Effects of Need for Cognition on Message Evaluation, Recall, and … · 2019-07-08 · Effects of Need for Cognition on Message Evaluation, Recall, and Persuasion John T. Cacioppo

Journal of Personality and Social Psychology1983, Vol. 45, No. 4, 805-818

Copyright 1983 by theAmerican Psychological Association, Inc.

Effects of Need for Cognition on Message Evaluation, Recall,and Persuasion

John T. CacioppoUniversity of Iowa

Richard E. PettyUniversity of Missouri—Columbia

Katherine J. MorrisUniversity of Iowa

This research concerns the relation among need for cognition, message processing,and persuasion. Pairs of subjects holding approximately the same attitude towardinstituting senior comprehensive exams and differing widely in their scores on theneed for cognition scale were recruited to participate in the first study. In Experiment1, subjects read a set of either strong or weak arguments supporting the recom-mendation that senior comprehensive exams be instituted. Results revealed thatargument quality had a greater impact on the message evaluations and sourceimpressions provided by individuals high than by those low in need for cognitionand that subjects high in need for cognition reported expending more cognitiveeffort and recalled more message arguments regardless of argument quality. Themajor findings in Experiment 1 were replicated in Experiment 2 with a differenttopic (i.e., raising student tuition) and cover story. In addition, the inclusion of apostcommunication attitude measure revealed that the attitudes of individuals highin need for cognition were more affected by argument quality than those of subjectslow in need for cognition. Together, these studies document a reliable differenceamong individuals in their tendency to derive information from and elaborate onexternally provided message arguments.

Social psychology is replete with theoriesbased on the premise that people, if not com-monly enjoying, are at least commonly en-gaging in effortful problem solving as they steertheir course through their social environment.The emphasis on the view that people activelyprocess information is apparent, for example,in the vintage theories of social comparison(Festinger, 1954), cognitive consistency (Abel-son et al, 1968), and attribution (Heider, 1958;Kelley, 1967) and in the recent interest in"social cognition." This view contrasts withthe notion that "behavior is actually accom-

This research was supported by a University FacultyScholar Award. The authors are grateful to John T. Lan-zetta, Barbara L. Andersen, and Joseph A. Sidera for theircomments on this research. We are also grateful to MartinHeesacker and Chuan Feng Kao for their assistance inpreliminary research, and to Hai Sook Kim for her as-sistance in Experiment 2.

Requests for reprints should be sent to John T. Cacioppo,Department of Psychology, University of Iowa, Iowa City,Iowa 52242 or to Richard E. Petty, Department of Psy-chology, University of Missouri, Columbia, Missouri65211.

plished much of the time without paying at-tention to the substantive details of the 'in-formative' environment" (Langer, Blank, &Chanowitz, 1978, p. 635).

The present research focuses on a particularand pervasive component of our social envi-ronment: persuasive communications. Will(1982) estimated that the average Americanis exposed to more than 1,500 persuasive mes-sages daily from national advertisers alone.People have neither the resources to think ex-haustively about every persuasive appeal towhich they are exposed nor the luxury (orapparently the inclination) of being able toignore them all. Elsewhere, we have suggestedthat situational factors can play an importantrole in enabling and motivating individuals tothink extensively versus superficially about themerits of an appeal. In one such study (Petty& Cacioppo, 1979), subjects were exposed toa counterattitudinal advocacy containing ei-ther strong arguments (i.e., they elicited pri-marily favorable thoughts about the appeal ina pretest) or weak arguments (i.e., they elicitedprimarily unfavorable thoughts about the ap-

805

806 J. CACIOPPO, R. PETTY, AND K. MORRIS

peal in a pretest). Some subjects were led tobelieve the recommendation would affect thempersonally (high issue involvement), and otherswere led to believe it had few personal ram-ifications (low issue involvement). Results in-dicated that subjects engaged in greater scru-tiny of the information in the message underhigh than under low issue involvement. Con-sequently, argument quality was a more im-portant determinant of the postcommunica-tion attitudes of subjects who believed the rec-ommendation was personally important thanof those who believed it was unimportant.Similarly, Petty, Wells, and Brock (1976) ex-posed subjects to a strong or a weak set ofpersuasive arguments under conditions of lowor high distraction. Distraction disruptedmessage evaluation, and, as a result, subjects'attitudes were affected by argument qualitywhen distraction was low, but not when dis-traction was high (see Petty & Cacioppo, 1983,for a review of how other situational factorsaffect people's motivation and ability to engagein issue-relevant thinking).

We reasoned that just as there are situationalfactors such as distraction and issue involve-ment that influence the likelihood that indi-viduals will think about and elaborate on theexternally provided message arguments, so toomust there be dispositional factors governingmessage processing and, indirectly, persuasion.Our reading of the literature suggested thatthe individual difference described by Cohen,Stotland, and Wolfe (1955; Cohen, 1957, Note1) as need for cognition was a particularlypromising factor to examine. Indeed, althoughit was overlooked at the time, Cohen et al.'sinitial study on need for cognition, in retro-spect, may have been the first to demonstratethat individuals high in need for cognitionscrutinize communications more, and aremore affected by its cogency, than are subjectslow in need for cognition. Subjects read a storyabout an interview between a student and apotential employer and then rated the storyin terms of how much they liked it, how in-teresting they found it, and so forth. Somesubjects read a well-organized story, whereasothers read an ambiguous story. As might beexpected, the ambiguous story was evaluatedmore negatively overall than the structuredstory, but this discriminating judgment seemed

only to characterize individuals high in needfor cognition.

The objective tests used to gauge individualdifferences in need for cognition were neverdescribed in detail or published and are ap-parently no longer available, so our initial ef-forts were necessarily aimed at developing andvalidating an assessment instrument (Cacioppo& Petty, 1982; Cacioppo, Petty, & Kao, inpress). In the first study, we generated a poolof questions concerning a person's reactionsto demands for effortful thinking in a varietyof situations (Cacioppo & Petty, 1982). Whengenerating these, we specifically included itemsdescribing a variety of situations in whichpeople could choose to garner information,analyze available evidence, abstract from pastexperiences, or synthesize ideas; explicitly ex-cluded were items dealing with inner brood-ings, reverie, mystical or religious experience,mind wandering, and artistic imaginings.Hence, the need for cognition scale (NCS) wasdesigned to distinguish between individualswho dispositionally tend to engage in and enjoyeffortful analytic activity and those who donot. Following pilot testing to identify ambig-uous items in need of rewording or deletion,the set of items was administered to two groupsof individuals presumed to differ substantiallyin their tendency to engage in and enjoy ef-fortful cognitive endeavors (i.e., members ofa university faculty vs. assembly line workersin the neighboring communities). Criteria ofambiguity, irrelevance, and internal consis-tency were employed to select the items forthe NCS, and a factor analysis of these itemsconfirmed that one factor was dominant. Asecond study, in which the NCS was admin-istered to a more homogeneous population(undergraduate students in an introductorypsychology class), validated the factor structureobtained in the first study. Finally, correlationswith what were expected to be related (e.g.,open-mindedness) and unrelated (e.g., testanxiety) constructs provided discriminant andconvergent evidence for the validity of the NCS(Cacioppo & Petty, 1982, in press).

