Journal of Personality and Social Psychology1981, Vol. 40, No. 4, 786-796

Copyright 1981 by the American Psychological Association, Inc.0022-3514/81 / 4004-0786500.75

The Role of Dispositional Empathy and Social Evaluationin the Empathic Mediation of Helping

Richard L. Archer, Rolando Diaz-Loving, Peter M. Gollwitzer, Mark H. Davis,and H. Clayton FousheeUniversity of Texas at Austin

Coke, Batson, and McDavis have proposed a two-stage model of empathy-me-diated helping, based on emotional arousal and perspective taking. We hypoth-esized that in addition, a dispositional factor—individual differences in empa-thy—and a situational factor—potential evaluation from others (demand)—should be included in the process. A study was conducted in which female subjectsreceived false galvanic skin response feedback, indicating that they had eitherhigh or low arousal during a broadcast of a person's need for help, as in the Cokeet al. experiment. In addition, subjects were led to believe that the experimentereither did or did not know their level of arousal (demand vs. no demand). Subjects'premeasured dispositional empathy constituted the third (continuous) variablein the design. The effect of greater help following high- rather than low-arousalfeedback found by Coke et al. was replicated. However, as predicted, this wastrue only for subjects higher in dispositional empathy in the demand condition.The implications of these results for a revised model of empathy-mediated helpingare discussed.

Coke, Batson, and McDavis (1978) re-cently proposed a two-stage model of em-pathy-mediated helping. These authors re-ject the view that empathy is primarily acognitive process of taking another's per-spective (e.g., Regan & Totten, 1975). In-stead they favor the position that empathyis the result of an emotional response basedon physiological arousal (e.g., Krebs, 1975),although they are unwilling to dismiss cog-nition as unimportant to the process. Build-ing on the suggestions of Stotland (1969),Feshbach (1975), and Krebs (1975), theymaintain that "(a) taking the perspective ofa person in need tends to increase one's em-pathic emotional response," and "(b) em-

The authors wish to thank Joe Ramirez for servingas the experimenter and John Trimble for performingas his assistant in conducting this study. We are alsoindebted to Thomas Cloud and Joseph Burleson, whooffered technical assistance in setting up the false-arousal feedback, and to Jeffrey Bertnan, who offeredstatistical advice concerning the data analyses. Finally,our thanks to Daniel Batson, who provided us with copiesof the stimulus materials from the Coke et al. experi-ments.

Requests for reprints should be addressed to RichardL. Archer, Department of Psychology, 330 Mezes Hall,University of Texas at Austin, Austin, Texas 78712.

pathic emotion in turn increases motivationto see that person's need reduced" (Coke etal., 1978, p. 753). Although we support thecontention of Coke et al. that both emotionsand perspective taking are bound up in theempathic process, other factors appear to beinvolved that have not been embraced by thetwo-stage model.

The first of these additional factors is theindividual's characteristic tendency to em-pathize with others: a stable dispositionalfactor. Mehrabian and Epstein (1972) de-veloped a scale to examine individual dif-ferences in empathy as an emotional re-sponse. Most of the items of this scale areaimed at vicarious emotional responses tothe experiences of another (e.g., "Seeingpeople cry upsets me"), but a few involveperspective taking (e.g., "It is hard for meto see how some things upset people somuch"). Mehrabian and Epstein demon-strated the effects of individual differencesas operationalized by the scale in two studiesinvolving different empathy-related behav-iors. In the first study, persons scoring highor low on the scale served as "teachers" ina variation of the Buss (1961) "aggressionmachine" paradigm. Subjects ostensiblyshocked a "learner" (in reality a confeder-

786

EMPATHY, EVALUATION, AND HELPING 787

ate) for his mistakes as he was being taughta task. High-empathy subjects who were inclose proximity to the learner—about 8 feet(2.4 m) away—gave the lowest level ofshock. High-empathy subjects in the adjoin-ing room and low-empathy subjects in bothproximity conditions gave higher levels. Inthe second study, subjects who had com-pleted the empathy scale met with their(confederate) partner to exchange infor-mation before participating jointly in a task.During the exchange, the confederate re-vealed her need for volunteers to completea class project and asked the subject if shewould participate. Scores on the empathyscale were significantly related to the amountof time volunteered. People who character-istically empathize, then, should be expectedto harm others less and to help them more.

A situational factor not addressed in thetwo-stage model is the potential for receivingevaluation from others. In public situationswhen others are aware that a person is re-acting emotionally to someone's plight, manyof the customary excuses for not helping,such as failing to notice (see Latane & Dar-ley, 1970), cannot be offered. An emotionalreaction betrays the observer's interpretationof the situation as distressing and apparentlycalling for aid. Not to take action in themidst of clearly detectable emotion may ap-pear irrational as well as reprehensible.Schlenker (1975) has demonstrated that un-der public performance conditions, subjectspresent themselves in a manner consistentwith their reported beliefs. Similarly, thepotential for social evaluation in helping sit-uations may increase helping among em-pathizing individuals.

So far we have argued that empathy asa personal factor) and potential-for-social-evaluation as a situational factor should bothincrease helping. But beyond their indepen-dent effects, individual differences in em-pathy and social evaluation may interactwith one another. Foushee, Davis, and Archer(1979) found that empathy, as measured bythe Mehrabian-Epstein scale, was stronglyrelated to the Masculinity-Femininity (M-F) Scale of the Personal Attributes Ques-tionnaire (PAQ; Spence, Helmreich, &Stapp, 1974). The individual items of theM-F Scale, with an emphasis not only on

emotionality but on emotional vulnerability(e.g., "excitable in a major crisis," "feelingseasily hurt," and "cries very easily"), seemedto account for most of the relationship. Ifhighly empathic individuals are both moreemotional and more emotionally vulnerable,then their behavior may be especially influ-enced by helping situations involving poten-tial social evaluation. Not only will theirfeelings be stronger, but they will be moresubject to the judgment of others. They willalso (perhaps as a result) be more fearful ofthe judgment when it comes.

