《社会心理学》课程教学资源（文献资料）Aggression——Influences of Gender Identity on Children’s Maltreatment of Gender-Nonconforming Peers - A Person X Target Analysis of Aggression.pdf

Influences of Gender Identity on Children’s Maltreatment of Gender-Nonconforming Peers: A Person Target Analysis of Aggression Rachel E. Pauletti, Patrick J. Cooper, and David G. Perry Florida Atlantic University We investigated whether gender identity influences preadolescents’ tendency to single out genderatypical peers for abuse. Data were gathered from 195 boys and girls (M age 10.1 years) in the fall and spring of a school year. Children self-reported multiple dimensions of gender identity (intergroup bias, felt pressure for gender differentiation, felt gender typicality, gender contentedness); peers assessed each other’s social behavior (gender nonconformity, aggression toward each classmate). Using multilevel modeling, we examined how children’s attacks on gender-nonconforming peers (relative to their attacks on other peers) changed over the school year depending on their gender identity. There was modest support for the hypothesis that overconfident, arrogant gender identity promotes abuse of gender-atypical peers but considerable support for the hypothesis that insecure, self-questioning gender identity fosters this tendency. Implications for issues central to contemporary personality theory (e.g., Person Situation interaction) are discussed. New and somewhat surprising information about the cognitive and behavioral characteristics of gender-nonconforming preadolescents is provided. Keywords: aggression, gender identity, gender nonconformity, Person Situation interaction, victimization Children, adolescents, and adults who exhibit salient crossgender characteristics are at heightened risk for rejection, harassment, discrimination, and abuse. As early as the preschool years, children who behave in gender-atypical ways elicit dislike, teasing, and even physical aggression from (some of) their peers (Fagot, 1977; Langlois & Downs, 1980). Negative judgments and harsh treatment of gender-atypical peers continue across middle childhood and into adolescence; boys are more likely than girls both to treat their gender-atypical peers badly and to be harassed for having gender-atypical qualities (Berndt & Heller, 1986; Kite & Whitley, 1996; Lobel, 1994; Moller, Hymel, & Rubin, 1992; Poteat, Espelage, & Green, 2007). Moreover, youths who are victimized for cross-gender behavior are at heightened risk for additional negative outcomes, including depression, substance abuse, truancy, social isolation, hopelessness, and violence (Kimmel & Mahler, 2003; Lampinen et al., 2008; Mishna, Newman, Daley, & Solomon, 2009; Toomey, Ryan, Diaz, Card, & Russell, 2010; Walls, Kane, & Wisneski, 2010). Understanding the roots of aggression toward gender-atypical persons is an important challenge. Harmful attitudes and actions toward gender-atypical persons are perpetrated by a minority of individuals, raising questions about the unique psychological factors that motivate them. Our knowledge of the motives underlying negative treatment of individuals with cross-gender characteristics comes mainly from studies of adolescents and adults who dislike, harass, or assault persons whom they presume to be gay or lesbian. An implicit (but untested) assumption of this research is that negative treatment of gender-atypical individuals derives from more than simple dispositional or trait aggression (i.e., the tendency to be more aggressive than other people across diverse situations, including different target types) but rather is motivated, at least in part, by cognitive factors (e.g., values, beliefs, expectancies, goals, identities) specific to gender. In other words, it is believed that perpetrators do not treat gender-atypical persons harshly simply because such targets constitute a convenient victim group (e.g., likely to hand over resources, unlikely to be defended by peers) but rather because the perpetrators’ gendered cognitions are stimulated by the targets’ gender-atypical behaviors in ways that arouse disgust, threat, anger, and aggression specifically toward these targets. This literature has identified several cognitive predictors of harsh treatment of gender-atypical others, though many studies are characterized by methodological weaknesses (e.g., concurrent research designs, exclusive use of self-reports, failure to test whether the cognitive structures predict harsher treatment of genderatypical persons than of other target types, lack of control for trait aggression). This work, which is mainly with adolescent and adult males, has highlighted two cognitive patterns associated with aggression toward gender-nonconforming others. The first pattern reflects an overconfident gender identity rooted in strong gender stereotypes and attitudes. That is, much harsh treatment of genderatypical males is perpetrated by males who polarize the sexes (believe that males and females should act very