570 5pages

ThirdCritique.pdf

Personal Responsibility and Altruism in Children

Geoffrey Maruyama University of Minnesota

Scott C. Fraser and Norman Miller University of Southern California

To assess the applicability of Zimbardo's deindividuation model to the study of prosocial behavior, groups of Halloween trick-or-treaters were asked to donate candy to hospitalized children under three conditions that manipulated the des- ignation of personal responsibility for the amount donated: no child identified as responsible; one child identified as responsible; and each child identified as responsible. As predicted, personal responsibility increased the likelihood of con- tributing as well as the number of candies donated. Examination of the contri- bution rates in different size groups indicated that assigning individual respon- sibility increased the number of candies donated by each child only if they were in small groups. The absence of a similar outcome among children in large groups suggests that the greater social support and/or the greater potential for the diffusion of responsibility that is available to those in larger groups effectively countered the effect of the manipulation of responsibility.

In 1970 Zimbardo proposed a three-stage model to explain the occurrence of behaviors that are normally inhibited by social norms and sanctions. The first stage focuses on an- tecedent conditions such as anonymity or social support from others that augment nor- mally restrained behavior. The second stage is concerned with the process that occurs when, in response to the cues provided by the facilitating antecedent conditions, indi- viduals are released from personal and social constraints. The release presumably reflects a reduced concern for self-evaluation and/ or beliefs that personal consequences of a negative evaluation by others can be mini- mized. The third stage, the emergence of behaviors that Zimbardo called deindivi- duated, results from the changes in internal state produced in the second stage.

The authors wish to acknowledge the assistance of the Social Science Research Institute, University of Southern California, in preparing this article for pub- lication.

Requests for reprints should be sent to Norman Miller, Department of Psychology, University of South- ern California, Los Angeles, California 90007.

Research examining the plausibility of this deindividuation model has focused on how deindividuation can increase the occur- rence of antisocial, aggressive types of be- havior.1 However, the model should also be able to account for the occurrence or non- occurrence of prosocial behaviors. The pres- ent study uses the Halloween paradigm (e.g., Diener, Fraser, Beaman, & Kelem, 1976) to extend the applicability of the model to prosocial behavior. Groups of children trick or treating on Halloween night were asked to donate candy to hospitalized children. Personal responsibility for the amount do- nated was manipulated by identifying either no child, one child, or all the children within each group as responsible. We assumed that when the children entered the houses se-

' Note that this article does not speak to another con- ceptualization of deindividuation that views it as a loss of personal identity (e.g., Dipboye, 1977; Snyder & Fromkin, 1980). Unlike Zimbardo's view, such a view argues that deindividuation is an unpleasant state to be avoided. In contrast Zimbardo's model views it as a situationally induced state that allows people to behave as they please without regard for social constraints.

658

PERSONAL RESPONSIBILITY AND ALTRUISM 659

lected as experimental sites, they were in a deindividuated state and that the designa- tion of responsibility should act to indivi- duate them. This, in turn, should make them more willing to give their own candy to hos- pitalized children.

To develop our rationale for the study, we will briefly examine the deindividuation model, discuss the relation between personal responsibility and prosocial behavior, and then, consider the implications of these ideas for the Halloween paradigm.

Deindividuation

Research examining the plausibility of Zimbardo's model generally has tended to validate it (e.g., Diener, Dineen, Endresen, Beaman, & Fraser, 1975; Donnerstein, Don- nerstein, Simon, & Ditrichs, 1972; Zabrack & Miller, 1972; Zimbardo, 1970); antisocial behavior has been found to increase directly with manipulations of anonymity, diffusion of responsibility, social support, and arousal. In addition studies concerned with children's transgressions such as stealing candy or money, provide further support for it (e.g., Diener et al, 1976; Diener, Westford, Di- ener, & Beaman, 1973). A few studies, how- ever, do report seemingly contrary findings (e.g., Diener, 1976; Diener, Westford, Di- neen, & Fraser 1973; Zimbardo, 1970).