Davis (1979) found a significant relation-ship between total scores on the Mehrabian-Epstein empathy scales and both the private(r = .61, /x.OOl) and public (r = .46,p < .001) subscales of the Self-Conscious-ness Inventory (Buss, 1980; Fenigstein,Scheier, & Buss, 1975), which supports thisline of reasoning. According to Fenigsteinet al., private self-consciousness refers to"an awareness of inner thoughts, feelings,and motives" (e.g., "I reflect about myselfa lot"). Persons high in private self-con-sciousness have been shown to provide moreaccurate self-reports (Scheier, Buss, & Buss,1978) and to react more intensely to positiveand negative affective stimuli (Scheier &Carver, 1977). Public self-consciousness, onthe other hand, reflects "an awareness of theself as a social object" (e.g., "I'm concernedabout what other people think of me"). Inresearch settings, persons high in public self-consciousness have demonstrated greaterconformity to group pressure (Froming &Carver, Note 1) and a more negative reac-tion to rejection by the group (Fenigstein,1979). On the basis of these relationshipswith self-consciousness, it would appear thatempathic persons are quite concerned withtheir own feelings and outcomes as well aswith the feelings and outcomes of others.

There is reason to believe that disposi-tional empathy and social evaluation mayhave contributed jointly and directly to thehelping effects obtained by Coke et al.(1978, Experiment 2) in their test of the two-stage model. In their experiment, subjectswere instructed to empathize with a grad-uate student who desperately needed vol-unteers to complete her master's thesis re-search. While subjects listened to the tape

788 ARCHER, DIAZ-LOVING, GOLLWITZER, DAVIS, FOUSHEE

recording of the graduate student's request,they received false feedback from a galva-nized skin response (GSR) machine. As theywatched, the machine registered either highor low arousal. The experimenter who hadbeen monitoring the subjects' GSR from thenext room then entered and obtained a self-report measure of their emotional state dur-ing the broadcast. Finally, he asked subjectsto read and respond to a letter from theneedy graduate student soliciting volunteersfor up to 4 hours of time. As Coke et al.predicted, the results showed that subjectsin the high-arousal feedback condition vol-unteered more time than subjects in the low-arousal feedback condition.

The experimenter appeared reluctant topresent the helping request and departedduring the period in which subjects re-sponded to it. But subjects may still have felttheir behavior was under scrutiny. The ex-perimenter told them that the professor su-pervising the project "thought it would benice to do something for (the needy graduatestudent)." Furthermore, he gave no assur-ance that responses to the request would re-main a secret. High-arousal subjects, in par-ticular, may have felt pressured to complywith the request. Not to do so when the ex-perimenter had witnessed their arousal mightmake them appear inconsistent or cast doubton the elaborate physiological measure. Con-versely, a desire to avoid inconsistency andto uphold the validity of the experimenter'smeasurement may even have pressured low-arousal subjects not to comply.

The account of the potential interactionbetween dispositional empathy and the eval-uative aspects of the situation presented ear-lier suggests further that high empathizersshould have been especially sensitive to theexperimental demand for help. Persons whousually empathize would have been not onlymore likely to experience empathy and thusperceive the feedback in the high-arousalcondition as veridical but also would havebeen more concerned about the experi-menter's view of their behavior in botharousal conditions. This analysis of the Cokeet al. (1978, Experiment 2) results wastested in the following study.

As in the original experiment by Coke etal., subjects received false GSR feedback

that they had high or low arousal during thebroadcast of a person's need for help. In ourstudy, however, half the subjects were ledto believe that the experimenter had knowl-edge of their arousal (demand). The otherhalf were led to believe that the experi-menter was blind to their arousal (no de-mand). In addition, all subjects were pre-measured for dispositional empathy on theMehrabian-Epstein scale, and their scoreswere included as a third (continuous) pre-dictor variable in the design. An interactionamong arousal, demand, and dispositionalempathy was predicted: Within the demandcondition for subjects with higher levels ofempathy, high arousal was expected to resultin a greater amount of helping than lowarousal.

A further issue that is raised by the resultsof the Coke et al. studies is the nature of theempathy-induced motivation to help. Thesource of the emotion may be a feeling ofempathic concern, an other-centered, al-truistic desire to reduce another's distress(Batson, Darley, & Coke, in press; Krebs,1975), and/or it may be a feeling of personaldistress, a self-centered, egotistical desire toreduce one's own arousal (Piliavin & Pilia-vin, Note 2). Coke et al. (1978) have arguedthat "empathic concern, and not personaldistress or some other emotion mediatedhelping" in their experiment (Experiment 2,p. 762). Our analysis of the helping situationsuggests otherwise. The effect of social eval-uation may be to increase feelings of self-consciousness for those who are high in dis-positional empathy (Davis, 1979). More-over, high empathy appears to be associatedwith emotional vulnerability (Foushee et al.,1979). Therefore, it was predicted thatgreater feelings of personal distress as wellas greater feelings of empathic concernwould be reported under high arousal ratherthan under low arousal among subjects highin dispositional empathy in the demand con-dition. In addition, both empathic concernand personal distress were expected to showa relationship with helping.