differently), who espouse hypermasculinity (believe that males should inhibit tender emotions, take dangerous risks, dominate others, etc.), and who appraise themselves as hypermasculine (Herek, 1988, 2000; Parrott, 2009; Parrott, Adams, & Zeichner, 2002; Pryor & Whalen, 1997; Stark, 1991; Thompson, Grisanti, & Pleck, 1985; Whitley, 2001). Also, women who endorse the traditional female role someRachel E. Pauletti, Patrick J. Cooper, and David G. Perry, Department of Psychology, Florida Atlantic University. Correspondence concerning this article should be addressed to David G. Perry, Department of Psychology, Florida Atlantic University, Boca Raton, FL, 33431. E-mail: perrydg@fau.edu This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Journal of Personality and Social Psychology © 2014 American Psychological Association 2014, Vol. 106, No. 5, 843– 866 0022-3514/14/$12.00 DOI: 10.1037/a0036037 843

times dislike and are angered by gender-atypical females (Parrott & Gallagher, 2008). Presumably, people with strong genderdifferentiating beliefs and identities harass gender-atypical others because they perceive them as threats to their worldviews, as morally disgusting, as challenges to the positive distinctiveness of their gender collective, as disloyal and untrustworthy members of the collective, or as expecting other people to reciprocate their gender-atypical behavior. A second cognitive pattern associated with maltreatment of gender-atypical persons reflects insecure gender identity— uncertainty and anxiety over one’s adequacy as a member of one’s gender collective. Insecurity over one’s gender fit is especially conducive to attacks on gender-atypical others for individuals who also possess strong gender-differentiating stereotypes and attitudes (in this regard, they resemble persons with overconfident gender identity). That is, males who subscribe to rigid standards of hypermasculinity but perceive themselves as failing to live up to the standards tend to treat gender-atypical others harshly (Bernat, Calhoun, Adams, & Zeichner, 2001; Eisler & Skidmore, 1987; Franchina, Eisler, & Moore, 2001; Kimmel, 1994; Parrott, 2009; Vandello, Bosson, Cohen, Burnaford, &Weaver, 2008). Such males presumably experience a disturbing ought self-discrepancy (a gap between what one feels one should be and what one really is). Ought self-discrepancies are conducive to frustration, agitated distress, guilt, shame, self-rejection, anger, and loss of selfregulatory control (C. S. Carver, Lawrence, & Scheier, 1999; Higgins, 1987; Swann & Bosson, 2008; Tangney, Wagner, Fletcher, & Gramzow, 1992). Males with a strongly gendered ought self-discrepancy are said to experience masculine gender role stress (Levant, 2011). Although they strive to be male typical, they either lack self-efficacy for self-prescribed male-typical behaviors (e.g., dominance, risk taking) or fear that in certain situations they will be unable to inhibit feminine behavior. Presumably, they perceive in gender-atypical others what they fear in themselves, and this perception activates latent feelings of insecurity (fears of gender inadequacy, of being perceived as gay, of rejection by same-sex others) that motivate them to lash out defensively at gender-atypical persons, thereby removing the threat and demonstrating that the gender inadequacy resides in other people and not themselves. Note that a common element of both the overconfident and insecure cognitive patterns is investment in gender-differentiating stereotypes and attitudes. When people adopt such standards and successfully emulate them, the result is an overconfident gender identity. When people hold the standards but fail to fulfill them, the result is an insecure gender identity. Perhaps either pattern disposes persons to harass gender-nonconforming others. The Present Study To our knowledge, no study has investigated the factors that lead preadolescent children to single out gender-nonconforming peers for abuse. The principal purpose of the present study was to investigate the role of gender identity in preadolescents’ maltreatment of gender-atypical peers. Gender identity is a multidimensional construct encompassing a person’s felt compatibility with, quality of motivation to fit in with, and evaluation of a gender collective (Ashmore, Deaux, & McLaughlin-Volpe, 2004; Egan & Perry, 2001; Tobin et al., 2010). Using a model of gender identity proposed by Egan and Perry (2001), we examined how four dimensions of gender identity might, individually and in combination with other variables, affect children’s tendencies to abuse gender-atypical peers. These dimensions of gender identity were intergroup bias (perceiving one’s own gender more favorably than the other), felt pressure for gender differentiation (internalized pressure to avoid other-sex behaviors), felt gender typicality (perceived similarity to same-sex others), and gender contentedness (satisfaction with one’s gender). The first two dimensions—intergroup