Journal of Personality and Social Psychology 1972, Vol. 24, No. 2, 207-21.?

PHYSICAL ATTRACTIVENESS AND EVALUATION OI< CHILDREN'S TRANSGRESSIONS ' KAREN K. DTONUniversity of Minnesota Preliminary evidence indicates that effects of a physical attractiveness stereotype may be present at an early childhood developmental level. Several of the mediating processes that may be hypothesized to be responsible for these effects presuppose thai, adults display differential treatment toward attractive and unattractive children in circumstances in which their behavior is identical. The present study is addressed to the lenabilily of (his assumption in a situation integral to the socialization process, that in which the child has committed a transgression and the socializing adult must evaluate the child's behavior. Within a 2 X 2 X 2 X 2 design (Attractiveness of Child X Severity of Transgression X Sex of Child X Type of Transgression), support was found for the hypothesis that the severe transgression of an attractive child is less likely to be seen as reflecting an enduring disposition toward antisocial behavior than that of an unattractive child. In addition, the transgression itself tends to be evaluated less negatively when committed by an attractive child. No differences in intensity of advocated punishment were found. These and additional findings arc discussed.

Accumulating research indicates that an individual's physical attractiveness is an important social cue used by others as a basis for social evaluation. It has been demonstrated, for example, that physically attractive young adults, both male and female, are assumed to possess more socially desirable personalities and to lead more successful and fulfilling lives than are those of lesser attractiveness (Dion, Berscheid, & Walstcr, 1972). There is also some suggestion that the effects of a physical attractiveness stereotype may be present at an early developmental level. Examination of sociometric choice behavior in a group of nursery school children revealed that a preschooler's level of physical attractiveness, as judged by adults, bears relation1 This study was completed in partial fulfillment of the requirements of the PhD degree in psychology at the University of Minnesota. The research was facilitated by National Institute of Mental Health Grant 16729 to Ellen Berscheid. The author wishes to express special thanks to her dissertation advisor Ellen Berscheid, and to committee members Karl Weick, Paul Rosenblatt, John Masters, and Raymond Collier for their helpful advice and suggestions. The author also wishes to thank Kenneth Dion for his useful comments and criticisms. 2 Requests for reprints should be sent to Karen K. Dion, who is now at Department of Psychology, University of Toronto, Scarborough College, West Hill, Ontario, Canada.

207

ship to the extent to which he is popular with his peers and to the extent to which he is perceived to exhibit certain types of behaviors in interaction with them (Dion & Berscheid, 1971). Although this preliminary evidence indicates that manifestations of the stereotype may be present at an early age, it is not clear what factors mediate the observed relationships. Among the variety of intervening processes that present themselves for consideration is the possibility that preschoolers have already absorbed the adult cultural stereotype and distort their perception of their peers to fit the stereotype. Another alternative is that affect toward attractive and unattractive peers is generated directly from differential behaviors displayed by these peers in interaction. Each of these possible mediating factors presupposes that adults display differential treatment toward attractive and unattractive children, even under circumstances in which their behavior is identical. The present study investigates the tenability of this assumption in a situation integral to the socialization process, namely, one in which a child has committed a transgression and the socializing adult evaluates the child. One might well expect that certain aspects of the transgression itself, such as its seriousness or

