distraction can enhance or reduce yielding to propaganda ...€¦ · distraction can enhance or...

Journal of Personality and Social Psychology1976, Vol. 34, No. 5, 874-884

Distraction Can Enhance or Reduce Yieldingto Propaganda: Thought Disruption

Versus Effort Justification

Richard E. Petty, Gary L. Wells, and Timothy C. BrockThe Ohio State University

Two experiments were conducted to test competing accounts of the distraction-persuasion relationship, thought disruption and effort justification, and alsoto show that the relationship is not limited to counterattitudinal communica-tion. Experiment 1 varied distraction and employed two discrepant messagesdiffering in how easy they were to counterargue. In accord with the thoughtdisruption account, increasing distraction enhanced persuasion for a messagethat was readily counterarguable, but reduced persuasion for a message thatwas difficult to counterargue. The effort notion implied no interaction withmessage counterarguability. Experiment 2 again varied distraction but the twomessages took a nondiscrepant position. One message elicited primarily favor-able thoughts and the effect of distraction was to reduce the number offavorable thoughts generated; the other, less convincing message elicited pri-marily counterarguments, and the effect of distraction was to reduce counter-arguments. A Message X Distraction interaction indicated that distractiontended to enhance persuasion for the counterarguable message but reducepersuasion for the message that elicited primarily favorable thoughts. Theexperiments together provided support for a principle having greater gen-erality than the Festinger-Maccoby formulation: Distraction works by in-hibiting the dominant cognitive response to persuasive communication and,therefore, it can result in either enhanced or reduced acceptance.

Festinger and Maccoby's (1964) well-known demonstration that distraction in-creases acceptance of propaganda assumedthat a person tends to engage in activecounterarguing when confronted with a per-suasive message with which he disagrees, andthat distraction could inhibit such counter-

This article is based on a master's thesis submittedto the Graduate School of The Ohio State Universityby the first author. The experiments were carried outwith the assistance of the second author under thesupervision of the third author. The help of thesiscommittee members Eric Knowles and Thomas M.Ostrom is acknowledged with thanks. Thanks arealso due Russ Dern, Marilyn Sosenko, and LynnOliver. Portions of Experiment; 1 were presented atthe annual meeting of the Midwestern PsychologicalAssociation, Chicago, May 1975; and Experiment 2at M.P.A., Chicago, May 1976.

Donald F. Roberts, Robert S. Baron, Robert B.Cialdini and John T. Cacioppo made valuable com-ments on an earlier draft of this manuscript.

Requests for reprints should be sent to Richard E.Petty, Social Psychology Doctoral Program, TheOhio State University, 404C West 17th Avenue, Co-lumbus, Ohio 43210.

arguing, thereby weakening resistance to themessage.

The present study had two aims. The firstaim was to examine a more general formula-tion than that proposed by Festinger andMaccoby (1964). We reasoned that if thedominant cognitive response to a communica-tion was counterarguing, then distractionwould lead to enhanced persuasion by inter-fering with the counterarguing process; but,if the dominant cognitive response to a com-munication was agreeing or favorable cogni-tive responses rather than counterarguments,distraction would inhibit these favorablethoughts and lead to lowered acceptance.1

The second aim was to evaluate two com-peting explanations of distraction-acceptanceeffects, namely thought disruption versuseffort justification. Numerous experiments em-

1 Festinger and Maccoby (1964) included condi-tions in their experiment in which agreeing commentsmight be expected, but no significant effects due todistraction were obtained for these groups, and nonewere hypothesized.

874

DISTRACTION AND PERSUASION 875

ploying a wide variety of distractions and adiverse list of message topics have replicatedthe original Festinger and Maccoby result(e.g., Insko, Turnbull, & Yandell, 1975; Keat-ing & Brock, 1974; Kiesler & Mathog, 1968;Osterhouse & Brock, 1970; Rosenblatt, 1966;Rule & Rehill, 1970; Silverman & Regula,1968; Zimbardo & Ebbesen, 1970; Zimbardo,Snyder, Thomas, Gold, & Gurwitz, 1970).However, the nature of the mediating processis still disputed by some. Osterhouse andBrock (1970), Keating and Brock (1974),and Insko et al. (1975) have provided thestrongest support for the thought disruptionhypothesis; using a procedure developed byBrock (1967) and Greenwald (1968) for thedirect measurement of cognitive responses,these studies showed that as the level of dis-traction increased, acceptance increased, whilethe number of counterarguments decreased.

Baron, Baron, and Miller (1973) recentlyreviewed many of the distraction-persuasionstudies mentioned above as well as some un-published data, and concluded that there wasonly one explanatory mechanism which couldrival the counterargument disruption hypothe-sis. This alternative explanation stems fromthe theory of cognitive dissonance (Festinger,1957). Baron et al. (1973) argued that in atypical distraction experiment, subjects per-ceive that they have chosen to participate inan experiment which requires that they exerteffort in order to attend to a discrepant mes-sage. Subjects can justify this effort expendi-ture by adopting the attitude expressed in thecommunication (Cohen, 1959; Wicklund,Cooper, & Linder, 1967). Baron et al. (1973)contended that all of the distraction manipu-lations to date have been confounded witheffort, and further that this confounding maybe unavoidable. They then suggested that "theprincipal means of circumventing this sort ofmethodological problem is to devise an experi-ment for which the contending explanationsmake competing predictions" (p. 320).

