Relations Among Gender Typicality, Peer Relations, and Mental Health During Early Adolescence

The current study examines whether being high in gender typicality is associated with popularity, whether being low in gender typicality is associated with rejection/teasing, and whether teasing due to low gender typicality mediates the association with negative mental health. Middle school children (34 boys and 50 girls) described hypothetical popular and rejected/teased peers, and completed self-report measures about their own gender typicality, experiences with gender-based teasing, depressive symptoms, anxiety, self-esteem, and body image. Participants also completed measures about their peers’ gender typicality, popularity, and likeability. Results indicated that popular youth were described as more gender typical than rejected/teased youth. Further, being typical for one’s gender significantly predicted being rated as popular by peers, and this relationship was moderated by gender. Finally, low gender typicality predicted more negative mental health outcomes for boys. These relationships were, at times, mediated by experiences with gender-based teasing, suggesting that negative mental health outcomes may be a result of the social repercussions of being low in gender typicality rather than a direct result of low typicality.


Introduction
Recent media coverage of teen suicides led to a widespread public discourse about the impact of teasing on adolescents, particularly when that teasing was related to teens 'not fitting in' as a typical boy or typical girl. In addition, media coverage has highlighted the ways that peers harass and ostracize boys and girls who are deemed atypical for their gender (see Russell, Kosciw, Horn, & Saewyc, 2010). Little empirical research, however, has examined how being atypical for one's gender is related to peer harassment. Thus, the goal of the current research is to examine how peer relationsboth positive peer relationships (being popular or liked) and negative peer relationships (being teased or harassed)-are associated with being either gender typical or gender atypical. disliked by many peers, possess positive and negative traits, and are highly visible and hold social power. Well-liked early adolescents, by contrast, are largely prosocial and have many high-quality friendships (see Mayeux, Houser, & Dyches, 2011).
Research at the intersection of gender, popularity, and likeability suggests that the degree to which boys and girls are gender typical may influence their positive peer relations (Rose, Glick, & Smith, 2011). Both experimental studies (Lobel, Bempechat, Gewirtz, Shoken-Topaz, & Bashe, 1993) and peer report studies (Egan & Perry, 2001) have shown that more typical children are more liked.
Indirect evidence suggests that gender typical children are also likely to be popular. For example, popular boys usually possess masculine (i.e., gender typical) characteristics, such as being athletic and tough, whereas popular girls usually possess feminine characteristics, such as being attractive, wearing expensive designer clothes, and gossiping (Adler, Kless, & Adler, 1992;Closson, 2009;Lease et al., 2002;Vaillancourt & Hymel, 2006). In addition, Kessels (2005) found that boys rated gender atypical hypothetical peers (e.g., girls who excelled in physics) as less popular than gender typical peers.
The current study extends previous findings by directly examining whether gender typicality is associated with popularity, over and above likeability. This is important for two reasons. First, children in early adolescence have been shown to highly value popularity, even more so than friendships (LaFontana & Cillessen, 2010;Merten, 2004). Thus, they may change and adapt their behaviors (in this case, their gendered behaviors) to gain or maintain popularity. Second, popularity is associated with social influence and control (Eder, 1985;Sandstrom, 2011). Thus, if popular teens are those with the most social influence, and they are the most gender typical, then the group norm of gender typicality is particularly likely to be maintained and enforced, regardless of whether they are liked or not.

Gender Typicality and Negative Peer Relations
Like positive peer relations, negative peer relations are also critically important for early adolescents (Hawker & Boulton, 2000). Whereas children who are high in gender typicality are more liked by their peers (Egan & Perry, 2001), children who are low in gender typicality are more likely to be teased or harassed by their peers (Young & Sweeting, 2004). In general, research on group norms indicates that children who do not meet the norms or fit the stereotypes of the group can be bullied, mocked, or ostracized by group members for not fitting in (Abrams, Rutland, Cameron, & Ferrell, 2007). Indeed, adolescents report that it is more acceptable to exclude or tease a gender atypical peer than a gender typical peer; this type of exclusion is a 'legitimate way of regulating behavior that they viewed as deviant or weird' (Horn, 2008, p. 185).