bias and felt pressure for gender differentiation—may be considered between-gender facets of gender identity in that they depend on perceiving, and likely exaggerating, differences between the sexes. Strong forms of these facets of gender identity may well contribute to the desire to interact with, fit in with, and emulate others of one’s gender, but they probably also encourage avoidance and disparagement of other-gender activities and persons. Felt pressure for gender differentiation is in fact assessed as the expectation of punishment (e.g., ridicule, alienation) for cross-gender behavior (Egan & Perry, 2001). Intergroup bias is part of a set of intergroup cognitions that also includes exaggeration of differences between the sexes, greater use of stereotypes, and homogenization of the other sex (Powlishta, 1995; Tajfel & Turner, 1979). By preadolescence, these gender-polarizing forms of identity are relatively immature and are associated with peer-relational difficulties, low selfesteem, internalizing problems (for girls), and externalizing problems (for boys; P. R. Carver, Egan, & Perry, 2004; P. R. Carver, Yunger, & Perry, 2003; Corby, Hodges, & Perry, 2007; Egan & Perry, 2001; Smith & Leaper, 2006; Yunger, Carver, & Perry, 2004). The other two dimensions of gender identity—felt gender typicality and gender contentedness—may be considered withingender facets of gender identity because they capture children’s feelings of compatibility with, and acceptance by, their own gender collective. They are largely a function of children’s selfperception of salient and valued same-gender attributes (e.g., same-sex friendships) and are not necessarily undermined by selfperception of a few other-gender attributes (Egan & Perry, 2001; Spence, 1993). These dimensions of gender identity are usually positively correlated with self-esteem, social skills (e.g., prosocial behavior), and peer acceptance (P. R. Carver et al., 2003, 2004; Corby et al., 2007; Egan & Perry, 2001; Smith & Leaper, 2006; Yunger et al., 2004). Because the four dimensions of gender identity typically show low correlations with one another (e.g., Egan & Perry, 2001; Yunger et al., 2004), they cannot be considered interchangeable reflections of a single, common underlying construct (e.g., gender schematicity, or a generalized tendency to view self and others through a pernicious gender lens; Bem, 1981). Moreover, as noted, the between-gender and the within-gender forms of gender identity usually relate to adjustment in opposite ways. Using measures of the four gender identity dimensions (and of other potentially relevant aspects of self-concept, described below), we examined whether the cognitive patterns found in previous research to be associated with maltreatment of gendernonconforming people by older persons (overconfident and insecure gender identity) also motivate preadolescents to attack gender-nonconforming peers. There are grounds for expecting both patterns to apply to preadolescents. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 844 PAULETTI, COOPER, AND PERRY

Children with overconfident gender identity might aggress toward gender-atypical peers for many of the same reasons that older persons do (e.g., moral outrage, distrust, threat to ingroup cohesiveness). Overconfident gender identity is perhaps most clearly represented when a strong form of between-gender identity (strong intergroup bias or felt pressure for gender differentiation) combines with strong felt compatibility with one’s gender (strong felt gender typicality or gender contentedness). Thus, we examined whether each form of between-gender identity interacts synergistically with each form of within-gender identity to spur children to attack gender-nonconforming peers. However, we also examined whether each between-gender form of gender identity combines with each of four gender-nonspecific self-appraisals to stimulate attacks on gender-nonconforming peers. These other selfappraisals were global self-esteem and perceptions of self-efficacy for dominance, popularity, and physical attractiveness. High selfesteem motivates people to act confidently on their self-perceived strengths, goals, identities, and values (Baumeister, 1998). When high self-esteem is paired with self-serving ideation (e.g., narcissism), it sometimes fosters aggression, for both adults and children (Baumeister, Smart, & Boden, 1996; Morf & Rhodewalt, 2001; Thomaes & Bushman, 2011; Thomaes, Bushman, Stegge, & Olthof, 2008). Thus, high self-esteem might encourage children with strong ingroup favoritism or felt pressure for gender differentiation to attack gender-nonconforming others. High selfefficacy for dominance, popularity, or attractiveness might as well. High self-efficacy in these domains reflects confidence in one’s social status and in one’s ability to satisfy demonstration goals, or desires to be admired by peers; children with such goals tend to display adjustment difficulties, including aggression (O’Moore & Kirkham, 