Recently, Diener (1979) revised the dein- dividuation model in order to integrate rel- evant findings and to provide a focus for fu- ture work. In particular, he examined the link between individuation and self-aware- ness (e.g., Beaman, Klentz, Diener, & Svanum, 1979; Diener, & Walbom, 1976; Ickes, Layden, & Barnes, 1978) and argued that it is the minimization of self-awareness that accounts for the occurrence of deindi- viduated behaviors. That is, deindividuation will occur only if the antecedent conditions described by Zimbardo act to block an in- dividual's self-awareness (and self-con- sciousness). Diener's arguments suggest that the instances in which the model is not sup- ported probably reflect the fact that the ex- perimental manipulations did not act to de- crease self-awareness. For example, Zim- bardo's (1970) attempt to deindividuate Belgian soldiers probably failed because

their uniforms already provided them with a source of deindividuation; the addition of cumbersome robes and hoods probably acted to increase self-awareness and thereby in- dividuate them rather than further deindi- viduate them. Thus, Diener shifts attention beyond simple examination of the relations between specific antecedent conditions and deindividuation and focuses instead on the underlying process events. His analysis im- plies that to successfully manipulate dein- dividuation or individuation, it is important to use manipulations that affect self-aware- ness.2

Diener (1979) also has attempted to clar- ify ambiguity about standards or norms. He argues persuasively against the view that deindividuated behavior reflects new stan- dards; in other words it does not result from a new set of deviant norms that rapidly de- velop and emerge from the specific setting. By arguing that the deindividuation is not likely to be mediated by a relaxation of ex- isting norms or by the introduction of new deviant norms, Diener avoids the dilemma of deciding which set among multiple sets of norms should be used to define deindivi- duation. Instead, it is defined in terms of those general social norms that will be made salient by self-focused attention regardless of situational circumstances.

Personal Responsibility and Prosocial Behavior

Prosocial behavior typically is seen as in- volving aspects of self-reward as well as re- ward that is given to others (e.g., Krebs, 1970). It even has been argued that prosocial behavior is primarily hedonistic (e.g., Cial- dini, Darby, & Vincent, 1973). People re- ward others because they expect to be re- warded in return either with tangible goods or by recognition for being kind. People ex- pect to receive such rewards in return only to the extent that they are recognized as being responsible for having provided the

2 It is important to note that Diener (1979) concep- tualizes deindividuation as more than simply a lack of self-awareness. He contends that "automatic," well- learned behaviors (e.g., driving to work) involve non- self-awareness, but not deindividuation.

660 G. MARUYAMA, S. FRASER, AND N. MILLER

necessary benefits to the person in need. In other words, insofar as receiving self-reward is tied to recognition, identification as the person responsible for providing the help is critical. In terms of this analysis, if a person is designated as personally responsible for the help given to another, the inclination to offer it should increase.

Turning to the implications of the dein- dividuation model, it should first be noted that societal norms generally promote altru- ism and that prosocial behavior is typically viewed as consistent with such norms. There- fore, from the standpoint of the deindivi- duation model, prosocial behavior should be reduced by deindividuation. When instead self-focus is increased, behavior that is con- sistent with altruistic norms should also in- crease. The designation of personal respon- sibility for the amount of help that is given to someone in need appears to be a direct means of increasing self-focus and thereby attenuating a person's level of deindividua- tion.

Halloween Paradigm

Halloween provides a good natural setting for testing the application of the deindivi- duation model to prosocial behavior. Chil- dren who are trick-or-treating should expe- rience many of the antecedent conditions for a deindividuated state. Their costumes gen- erally make them anonymous; they are ex- cited (high arousal); and, since they typically trick-or-treat with friends, they have social support and can readily diffuse responsibility for any antisocial behavior in which they engage. Certainly, the many occurrences of petty vandalism on Halloween, such as toilet papering trees and houses or soaping win- dows and cars, provide strong anecdotal ev- idence for a changed mental state. Within this setting the deindividuation model can be tested by removing one or more of the ingredients for a deindividuated state, thereby restoring children to a more indi- viduated state. As argued above, making children personally responsible should in- crease their self-awareness and thereby make them more inclined to judge their behavior against normative standards. This, in turn, should increase the occurrence of prosocial behavior.