MethodSubjects

One hundred and twenty-three female undergradu-ates from the University of Texas at Austin participated

EMPATHY, EVALUATION AND HELPING 789

as subjects in partial fulfillment of a course requirementfor introductory psychology. They were part of a largerpool of 610 women who had been premeasured for dis-positional empathy on the Mehrabian-Epstein (1972)scale. To ensure a representative distribution of scores,equal numbers of subjects from above and below thescale median (Mdn = 47.51, SD = 27.00) were assignedby a randomized blocks procedure to each of the ma-nipulated conditions of the 2 (arousal) X 2 (demand)design. During the course of the study, three subjectswere eliminated from the design. Two subjects had beenin previous studies that used the same false feedbackmanipulation of arousal, and an equipment failure oc-curred during the testing of a third subject. All threewere replaced in the design, so that there were 30 sub-jects in each manipulated condition.

Apparatus

A display meter (a Micronta DC meter in a slope-front metal cabinet), seemingly connected to a galvanicskin response monitor (a C. H. Stoetting Psychogalvan-oscope) in the next room, was modified to deliver false-arousal feedback. The needle on the display meter wasin reality driven by electrical current from the speakeroutlet of one channel of a two-channel (Sony TC 252)tape recorder, as in the Coke et ai. (1978, Experiment2) study. While the subject listened to the voice re-cording on the second channel, variations in the volumeof a tone recorded on the first channel altered the elec-trical output signal. The signal was then registered bythe display meter as arousal. The display meter wasprovided with a 30-point scale (-15 to +15). The lowend of the scale was labeled "low arousal" and the highend, "high arousal."

Procedure

Subjects for whom dispositional empathy scores hadbeen obtained were recruited by means of eligibility lists:one composed entirely of persons with scores above themedian, the other entirely of persons below the median.The experimenter was kept blind to the existence of thetwo lists as well as to the hypothesis for the study.

Most of the procedure, excepting the demand manip-ulation, was taken directly or adapted from Coke et al.(1978, Experiment 2). When each subject arrived at thedesignated waiting place, she was met by a male ex-perimenter and escorted to the suite of experimentalcubicles. The subject was first ushered into the "controlroom," which contained a GSR monitor and tape re-corder. The experimenter explained that during thestudy the subject would listen to and evaluate a pair ofbroadcasts. He showed her the GSR monitor and de-scribed its operation and role in the study in one of twoways.

The demand manipulation. For half the subjects,the description of the role of the GSR monitor in thesession was essentially that provided by Coke et al. Inthis, the demand condition, the experimenter told sub-jects that he would monitor their responses to the broad-cast and showed them the built-in arousal display meteron the GSR machine.

For the other half of the subjects, the operation of the

GSR monitor was described differently. In this, the no-demand condition, the experimenter told subjects thatthe GSR machine would internally record their re-sponses to the broadcasts so that he could remain blindto their reaction. He showed them how the display meteron the machine had been taped over to prevent the read-ing from influencing his behavior toward subjects.

Following the demand manipulation, the subject wasconducted to the next cubicle and seated before a deskon which was displayed the false-feedback GSR mon-itor. She was asked to read the following written expla-nation of the study:

In this experiment your job will bt to listen to andevaluate two radio broadcasts. These broadcasts wererecorded as pilots for two new programs planned byKUT, the University radio station. The first programinvolves an announcement about upcoming campusevents; the second involves appeals for student help.Although authentic, neither pilot has been or will beaired.

When the subject had finished reading, the experi-menter further explained what would take place duringthe session. Each broadcast, he said, would be playedon a tape deck in the control room connected to a speakerin the subject's cubicle. In addition to the GSR machinein the control room, the monitor on her desk would alsoregister her arousal during the broadcast. This secondmonitor was supposedly set up because subjects mightfind it interesting to see how they were reacting. Theexperimenter then described the working of the monitorand how to read it. He also noted that research hadshown that GSR was impossible to consciously control.In the no-demand condition, the experimenter also em-phasized that the subject was not to tell him what themonitor registered.

After the subject agreed to participate in the study,the experimenter attached electrodes to the first andthird fingers of her nondominant hand. In keeping withthe procedure of Coke et al., she was then given writteninstructions describing her evaluation task during thefirst broadcast:

In order to provide a more thorough evaluation of thetapes, different subjects are asked to listen to eachbroadcast from a different point of view or perspec-tive. By random assignment, you were selected to re-ceive the following instructions. Please attend to thebroadcast techniques while listening to the first broad-cast. Give them full attention and consider them care-fully.

Having handed these instructions to the subject, theexperimenter left the room, allegedly to turn on the GSRmachine and start the broadcast. In fact, an assistantto the experimenter entered the control cubicle andplayed the broadcast recordings so the experimentercould remain blind to the arousal manipulation later.

The first broadcast was designed to serve as a pointof reference for interpreting the false-arousal feedbackto come later. It took the form of a bland announcementof an upcoming anthropology lecture. During the an-nouncement, the subject observed the needle on herarousal display meter fluctuate only slightly between-3 and +3 on the 30-point scale. When the recording

790 ARCHER, DIAZ-LOVING, GOLLWITZER, DAVIS, FOUSHEE

ended, the experimenter returned and administeredtwo bogus questionnaires purportedly assessing percep-tions of the "theme of the broadcast" and its "technicalquality."

When the subject had finished with the questionnairesconcerning the first broadcast, the experimenter handedher written empathy instructions for her evaluation ofthe second recording:

While listening to the second tape you should try toimagine how the person in the broadcast feels. Imag-ine how you yourself would feel if you were subjectedto the same experience. In your mind's eye, perhapsyou can visualize how it would feel for you to be inthis situation.