Evidence that individuals high in need forcognition do indeed enjoy relatively effortfulcognitive tasks even in the absence of feedbackabout performance was obtained in Study 4of Cacioppo and Petty (1982). Subjects used

NEED FOR COGNITION, MESSAGE PROCESSING, AND PERSUASION 807

either simple or complex instructions whileperforming a boring number-circling task, Af-terward, subjects expressed their attitudes to-ward the task. It was not surprising that sub-jects generally disliked the task, but a signif-icant interaction revealed that people high inneed for cognition tended to prefer the com-plex to the simple task whereas people low inneed for cognition tended to prefer the simpleto the complex task.

Having achieved a means for assessing needfor cognition, we were in a position to returnto the question that initiated our interest inthis disposition: Do individual differences inneed for cognition predict the likelihood thatrecipients will extract and cognitively elaborateon information from a persuasive commu-nication? Experiment 1 was designed as a con-ceptual replication of Petty, Harkins, and Wil-liams (1980), with the major difference beingthat we examined the role of a dispositional(need for cognition) rather than a situational(group size) factor on cognitive effort andmessage evaluation. Experiment 2 was sub-sequently conducted to examine the effects ofneed for cognition and argument quality onpersuasion.

Experiment 1

In Experiment 1, we varied the cogency ofthe message arguments supporting the rec-ommendation that seniors be required to passa comprehensive exam in their major as a re-quirement for graduation (a counterattitudinaladvocacy). Subjects, who were selected follow-ing preliminary testing on the basis of theirscores on the NCS and their initial attitudestoward the recommendation, were told thattheir task was to evaluate an editorial writtenby a journalism student. They then read eithera strong or a weak set of arguments supportingthe recommendation to institute senior com-prehensive eMms. We expected the resultswould be reminiscent of those obtained whensituational factors such as issue involvement(Petty & Cacioppo, 1979), group size (Pettyet al, 1980), or distraction (Petty et al., 1976)were manipulated. Specifically, although weexpected the version of the message containingstrong arguments to be evaluated more posi-tively than the version containing weak ar-

guments by all subjects, this difference wasexpected to be stronger for individuals whocarefully scrutinize the message (i.e., subjectshigh in need for cognition) than for those whodo not (i.e., subjects low in need for cognition).

Method

Preliminary TestFive hundred twenty-seven undergraduate students in

introductory psychology classes participated in a prelim-inary testing session during which they completed a batteryof tests. Included in the battery were the 34-item NCS(Cacioppo & Petty, 1982) and an attitude survey regardingcampus issues (e.g., raising student tuition, instituting seniorcomprehensive exams). Students were classified as beinglow, moderate, or high in need for cognition based ontheir responses to the NCS, and pairs of students whodiffered dramatically in need for cognition (i.e., who fellin the bottom or the top tripartite of scores) and whoreported approximately the same attitude toward institutingsenior comprehensive exams were recruited for possibleparticipation in Experiment 1. Students did not knowwhat facet of the preliminary testing was responsible fortheir eligibility, and the experimenter did not know thesubjects' levels of need for cognition when they arrivedat the laboratory.

Design and ProcedureOne hundred fourteen undergraduates (S3 men and 61

women) ultimately participated in a 2 (Need for Cogni-tion) X 2 (Argument Quality) factorial. Subjects were testedin groups of 5 to 16 in cubicles constructed so that nosubject could have visual or verbal contact with any othersubject. It was possible, therefore, to run all experimentalconditions simultaneously.

At the beginning of the experimental session, subjectswere instructed that the University of Iowa's branch ofthe National Society of Journalism Teachers was under-going an evaluation of its programs and that the psychologydepartment was assisting by studying the subjective impactof various journalistic styles. The subjects were furthertold that they would be reading an editorial prepared bya journalism student for possible broadcast and publication.Subjects were asked simply to provide an evaluation ofthe editorial.

Immediately following these background comments, thesubjects read either a strong or a weak set of argumentsfavoring the recommendation that senior comprehensiveexams be instituted as a prerequisite for graduation attheir university. Afterward, subjects completed a dependentvariable booklet and participated in another study that isirrelevant to the present discussion. At the conclusion ofthe study, subjects were debriefed, thanked, and dismissed.

Independent VariablesNeed for cognition. Subjects' classification as being high

or low in need for cognition was based on their responsesto the NCS, obtained several weeks earlier in the prelim-


inary test (see above). Although the separation of the pre-liminary test from Experiment 1 by several weeks mayhave weakened this factor, we reasoned that need for cog-nition would not be an important recipient factor to con-sider should it not represent a fairly stable individual dif-ference. Students whose NCS scores fell within the middletripartite were not recruited to participate in Experiment1 to minimize any errors in classification that could resultwhen, as in our preliminary test, there are clusters ofindividuals scoring at and near the median value.

Argument quality. All subjects read an editorial thatthey believed had been prepared by a journalism studentfor possible publication in the local newspaper. The edi-torials were approximately 300 words long and containedeither six strong or six weak arguments for instituting seniorcomprehensive exams. These sets of arguments, whichwere essentially those described as the "strong" and "veryweak" communications in Petty et al. (1980), were pretestedand adjusted as necessary to assure that the strong versionelicited primarily favorable thoughts and the weak versionelicited primarily unfavorable thoughts from both subjectshigh and subjects low in need for cognition.1

Dependent Variables

Message evaluation. We postulated that subjects high,in contrast to those low, in need for cognition would dis-criminate between the strong and the weak versions of themessage arguments when evaluating the editorial. In orderto assess message evaluation, subjects were asked to ratethe editorial, using the following 9-point scales: (a) "Towhat extent do you feel the communication made its pointeffectively?" (1 = not at all, 9 = completely); (b) "To whatextent did you like the communication?" (1 = not at all,9 = very much); (c) "To what extent do you feel thecommunication was convincing?" (1 = not at all, 9 = veryconvincing); (d) "Considering both content and style, howwell written was the communication?" (1 = poorly written,9 = very well written); and (e) "Would you judge thereasons given for supporting the recommendation in thepreceding message as being . . ." (1 = very poor anduncompelling, 9 = very good and compelling). The averageof each subject's responses to these five questions servedas the general index of message evaluation.

Cognitive effort. We also postulated that people high,in contrast to those low, in need for cognition would bethinking more extensively about the message arguments.Two questions were embedded in the postexperimentalquestionnaire to test this hypothesis: "To what extent wereyou trying hard to evaluate the student's performance asa journalist?" (1 = not at all, 9 5» very much) and "Howmuch effort did you put into evaluating the communi-cation?" (1 = none, 9 = very much). The average of eachsubject's responses to these two questions served as theindex of cognitive effort.

