Effects of the volunteer subject, choice, and fear arousal on attitude change

JOURNAL

OF

ffects

EXPERIMENTAL

SOCIAL

PSYCHOLOGY

of the Volunteer Arousal

6,

Subject,

on Attitude

IRWIN A. HOROWITZ

293-303

AND

The University

Choice,

an

Change

WILLLAM

of

( 1970)

E. GUMENIK

Tokdo

Subjects who had volunteered (volunteers) and subjects who had refused to volunteer (nonvolunteers) for a previous experiment which was not run were recruited for the present experiment as part of a class requirement. Upon reporting, both volunteers and nonvolunteers were either given or not given choice as to participation in, and selection of experiments. The volunteering and choice variables were combined factorially with high or low fear arousal. Volunteers and subjects allowed choice exhibited greater acceptance of the recommendations of the appeal than did nonvolunteers and subjects allowed no choice, respectively. Nonvolunteers allowed choice and volunteers exhibited greater acceptance of the recommendations of the high fear appeal than of the low fear appeal. The reverse occurred for nonvohmteers allowed no choice. Results were interpreted in terms of reactance theory,

Recent evidence has indicated that volunteers and nonvolunteers are differentially affected by fear appeals in attitude change studies (Horowitz, 1969). Horowitz ( 1969) reported a positive relationship between fear and attitude change for volunteers, while a negative relationship was observed when the subjects were nonvolunteers. It seems plausible that some of the contradictory findings in the area of fear arousal and attitude change can be considered a function of the volunteering dimension While the results of the Horowitz (1969 j study can be reasonably interpreted in terms of population differences between subjects who volunteer for experiments and subjects who do not, the design did not allow for a distinction between these differences and differences attributable to different recruitment techniques employed to get vohmteer and nonvolunteer subjects. Volunteer subjects come to the laboratory of their own volition, while nonvolunteer subjects come to the laboratory to fuM1 a mandatory course requirement to participate in ~sycho~o~~ca~ research TQ state it &ffgently, nonvohmteers were coerce 295

294

HOROWITZ

AND

GUMENIK

while volunteers were not. Not surprisingly, volunteers’ perceived degree of felt coercion was significantly less than nonvolunteers, and the degree of felt coercion affected attitude change, It is therefore possible that Horowitz’s ( 1969) findings are attributable to manipulation of a coercion variable rather than to inherent differences between subjects who volunteer and subjects who do not. While volunteers and nonvolunteers clearly differ on several crucial dimensions (Rosenthal & Rosnow, 1969), these differences may be mitigated by experimental manipulations which allow subjects to perceive that they have freely chosen to participate and which therefore elicit a feeling of commitment on the part of all subjects to the experiment. It is therefore possible that if the freedom of choice of nonvolunteer subjects were less constricted, their responses would be more similar to volunteer subjects. The present experiment is designed to test, independently, the effects of differences between volunteer and nonvolunteer subject samples and two degrees of behavioral choice, high and none, on attitude change in a fear arousal situation. The volunteer and nonvolunteer samples were identified prior to, and independently of, the present experimental situation and were exclusive (subjects either volunteered or refused to volunteer for an unrelated experiment). Both subject samples were then drafted to appear in the present experiment to fulfill a class requirement. Upon appearing at the appointed place, both types of subjects were given or not given freedom of choice as to whether they wished to participate, and a choice of experiments in which they could participate. Subjects in the no-choice condition were made aware that another group of subjects had greater freedom of behavioral choice. Whenever a subject is drafted for an experiment, some constraints have been placed upon his behavior. Brehm (1966) has theorized that when a person’s behavioral freedom has been threatened or reduced, he will become motivationally aroused. The ensuing motivational state is called This motivational arousal is “. . , directed psychological reactance. toward the reestablishment of whatever freedom had already been lost or threatened” (Brehm, 1966, p. 2). In the sense that both volunteers and nonvolunteers were required to participate in the present experiment, all subjects should experience a degree of reactance initially. Subjects did not have the freedom to decide not to participate. Nevertheless, it is reasonable to assume that drafted-nonvohmteers experience greater reactance than drafted-volunteers. The drafted-volunteer should not feel as coerced as the drafted-nonvolunteer because he is being asked to do what presumably he would choose to do for reasons ,of curiosity, interest, or other personal [email protected]!v&ons, Green (1963) has