208

KAUKN K. DION

severity, would influence an adult's reactions to the child. Indeed there is some evidence that adults assign greater responsibility for an action to a child if the outcomes of his behavior are negative (cf. Shaw & Sulzer, 1964). It seems plausible, however, that a child's personal characteristics may also influence an adult's evaluations of his behavior. Some derivations from the principle of cognitive consistency suggest that individuals strive for internally consistent impressions of others. Heider (1958), for example, speculated that knowledge of a person's general character may influence how others interpret his behavior. People may interpret an individual's actions in a manner consistent with their knowledge or expectations about his personal dispositions. Thus, if adults believe that children differing in attractiveness typically display different personal characteristics, these expectations may affect their evaluation of an attractive versus an unattractive child's behavior. In the present study, it was assumed that adults hold a physical attractiveness stereotype of children similar to that held for other adults. If so, adults should expect that physically attractive children typically engage in more socially desirable behavior than do unattractive children. The knowledge that an attractive child has committed a harmful act is obviously inconsistent with these expectations. Reasoning from a general consistency principle, therefore, adults should attempt to resolve the discrepancy between their expectations and the child's behavior. Various modes of resolution might conceivably be used to reduce the inconsistency, depending on the severity of the transgression. If the offense is relatively mild, judgmental dimensions, such as evaluations of the transgression's undesirability, may be primarily affected. A minor transgression committed by an attractive child may be assimilated into the generally positive stereotype for physically attractive children, and thus be regarded as less serious or less undesirable than the same offense committed by an unattractive child. It might, however, be more difficult to interpret an action that has serious conse-

quences so that it can be assimilated without difficulty into the prevailing stereotype of attractive children. Presumably a severely harmful act evokes stronger negative reactions than a mild offense; there should be less ambiguity about its undesirability. Consequently, when harm-doing is severe, attributional inferences (cf. Heider, 1958; Jones & Davis, 196S) 3 may be primarily influenced by the child's attractiveness. In. the present study, attributional inferences refer to the subjects' expectations concerning the child's character, that is, whether his harmful act is interpreted as reflecting a chronically antisocial disposition or, on the other hand, merely a temporary bad mood. When a .serious harmful act occurs, adults may interpret this act as less characteristic of attractive than unattractive children. In, such a way, even though the transgression is serious, the positive stereotype of physically attractive children may be retained since the behavior is interpreted as more atypical of the attractive child and less likely to reoccur than when the child is unattractive. Regardless of the severity of the transgression, less punishment should be advocated for attractive than for unattractive children. If the offense is relatively minor, it should be perceived as less undesirable when the child is attractive than when he is unattractive. When the transgression is severe, attractive children should be viewed as less chronically antisocial than unattractive children. Both of these tendencies should result in adults advocating less punishment for attractive children. On the basis of these rationales, the following hypotheses were formulated: 1. An attractive child who commits a harmful act will be perceived as less likely to exhibit chronically antisocial behavior than an unattractive child, primarily when the offense is severe. Thus, adults evaluating an attractive child should perceive him as less likely to have committed a similar transgression in the past and less likely to commit one in the future than an unattractive child. 2. A transgression committed by an attractive child will be evaluated as less socially 3

This discussion should not be taken as implying that the present study tests Jones and Davis' (1965) theory of attribution.

CHILDUEN'S TRANSGRESSIONS undesirable than the same act committed by an unattractive child, particularly for mild offenses. In addition, adults should advocate less punishment for an attractive child's offense, regardless of the severity, than they do for an unattractive child committing the same transgression. To test these hypotheses, undergraduate females received written information describing either a mild or a severe transgression reputedly committed by a 7-year-old child. This sample was thought to be particularly appropriate, since women of approximately this age group generally constitute the primary socializing influence in the home and in the elementary school. Pictures of both male and female children were used to assess the generality of the predicted effects across sex of children. In addition, impersonal and interpersonal aggression were both represented to assess the generality of effects across more than one type of transgression. No predictions were made for type of transgression, and no differences were expected. The aforementioned hypotheses were tested within a 2 X 2 X 2 X 2 factorial design in which the independent variables were (a) physical attractiveness of the child (attractive versus unattractive); (b) severity of the offense (mild versus severe); (c) sex of the child (male versus female); and (d) type of offense (impersonal versus interpersonal). METHOD Subjects Two hundred mid forty-three undergraduate females at the University of Minnesota participated in this study. One hundred and eighty-three subjects were recruited from the subject pool of the psychology department and received experimental points in return for participation. Sixty subjects were obtained from introductory sociology classes and were paid $1.50 for participation.