The experimenter again left the room during the secondbroadcast while the unseen assistant played the record-ing manipulating the false-arousal feedback.

Arousal manipulation. The second broadcast wasan appeal for help from a graduate student in the Col-lege of Education. Her need was for students to partic-ipate in her master's thesis research. She could not af-ford to pay for participation, so she was asking forvolunteers. During the playing of the appeal, subjectsin the low-arousal condition saw the needle of the GSRmonitor remain within the same 6-point range (-3 to+3) as it had during the first recording. But for subjectsin the high-arousal condition, the needle climbed slowlyuntil it reached a point between +12 and +15, whereit remained until the end of the broadcast.

Measurement of emotional state. When the secondbroadcast ended, the experimenter returned with morequestionnaires. Two of the questionnaires were the bo-gus "theme of the broadcast" and "technical quality"instruments. The experimenter placed the bogus ques-tionnaires on the table and asked the subject to completethem first. In the demand condition, subjects next re-ceived an "emotional reactions" questionnaire. Subjectsin the no-demand condition, however, did not completethe emotional reactions questionnaire until after themeasure of helping had been taken. This switch in theorder of the measures was necessary to prevent subjectsin the no-demand condition from inferring that the ex-perimenter had learned of their arousal by examiningtheir self-ratings.

The emotional reactions questionnaire was composedof 22 of the 23 adjectives used by Coke et al. (1978,Experiment 2) to tap emotional state. For each adjec-tive, the subjects were to indicate the degree to whichthey had experienced the emotion on a scale from 1 (notat all) to 7 (very). All 13 of the original personal distressadjectives (alarmed, perturbed, disconcerted, bothered,irritated, disturbed, worried, uneasy, distressed, trou-bled, upset, anxious, and grieved) were included. How-ever, the adjective empathic was deleted from the orig-inal empathic-concern adjectives, leaving seven (moved,softhearted, sorrowed, touched, warm, concerned, andcompassionate).1 The two filler adjectives used by Cokeet al., intent and intrigued, that were not part of eitherset were also included.

Measurement of helping. After the subject had com-pleted the first of the bogus questionnaires, the exper-imenter placed the other before her. Lifting the second

questionnaire from the table, he revealed a letter. Ap-pearing surprised and a little dismayed (cf. Coke et al.,1978, Experiment 2), as though he had picked up theletter by mistake, he explained to the subject:

Since I was going to give this to you at some pointanyway, I guess I should go ahead and explain it now.This is a letter from the woman you just heard. Thereason I am giving it to you is that the professor incharge of this study, Dr. , had to ask her forpermission to use her broadcast in the study. Dr.

thought it would be nice to do something forher. For that reason he asked me to give this letterto all the people that participated in our study.(Pause) Let me ask you to go ahead and read theletter and decide what you want to do about it whileI get the [last/third] questionnaire.

The letter restated the graduate student's need forvolunteers. A response scale with 30-min. incrementswas provided on which subjects were to indicate howmuch time they were willing to give.

In the demand condition, the experimenter also gavethe subject the emotional reactions questionnaire whilethe helping request was before her. Then he left thecubicle long enough for her to complete the question-naires and the helping measure. When he returned hepersonally collected the request for help along with thequestionnaires.

In the no-demand condition, before departing fromthe room, the experimenter added:

No matter what you decide to do about the request,that is if you fill in any information or leave it blank,please enclose and seal it in one of those envelopesaddressed to [the graduate student]. Then put it inthat box where those other envelopes are. We willsend them over to the College of Education.

Upon his return, the sealed request was lying in the box(along with two other envelopes). Not until this pointdid he give the subject the emotional reactions' ques-tionnaire.

Manipulation checks. Before debriefing the subject,the experimenter gave her a sheet with three items em-ployed by Coke et al., as checks on the arousal manip-ulation. Subjects were asked to indicate how arousedthey had felt while listening to the broadcast (from notat all to extremely), how accurately the GSR monitorhad measured their level of arousal (from not at allaccurate to very accurate), and the magnitude of thegraduate student's need (from very little to very great).All three items were in the form of 8-point scales.

Results

Analyses

The major analyses performed used thesimultaneous multiple regression program

1 The adjective empathic was deleted because its closerelationship to the wording of the empathy-listening in-structions might have led to demand effects.

EMPATHY, EVALUATION, AND HELPING 791

from the SPSS (Statistical Package for theSocial Sciences; Nie, Hull, Jenkins, Stein-brenner, & Bent, 1975). The continuousvariable—dispositional empathy score—andthe two dichotomous variables—arousal anddemand condition—were the predictors. Si-multaneous multiple regression and effectcoding of the dichotomous variables wereused as the best approximation of analysisof variance when cell ns are unequal (Carl-son & Timm, 1974; Overall, Spiegel, &Cohen, 1975). Values of the continuous vari-able were coded in terms of their deviationfrom the mean to reduce the correlation be-tween main effect and interaction terms(Althauser, 1971; Kenny, 1979).

Separate simple effects tests were per-formed when interactions were obtained.These simple effects tests were also run assimultaneous multiple regressions (i.e., one-degree-of-freedom contrasts), using theoverall residual from the main analysis asthe error term.