To meet these two aims, two experimentswere conducted. Experiment 1 employed dis-crepant messages and was primarily designedto deal with the effort challenge to thethought disruption hypothesis by devising asituation in which the two theories would

make competing predictions.2 Experiment 2was primarily designed to test the more gen-eral distraction formulation: that distractionworks by disrupting the dominant cognitiveresponse to propaganda, and thus could eitherenhance or reduce persuasion.

EXPERIMENT 1

Two discrepant messages were specificallyconstructed for Experiment 1. Both messagesargued that the tuition at Ohio State Univer-sity should be increased by 20%, from thethen current $266 per quarter to $320 perquarter, but the messages differed in theirpresentation of five key arguments. One mes-sage was designed to contain points that werelogically sound, easily defendable, and morecompelling than previous tuition increasepropaganda (e.g., Osterhouse & Brock, 1970)and thus be difficult to counterargue. Another,easy-to-counter argue message, was designed tobe more open to refutation and skepticism,and be less compelling than previous mes-sages.

Given that the messages differ predomi-nantly in how easy or difficult they are tocounterargue, it becomes possible to evaluatethe two major explanations of the distraction-acceptance effect. The effort justification ex-planation, as derived from dissonance theory,states that with increasing levels of distrac-tion (and thus effort), there is increasedacceptance of both messages since for eachmessage, subjects have to justify their effortexpenditure. Besides this main effect predic-tion favored by Baron (Note 1), a differenteffort prediction is possible, if one considersthe quality of the message arguments in con-junction with the effortful distraction. Sincehearing good arguments advanced againstone's position should be more dissonancearousing than hearing poor arguments (Klein-hesselink & Edwards, 1975; Lowin, 1967), itcan be predicted that subjects hearing thedifficult-to-counterargue communication will

2 The "competing predictions" approach was takenearlier in an experiment by Kiesler and Mathog(1968), but see the Baron et al. (1973, p. 320) cri-tique arguing that both theories could handle theobtained results.

876 R. PETTY, G. WELLS, AND T. BROCK

experience more difficulty justifying theireffort than those hearing the easy-to-counter-argue message. This line of reasoning predictsthat the slope of agreement as a function ofdistraction (effort) will be steeper for thedifficult-to-counterargue message. In any case,effort hypotheses expect increased acceptancewith increased distraction for both messages.

The counterargument disruption hypothesis,on the other hand, would expect an enhanc-ing effect only for the easy-to-counterarguemessage, because presumably only for thismessage are counterarguments generated thatcan be disrupted by distraction. In an experi-ment designed to test an unrelated hypothesis,Regan and Cheng (1973) obtained data rele-vant to the current hypothesis. Regan andCheng constructed two messages on each oftwo topics and found that distraction en-hanced persuasion for "simple" (short, easilyunderstood, but unconvincing) messages, butdecreased persuasion for "complex" (long,difficult to understand, but convincing if un-derstood) messages. However, due to the con-founding of at least three variables (lengthof message, ease of understanding, and per-suasiveness) the results offer only tentativesupport for the argument that distractiontended to enhance persuasiveness for the sim-ple messages, because counterarguments weredisrupted; and reduce persuasion for the com-plex messages, because favorable thoughtswere disrupted.

The following predictions were made forExperiment 1. More counterarguments wouldbe generated to the easy-to-counterargue mes-sage than to the difficult-to-counterargue mes-sage, distraction would inhibit counterargu-mentation only for the easy-to-counterarguemessage, and a significant Message X Dis-traction interaction would be found on theattitude dependent variable indicating thatdistraction increased acceptance for the easy-to-counterargue message, but not for thedifficult-to-counterargue message. Thus, theresults were expected to support thought dis-ruption rather than effort justification.

MethodProcedure

One hundred and thirty-two introductory psychol-ogy students at The Ohio State University (O.S.U.)

chose to participate in the experiment by selecting itfrom over IS other possible experiments on a "sign-up" bulletin board. Students are informed in theirintroductory classes that they have the right toleave any experiment in progress and still receivecredit that partially fulfills a course requirement.Additionally, subjects may fulfill the course require-ment by writing a term paper rather than by experi-mental participation. Thus, the level of choice feltby experimental subjects was at least as high (orhigher) than that for subjects participating in pre-vious distraction-persuasion experiments.

The design was a 4 X 2 X 2 factorial with fourlevels of distraction (no, low, medium, and high),two levels of message counterarguability (easy anddifficult to counterargue), and two orders of thedependent variable booklet (list thoughts before ex-pressing attitude, and express attitude before listingthoughts). Subjects were run in groups of 6 to 12 incubicles constructed so that no subject could havevisual contact with any other subject. During anyone session, in which one level of distraction wasrun, half of the subjects heard the easy-to-counter-argue message over headphones, while half of thesubjects heard the difficult-to-counterargue message.After listening to the communication under theappropriate distraction condition and completing thedependent variable booklets, subjects were debriefed,thanked, and dismissed.

