Journal o1972

nal of Personality and Social Psychology, Vol. 23, No. 1, 105-111

EMOTIONAL ROLE PLAYING, ATTITUDE CHANGE, ANDATTRACTION TOWARD A DISABLED PERSON1

GERALD L. CLORE 2 AND KATHARINE McMILLAN JEFFERY

University of Illinois

An experiment explored the effects of emotional role playing on interpersonalattitudes toward the disabled. The role playing consisted of traveling aboutthe campus in a wheelchair for an hour, and the design involved three treat-ment groups (role players, vicarious role players, and controls). Compared tothe control experience, both direct and vicarious emotional role playing led tomore positive responses (a) to a specific disabled person (the experimenter),(b) to a series of issues concerning disabled students in general, and (c) to adisguised attitudinal measure given by telephone 4 months later. The conceptof empathy appeared more adequate than dissonance for understanding theresults. The potential of role playing and vicarious experience for increasingtolerance and social maturity is considered.

According to Sarbin (1964) and other roletheorists, the technique of role playing is aneffective agent of attitude and behaviorchange. Although little systematic researchhas been conducted to assess the effects of roleplaying, several pieces of anecdotal evidenceare intriguing: (a) A Los Angeles area policedepartment instituted a role-playing programto give its recruits empathy for those theyarrest. The policemen learned what it is liketo spend the night in jail, played the role of abum on skid row, and enacted the hippie roleat a city park (Newsweek, December IS,1969). (b) Another example of role playingwas reported by John Howard Griffin, afree-lance writer who artificially pigmentedhis skin and toured the South by bus. In hisbook, Black Like Me, he records his experi-ences and the ways in which he began to con-form to the expectations that others had ofblacks (Griffin, 1960). (c) Some speech pa-thology departments have found it useful forspeech therapy students to play the role of a

!This study was supported by Grant MH 14510-03 from the National Institute of Mental Health.Portions of this study were presented at a symposiumon interpersonal attraction, Midwestern PsychologicalAssociation, Detroit, May 1971. Thanks are due toT. J. Nugent, Director, Division of Rehabilitation-Education Services, University of Illinois, for lendingwheelchairs for the experiment. The data were ana-lyzed by Howard McGuire and Ted Cottingham.

2 Requests for reprints may be sent to Gerald L.

Clore, Department of Psychology, University ofIllinois, Champaign, Illinois 61820.

8 Now at Stanford University.

stutterer (C. Van Riper, personal communi-cation, September 14, 1970). By stutteringwhen waited on in stores and other publicplaces, students get intense firsthand knowl-edge of the social problems encountered bytheir patients, (d) Recently, a grade schoolteacher in Iowa taught her third-grade stu-dents about prejudice by segregating them intobrown- and blue-eyed groups. By turns, eachgroup spent a whole day in the role of in-feriors. Reportedly, the role playing was anintense experience for the children (Peters,1971), but the nature of the results have notbeen objectively assessed.

In each of the examples cited, the objectof the role playing seems to have been thesameto create empathy for stigmatized per-sons. The experience appears to be capable ofproducing significant and lasting changes inone's perception of the stigmatized role, butuntil such consequences are studied, exactlywhat changes occur remains obscure.

Mann and Janis (1968) have previously re-ported research on emotional role playing asa technique of inducing fear. They had col-lege women who were heavy smokers play therole of cancer victims. For an hour, each sub-ject pretended to be a patient, while the ex-perimenter played the physician's role. Thephysician showed the patient a chest X rayand informed her that she had lung cancer.Together they acted out several scenes in thedoctor's office, focusing on the pain of the ill-ness, hospitalization, and early death. Later

105

106 GERALD L. CLORE AND KATHARINE MCMILLAN JEFFERY

these subjects were compared to a secondgroup that heard a tape recording of one ofthe role-playing sessions. In comparison withan untreated control group, the role-playingand passive control groups both reportedsmoking significantly fewer cigarettes per daywhen interviewed by telephone 18 monthslater. Mann and Janis thus demonstrated thatsubjects' feelings of invulnerability to cancercould be broken down through emotional roleplaying, and that the consequent fear couldmotivate changes in self-reported smokingbehavior.

The present research also investigated emo-tional role playing, but its focus was thecreation of empathy rather than fear. Thestudy grew out of repeated attempts to findobjective evidence for Goffman's (1963) ob-servation that people sometimes talk tooloudly to physically disabled and other stig-matized individuals.4 In three different studies,the voice loudness data taken from tapedconversations proved rather uninformative.However, the stooges made some provocativecomments after they feigned disability orblindness in the experiments. Some of theirremarks suggested that a convincing enact-ment of the disabled role might have greaterimpact on the role player himself than onthose he encounters.