Recall of message arguments. In pilot research (Petty,Note 2), subjects high in need for cognition recalled moremessage arguments than did subjects low in need for cog-nition. In order to determine whether this was a reliablefinding, subjects in the present study were asked to list asmany of the message arguments as they could remember.Subjects, who had not expected this recall test, were given5 minutes to list arguments. Subsequently, two judges in-dependently determined the number of arguments each

subject correctly recalled. Judges were unaware of the ex-perimental hypotheses and any subject's level of need forcognition, though of course judges knew whether they werescoring recall from the strong or the weak version of themessage. An item listed by a subject was counted correctlyrecalled if it expressed one of the six arguments containedin the appropriate message. Arguments listed twice by asubject were counted only once. The interrater reliabilitywas .74, and the judges' counts were averaged prior toanalyses to derive the index of message recall.

Ancillary questions. Several additional items wereembedded in the questionnaire to maintain the cover storyand for exploratory purposes. Subjects were asked, "Wouldyou judge the source of the preceding message as being. . ." (1 = very low in expertise, 9 = very high in expertise);"Would you judge the source of the preceding message asbeing. . ." (1 = very untrustworthy, 9 = very trustworthy);"Would you judge the recommendation made in the pre-ceding message as being . . ." (1 = very unimportant toyou personally, 9 = very important to you personally);and "Would you characterize the preceding message asbeing. . ." (1 = not at all worth thinking about, 9 = verymuch worth thinking about).

Results

Because not every student we recruited toparticipate in Experiment 1 appeared, a one-way analysis of variance (ANOVA) was per-formed first to determine whether the subjectswho participated in Experiment 1 differed intheir initial attitude toward instituting seniorcomprehensive exams. The results confirmedthat the attempt to match subjects on thismeasure was successful (Ms = 4.02 and 4.06on a 9-point scale for subjects low and thosehigh in need for cognition, respectively; F <1). Next, subjects' responses in Experiment 1

1 Argument quality has been defined operationally suchthat "strong" arguments elicit more favorable than un-favorable statements, as assessed by the thought-listingtechnique (cf. Cacioppo & Petty, 1981) and "weak" ar-guments elicit more unfavorable than favorable thoughtsabout the recommendation (e.g., see Petty & Cacioppo,1979; Petty et al., 1976). In pretests for the present study,we observed that a "strong" message that we had usedpreviously again tended to elicit more favorable than un-favorable thoughts when need for cognition was ignored.However, when the profile of cognitive responses was ex-amined within each need-for-cognition group, we foundthat people high in need for cognition were actually gen-erating slightly more unfavorable thoughts (e.g., counter-arguments) than favorable thoughts to the strong message.Modifying the strong arguments (e.g., by fabricating moreimpressive and believable statistics) proved sufficient toelicit primarily favorable thoughts from both individualshigh and those low in need for cognition.


were submitted to 2 (Need for Cognition) X2 (Argument Quality) ANOVAs.2

Cell means for message evaluation are de-picted in Figure 1. As expected, the ANOVAyielded two significant effects on this measure:a main effect for argument quality, F(l,110) = 160.86, p < .001, which simply re-fleeted that the strong arguments were judgedmore positively (M = 5.86) than the weak ar-guments (M = 2.26), and a Need for Cogni-tion X Argument Quality interaction, F(l,110) = 22.45, p< .001 (see Figure 1). Pairwisecomparisons by the Duncan multiple-rangetest revealed that individuals high in need forcognition provided more discriminating judg-ments of these externally provided argumentsthan did individuals low in need for cognition.

, We have found that people scrutinize a per-suasive message more when they believe therecommendation is personally relevant thanwhen they believe it is irrelevant (e.g., Petty& Cacioppo, 1979). In the present study, wereasoned that individuals high in need for cog-nition were more likely to cogitate generallyand about persuasive communications specif-ically, but it is possible that they viewed thepersuasive appeal about senior comprehensiveexams as more personally relevant. If the latterwas the case, then a main effect for need forcognition should obtain on the appropriateancillary measure included in this study.Analysis of the measure regarding issue in-volvement ruled out this alternative interpre-tation. However, a main effect for argumentquality, F(l, 110) = 4.46, p < .05, indicatedthat the recommendation was considered moreimportant when supported by strong ratherthan weak arguments, and a Need for Cog-nition X Argument Quality interaction, F(l,110) = 4.47, p < .05, reflected that subjectshigh in need for cognition rated the recom-mendation as being more important when itwas supported by strong rather than weak ar-guments, whereas argument quality did notalter the ratings of subjects low in need forcognition (see Table 1).

, The results of analysis of the recall data,which revealed one significant effect, were alsoconsistent with the notion that individuals highin need for cognition extracted more from,and thought more about, the message argu-ments. Subjects high in need for cognition re-called substantially more arguments (M =

Positiver.o

6.0

I 5.0

\ 4.0

1 3.0S

2.0

1.0Negative

3 StrongArguments

•WeakArguments

low needfor cognition

high needfor cognition

Figure 1. Message evaluation as a function of argumentquality and need for cognition.

3.98) than their counterparts (M = 2.91), F( 1.,110) = 13.62, ;»< .01. Analysis of the measureof cognitive effort also revealed that subjectshigh in need for cognition reported expendingmore effort thinking about the editorial (M =5.93) than did subjects low in need for cog-nition (M = 4.98), F(\, 110) = 9.61, p < .01.A main effect for argument quality signifiedthat subjects reported thinking more aboutthe weak than the strong message arguments,F(l, 110) = 6.78, p < .05, but a Need forCognition X Argument Quality interactionindicated that this effect primarily character-ized subjects high in need for cognition, F(l,110) = 6.48, p < .05 (see Table 1).

Yet another suggestion that individuals highin need for cognition were thinking more aboutthe message arguments comes from their judg-ments of the communicator. Recall that sub-jects knew nothing about the communicatorother than that he or she was a journalismstudent who had written an editorial that theyhad read. The subjects' impressions of thecommunicator are summarized in Table 1, Itis not surprising that subjects perceived thecommunicator to be more expert, F(l,110) = 84.38, p < .01, and trustworthy, F(l,110) = 135.69, p < .001, after they had beenexposed to the strong rather than the weakarguments. However, these effects were clearly

2 Initial attitudes were also equivalent across conditionswhen a 2 (Need for Cognition) X 2 (Argument Quality)analysis of variance was performed (all F.s < 1).


Table 1Ratings of the Persuasive Communication as a Function of Need for Cognition and Argument Quality

Low need for cognition High need for cognition

Variable

Cognitive effortImportance of recommendationTrustworthinessExpertiseWorth considering further

Weakargument

5.00.5.80.,b3.64b

2.88b4.64,

Strongargument

4.96.S.67.J,5.08C

4.83C

5.96b

Weakargument

6.60b

4.90.1.83.1.07.3.62.

Strongargument

5.25.6.53b5.86C

5.58C

6.89b

Note. Means in a given row with dissimilar subscripts differ at p < .05 by the Duncan multiple-range test.

stronger for individuals high than those lowin need for cognition. The Need for Cogni-tion X Argument Quality interactions weresignificant for both trustworthiness, F(\,110)= 18.05, p < .01, and expertise, F(l,110) = 21.00, p < .01. Means and pairwisecomparisons for these measures are summa-rized in Table 1.