.Because children trick-or-treat in groups that vary in size, the effects of group size warrant additional consideration. Prior stud- ies on both deindividuation and prosocial behavior have found that subjects in larger groups more readily diffuse responsibility among group members (e.g., Diener, 1979; Krebs, 1970). One interpretation of this finding is that children in larger groups are less self-focused and have less concern about negative evaluation by others. In addition, however, because groups are self-selected, their members should provide social support to one another. Thus, the larger the group, the more social support available to any in- dividual member. These considerations ar- gue that children in larger groups should exhibit less prosocial behavior than those in smaller ones.

Method Subjects were 177 children who were trick-or-treating

in 56 groups within the Los Angeles metropolitan area. They ranged in estimated age from 4 to 13 years. Groups generally had two to five persons, although some single groups had six, seven, and eight children. Eighty-nine children were identified as male, 72 as female; the sex of 16 children could not accurately be determined. Fe- male experimenters and male raters from research classes at the University of Southern California were assigned to each of six houses chosen as experimental sites. In the first room of each house, a female met children who were trick-or-treating on Halloween night. She was the only person visible in each house. In the front room of each house, hospital posters and pictures of hospitalized children were prominently placed. The female greeted the children and allowed them to enter, leaving the door open. She individually complimented each child on his or her costume. She then informed the children that she was sorry but that she was completely out of candies. Pointing to a large poster containing pictures of hospitalized children, she then told them that she was trying to collect candies for hospitalized children who were unable to go trick-or-treating. She asked the children to donate candies to the hospitalized children by putting them into a box.3 The specific instructions about the donation comprised the manipulation of re- sponsibility. The three conditions were as follows:

1. No designated responsibility. Pointing to the con- tribution container, the experimenter said, "Please give them as many as you want by putting the candies in the white box on the table. Then you can leave."

2. One child responsible. One child was arbitrarily selected by the experimenter. She pointed at the child and said, "I will put you in charge of the group here." Looking in turn at each of the other children and again

3 All donated candies were delivered to Children's Hospital in Los Angeles.

PERSONAL RESPONSIBILITY AND ALTRUISM 661

pointing, she said, "She/He is the leader of the group, and I will put her/his name on the bag of candy I give tomorrow to the hospitalized children. She further said, "Here is a badge for you to be head of the group," and placed a badge on the selected child, continuing, "Please give as many candies as you all want by putting the candies in the white box on that table. Then you can leave."4

3. Each child responsible. Pointing to each child in turn, the experimenter said, "I will be counting on you and you and you . . . I will put each of your names on the bag of candy I give tomorrow to the hospitalized children. Here is a badge for each of you to be in charge." She than placed a badge on each child, and after each badging, she said, pointing to the appropriate container, "Please put as many candies as you want in the (color) box on that table. Then you can leave." In this condition a distinctively colored box was available for each individual child.

After administering the instructions, the adult left "to finish my work in the other room." The boxes were constructed so that candy contributed by any one child was not visible to other children. In no condition did the experimenter specify which child should go to the box(es) first. Also, at no time during her contact with them did she ask any of the children their names. A male rater, hidden behind a partition, unobtrusively observed the behavior of the children, recorded the do- nations of each child (whether the child contributed any candy and the amount of any donation), and estimated demographic characteristics of each child, such as age and sex. After the children had time to contribute and were about to leave the house, the adult returned, told the children that she had just found some more candies, and gave some to each of them. If any child'asked whether the experimenter was going to write down his/ her name to put on the candy, the experimenter wrote down the name of the child on a piece of paper. Between experimental sessions the rater emptied 'the boxes and validated his observations against the actual amount of candy given whenever possible. If the estimate could not be reconciled with the actual amount of candy, the data was discarded (scores for five individuals or two groups had to be discarded). At each house the first replication of each condition was used as an additional training session and, therefore, was'omitted from the analyses; the sample size of 177 subjects excludes 20 omitted groups (i.e., three groups from each of six sites) plus 2 with missing data.

