cognitive affective theory

10
Journal ol Personality an d Social Psychology 1975, Vol. 32, No. 1, 83-91 Affective-Cognitive Consistency, Attitudes, Conformity, and Behavior Ross Norman University of Western Ontario, London, Canada Three studies were done to investigate th e relationship between affective- cognitive consistency and the strength of the attitude-behavior relation. In each study the affective and cognitive components of students' attitudes toward volunteering as a subject for psychological research were assessed, and an op- portunity was later presented for these students to actually volunteer as sub- jects. It was found as hypothesized that students showing high affective- cognitive consistency were more likely to act in accord with stated attitude than those showing- lo w intraattitudinal consistency. It was also hypothesized that students showing high affective-cognitive consistency would have less tendency than low-consistency subjects to conform to the actions of others in attitude-relevant behavior. The data did not provide strong support f o r this prediction. Overall, th e results indicate th e importance of assessing both th e affective and the cognitive components o f attitude in identifying those whose verbal attitude reports have consequences for their behavior. Attitude formation and change have long been prominent research topics in social psy- chology. Implicit in most of this research ap- pears to be the assumption that an individ- ual's attitudes ar e significant determinants of his subsequent behavior. However, despite early warnings against assuming a strong rela- tion between verbal attitude measures an d other forms o f behavior (e.g., Corey, 1937; LaPiere, 1934), only in recent years h as there been a concerted effort t o study t h e nature o f th e attitude-behavior relation. Recent approaches to the study of attitudes and behavior (e.g., Fishbein, 1967; Wicker, 1971) have emphasized that behavior is a function o f many factors (norms, situational pressures, etc.) in addition to attitude. When these "other variables" are taken into con- sideration in addition to attitude, behavioral predictions are significantly improved (e.g., This paper is based, in part, on a dissertation submitted by the author in partial fulfillment of the doctoral requirements at the University o f Michigan. The author would particularly like to express his thanks to his supervisor, Melvin Manis, and Eugene Burnstein, James Jackson an d Howard Schuman for their advice in the execution o f this research; and to Richard Sorrentino for his comments on an earlier draft of this article. Requests f o r reprints should b e sent to Ross Norman, Department o f Psychology, University o f Western Ontario, London, Ontario, Canada. Copies o f th e scales used in this article m ay also b e obtained from t h e author. Ajzen & Fishbein, 1973; Wicker, 1971). That other factors beside attitude influence behav- io r is hardly surprising. Although th e develop- ment o f refined methodology for the assess- ment of these additional influences is a wel- come result of recent research, the fact that various normative an d situational pressures c an influence an individual's tendency to behave in accord with his attitudes is, of course, to be expected. T he present research h as been directed toward identifying structural characteristics of attitudes which may help predict the strength of their relation to behavior. In par- ticular, this work focuses on Rosenberg's hy- pothesis (1960, 1968) that th e greater th e consistency between the affective and cogni- tive components of an attitude, the truer is th e disposition implied b y that attitude. T o quote Rosenberg (1968): . . . for any particular social issue o r object a s con* fronted b y a sample drawn from some fairly uniform sector of the population (e.g., college students re - cruited for attitude change experiments, o survey respondents recruited from some particular sector o f th e total national population) one can reasonably that those who show less intra-attitudinal consistency are, on the average, less invested in the issue and less likely to have a presently stable orientation toward it. (p. 88) Rosenberg (1968) suggested two reasons why attitudes characterized by low affective- 83

Upload: gabi-landoy

Post on 10-Apr-2018

231 views

Category:

Documents


0 download

TRANSCRIPT

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 1/9

Journal ol Personality an d Social Psychology1975, Vol. 32, No. 1, 83-91

Affective-Cognitive Consistency, Attitudes, Conformity,and Behavior

Ross NormanUniversity of Western Ontario, London, Canada

Three studies were done to investigate th e relationship between affective-cognitive consistency and the strength o f the attitude-beh avior relation. Ineach study the affective and cognitive components of students ' at t i tudes towardvolunteering as a subject for psychological research were assessed, and an op-portunity was later presented for these students to actually volunteer as sub-jects. It was found as hypothesized that students showing high affective-cognitive consistency were more likely to act in accord with stated att i tudethan those showing- lo w in traat t i tudinal consistency. It was also hypothesizedtha t students showing high affective-cognitive consistency would have lesstendency than low-consistency subjects to conform to the actions of others inattitude-relevant behavior. The data did not provide strong support for thisprediction. Overall, th e results indicate th e importance of assessing both th eaffective and the cognitive components of att i tude in identifying those whoseverbal attitude reports have consequences for their behavior.

Att i tude formation and change have longbeen prominent research topics in social psy-chology. Implicit in most of this research ap-pears to be the assumption that an individ-ual's attitudes are significant determinants

of his subsequent behavior. However, despiteearly warnings against assuming a strong rela-tion between verbal attitude measures andother forms of behavior (e.g., Corey, 1937;LaPiere, 1934) , only in recent years has therebeen a concerted effort to s tudy the natureo f the attitude-behavior relation.

Recent approaches to the study of attitudesand behavior (e.g., Fishbein, 1967; Wicker,1971) have emphasized that behavior is a

function of many factors (norms, s i tuationalpressures, etc.) in addition to attitude. Whenthese "other variables" are taken into con-sideration in addition to attitude, behavioralpredictio ns are significantly im proved (e.g.,

This paper is based, in part, on a dissertationsubmitted by the author in part ial fulf i l lment of thedoctoral requirements at the University of Michigan.The author would particularly like to express histhanks to his supervisor, Melvin Manis , and EugeneBurns te in , James Jackson an d H o w a r d Schuman for

their advice in the execution of this research; and toRichard Sorrent ino for his comments on an earlierdraf t of this article.

Requests for reprints should be sent to RossNorman, Depar tment of Psychology, University ofWestern Ontario, London, Ontario, Canada. Copieso f the scales used in this article m ay also be obtainedfrom the au thor .

Ajzen & Fishbein, 1973; Wicker , 1971). That

other factors beside attitude influence behav-io r is hardly surprising. Although the develop-ment of refined methodology for the assess-ment of these addit ional influences is a wel-

come result of recent research, the fact thatvarious normative an d situational pressurescan influence an individual's tendency tobehave in accord w ith his attitud es is, ofcourse, to be expected.

The present research has been directedtoward identifying structural characteristicsof attitudes which may help predict thestrength of their relation to behavior. In par-

ticular, this work focuses on Rosenberg's hy-

pothesis (1960, 1968) that the greater th econsistency between the affective and cogni-

tive components of an attitude, the truer is

th e disposition implied by that att i tude. To

quote Rosenberg ( 1 9 6 8 ) :

. . . for any particular social issue o r ob jec t as con*fronted by a sample draw n from some fairly uniformsector of the population (e.g., college students re-cruited for atti tude change experiments, or surveyrespondents recruited from some particular sector ofth e to tal nat ional populat ion) one can reasonablyexpect that those who show less in tra-at t i tudinalconsistency are, on the average, less invested in theissue and less likely to have a presently stableorientat ion toward it. (p. 88)

Rosenberg (1968) suggested two reasons

why attitudes characterized by low affective-

83

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 2/9

84 Ross N O R M A N

cognitive (structural) consistency are likelyto be unstable.

1. The procedure used for assessing the af-

fective and cognitive att i tude componentsforce the respondent to examine inconsist-

encies between his thoughts an d feelings ofwhich he was previously unaware. Once theinconsistency is salient, th e affective and/or

cognitive components may be changed tobring about greater consistency, an d suchchanges result in instability of both compo-nents in those attitudes which show initiallylo w structural consistency.

