYLE AS A BASIS FOR MIS AND DSS
DESIGNS: MUCH ADO ABOUT NOTHING?*
GEORGE P. H u b e r t
It is conmonly believed that the user's cognitive style should bc considered in the design of
Management Information Systems and Decision Support Systems. In contrast, an examination
of the IitcrJrurc and a consideration of some of the broader issues involved in MIS and DSS
design lead to the conclusions that: (I) the currently available literature on cognitive style is an
unsatisfactory basis for deriving operational design guidelines. and (2) further cognitive style
research is unlikely to provide a satisfactory body of knowledge from which IO derive such
guildelines. The article presents six specific basies for these two conclusions.
From a manager's perspective, the outcome of the study is a suggestion: maintain a health)
skzptlcism if it is suggested that paper and pencil assessments of the user’s cognitive style
should be used as a basis for M I S or DSS designs. From a researcher’s viewpoint. the study
raises two questions: (I) If our research interest is MIS and DSS design. does it seem that
further research in cognitive style is a wise allocation of our research rcsourccs? (2) If our
rec;lrch inwrcsl is cognitive st)le. dots it seem that IIIC USC of cognitive SIJIC as a basis fur
MIS and DSS &signs uill hwome an important application arcs?
(hf.\Xs’hCiEMENT INFORMATION SYSTEMS; DECISION SUPPORT SYSTEMS; COGNIl.I\‘E s I1’I.E)
1. Introduction
D u r i n g the p;~st half dozen years, there has been a surge of studies concerning the
effect of cognitive style o n d e c i s i o n m a k i n g behavior or performance.’ Many of thcsc
stud& u’crc nlolivntccl by or linked IO the hclicf that the user’s c o g n i t i v e s t y l e s h o u l d
be con~idcrcd in 111-e designs of Mtinagcmcnt Information Systems (MIS) and Decision ’
S u p p o r t Sys~cms (DSS) ([I]. 141. [ 121. [ 141, [25], (381. and [Sl]).’
WI:II conclusions can we draw, at this time, concerning the usefulness of cognitive
sijle as a basis for DSS designs? An examination of the literature and a consideration
of swnc of the larger ‘issues involved in D S S d e s i g n l e a d t o t w o d i s c o m f o r t i n g
conclusions.
COS~LIAIOS N o . I. Tile crlrrelrr!r available literature 011 cognitive styles is an
unsari$rcto<~* hnsi.r jor dcrking operational guidelines for AfIS and DSS designs.
Cosc~.us~o~ NO . 2. Furlher cognirive syle research is unlikely to lead to operational
~idelitws jar MIS and DSS designs.
Taken together, these conclusions are at variance with some widely held beliefs. It is
especially important, therefore, that the conclusions be correctly interpreted. In particular, they do not state that all cognitive style research is without merit. II may be that
cognitive style research has been or will be useful for career counseling. selection and
‘Acccp~cd by Aric Y. Lcwin; rcccivcd November II. 1981. This paper has been with the author 4 months
!or 2 rcvtsions.
’ Uuivcrsity of Wixonsin. M.tJium. Visiting Schobr. Exccucom Systems Corp.. Awin. Texas.
‘“Cognlrivc slylc rrfcrs IO IIIC procca behavior that individuah cxhihir in the formul;llion or acquisition.
Injlysls. ;InJ inlcrpretation of information or data of presumed value for decision making” [Sl]. “The
qnitiw sr)lc pxadtgm emphasizes the problem-solving process rather than the cognili\e structure and
;I:xi&. II catcpizcs individual hahils and strategies at a fairly broad level and csscnlially v i e w s
r ~rohlcm-wiving hchnvior as a personality variJhlc” 125. p. 741. In particular. ir should be nolcd thai
. cognili\c qlc cmphr\\lzcs an approach rather than an ability.
‘For ihe sake of parsimony. the acronym DSS will hereafter be used to refer to both h(anagemen:
IInlormaiion Systems and Decision Support Systems.
*
5 6 7
i
:
OOZ.c-1909/83/2905/0567501.25
I
CopyriShl C 1913. The Inrwuw ‘4 \I.~na~cmcnl Socnccr
placement, coaching and training. choosing procedures for designing DSS. or for L&’
related purposes. The two conclusions simply do no speak lo the issue of whether &C -1
output of cognitive style research will be useful for such purposes. They do speak w .’
the issue of whether it will be useful as a basis for DSS designs, as is widely bclievcd, 1
a n d state that in all likelihood it will not.
$52 and 3 of the article present six bases for the first of Ihc above conclusions. Two .
of these bases apply only to the first conclusion. The remaining four bases apply to
both rhe first qnd second conclusions. In a more positive vein, 64 suggests some new
directions for cognitive style research that could lead to important contributions to the
fields of management in general and DSS in particular. $5 contains a brief summary. *
2.