Procedure After greeting each subject and seating her in an individual, partitioned cubicle, the experimenter gave her a large envelope containing stimulus materials appropriate to 1 of the 16 cells of the design. The subjects were randomly assigned to conditions. The first material presented to all of the subjects was an introduction to the study, which was described as one concerning person perception. The subjects were told that the present investigation

209

focused on "adults' evaluations of children's behavior, specifically the behavior of 7-year-old children." The behavioral descriptions that were to be provided for their evaluation were alleged to have been randomly selected from teachers' daily journal reports in which various types of classroom and playground disturbances were routinely described. Thus, these descriptions were said to have been formulated by those who were present when the behavior actually occurred. As a rationale for the evaluation procedure, the subjects were further told that some theorists hypothesized that observers who are actually immersed in real-life situations tend to give "richer judgments of the behavior segments observed." On the other hand, other theorists had hypothesized that those not present when the behavior segment actually took place "tend to gain a perspective that adds an extra dimension to their judgments." Accordingly, the subjects were informed that the purpose of the present study was to "determine the dimensions along which judgments in these two situations (i.e., direct observation of a behavior segment versus written information about a behavior segment) are likely to differ." All subjects were to take part, of course, in the "written information" condition. Following this introduction, the subjects were instructed to remove the behavioral description page from the envelope, to read it carefully once, and to then place it back in the envelope. They were told that this procedure simulated real-life observation of events, where the "instant replay" of behavior segments afforded by television is not possible. The behavioral description page included, in addition to the written description of the child's behavior, his name, age, and a small photo glued to one corner, ft thus presumably included all of "the information the subject would have had had she been actually present when the act occurred." The black and white photographs of second-grade boys and girls which accompanied the behavioral descriptions were selected from a larger group on the basis of judges' ratings of their physical attractiveness.1 On the basis of these ratings, each photo was categorized as depicting a child of cither high or low physical attractiveness. Four photographs of each sex at each attractiveness level were used in an attempt to ensure that any effects obtained for the attractiveness variable would not be attributable to isolated stimulus characteristics (e.g., light hair, darkeyes, etc.). All of the photographs depicted children of "normal" appearance (i.e., the unattractive category did not include any with physical defects or deformities). No child in any photo wore glasses. The behavioral description consisted of a very brief written account of a child's transgression. After pretesting with 90 undergraduate females, two types of behavior—interpersonal physical aggression * The photos were rated on a S-point scale by nine graduate student judges (five females and four males). The scores ranged from 1 (very unattractive) to 5 (very attractive).

210

KAKEN K. DION

(toward another child) and impersonal physical aggression (toward an animal)—were chosen from a larger group of behaviors.5 Both a mild and a severe version of each behavior were used. For example, in the severe-impersonal condition, the subjects were told: At one corner of the playground a dog was sleeping. Peter stood a short distance from the dog, picked up some sharp stones from the ground, and threw them at the animal. Two of the stones struck the dog and cut its leg. The animal jumped up yelping and limped away. Peter continued to throw rocks at it as it tried to move away from him. J n the mild-impersonal condition, the child was said to have walked up quietly behind a sleeping dog and stepped on its tail causing it to yelp. Jn the severe-interpersonal condition, the child was said to have packed a sharp piece of ice into a snowball and have aimed it at another child's head resulting in a deep and bleeding cut, while in the mild-interpersonal condition, the child was said to have thrown an ordinary snowball which resulted in a sling to another child's leg. A f t e r returning the stimulus materials to the envelope, the subjects completed the response questionnaire which contained a series of ungraded, 17centimeter scales with anchor words at each end. Two of these scales assessed the following atlributional inferences: (a) the likelihood that the child had committed a similar harmful act in the past (ranging from "very unlikely" to "very likely") and (ft) the probability that, he would commit a similar act in the future (ranging from "very improbable" to "very probable"). Two scales assessed the judgment of the child's offense: (a) the undesirability of the act itself (ranging from "not undesirable at all" to "extremely undesirable"), and (ft) the intensity of punishment advocated for the child (ranging from "very mild" to "very strong"). In addition, the subjects were also asked to rale the child on 16 personality trait dimensions. Six of these dimensions were chosen a priori for subsequent analysis: good-bad, aggressive unaggressive, pleasant-unpleasant, kindcruel, honest-dishonest, and nice-awful. A seventh dimension, physically attractive-physically unattractive, was included as a manipulation check. The others served as filler items. Finally, on open-ended questions, the subjects were asked to estimate in their own words (a) why the child had committed the h a r m f u l act and (ft) how the child usually behaved on a typical day. After completing this questionnaire, the cxpcri5 The original group of behaviors pretested included impersonal physical aggression, interpersonal physical aggression, verbal aggression, personal damage (to another child's property), stealing, and ridicule. To be selected for inclusion in this study, the behavior had to be perceived as intentional and unjustified (i.e., not immediately the result of another's behavior toward the child).