2 . Rosenberg also noted that individuals

who are not particularly interested in an issuerarely have well-articulated, well-thought-outatti tudes. Often such individuals will not re-veal their apathy by giving a neutral responsebut rather will take a position that is areaction to situational pressures or that repre-sents the consensus of some salient individualor imp ortant reference group (see also Con-verse, 1970). Such a responder may beregarded as presenting a "vacuous" or "in-

authentic" att i tude toward the issue. For thevacuous responder, neither statements con-cerning his beliefs nor his feelings are reflec-

tive of a true an d stable disposition, and hisattitude responses are apt to show con-siderable and apparently random fluctuat ionover time.

Previous research by Rosenberg (1968)has supported the notion that affective-cogni-tive consistency is associated with att i tudinalstability and resistance to persuasion at-

tempts . Given that attitudes characterized byaffective-cognitive consistency have greaterstability than less consistent attitudes, it

seems reasonable to anticip ate that they willhave greater validity as predictors of subse-quent behavior. While there is evidence thataffective-cognitive consistency results in rela-tively stable verbal attitudes, there is at

present no empirical support for the idea thatstructural consistency is related to predictive

validity. Such a relation, if it does exist,would be impor tant in ident i fying individualsw ho will act in accord with their attitudesfrom those whose stated attitudes are rela-tively superficial an d have few behavioralramifications. The above considerations leadto the following two hypotheses: (a) H y-pothesis 1—An affective or evaluative index

of an individual 's att i tude will be more pre-dictive of behavior when it is consistent withthe cognitive component of the attitude struc-

ture than when there is inconsistency betweenth e two. (b ) Hypothesis 2 — A cognitive indexo f an individual's attitude will be a betterpredictor of behavior when it is consistentwith th e affective component of the att i tudethan when there is inconsistency betweenthe two.

A series of three experiments were under-taken to test the importance of affective-

cognitive consistency. The first experimenttested the relation between structural con-

sistency and the predictive validity of verbalat t i tudes. The second was an a t tempt to repli-cate the first and also tested the relation ofstructural consistency to att i tude stability.The final experiment examined whether a f f e c -

tive-cognitive consistency also moderates sub-jects' tendency to conform to others in atti-tude-relevant behavior. Such an effect mightbe found if, as Rosenberg suggests, structu ralconsistency is related to an individual 's

involvement and certainty conc erning an issue.Volunteering to act as a subject for psycho-

logical research was chosen as the attitudeobject. This was an issue from which itseemed possible to obtain measures of at t i tudeand behavior u nd er c ircumstances whichwould not appear too unusual to the subjects.

METHOD

Development o f the Attitude Instrument

Att i tude theor is t s have t rad i t ional ly made a dis-

t inct ion between th e affect ive an d cogni t ive compo-nents of a t t i tude (e.g., Bern , 1970; Katz & Sto t land ,1959; Newcomb, 1959; Rosenberg & Hovland , 1960;Triandis, 1971), and several recent studies using

f a c t o r analy t ic techniques s u p p o r t this long-standingconceptual d i s t inct ion (Kernan & Tr e bb i , 1 9 7 3 ;

Kothandapani , 1971; Ost rom, 1969) . The affect ivecomponent in these theories refers to the individual 'sgeneral level of positive or negative feeling concern-

ing the issue. The cognit ive co mp onen t consists ofth e individual 's bel iefs concerning th e issue . W hen

the at t i tude issue is an act ion or behavior , therelevant cognit ive structure is considered to be the

individual 's bel iefs about the ins t rumental ut i l i ty o fth e act ion for the a t t a i n m e n t or b l o ck i n g of his or

her goals weighted by the value placed on such goals.In the first two studies a 9-point rat ing scale o f

overal l favorabil i ty t o w a r d act ing as a su b j ec t inpsychological research w as used as the affective

i ndex. This scale was anchored a t one ex t reme bythe term Very Favorable and at the o t h e r ex t r em eby the term Very Unfavorable, and a neutral point

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 3/9

ATTITUDES, CONFORMITY, AND BEHAVIOR 85

was explicitly indicated. Similar scales have beenused by K o t h a n d a p a n i ( 1971) , O s t r om ( 1969) , a ndRosenberg (1968) to measure the affective compo-

n e n t of the a t t i t ude . In the th i rd s tudy a 16-itemsemantic differential evaluative scale was used in

addit ion to the self-rating scale in or de r to m a k e th ecognitive an d affective scales of comparable length.