2001; Rodkin, Ryan, Jamison, & Wilson, 2013; Rudolph, Abaied, Flynn, Sugimura, & Agoston, 2011; Salmivalli, 1998). It is conceivable that high self-efficacy in these domains empowers children with strong gender-polarizing identity to strive for admiration and status by discriminating against gendernonconforming peers. Overconfident gender identity may not always require strong between-gender identity, however. Strong within-gender identity— high felt gender typicality or gender contentedness—might also sometimes take on shades of overconfidence or narcissistic arrogance and promote attacks on gender-nonconforming peers. Perhaps this occurs when strong within-gender identity is accompanied by high selfesteem or self-efficacy for dominance, popularity, or attractiveness. Thus, we also examined whether each form of within-gender identity interacts with self-esteem or a measure of self-efficacy to promote harassment of gender-atypical peers. Children with insecure gender identity might also discriminate against gender-nonconforming others for many of the same reasons that older persons do (e.g., defensive lashing out at those who arouse their inner conflict, attempting to show that the deficiency lies in others and not themselves). Insecure gender identity is perhaps most clearly represented when a strong form of betweengender identity (intergroup bias or felt pressure for gender differentiation) combines with strong felt incompatibility with one’s gender (low felt gender typicality or low gender contentedness). Such pairings constitute clear-cut gender self-discrepancies. However, children with strong between-gender identity might also feel self-discrepant and anxious if they have low self-esteem or feel incapable of displaying dominance, popularity, or good looks. These self-appraisals do not necessarily reflect felt incompatibility with one’s gender collective, but at this age, children tend to judge their self-worth and competencies by comparing themselves to same-gender others (Harter, 2006), and self-appraisals on these dimensions are therefore likely to reflect how well children feel they are stacking up against same-sex others. Moreover, children with strong between-gender identity might focus on satisfying demonstration goals (e.g., feeling superior, being admired, avoiding ridicule), and if they view themselves as valueless or as incapable of fulfilling these goals, their frustration and insecurity may be acute. Although research on self-discrepancies in children is scant, children who ardently strive for status and admiration in the peer group but perceive themselves as falling short do tend to be aggressive (Barry, Frick, & Killian, 2003; Harter & McCarley, 2004; Pauletti, Menon, Menon, Tobin, & Perry, 2012); it is plausible that between-gender identity channels the aggression of these frustrated children toward gender-atypical targets. Thus, the insecure pathway to aggression toward gender-nonconforming peers would be supported if for children with a strong form of betweengender identity, lower felt gender typicality, gender contentedness, self-esteem, or self-efficacy for dominance, popularity, or attractiveness were to motivate attacks on gender-nonconforming peers. However, like overconfident gender identity, insecure gender identity may not always require strong between-gender identity. Children who feel incompatible with their gender (who feel gender atypical or discontent) probably experience some uncertainty over their adequacy as members of their gender collective even if they lack strong gender-differentiating stereotypes and attitudes; moreover, their anxiety may be exacerbated if they have low selfesteem or feel incapable of displaying attributes needed for status in their peer group. Thus, low self-esteem or low self-efficacy for dominance, popularity, or attractiveness might potentiate a tendency on the part of children who feel gender incompatible to demonstrate their gender adequacy by hassling gendernonconforming peers. The overconfident and insecure pathways may well be alternate routes to preadolescents’ harassment of gender-nonconforming peers: Some children might harass gender-atypical peers because of overconfident gender identity, others because of insecure gender identity. We allowed for this possibility in our analytic strategy by probing for a quadratic effect of one cognitive variable (e.g., self-esteem) at different levels of another cognitive variable (conceptualized as the moderator; e.g., intergroup bias). For example, it might be that for children with high intergroup bias (but not for children with low intergroup bias), maltreatment of genderatypical peers is greater for children with either high or low self-esteem than for children with intermediate self-esteem. Such a pattern would constitute evidence for both pathways. The Present Study in Relation to Contemporary Personality Theory Our approach to conceptualizing and evaluating cognitive pathways to harassment of gender-atypical others is grounded in contemporary cognitive metatheories of personality (Cervone, 2004; Mischel & Shoda, 1995, 2008; Read et al., 2010; Zakriski, Wright, & Underwood, 2005). These theories offer conceptual strategies for generating hypotheses linking people’s stable cognitive processing structures—their values, expectancies, goals, beliefs, idenThis document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. AGGRESSION TOWARD GENDER-NONCONFORMING PEERS 845