BOWS for the First Conclusion
One basis for the first conclusion is that the cognitive style literature is, in aggregate,
both weak and inconclusive. “Among the weaknesses are: (I) inadequately d e v e l o p e d
theory o f c o g n i t i v e styles. (2) multilude of mc;i\uring inhtrumcnts w i t h inadequately
cr~;tblishcd psychomelric p r o p e r t i e s ( e . g . , rcil;lhility a n d v a l i d i t y ) , a n d ( 3 ) f a u l t y
research designs used in empirical investigations of cognitive styles” [60). Therefore,
“Dcspirc the potential importance of cognitive SI) Its for management decision-making
and information use. conceptual and me~hod~~logical wcskncssss have sharply limited
rht p a y o f f s f r o m this line of rexarc!.” 1611.
Vv‘lth regard to the mailer of the hrerature’s being i n c o n c l u s i v e , Zmud [73] r e c e n t l y
nored that “In contrasting the decision behavior of heuristics and systematics h&
consistent and inconsistent findings have been obscrvcd. Syslematics have been seen 10
consisMtly prefer more quantitntivc information 1221. and require more decision time
[-to]. (311. (661 than h e u r i s t i c s . f lowcvcr, syslrmalics have hccn f o u n d 10 p r e f e r more
[IS]. [40]. 166) and less 121. [48] information and I O p r e f e r arzgregated [2] and raw [3j,
[48]. [57) d a t a w h e n c o m p a r e d w i t h h e u r i s t i c s ” 1731. Similarly, in their rmiews, Libby :
and Lewis (331 and Taylor and Benhasat [6l] identified a number of inconsistencies in
the findings from various studies.
It may be that the literature’s problems of methodological weakness and inconsistency should be trcarcd as I W O s e p a r a t e bases for the first conclusion. On the other
hand, it may be that the methodological problems circd earlier arc the cause of at least
some of the reported inconsistencies, and so I have treated the IWO problems as one.
Of course another possible explanation for the apparent inconsistencies is that we do
nol yet have an adequately developed theory of cognitive styles with which to interpret
our findings. But if this is S O, if we do not have an adequate basis for interpreting our
findings, then we must seriously question whether or not WC have an adequate basis for
deriving DSS designs.
A third possible reason for the apparent inconsisrcncics is that the multitude of
measuring instruments for assessing cognitive sr),le (cf. 124). 1491, [61 I), in combination
with the variety of behaviors and performance measures that have been studied, causes
there to be a limited number of precisely comparable studies. This. in turn. may force
those who summarize the literature IO wmparc. cm occasion, the results of swdics that
are not as comparable as the summarizer would wish they were. Thus the literature
must be labeled as inconclusive, since either (I ) Ihe sparseness of comparable studies
prohibits precise comparisons, or (2) the aggregation of noncomparable srudics leads
to the idcnGfication of apparent incon&tencics.
Let us turn now IO the second basis for the fir31 conclusion. This basis is concerned
with the relative magnitude of rhe effects of copnitive style. Almost all of the studies
contained in rhe literature report statiblical significance. Relatively few report, as a
measure of the practical significance. the proportion of the variance in the decision-
;.
1
C(K;SlTI\‘E STI’LE AS A 8.4515 FOR XllS ASD DSS DESIGSS
c)r other
1hcr lhe
peak l o
clievcd.
maker’s behavior or performance tha1 is attributable to cognitive style. When rhey do.
brically only 10% of 1he total variance is so explained. minus or plus a few percent. Of
course this mcasurc of practical significance is not available when small sample sizes
or scalability problems force the resear&er to use nonparametric statistics, b u t
1s. Two
whatever the reason. the problem remains- it is difficult to view the literature as a
whole and IO conclude that cognitive style is of practical significance. This is especially
PP!P to
n c new
; to the
iUXU-y.
rcga k,
:lOpC~i
UillCl_~
f2uity
Xfore,
!nking
mitcd
:cn tly
both
:cn to
lime
1110rC
V.’ [3).
_ihhJ
lcs in
>nsisJther
least
enc.
‘C d o
iprCt
our
.5
569
for
c of
lion
USCS
3rce
that
turc
dies
YIils
ncd
3ies
IS 3
on-
when one considers the facts that (I) the reported studies generally controlled
sources of variance that would be largely uncontrolled in actual practice. and (2)
itudics where little variance was explained lend to go unreported.‘*’
Ihe case
Many researchers involvcJ in the analysis of individual characteristics (e.g.. cognilivc ~t)lcI as
prcJicws of human behavior have reached a troublesome conclusion. Specifically. they have
conrlwlcd th;lt thc study of indtviclual dlffcrrnccs as prcdicrors of human hchavior/pcrformancc have hccn h;lsic;llly unsuccessful. Researchers have nor been ahlc 10 predict consiswnrl! hch~vi~~r/~rforn~an~c on the h3s1s o f i n d i v i d u a l personality characlcristics. Rather.
beh:lvior rrppz;lrs IO hc (IO s wry hrpc Jcgrcc) determined by the characteristics of ths 1~4. m
vlrlch ~hr ~~JI\IJu:II IS invol\cJ [I I].
AIthou!Ji Chcrvany and Dickson (I I] do not indicate that their statemenl f o l l o w s
‘ram a sjslcmutic Jilerslure r e v i e w . it does parallel the thinking of authorities in the
‘ield of hchavioral decision theory (cf. 1511. [55. p. 148)). This view. thar p e r s o n a l i t y
t~mcttristics accoun1 f o r r e l a t i v e l y l i t t l e v a r i a n c e i n c o m p l e x d e c i s i o n tasks, is
upporrccl by studies where a number of personality characteristics are examined (cf.