D e v e lopme nt of the cognit ive index w as some w ha tmore complex. Goals that appeared possible relevantto the att i tud e issue w ere selected by the- au tho rf rom lists presented by Carlson (1956), Rokeach(1968) , an d Rosen berg (1 9S 6). These goals weresubmi t t e d to a g r oup o f judges (4 psychology gradu-a te s tude n t s a nd 4 0 i n t r oduc tor y psyc ho log y s tu -dents) who were asked to indicate whether each goal

was relevant to the att i tude issue, whether therewere marked redundancies be tween the goals, an dwhether were goals relevant to the att i tude issuewhich were not included in the list. Based on the

j u d ges ' responses, th e n u m b e r o f goals w as reducedto 12 . Examples of goals in the final l ist include:help ing others, advancing se l f -know ledge , hav ingnew an d interesting kinds of experience, an d having

lots of free t ime .In the f inal version of the att i tude questionnaire,

these 12 goals were rated on two 9-point scales. O none the respondent was asked to indicate the extent

to which he believed each goal would b e achieved(o r b loc ke d) b y volunteer ing as a subject. This scalew as anchored by the phrase completely achieved at

one end (scored as +4) through a neutral point (0 ) ,to completely blocked ( — 4 ) at the other end. Theother scale w as used to evaluate each goal; this scalew as similar to the "belief scale" described above, bu tit was anchored by the terms extremely positive goal( + 4 ) a nd extremely negative goal ( — 4 ) . The over-al l cogni t ive index w as calculated by f inding thep r o d u c t of the two ratings fo r each goal an dsumming across goals (see Rosenberg, 1960). Spacewas also included at the end of the questionnairefo r subjec t s to inc lude an d rate on the two scalesgoals they con sidered relevant to the issue whichwere not in the list of 12 presented to t he m.

Nin e-po int ratin g scales were also provided in thethi rd s tudy for respondents to indica te how cer ta inthey felt a bou t the i r a t t i t ude : how impor ta n t the yconsidered the issue and how much thought theyhad given to the issue in the past.

Subjects

A to ta l of 2 4 2 s tude n t s f rom i n t r oduc tor y psyc ho l -ogy courses at the Universi ty o f Michigan were usedas subjec ts in the three studies. O f these, 48 subjec tswere dropped f rom var ious ana lyses in the studiesbecause of their absence at the t ime of an a t t i tudina l

or behaviora l measure , or suspic ion con cerning someaspect of the exper imenta l proc e dure .

Procedure

In a l l s tudies the ini t ia l a t t i tude quest ionna ire wasadministered during regular class meetings by the

inst ruc tors . The quest ionna ire w as described as part

of "a general survey being done within th e depar t -

m ent of psychology." Su bjects were asked to put

their names on the completed forms, since there w assome possibility that the department might want to

relate their answers to a later questionnaire. In Ex-

periment 2 -the attitudes scales were readministered

3 weeks af ter th e initial assessment.In the first two studies the behavioral measure

was taken 3 weeks af te r presentation of the att i tud equestionnaire (i n Exper iment 2 after presentation of

the second quest ionna ire ) . A n experimenter, intro-

duced as a graduate student in .psychology, was given

S minu te s at the beginning of a class session to re-cru i t volunteer subjects for a research project. Thestudents were -told that the study w ould take ap -proximately 1 hour of their time and would hope-