inenter asked all of the subjects to respond to a separate questionnaire which contained several items designed to identify those who misunderstood the instructions or who were suspicious about the true purpose of the study. Three subjects were discarded for failing to follow the experimental instructions, leaving an N of 240.

RKSULTS AND DISCUSSION Manipulation Checks As mentioned earlier, one item on the response questionnaire asked the subjects to rate the stimulus child's physical attractiveness. Analysis of variance and simple effects analyses (cf. Winer, 1962) performed on these data indicated that the manipulation of physical attractiveness was successful. Children in the attractive conditions were rated as more attractive than children in the unattractive conditions (F - 93.45, df = 1/224, p < .001 for females; F = 45.31, df ~ 1/224, p< .001 for males).11 Analyses also indicated a Sex X Attractiveness interaction (F ~ 4.29, df — 1/224, p < .05) in which unattractive females were rated as more unattractive than unattractive males (F = 19.38, df = 1/224, j f r < . 0 1 ) , 7 while no difference in perceived attractiveness was found between attractive males and females. Since no predictions were made within unattractive conditions, this finding did not affect test of the major hypotheses. Because severity of transgression was manipulated across two types of offense, it was important that the subjects perceived the severe form of both types to be more undesirable than the mild form. Analyses of responses to an item assessing the perceived undesirability of the transgression revealed that, as expected, the severe form of both types of offense was perceived as more undesirable than the mild form (F — 20.74, p < .001 for impersonal aggression; and F -62.82, p < .001 for interpersonal aggression). However, analyses also revealed a Severity X Type of Offense interaction (F = 5.62, p < .05). Mild impersonal aggression was rated more negatively than mild interpersonal aggression (F = 14.37, p<.00l). Since no 0 These F values refer to the simple effects analyses of the Attractiveness X Sex interaction. ''Unless otherwise noted, df ~ 1/224 on all subsequent results.

211

CHILDREN'S TRANSGRESSIONS predictions were made within the mild condition, this significant contrast did not affect interpretation of the results. Tests of Hypotheses Attributional inferences. ]t was hypothesized that an attractive child who committed a transgression would be less likely to be perceived as chronically antisocial than an unattractive child who committed the same act, particularly when the offense was severe. Several response measures were relevant to this differential attribution hypothesis. First, it will be recalled that subjects gave their personal estimates of how the child usually behaved on a typical day on an openended questionnaire item. These descriptions were categorized by four naive judges as being predominantly "prosocial," "mixed prosocial and antisocial," or "predominantly antisocial." 8 Two chi-square analyses of the resultant data, one within each level of harm, were performed." No differences were observed in attributional inferences between attractive and unattractive children within the mild harm-doing condition (^2 = 3.30, dj = 2, ns). Within the severe transgression condition, however, a significant difference in the predicted direction did emerge as a function of the physical attractiveness of the child. Specifically, more antisocial inferences implying a chronic behavioral disposition were attributed to unattractive than to attractive children (x2 = 6.26, df = 2, p < .05). In addition, the subjects also estimated the likelihood that the child would transgress in 8