63). 1701) or where a comparison can be made between the variance explained by task
hancre&tics and the variance explained by individual differences (cf. (431. [45], 1671.
651,.
To summarize what has been said so far. the facts (I) (hat the literature on cognitive
!!lcs is i~c3.k and inconclusive. a n d (2) t h a t t o d a t e t h e p r e p o n d e r a n c e o f e v i d e n c e
ldicntcs that I!IC p r a c t i c a l s i g n i f i c a n c e o f co_rnitive styles is relatively small, suggest
UI the currently nvailahle lilera1urc i s n o t a s a t i s f a c t o r y b a s i s f r o m wllich IO &rive
pcr;l1ion;ll D S S design guidclincs.’
But. U!IJI of the second conclusion? Is it not likely that more and bet1cr r e s e a r c h
P would provide a satisfactory basis? One would certainly like to think so. but a n u m b e r
0 I con&!cmtions suggest that the answer to this question is negative. The firs1 and
‘Of course. a few pcrccnt ol a large numhcr can be a number of some impor1ancc. In such ~3~s. however.
:.. in cazc’~ uhcrc a de&ion can have a consiL(crablc impact. WC expect IO obscrvc (a) multlplc participants
the Jcciwm process anJ (h) extensive organiutional ,wdy and review of both 1hc prohlcm and the choice
1.153!.!39!). Thus. in those cases u-here the user’s cognitive style might bc of some importance. the cffcc~ of
I) one JcciGon m&r’s cogni1ivc s~ylc tends IO hc dllutcd.
‘A WQ few studies report 1h;l1 a large proporlion of variance was explained. in this regard WC should note
it. if Ihc cqwimcntal 134 is wry similar to ilcms in the assessment inrlrumcnl. i1 is incvitdhlc that a high
trrclarion UIII hc ohwncd. This is c\pccially true when 1hc dcpcndcnt variahlc is pcrform~lncc ancl when
c inswumcnr tends IO mcawrc pcrccptual or cognitive ohili!r rather than cqnitivc +/t*. An cxccllcnl
ample is Ihe vcv inrcrcsling ,~udy hy ROUX anrl Rouse 1501. uhcrc an insrrumcnl Ihat INS 1hc suhjcc1’s
#iMy IO match complex images prcdic1cd up IO 50'2 of the variance in pcrformancc on B ~;lsl; of finding
aul15- in graphirally porrraycJ mechanical sys1cms.
‘The wnccpt trf w~nilive style 3s a prcrlictor ol decision-rclatcd hchavior has high law talidily. hItin>
oplc. inrltrrlrny the ;lu1hor. can recall ins1ances crhcrc lhcir suhjcc1ivc a\w>.tmcn1 ol ;rn,)rhrr person’>
Fnilivc style scrmcJ IO h c h i g h l y cxplJna1ory ol the person’s decision-rclatcrl bchavicrr. lluw c a n WC
plain this discrqwncy hetuccn what WC “know” from cxpcricncc and whJc UC mus1 conclude from an
:cs~rwnl 14 ~hr hlcr.llurc? One a n s w e r . I bchcvc. is that our inadvcrlcnt cmploymcnl t,f the r~ndohilir~~
Nwr:ir 1641 CJUWS us IL) juJgc the strength o f t h e cogni1ivc srylc decision-hchavior rcl~lttmship h y the
qucncy ol “hits” in the conflrminy cell of 1hc 2 X 2 table. while ignorine (or not storing) J~IJ in the other
‘cc ccll~ Tuo rcccn! rcvicu’s on ovcrconfirlcncc in jotlgmcnr offer supior1 for this view (15. p. 71. [Q. p,
21. Other answers may he our USC of the rr~rewrtrtiti\tnz~ hcurirrir !&I! (parGcularly as this is rcalizcd
w;h “the illu&m ol valiclity” [61!). o r o u r susscp:ihillty to the himftighr br,rc [>I]. althtru:h I f i n d IL-S
y-wr( for ~hrsr cxpl.~n;ltions than for the prc\ious.
GEORGE P. HUBER
third of these considerations are related principally IO scientific matters. The seak’
. .A
and fourth are related to the practice of DSS design.
:..
3.
-
>
c
. .
:.
z.2
.“
I.
‘- : ‘.
Bases for the Second Conclusion
_.