fu l ly prove to be interesting and educationally bene-

f icial. They were also told that they could not be

paid fo r their time an d that their par t ic ipa t ion inthe experiment would not c oun t toward the 3 hours

of "subject t ime" that was required as part of theircourse. Sign-up sheets were then passed around th eclasses so that students w ho were interested involunteering could indicate when they would be able

to come to the designated location for the experi-ment. Volunteers were given an oppor tun i ty to signu p for any t im e betw een 9 a.m. to S p.m. on anyone of three designated days. Students who signed

u p and d id indeed show up a t the proper place an dtime participated in an exper iment unre la ted tothis study.

The behavioral measure in Experiment 3 differedfrom that in the f irst two studies. The change w asmade in order to better co nt rol si tuation al pressuresrelevant to the behavior. Sub jects were recruitedfrom 'their classes to take part individually in an

exper iment as part of their course requirement. Theywere given a wide choice of times to sign upranging f rom 9 a .m. to 9 p .m. M ond ay throughFriday of a given week. Of the students who had

answered th e ini t ial at t i tude questionnaire, ID S

signed up in class as subjects for the study. Anaddi t iona l 29 s tudents were contac ted by te lephoneand scheduled to participate in the experiment. Of

the s tudents who had f i l led out the earlier atti-tude q uest ionna ire , 12 7 ac tua lly showed up for theexper iment .

When each subject arrived at the designated lab,he or she was seated in a waiting area to awaitbeing called by the experimenter. Another subject(i n reality an experimental confederate) also ar -rived in the lab at the same time. The confederatewas always of the same sex as the subject . B othth e subjec t and the confederate were greeted in thesame fashion by the experimenter and were taken toan area of the lab where numerous pieces of equip-

ment had been placed. A f t e r seating the subject andconfederate , the exp erim enter asked th em w hetherthey would be wi l l ing to re turn for an addi t iona ltw o experimental sessions after the first one had

been completed. The confederate was f irst askedwhether he or she was will ing to re turn, an d then

th e real subject was asked.

Under som ewhat similar circum stances Blake an d

M o u t o n ( 1 9 S 7) f o u n d that a naive subject was mor e

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 4/9

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 5/9

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 6/9

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 7/9

A T T I T U D E S , CONFORMITY, A N D B E H A V I O R 89

cumstances low-consistency subjects, because

of their presumed uncer ta inty and vacuity,should be more likely than high-consistency

subjects to follow the lead of the confederate.Whe n th e confederate's response is congruen twith the subject 's stated att i tude, both si tu-ational and attitudinal pressures imply the

same behavior, and there should therefore bea strong relation between the confederate 'san d subject's responses for all subjects re -gardless of whether they are acting primarilyon th e basis of their ow n a t t i t ude or conform-

ity. The summary of data relevant to thisconjecture is presented in Table 4. The rele-

vant chi-square value fo r test ing whether the

difference in conformi ty between high- an d

low-consistency subjects is greater under con-di t ions of incongruency than congruencyagain only approaches significance at the .10level, X

2( l ) =2.35.

Overall, the results of these analyses donot provide statistically significant support

fo r th e prediction that affective-cognitiveconsistency moderates conformi ty . However ,

because the results tend to approach signifi-cance, it might be advisable at this point to

withhold a final conclusion on the validity of

this hypothesis.

Necessity of the Affective-CognitiveDistinction

I t might b e argued that the dist inction b e-in g made between th e affective an d cognitivecomponents of a t t i t ude is meaningless or ir-

relevant. Both types of scales could be mea-suring th e same aspect of a t t i t ude and theaffective-cognitive consistency index might,according to this view, simply indicate th e

general reliability of the subjects' responsesacross a set of relatively homogeneous items.Consistent responding across test items hast radi t ional ly been considered an impor tant

index of test reliability an d might plausiblyfacil i tate accurate behavioral predictions.