Prosocial behavior was defined as exhibiting cooperative and positive relations with others (i.e., the child was seen as usually cheerful, friendly, helpful, pleasant, well-adjusted, etc.). In contrast, antisocial behavior was defined as the exhibition of negative reactions to others (i.e., the child was perceived as typically hostile, nasty, difficult, unfriendly, spiteful, maladjusted, etc.). 8 For the chi-square analyses, « = 82 for the mild condition and n = 90 for the severe condition. Subjects were not included in these analyses if they failed to provide sufficient information to permit classification of a child's social behavior. For example, in spite of instructions to describe the child's behavior, some subjects focused on details irrelevant to the present classificatory scheme, such as what the child's family was like, how many siblings he had, etc.

MKAN RATINGS BY CONDITION HJR I'KRCKIVKD LIKELIHOOD OK I''UTURE TRANSGRESSIONS Condition

Unattractive child Attractive child

Mild tiansgression

Severe transgression

12.19 11.73

13.12 10.70

Mole.— The above scale rating range from 0 (very unlikely) to 17 (very likely).

the future as well as the likelihood that he had done so in the past. Analysis of variance conducted on estimates of the probability of future offenses revealed the predicted interaction between attractiveness and severity of offense (F = 4.33, p < .05). Table 1 presents the means relevant to this interaction. An analysis of the simple effects revealed only one significant comparison. As predicted, within the severe condition, attractive children, were seen as less likely than unattractive children to commit a similar transgression in the future (F = 13.78, p < .001). An analysis of variance on estimates of the child's past offenses revealed a slight and nonsignificant main effect trend for attractiveness. Subjects tended to see attractive children as less likely to have similarly transgressed in the past whether the offense was mild or severe (F = 2.01, p- .15). One may ask why attributional inferences occurred on the subjects' estimates of the child's future behavior but not on their estimates of his past actions. It is possible that subjects may be more reluctant to make strong inferences about events that reputedly have already occurred, since the accuracy of their inferences could presumably be easily verified. They might wish to be seen as reasonably accurate assessors of the child's character. Estimates of future behavior provide a more ambiguous frame of reference which cannot as immediately be subjected to empirical verification. On the whole, however, the above results support the hypothesis that adults' evaluations of a child who commits a serious transgression differ as a function of the child's physical attractiveness. Specifically, adult evaluators are less likely to attribute a chronic, antisocial behavioral disposition to

212

KAREN K. DION

attractive than to unattractive children. This result occurred both on the subjects' free descriptions of the child's typical behavior and on their estimates of the likelihood of future offenses. Trait ratings. Additional evidence concerning adults' inferences about the character of attractive and unattractive children was provided by responses to the six personality scales, described earlier, which tapped dimensions potentially relevant to evaluations of a child who commits a transgression. Analyses of these data revealed main effects for attractiveness on two dimensions—honest—dishonest and pleasant-unpleasant. Unattractive children who transgressed were perceived as being more dishonest (F — 9.70, p < .01) and more unpleasant (F = 4.28, p < .OS) than attractive children. The effect for honesty is particularly interesting. Given the ambiguous circumstances of many transgressions, an unattractive child may operate at some disadvantage in the "assignment of blame" process. If the unattractive child is seen as more likely to lie than an attractive child, establishment of his innocence may not be as easy. These two main effects for attractiveness are in accord with a general stereotype that physically attractive people, adults or children, are assumed to possess more socially desirable personalities than less attractive persons. It is possible that the results of these dimensions might have occurred even if children differing in attractiveness engaged in more neutral (i.e., not harmful) behavior. On the other hand, it is also conceivable that certain types of inferences such as those about general honesty arc elicited most strongly in transgression situations. On a third personality dimension, goodbad, a three-way interaction, Severity X Attractiveness X Sex, was observed (F =5.82, p < .05). Simple effects analyses revealed significant contrasts within the severe condition. Attractive males were rated more negatively for a severe offense than either unattractive males (F = 10.29, p < .01) or attractive females (F — 4.34, p < .05). Also, unattractive females who committed a severe offense received a more negative evaluation than did unattractive males (F — 4.49, p < .05). These results suggest the possibility of