tities, and the like—to their social behavior. Because our research questions and paradigm were stimulated by these theories and because the theories depart in critical ways from other approaches to investigating human aggression, we describe certain features of the theories and tell how they guided the present study. Cognitive theories emphasize that many persons display considerable situational specificity in a social behavior (e.g., aggression) owing to interactions between their relatively enduring processing structures and features of the situations they encounter. That is, people’s unique cognitive structures cause them to be sensitive to, and to react in particular ways to, aspects of situations that are psychologically relevant to them. These theories do not deny that aggression also possesses the qualities of a personality trait (i.e., that some people are generally more aggressive than others), but they suggest that much can be learned about aggressive motivation by studying how individuals organize their aggression around particular eliciting circumstances. Although data are still accumulating, it is clear that many children do possess unique aggression signatures (Mischel & Shoda, 1995)— distinctive and stable personal profiles indicating how their aggression varies over different situations (relative to other children’s aggression in the same situations). To determine children’s signatures, children are observed for aggression in several situations. For each situation, each child’s aggression is converted to a z score indicating the child’s aggression relative to other children, thus removing the nomothetic influence of the situation. A child’s idiographic profile of z scores across the situations is the child’s aggression signature. The child’s average z score is a measure of trait aggression (Fournier, Moskowitz, & Zuroff, 2008; Mischel & Shoda, 1995). Many children differ markedly across situations in their aggression, and this variability cannot always be accounted for by trait aggression. In other words, after the normative influences of situations and trait aggression are removed, considerable intraindividual differences across situations (signatures) often remain to be explained. There has been some effort to identify the salient features of situations that organize within-child variability in children’s aggression. Chief among these situational features are the characteristics and behaviors of a child’s interaction partners (e.g., peer vs. adult, male peer vs. female peer, liked peer vs. disliked peer; Coie et al., 1999; Hodges, Peets, & Salmivalli, 2009; Matthys, Maassen, Cuperus, & Van Engeland, 2001; Mischel & Shoda, 1995; Peets, Hodges, & Salmivalli, 2011; Rodkin & Berger, 2008; Veenstra et al., 2007; Zakriski et al., 2005). These data support Fournier et al.’s (2008, p. 533) observation that “the most salient psychological features of the situation are found in the behavior of the individual with whom one is interacting.” Although there has been progress in identifying some features of situations (e.g., target characteristics) that underpin within-child situational variability in aggression, a central tenet of cognitive theories—the idea that individual differences in within-child situational variability in aggression are predictable from enduring cognitive processing structures— has, to our knowledge, yet to be investigated. The present study illustrates a paradigm for researching this issue, by examining whether children’s cognitive structures predict their tendency to aggress more toward genderatypical peers than toward other peers. Thus, we conceptualize the degree to which a child’s interaction partner is characteristically gender atypical as a situational variable that interacts with children’s processing structures to affect aggression. Our paradigm therefore is one of Person Target interaction. To examine cognitive pathways to children’s harassment of gender-atypical peers, we used multilevel modeling. In both the fall and spring of a school year, we collected measures of children’s cognitions, of their aggression toward each classmate, and of their gender-atypical behavior. For each child, a beta coefficient (slope) was calculated indicating how the child’s aggression toward his or her classmates changed over the school year as a function of the classmates’ gender-atypical behavior in the fall (within-subjects analyses). To test the hypotheses, these within-child slopes were predicted from measures of participants’ cognitions (between-subjects