EFFECTS oiv ATTITUDE C~IANCE

295

pointed out that volunteers are interested and curious about things they are asked to participate in while nonvolunteers do not chose to par”~ ticipate because of lack of interest or antipathy to such things. The initial interest and curiosity of volunteer subjects would suggest that the freedom not to participate is less important for him than it is for the nonvolunteer subject, who, after all, has declared that given a choice, he would not participate. The drafted-nonvolunteer may attempt to reestablish his freedom of action by reacting against the agent of constraint, the experimenter. The drafted-nonvolunteer may therefore regain his freedom by resisting the experimental inductions. The greater the threat to the subjects’ freedom, the greater the reactance, and in the present experiment, the less the attitude change. The experimental induction of fear is a further source of reactance for the nonvolunteer subject. A high fear appeal threatens to eliminate even more freedom by typically requiring some specific action (i.e., stop smoking) or else some dire state of affairs (i.e,, cancer) will ensue. A high fear appeal is likely to be perceived as an attempt to induce a specific action on the part of the subject and should therefore elicit greater reactance and hence less attitude change than the low fear appeal, The act of volunteering is one of personal choice and commitment (de Charms, 1968). Presumably the draft-volunteer would have chosen to participate in the present experiment. It is reasonable to suppose that the arousal of reactance on being drafted would be subordinated to the feelings of commitment, involvement, and interest that the volunteer subject brings to participation in an experiment (Rosenthal & Rosnow, 1969). In this instance the more powerful the experimental inductions; the greater the attitude change on the part of the volunteer subject. Therefore, in line with previous findings (Horowitz, 1969) we predict that high fear appeals will have a positive affect on the attitude scores of volunteer subjects, If one can induce the feeling of freedom of choice, or relative lack of constraint, in the Iaboratory it should be possible to vitiate, to some degree, the original reactance aroused in all subjects by compulsory recruitment for the present experiment, It is very difficult to duplicate, by experimental manipulations, the feeling of freedom of behavioral choice that a subject experiences outside the laboratory in deciding, anonyII~OUS~Y, whether ,or not he wishes to participate in an experiment jdc Charms, 1968). If, however, the experimental manipulations are such that the subject perceives that he is in a condition involving fewer constraints (relatively higher choice) than subjects in a comparable group and he can in fact, within limits, make a choice as to whether or not he wishes to participate and in which of two experiments he would par-

296

HOROWITZ

AND

GUMENIK

ticlpate, then we expect a consequent reLiction in reactance. We, therefore, expect experimentally induced choice to increase attXude change. We further expect choice to increase the subject’s felt commitment and interest in the experiment and cons2quent.y to make the subject more amenable to the experimental inductions. Experimental manipulation of choice should increase the effectiveness of the high fear appeal. The no-choice condition should arouse reactance in all subjects. In this condition subjects are aware cf greater constraints placed on their behavior than on their c?oicp counterparts. In addition to being drafted, a degree of freedom of choice available to others is eliminated for them. In line with previous arguments, we expect greater reactance arousal in nonvolunteer subjects than in volunteer subjects. In the no-choice condition we have drafted-volunteers and drafted-nonvoluntezrs compelled to engage in behaviors chosen by the experimenter. For volunteer subjects, their initial interest, curiosity, and commitment should overcome, to a degree, the reactance aroused by the implied negation of choice. Nonvolunteer subjects, more negativistic to begin with, further threatened by implied loss of freedom, will be less amenable to the experimental inductions (i.e., high fear appeals) than their choice condition counterparts. In summary, we predict that allowing nonvolunteers choice and restricting volunteers’ choice would increase attitude change by decreasing reactance and decrease attitude change by increasing reactance, respectively. Manipulation of choice allows for less reactance and hence greater attitude change for all subjects. By reducing reactance and increas!ng commitment to the experiment, the choice manipulation permits the high fear appeal to be effective, and we therefore predict tl?at under high choice conditions high fear appeals will be positively related to attitude change. Based on previous findings (Horowitz, 1969), we predict that volunteers will exhibit greater attitude change than nonvolunteers. METHOD Subjects. The subjects were 120 students enrolled in the basic psychology course at the University of Toledo. Subjects populations were identified prior to the experiment and independently of the procurement technique. Several weeks prior to the experiment a request was made to volunteers for an interesting psychological experiment. The students were told that the present method of subject procurement was being evaluated and it was necessary to know how many people would volunteer for an experiment if it were not a course requirement. It was emphasized that students would not receive any credit for participation. They would be gratis volunteers. A sign-up sheet was then distributed and students were asked to indicate whether or not they would be willing to volunteer for this current experiment. The volunteers were also asked to indicate the hours they would be available. The experiment was