One basis for believing that further cognitive style research will n o t l e a d t o
operational DSS design guideline:4 is associated with the fact that cognitive style is only
one of many individual differences that may be relevant to designing D S S . F o r
e x a m p l e . r e s p o n s e IO uncertainty 1321. 1701, cognitive complexity 1651. [69]. rcsponsivencss to organizational norms of rationality [37], need for achievement, r i s k - t a k i n g
propensity 1621. 1631. academic background, intelligence [62]. [63] previous training or
familiarity with the particular task or instrumentation 161. [36]. [54], d o g m a t i s m [34],
1621. [63] and a number of other individual differences (591, [70] arc related IO h u m a n
information processing and decision-making behavior. Since it is highly unlikely that
all individual diffcrcnccs would bc positively correlated with one another, either in
general or especially in the case of any one decision maker, it is also highly unlikely
that D S S design g u i d e l i n e s t h a t m i g h t f o l l o w f r o m a p a r t i c u l a r d e c i s i o n - m a k e r ’ s
cognitive style would not be in conflict with at least some of the guidelines that might
follow from some other of the decision-maker’s characteristics. flow are such dilemmas to be resolved? Arc we to build a master multivariate stntistictil model that draws
u p o n the intcrcorrclations a m o n g a l l kno\vn-to-bc-rclrvant i n d i v i d u a l diffcrcnces in
orclcr IO gcneratc a set of “best-fitting” design guidelines? Of course not. Not only is
simultaneously accounting for all individual differences likely to be infeasible, but the
empirically derived correlations would be necessarily based on data from samples of
intli\~idu:rls and would therefore not necessarily apply to the particular individual for
11 horn NY might be designing a D S S .
T o s u m m a r i z e t h i s p o i n t , f u r t h e r c o g n i t i v e s t y l e r e s e a r c h i s unlikily to lead to
operational design guidclincs because (I) there are many individual diffcrcnccs related
IO decision marking behavior and the task of constructing an empirically-h;lscd normati1.e design mocfcl 11131 accounts for all of their effects is ovcrwhclming. and (2) even if
N’C c o u l d b u i l d s u c h a model. it \vould be inapplicahlc to any one decision m a k e r
hcc~usc there arc individual differences in the nature and extent of association among
inilividunl diffcrenccs.
A second basis for skepticism concerning the usefulness of further cognitive style
research for developing DSS guidelines follows from a consideration of what would be
admissible and credible DSS design advice in the eyes of a practicing manager. Many
would agree that it is the professional responsibility of a DSS designer, in his or her
role as a decision analyst, to tactfully challenge apparent deviations between what on
the one hand seems to be the most appropriate decision process, based on the decision
task and setting. and what on the other hand seems to be the process implicit in the
decision-maker’s expressed preferences for data, decision aids, and other components
of the DSS. Some managers would respect challenges to their expressed preferences for
DSS features if the challenge followed from what the designer/analyst saw as the
correct process. But it seems highly unlikely that very many manngcrs would accept as
ndmissiblc or crcdihle chnllcngcs to t h e i r p r e f e r e n c e s i f rhc challcn~cs follo\vcd f r o m
the dcsigncr’s opinion of what DSS features the manager should prcfcr, based on the
dcsigncr’s assessment of the decision-maker’s cognitive style.
The third reason to question the value of further c o g n i t i v e SIJ-lc r c s c n r c h . f o r t h e
purpose of dcvcloping D S S design g u i d e l i n e s . i s associ;ttcd with the fact 11131 Jcvelopmcnt o f e m p i r i c a l l y - b a s e d b o d i e s o f knotvledge is painfully slow. T h e p r o b l e m s
associated with the existing cognitive style literature are so great that, by the time a
scicnlifically satisfactory data base of empirical findings could hc cstsblishcd, “ D S S
g e n e r a t o r s ” (561 may well be so flexible and friendly and data accessing t e c h n o l o g y
y-g y3 ;fi&: :v,
-__
_:: . - 2;
L-rL :‘_-.ve_L ‘_Lir -L. -dt
scientists aspiring ti
deriving operations
puter scientists W(
technology as ad3r
bet on the latter.
Finally, as a fat
research, again Jo
question. “Even if
decision maker’s c
did design such a
reinforcing his 01
examination of th
that predispositior
( c f . 1211. 1511. [64]’
conforms to the u.
or the organizatio
the effect of dysfl
the USC of the cog
support altcrnati~
t h e less 3ppropri3
a D S S uhose feat
implcment3li0n 0
of the DSS or. if I
through use of a
To summarize
user’s cognitive St
analysis of this m
the issue [ 131.
As a conscque
unlikely lo result
sufficiently flcxil
predispositions a
to direct cogniti\
how they s h o u l d
employ the DSS
F o u r argumen
that further co:
guidelines. IVhili
th:~t if only one
have to bc valid
Where should
hl:IS/DSS field’:
At the outset
that cognitive S’
useful are ( I) c:
COGNITI\‘E STYLE AS A B.\SIS FOR MIS ASD DSS DESIGNS
!cond
.d IO
Only
For
Isivcil;in_g
I” or
1=34].
man
lhi1:
:r i n
1kcly
kcr’s
light
IL711r;1u’s
:s ii1
ly is
: the
:S of
; for
.tylt?
be
any
her
On
<ion
the
rnts
for
the
t as
om
the
I
571
may well hc so adv;lnccd that the ides of a stable DSS design may be o b s o l e t e .
lnstcad. users are Iikcly IO be creating. modifying, and discarding *‘specific DSS” [56]
according to their weekly needs a n d w h i m s .