Analysis of the ancillary question regardinghow worthwhile it would be to think furtherabout the recommendation yielded two sig-nificant effects. A main effect for argumentquality indicated that subjects viewed the rec-ommendation as deserving more considerationwhen it was supported by strong (M = 6.43)than rather than weak (M = 4.13) arguments,F(l, 110) = 44.04, p < .001, and a Need forCognition X Argument Quality interactionagain indicated that this effect was more ap-parent for subjects high than for those low inneed for cognition, F(\, 110) = 6.81, p < .05(see Table I).3

Experiment 2

The most common strategy employed ininvestigations of the role of recipient factorsin persuasion is to search for a particular per-1

sonality trait that renders people generally sus-ceptible or resistant to influence (e.g., Janis &Field, 1959). Eagly (1981) referred to this asthe "personality strategy." A second, termedthe "attitude change strategy," involves deriv-ing hypotheses about individual differences ingeneral persuasibility from postulates in atti-tude theories (e.g., Eagly & Teelak, 1972).Eagly suggested that neither strategy has beenparticularly successful: The former fails toconsider the underlying cognitive processes

postulated to be responsible for persuasion,and the latter fails to consider stable individualdifferences in cognitive ability, motivation, andstyle. The present research illustrates a thirdhybrid approach wherein the focus is on in-dividual differences in attitudinal processing.The results of Experiment 1 showed that sub-jects high in need for cognition were morelikely to (a) discriminate between the strongand the weak versions of the message argu-ments when evaluating the merits of the sup-portive arguments, (b) be affected by the qual-ity of these externally provided message ar-guments when forming impressions of thecommunicator, (c) recall more message ar-guments regardless of their cogency, and (d)report expending more cognitive effort thanwere individuals low in need for cognition.

There is, however, a potential problem withthe generalizability of the results in Experiment

3 Reanalysis of these data with sex of subject as a factordid not change any of the conclusions reported above. ASex of Subject X Need for Cognition interaction was ob-tained on the measure of cognitive effort, a result indicatingthat although people high in need for cognition reportedexpending more cognitive effort than individuals low inneed for cognition, this difference was stronger for malesthan for females, F( 1,106) = 6.63, p < .05. Two additionalfindings involving the factor of sex of subject were obtainedin the analyses of the ancillary measures. First, a maineffect for sex of subject indicated that female subjects judgedthe recommendation to be more important personally(M = 6.30) than did male subjects (M = 5.17), 7^1,106) =7.74, p < .05. Second, a Sex of Subject X Need for Cog-nition X Argument Quality interaction on the measure oftrustworthiness indicated that the two-way interactionnoted above was more apparent for males than for females,F(\, 106) = 6.13, p < .05. No other tests involving thisfactor were significant.


1. Analyses of the attitude surveys that the527 students completed during the preliminarytesting suggested that individuals high in needfor cognition possess more positive attitudestoward intellective issues (i.e., recommenda-tions with implications for effortful thinking)but similar attitudes toward nonintellective is-sues, when compared with students low in needfor cognition. Analyses of the data from these527 students revealed that those high in needfor cognition held more positive attitudes to-ward instituting senior comprehensive exams(high, M = 4.05; low, M = 3.38; p < .05) andless positive attitudes toward adopting nar-rower college curricula (high, M = 1.40; low,M = 2.99; p < .05), whereas their attitudestoward sleeping fewer hours in a day (high,M = 3.52; low, M = 3.11; ns), raising studenttuition (high, M = 2.86; low, M = 2.65; ns),and extending international boundaries at sea(high, M = 3.48; low, M = 3.54; ns) did notdiffer. Students high and students low in needfor cognition were selected for Experiment 1in part on the basis of their initial attitudestoward senior comprehensive exams to avoidconfounding need for cognition with subjects'initial attitudes toward the recommendation,an effort that was accomplished successfully(see Footnote 2). This subsequent analysis ofthe attitude data from our preliminary testsuggests, however, that the subjects classifiedas high or low in need for cognition who par-ticipated in Experiment 1 may not be entirelyrepresentative of their constituencies. In orderto eradicate any concern about the validity ofthe results obtained in Experiment 1, a partialreplication was conducted by using a rec-ommendation toward which students high andthose low in need for cognition felt equallyantagonistic: raising student tuition. Strongand weak versions of a persuasive message weredeveloped in pretests during the semester inwhich Experiment 2 was conducted. The re-sponses of both students who scored high andstudents who scored low on the NCS weremonitored in pretests to assure that the strongand the weak arguments were effective.

A second and important aim of Experiment2 was to determine the effects of argumentquality and need for cognition on attitudestoward the recommendation rather than to-ward the communication per se. Pretestingreplicated previous surveys showing that stu-

dents high and those low in need for cognitiondid not differ in their attitudes toward raisingstudent tuition at their university, so onlypostcpmmunication measures of attitude wereadministered in Experiment 2. This allowedthe data to be collected from each subject ina single session without the potentially obfus-cating effects of having subjects specify theirattitudes toward an issue, read a persuasivecommunication, and express their attitudes onthe issue again.

Finally, data from subjects falling in themiddle third of the distribution of scores onthe NCS were deleted prior to analyses, sothat, as in Experiment 1, we minimized themeasurement error in classifying subjects ashigh or low in need for cognition.4

Method

Design and Procedure

One hundred ten female undergraduates participatedin a 2 (Need for Cognition) X 2 (Argument Quality) fac-torial. Subjects were tested in groups of 8 to 16 in cubiclesconstructed so that no subject could have visual or verbalcontact with any other subject, a condition making it pos-sible to run all of the experimental conditions simulta-neously.

Subjects were told that the Office of Student and Ac-ademic Affairs at the University of Iowa was Devaluatingits policies and was seeking recommendations about pos-sible changes. Subjects were told that policy statementshad been prepared for possible broadcast and publicationin the local media to inform and to obtain the reactionsof people in the university'community. Subjects were askedto read one of the policy statements, rate it for readability,and respond to the questions about the policy statementwhich followed it in their booklet.

Following the reading of this background material, sub-jects read either a strong or a weak set of arguments favoring

4 The data from subjects whose scores on the NCS fellnear the median were not analyzed in the initial tests ofthe hypothesis because post hoc analyses of the pilot studiessuggested that subjects from a homogeneous population(e.g., from introductory psychology classes) whose NCSscores fell near the median responded to a variety of ex-perimental tasks in a similar fashion whether their scorewas marginally above or below the median. It was assumeda priori that by deleting the data from these subjects inthe present study, the responses to the persuasive com-munication attributable to need for cognition would bemore apparent. As a follow-up of this reasoning, we con-ducted reanalyses, including the data from subjects scoringin the middle tripartite on the NCS as a group with mod-erate need for cognition, and found the same pattern ofresults, with a few F ratios involving the factor of needfor cognition slightly reduced.


the recommendation that student tuition be increased atthe University of Iowa, completed a dependent variablebooklet, were debriefed, thanked, and dismissed.

Independent Variables

Need for cognition. The NCS was included last in thedependent variable booklet to allow for the subsequentselection of subjects who were high or low in need forcognition. Following the completion of data collection,subjects were ranked on the basis of their scores on thisscale, and the data from those in the upper or lower tri-partite were retained for analysis (see Footnote 4).