EFFECTS

ON

ATTITUDE

ClLkXGE

297

described as a very interesting and informative psychological experiment, No further information was given. Two weeks after the identification of volunteers and nonvolunteers, these students were informed that due to technical difficult-es the experiment had to be postponed, and the volunteers would not be needed. Design a72d procednre. Both volunteers and nonvolunteers were then recruited for the experiment in the same manner. They were given a stardard subject card requiring that they appear at the laboratory at an appointed time for a psychological experiment to fulfill a course requirement. This is the normal procedure for recruiting subjects for experiments at the University of Toledo, All subjects from the basic psychology course are recruited in this manner. Upon arrival in the laboratory subjects were assigned to one of two levels of fear arousal, high or low, and to a choice or no-choice condition. Therefore, volunteering, fear arousal; and choice were combined in a 2 X 2 >: 2 factorial design. Choice munip&tion, In the choice condition, subjects were told that it was the policy of the experimenter to give people a choice of experiments in which they might participate. The experimenter noted that several experiments were ongoing and suggested the possibility of the subject volunteering for one of two current expipsriments. Experiment A was described as “a very interesting psychological experiment,” the same description as the original, unrelated experiment which wa; employed to distinguish the volunteer and nonvolunteer samples, while B was described as being concerned with psychophysics. Subjects were informed that if neither A nor B was to their liking, they did not have to volunteer and could leave, but, of course, they would not receive credit for participation. The experimenter noted at this point ,that not all subjects had this freedom of choice and the subject had come to the laboratory at a time when several experiments were b,eing conducted. Seven nonvolunteers decided not to participate, while two subjects previously identified as volunteers did not participate. These defections were individually replacpd by their appropriate counterparts. Therefore, while 129 subjects were called to the laboratory, only 120 participated. Fifteen subjects were run in each of the eight ceils of the factorial design. Subjects assigned to the no-choice condition were told that while the experimenter understood that the subjects did not have any choice in the selection of the experiment because previous subjects had freedom to choose and had completed the subject requirements for most of the ongoing experiments, it was hoped that they would enjoy “this very interesting psychological experiment” for which “we had to grab you.” Subjects were run in groups ranging from two to five each. FGUT arousal manipulation. The fear arousal materials were presented, in part, in several pamphlets entitled “The Uses and Abuses of Drugs.” All pamphlets contained passages on the recourses citizens had for controlling and melioratiug the deleterious effects of drug abuse. Each pamphlet offered three basic recommendations for elimination of these abuses. The recommendations pertained to greater control of and increased ethical standards in drug research, increased control over the labeling and advertising of drugs, and more careful prescription and distribution of harmful drugs. High fear. Subjects in the high fear (I-IF) cond;t:on were given materials to read which emphasized case histories and vivid descriptions of the effects of drug misuse. The information and descriptions were supplied by materials published by the United States Department of HeaIth, Education, and Welfare.