This likelihood of 1hc possibility is, of course, a matter of opinion. Given that it is. I
would Iikc to m:lke it clesrcr. It seems 10 mc that in the race between (1) management
.scicnlists aspiring to drvclop a cognitive style literature that is a satisfactory basis for
&riving opcr;\tioncll DSS design guidclincs, and (2) management scientists and compulcr scientists w o r k i n g togc‘thcr IO develop’ D S S generators a n d d a t a a c c e s s i n g
technology 3s ad~pt~hlc as those suggested above. a person would be well advised lo
bet on rhc latter.
Finally. as a fourth basis for challenging the usefulness of further cognitive qlyle
rc?carch. ;Ig;Iin for rlre pwpose 01 dcrivirrg D S S grridelittes. c o n s i d e r t h e f o l l o w i n g
I;uc\lion, “Even if NY knc\v enough that H’C could design a DSS conforming to the
dcci&n maker’s cognitive style. should W C? ” I am not so sure that we should. If W C
diJ design such ;I system, WY uould provide the decision maker a powerful tool for
reinforcing his or her i d i o s y n c r a t i c predispositions. Do we w a n t IO d o t h i s ? A n
e~;tlninalic~n of 1l1c litcrclturc on the acquisition and use of information makes clear
III;I~ prcdisposilions are often dysfunctional, at least in certain decision environments
(cf. [?I). (Si]. ](;&]I. Given this, iris not at all clear that providing a D S S w h o s e d e s i g n
~.inf~~rnl> IO IIN u~r’s co;nitivc style is i n the best intcrcst of cirhcr t h e dccisicbn mnkcr
or 11x org;tni~ti\i~m on \\hosc hchslf hc or she i s m a k i n g Jccisions. as it m;ly heighten
hc dfcct of dysfunctional predispositions. That is, if we design a DSS th;lt s u p p o r t s
the use of the cognitive style IO svhich the decision maker is predisposed. hut dots not
s;rpporl ;Ilrcrn;lti\.c cogniti\.c s~ylc(s). WC may hc r e i n f o r c i n g the USC of the SIJ.IC Ihat is
:!ic lc<s Ilpproprialr, given the Jccision sclting and [ask. On t h e o t h e r h a n d . p r o v i d i n g
a DSS whose ‘fcaturcs complement the user’s cognitive style. i.e., \vhose fcaturcs allon
implementation of only the alternative cognitive style(s). would likely result in nonuse
of hc DSS or, if use is mandatory, may result in lower performance than th;lt achicvcd
:hrouSh use of a DSS that does conform to user predispositions [5].
To summarize the point, WC do not know if DSS designs should (I) conform to the
ucer’?: cognirivc style or (2) complement the user’s cognitive style. De Waelc’s excellent
an:ilysis of this m;tttcr makes char that sound arguments can be made on both sides of
the issue [ 13 J.
As a conscqucnce of this, it seems that knowledge of the user’s cognitive style is
unlikely to result in operational DSS design guidelines. More likely, future DSS will be
sufficiently flexible that they will enable the users to employ or complcrnent their
predispositions 31 will, as argued above. If this is true, it suggests that it may be useful
IO direct cognitive style research toward the coaching and training of users concerning
hoiv they should think about and through their decision process. and how they can
employ the DSS in that process. We will return to this thought in the following section.
Four arguments have been put forth in this section. Each supports the conclusion
that furrhcr cognitive style research is unlikely to lead to operational D S S d e s i g n
guidelines. While I bclicve that each of the arguments has merit, it is important to note
that if only one or two has merit, the conclusion has merit; each argument does not
hnvc to hc valid in or&r for the conclusion IO be v a l i d .
Whcrc s h o u l d WC go f r o m hcrc ? C a n c o g n i t i v e s t y l e r e s e a r c h b e useful in the
XiIS/DSS field? Thcsc questions arc addressed in the next section.
4.
Canunrnts and Suggestions for Future Hcswrch
Al 111~ outset w should m;lke clear that the arguments of this article do not suggest
Ihat cognitive style rcscnrch is of no use. Among the arcss \vhcrc its results may bc
u s e f u l arc ( I) carter counseling 124). [28]. [29],]33]. (2) personnel sclccrion IS]. 120). 126)
‘>
,
t
:
:
:
‘
:
‘
i.
,
_-:
i;
.’
‘.
1
i
572
GEORGE P. IIUBER
and placement 1351. [50], and (3) coaching and training [54]. [72]. We discuss th&r
possibilities and other matters in this section. In particular, we discuss the need and
opportunity for additional cognitive style research.
Thcrc are studies where a significant degree of association was found between
cognitive style and some occupation (and. also, where no significant degree d
association was found between cognitive style and other occupations) 1241, 128). 129)
Bcforc WC can USC such findings in career counseling. however, we must address four
questions. (I) How rcplicable are the findings? (W C note here that there is some
cvidcncc (15;1]. [ l5h] that these findings would not hc rcplicnhlc.) (2) What is the
(Iom;~in of such findings? (3) Do the apparent associ;ttions exist hccausc people of that
s~ylc ~I\OOX th;lt occupiltion, or bccausc pcoplc of thaw o c c u p a t i o n arc trained and
“conditioned” to use that style? (While both explanations may be valid. the counseling
implications may he quite different.) (4) Do the outstandingly successful people in a
given occupation conform IO the prevalent cognitive style. or Jo they hold IO it even
more strongly than others in the occupation. or do the! manifest a style different from
the run-of-the-mill member of the occupation. or have they modified their style after
thc~, become more successful? (We note here Keen and Bronsema’s finding (241 that
top managers had cognitive st~lcs different from those of middle mansgcn.)