This type of gender-based teasing, teasing because the individual is acting atypical for his or her gender, is disproportionately directed at boys. Although peers make negative comments to girls when they engage in typically male activities like athletics and computers (Leaper & Brown, 2008), boys who appear feminine with poor athletic abilities face even harsher repercussions from their peers (Pascoe, 2007). Further, low gender typicality is associated with negative peer relations for boys but not girls (Lee & Troop-Gordon, 2011). This asymmetry, in which boys are more heavily sanctioned for low levels of gender typicality than girls, is supported by findings showing that girls exhibit gender atypical behaviors much more frequently than boys (Sandberg et al., 1993).

Negative Peer Relations Mediate Link between Gender Typicality and Mental Health
Gender typicality has not only been linked with positive and negative peer relations, but with mental health outcomes as well. Children who are low in gender typicality, or considered gender atypical, are more likely to have lower feelings of self-worth, more likely to be perceived by others as depressed and anxious, and at greater risk for suicide (Carver, Yunger, & Perry, 2003;Russell, Kosciw, Horn, & Saewyc, 2010;Yunger, Carver, & Perry, 2004). At its most severe, extreme gender atypicality is so strongly associated with depression and anxiety that it is labeled a psychological disorder (gender identity disorder) in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) (Fraser, Karasic, Meyer, & Wylie, 2010).
Although rarely addressed, the negative mental health outcomes associated with low gender typicality may be due to the accompanying negative peer relations (Smith & Leaper, 2005;Yunger et al., 2004). In other words, negative mental health may not be due to being atypical, but being teased for it (Haldeman, 2000). Peer teasing related to gender or appearance has been consistently linked with negative mental health outcomes. For example, peer teasing about one's appearance is associated with overall lower global self-esteem among adolescents, even after controlling for earlier selfesteem (Jones & Newman, 2009). Further, after gender-based peer teasing, children are more likely to feel self-conscious, embarrassed, hold a negative body image, experience a reduction in appetite, difficulty paying attention, and express a desire to avoid school (Cash, 1995;Harris Interactive, 2001).
Although research has shown that being low in gender typicality is associated with more negative mental health outcomes (e.g., Yunger et al., 2004), and that genderbased teasing accompanies low typicality (Horn, 2008) and is itself associated with more negative mental health outcomes (Harris Interactive, 2001), research has not explored gender-based teasing as a mediator between low typicality and negative mental health. This possible mediation is supported by Smith and Leaper (2005), who found that adolescents who perceived themselves to be accepted by their peers, despite feeling gender atypical, had higher self-worth than adolescents who were equally atypical but felt low peer acceptance. The current study explores whether being low in gender typicality predicts being teased because of that atypicality, and whether this gender-based teasing in turn predicts more negative mental health outcomes.

Current Study and Hypotheses
In the current study, children in early adolescence were asked to describe hypothetical popular and rejected/teased peers. We assessed whether they spontaneously associated popularity with gender typicality and rejection/teasing with atypicality. We hypothesized that descriptions of popular boys and girls would include more gender typical descriptions than those of rejected/teased boys and girls, and conversely that descriptions of rejected/teased boys and girls would include more gender atypical descriptions than those of popular boys and girls.
Second, children rated their peers based on popularity, likeability, and typicality. We hypothesized that individual children's gender typicality would predict their popularity among their peers, over and above their likeability. We also hypothesized, because of the asymmetry in the enforcement of norms for boys and girls (Pascoe, 2007), that gender would moderate this association such that boys' typicality would be more closely tied to their popularity than girls' typicality.

Jennifer A. Jewell and Christia Spears Brown
Third, children completed measures of mental health outcomes. We hypothesized that low gender typicality would be associated with greater depressive symptoms and anxiety, and lower self-esteem and body image. We also hypothesized that genderbased teasing would mediate this association in that being low in gender typicality would be associated with being teased more frequently, which in turn would be associated with more negative mental health outcomes.
The current study focused on gender typicality and peer relationships in middle school. We chose this age group for three reasons. First, gender intensification in middle school leads to greater gender conformity (Galambos, 2004;Hill & Lynch, 1983). Second, peer relationships take on a renewed importance during middle school, especially in relation to the protection that social support affords from bullying and/or being a bully (Demaray & Malecki, 2003;Rigby, 2000). Third, middle school-aged students are the most likely of any age group to be victims of teasing and bullying (Juvonen, 2001).