Argument quality. Though all subjects read an ap-proximately 300-word message justifying the proposal thatstudent tuition be increased, half of the subjects read aset of four strong arguments, and half read a set of fourweak arguments. Briefly, in the strong version of the mes-sage, it was noted that despite increasing enrollments, in-flation and interest rates were responsible for the erosionof the university's buying power, which had now reachedits lowest level in 20 years. Second, it was stated that facultyand staff had already sacrificed a great deal in acceptingsmall annual increases in salary. It was argued that severaltop faculty members, researchers, and creative artists hadrecently left the university for better paying positions atother universities (a well-publicized fact on campus) andthat others were threatening to leave if their salaries werenot maintained at levels at least comparable with thoseat other financially stricken peer universities. Third, sta-tistics were cited indicating that the number of recruitingcompanies and the average starting salaries of undergrad-uates were a direct function of the prestige of the universitygranting their degree. The implications of an eroding fi-nancial base and the loss of quality faculty for the qualityof teaching, the intellectual atmosphere at the university,and the prestige of the university were discussed briefly.Finally, it was noted that increasing undergraduate en-rollments, coupled with the hiring freeze imposed on theuniversity (at that time), had resulted in the overcrowdingof classes and the reduction in the quality and diversityof instruction offered to students. Increasing student tuitionwas deemed a necessary means of dealing with these prob-lems.

By contrast, in the weak message it was argued that thetaxpayers of Iowa had been bearing the burden of highereducation too long. It was suggested that tuition shouldbe increased in order to take the unfair burden off taxpayersand place the burden on the shoulders of those who chosehigher education as a priority in these financially difficulttimes. Second, it was stated that the multimillion-dollarsports arena under construction at their university hadbeen funded primarily by contributions from residentsand businesses in the state and that students would be theprimary benefactors of the new arena. Hence, it was arguedthat state residents had already contributed significantlyto students at the university and that students should con-tribute more to cover the cost of their own education.Third, it was argued that social programs, which werebeing cut by the federal government, were a more importantpriority for limited state resources (e.g., being of moreuse to the state residents) than was the university. It wasnoted that the average faculty salary was considerably higherthan the average salary of a state resident "not on the

government payroll" and that if students felt that qualityfaculty and staff were leaving for higher salaries at com-parable universities, then the students should contributethe lion's share of the funds needed to keep faculty salarieson a par with those in peer institutions. Finally, it wasreasoned that at least part of the money from tuition in-creases should be spent on improving the physical ap-pearance of the university. It was suggested, for instance,that funds were needed to improve the roadways to andfrom the new arena to minimize the traffic congestionthat was expected to accompany sports events at the uni-versity.

Dependent VariablesAttitude index. Immediately after reading the editorial,

subjects read the following:

Since your personal views on the desirability of the policyrecommendation about which you read might influencethe way you rate these materials, a measure of yourown opinion is desired.

Subjects used four semantic differentials (good/bad, ben-eficial/harmful, wise/foolish, favorable/unfavorable) toindicate their own feelings about the recommendation toraise student tuition. Nine-point scales were used through-out the dependent variable booklet. In some scales theendpoint designating a positive response was labeled witha positive attribute (e.g., beneficial), and in some scalesthis endpoint was labeled with a negative attribute (e.g.,foolish). Prior to analyses, responses to scales were trans-formed so that 1 signified a negative (or low level of an)attribute and 9 designated a positive (or high level of an)attribute. Each subject's responses to the attitude measureswere averaged to form an attitude index.

Message evaluation. Subjects were asked to "rate thequality of the arguments used to support the policy rec-ommendation" (1 = poor and unconvincing arguments,9 = good and convincing arguments), which served as themeasure of message evaluation.

Ancillary measures. Subjects' responses to two itemsembedded in the questionnaire were averaged to determinethe extent to which the subjects felt the recommendationwas personally important. These items were as follows:"How likely is it that the University of Iowa will institutethe policy recommendation about which you read?" (1 =very unlikely, 9 = very likely); and "How personally relevantor important did you find the policy recommendation?"(1 = not at all relevant, 9 = very relevant). Subjects pro-vided data on their impressions of the communicator byresponding to the question "Regardless of how you feltabout what the author had to say, how qualified did youthink the author was to speak on the topic?" (1 = not atall qualified, 9 = very qualified). Finally, three items wereincluded immediately following the attitude measure inorder to maintain the cover story, assure that the strongand the weak sets of message arguments did not differ intheir peripheral cue value, and ensure that the argumentswere just as easy to comprehend and follow by subjectslow as by those high in need for cognition. Subjects wereasked to rate the readability of the policy statement theyhad read on scales labeled "difficult to understand/easyto understand," "hard to follow/easy to follow," and "con-tains complex structure/contains simple structure" (seePetty, Cacioppo, & Goldman, 1981). No differences were


Table 2Ratings of the Persuasive Communication as a Function of Need for Cognition and Argument Quality

Low need for cognition High need for cognition

Variable

Message evaluationQualification of source

Weakargument

6.94b

6.38b

Strongargument

7.63b

6.00b

Weakargument

3.94.4.18,

Strongargument

7.26b

6.42b

Note. Means in a given row with dissimilar subscripts differ at p < .05 by the Duncan multiple-range test.

found on these items, and hence they are not discussedfurther.

Results. An ANOVA of message evaluation yieldedthe expected main effect for argument quality,F(\, 70) = 18.32, p < .05, a result indicatingthat the strong arguments were rated morepositively (M = 7.45) than the weak arguments(M = 5.49). A main effect for need for cog-nition was also obtained, F(\, 70) = 12.25,p < .05, due to the fact that individuals highin need for cognition rated the message morenegatively overall (M = 5.69) than individualslow in need for cognition (M = 7.30). Thesemain effects were qualified by the expectedNeed for Cognition X Argument Quality in-teraction, F( 1,70) = 8.70, p < .05. A summaryof pairwise comparisons by the Duncan mul-tiple-range test are presented in Table 2. Thesetests revealed that, as in Experiment 1, ar-gument quality had a larger impact on theevaluations of subjects high than of those lowin need for cognition. This result again suggeststhat the former individuals scrutinized morethe message arguments that were provided.

The analysis of the attitude index yielded amain effect for argument quality, F(l, 70) =27.53, p < .01, which shows that the strongarguments were more persuasive (M = 6.36)than the weak arguments (M = 3.93) and amarginally significant Need for Cognition XArgument Quality interaction, F(l, 70) = 3.54,p < .07. In order to provide a more sensitivetest of the specific experimental hypothesis, acontrast was calculated to test whether the dif-ference between attitudinal responses to thestrong and those to the weak message was largerfor the high than for the low need for cognitiongroup.5 This directional contrast was signifi-cant, t = 1.91, p < .03, one-tailed (see Fig-ure 2).