Effectiveness of the Manipulations

The multiple regression analysis per-formed on subjects' self-reported arousalshowed only one significant effect, a maineffect for the arousal variable, F(l, 112) =15. 79, jtK.OOl. The false feedback waseffective in eliciting higher reports of arousalfrom subjects in the high-arousal conditionthan from subjects in the low-arousal con-dition (5.09 vs. 3.95). The analysis of theitem assessing subjects' perceptions of themagnitude of the graduate student's needalso revealed an arousal main effect, F(\,112) = 4.73, p < .04, but it was stronglyqualified by an interaction between all threeof the predictors, F(l, 112) = 6.79, p< .01.High-arousal subjects tended to perceivemore need on the part of the graduate stu-dent (6.52 vs. 5.97 in the low-arousal con-dition). Consistent with our predictions,however, arousal interacted with demand forsubjects high in dispositional empathy, F( 1,112) = 5.45, p<.03, but not for subjectslow in dispositional empathy, F(l, 112) =2.51, p> . l l . Among high-empathy sub-jects, the difference between high and lowarousal was significant in the high-demandcondition (7.00 vs. 5.73), F(l, 112) = 5.01,

p < .03, but was (nonsignificantly) reversedin the low-demand condition (6.00 vs. 6.6),F(l, 112)= 1.13, p> .29.

The results of the regression analysis per-formed on subjects' perceptions of the ac-curacy of the GSR feedback also reflect theinfluence of the demand manipulation. Amarginal interaction between arousal anddemand was found, F(l, 112) = 3.32, p <.08, because high-arousal feedback led toslightly (and nonsignificantly) higher per-ceptions of its accuracy in the high-demandcondition than low-arousal feedback (5.53vs. 5.17), F< 1, but somewhat lower per-ceptions of accuracy in the no-demand con-dition (5.04 vs. 5.97), F(l, 112) = 3.98,p < .05.

Overall, these checks suggest that both thearousal and the demand manipulations hadtheir intended effects. Furthermore, therewas an indication that dispositional empa-thy, the individual difference variable, mag-nified the effects of the manipulations asexpected.

Effects of the Predictor Variables onEmotions

The major analyses to examine subjects'emotional reactions were conducted on anempathic concern index and a personal dis-tress index. The empathic concern index wascomposed of responses to the adjectives soft-hearted, warm, concerned, and compassion-ate, summed and averaged for each subject(a = .82). Except for the absence of the ad-jective empathic, this index was identical tothe empathic concern index constructed byCoke et al. The personal distress index wascomposed of the adjectives upset, alarmed,and troubled, again summed and averagedfor each subject (a = .80). The correspond-ing personal distress index of Coke et al. wasformed from these same adjectives.2

2 The adjectives for the empathic concern and per-sonal distress indices were, except for the missing em-pathic, the same ones used by Coke et al., in order tofacilitate comparison between the studies. A factor anal-ysis was also performed on subjects' responses to theentire set of emotions adjectives. Since the factors werenot expected to be orthogonal, an oblique rotation wasused. Sizeable intercorrelations between the items (.35-.77) led to the employment of the PA1 option from theSPSS program package (Nie et al., 1975). A three-fac-

792 ARCHER, DIAZ-LOVING, GOLLWITZER, DAVIS, FOUSHEE

Table 1Mean Self-Reported Emotion for SubjectsAbove and Below the Median in DispositionalEmpathy in Each Condition

Low empathy High empathy

Condition

DemandEmpathic

concernPersonal

distress

No demandEmpathic

concernPersonal

distress

Lowarousal

4.73

2.62

4.50

2.62

Higharousal

4.87

2.60

4.85

2.15

Lowarousal

4.47

2.61

4.89

2.42

Higharousal

5.35

3.89

4.72

2.76

Note.N= 15.

The multiple regression performed on theempathic concern index yielded an interac-tion among dispositional empathy, arousal,and demand, F(l, 112) = 4.20, p < .05. Asmay be seen from the Table 1 means, a(borderline) interaction between arousal and

tor solution emerged. Within this solution troubled, up-set, anxious, and grieved all loaded highly (load-ings > .60) on the first factor (exclusively), a sadpersonal distress factor. Moved, softhearted, touched,warm, concerned, and compassionate all loaded highly(loadings > .60) on the second factor (exclusively), em-pathic concern. On the third factor, which accountedfor less than 6% of the variance, perturbed, bothered,irritated, and disturbed loaded highly (loadings > .60)and exclusively. It seemed to represent an angry formof personal distress.

Subjects' responses to these items with loadings >.60 were summed and averaged for each factor to forman empathic concern index, a personal distress-sad in-dex, and a personal distress-angry index. When thisexpanded version of the empathic concern index and thepersonal distress-sad index were substituted in the mul-tiple regression analyses for the indices based on Cokeet al.'s selection, the effects were weaker but similar.For the expanded empathic concern index, the maineffect for arousal, F(\, 112) = 3.72, p < .06, and thehint of the interaction among dispositional empathy,arousal, and demand, F(\, 112) = 3.09, p < .09, werefound. For the personal distress-sad index, significantdispositional empathy, F(\, 112) = 4.81, p < .03, anddemand, F(\, 112) = 3.76,p < .06, main effects emergedalong with an Empathy X Arousal interaction, F(\,112) = 5.30, p < .03. The analysis for the personal dis-tress-angry index yielded only a dispositional empathymain effect, F(\, 112) = 9.48, p < .003.

demand was found among subjects high indispositional empathy, F(l, 112) = 3.73,p = .056, but not among subjects low in dis-positional empathy (F< 1). For subjectswith higher empathy scores, high arousal ledto significantly greater self-reports of em-pathic concern under high demand, F(\,112) = 5.90, /><.02, but the effect wasslightly (and nonsignificantly) reversed un-der low demand (F < 1).