Participants
The participants in this study were 84 students from a public middle school (34 boys, 50 girls); by school year, 32 were in sixth grade (14 boys, 18 girls), 26 were in seventh grade (9 boys, 17 girls), and 26 were in eighth grade (11 boys, 15 girls). The participating school has 1081 students (in grades 6-8), with 171 enrolled in a required health class at the time of data collection; all 171 students received consent forms during their health class. One week after consent forms were sent home, 94 were returned with parental signatures. Of the students with parental consent, eight students were absent on the day of data collection, one female participant opted not to participate, and the data from one male participant were not included in analyses because of a large amount of missing data.
Of the 84 participants, all were between 11 and 15 years old (M = 12.46, SD = 1.02). Most were White (71.4 percent), with Hispanic/Latino (8.3 percent) and African-American (7.1 percent) being the next largest groups. These numbers are representative of the school and the community it serves (population approximately 300 000). Of the students at this school, 26 percent qualify for free and reduced lunch. Of the participants' parents, 28.4 percent of mothers and 20.3 percent of fathers had a high school degree or less, 39.5 percent of mothers and 44.3 percent of fathers had some college or a college degree, and 32.1 percent of mothers and 35.4 percent of fathers had a graduate degree.

Procedure
Trained research assistants administered the questionnaires to the class. Only the directions were read aloud; participants were instructed to complete the questions individually and stop when they had completed each section. Upon completion of all questions, students were given a university logo folder.

Hypothetical Popular and Rejected/Teased
Children. This open-ended questionnaire assessed perceptions of popular and rejected/teased peers. Participants wrote five words that described a popular boy, five words that described a popular girl, five words that described a boy who gets picked on or left out, and five words that described a girl who gets picked on or left out (based on LaFontana & Cillessen, 2002;Xie, Li, Boucher, Hutchins, & Cairns, 2006). They were instructed not to write the names of specific children, but to describe the children using adjectives and behaviors.
To code the descriptions, initially they were coded as being either stereotypical or counter-stereotypical for boys, stereotypical or counter-stereotypical for girls, or neither. Coding was based on two indices. First, descriptions were categorized based on the adjectives and behaviors listed on a validated gender stereotype measure, Child-Occupations, Activities, Traits (C-OAT, Liben & Bigler, 2002). For example, 'strong' is a masculine stereotyped item in C-OAT (Liben & Bigler, 2002), and thus was coded as stereotypical for boys in this study and counter-stereotypical for girls. Second, if the description was not included in C-OAT (N = 115 descriptions), it was independently rated by eight research assistants blind to the study. Each research assistant rated each description on a scale from highly masculine (-10) to neutral (0), to highly feminine (+10). This allowed us to assess which descriptions were closely (rather than loosely) associated with a stereotype. Descriptions that yielded a mean absolute value of greater than 5 were categorized as consistent with a stereotype (e.g., 'player' had a mean rating of -5.43 and was coded as masculine; 'talks behind backs' had a mean rating of 5.63 and was coded as feminine), and items with a mean rating between -5 and +5 were categorized as neutral. We also examined whether any descriptions could be included in both genders' stereotypes if the specific wording was altered (e.g., 'snobby' for girls and 'arrogant' for boys). This did not apply to any descriptions given by participants. The three most common stereotypical descriptions for boys were 'athletic', 'tall/ strong', and 'popular with girls'. The three most common stereotypical descriptions for girls were 'pretty', 'nice', and 'stylish/wears nice clothes' (a complete list can be obtained from the authors).
Typicality was calculated by summing the total number of stereotypical words (i.e., the number of masculine stereotypic words used to describe boys and feminine stereotypic words used to describe girls) given per participant for each hypothetical child. Atypicality was calculated by summing the number of counter-stereotypical words (i.e., the number of feminine stereotypic words used to describe boys and masculine stereotypic words used to describe girls) given per participant for each hypothetical child. Because participants were asked to give five descriptions for each hypothetical peer, numbers could range from 0 to 5.