Independent Variables

Message counterarguability. In brief, the difficult-to-counterargue message stated that the tuition hikewas recommended after an intensive 2-year investi-gation, that Ohio was 50th in state expenditures oneducation, that improvements were needed in thelibrary system, that a mass transit system could beinitiated, and studies had proven that if O.S.U. couldupgrade its quality, the average salary of its gradu-ates would increase substantially. The easy-to-counterargue message stated that the tuition increasewas recommended after a 2-month study, that Ohiowas 10th in expenditures on education, that moretrees should be planted on campus, that the univer-sity should start a bus system, and that improvedlighting was needed in classrooms to cut down onstudent headaches. Each message was approximately3 minutes long. Subjects' ratings in a pretest indi-cated that the messages did not differ in the extentto which they were "difficult to understand," were"hard to follow," or possessed "complex structure."

Distraction and effort. On arrival, each subject washanded a "monitor recording form." The distractiontask required subjects to monitor a large screenwhich hung from the ceiling in front of the room.The screen was from 2 to 5 m in front of the sub-jects, which required them to view it at an angle ofapproximately 45°. The subjects were informed viaprerecorded instructions that the experiment con-cerned the ability to do two things at once. Thesubjects were told that while they listened to themessage, an X would flash periodically in one of


the four quadrants of the screen. Their task was torecord on "monitor recording forms" in whichquadrant the X appeared. If the X appeared in theupper left-hand quadrant, they were to record theletters "UL" to signify upper left; if upper right,"UR"; if lower left, "LL"; and if lower right,"LR."

Twenty-five slides were prepared for each quad-rant, and the slides were randomly ordered whenplaced in the projector. For subjects in the low-dis-traction conditions, the Xs flashed at 15-sec intervalsthroughout the message. For medium distraction, theXs flashed at 5-sec intervals, and in the high-distrac-tion conditions the Xs flashed at 3-sec intervals.Each slide remained viewable until the next slideappeared. Subjects in the no-distraction conditionswere given instructions identical to those in other,conditions, but were told that "For now, no Xs willflash while you are listening to the message." Sub-jects in the other distraction conditions were not toldhow many Hashes to expect during the message. Allsubjects were instructed to try to pay close atten-tion to the message.

It is important to note that the manipulation ofdistraction was also intended to be a manipulationof effort expenditure. Thus, the conditions mightalternatively have been called no, low, medium, andhigh effort.

Order of the dependent variable booklets. For halfof the bookk-ts, subjects answered the communicationacceptance ratings before they recorded their thoughtsand ideas on the message; for the other half, thisorder was reversed. This was done to check on thepossibility raised by Miller and Baron (1973) thatthe awareness of one's attitude, as made salient bythe attitude scales, would influence subsequentcounterargument production.

Dependent Variables

Measures of attitude. Two measures of attitudetoward the proposed tuition increase were included.The first question was general and asked, "In gen-eral, to what extent did you agree with the speaker'srecommendations?" Subjects were to respond on a7-point scale where 7 indicated "completely agree."The second question was specific and asked, "Thecurrent tuition at O.S.U. is $266. What do you thinkthe appropriate level should be?" Subjects were torecord a dollar figure in a blank beside the question.Subjects' responses to the two attitude questionswere converted to standard scores and summed.

Measures of cognitive responses. Employing a pro-cedure adapted from Brock (1967) and Osterhouseand Brock (1970), subjects were given 2\ minutes towrite their thoughts on the topic of increasing tui-tion after they had listened to the message.3 Afterrecording their thoughts, subjects were instructed torate their ideas in a manner adapted from Cullen(1968) and Cialdini, Levy, Herman, Kozlowski, andPetty (in press). Subjects rated their ideas aseither + (in favor of a tuition increase), — (op-posed), or 0 (neutral). Each idea that a subject

wrote was submitted to 2 judges for scoring as eithera counterargument or a favorable thought.4 Thejudges agreed on 92% of the statements. In cases ofdisagreement between judges, the subject's rating wasemployed. Thus, each subject was assigned a counter-argument and a favorable thoughts score by thejudges. Only the judge-assigned scores were analyzed.

Measures of recall. When subjects had completedboth the attitude- and thought-listing sections of thebooklet, they were given 2 minutes to attempt torecord as many message arguments as they couldremember. Each booklet was rated by two judges(r = .86). An argument had to correctly summarizeone of the arguments that appeared in the appro-priate message to be counted. Repetitions of the sameargument were not counted. When judges disagreedin their independent ratings, ratings were obtainedfrom two additional judges and a majority decisionwas reached among the four.

Unequal n and Missing Data

Unequal cell size resulted from subjects failing tokeep appointments. Cell ws ranged from 5 to 11. Un-expectedly, 10 subjects failed to respond to one ormore questions. Rather than eliminate these sub-

3 Roberts and Maccoby (1973) suggested thatmeasures of counterarguing taken during presentationof a message "are more easily viewed as possiblemediators" of attitude change "since they precedesubjects' taking an opinion position" (p. 294). Sincethe current study involved distraction, and it washypothesized that the distraction would interfere withsubjects' processing of the message, it would havedefeated the purpose of the distraction to havegiven subjects an opportunity to collect and recordtheir thoughts during the presentation of the mes-sage. In any case, half of the subjects were given anopportunity to express their thoughts after the mes-sage, but before they responded to the attitudescales.