In the experiment to be described here, thesubjects spent an hour on the campus in awheelchair, and we hypothesized that theexperience would have a positive effect ontheir attitudes (a) toward a specific disabledindividual as well as (b) on issues involvingdisabled people generally. In addition, thesubjects in a second group were asked to ob-serve the role player's experience. Mann andJanis had found that passive exposure to emo-tional role playing was effective in reducingsmoking, and on this basis we hypothesizedthat the interpersonal attitudes of observersas well as those of role players would beaffected. Some previous work has been re-ported on the effects of such vicarious experi-ences. Berger (1962) has shown that thegalvanic skin response can be elicited if asubject merely believes that someone else is

4 Unpublished studies conducted in collaboration

with Nancy Wiggins of the University of Illinois.

being shocked. Also, Stotland (1969) has in-vestigated palmar sweating and heart ratechanges as measures of empathy. His subjectswatched others receive either painful orpleasurable sensations, and they displayedmore physiological responsiveness when in-structed to empathize with the other personthan when simply watching him.

METHODSubjects

Seventy-six University of Illinois students (31 male,45 female) served as subjects. They were randomlyselected by computer within available hours froman introductory psychology pool. After reporting tothe experimental room in groups of 6, they wererandomly assigned to one of three conditionsroleplaying, vicarious role playing, or control. The ex-perimenter was a female graduate student who wasseated in a wheelchair. By appearing to be dis-abled, she could serve as a target person for laterattraction ratings.

Experimental ConditionsPrinted instructions asked each subject in the role-

playing group to play the role of a person confinedto a wheelchair. The subject was to imagine thathe had recently been in an automobile accidentwhich had severed the lower part of the spinal cordleaving his legs permanently paralyzed. He was askedto pretend that it was the first day back on campusafter the accident. A 25-minute wheelchair trip wasthen outlined that consisted of going from the fourthfloor of the psychology building to the basement ofthe student union. There he was to buy a cup ofcoffee or a coke, to drink it at one of the tables inthe snack bar, and finally to return to the laboratory.The experience involved a hundred yards of slightlyuphill sidewalk, four elevator rides, several rampsand doors, and a complicated procedure for gettingcoffee.

Vicarious role players were instructed to walkbehind the role players at a distance of 20 feet. Theywere informed about the role being played by thosein the wheelchair including the details of the autoaccident. The task of the vicarious role player wasto observe the role player's experiences, doing every-thing he did; however, he was not to interact withhim, ride in the same elevator, or help him inany way.

Finally, the control subjects were asked to spendan equivalent amount of time (25 minutes) walkingon the campus and having coffee at a differentspecified location. All of the subjects including thecontrols were asked to tell friends they happened toencounter that they were in an experiment and wouldexplain later. Six subjects who refused to take thewheelchair role were given the control instructionsinstead, but since they were self-selected members of

ATTRACTION TOWARD A DISABLED PERSON 107

the control group, their data were not included inthe analyses.

Dependent MeasuresAfter returning from their routes, the subjects were

given a blank sheet of paper and asked to describetheir experience in 100 words. The experimenter,still in her wheelchair, then gave them the threesets of rating scales described below, asking themto rate their feelings during the experiment, theirattitudes toward problems involving disabled stu-dents, and their evaluation of the experimental pro-cedures and the apparently disabled investigator.

Content analysis. The 100-word descriptions werecontent analyzed as one measure of the emotionalresponse of the subjects to their experiences. Tworaters,5 unaware of the hypothesis being tested,independently scored all of the descriptions by count-ing the number of clauses with emotional verbs,modifiers, descriptive statements, and nouns. Theoverall score was simply the sum of the scores ineach category. The results from the two scorers wereaveraged.

Affective rating scales. There were twenty 7-pointsemantic-differential-type items intended to tap theaffective character of their experience. At the topof the page was the phrase "During this experimentI felt," and beneath the heading were 20 items com-posed of $ potency and 5 evaluative items fromOsgood, Suci, and Tannenbaum (1957) along withS anxiety (e.g., tense-relaxed, anxious-calm, etc.),and S empathy items (e.g., apathetic-compassionate,concerned-indifferent, etc.) that were rationallyderived for this study.

Attitude scale. Next, the subjects filled out asurvey of attitudes composed of twelve 6-point itemsconstructed especially for this study. The questionsincluded such topics as the following: whether torequire new students to attend an orientation sessionabout the rehabilitation center, whether to includelectures on the disabled in introductory psychology,whether to favor foreign students or students con-fined to wheelchairs in giving new scholarships, andwhether the approval of the rehabilitation centershould be required on the plans for new campusbuildings. Responses to the items were summed toyield an attitude measure with a possible rangeof 12-60.