Peer Ratings of Popularity, Likeability, and Typicality. Children were given a roster of all of the students in their grade who were enrolled in health class (this resulted in 76 names for sixth graders, 48 names for seventh graders, and 47 names for eighth graders). Because participants could bring the consent form back until the day of administration, we were unable to include only consented participants. Thus, all students were originally rated; however, only the names of participants who returned signed consent and assent forms were used in the analyses (as either raters or ratees). The institutional review board approved this methodology, and students were reminded that their ratings were confidential and that they were not to discuss any of their ratings with anyone except their parents.
First, each participant was asked to assign a rating to each of his or her peers on how popular he or she is (see Cillessen & Marks, 2011;French, 1988;Ladd, 1983;Terry & Coie, 1991). They read, 'Here is a list of some of the kids in your school. We know that some of them are popular and some of them are not popular. Please circle the number that shows how popular each kid is'. They rated each student in their grade from Not at all popular (1) to Very popular (4). They next rated their peers on likeability (see Cillessen & Marks, 2011;French, 1988;Ladd, 1983;Terry & Coie, 1991). Specifically, they read, 'Here is a list of some of the kids in your school. We know that some of them are liked by lots of kids and some of them are not liked by many kids. Please circle the number that shows how much you like each kid'. They rated each student in their grade from Not at all liked (1) to Liked a lot (4). Finally, participants rated their peers on gender typicality. Specifically, they read, 'Here is a list of some of the kids in your school. We know that some of the boys are very typical of boys (meaning the boys are very "boy-ish") and some of the girls are very typical of girls (meaning the girls are very "girl-ish"). But some of the boys are not very typical of boys (meaning they have some qualities that girls also have) and some of the girls are not very typical of girls (meaning they have some qualities that boys also have). Please circle the number that shows how typical of a boy or girl each kid is'. They rated each student in their grade from Not at all typical (1) to Very typical (4). If participants did not know a peer, they responded, 'Do not know'.
Self-Reported Gender Typicality. Based on Egan and Perry's (2001) gender identity measure, and adapted by Leaper and Brown (2008), we assessed participants' assessment of their self-reported typicality. Specifically, children were told, 'These are questions about how typical you feel that you are for your gender (as a typical boy or a typical girl). Please select how much each statement describes you'. Participants' self-reported typicality (e.g., 'I feel I am a good example of being a boy [girl]', 'I believe that I am a typical boy [girl] at my school') was measured using a six-item measure (one subscale from the original Egan & Perry, 2001, measure). Participants responded on a 4-point scale, ranging from Not at all like me (1) to A lot like me (4). A mean self-reported typicality rating was calculated for each participant, with a higher score indicating greater self-reported typicality. This scale was adapted to be gender-specific, and each participant received the scale corresponding with his or her gender. Internal consistency was good for both boys (a = .88) and girls (a = .87).
Mental Health Outcomes. Participants' depressive symptoms and anxiety were assessed using a 22-item measure consisting of two subscales from the behavior assessment system for children (BASC-2; Reynolds & Kamphaus, 2004). Sample items included the following: 'Little things don't bother me' and 'I am afraid I might do something bad' (reverse-coded). Participants answered these questions using a 4-point scale, ranging from Never (1) to Almost always (4). Participants' self-esteem and body image were also assessed using a 13-item measure used by Leaper and Brown (2008), adapted from Rosenberg (1979) and McKinley and Hyde (1996). Sample items included the following: 'I am able to do things as well as most other people' and 'I am not good-looking' (reverse-coded). Participants indicated their agreement with each statement using a 4-point scale, ranging from Disagree strongly (1) to Agree strongly (4). Internal consistency was acceptable or good for all measures [depressive symptoms, a (boys) = .76, a (girls) = .74; anxiety, a (boys) = .73, a (girls) = .76; self-esteem: a (boys) = .82, a (girls) = .82; body image: a (boys) = .76, a (girls) = .79].