Table 1 also indicates that the scores onthe personal distress index were arrayed ina similar pattern to the empathic concernscores. The multiple regression analysis ofthese scores, however, revealed a main effectfor dispositional empathy, F(l, 112) = 5.21,p < .03, and a main effect for demand, F(l,112) = 4.17, p < .05. Both higher empathyscores and the demand manipulation addi-tively increased subjects' reports of self-ori-ented personal distress.

The close parallel between the results forthe analysis of the empathic concern indexand those for the personal distress index sug-gest that the two emotions are interrelated.3

Subjects with higher dispositional empathyscores exposed to the demand manipulationfelt not only more concern for the person inneed but also more concern for themselves.

Effects of the Predictor Variables onHelping

Responses to the request for help werecoded for analysis using the followingscheme: no help = 0; 30 min. = 1; 60min. = 2; 90 min. = 3; 120 min. = 4; andmore than 120 min. = 5. When these codedresponses were subjected to a multipleregression analysis, a main effect fordispositional empathy emerged, F(l,112) = 5.01, p < .03, strongly qualified bythe predicted interaction among all threevariables (empathy, arousal, and demand),F(l, 112) = 6.25, p <. 02. As may be seenin Figure 1, subjects with higher disposi-tional scores generally offered more help.But the pattern of volunteering also clearlyshows the interactive effects of arousal and

3 The correlation between the empathic concern indexand the personal distress concern index was .42, p <.001

EMPATHY, EVALUATION, AND HELPING 793

demand on high-empathy subjects, F(l,112) = 4.13, p < .04, but not on low-empa-thy subjects (F< 1). High arousal led tomore volunteering than low arousal whensubjects had higher dispositional empathyscores and received the demand manipula-tion, F(l, 112)= 10.26, p<.001. Interest-ingly enough, this difference was producedprimarily by an extremely low level of help-ing in the low-arousal condition rather thana particularly high incidence of helping un-der high arousal. There was a slight reversalof the pattern in the no-demand conditionin which subjects higher in empathy volun-teered (nonsignificantly) more help underlow arousal than high (F < 1). However, forsubjects with lower empathy scores, the ma-nipulations had little effect on their willing-ness to help. Thus, as predicted, the originalfinding by Coke et al. of greater help afterarousal feedback was replicated, but onlywhen a particular combination of disposi-tional empathy and situational demand waspresent.

Empathic Concern and Personal Distressas Mediators of Helping

To investigate the relationship among em-pathic concern, personal distress, and help-ing, a multiple regression on helping wasperformed in which subjects' scores on theempathic concern and personal distress in-dices were entered as the predictors. In thisanalysis, subjects' dispositional empathy,arousal condition, and demand conditionwere not entered as predictors of helping.

The main effects were generally consistentwith the findings of Coke et al. Empathicconcern significantly affected helping, F(l,116) = 23.28, p<.001, whereas personaldistress did not (F < 1). But the interactionbetween empathic concern and personal dis-tress was also significant, F(l, 116) = 5.61,p < .02. As Figure 2 shows, for subjects highin empathic concern, personal distress didaffect helping, F(l, 116) = 4.31, p< .04,whereas for subjects low in empathic con-cern it did not (F < 1).

The results of this analysis undercut theargument of Coke et al. that personal dis-tress plays no mediating role in helping inthis situation. Moreover, they provide sup-

zIon

a. u.-i o£ P

1-

LEGENDA HIGH EMPATHY-DEMAND

• HIGH EMPATHY-NO DEMAND

A LOW EMPATHY-DEMANDO LOW EMPATHY-NO DEMAND

LOW HIGHAROUSAL

Figure 1. Mean help volunteered for subjects above andbelow the median in dispositional empathy in each con-dition.

port for our contention that helping is influ-enced by both empathic concern and per-sonal distress—emotions that were strongestfor those high in dispositional empathy whowere subjected to demand characteristics.

Discussion

Our hypothesis stated that among subjectshigh in dispositional empathy in the high-demand condition, high arousal should resultin greater helping than low arousal. Clearly,the results from the helping measure confirmit. Empathy as a dispositional factor andsocial evaluation as a situational factor in-teracted with arousal feedback to predicthelping.

The two-stage model of empathy-me-diated helping proposed by Coke et al.(1978) treats only the situational factors ofperspective taking and emotional response.However, they recognized the possibilitythat demand characteristics (Orne, 1962)might have influenced their results (Coke etal., 1978, p. 763). Furthermore, the likeli-hood that the process would be influencedby individual differences seems clear fromprevious dispositional empathy research (e.g.,Mehrabian & Epstein, 1972). What mightnot have been anticipated was the intriguingmanner in which these factors interact witharousal in the helping situation. The originalCoke et al. finding of more help volunteeredafter high-arousal than low-arousal feed-back, was replicated only for high-empathy

794 ARCHER, DIAZ-LOVING, GOLLWITZER, DAVIS, FOUSHEE

z5omLi.

Ov>Hi

z

2 —

1 —

LEGENDA. HIGH EMPATHIC CONCERN• LOW EMPATHIC CONCERN

I I

LOW HIGH

PERSONAL DISTRESS

Figure 2. Mean help volunteered by subjects' self-re-ported empathic concern and personal distress.

persons subjected to the demand manipu-lation.

Coke et al. interpreted their helping re-sults as purely the product of empathic con-cern for another. We hypothesized andfound that personal distress would also beaffected by the unique combination of dis-positional empathy, arousal, and demand.Moreover, the analysis employing empathicconcern and personal distress as predictorsof helping found that these emotions bearan interactive relationship to volunteering.This interaction between empathic concernand personal distress corresponds well withthe correlations found by Foushee et al.(1979) between empathy and emotional vul-nerability. Persons who become the mostconcerned over the welfare of others alsoseem to experience the most personal upset.Thus, it seems premature to dismiss egoisticmodels of helping motivation (e.g., Piliavin& Piliavin, Note 2) where empathy is in-volved. Interestingly enough, among personswho experience stronger altruistic feelings,part of the motivation to aid another maystem in part from the egoistical desire to endone's own aversive arousal. To fully disen-tangle the respective roles of personal dis-tress and empathic concern in helping situ-ations, it seems likely that it will be necessaryto develop a multidimensional measure ofempathy along the lines undertaken by Davis(1979).