Results

Because the manipulations were imposed on groups of children, we will-first examine the group data, presenting individual data to assess effects of age (and sex) on contri- bution rates and to augment findings of group data. The dependent measures are (a) whether children contributed candy and (b) the amount of candy contributed. In order to examine the effects of group size, groups were classified as "small" (two or three chil- dren) or "large" (four or more children).

Preliminary examination of the data ad- dressed two issues: the overall rate of oc- currence of prosocial behavior and the pos- sibility that the average group size differed between conditions. First, the overall rate of prosocial behavior was high. In over 86% of the group, each child donated at least one candy. Over 92% of the children donated one or more candies, and over 40% of the chil- dren donated 3 or more candies. Second, there were virtually no differences in average size of the groups in each condition, F(2, 50) = .83, and no interaction of condition by group size (large/small), F(2, 50) = 2.12. Therefore, group size cannot account for any between-condition differences.

Because virtually all children contributed candy, our analyses focus on the number of candies contributed; data were analyzed us- ing a 3 (condition) X 2 (group size) analysis of variance. In order to reduce somewhat the large between-condition variance caused by the positive skew of the data, the dependent measure for this and all other group analyses was the square root of 1 plus the average contribution per child. (To aid interpreta- tion, however, we report untransformed con- tribution rates.)5 This analysis yielded main effects for the responsibility manipulation, F(2, 48) = 3.66, p < .05, and for group size, F(l, 48) = 8.87,/x .01, and no interaction. When each child was made responsible, chil- dren donated an average of 5.0 candies, com- pared with 3.3 candies when one child was designated as responsible and 2.2 when no child was designated as responsible.6 Chil-

4 Experimenters were instructed to randomize their selection of leaders as much as possible, that is, not to systematically select the largest or smallest child or the natural leader. Note that the designation as leader only referred to whether a child was told his/her name was to be put on the bag of candy that was to be given to the hospitalized children. The three authors inspected and observed all sites to assure that the procedures were being properly followed. Because random selections of leader were not assured, however, the results are sus- ceptible to aternative interpretations that rest on how or why selection of leaders may have affected contri- bution rates.-

5 In similar analyses of the untransformed data, the effects reported below are stronger.

6 When the authors used a modified least square dif- ference approach that yields exact tests for unequal cell size, they found that in post hoc analyses, only the two extreme groups were significantly different from one another.

662 O. MARUYAMA, S. FRASER, AND N. MILLER

dren in large groups donated an average of only 2.0 candies, compared with 4.0 candies by children in small groups.

Individual level data found not only main effects for conditions, F(2, 165) = 6.75, p < .01, and group size, F(l, 165) = 18.6, p<.001, but an interaction as well, F(2, 165) = 5.64, p < .01. As responsibility was increased from none to one to each child, contributions in small groups increased; mean number of candies donated was 2.44, 3.77, and 6.96, respectively. In large groups, however, the conditions did not differ; means were 1.76, 2.30, and 1.92, respectively.7 Fur- ther analyses comparing responsible and nonresponsible children within the one-child- responsible condition found no differences between them.

Recently, Kenrick, Baumann, and Cial- dini (1979) have provided a theoretical framework for examining the socialization of prosocial behavior (see also Cialdini & Kenrick, 1976). They argue that for adults personal identification is not necessary be- cause they have internalized societal stan- dards and consequently find prosocial be- havior personally rewarding even when its occurrence is not public. Further, these au- thors have suggested that internalization of society's standards for prosocial behavior occurs during the latter primary grades (ages 10+). Thus, young children, who have not yet internalized standards for prosocial behavior should behave prosocially only in instances in which they are made responsible for the help that another receives and can be identified as having provided it. Only then can others reward them for their behavior. When applied to the present data, their po- sition argues that among young children pro- social behavior should occur only when they can be identified as responsible for it; when children are not identified as responsible, lit- tle or no prosocial behavior should occur. In contrast, for older children prosocial behav- ior should occur irrespective of whether or not personal responsibility has been desig- nated. To test these ideas, additional anal- yses examined the effects of age in conjunc- tion with condition and group size. Age was trichotomized and produced groups of ages 4-7, 8-9, and 10-13. We found no age main

effect and no interaction of age with either group size or condition. In conjunction with the relatively high contribution rate found even in the no-responsibility condition, the results offer no support for the developmen- tal model of Kenrick et al. (1979).8