298

HOROWITZ

AND

GUMENIK

The pamphlet topics included drug catastrophes, such as the thalidomide case; the indicated relationship between oral birth control pills and thrombophelebitis and pulmonary ,embolisms; the dangers of amphetamines and barbituates; the consequences of taking hallucinogenic drugs; and the sometimes near-fatal, or fatal, reactions to aspirin, penicillin, and less exotic drugs. In addition, subjects in the HF condition were shown one of two films, depending on a predetermined schedule. One film was entitled “The Mind Benders” which explored the uses and hazards of LSD and other hallucinogens. The second film pertained to amphetamines and barbituates, and included interviews with actual victims of “pill-popping.” Both films are products of the United States Public Health Service. Low fear. The low fear (LF) materials were concerned with the same topics as employed in the HF conditions differing only to the extent that the vivid descriptions of death and disability were excluded. The incidence of death and disability were included. The LF materials were written with the intent of conveying as persuasively as the HF materials the dangers inherent in drug abuse. Subjects in the LF condition did not see ‘either of the two films. Dependent measures. The degree of attitude change was obtained by having subjects respond to three lo-point scales indicating the extent of agreement with the three basic recommendations of the experimental communications. Attitude change scores were obtained by averaging the three responses for each subject. Subjects were requested to respond to a postexperimental questionnaire item which asked “To what degree did you feel coerced or forced to participate in the present experiment?” The lo-point scale was presented in the following manner: 1 2 no coercion

3

4

mild

5

coercion

7

6

moderate

8

strong

9

coercion

10

severe

coercion

To test the effectiveness of the fear manipulation, a self-report item was included on which the subjects indicated the degree of emotional arousal they felt during the experiment. This lo-point scale ranged from no arousal (1) to mild, moderate, and highly aroused ( 9 and 10). Subjects were also given two personality measures, both of which have been found to discriminate between volunteers and nonvolunteers. One measure was McDavid’s Social Reinforcement Scale ( SRS) ( 1965), the second was the California F Scale (Rosen, 1951). This was done in part to ensure that the original identification of subject samples was correct and, in part, to corroborate these previous findings. RESULTS

Efectiveness reported

arousal

ff = 5.31, greater

the

2.31,

choice

nor

revealed their

group

the

fear

=

that (F

volunteers = 17.57,

any

interactions

in

on

1 and

felt 1 and were

112

df,

significantly 112 df,

p

significant.

felt

felt

more

p

<

less .OOl).

fear 119

4.61,

coercion .Ol),

of

high

2.02, x =

the

<

analysis

the

t =

condition

9.95,

An

counterparts,

performed

no-choice

(F

subjects

significantly, low

of variance in

that

reported

than

analysis subjects

revealed volunteers effect

scores

o =

arousal

An that

of experimental manipulations.

self-

condition, df,

u = scores

coerced

than

The

analysis

coerced Neither

the

than the

p < .05, 1.74. revealed those further did fear

nonmain

in

EFFECTS

ON

ATTITUDE

CHANGE

292

Analysis of the &ta on the personality measures indicated that while volunteers and nonvolunteers did not significantly differ on the F Scab, the social reinforcement scale did successfully discriminate betwee ( w = 52.52, SD volunteers (2 = 61.70, SD = 12.31) and nonvolunteers 10.63) (t = 4.37, p < .Ol). The effects of the independent variables on attitude scores are PIGsented in Figure 1, Xnspection of Figure 1 reveals that the mean attitude scores of volunteer subjects and subjects given choice were higher (mo in accord with the recommendations of the messages) than those of noi volunteer subjects and subjects given no choice, respectively. High fear arousal resulted in higher mean attitude scores than low fear arousal for volunteer groups and for the nonvolunteer choice group, but the relationship was reversed for the nonvolunteer no-choice group. An analysis of variance indicated that the voluntcerinng main e F = 21.50, 1 and 112 df, p < .OOl, the choice main effect, F = 6.48, 1 and 112 elf, p < .05, the Fear Arousal X Volunteering interaction, F =

8.0 9OLUMIEER:CHOICE

7.a 9OLUNTEER:NO

g 5:

CHOICE

6.C

x 3 I= 5

5.c

s Y

4s ,

NON-VOLUNTEER

I

:NO CHOICE

HIGH

LOW

FEAR AROUSAL FIG.

choice

1. Mean attitude for low and high

scores for volunteers fear arousal.