The anstrcrs to such questions are important IO mannfers and manngcmcnt scientists, as members of these occupations are often involved in the career counseling of
subordinates or students. Clearly there is a nwd for cognitive style rcscarch hearing on
IIWW questions. Whcrhcr man;lgcmcnt scicntihts should Icad the way is ;ln issue best
left for another forum.
One way that organizations attempt IO increase the quality of decision making is
through personnel selection and placement. This may home an important area of
application for cognitive style research. “The most direct application of this work (on
cognitive style) appears to be in matching managers to decision situations where their
naturcll styles are most effective” 1581. Especially in \.icw, however. of our earlier
remarks concerning the large number of other relevant individual differences, it is
clear that the instruments for assessing cognitive style must be highly predictive of
decision-making behavior and performance in order to he of practical value. To date,
the proportion of empirical studies that have found copitive style IO be highly
predictive is very small. Let us consider the nature of the research that might lead :o
an increase in the predictive validity of cognitive style astcssmcnt instruments.
There arc a variety of instruments for assessing cognitive style [24], [49], [al].
Unfortunately, management scientists have viewed this almost exclusively as a prob
lem rather than. at least in part, an opportunity. The facts that the intercorrelations of
scores on these instruments tend to be low 1661. 1731 and that some of the individual
inslrumcnts assess multiple constructs [Sl] indicate th3t cognitive style is a generic
construct possessing a number of subconstructs. Thus, it is somewhat surprising that
thcrc arc very few studies where the prcdicti1.e validity of various instruments is
comp:lrcd. and I am aware of no instances \vhcrc it is csylicitly rccommcndcd that
cognitive style instruments shc~~ld hc comp;lrrd ;~nd a cl1oil.c m;lJc b;~sc~I on a match
hct\vccn instrument features and the nnturc of the ticcision setting ;~nll t a s k . II seems
th:~t mt~ch of prtictical vuluc could h c gained if rcscarch \\crc unrlcrtakcn IO cxaminc
IJIC validity of the various instruments for prcJicting behavior or performance in
diffcrcnt types of decision situations. What is needed is the dcvclopmcnt of a
‘,
con tingenc
empirical r
tasks!
Related
informatio
Relatively
identificati
more imp<
ent decisii
were used
Let us .c
that, hold*
sclcction :
constant,
predictive
Predictive
the direct
types of i
&UChit?g
A num
task con:
making t
appropri;
narrow a
natwall>
To an
processe:
of coach.
calic for
style). A
larger re
explicit :
dffer suy
and the!
ally dep
should t
their SIJ
If thi!
making.
implicat
follo\vin
(I) T
cognilit
coaches
propens
(2) T
cofnili\,
COG+II~‘I\‘~:. SI7’I.E AS A BASIS FOR hllS AND DSS DESIGNS
573
contingency theory of cognitive style subconstructs, a theory or at least a body of
empirical research that relates subconstructs and instruments to decision settings and
tasks.6
Related to this point is 1hc ohscrvation that today’s DSS are helpful primarily in the
information-retrieval and choice-making phases of the overall problem solving effort.
Relatively less support is given to the problem Formulation and ncedcd-inFormation
identification phases. It may be that it is in these phases that cognitive style plays a
more important role. Rcse3rch that treated diFFeren problem-solving phases as different decision tasks could tell us if this were so, especially if a variety of inslrumcnls
were used lo aid in making this judgment.
Let us summ3riz.c what h3s been said here with the Following. (1) It can be argued
thnt, Itoltling it11 clsc const;tnt, knovvlcdgs of a person’s cognitive style may hc useful i n
sclcction 3nd plxemcnl. (2) Since in almosl any real situation, “311 else” will not h e
constant, cognitive s1yle will not be useful in selection and placcmcnt ,unlcss 1hc
prcdictivc vnlidity of the cognitive style assessment instrument is quite high. (3)
Predictive validity will, in general, no1 be high unless cognitive style rcscarch moves in
the direction Of rel3ting the predictive validity of p3rticular ins1rumcnts to parricular
IypCS of decision siluaiions.
A numher of studies (CF. [43], (671. (6S]) and aurhorities [Sl), [55].]58) indicate (ha1
task consider3tions tend to dominate style considerations as predictors of dccisionmaking hehnvior. In addition, For at least some tasks, rhis dominance may be quite
appropriarc. as the range of functional approaches to the decision t3sk may be quite
narro\\’ and may not include approaches that are congruent with 1he decision-maker’s
nntumlly cognilive style 15). 16). [ 12).