Discussion

The results are consistent with the dein- dividuation model. They argue that pro- cesses that spontaneously occur in large groups diminish individual prosocial behav- ior. Furthermore, they suggest that when individuals are embedded in large groups, they are not affected by procedures that at- tempt to increase their contribution rates by identifying them as personally responsible; the effect of this manipulation was apparent only in smaller groups. As argued in the in- troduction, one plausible interpretation of this effect is that responsibility is more readily diffused in larger groups. Diener, Lusk, DeFour, and Flax's (1980) finding that self-awareness is inversely related to group size fits nicely with this explanation. On the other hand, as noted earlier, in self- selected groups it seems particularly likely that social support will covary directly with group size and thus provide an alternate ex- planation of this effect.

As well as demonstrating generalizability of the deindividuation model to prosocial behavior, the present application may make clearer its conceptual base with respect to the terms normative and normal restraints. Although Zimbardo used the two terms in- terchangeably, a distinction between them becomes important when examining proso-

7 Although this interaction was not significant in the group data, the means do parallel closely those of the individual data. In small groups, as responsibility in- creased from none to one to each child, the means were 2.35, 3.85, and 6.57, respectively, whereas in large groups they were 1.69, 2.33, and 1.88. The failure to obtain an interaction in the group data to some degree reflects the reduced degrees of freedom in the group level analysis.

8 Similar analysis of sex failed to yield either a main effect for sex or any interactions involving sex. (For the last two sets of analyses, it is important to remember that age and sex classifications were estimated by adult raters and are probably somewhat unreliable.)

PERSONAL RESPONSIBILITY AND ALTRUISM 663

cial behaviors. For example, a prosocial be- havior such as contributing to a charity, al- though normatively, encouraged, may occur with low frequency and consequently might be viewed as "normally restrained." When this latter aspect of charitable contributions is emphasized, it makes the model appear to argue that such behavior will increase under conditions of deindividuation. Clearly, however, the term normative rather than normal is the one that is consistent with the intent of the deindividuation model; regard- less of initial frequency of a given behavior, a deindividuated state should lead to a de- creased occurrence of normatively sanc- tioned behavior. In the present instance the uncurbed deindividuated state produced by Halloween reduced charitable donations, whereas individuating children by assigning personal responsibility to them increased their compliance with social responsibility and helping norms.

The bearing of the deindividuation model on the broader range of prosocial behaviors also warrants consideration. Although the most common forms of prosocial behavior do seem to be supported by social norms, nonnormative instances of prosocial behav- ior, such as one male hugging another to express his joy in seeing him, do occur. Dein- dividuation should facilitate rather than in- hibit the occurrence of such nonnormative instances of prosocial behavior.

Finally, the fact that our own particular manipulation of responsibility did effectively increase the occurrence of normatively sanc- tioned behavior is interesting because it de- fines a boundary condition for countering the effect of diffusing responsibility. Diener et al. (1979) found that appointing an anony- mous individual to be responsible increased the antisocial behaviors of the group. The assignment of responsibility in the present study, which in addition told subjects that they would be identified by name, produced the opposite effect, that is, it increased rather than decreased normatively sanctioned be- havior. When taken together, these findings suggest that personal identification is a nec- essary ingredient if one wishes to induce greater helping by assigning individual re- sponsibility; the mere assignment of respon-

sibility (Diener et al., 1976) did not increase normative behavior.

In conclusion the results provide evidence that the deindividuation model can be ap- plied to prosocial behaviors. Not only does being identified as responsible increase the occurrence of normative prosocial behavior, but in addition its effects operate primarily in smaller groups.