and

nonvolunteers

with

and

without

300

HOROWITZ

AND

GUMENIK

4.62, 1 and 112 df, p < .05, and the triple order interaction, F = 19.68, 1 and 112 df, p < ,001 were significant. No other main effects or interactions were significant. Further analysis of simple effects revealed that for the volunteer and choice conditions the fear arousal effect was significant, F = 9.23, 1 and 112 df, p < .Ol and F = 5.85, 1 and 112 df, p < .05, respectively, while interactions with fear arousal were not. In contrast, for the nonvolunteer condition the Fear Arousal x Choice Interaction, F = 10.12, 1 and 112 df, p < .Ol, and for the no-choice condition, the Fear Arousal x Volunteering interaction, F = 5.85, 1 and 112 df, p < .05, were significant, while the fear arousal effect was not, Also the choice effect was not significant for the volunteer condition, F = 3.23, 1 and 112 df, p > .05, or for the low fear arousal condition F = 1.72, 1 and 112 df, p > .05. DISCUSSION

Volunteers and subjects allowed choice gave attitude scores which were more in accord (higher) with the recommendations of the experimental message than nonvolunteers and subjects not allowed choice, respectively. Furthermore, for volunteers and/or subjects allowed choice, high fear arousal resulted in higher attitude scores than did low fear arousal. The positive effect of high fear arousal on attitude scores, however, did not occur for nonvolunteers who were not allowed choice. It appears that allowing nonvolunteers choice resulted in behavior more similar to volunteers in that attitude scores were raised and high fear arousal elicited a positive effect on attitude scores. The reported felt coercion score results indicated that subjects having no choice and nonvolunteers reported more felt coercion than subjects having a choice and volunteers. In the case of the choice variable, the explanation of these results is self-evident. However, while volunteers had volunteered for an unrelated experiment which was not run, they were recruited for the present experiment in the same manner as nonvolunteers, i.e., told to report as part of a classrequirement. It is probable that volunteers even when they are drafted for an experiment participate more willingly than nonvolunteers. It should be noted that the nonvolunteer sample was drawn from students who initially refused to participate in an experiment. It appears plausible that nonvolunteers who were allowed no choice, being most coerced, responded with the greatest degree of reactance. Th? reactance was probably also increased by the implication that choice had be-n allowed previously and was eliminated for these subjects. It also appears reasonable that the high fear appeal represented a greater threat to the subjects’ freedom than the low fear appeal. The nonvolunteer no-choice subject may have reestablished his freedom by not

EFFECTS

ON

ATTITUDE

CHANGE

301

acting in accordance with the recommendations of the high fear appeal. Allowing nonvolunteers choice, on the other hand, may have mitigated the reactance, resulting in higher attitude scores and a positive effect ot high fear arousal on attitude scores, i.e,: in their acting more like volunteer subjects. Thus, optimal levels of fear arousal differ between vounteers or subjects havng choice and nonvolunteers having no choice. A comparison with previous studies must be conjectural because of differences of raciuitment. The present volunteer choice condition is analogous to the techniques usually used by investigators in recruitirrg volunteer subjects, that is, volunteers participated in the experiment of their own volition. Studies using volunteer subjects (Dabbs & Leventhal, 4968; Horowitz, 1969; Leventhal & Singer, 1966) found a positive relation between fear arousal and attitude change. The nonvoluntesr sample used in this experiment was drawn from students who had previously refused to participate in a research project. Typically, nonvolunteer subjects are required to participate to meet a class requirement without being asked if they wished to vohmteer or not. No doubt, this typical sample includes subjects who would vohmteer if given the opportunity as well as those who would not. The present nonvolunteer no-choice condition is, however, akin to a condition used by Horowitz ( 1963) in which he found a negative relation between fear arousal and attitude score. Therefore the results of oxur two extreme groups, the volunteer choice and nonvolunteer no choice, ar-e in agreement with the previous findings in the literature on the relationship between fear arousal and attitude change. While the reactance approach appears to us to be the most plausib!e manner of explaining the present Endings, it should be noted that cognftive dissonant e may offer an alternative explanation, Since vohmteering implies commitment (Rosenthal, 1965), it is possible that volunteers and nonvolunteers in the choice condition experience dissonance, and the higher the fear arousal, the greater the dissonance. That is, since subjects in the choice condition essentially volunteered to expose themselves to the communication, they can reduce the dissonance b!accepting the recommendations of the message. Nonvolunteers in the no-choice group, not having made a commitment, should not be in a state of dissonance and hence higher fear does not elicit greater acceptance of the recommendations. However, the volunteers in the no-choice condition did not make an overt choice (i.e., a commitment) to participate and therefore, again, should not be in a state of cognitive dissonance. Nevertheless, high fear elicits greater acceptance of the recommendations for them than low fear. It is difficult to see how the dissonance explanation can be applied to the latter group. Several methodological problems should be noted at this point. 11: the