Demand and attraction measures. Finally, the sub-jects were told that the session was finished, butbefore leaving they were given an experimentalevaluation form which was to be filled in, sealed inan envelope, and deposited in the mail shute in thesupervising professor's office door. The form con-sisted of two parts. The portion assessing the demandcharacteristics of the experiment asked the subjectto state the hypothesis under study and also askedon a 7-point scale how strongly he felt he had beenexpected to indicate sympathetic or unsympathetic

5 The content analyses were done by Ted Cotting-

ham and Charla Denton.

attitudes toward handicapped students. The attrac-tion part of the form asked how much the subjectliked the experimenter (who appeared to be confinedto a wheelchair). The attraction scales were pre-sented as a confidential communication intendedonly for the supervising professor. They consisted ofthree 7-point scales asking how much the subjectwould like (or dislike) being in a different studywith the same experimenter, how much the experi-menter impressed him as a pleasant (or unpleasant)person, and finally what grade she should get forher performance as experimenter (from A to E). Asa check on the validity of these previously unusedattraction items, a subset of 54 subjects also answereda two-item measure of attraction taken from Byrne's(1961) Interpersonal Judgment Scale. These two7-point scales asked how much the subject liked (ordisliked) the target person and how much he wouldlike (or dislike) working with her in an experiment.

Volunteering. A month after the experiment, thesubjects were telephoned by a bogus representativeof a well-known campus service organization. Herequested volunteers to help show prospective stu-dents who were handicapped around the campus foran hour.

Follow-up attitude measure. Near the end of thesubsequent semester, 4 months after the experiment,the follow-up attitude assessment was also conductedby telephone. A caller identifying herself as someonefrom the Undergraduate Students Association statedthat she was sampling the student body to determinehow they would favor spending leftover funds in theUndergraduate Students Association treasury. Sub-jects were asked to rate how they felt about eachof four alternatives on a S-point scale. The alterna-tives were to spend the money on scholarships forblack students, increased parking space, increasedfacilities for disabled students, and increased intra-mural sports facilities. Since the initial attitude assess-ment might be affected by demand characteristics orby the subject's knowledge of the other treatmentconditions, this disguised 4-month follow-up wasconsidered the primary test of treatment effectiveness.

RESULTSThe major analyses were a series of

2 X 3 X 2 analyses of variance with sex,treatment condition, and volunteering as thefactors. Volunteering referred to whether ornot the subject volunteered to show a dis-abled student around the campus when asked1 month later.

AffectA content analysis of the subjects' 100-

word descriptions of their experience pro-vided one measure of affective response. Al-though the interrater reliability was not high( r= .55) , the analysis of variance yielded

108 GERALD L. CLOEK AND KATHARINE MCMILLAN JEFFERY

TABLE 1AFFECTIVE RATING SCALES : MEANS AND ANALYSES OF FIVE-ITEM SCALES DESCRIBING

How THE SUBJECT FELT DURING THE EXPERIMENT

Scale

PotencyEvaluationAnxietyEmpathy

Treatment condition

Role-playing

14.5521.2422.0725.62

VicariousX

19.7720.5421.3125.86

ControlX

21.5722.9316.9218.14

Analysis of variance

if

2/582/582/582/58

MS

274.8231.40

159.29397.28

F

10.95**2.554.39*

14.42**

% variance

254

1031

*t < .05.**p < .001.

a significant treatment effect (F = 4.88,p < .05), accounting for IS % of the variance.A Scheffe" test with significance set at the.OS level showed that the role players(^ = 2.87) were significant different fromthe controls (. = 1.16), but that the vicari-ous group (^ = 1.95), with an intermediatelevel of emotional response, was not signifi-cantly different from either of the othergroups.

The second measure of affect was the set ofaffective rating scales. Table 1 shows themeans and analyses of the responses to thescales. A significant treatment effect was foundfor the potency, anxiety, and empathy scalesbut not for the evaluative scale. Generally,the role players said they felt weak, bad,anxious, and empathic after their experience.The vicarious role players, who merely ob-served those in the wheelchair, describedtheir experience in similar terms. They feltslightly but not significantly less anxious,more negative, and more empathic than therole players. Only on the potency scale didthe vicarious group resemble the controlsrather than the role players.

There were also sex differences accountingfor 10% of the variance of the potency(F = 10.95, p < .01) and the anxiety scales(F = 8.43, p < .01). Females described them-selves as both weaker and more anxious thandid males.

AttitudeThe analysis of variance of scores on the

survey of attitudes concerning disabled stu-dents yielded a significant treatment effect(F= 13.34, p< .001), accounting for 25%

of the variance. A Scheffe test with signifi-cance set at the .05 level showed both therole-playing (. = 53.61) and the vicarious(^ = 50.29) groups to be significantly dif-ferent from the control group (% = 44.03)but not different from each other. The meansare diplayed graphically in Figure 1.

The only other significant effect was forsex (F = 9.06, p < .01), accounting for 9%of the varance. The females (^ = 51.63) weremore favorable than the males (X = 47.00)on issues concerning disabled students.

DemandThe attitude assessment made at the time

of the experiment was not considered a criticaltest of the treatment effects because of thepossible presence of demand characteristics.However, the analysis of variance of responsesto the 7-point scale asking how strongly thesubjects felt they had been expected to indi-cate sympathetic attitudes toward disabledstudents showed a significant effect only forthe volunteering variable (F = 6.56, p