References

Beaman, A. L., Klentz, B., Diener, E., & Svanum, S. Self-awareness and transgressions in children: Two field studies. Journal of Personality and Social Psy- chology, 1979, 37. 1834-1841.

Cialdini, R. B., Darby, B. L., & Vincent, J. E. Transgression and altruism: A case for hedonism. Journal of Experimental Social Psychology, 1973, 9, 502-516.

Cialdini, R. B., & Kenrick, D. T. Altruism and hedon- ism: A social development perspective on the rela- tionship of negative mood state and helping. Journal of Personality and Social Psychology, 1976, 34. 907- 914.

Diener, E. Effects of prior destructive behavior, anon- ymity, and group presence on deindividuation and aggression. Journal of Personality and Social Psy- chology, 1976, 33, 497-507.

Diener, E. Deindividuation, self-awareness, and disin- hibition. Journal of Personality and Social Psychol- ogy, 1979, 37, 1160-1171.

Diener, E., Dineen, J., Endresen, K.., Beaman, A. L., & Fraser, S. C. Effects of altered responsibility, cog- nitive set, and modeling on physical aggression and deindividuation. Journal of Personality and Social Psychology, 1975, 31, 328-337.

Diener, E., Fraser, S. C., Beaman, A. L., & Kelem, R. T. Effects of deindividuation variables on stealing among Halloween trick-or-treaters. Journal of Per- sonality and Social Psychology, 1976, 33, 178-183.

Diener, E., Lusk, R., DeFour, D., & Flax, R. Deindi- viduation: The effects of group size, density, number of observers, and group member similarity on self- consciousness and disinhibited behavior. Journal of Personality and Social Psychology, 1980, 39, 444- 459.

Diener, E., & Wallbom, M, Effects of self-awareness on anti-normative behavior. Journal of Research in Personality, 1976, 10, 107-111.

Diener, E., Westford, K. L., Diener, C., & Beaman, A. L. Deindividuation effects of group presence and arousal on stealing by Halloween trick-or-treater. Proceedings of the 81st Annual Convention of the American Psychological Association, 1973, 8, 219- 220.

Diener, E., Westford, K. L., Dineen, J., & Fraser, S. C. Beat the pacifist: The deindividuating effects of anonymity and group presence. Proceedings of the 81st Annual Convention of the American Psycholog- ical Association, 1973, 8, 221-222.

664 G. MARUYAMA, S. FRASER, AND N. MILLER

Dipboye, R. L. Alternative approaches to deindividua- tion. Psychological Bulletin, 1911,84. 1057-1075.

Donnerstein, E., Donnerstein, M., Simon, S., & Di- trichs, R. Variables in interracial aggression: Anon- ymity, expected retaliation, and a riot. Journal of Personality and Social Psychology, 1972, 22, 236- 245.

Ickes, W., Layden, M. A., & Barnes, R. D. Objective self-awareness and individuation: An empirical link. Journal of Personality, 1978,45, 146-161.

Kenrick, D. T., Baumann, D. J., & Cialdini, R. B. A step in the socialization of altruism as hedonism: Ef- fects of negative mood on children's generosity under public and private conditions. Journal of Personality and Social Psychology, 1979, 37, 747-755.

Krebs, D. L. Altruism—an examination of the concept

and a review of the literature. Psychological Bulletin, 1970, 73, 258-302.

Snyder, C. R., & Fromkin, H. L. Uniqueness: The hu- man pursuit of difference. New York: Plenum, 1980.

Zabrack, M., & Miller, N. Group aggression: The ef- fects of friendship ties and anonymity. Proceedings of the 80th Annual Convention of the American Psy- chological Association, 1972, 7, 211-212.

Zimbardo, P. G. The human choice: Individuation, rea- son, and order versus deindividuation, impulse, and chaos. In W. J. Arnold & D. Levine(Eds.), Nebraska Symposium on Motivation (Vol. 17). Lincoln: Uni- versity of Nebraska Press, 1970.

Received June 22, 1981 •