Gender Typicality in Adolescence 143
Gender-based Teasing. Participants were also asked whether 'anyone had made discouraging comments, "put them down", or made fun of them for not being a typical or "regular" girl or boy' (adapted from Leaper & Brown, 2008). Participants were asked whether they heard negative comments about not being a typical boy or girl from nine different sources: coaches/teachers, mother, father, close female friends or sisters, close male friends or brothers, other family members, neighbors, other girls, other boys, and anyone else not described above. Participants responded for each source using the following scale: No (1), Yes, once or twice (2), Yes, a few times (3), and Yes, several times (4). A mean score was calculated. This scale was gender-specific, and each participant received the scale that corresponded with his or her gender. Internal consistency was good for both boys (a = .80) and girls (a = .86). Table 1 shows the overall means and SDs of all variables. There were no gender differences in depressive symptoms or anxiety. Boys, however, had higher self-esteem than girls [M (SDs) = 3.65 (.42), 3.40 (.51), respectively], t(78) = -2.50, p < .01. Boys also had more positive body image than girls [M (SDs) = 3.38 (.64), 3.09 (.66), respectively], t(80) = -1.99, p = .05. Boys were rated by their peers as significantly more typical than girls [M (SDs) = 3.33 (.38), 3.10 (.54), respectively], t(82) = -2.21, p < .05. There were no gender differences in peer ratings of popularity or likeability.

Preliminary Analyses
All correlations between variables were tested for gender differences and are reported in Table 1. Body image correlated positively with peer ratings of popularity and typicality for boys but negatively for girls. Anxiety correlated negatively with peer ratings of typicality for boys but positively for girls. Anxiety and self-esteem correlated more negatively with self-reported typicality for boys than girls.
Preliminary analyses also revealed that 69 percent of participants reported experiencing gender-based teasing at least once. There were no gender differences in reported teasing: 68 percent of girls and 70.6 percent of boys reported experiencing it at least once. Eleven boys reported being teased by girls at least once whereas 17 boys reported being teased by other boys at least once. Twenty-four girls reported being teased by boys at least once whereas 22 girls reported being teased by other girls at least once. There was considerable overlap between who each gender was teasing: 11 boys reported being teased at least once by both boys and girls whereas 17 girls reported getting teased at least once by both boys and girls.
Hypothesis 1: There is an association between popularity and gender typicality, and rejection/teasing and gender atypicality.
To examine whether children generally associated (1) popularity with gender typicality and (2) rejection/teasing with gender atypicality, descriptions of hypothetical popular boys and girls and hypothetical rejected/teased boys and girls were compared. The total number of typical (i.e., stereotypical for the gender) and atypical (i.e., counter-stereotypical for the gender) descriptions for each hypothetical child were examined. As can be seen in Table 2, popular boys and girls were described with more typical descriptions and fewer atypical descriptions than predicted by chance. By contrast, rejected/teased boys and girls were described with fewer typical and atypical descriptions than predicted by chance.  To more closely examine participants' descriptions of hypothetical peers, two different 2 (target popularity status: popular, rejected/teased) ¥ 2 (target gender: boys, girls) ¥ 2 (participant gender: boys, girls) mixed analyses of variance (ANOVAs) were conducted, in which the first two variables were within-subjects variables. One ANOVA was conducted for the number of typical descriptions, and one ANOVA was conducted for atypical descriptions. Means are presented in Table 3.
In terms of typical descriptions, as predicted, there was a main effect of target status such that popular children were described with more typical descriptions than rejected/ teased children, F(1, 83) = 440.47, p < .001. There was also a main effect of target gender such that girls were described with more typical descriptions than boys, F(1, 83) = 11.78, p < .01. There were no significant effects involving participant gender.
In terms of atypical descriptions, as predicted, there was a main effect of target status such that rejected/teased children were described with more atypical descriptions than popular children, F(1, 83) = 7.60, p < .01. There was also a main effect of target gender such that boys were described with more atypical descriptions than girls, F(1, 83) = 10.52, p < .01. However, this main effect of target gender was superseded by the interaction between target gender and status, F(1, 83) = 43.12, p < .001. Test of simple effects indicated that, as predicted, rejected/teased boys were described with more atypical descriptions than popular boys, t(83) = -6.15, p < .001. Counter to expectations, however, popular girls were described with more atypical descriptions than rejected/teased girls, t(83) = 4.03, p < .001. Follow-up analyses indicated that this effect was driven by the inclusion of 'athletic' and 'independent' (i.e., counterstereotypical descriptions for girls) as descriptions for popular girls. There were again no significant effects involving participant gender.
Hypothesis 2: Gender typicality predicts popularity among peers, controlling for likeability and moderated by gender.