It is of interest to note that the pattern ofmeans on the attitude measure was similar,though not identical, to that found for messageevaluation (r = .52, « = 74, p < .05). Wereasoned that if subjects high in need for cog-nition were more likely to derive their attitudethrough a considered evaluation of the argu-ments central to the recommendation, thenthere should be a stronger association betweenmessage evaluations and attitudes in subjectshigh than in those low in need for cognition.In order to evaluate this notion, a separatecorrelation was calculated between these mea-sures in the high and the low need for cognitiongroups. As expected, this correlation was sig-nificantly larger in the high need for cognitiongroup (r = .70, « = 37) than in the low needfor cognition group (r = .22, n = 37; Z =2.68, p < .05).

To further test the notion that it was therelative insensitivity of individuals low in needfor cognition to variations in argument qualitythat determined their failure to show the at-titude effects, we conducted two analyses ofcovariance (ANCOVAS). In the first, postcom-munication attitudes served as the criterionmeasure, and message evaluation served as thecovariate; in the second, message evaluationserved as the criterion, and postcommunica-tion attitudes served as the covariate. The re-sults revealed that the significant Need forCognition X Argument Quality interaction formessage evaluation remained significant whenattitudes served as the covariate, F(l, 68) =5.24, p < .03, whereas any hint of a significantNeed for Cognition X Argument Quality in-

5 This contrast was based on past research focused onthe effects of argument quality and situational factors (e.g.,distraction, issue involvement). A recent review of thisresearch can be found in Petty and Cacioppo (1983).


Positive75

65

5,5

3.5

Negative

3 StrongArguments

low needfor cognition

high needfor cognition

Figure 2. Postcommunication attitudes as a function ofargument quality and need for cognition.

teraction on the attitude measure was elimi-nated (F < 1) when the measure of messageevaluation served as the covariate. The resultsof these analyses too are consistent with theview that need for cognition affects messageprocessing and thereby affects people's sus-ceptibility or resistance to persuasion.

Analyses of subjects' ratings of how per-sonally involving was the recommendationyielded one main effect. Subjects rated the rec-ommendation as more involving after readingthe strong (M - 3.08) than after reading theweak (M = 1.78) version of the message, F(l,70) = 9.12, p < .01. Finally, analyses of thesubjects' ratings of the qualifications of thecommunicator revealed that the Need forCognition X Argument Quality interaction wasreplicated, F(l, 70) = 5.75, p < .05. As inExperiment 1, the impressions formed by in-dividuals high, but not low, in need for cog-nition about the communicator were affectedby argument quality (see Table 2). No othertest was significant.

General Discussion

We began the present investigation withthree observations: (a) A premise in many so-cial psychological theories, and particularly inmany theories of attitude change, is that peoplecommonly deliberate about information intheir social environment (cf. Insko, 1967; Shaw& Costanza, 1982; West & Wicklund, 1980);(b) persuasive communications are so perva-sive that people cannot think extensively aboutall to which they are exposed (Miller, Maruy-ama, Beaber, & Valone, 1976; Petty & Ca-

cioppo, 1981); and (c) recipient factors, suchas those described by Cohen et al. (1955),Murray (1938), Murphy (1947), and otherearly personality and social psychologists, maybe predictive of the extent to which externallyprovided arguments for a recommendation arescrutinized. The present investigation was de-signed to explore this third postulate.

Social psychologists have long noted thatthere are not only situational factors that affectpeople's tendencies to deliberate about infor-mation from their social environment but alsodispositional factors that govern the likelihoodthat individuals will scrutinize social infor-mation (cf. Maslow, 1943; Murphy, 1947;Murray, 1938). Indeed, it is noteworthy in ret-rospect that early evidence for the third pos-tulate listed above can be found in the first(and, heretofore, the only) experiment pub-lished on the relation between need for cog-nition and persuasion (Cohen, 1957). In Coh-en's study, 35 undergraduates indicated theirattitudes toward instituting the stricter scoringprocedure of grading on a curve, and eachstudent's level of need for cognition was mea-sured. Approximately 1 month later, the stu-dents were introduced to a communicator whowas described as a faculty member who hadrecently chaired the departmental committeeinvestigating grade inflation at their university.Students were told that he was there to speakabout the current grading problem at the uni-versity and to obtain their views on the prob-lem. About half of the students heard the com-municator outline the grading problem andthen provide information about how gradingon the curve was a judicious solution; the re-maining students heard the arguments forgrading on the curve before the grading prob-lem was described. Cohen reasoned that out-lining the problem prior to providing the pos-sible solution would result in more attitudechange than the reverse order of presentation;he also reasoned that subjects high in need forcognition would be more motivated than theircounterparts to think about the issue and,consequently, would evidence a weaker orderof presentation effect following a short delay.His results supported both predictions.

The present studies extend Cohen's obser-vations in several important respects. Messageevaluation was examined in Experiments 1and 2, recall was obtained in Experiment 1,

NEED FOR COGNITION, MESSAGE PROCESSING, AND PERSUASION •815

impressions of the communicator were col-lected in Experiments 1 and 2, and postcom-munication attitudes were assessed in Exper-iment 2. Various topics, messages, and mea-sures have been used in these studies, and theneed for cognition of subjects was assessed ina purportedly unrelated investigation in Ex-periment 1 and at the conclusion of Experi-ment 2. For all these variations, the patternsof results are quite consistent. The manipu-lation of argument quality had a larger impacton the message evaluations, impressions of thecommunicator, and attitudes of individualshigh in need for cognition than of those lowin need for cognition. Individuals high in needfor cognition also recalled more message ar-guments from both the strong and the weakversions of the message, and they reported ex-pending more cognitive effort in deliberatingabout the message to which they were exposedthan did their counterparts. Moreover, thecorrelation between message evaluation andpostcommunication attitudes was significantlyhigher for individuals high than for those lowin need for cognition, and statistical,analysesto examine the causal sequence of messageevaluation and attitudes supported the notionthat the scrutiny and evaluation of the messageinfluenced postcommunication attitudesrather than vice versa. Finally, it should benoted that in a study in which we used au-diotapes rather than printed communications,we again found that subjects high, in contrastto subjects low, in need for cognition providedmore discriminating evaluations of the strongand weak versions of the message argumentsand recalled more message arguments regard-less of their quality (see review by Cacioppo& Petty, in press). Collectively, these data sug-gest that individuals high in need for cognitionare more likely to extract information fromand think about externally provided messagearguments than individuals low in need forcognition.

There are several limitations that should benoted, however. Need for cognition may con-stitute one factor that affects the likelihoodthat an individual will deliberate about themerits of the arguments for a recommendation,but we do not mean to suggest it is the only,or perhaps even the most important, factor.Situational factors are particularly powerfuldeterminants of individuals' motivation and

ability to consider the meaning, implications,and nuances of persuasive arguments (cf. Petty,Cacioppo, & Heesacker, in press). Moreover,other dispositional factors, such as intelligence(cf. McGuire, 1968) or cognitive development(Inhelder & Piaget, 1958; cf. Neimark, 1982),may also prove to be predictive of elaborationlikelihood in persuasion contexts. It has been'our working assumption that need for cog-nition represents a motivational tendencywhich, though not invariant, develops throughan individual's experiences with complex cog-nitive endeavors and, for the most part, reflectsthe intrinsic rewards the individual derivesfrom these types of efforts. Intelligence, incontrast, may perhaps best be viewed as anability factor that delimits the boundaries ofan individual's deliberations. The factor ofprior knowledge about an issue, at least if ex-perimentally manipulated, may function as yetanother, more limited ability factor governingthe extent to which an individual can thinkabout and elaborate on arguments for a rec-ommendation (e.g., Cacioppo & Petty, 1980;cf. Wood, 1982). Therefore, need for cognitionis viewed here as a contributory rather thana necessary or sufficient cause for the likeli-hood of high elaboration.