302

HOROWITZ

AND

GUMENIK

choice condition there were seven defections from the nonvolunteer group (out of 37) (n = 30) and two defections from the volunteer group (out of 32). While these defections give us greater confidence that our choice manipulation was successful, it is probable that those subjects who chose to leave the experiment were those least likely to be effected by the experimental manipulations and hence less likely to exhibit attitude change as a function of these inductions. The gain in experimental credibility (choice) in this instance may have been vitiated to an extent by loss of control. A second problem involves the assumption on the part of the experimenters that we have clearly identified the volunteer population. Since all subjects were recruited the same way, i.e., drafted, we have no assurance that subjects who volunteered for the unrelated experiment would willingly volunteer for the present experiment. However, it should be noted that those subjects designated as volunteers did not originally volunteer for an experiment in a specific area. The unrelated experiment was described as a very interesting and informative psychological experiment, which might, presumably, have involved anything within the area of psychology. Furthermore the volunteer sample actually used in the current experiment did differ significantly in their responses to the McDavid Social Reinforcement Scale (1965) from their nonvolunteer counterparts. McDavid (1965) has reported that the SRS scale has distinguished between volunteer and nonvolunteer populations. A third problem involves the manipulation of fear. The attempt was made to separate the fear arousal properties of the experimental messages from their persuasibility and interest properties. Differences between the high and low fear materials should, of course, be only in terms of arousal properties. However, the high fear group was shown two movies not seen by the low fear group concerning the abuses of drugs. While these films no doubt enhanced the fear arousal properties of this condition they also, almost certainly, introduced extraneous nonfear factors as well. The films no doubt were interesting, informative, and persuasive. Unfortunately, the independent check of arousal, the self-report scales, is not very helpful. While the data indicates that the high fear group was more aroused than the low fear, those results may be due to interest, persuasiveness, fear, or in some cases, possibly reactance. REFEREN’CES BREHM, DABBS,

J. W. A theory of psychological reactance. New York: Academic J. M., & LEVENTHAL, H. Effects of varying the recommendations

Press, 1966. in a fear

EFFECTS

ON

ATTITUDE

CHANGE

303

arousing communications. Journal of Personality and So&al Psychotogy, 1966> 4, 525-531. DE CHARMS, R. Personal causation. New York: Academic Press, 1968. GREEN, D. R. Volunteering and the recall of interrupted tasks. Journal of Abnormal: and Social Psychology, 1963, 66, 397-401. HOROWITZ, I. A. Effects of volunteering, fear arousal, and number of communications on attitude change. Journal of Personality and Social Psychology, 1969, II, 34-37. LEVENTIIAL, I-I., & SINGER, R. Affect arousal and positioning of recommendations in a persuasive communication. Journal of Personality and Social Psycholngy, 1966, 4, 137-146. MCDAvID, W. S. Approval-seeking motivation and the volunteer subject. Journal of Personality and Social Psychology, 1965, 2, 114-117. ROSEN, E. Diiferences between volunteers and nonvolunteers for psychological studies. Journal of Applied Psychology, 1951, 35, 185-193. ROSENTHAL, R. The volunteer subject. Human Relations, X%5, 18, 389-406. ROSENTHAL, El., & ROSNOW, R. The volunteer subject. In R. Rosenthal and R. Rosnow (Eds.), Artifact in Behavioral Research. New York: Academic Press, 1969. (Received

May

9, 1969)