We predicted that children rated as high in gender typicality by their peers would also be rated as high in popularity by their peers (over and above being liked), and gender would moderate this association. Hierarchical multiple regression analysis was used to test this hypothesis. To predict peer ratings of popularity (the average peer rating received by each participant), gender (dummy-coded) and average peer ratings of likeability were entered in the first step, followed by average peer ratings of typicality in the second step. To test moderation, the interaction term between gender and typicality was entered in the third step. The overall model predicting popularity was significant, F(4, 79) = 20.74, p < .001. Within the first step, gender was not a significant predictor of popularity, but peer ratings of likeability were significant, b = .28, p < .01; first step R 2 = .35, p < .001. Students who were liked more by their peers were rated as more popular. The second step significantly predicted peer ratings of popularity above and beyond the first step (DR 2 = .14, p < .001), with peer ratings of typicality being a significant predictor, b = .40, p < .001. Controlling for gender and peer ratings of likeability, students higher in gender typicality were rated as more popular.
As predicted, this effect was moderated by gender, interaction b = .29, DR 2 = .06, p < .01. Analysis of the interaction revealed that peer ratings of typicality predicted Gender Typicality in Adolescence 147 popularity more strongly for boys (b = .78, R 2 = .61, p < .001) than for girls (b = .58, R 2 = .34, p < .001). A Sobel test indicated that this is a significant difference, z = 4.08, p < .001. This interaction can be seen in Figure 1. When the model was run separately for boys and girls without including gender as a predictor, the model accounted for almost twice as much variance in peer ratings of popularity for boys than girls (61 percent vs. 34 percent). Follow-up analyses confirmed that the relationship between peer ratings of typicality and popularity are linear, and not curvilinear (results available upon request).
Hypothesis 3: Typicality is associated with mental health outcomes and mediated by gender-based teasing.
The third hypothesis was that lower gender typicality would be associated with more negative mental health outcomes (i.e., more depressive symptoms and anxiety, and more negative self-esteem and body image). To test this hypothesis, a series of hierarchical multiple regression analyses were conducted on depressive symptoms, anxiety, self-esteem, and body image. Self-reported typicality was always entered in the first step, along with gender. To test moderation effects of gender, the interaction terms between gender and typicality were always entered in the second step. We also predicted that gender-based teasing would mediate these associations. Thus, using the definitions outlined by Muller, Judd, and Yzerbyt (2005), we examined whether there was (1) moderated (by gender) mediation or (2) mediated moderation. We tested mediation with a series of regressions recommended by Baron and Kenny (1986).
Body Image. In the first regression, typicality alone did not significantly predict body image, b = .17, NS. Instead, the interaction between typicality and gender was significant, b = .27, p < .05; final F(3, 80) = 3.41, p < .05, R 2 = .12. As with anxiety and self-esteem, typicality positively predicted body image for boys (b = .43, p < .05) but not girls (b = .03, NS). Thus, mediation was only tested for boys. When gender-based teasing was entered in the first step and typicality entered in the second step, typicality was no longer a significant predictor of body image for boys, b = .28, NS. A Sobel test indicated that this was a significant decrease in association between typicality and body image, z = 3.05, p < .01. In other words, gender-based teasing fully mediated the association between low typicality and more negative body image for boys (i.e., mediated moderation). Girls' typicality was unrelated to their body image.
Alternative Models. It is reasonable to hypothesize that gender typical children are more popular or more liked, and this positive peer status predicts positive mental health outcomes. To rule this out, alternative models were also examined testing both popularity and likeability as potential mediators. Results indicated that none of these mediational relationships were significant (complete analyses and null results available upon request).

Discussion
The current study examined the links among gender typicality, peer relations, and mental health outcomes. Overall, children in early adolescence thought of popular peers as very gender typical, indicated by both the descriptions of hypothetical popular peers and the ratings of the participants themselves. Importantly, the links between being gender typical and being popular were not simply a reflection of gender typical children being liked more, although they were liked more (consistent with Egan & Perry, 2001). Hypothetical popular children were described as gender typical, but not necessarily prosocial (e.g., boys were described as 'having a temper'); in addition, typicality predicted popularity, even when likeability was controlled. This supports the peer popularity literature that has consistently shown that being popular is not synonymous with being well liked or having positive traits (e.g., Adler & Adler, 1998;Adler et al., 1992;Rose et al., 2011). Rather, gender typicality was associated with social status and prestige among peers (consistent with Adler et al., 1992;Lease et al., 2002;Vaillancourt & Hymel, 2006). Considering the importance placed on popularity by early adolescents (LaFontana & Cillessen, 2010;Merten, 2004), it is likely that children feel, at least implicitly, pressure from their peers to be gender typical to gain or maintain social status.