TO 3n cxtcnt thcsc f3c1s are often the consequence Of coaching and training. IWO
proccsscs that organizations use to increase performance. Although the intcndccl goal
of co:tching and training is in general to direct behavior, in many inslclnces il. in cffcct.
calls for a modificstion of behavioral propensiries (one of which may bc cognirive
style). An outcome OF coaching and training, then, is th3t those [mined will have a
larger repertoire Of behaviors upon which to dr3w. A number of authorities 3rc quite
explici1 about the desirability of such an outcome. Ta8gar1 and Rohey, for example,
offer supporting quotes From Chester Barnard, Harold Leavi11. and JJenry Minlzberg,
and they themselves slate that managers should be able to employ “flexible, situa1ionally dependent styles and slrategies for decision making” 158, p. 1881. “Managers
should be flexible in processing style . . . they are more effective if they c3n change
their style IO fit their problems** 158, p. 1911.
If this is true, if managers should have (I) a reperroire of approaches IO d e c i s i o n
making. and (2) the ability to appropriately select From this repertoire, wh3t are the
implications for cognitive style research? it seems that the implicstions include the
following:
(I) Thcrc is 3 r&cd For more and hettcr research concerning the assessment o f
cognitive s t y l e . a s suggcs~cd b y T a y l o r nnd Bcnh3sit1 1611 a n d Zmud 1731, so that
conches, tritincrs. and decision m a k e r s vvill have an accurate asxssmcnt of **n:tturaj”
propcnsilics:
( 2 ) ‘IIlcre i s a need f o r conlingcncy-fc,cliscil rcscarL*h. rrscarch rclilling p;~fticular
cognilivc style s u b c o n s t r u c t s t o different decision sctlings anil t;~\ks. ;IS s~~y~cstcd
COGS
y
*.
:
<
.. :.
c;
.’ _
; .
2;
carlicr anJ by Payne [44]. so that decision makers will have avsil;lblc a taxonomy of
decision situations 11131 is useful for matching s~ylcs IO situations: and
(3) Thcrc is a need for rcscnrch that will show how decision mrrlcrs can hc trained lo
employ cognitive: styles other than that IO which they arc naturally prcdisposcd. It may
hc useful. bclors initiating such research. to examine IWO sets of studies: (1) studies
\vhcrc training was effcclive in enhancing cognitive complexity (cf. 116). [ 171. [sl]), and
(2) studies e x a m i n i n g c u l t u r a l d i f f e r e n c e s i n c o g n i t i v e style (cf. (191, [3I], [471X
cspccia1ly studies rrlaling acculturation to changes in cognitive style (cf. 191. [27]. [461,
1711.)
It seems rcasonnhlc IO b e l i e v e t h a t when the range of decision tasks is small,
cognitive style will explain a larger proportion of the variance in behavior or perform;lncc than when the range of [asks is large. This fact. pius what w e k n o w a b o u t
hchavior modifictition in applied s e t t i n g s 123). suggests what the mcasurablc success of
co;lching and training in allcrnative cognitive styles wilt he a fun&n of the r a n g e o f
decision situations involved. The transfer of learning (to use orhcr than the natural
cognitive s~ylc) \vill bc grcatcr if the application situarion is similar to [hc l e a r n i n g
.\ilu;ltion than if the situations are dissimilar. This ohscnation rcinforccs the riced for
rcscarch s u c h as [haI noted in (2) a n d ( 3 ) a b o v e . C l e a r l y , there a r c n e e d s a n d
opporlunitics for further rcscurch on cognitive slylc.
The tihovc discussion focused on an outcome, lhc cnl;irgcmcnl of a decision m a k e r ’ s
rcpcrtoirc of approaches IO decision situations. Let us (urn now to a focus on process.
I;pccifically, let us consider the use of cognitive style in designing training programs for
Jccision makers in general and DSS users in particular.
Al!hough “Icarning style” h a s received more than a modcratc amount of artenrion
from psychologists [Xl, 1291. 130). a n d while Taggari a n d Robe\ have suggeslcd that
managcmen~ education should be directed toward the enhancenwnr of nonanalytic as
~~11 a s a n a l y t i c d e c i s i o n s t y l e s [SS]. m a n a g e m e n t scicntisls have appnrcnlly not
conAxc~l lhc possibility that training in the U S C of decision aidins tcchniqucs (e.g.,
opcrntions rcscarch) or decision a i d i n g devices (e.g.. D S S ) s h o u l d b e d i f f e r e n t f o r
p e o p l e o f d i f f e r e n t c o g n i t i v e s t y l e s . It is not at all clear, f o r e x a m p l e . t h a t t h e
proccdurc f o r tc;lching intuitives I O p e r f o r m anslyscs s h o u l d bc lhc same a s t h e
proccdurc used IO reach a n a l y t i c s , o r t h a t t h e career manngcmcnt program IO develop
:Inaly[ics into gencr3l m3n3gers should be the same as that used IO develop intuitives
into general m a n a g e r s .
llisroricnlly. m a n a g e r s , managcmenf e d u c a t o r s . a n d instruclors in the arca o f
computing have addressed this matter as if it were one of l
‘sIow learners” and “fast
Icarncrs.” They had no choice. There were no research resulrs IO use as a basis for
dcsi_cning different training programs. This is another arca. then. lvhere there is a need
for cognitive style rese3rch.7 In psrticular. (I) there is a need for more and better
rcscnrch concerning the assessment of cognitive style, so that the co3chcs and trainers
of decision makers in general and DSS users in particular will have accurate assessmcnts of natural propensities. and (2) (here is a need for research relating p a r t i c u l a r
cognitive style subconstrucls lo different decision situations anJ le;lrning situations.