different evaluative standards for attractive males and female children who unambiguously violate adult expectations by committing a severe transgression. This interpretation is speculative, however, since these results occurred on only one personality item. None of the other personality items yielded Severity X Attractiveness X Sex interactions, nor were there any Sex X Attractiveness interactions.10 Judgment dimensions. It was also hypothesized that the subjects would evaluate an attractive child's transgression less negatively than that of an unattractive child, particularly for a mild offense. In addition, the subjects were expected to advocate less intense punishment for attractive than for unattractive children, regardless of the severity of the transgression. An analysis of variance performed on the subject's ratings of the undesirability of the child's transgression indicated an attractiveness main effect (F — 3.60, p = .06). A transgression, whether mild or severe, when committed by an attractive child, tended to be perceived as less undesirable than the same act committed by an unattractive child. No support was found for the hypothesis that the intensity of punishment advocated for a given transgression would vary as a function of physical attractiveness (F = 1.98, m).11 Conceivably, parameters of punishment other than intensity (e.g., type of punishment) may be influenced by the child's appearance. Also, in the present study, the subjects may have assumed that the term "intensity of punishment" implied physical punishment. Differences might have been ob10 On the aggressiveness dimension, a four-way interaction occurred. Simple effects analysis revealed an ambiguous pattern of results that could not be definitively interpreted either for or against the attractiveness hypothesis. No significant differences were found as a function of attractiveness on the cruel-kind or nice-awful dimensions. 11 Though not of primary interest for the present study, a Sex X Severity interaction occurred on both evaluation of the offense (F = 7.49, p < .01) and on intensity of punishment advocated (F — 6.48, p < .05). Mild transgression by females was perceived as more undesirable than mild transgression by males (F = 6.48, p<.05). The subjects advocated less intense punishment for females than for males who committed a severe offense (/ i '~-3.96, / > < . 0 5 ) .

CHILDKKN'S TRANSGRESSIONS tained if items concerning intensity had been paired with various disciplinary methods, for example, how strongly should the child be scolded, etc. In any event, it would seem unusual if adults' inferences concerning the behavioral dispositions of attractive versus unattractive children who commit a transgression do not affect the process of punishment administration on at least some dimensions. CONCLUSIONS AND IMPLICATIONS The present study finds that the physical attractiveness of a child who commits a transgression does indeed influence adults' evaluations of him. Attributional inferences are affected both by the severity of a transgression and the attractiveness of the child who commits the offense. Children's physical attractiveness also tends to influence judgments of the transgression's undesirability, whether the offense is mild or severe. Given the numerous occasions when adults must deal with children's transgressions, the implications of these results merit further investigation. It seems plausible that adults' assumptions about a child's character, partic-

ularly the type of behavior expected from him, may ultimately be communicated to the child, potentially influencing his self-evaluation. If so, one might find differences between attractive and unattractive children in areas such as reactions to their own transgressions as well as the type and frequency of transgression committed. These issues might be particularly interesting for future research. REFERENCES DION, K., & BEKSCUEIU, E. Physical attractiveness and sociometric choice in young children. University of Minnesota, 1971. (Mimeo) DION, K., BERSOIIEH), E., & WALSTER, E. What is beautiful is good. Journal of Personality and Social Psychology, 1972, in press. HKIDEK, F. The psychology of interpersonal relations. New York: Wiley, 1958. JONES, E. E., & DAVIS, K. E. From acts to dispositions: The attribution process in person perception. In L. Bcrkowitz (Ed.), Advances in experimental social psychology. Vol. 2. New York: Academic Press, 1965. SHAW, M. E., & SUEZER, J. L. An empirical test of Heider's levels in attribution of responsibility, Journal of Abnormal and Social Psychology, 1964, 69, 39-46. WINER, B. J. Statistical principles in experimental design. New York: McGraw-Hill, 1962. (Received July 14, 1971)