High typicality was associated with positive peer relations, while low typicality was associated with negative peer relations. Children low in gender typicality reported more frequent gender-based teasing than more typical children (consistent with previous work, e.g., Young & Sweeting, 2004). Ironically, a majority of children experienced teasing for being atypical at least once. This supports previous research indicating that many children feel, at times, atypical (Smith & Leaper, 2005). Thus, children in early adolescence are not only feeling implicit pressure from peers to be gender typical, but explicit pressure as well.
Because it is unclear from the current study, future research should examine the exact nature of teasing and should address how much atypicality children can express before being teased. Future research should also examine whether some atypical behaviors are associated with teasing more than other behaviors, and whether some atypical children are excluded from teasing because of other protective factors (e.g., an atypical, but attractive girl). Regardless, because of the overall high frequency of gender-based teasing, schools and parents should more explicitly monitor the teasing and bullying that occur among children. Teasing is routinely associated with negative mental health outcomes (Jones & Newman, 2009) and should not be considered a normative part of development.
As predicted, however, gender played a critically important moderating role. Specifically, gender typicality was more closely linked with popularity for boys than girls. For example, rejected or teased boys were much more likely to be described with atypical traits than rejected or teased girls, and gender atypical descriptions (such as 'athletic' and 'independent') were used to describe popular girls yet never popular boys. In addition, the link between levels of typicality and popularity was considerably stronger for boys than girls.
This asymmetry in the association between gender typicality and popularity suggests that boys are more heavily sanctioned for low levels of gender typicality than girls are. This is supported by the results indicating that boys were rated overall as more gender typical than girls. Previous research has shown that boys feel more pressure to conform to gender stereotypes from peers and parents than girls do (Brown, Alabi, Huynh, & Masten, 2011;Egan & Perry, 2001;Leaper, 2002;Smith & Leaper, 2005). Further, the consequences of breaking traditional gender roles are harsher and more consistent for boys than girls (Leaper, 1994).
This asymmetry also may partially reflect changing gender norms for girls. Specifically, the stereotypes about girls' ability to be athletic and independent may be changing. For example, although only 1 in 27 high school girls played sports prior to Title IX, now it is close to half of girls (Heywood & Dworkin, 2003;National Federation of State High School Associations, 2004). Likewise, it may be more acceptable today for girls to assert themselves as independent and self-motivated than in the past. Research in the field of counseling psychology has similarly documented changes in gender stereotypes for women, such that women with good mental health are now associated with traits like 'independent', 'strong', and 'enjoys a challenge' (Seem & Clark, 2006). Given that masculine traits have been associated with positive outcomes for women since the early 1980s, it is possible that this shift in acceptable traits for females has been developing for many years (Taylor & Hall, 1982).
There were not only gender differences in the links between gender typicality and popularity, but also the associations between typicality and mental health outcomes were markedly different for boys and girls. For boys, low self-reported gender typicality was associated with greater depressive symptoms, greater anxiety, more negative self-esteem, and more negative body image. By contrast, girls' self-reported gender typicality was only related to their depressive symptoms. This asymmetry may be due to the greater social importance of typicality for boys relative to girls, such that perceiving oneself to be low in typicality is more detrimental for boys' mental health. There could, however, be underlying factors that lead boys to be both low in gender typicality and have more negative mental health outcomes, such as hormonal deficiencies or pubertal timing.
The current study predicted that gender-based teasing would mediate this link between typicality and mental health, and this prediction was partially supported. Gender-based teasing fully mediated boys' association between low typicality and greater depressive symptoms and more negative body image. Boys who were low in gender typicality were teased more on the basis of gender, and in turn expressed more depressive symptoms and felt worse about their bodies (as in Smith & Leaper, 2005). Regardless of gender-based teasing, however, boys who were low in typicality had greater anxiety and lower self-esteem (teasing accounted for some of the low selfesteem, but not much).