4 Counted as counterarguments were statementsdirected against a tuition increase which mentionedspecific unfavorable consequences of raising the tui-tion, statements which suggested alternative methodsof raising money instead of increasing tuition, state-ments which challenged the validity of argumentsraised in the message, and statements of affect op-posing the tuition increase. Counted as favorablethoughts were statements in favor of the increasewhich mentioned specific favorable consequences ofraising the tuition, statements ruling out alternativeways of raising the money, statements which sup-ported the validity of arguments raised in the mes-sage, and statements of affect supporting the tuitionincrease. Similar examples of the same statementwere counted once. The correlation between judges'ratings of counterarguments and subjects' ratingswas .79 (p < .01)., and between judges' ratings offavorable thoughts and subjects' ratings, .82, (p <.01).


jects from all analyses, it was decided to eliminatea subject from an analysis only for those measuresfor which no response was recorded.

Results


Order of the dependent variable booklet. Itwas not expected that the order of the book-lets would make any difference in the patternof results. A 4 X 2 X 2 unweighted-meansanalysis of variance on each dependent varia-ble generally confirmed this expectation. Maineffects and interaction effects involving orderwere generally uninteresting and irrelevantfor the purposes of this study (Petty, 1975).Therefore, on dependent variables for whichorder had no effects, the data were collapsedacross order in subsequent analyses.

Message counterarguability. An average of2.06 counterarguments was generated to thedifficult-to-counterargue message, while anaverage of 2.89 counterarguments was gen-erated to the easy-to-counterargue message,F(l ,124) = 5.81, p<.02. Furthermore, ofthe total number of written thoughts in theno distraction condition, 61% were coded ascounterarguments in the easy message condi-tion and 37% in the difficult message condi-tion.

An average of 2.21 favorable thoughts weregenerated to the difficult-to-counterargue mes-sage, and 2.06 favorable thoughts were gen-erated to the easy-to-counterargue message.There were no significant differences on thismeasure or on the total number of thoughtsgenerated. No interactions on the cognitiveresponse measures were significant. In sum,Experiment 1 conditions produced differentialcounterarguing, and in the easy message con-dition, the dominant cognitive response wascounterarguing.

The manipulation of distraction and effort.The analysis of variance on the distractionmanipulation check yielded a significant F(3,116) = 19.84, p < .001 for the distractionfactor. On a 7-point scale where 7 indicatedthat a subject was "completely distracted"from paying attention to the message, themean responses for the no-, low-, medium-,and highly distracted subjects were 2.21, 3.30,4.16, and 4.81, respectively.

Because of unequal cell frequencies, andunequal spacing of the levels of distractionmeasured in slides per minute, all linear trendtests in this report were conducted accordingto Gaito (1965). A significant linear trend wasfound for the ratings of distraction, F ( l , 116)= 52.80, p < .001.

Responses to the question, "To what extentdid you find the task effortful?" showed thatthe distraction manipulation was successful invarying perceived effort. On a 7-point scalewhere 1 indicated "completely effortful," themeans for the no-, low-, medium-, and highlydistracted subjects were 4.62, 4.80, 3.68, and2.64, respectively. Again, an analysis of vari-ance for the distraction factor yielded a sig-nificant F ratio, F(3, 116) = 16.28, p < .001,and a significant linear trend, 77(1,116) =42.67, p < .001. In addition, a Message XDistraction interaction, F(3, 116) = 2.85, p< .05, was obtained. Although the linear trendinteraction was not significant, the pattern ofmeans indicated that rated effort increasedmore sharply with distraction for the difficult-than for the easy-to-counterargue message. Inany case, these results clearly indicate thatthe faster the Xs flashed on the screen, themore distracted subjects reported feeling andthe more effortful they rated the task.

Dependent Variables

Communication acceptance. An unweighted-means ANOVA on the summed standardscores yielded two effects: a main effect formessage, F ( l , 119) = 5.23, p < .02, indicat-ing that the difficult-to-counterargue message(X = .33) produced more agreement than theeasy-to-counterargue message (X = —.31);and, the predicted Message X Distraction in-teraction, F(3,119) = 2.97, p < .03. The toppanel of Figure 1 indicates that the easy-to-counterargue message shows the much repli-cated effect of increasing communication ac-ceptance with increasing levels of distraction.A test of linear trend on the means ( — .84,— .46, —.17, and .23) proved significant, F ( l ,119) = 3.96, p < .05. The difficult-to-counter-argue message shows an opposite pattern—decreasing acceptance with distraction. A testof linear trend on the means (.66, .95, —.19,and —.11) indicated a significant negative

1.0 h

HI6HAGREEMENT

DISTRACTION AND PERSUASION

EXPERIMENT I

.difficult message

879

LOWAGREEMENT

20

NUMBER OF FLASHES PERMINUTE

HIGHAGREEMENT

1.2 -

.6 "

-.6

LOWAGREEMENT

-1.2

EXPERIMENT 2

difficult message

easy message

12NUMBER OF FLASHES PER

MINUTE

FIGURE 1. Top panel: mean attitude scores in relation to message and level of distraction forExperiment 1. Bottom panel: mean attitude scores in relation to message and level of distractionfor Experiment 2.

slope, F(l, 119) = 4.49, p < .05. In additionthe linear trend interaction was highly signifi-cant, F ( l , 119) = 8.39, p < .01.