However, the theory under lying the cur ren tresearch assumes that th e consistency betweenth e affective and cognitive responses (consid-ering the two as conceptually separate) givesmore in format ion about the predictive poten-t ial of verbal at t i tudes than would an index

of reliability based on a random division of

items. In order to check this assumption, in

TABLE 4

S U B J E C T S ' TENDENCY T O C O N F O R M T O C O N F E D E R A T E ' SR E S P O N S E U N D E R C O N D I T I O N S OF C O N G R U E N C Y

A N D I N C O N G R U E N C Y , S T U D Y 3

Sub jec t ' s response

ConformedDidn't c onf orm

ConformedD i d n ' t conform

High-consistencysubjec t s

Low-consistencysubjects

Congruent

2 19

2 29

Incongruent

1117

199

Note. Cong ruenc y defined by relation of confederate 's re-sponse to averaged affec t ive/cogni t ive index.

each of the three studies subjects were classi-fied as consistent or inconsistent in respond-in g across randomly divided sets of scales( ignoring the affective-cognitive dis t inc t ion) .N o significant differences in strength of the

att i tude-behavior relation or tendency to con-form were found between subjects classifiedas high and low in consistency on the basis

of random splits. This suggests that whileaffective-cognitive consistency is a conceptwith behavioral implications, in these studies

overall consistency (based on random group-ings of i tems) is not .

Consistency: Issue Specific or Response

Style?

A subject w ho generally gives little thoughtto the responses he makes on an attitude

questionnaire would probably score lower onour index of affective-cognitive consistencythan one who gives more careful considera-tion to his responses. If such a general re-

sponse style is substantially influencing th eindex of structural consistency, we mightexpect subjects to show significant correla-tions across issues in the amount of affective-cognitive consistency that they exhibit. In

order to check for the possibility of such ageneral response style, in the second study

affective and cognitive scales similar to thosealready described were developed an d admin-istered concerning tw o issues in addition tovolunteering: (a) prohibition of cigarette

smoking within classrooms and (b) a pro-posal that all students should be requestedto take at least one psychology course in theirprogram of studies. These additional scales

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 8/9

90 Ross N O R M A N

were administered just once at the t ime of thesecond testing session.

There was little relation between the af-

fective-cognitive consistency that the indi-vidual respondents displayed in reacting to

the various attitude issues. The correlationsvaried between — . 2 0 an d .18, none of themapproaching statistical significance. Thus

there is little evidence that the affective-cognitive consistency index is significantly

influenced by general response style.

Relation oj Affective-Cognitive Consistency

to Other Self-Ratings

Rosenberg (1968) suggested that degree of

affective-cognitive consistency is associatedwith a subject 's certainty and involvementconcerning an attitude issue. In Exper iment3, however, no significant relation w as found

between subjects' affective-cognitive consis-tency and their ratings of certainty, issue im-portance, or how much thought they hadgiven to the issue in the past (rs ranged be-

tween .03 and .14).

Order of Scales

In the questionnaire used in the first twostudies the affective scale items preceded the

i tems of the cogn itive scale. B oth possibleorders were used in E x p e r i m e n t 3 to see ifthere would be greater consistency betweenreported beliefs an d feelings using one se-quence rather than the other. There was no

evidence of a significant difference; the affec-

tive-cognitive correlation was .45 when the

cognitive items preceded the affective scaleand .54 when the reverse order ing was used.