analyses). The longitudinal design was used to help rule out certain alternative directions of influence. For example, it is conceivable that children’s cognitions are reactive to their aggression toward gender-atypical others (e.g., boys who attack boys for playing with girls may afterward feel more gender typical or more dominant). It is important to note that our hypotheses address only the degree to which the cognitive variables affect children’s tendencies to discriminate against gender-atypical peers, that is, to direct more aggression toward gender-atypical peers than toward gender-typical peers. Our hypotheses do not address the absolute severity of children’s aggression toward their victims. Two children may be similarly inclined to victimize gender-atypical peers more than gender-typical ones (i.e., have similar positive within-child betas), but one child may aggress generally (across all peer targets) at a higher level than the other and therefore show a more generally elevated pattern (i.e., a higher intercept). Sensitivity to a target feature (captured by slope) and average aggression across targets (i.e., trait or general aggression, captured by intercept) are distinct constructs, and our hypotheses address the impact of social identity only on the former. Characteristics of Gender-Nonconforming Children Although the focus of this investigation was on the cognitive determinants of children’s aggression toward gender-atypical peers, the measures we collected afforded an opportunity to learn more about the children whose victimization was the focus of study— gender-nonconforming preadolescents. We describe the self-concepts and social-behavioral qualities of these children. We examine whether gender nonconformity predicts, or is predicted by, each other study variable over time. Although this aspect of our study was largely exploratory rather than hypothesis driven, we examined whether gender-nonconforming children are at greater risk for adjustment difficulties (e.g., peer victimization, internalizing problems, low selfesteem) if they possess a problematic form of gender identity (high between-gender identity, low within-gender identity) than if they do not. For example, gender-nonconforming children who feel strong pressure for gender differentiation or feel disconnected from their own gender may be particularly distressed. Method Participants and Procedure All children in the fourth through seventh grades of a university laboratory school were invited to participate. Participants were 101 boys (M age 10.2 years) and 94 girls (M age 10.0 years) who received written informed parental consent; they represented 75% This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 846 PAULETTI, COOPER, AND PERRY

of the children in their grades. The numbers (and mean ages in the fall) of children in Grades 4 through 7, respectively, were 52 (8.7 years), 56 (9.7 years), 43 (10.7 years), and 44 (11.7 years). Four additional children were tested in the fall but did not participate in the spring; participation in the spring was limited to children who had participated in the fall, and thus spring participants represented 98% of fall participants. The sample was approximately 51% White, 21 % Black, 20% Hispanic, and 8% other. The admissions procedures of the school ensure that the student body is representative of the entire population of Florida in terms of ethnicity/race and socioeconomic level. Children were tested individually in a library cubicle in both the fall and spring of the school year. At the start of each session, children signed an assent form describing the questionnaires and stating that they were free to leave blank any question they preferred not to answer. At the conclusion, children were asked not to discuss their responses or the study with other children. Assessment of Cognitive Variables Gender identity. Four dimensions of gender identity were assessed. Intergroup bias was calculated from children’s ratings of how much they liked each of their participating classmates (“How much do you like each kid?”). Ratings were on a 4-point scale with 1 indicating Not at all and 4 indicating A lot. A child’s rating of a peer was converted to a z score indicating the child’s liking of the peer relative to all children’s liking of that peer. The average other-sex z was subtracted from the average same-sex z to index intergroup bias, or ingroup favoritism. The other three facets of gender identity were assessed with modified versions of Egan and Perry’s (2001) scales; the scales were shortened to retain items with high item-total correlations in previous studies. Felt pressure for gender differentiation was assessed with seven items. Children rated statements on 4-point scales with 1 indicating Not at all true for me and 4, Very true for me. Items described possible reactions for gender-atypical behavior from parents (two items; e.g., “My parents would be upset if they saw me acting like a girl”), same-sex peers (two items; e.g., “The boys I know would be upset if I wanted to learn an activity that girls usually do”), and the self (three items; e.g., “I wouldn’t like myself if I heard myself talking or laughing like a girl”). Scale totals were averages of the seven items. Cronbach’s alpha was .81. Items on the felt gender typicality and gender contentedness scales were formatted alike. Each had five items, with about half reverse-scored. For each item, children read descriptions of two kinds of children— ones depicted as high on the gender identity dimension and ones depicted as low—and chose which kind of children they resembled more; they then indicated whether this choice was Very true or Sort of true for them. This yielded a 4-point scale, with higher scores indicating greater gender identity. Scale scores were item averages. Items assessing gender typicality included (from the girls’ form) “Some girls feel they are different from other girls BUT Other girls feel they are similar to other girls” and “Some girls have the same feelings that other girls have BUT Other girls don’t have the same feelings that other girls have.” Gender contentedness items included “Some girls are happy that they were born a girl BUT Other girls are not happy that they were born a girl” and “Some girls wish they didn’t have to be a girl all their life BUT Other girls are glad they’ll be a girl all their life.” Cronbach’s alpha was .68 for gender typicality and .59 for gender contentedness. The marginal reliability of the gender contentedness scale indicates that results with this measure should be interpreted cautiously. Gender-nonspecific self-appraisals. Self-esteem was assessed with Harter’s (1985) six-item global self-worth scale; each item is scored from 1 to 4 (averaged for the scale total); Cronbach’s alpha was .73. Self-efficacy for dominance, popularity, and attractiveness was measured by having children rate how hard versus easy it was for them to enact each of 18 behaviors. Children rated each behavior on a 4-point scale, with 1 indicating HARD! and 4 indicating EASY! Dominance self-efficacy was assessed with eight items (e.g., “Bossing others around is ____ for me,” “Making others feel like I am in charge is ____ for me”). Popularity self-efficacy was assessed with six items (e.g., “Being popular is ____ for me,” “Being cool is ____ for me”). Attractiveness selfefficacy was assessed with four items (e.g., “Looking good in a bathing suit is ____ for me,” “Being good looking is ____ for me”). Cronbach’s alphas for these scales, respectively, were .79, .84, and .87. Assessment of Social Behavior The study required assessments of children’s gender-atypical behavior as well as of their aggression toward each classmate. These measures came from separate peer nomination instruments, described in turn. Only the names of children with parental permission to participate were included on these instruments. Assessment of gender nonconformity (and other social behaviors) in the peer group. Fourth and fifth graders provided peer nomination data for all of their classmates. Because sixth and seventh graders did not have a homeroom class but knew all the children in their grade very well, each child provided nominations for 14 –15 grade-mates (a randomization procedure ensured that each child could nominate, and be nominated by, a similar number of peers). We use the term classmates to refer both to fourth and fifth graders’ classmates and to sixth and seventh graders’ grademates. The mean number of peers serving as nominators for a child was 16.1 (minimum 14). A 23-item peer nomination inventory was administered. Items covered a broad array of social behaviors in the peer group. Most items came from previous studies (e.g., Pauletti et al., 2012), but four that assessed gender-atypical behavior were written for the present study (from the boys’ form: “He’d rather play with girls than with boys,” “He likes to do things that girls usually do,” “He acts like a girl,” and “He always plays with girls”). Children checked off the names of classmates who fit each item (unlimited nominations); children crossed their own name off the list so they did not nominate themselves. A child’s score on each item was the proportion of classmates (of either sex) who nominated the child for the item. A principal components analysis (with varimax rotation) on the data of all children in the fall yielded five components, one of which had very high loadings for all four gender-atypicality items (.83 to .92) and very low loadings for the remaining 19 items (.16 to .23); it was labeled gender nonconformity. The other components were labeled social prominence (e.g., popularity, sports competence), prosocial qualities (e.g., helpful behavior, physical attractiveness), coercion (e.g., dominance, manipulation), and internalizing problems (e.g., sadness, anxiety). A similar analThis document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. AGGRESSION TOWARD GENDER-NONCONFORMING PEERS 847