Distraction and thought production. In theeasy-to-counterargue condition, the mean num-bers of counterarguments for the no-, low-,medium-, and highly distracted groups were3.75, 3.39, 2.65, and 1.72, respectively. Thelinear trend on these means was marginallysignificant F(l, 124) = 3.49, p < .10. In thedifficult-to-counterargue condition, the corre-sponding means were 2.43, 2.22, 1.68, and1.90. The linear trend on these means was notsignificant (F < 1). The main effect for dis-traction was reliable, F(3,124) = 2.95, p <.05. Using the Dunn multiple comparison

procedure (Kirk, 1968, p. 79), the amount ofcounterarguments generated at no and highdistraction was compared for both easy- anddifficult-to-counterargue messages. The testsindicated that for the easy-to-counterarguemessage, the difference between no and highdistraction was reliable (p < .05), but therewas no significant difference for the difficult-to-counterargue message in number of counter-arguments.

In the difficult-to-counterargue condition,the mean numbers of favorable thoughts forthe no-, low-, medium-, and highly distractedgroups were 2.56, 2.S3, 2.47, and 1.00, re-spectively. The linear trend for these meansonly approached reliability F(l, 124) = 2.46,

5 -

0 -

5 -


p < .15. In the easy-to-counterargue condi-tion, the corresponding means were 1.62, 2.22,2.21, and 2.18. The linear trend on thesemeans was not reliable (p > .25). Comparingthe number of favorable thoughts generatedat no and high distraction for both messagesusing the Dunn procedure indicated a nearsignificant drop in favorable thoughts for thedifficult message (p < .10), but not for theeasy message. In sum, the number of counter-arguments elicited by the easy-to-counterarguemessage and the number of favorable thoughtselicited by the difficult-to-counterargue mes-sage tended to be inhibited by the distractionmanipulation as expected by the thought dis-ruption hypothesis.

Recall of the persuasive message. The ma-nipulation of distraction produced a signifi-cant effect on the number of message argu-ments recalled, F(3,124) = 9.07, p < .001.The pattern of means for the no-, low-, me-dium-, and highly distracted groups was 3.8,4.0, 3.7, and 2.5, respectively (a perfect scorewould be 5 arguments recalled). Testing allpairwise comparisons on these four meansusing the Dunn multiple comparison pro-cedure indicated that the recall scores for theno, low, and medium groups did not differfrom each other, but the high-distractiongroup had significantly less recall than anyof the other groups (all ps < .05). There wereno interactions on this measure.

Ancillary measures. There were no maineffects nor interactions on the measures ofspeaker credibility, persuasive intent of thespeaker, or enjoyment of the experiment.

Discussion

Distraction and Recall

The highest level of distraction unexpect-edly led to a significant drop in the number ofarguments that subjects could recall. Yet, forthe easy-to-counterargue message, the amountof acceptance produced under the highestlevel of distraction was slightly greater thanthat produced under medium distraction, andclearly greater than that produced under nodistraction. (See Insko et al., 1975, for asimilar result.) Previous researchers (Fest-inger & Maccoby, 1964; Osterhouse & Brock,1970; Vohs & Garret, 1968) suggested that

"distraction serves to facilitate acceptance ofa counterattitudinal message only when thedistraction is not so severe as to inhibit recep-tion of the arguments contained in the mes-sage" (Osterhouse & Brock, 1970, p. 355).Certainly at some high level of distraction,reception of the message could be so severelyimpaired that the decrement in message learn-ing would outweigh all other factors. How-ever, Insko et al. (1975) suggest that there isa relatively flat slope relating recall and atti-tude change and, therefore, it takes a rela-tively large change in recall before there isany associated attitude change.

Distraction and Thought Production

The findings for the easy-to-counterarguemessage replicated the patterns found byOsterhouse and Brock (1970), Keating andBrock (1974), and Insko et al. (1975). Therewas more agreement with the message as thelevel of distraction increased, and addition-ally, there was a parallel drop in the numberof counterarguments generated when the no-and high-distraction cells were compared.

Given a message which elicits predomi-nantly counterarguments, the effect of distrac-tion is to interfere with counterargument pro-duction. The logical parallel, as we suggested,is as follows: given a message which elicitspredominantly favorable thoughts (here, thedifficult-to-counterargue message), the effectof distraction might be to interfere with fa-vorable thought production, and thus decreasemessage acceptance. In point of fact, underconditions of no distraction, the difficult mes-sage did elicit more favorable thoughts (38%of total thoughts) than counterarguments(36%). Of course, this difference was unreli-able, but having favorable thoughts disruptedby distraction could have accounted for theobserved decline in acceptance (see Figure 1).

EXPERIMENT 2

Since Experiment 2 was designed to morefully explore the possibility that for a difficult-to-counterargue message, distraction wouldinterfere with favorable thought production,the difficult-to-counterargue message in Ex-periment 2 advocated a proattitudinal posi-tion and used arguments that were logically


sound, and easily defendable. The easy-to-counterargue message also advocated a pro-attitudinal position, but as in Experiment 1,used arguments that were open to refutationand skepticism. This is in line with Robertsand Maccoby's (1973, p. 303) suggestion thatdistracting a receiver from generating agree-ing comments about a proattitudinal appealshould result in less acceptance of the appeal.

Although Experiment 2 was primarily de-signed to explore the more general distractionformulation, it might also be viewed as an-other test of "effort justification." Besides theneed for a proattitudinal message to maxi-mize favorable thought production, Baron etal. (1973) have argued that "Under certaincircumstances, the effort hypothesis predictsdistraction effects even on proattitudinal mes-sages. If the effort involved is particularlyhigh . . . and there are few other justifica-tions . . . enough dissonance could remain to. . . make one more extreme in one's favorableviews" (p. 317).