GENERAL D I S C U S S I O N

O n the w hole, the results are supp ortive of

th e hypotheses that th e predictive validity ofboth cognit ive and affective scales is mod-erated by affective-cognitive consistency. All

three studies have shown th e predictive valid-i ty of the affective scale to be a funct ion of

it s consistency with an individual 's beliefstructure. The strongly suggestive finding ofthe first study and the statistically significantevidence of the third study indicate that th estrength of the relation between the cognitivescale and behavior is also a function of af-

fective-cognitive consistency.For the moment, however, several of the

results obtained in the second study remainsomewhat anomalous. In that study it was

found that the predictive validity of cognition

was not a function of its relation to affect.While the difference in this regard betweenth e second an d third studies might be ex-plained by the use of different affective in -dices, such an explanation does not accountfo r the somewhat different pattern of resultsbetween the first and second studies. In termsof design, th e only difference between th e

lat ter tw o studies is the use of an at t i tuderetest in the second. There is no apparent

reason why such a difference in design should

affect the overall validity of the cognitivescales; but a replication of the test-retest

design using a mult i - i tem measure of affect

might help clarify some of the issues involved.

Exper iment 3 does no t o f f e r statisticallysignificant suppor t for the prediction that

degree of affective-cognitive consistency af -

fects th e extent to which subjects will con-

form to others in behaviors that are relevantto the att i tude issue. This hypothesis was

derived from Rosenberg's (1968) suggestionthat structural consistency is related to thecertainty and involvement an individual feelswith regard to an attitude issue. The fai lureto f ind a relation between th e consistencyindex an d self-rating of uncer tainty or issueimportance somewhat compromises this postu-late. It may be that subjects found the mean-ing of the terms certain or important to beambiguous in the context in which i t wasused, and the measure w as therefore a poor

one. Possibly other techniques of measur ingcertainty or involvement (e.g. , th e "owncategories" technique suggested by Sherif &

Sherif , 1 9 6 7 ) m a y , in the fu tu re , be found tobe related to affective-cognitive consistency.

In the three experim ents reported there w asa 3-4 week delay between at t i tudinal and

behavioral measures. This was done for two

related reasons: (a) to m inim ize any experi-mental demand characterist ics that would in-

fluence subjects ' tendency to act in accordwith their stated attitudes and (b) becausethe major challenge to attitude researchers

appears to be the prediction of later behaviorin situations at least somewhat different from

the a t t itude m easuremen t se tt ing . How ever,

the design used leaves some ambiguity as to

whether the low predictive validity of the

8/8/2019 Cognitive Affective Theory

http://slidepdf.com/reader/full/cognitive-affective-theory 9/9

ATTITUDES, CONFORMITY, AN D BEHAVIOR 91

stated at t i tudes of subjects showing low af-fective-cognitive consistency is simply a func-

tion of the instability of these attitudes overtime. The possibili ty exists that att i tude mea-

sures for low-consistency subjects, while notpredictive of fu ture behavior, m ay predictbehavior displayed at a b o u t the same time as

the atti tude measure. Such a possibility re-

m ains to be investigated. There are, of course,great difficulties in get t ing contemporaneousatti tude and behavior measures in credible

circumstances that would not involve strongexperimental demand characteristics.

Despite these problems and unresolved

issues, th e results of the present studies indi-cate the value of assessing both the affective

and cognitive components of an att i tude be-fore attempting to make behavioral predic-tions. Consideration of affective-cognitiveconsistency is intended to supplement issuesemphasized in most past research on theattitude-behavior relation. Such research hasfocused on the importance of other disposi-tions and situational factors in addition toattitudes in the determination of behavior.

These "other variables" approaches havemade an important contribution, but it seemsprobable that through consideration of the

structural characteristics of att i tude, the pre-diction of social behavior can be improvedfurther. The most important contr ibut ion ofthe current research is in providing evidencethat the importance of attitudinal factors in

the determination of behavior may be a func-

tion of the structural characteristics of the

relevant attitudes.

R E F E R E N C E S

Ajzen, I., & Fishbein, M. At t i tud ina l and no rmat ivevariables as predictors of specific behaviors. Jour-nal of Personality and Social Psychology, 1973, 27,

41-57.

Bern, D. J. Beliefs, attitudes, and human affairs.

Belmont , Cal i f . : Brooks/Cole , 1970.