It is unclear why gender-based teasing fully mediated associations with depressive symptoms and body image, but only partially mediated associations with self-esteem and did not mediate associations with anxiety. All explanations would be post hoc and speculative. For example, perceiving unfair treatment because of gender impacts children's self-esteem in complex ways, dependent on the degree to which gender is valued (Brown, Bigler, & Chu, 2010). Additional mediators and moderators that may be relevant for self-esteem and anxiety were not necessary for depression and body image. Additional work needs to be conducted to examine exactly how gender-based teasing impacts boys' mental health.
For girls, however, there are unique challenges to positive mental health. Overall, girls reported lower self-esteem and more negative body image than boys (consistent with Ah-Kion, 2006;Graham, Eich, Kephart, & Peterson, 2000). It is likely that cultural images of thin girls and women conveyed in the media, as well as the societal emphasis placed on women's bodies in general, impacted the girls as a whole. Further, Gender Typicality in Adolescence 151 across our middle school girls, girls have more negative body image with age, likely a function of developing pubertal status during middle school (Angold, Costello, & Worthman, 1998).
Gender typicality was related to girls' mental health outcomes, but in ways different from boys. As with boys, girls who were low in self-rated gender typicality were teased more on the basis of gender, and in turn expressed more depressive symptoms. Anxiety and body image, however, showed different trends. Self-reported typicality was unrelated to these outcomes, but peer-rated typicality was. Girls who were rated as more typical and popular by their peers held more negative body image and had more anxiety than girls who were rated as less typical and popular. This finding parallels other research with adolescent girls indicating that popular girls feel more pressure to be thin than less popular girls (Rancourt & Prinstein, 2010). Future research should explore how other people's perceptions of typicality differ from self-perceptions of typicality. Girls seem to be more impacted by others' perceptions of typicality, but the exact mechanisms involved are unclear.
Thus, there is an important gender distinction in the current study. For boys, being perceived by peers as high in gender typicality, and perceiving oneself to be high in gender typicality, was associated with more popularity and likeability, less genderbased teasing, and more positive mental health outcomes. For girls, being perceived by peers as high in gender typicality was associated with more popularity and likeability, but more negative body image and greater anxiety. Perceiving oneself to be high in typicality was only associated with fewer depressive symptoms, and this was fully mediated by gender-based teasing. It appears, then, that gender typicality operates differently for girls and boys. For boys, high typicality is overwhelmingly positive (and conversely, low typicality is overwhelmingly negative); for girls, however, there is a double-edged sword of typicality, as it is associated with positive social outcomes but more negative mental health outcomes.
As with all research, there are limitations to this study. The most important limitation is the cross-sectional, correlational nature of the study. Future longitudinal research should be conducted to examine whether gender typicality at one time point predicts mental health at a later time point. In addition, the sample was relatively small and European American. This limits the ability to generalize the findings to other populations of adolescents. Future research should explore gender typicality in more racially, ethnically, and economically diverse children. For example, research has shown ethnic group differences in body image, such that there are differences in the ideal body type among European American, African-American, and Latino/Latina adolescents (see Barroso, Peters, Johnson, Kelder, & Jefferson, 2010;Schooler, 2008). How these ideal body types interact with popularity and typicality (particularly for children who may be in the minority at their school and thus atypical across other domains) are complex and need to be addressed in future work.
In addition to cultural differences, it is likely that different schools place a different emphasis on gender typicality. The relationships above might not exist in schools that encourage students to express themselves regardless of gender stereotypes. Future research should explore whether typicality might be strongly associated with popularity if the school teaches and enforces gender stereotypes, or more weakly associated if the school encourages self-expression regardless of gender stereotypes.
Finally, future research should examine the implications of being either low or high in typicality, as there may be different ideal thresholds for boys and girls. Specifically, there may be a lower threshold for boys, such that too much atypicality is associated with negative ramifications. There may also be an upper threshold for girls, such that too much typicality is similarly associated with negative ramifications. Regardless of gender, however, it is clear that gender typicality is important for both peer relations and mental health. The importance of typicality for children in early adolescence is not likely to wane in the near future, and thus should continue to be explored, not as a psychological disorder but as a consequence of the social world in which we live.