Briefly, two other changes were made inExperiment 2: (a) Only one order of thedependent variable booklet was used sinceorder had minimal effects in Experiment 1,and (b) only two levels of distraction wereused (low and medium) since these levels didnot affect recall in Experiment 1.

The following predictions were made. Morecounterarguments would be generated to theeasy-to-counterargue message than to the dif-ficult-to-counterargue message, but more fa-vorable thoughts would be generated to thedifficult-to-counterargue message. There wouldbe a significant Message X Distraction inter-action on the dependent variable measuringacceptance: Distraction was expected to in-crease acceptance for the easy-to-counterarguemessage, but reduce acceptance for the diffi-cult-to-counterargue message. Distraction wasexpected to inhibit counterargument produc-tion for the easy-to-counterargue message,and inhibit favorable thoughts for the diffi-cult-to-counterargue message.

MethodProcedure

Fifty-four introductory psychology students at TheOhio State University were run in a 2 X 2 factorialdesign with 2 levels of distraction (low and me-

dium) and 2 levels of message counterarguability(easy and difficult to counterargue). Thirteen sub-jects were run in each cell, and subjects were run ingroups of from 4 to 9 members. The attitude ques-tions preceded the thought listing section in thebooklets. After completing the booklets, subjectswere debriefed, thanked, and dismissed.

Message counterarguability. As in Experiment 1,two messages were constructed for this study. Bothmessages argued that the tuition at O.S.U. should bereduced by 50% from the then current $270 perquarter to $135 per quarter. The difficult-to-counter-argue message was designed to minimize counter-argumentation, and maximize favorable thoughts.The difficult message claimed that Ohio was 48th inper capita spending on higher education, and if Ohiocould move to 30th place, fees could be cut in half.Other revenue sources discussed in the message were:a $22 million surplus in the state budget, the statesales tax, and profits from the new state lottery.The message also argued that tuition needed to bereduced because the high price of an education inOhio had led the state to be 49th in the percentageof high school graduates going on to college.

The easy-to-counterargue message was designed tominimize favorable thoughts and to maximizecounterarguments, and thus took a different approachto the idea of reducing tuition. The message calledfor replacing high-prestige faculty with lower pres-tige, lower paid faculty; for cutting back on bookand magazine purchases in the library; for institutinga 5$ per ride charge on campus bus rides; for in-creasing class sizes; and for exerting more controlover courses taken by students to insure maximallyefficient use of instructors. Each message was ap-proximately 3 minutes in length.

Distraction and effort. The distraction task wasidentical to that employed in Experiment 1. Forthe low-distraction group the Xs flashed at the rateof 1 every 15 sec, and for the medium distractiongroup, the rate was 1 every 5 sec. Again, this manip-ulation was also designed to vary effort.

Dependent Variables

Attitude and thought measures. The specific tuitionattitude measure employed in Experiment 1 was usedagain in conjunction with a question which asked,"In general, to what extent did you agree with thespeaker's recommendations about lowering the tui-tion?" As in Experiment 1, these two attitude mea-sures were converted to standard scores and summedfor each subject. The thought collection and thoughtscoring procedure paralleled that employed in Ex-periment 1. Manipulation checks and ratings ofspeaker credibility, persuasive intent, and enjoymentof the experiment were also taken. All questionsexcept the specific tuition measure were respondedto on 12-point Likert-type scales.

Measure of recall. The last question in the book-lets asked subjects to try to recall as many argu-ments as they could from the message. The argu-ments recalled were scored in a manner identical tothat in Experiment 1.


TABLE 1MEAN NUMBER OF COUNTERARGUMENTS AND

FAVORABLE THOUGHTS IN RELATION TOMESSAGE COUNTERARGUABILITY AND

LEVEL OF DISTRACTION

Measure

Easy tocounterargue

Low Mediumdistrac- distrac-

tion tion

Difficult tocounterargue

Low Mediumdistrac- distrac-

tion tion

Counterarguments 3.92a 2.61b .38O .48O

Favorable thoughts 1.15a 1.77a 3.69b 2.31O

Note. Cells in any row having a common subscript are notsignificantly different at the .05 level by the Duncan multiplerange test.

In the low-distraction condition, subjects received 4 flashesper minute.

In the medium-distraction condition, subjects received 12flashes per minute.

Results


Message counterarguability. Significantlymore counterarguments (see Table 1) weregenerated to the easy message (X = 3.27)than to the difficult message (X = .42); F(l,43) = 52.49, p < .0001.5 Sixty percent of thethoughts generated to the easy message werecoded as counterarguments, while the corre-sponding figure for the difficult message was8%.

Significantly more favorable thoughts weregenerated to the difficult-to-counterargue mes-sage (X = 3.00) than to the easy-to-counter-argue message (X = 1.46); F ( l , 48) = 14.84,p< .001 (cell means in Table 1). In addi-tion, favorable thoughts predominated in re-sponse to the difficult-to-counterargue mes-sage: 62% of the total thoughts were codedas favorable. In the easy-to-counterargue mes-sage, which used weak arguments to supportan agreeable position, the percentage of fa-vorable thoughts was 27%. There were nosignificant differences between groups in totalthoughts generated.

The manipulation of distraction and effort.Subjects in the medium distraction conditionsreported feeling more distracted (X — 6.39)than subjects in the low-distraction conditions(X = 3.93); F( l ,48) = 9.07, p < .005. Also,subjects in the medium-distraction conditionsperceived the task as more effortful (X —

6.70) than subjects in the low-distraction con-ditions ( ^ = 5.00); F( l ,48) = 3.91, p < .05.