Although our prediction was simply statedin terms of arousal differences among high-

empathy subjects in the demand condition,we were surprised that most of this differ-ence was due to low arousal. It was primarilythe extreme decrease in helping for high-em-pathy subjects in the low-arousal demandcondition rather than the slight increase inhelping in high-arousal demand that pro-duced the interaction. Since the originalCoke et al. (1978, Experiment 2) study in-cluded only a low- and a high-arousal con-dition and no other control groups, theirarousal differences may have resulted froma similar decrease. Our account of the forcesbearing on highly empathic persons wouldsuggest that in the high-arousal condition,both their personalities and the meter feed-back would coincide, whereas in the low-arousal condition, the feedback would bediscrepant with their disposition. Social eval-uation might be expected, then, to elevatehelping in the high-arousal condition muchmore than it would depress it in the high-arousal condition. But the form-of-actual-helping results argue that the impact of so-cial evaluation was not to magnify the in-fluence of dispositional empathy; rather, itwas to magnify the influence of situationalarousal feedback. To put it another way, itseems as though the tendency toward publicself-consciousness of our high-empathy sub-jects, rather than their tendency toward pri-vate self-consciousness, produced this reac-tion to the demand manipulation (see Buss,1980).

Dispositional empathy is, to some extent,an index of the tendency to experience vi-carious arousal. Persons high on this indexare presumably used to experiencing em-pathic concern and personal distress and la-beling them as such. But the results of thisand some other studies examining individualdifferences in emotional empathy (Archer,Foushee, Davis, & Aderman, 1979; Mehr-abian & Epstein, 1972) provide a basis forspeculating that dispositional empathy andresponsiveness to social influence go hand inhand. Mehrabian and Epstein (1972) foundthat high-empathy subjects were less ag-gressive than low-empathy subjects, but onlyin the immediate condition when they wereactually in the room with their victim, facedwith his facial grimaces, gasping, and armjerking. In the nonimmediate condition, the

EMPATHY, EVALUATION, AND HELPING 795

aggression scores were slightly (and nonsig-nificantly) higher for the high-empathy sub-jects than for the low-empathy subjects.Similarly, Archer et al. (1979, ExperimentII) found in a simulated trial that subject-jurors high in dispositional empathy ratedthe defendant less guilty and assigned lessof the blame for the hypothetical stabbingto him. However, these benevolent reactionsappeared only for high-empathy subject-ju-rors who heard the defense attorney deliveran appeal to empathize—a type of demandcharacteristic. In fact, high-empathy sub-ject-jurors who heard the neutral, fact-ori-ented (nonempathy) appeal rated the defen-dant more severely than low-empathysubject-jurors. Finally, in the present study,the demand manipulation produced thehighest perceptions of the graduate student'sneed and the most volunteering among high-empathy subjects in the arousal condition,but it also led to the lowest perceptions ofneed and the least volunteering among thesesubjects when arousal was low.

In all three studies cited above, disposi-tional empathy is associated with a chame-leonlike responsiveness to the salient situa-tional variables (i.e., presence of victimfeedback, nature of the attorney's appeal,degree of arousal on the experimenter's dis-play meter). As Archer et al. and Fousheeet al. suggest, it may be necessary to reviseour traditional view of the empathic individ-ual as an unfailing altruist. Apparently em-pathy can be counted on to produce helpingonly when the social situation clearly callsfor it. As we suggested earlier, a more thor-ough study of empathy as a characteristic(or set of characteristics) of personality iscalled for.

It might be argued that social evaluation,in general, and our demand manipulation,in particular, bear a more complex relation-ship to the helping situation than we haveso far considered. By warning subjects thatthe experimenter would be monitoring theirarousal, we may have induced self-awareness(self-focus of attention; Duval & Wicklund,1972; Wicklund, 1975). In a recent reviewof the experimental literature, Wicklund(1978) concludes that "especially when thecue for helping is a prominent feature of thesituation, self-focused attention acts to bring

behavior into line with the norm of helping"(p. 516). Furthermore, our intentional con-founding of order of measurement of emo-tions with monitoring of the subject's arousalby the experimenter might have strength-ened differences in self-awareness betweenthe demand and no-demand condition. Asin the Coke et al. (1978, Experiment 2) ex-periment, demand-condition subjects wereasked to report, and thus focus, on theiremotions before the opportunity to help oc-curred, whereas no-demand condition sub-jects did not report their emotions until afterthe opportunity occurred. However, if self-focus had been manipulated through the or-der of measurement, it seems unlikely thathigh-empathy subjects exposed to the "be-fore" order (demand) would reduce theirhelping under low arousal more than theyincreased it under high arousal—or, for thatmatter, that they would reduce it at all.Based on the obtained results, the confound-ing does not appear to have had much of aneffect.