Blake, R. R. , & M ou ton , J . S . The s tudy of socialconduct wi th in th e f r a m e w o r k o f adaptation-level

theory . In M. Sherif & M. O. Wilson (Eds.),Emerging problems in social psychology. Nor-man : University of Oklahoma Book Exchange,

19S7.

Carlson, E. R. Att i tude change through modif icat ionof at t i tude s tructure . Journal of Abnormal an dSocial Psychology, 1956, 52 , 2S6-261.

Converse , P. E. At t i tudes and non-a t t i tudes : C on -t inua t ion of a dialogue. In E. R. T u f t e (Ed.), The

quantitative analysis of social problems. Reading,

Mass.: Addison-Wesley, 1970.Corey, S . M . Professed atti tudes and actual behavior.

Journal of Educational Psychology, 1937, 28, 271-280.

Fishbein, M. Attitude and the prediction of be-havior . In M. Fishbein (Ed.), Readings in atti-

tude theory and measurement. N ew York: Wiley,1967.

Katz, D., & St o t l a nd , E . A preliminary s ta tement toa theory of at t i tude s tructure and change. In S.

K o c h (Ed.), Psychology: A study of a science

(Vol . 3). New Y ork: McGraw-Hil l , 1959.

Kernan , J. B., & Trebbi, G. C. Att i tude-dynamics as

a hierarchical structure. Journal of Social Psy-chology, 1973, 89 , 193-202.

Ko thandapan i , V . Validat ion of feeling, belief, an din ten t ion to act as three com ponents of a t t itude ,and their contr ibut ion to the predic t ion of contra-ceptive behavior. Journal of Personality and So-cial Psychology, 1971, 19 , 321-333.

LaPiere, R. T. At t i tudes vs. actions. Social Forces,1934, 13, 230-237.

Newcomb, T. Individual systems of orientation. In

S. Koch ( E d . ) , Psychology: A study o f a science(Vol. 3). New Y o r k : McGraw-Hill, 1959.

Osgood, C. E., Suci, C. J., & T a n n e n b a u m , P. H.

Th e measurement of meaning. Urbana: Univers i tyof Illinois Press, 1957.

Os t rom, T. M. The relationship between the affective,

behavioral, and cognitive components of atti tudes.Journal o f Experimental Social Psychology, 1969,5, 12-30.

Rokeach , M. Beliefs, attitudes and values. S anFrancisco: Josey-Bass, 1968.

Rosenberg, M. J. Cognit ive s t ructure and attitudinal

affect . Journal o f Abnormal an d Social Psychol-ogy, 1956, S3 , 367-372 .

Rosenberg, M. J. A s tructural theory of attitude

dynamics . Public Opinion Quarterly, 1960, 24, 319-341.

Rosenberg, M. J. Hedon ism, inauthent ic i ty , and

other goals toward expansion of a consistency

theory . In R. P. Abelson et al. (Eds.), Theories ofcognitive consistency: A sourcebook. Chicago:Rand McNally, 1968.

Rosenberg, M. J., & Hovland , C. I. Cognitive, af-

fective, and behavioral co m ponen ts of attitudes.

In M. J. Rosenberg et al. (Eds.), Attitude organi-

zation and change. New Haven, Conn.: YaleUniversity Press, 1960.

Sherif , M., & Sherif , C. W. The own categoriesprocedure in attitude research. In M. Fishbein(Ed.), Readings in attitude theory and measure-ment. N ew Y ork: W iley , 1967.

Triandis, H. C. Attitude and attitude change. New

York: Wiley, 1971.Wicker , A. W. An examinat ion of the "other varia-

bles" explanation of attitude-behavior inconsis-tency. Journal of Personality an d Social Psychol-ogy, 1971, 19, 18-31.

Winer , B. J. Statistical Principles in ExperimentalDesign. N ew Y ork: M cGraw-Hil l, 1962 .

(Received March 6, 1974)