Dependent Variables

Communication acceptance. Subjects whoheard the difficult-to-counterargue messageexpressed more agreement with the message(X = .68) than subjects hearing the easy-to-counterargue message (X = —.72); 7^(1,48)= 14.10, p < .001. Also, the predicted Mes-sage X Distraction interaction was significant;7^(1,48) = 6.89, p < .02 (see bottom panelof Figure 1). Pairwise comparisons employingthe Duncan multiple range procedure indi-cated that for subjects hearing the easy-to-counterargue message, those under mediumdistraction ( Z = . 0 1 ) showed significantlymore agreement with the speaker than sub-jects under low distraction (X = —1.45), p< .05. Although not statistically significant,for subjects hearing the difficult-to-counter-argue message, the pattern was reversed (forlow-distraction subjects, X = .95, for medium-distraction subjects, X = .42).

Thought production. A marginally signifi-cant Message X Distraction interaction F(l,48) = 3.10, p < .08, was obtained on thenumber of counterarguments generated. Dun-can's multiple range procedure (Table 1) in-dicated that, as predicted, the number ofcounterarguments generated under low dis-traction for the easy-to-counterargue messagewas significantly greater than the numbergenerated under medium distraction. Thenumber of counterarguments generated to thedifficult message did not differ as a functionof distraction. As in Experiment 1, increaseddistraction reduced counterargumentation forthe easy message only.

A significant Message X Distraction inter-action was obtained on the number of favor-able thoughts generated F ( l , 48) = 6.27, p< .05. The Duncan multiple range procedure(Table 1) indicated that significantly morefavorable thoughts were generated to the dif-

5 The reader may wonder why a message advocatinga proattitudinal position would generate any counter-arguments at all. Recall that the easy-to-counter-argue message discussed specific consequences thatwould result if the tuition were reduced (e.g., initia-tion of bus fares, reduced books in the library, etc.)that most students viewed as undesirable.


ficult message under low distraction thanunder medium distraction. For the easy-to-counterargue message, there was no differenceas a function of distraction. The patterns ofdifferences on the thought measures are ex-actly congruent with the thought disruptionhypothesis.

Recall. The manipulations produced no sig-nificant differences on the number of messagearguments that could be recalled. The averagenumber of arguments recalled across all cellswas 3.77 out of 5 possible.

DISCUSSION

The pattern of means on attitude, counter-argument, and favorable thought measures inboth Experiments 1 and 2 supported the gen-eral thought disruption hypothesis over theeffort justification notion. As the level of dis-traction went up, the number of counterargu-ments went down for the easy-to-counterarguemessage, but remained unchanged for thedifficult-to-counterargue message. Likewise, asthe level of distraction went up, the numberof favorable thoughts generated went downfor the difficult-to-counterargue message, andremained unchanged for the easy-to-counter-argue message. The role of distraction appearsto be selective. The present evidence suggeststhat distraction inhibits the dominant cogni-tive response to a persuasive appeal. If thedominant cognitive response is counterargu-mentation, then distraction will tend to in-hibit this and result in increased persuasion;but if the dominant cognitive response isfavorable thoughts, distraction will tend toinhibit this and result in decreased persuasion.

Following Osterhouse and Brock (1970)and Insko et al. (1975), several analyses ofcovariance were conducted using the stand-ardized attitude index as the criterion in allcases. In Experiment 1 the crucial Message XDistraction F of 2.97 is reduced to a nonsig-nificant 2.49 when counterarguments are thecovariate, to a nonsignificant 2.51 when fa-vorable thoughts are the covariate, and to anonsignificant 2.28 when both are used as co-variates. This same result occurs in Experi-ment 2 in that a significant Message X Dis-traction F of 6.89 is reduced to a nonsignifi-cant 3.85 when counterarguments are co-

varied, to 2.03 when favorable thoughts arecovaried, and is reduced substantially to 1.48when both are used as covariates. These re-sults are consistent with the notion that thesubjects' cognitive responses mediate the in-teraction.

CONCLUSIONS

Why not conclude that the distraction ef-fect is mediated by both thought disruptionand effort justification? Why insist on a mo-nistic explanation? The answer lies in thepredictive power of the competing explana-tions. The effort justification prediction isthat the more effortful the distraction, themore successful it will be in producing ac-ceptance. Yet the studies reported here indi-cate that the distraction effect appears to bedependent not upon the amount of effort pro-duced, but on the message content and thenature of the thoughts elicited by the mes-sage. The thought disruption hypothesis al-lows the specific prediction that distraction ismost likely to lead to enhanced persuasionwhen a message presents poor arguments (i.e.,arguments that are open to refutation andcounterargumentation) and to reduced per-suasion when a message presents very goodarguments (i.e., arguments that are likely toelicit favorable thoughts). The effort justifica-tion hypothesis fails because it is unable tomake such differential predictions.

In sum, the results of the present experi-ments appear to have extended the role ofdistraction beyond the confines of counter-attitudinal persuasion. Not all of the perti-nent comparisons were statistically reliable,but the replicated patterns of differences makeit likely that subsequent investigators will beable to determine the dominant cognitive re-sponse to a communication, estimate thecumulative effect of these responses on ac-ceptance, and control much of the variance inacceptance by systematic manipulations ofdistraction. The current research suggests thatdistraction can enhance or reduce yielding topropaganda depending on the nature of themessage (whether easy or difficult to counter-argue) and the extent to which appropriateresponses (favorable thoughts or counterargu-ments) are evoked.