It may be questioned whether individualshigh and those low in need for cognition differin terms of their general intelligence. To assessthis, we administered the NCS and the Shipley-Hartford (1940) scale to more than 100 un-dergraduates drawn from the same subjectpopulation that was employed in this research.The results revealed that need for cognitionwas unrelated to both abstract reasoning (r -—.03) and verbal reasoning (r = .15). In asubsequent replication by Morris, Bachman,Bromwell, and Sterling (Note 3), again in-volving more than 100 students, need for cog-nition was again found to be unrelated to ab-stract reasoning (r = .12) and weakly relatedto verbal reasoning (r = .21). Hence, althoughit is reasonable that more intelligent individ-uals are more likely to experience success indifficult cognitive endeavors and, therefore,come to be characterized by high levels of needfor cognition, there appear to be at most onlysmall differences in intelligence <between in-dividuals who are classified as high and thosewho are classified as low in need for cognitionin this subject population.


Although further research in which mea-sures of need for cognition and mental abilitiesare both obtained would be informative, itseems unlikely that the small potential differ-ences in intelligence between individuals clas-sified as high and those classified as low inneed for cognition could account for the ob-tained differences in message evaluation,source evaluation, reported cognitive effort,and attitude change. Eagly and Warren (1976)found that verbal intelligence could influenceboth a recipient's ability to understand a mes-sage and his or her susceptibility to persuasion.In their study, subjects high, in contrast tothose low, in intelligence understood andtended to agree more with complex arguments,whereas the opposite pattern emerged for sim-ple arguments. In Experiment 2 of the presentresearch, we specifically asked subjects to ratehow easy or difficult the message was to un-derstand, how easy or difficult it was to follow,and the extent to which the message was char-acterized by a simple or a complex structure.No differences were found on any of thesemeasures, results suggesting that the high- andthe low-need-for-cognition subjects wereequally able to understand and follow the mes-sage arguments.

A second caveat concerns the generaliza-bility of these data. In these studies, includingCohen's (1957) study, students were con-fronted with relatively involving, counterat-titudinal appeals. As the perceived personalimportance of a recommendation decreases,people tend to move from careful deliberationsof the merits of the arguments for a recom-mendation to simpler, less cognitively taxingmeans of evaluating the recommendation (Ca-cioppo & Petty, 1980; Chaiken, 1980; Petty &Cacioppo, 1979, in press; Petty et al., 1981;Petty, Cacioppo, & Schumann, in press). It isonly reasonable that lowering issue involve-ment would reduce the elaboration likelihoodof both individuals high and those low in needfor cognition, though at any given level of issueinvolvement it is individuals high in need forcognition who should think more about anissue. However, exceptions to this principleshould include appeals that are so importantthat individuals high and those low in needfor cognition would think as much as possibleabout them and appeals that are so unim-

portant that neither group would think aboutthem.

Finally, this research raises several questionsabout the relation between need for cognitionand yielding. The data suggest that individualshigh in need for cognition are more likely tothink about the arguments presented whenderiving their attitudes toward an issue, but itis not clear from these studies whether theywould also think more about variations in pe-ripheral cues. Petty et al. (1981) exposed sub-jects to either strong or weak persuasive ar-guments emanating from either an expert oran inexpert source. In addition, the advocacywas presented to some subjects as being per-sonally consequential and to others as beinginconsequential. The results showed that theattitudes of subjects who believed that the rec-ommendation would affect them personallywere influenced by the quality of the argu-ments presented in support of the recom-mendation, whereas the attitudes of subjectswho believed the advocacy was not particularlyinvolving were influenced by the expertise ofthe source rather than the quality of the ar-guments. Similarly, Petty and Cacioppo (inpress) demonstrated that message factors suchas the apparent number of arguments sup-porting a recommendation could also serve asa persuasion cue (e.g., the more arguments,the better the recommendation) when moreobvious cues were held constant across con-ditions. Subjects were exposed to a set of eitherthree or nine arguments that were either strongor weak in quality. In addition, half of thesubjects believed that the recommendation waspersonally important, and the remaining sub-jects believed that it was unimportant. As ex-pected, the attitudes of subjects who believedthe recommendation was involving were de-termined by the quality of the arguments pre-sented, whereas those of the remaining subjectswere determined by a feature of the appealthat was apparent following a more superficialanalysis of the persuasive communication: thesimple number of arguments.

These data, which were obtained in responseto the manipulation of situational factors inthe persuasion context, support the existenceof central and peripheral persuasion processes(see Petty & Cacioppo, 1981). It remains forfuture research to determine exactly how the


dispositional factor of need for cognition bearson our analysis of persuasion, however. Sub-jects low in need for cognition, for example,may. be more likely to travel the peripheralroute to persuasion because, in their effort toderive a "reasonable" position on the issuewhile minimizing the expenditure of cognitiveresources, they utilize obvious cues in the per-suasion context (e.g., the "apparent" numberof arguments for the recommendation).. It isalso possible, however, that individuals low inneed for cognition are less likely to think aboutchanges in superficial as well as substantivefeatures of a persuasive appeal, and this wouldmake them no more likely to travel the pe-ripheral route to persuasion than individualshigh in need for cognition.

Reference Notes

1. Cohen, A. R. Need for cognition and the structure ofsocial situations. Paper presented at the InternationalConference on Human Relations, Berg-en-dal, TheNetherlands, September 1956.

2. Petty, R. E. The need for cognition: Application to per-suasion. Paper presented at the meeting of the AmericanPsychological Association, Los Angeles, August 1981.

3. Morris, K., Bachman, E., Bromwell, G., & Sterling, S.The need for cognition: Relationship to other personality

factors. Unpublished manuscript, University of Iowa,1982.

References

Abelson, R. P., Aronson, E., McGuire, L., Newcomb,T. M., Rosenberg, M. J., & Tannenbaum, P. H. (Eds.).Theories of cognitive consistency: A sourcebook. Chicago:Rand McNally, 1968.

Cacioppo J. X, & Petty, R. E. Sex differences in influ-enceability: Toward specifying the underlying processes.Personality and Social Psychology Bulletin, 1980, 6,651-656.

Cacioppo, J. T., & Petty, R. E. Social psychological pro-cedures for cognitive response assessment: The thought-listing technique. In T. Merluzzi, C. Glass, & M. Genest(Eds.), Cognitive assessment. New \brk: Guilford, 1981.

Cacioppo, J. T, & Petty, R. E. The need for cognition.Journal of Personality and Social Psychology, 1982, 42,116-131.