In sum, the results of our study pose someserious questions for the Coke et al. two-stage model. Strictly speaking, they do notdisconfirm the model. Indeed, the arousal-feedback effect on helping was reproduced,and empathic concern was again implicatedas a mediating variable. But apparently themodel predicts helping only for would-bebenefactors who are high in dispositionalempathy and who find themselves in sociallyevaluative circumstances. It would be pre-mature to attempt to finalize our speculationas a new model of empathy-induced helpinguntil dispositional empathy and social eval-uation have been investigated more fully.The role of public and private self-conscious-ness, in particular, deserves a closer look.However, it seems certain that the Coke etal. two-stage model must be expanded to in-corporate these factors in some form. Bothhave important motivational and cognitiveconsequences and appear to deserve equalstatus with emotional arousal and perspec-tive taking in a fully interactive model.

Reference Notes

1. Froming, W. J., & Carver, C. S. Divergent influencesof private and public self-consciousness in a com-

796 ARCHER, DIAZ-LOVING, GOLLWITZER, DAVIS, FOUSHEE

pliance paradigm. Unpublished manuscript, 1979.(Available from Department of Psychology, Univer-sity of Florida, Gainesville, Florida 32601.)

2. Piliavin, J. A., & Piliavin, I. M. The good Samaritan:Why does he help? Unpublished manuscript, 1973.(Available from School of Family Resources andConsumer Sciences, University of Wisconsin, Mad-ison, Wisconsin 53706.)

References

Althauser, R. P.. Multicolinearity and nonadditiveregression models. In H. M. Blalock (Ed.), Causalmodels in the social sciences. Chicago: Aldine-Ath-erton, 1971.

Archer, R. L., Foushee, H. C., Davis, M. H., & Ad-erman, D. Emotional empathy in a courtroom simu-lation: A person-situation interaction. Journal ofApplied Social Psychology, 1979, 9, 275-291.

Batson, C. D., Darley, J. M., & Coke, J. S. Altruismand human kindness: Internal and external determi-nants of helping behavior. In L. Pervin & M. Lewis(Eds.), Perspectives in interactional psychology. NewYork: Plenum Press, in press.

Buss, A. H. The psychology of aggression. New York:Wiley, 1961.

Buss, A. H. Self-consciousness and social anxiety. SanFrancisco: Freeman, 1980.

Carlson, J. E., & Timm, N. H. Analysis of nonorthog-onal fixed-effects designs. Psychological Bulletin,1974, 81. 563-570.

Coke, J. S., Batson, C. D., & McDavis, K. Empathicmediation of helping: A two-stage model. Journal ofPersonality and Social Psychology, 1978, 36, 752-766.

Davis, M. H. Individual differences in empathy: Amultidimensional approach. Unpublished doctoraldissertation, University of Texas at Austin, 1979.

Duval, S., & Wicklund, R. A. A theory of objectiveself-awareness. New York: Academic Press, 1972.

Fenigstein, A. Self-consciousness, self-attention, andsocial interaction. Journal of Personality and SocialPsychology, 1979, 37. 75-86.

Fenigstein, A., Scheier, M. F., & Buss, A. H. Publicand private self-consciousness: Assessment and the-ory. Journal of Consulting and Clinical Psychology,1975, 43. 522-527.

Feshbach, N. D. Empathy in children: Some theoreticaland empirical considerations. The Counseling Psy-chologist, 1975, 5, 25-30.

Foushee, H. C., Davis, M. H., & Archer, R. L. Em-pathy, masculinity, and femininity. JSAS Catalog ofSelected Documents in Psychology, 1979, 9, 85, (Ms.No. 1974).

Kenny, D. A. Correlation and causality. New York:Wiley-Interscience, 1979.

Krebs, D. L. Empathy and altruism. Journal of Per-sonality and Social Psychology, 1975, 32, 1134-1146.

Latane, B., & Darley, J. M. The unresponsive by-stander: Why doesn't he help? New York: Appleton-Century-Crofts, 1970.

Mehrabian, A., & Epstein, N. A measure of emotionalempathy. Journal of Personality, 1972, 40, 525-543.

Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner,K., & Bent, D. H. Statistical package for the socialsciences (2nd ed.). New York: McGraw-Hill, 1975.

Orne, M. T. On the social psychology of the psycholog-ical experiment. American Psychologist, 1962, 17,776-783.

Overall, J. E., Spiegel, D. K., & Cohen, J. Equivalenceof orthogonal and nonorthogonal analysis of variance.Psychological Bulletin, 1975,82, 182-186.

Regan, D. T., & Totten, J. Empathy and attribution:Turning observers into actors. Journal of Personalityand Social Psychology, 1975, 32, 850-856.

Scheier, M. F., Buss, A. H., & Buss, D. M. Self-con-sciousness, self-report of aggressiveness, and aggres-sion. Journal of Research in Personality, 1978, 12,133-140.

Scheier, M. F., & Carver, C. S. Self-focused attentionand the experience of emotion: Attraction, repulsion,elation, and depression. Journal of Personality andSocial Psychology, 1977, 35, 625-636.

Schlenker, B. R. Self-presentation: Managing theimpression of consistency when reality interferes withself-enhancement. Journal of Personality and SocialPsychology, 1975, 32, 1030-1037.

Spence, J. T., Helmreich, R., & Stapp, J. The PersonalAttributes Questionnaire: A measure of sex-role ste-reotypes and masculinity-femininity. JSAS Catalogof Selected Documents in Psychology, 1974, 4, 43.

Stotland, E. Exploratory investigations of empathy. InL. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 3). New York: Academic Press,1969.

Wicklund, R. A. Objective self-awareness. In L. Ber-kowitz (Ed.), Advances in experimental social psy-chology (Vol. 8). New York: Academic Press, 1975.

Wicklund, R. A. Three years later. In L. Berkowitz(Ed.), Cognitive theories in social psychology. NewYork: Academic Press, 1978.

Received April 28,1980