REFERENCE NOTE

1. Baron, R. S. Personal communication, October9, 1975.

REFERENCES

Baron, R. S., Baron, P. H., & Miller, N. The rela-tion between distraction and persuasion. Psycho-logical Bulletin, 1973, 80, 310-323.

Brock, T. C. Communication discrepancy and in-tent to persuade as determinants of counterargu-ment production. Journal of Experimental SocialPsychology, 1967, 3, 296-309.

Cialdini, R. B., Levy, A., Herman, C. P., Kozlow-ski, L. T., & Petty, R. E. Elastic shifts of opinion:Determinants of direction and durability. Journalof Personality and Social Psychology, in press.

Cohen, A. R. Communication discrepancy and atti-tude change: A dissonance theory approach. Jour-nal of Personality, 1959, 27, 386-396.

Cullen, D. M. Attitude measurement by cognitivesampling. Unpublished doctoral dissertation, OhioState University, 1968.

Festinger, L. A theory of cognitive dissonance.Evanston, 111.: Row, Peterson, 1957.

Festinger, L., & Maccoby, N. On resistance to per-suasive communications. Journal of Abnormal andSocial Psychology, 1964, 68, 359-366.

Freedman, J . L., & Sears, D. O. Warning, distrac-tion, and resistance to influence. Journal of Per-sonality and Social Psychology, 1965, 1, 262-266.

Gaito, J. Unequal intervals and unequal n in trendanalysis. Psychological Bulletin, 1965, 63, 125-127.

Greenwald, A. G. Cognitive learning, cognitive re-sponse to persuasion and attitude change. In A. G.Greenwald, T. C. Brock, & T. M. Ostrom (Eds.),Psychological foundations of attitudes. New York:Academic Press, 1968.

Insko, C. A., Turnbull, W., & Yandell, B. Facilitat-ing and inhibiting effects of distraction on atti-tude change. Sociometry, 1975, 4, 508-528.

Keating, J . P., & Brock, T. C. Acceptance of per-suasion and the inhibition of counterargumenta-tion under various distraction tasks. Journal ofExperimental Social Psychology, 1974, 10, 301-309.

Kiesler, S., & Mathog, R. The distraction hypothe-sis in attitude change: Effects of effectiveness.Psychological Reports, 1968, 23, 1123-1133.

Kirk, R. E. Experimental design: Procedures for thebehavioral sciences. Belmont, Calif.: Brooks/Cole,1968.

Kleinhesselink, R. R., & Edwards, R. E. Seeking andavoiding belief-discrepant information as a func-

tion of its perceived refutability. Journal of Per-sonality and Social Psychology, 1975, 31, 787-790.

Lowin, A. Approach and avoidance as alternatemodes of selective exposure to information. Jour-nal of Personality and Social Psychology, 1967, 6,1-9.

Miller, N., & Baron, R. S. On measuring counter-arguing. Journal for the Theory of Social Behav-iour, 1973, 3, 101-118.

Osterhouse, R. A., & Brock, T. C. Distraction in-creases yielding to propaganda by inhibitingcounterarguing. Journal of Personality and SocialPsychology, 1970, 15, 344-358.

Petty, R. E. Distraction can enhance or reduce yield-ing to propaganda by interfering with cognitiveresponses. Unpublished master's thesis, Ohio StateUniversity, 1975.

Regan, D. T., & Cheng, J . B. Distraction and atti-tude change: A resolution. Journal of ExperimentalSocial Psychology, 1973, 9, 138-147.

Roberts, D. F., & Maccoby, N. Information process-ing and persuasion: Counterarguing behavior. InP. Clark (Ed.), New models for mass communica-tion research (Vol. 2) . Beverly Hills, Calif.: Sage,1973.

Rosenblatt, P. C. Persuasion as a function of varyingamounts of distraction. Psychonomic Science, 1966,5, 85-86.

Rule, B. G., & Rehill, D. Distraction and self-esteemeffects on attitude change. Journal of Personalityand Social Psychology, 1970, 15, 359-365.

Silverman, I., & Regula, C. Evaluation apprehension,demand characteristics, and the effects of distrac-tion on persuasibility. Journal of Social Psychol-ogy, 1968, 75, 273-281.

Vohs, J. L., & Garrett, R. L. Resistance to per-suasion: An integrative framework. Public OpinionQuarterly, 1968, 32, 445-452.

Wicklund, R. A., Cooper, J., & Linder, D. E. Effectsof expected effort on attitude change prior toexposure. Journal of Experimental Social Psychol-ogy, 1967, 2, 416-428.

Zimbardo, P. G., & Ebbesen, E. B. The experimentalmodification of the relationship between effort,attitude and behavior. Journal of Personality andSocial Psychology, 1970, 16, 207-213.

Zimbardo, P. G., Snyder, M., Thomas, J., Gold, A.,& Gurwitz, S. Modifying the impact of persuasivecommunications with external distraction. Journalof Personality and Social Psychology, 1970, 16,669-680.

(Received February 12, 1976)

distraction can enhance or reduce yielding to propaganda ...€¦ · distraction can enhance or...

Documents