Cacioppo, J. T, & Petty, R. E. The need for cognition:Relationship to attitudinal processes. In R. P. McGlynn,J. E. Maddux, C. D. Stoltenberg, & J. H. Harvey (Eds.),Social perception in clinical and counseling psychology.Lubbock, Tex.: Texas Tech University Press, in press.

Cacioppo, J. T, Petty, R. E., & Kao, C. F. The efficientassessment of need for cognition. Journal of PersonalityAssessment, in press.

Chaiken, S. Heuristic versus systematic information pro-cessing and the use of source versus message cues in

persuasion. Journal of Personality and Social Psychology,1980, 39, 752-766.

Cohen, A. R. Need for cognition and order of commu-nication as determinants of opinion change. In C. I.Hovland (Ed.), The order of presentation in persuasion.New York: Yale University Press, 1957.

Cohen, A. R., Stotland, E., & Wolfe, D. M. An experi-mental investigation of need for cognition. Journal ofAbnormal and Social Psychology, 1955, 57, 291-294.

Eagly, A. H. Recipient characteristics as determinants ofresponses to persuasion. In R. E. Petty, T. M. Ostrom,& T. C. Brock (Eds.), Cognitive responses in persuasion.Hillsdale, N.J.: Erlbaum, 1981.

Eagly, A. H., & Teelak, K. Width of latitude of acceptanceas a determinant of attitude change. Journal of Person-ality and Social Psychology, 1972,2.?, 388-397.

Eagly, A. H., & Warren, R. Intelligence, comprehension,and opinion change. Journal of Personality, 1976, 44,226-242.

Festinger, L. A theory of social comparison processes. Hu-man Relations, 1954, 7, 117-140.

Heider, F. The psychology of interpersonal relations. NewYork: Wiley, 1958.

Inhelder, B., & Piaget, J. The growth of logical thinkingfrom childhood to adolescence. New York: tiasic Books,1958.

Insko, C. A. Theories of attitude change. New York: Ap-pleton-Century-Crofts, 1967.

Janis, I. L., & Field, P. B. Sex differences as a personalityfactor related to persuasibility. In C. I. Hovland & I. L.Janis (Eds.), Personality and persuasibility. New Haven,Conn.: Yale University Press, 1959.

Kelley, H. H. Attribution theory in social psychology. InD. Levine (Ed.), Nebraska Symposium on Motivation(Vol. 15). Lincoln: University of Nebraska Press, 1967.

Langer, E., Blank, A., & Chanowitz, B. The mindlessnessof ostensibly thoughtful action: The role of "placebic"information in interpersonal interaction. Journal ofPersonality and Social Psychology, 1978, 36, 635-642.

Maslow, A. H. A theory of human motivation. Psycho-logical Review, 1943, 50, 370-396.

McGuire, G. Personality and attitude change: An infor-mation processing theory. In A. G. Greenwald, T. Brock,& T. Ostrom (Eds.), Psychological foundations of atti-tudes. New York: Academic Press, 1968.

Miller, N,, Maruyama, G., Beaber, R. J., & Valone, K.Speed of speech and persuasion. Journal of Personalityand Social Psychology, 1976, 34, 615-625.

Murphy, G. Personality: A biosocial approach to originsand structure. New York: Harper, 1947.

Murray, H. A. Explorations in personality. New Vbrk: Ox-ford University Press, 1938.

Neimark, E. D. Adolescent thought: Transition to formaloperations. In B. Wolman (Ed.), The handbook of de-velopmental psychology. Englegood Cliffs, N.J.: Prentice-Hall, 1982.

Petty, R. E., & Cacioppo, J. T. Issue involvement canincrease or decrease persuasion by enhancing message-relevant cognitive responses. Journal of Personality andSocial Psychology, 1919,37, 1915-1926.

Petty, R. E., & Cacioppo, J. T. Attitudes and persuasion:Classic and contemporary approaches. Dubuque, Iowa:W. C. Brown, 1981.


Petty, R. E., & Cacioppo, J. T. Central and peripheralroutes to persuasion: Application to advertising. In L.Percy & A. Woodside (Eds.), Advertising and consumerpsychology. Lexington, Mass.: D.C. Heath, 1983.

Petty, R. E., & Cacioppo, J. T. The effects of involvementon responses to argument quantity and quality: Centraland peripheral routes to persuasion. Journal of Person-ality and Social Psychology, in press.

Petty, R. E., Cacioppo, J. T, & Goldman, R. Personalinvolvement as a determinant of argument-based per-suasion. Journal of Personality and Social Psychology,1981, 41, 847-855.

Petty, R. E., Cacioppo, J. T, & Heesacker, M. Central andperipheral routes to persuasion: Application to coun-seling. In R. P. McGlynn, J. E. Maddux, C. D. Sto'l-tenberg, & J. H. Harvey (Eds.), Social perception inclinical and counseling psychology. Lubbock, Tex.: TexasTech University Press, in press.

Petty, R. E., Cacioppo, J. T, & Schumann, D. Centraland peripheral routes to advertising effectiveness: Themoderating role of involvement. Journal of ConsumerResearch, in press.

Petty, R. E., Harkins, S. G., & Williams, K. D. The effectsof group diffusion of cognitive effort on attitudes: An

information processing view. Journal of Personality andSocial Psychology, 1980,38, 81-92.

Petty, R. E., Wells, G. L., & Brock, T. C. Distraction canenhance or reduce yielding to propaganda: Thought dis-ruption versus effort justification. Journal of Personalityand Social Psychology, 1976, 34, 874-884.

Shaw, M. E., & Costanza, P. R. Theories of social psy-chology (2nd ed.). New York: McGraw-Hill, 1982.

Shipley, W. C. Shipley-Hartford Scale: Manual of directions.Hartford, Conn.: Neuro-Psychiatric Institute of theHartford Retreat, 1940.

West, S. G., & Wicklund, R. A. A primer of social psy-chological theories. Monterey, Calif.: Brooks/Cole, 1980.

Will, G. F. But first, a message from . . . Newsweek, May10, 1982, p. 98.

Wood, W. Retrieval of attitude-relevant information frommemory: Effects on susceptibility to persuasion and onintrinsic motivation. Journal of Personality and SocialPsychology, 1982, 42, 798-810.

Received December 10, 1982Revision received March 11, 1983

Search Opens for JPSP Section Editor:Interpersonal Relations and Group Processes

The Publications and Communications Board has opened nominations for the edi-torship of the Interpersonal Relations and Group Processes section of the Journal ofPersonality and Social Psychology for the years 1986-1991. Robert Helmreich, thecurrent editor-elect, will serve as editor from 1984 to 1985. Candidates must bemembers of APA and should be available to start receiving manuscripts in early 1985to prepare for issues published in 1986. To nominate candidates, prepare a statementof one page or less in support of each nomination. Submit nominations no later thanFebruary 1, 1984, to:

Jacqueline GoodchildsDepartment of PsychologyUCLALos Angeles, California 90024

Documents

Effects of Need for Cognition on Message Evaluation, Recall, and … · 2019-07-08 · Effects of Need for Cognition on Message Evaluation, Recall, and Persuasion John T. Cacioppo