N o t h i n g solid in this section h3s c o n t r a d i c t e d t h e IWW conclusic>n\ pur forth a1 the
hcginning of [hi.\ ;1rliclc. \Vh;lt h a s been s a i d . ho\\,ct.cr. dots ha\x iml~liccllions for DSS
designs, O n e implica,
o f approaches to tht
design effort should
that provides a varic
determination of the
becomes largely irrel
The second implic.
s u p p o r t o f the sect
employment of his c
of an alternative St>
constraining the nar:
that if a DSS is lo SC
enforce or capture a
style should be ace\
made earlier, there
that m a n y D S S ha\
incumbents move 11
constrained design :
be made using a D
selecting approachc
q u a l i t y i f t h i s str3
n o n - D S S related d.
coaching and trair
style-selecting strat
through D S S dcsi:
In closing this ~1
in the near future.
d e s i g n t h a t Tag:.
technological ad\;
features must be rr
style. be considere
Let us be clear
t h a t i n d i v i d u a l d:
They undoubtedI>
often suggest l o
processes, and dec
ing, at the very le
will be the result c
H o w c a n rhe DSZ
not, as we noted (
model.
There seem to
t h e designer/con
components or fC
through experienc
a n d i n l i g h t of h
designcr/consu!t:
what the user thi
of the approache
her role as a deci
g o o d candidates
576
CoGNlTI\‘l
GEORGE P. IIUBER
CAR,_~OU. E.. “An APPro.
(Wmlcr 1979).
CIIII,,v~sy. S. I_. ASD D,r
i n ‘The hllnnccola EXP\
1091-IO’)?.
D,CYMS, G. W.. ST.>Y. J.
The Minnc\oLI hpcfim
Dr. WAr1.l.. X1.. ‘htan3fc’
seem incompatible with other information. such as the nature
the decision CIA or the proven usefulness of certain data or decision aids. Bo,
approaches allow rhc user’s personal characteristics (i.e., individual differences)
affc’ct the DSS design. bum only to the degree Ihal the user feels th;l~ they should, give
his or her perception of their relative importances in his or her decision situarion.
In employing either approach, we arc essentially taking the posirion that a larl
number of user’s chnracleristics may be important and that the idiosyncratic an
intcrxtivc import;lncc of these individual diffrrcnccs IO the DSS tlcsign is more readii
dctcrmincd by the user rather than by the designer/consultani. \\‘hclher the user
numerical score on a marginally predictive cognitive style assessment instrumel
Lvould be of much value IO either the user or the designer/consultant seems high1
problematic. fnstcad, it seems that task considerations and the user’s expresse
prcfcrcnccs for specific DSS features should be the factors that dtwrmine DSS design
ahout thcw fcaturcs
5 . Summ~
pp. s- 13.
C
16.
8
in sumwry. this article sugx~s that (I) the study of cognitive srylr as a basis fc
deriving opcr~tional guidelines for XlIS and DSS designs has no1 been fruitful and (2
such study is likely not IO prove fruitful. Six bases for the firer conclusion wer
idcnlifird. four of \vhich are also bases for the second conclusion. In ag_crcgnlc Ihe,
offer suppw for the Micfs ( 1) th3l the designs of fulurc DSS will nol hc dcierrninel
from ~hc dcsi_encr’s knowledge of the user’s score on a c0gnitit.c style assessmen
inslrumcnf. and (2) that our current fancy. of focusing such a large proportion of ou
rcwarch rcwurces on studying the user’s cognitive style as a basis for MIS and DS!
designs. is quite possibly a misallocation of these resources.
In p:lrt as a consequence of (he above, the article goes on IO su_r_cesr some nev
dircclions fur cognitive st~lc research that could lead IO imporlan( contributions IO the
ficlils of management in general and DSS in particular.8
17.
18.
19.
20.
Ii.
22.
23.
24
25.
‘1 uould like ,I) thank lzak Bcntwal. S~anlcy Bigpr. Dennis Fphack. John Ilcndcr>on. Jahn P3\nc. John
\vhlrnc?. R&r, ZmuJ. 3rd Ihr rrwwr, f o r their wry h e l p f u l ctrmmen,s ,)n c:l:licr tlr& of this
~;~IIIIw~~;~,.
‘Ihi> rw3rch ~‘3s suppor,d i n p3r, h) thr Arm) Rrw3rch Inswule
f o r lhr St)&1
a n d Dch3vioral
SCKI~CS~ and in par, hy Ihr N;t,iond Scicncc FounJa,ion’s Di\l%ion of Inl,umatiLm Scisncc ;InJ Tcrhnology,
!
I
i
I
26
27
28.
29.
30.
-.
Rcnf\;
1. \l.
A
cd.. PrznlKc-ll2ll. f
31. ~~0s~. L. xl. AYD Dr
and Dual:slic cor?.C