1 Introduction

In examining children’s well-being, the peer relationships context is important. In this context, school bullying can be a vital contributor to psychosocial adjustment (van den Eijnden et al. 2014). Experiences of bullying and victim behaviors have been found to be related to negative effects in children’s well-being, such as low self-esteem, depression or behavioral problems (Shetgiri et al. 2013). Bullying refers to a subset of aggressive behavior toward a victim who cannot easily defend him/herself, by one perpetrator or more, characterized by repetition over time, imbalance of power, and occurring without provocation (Smith and Brain 2000). Bullying can be indirect (e.g., rumor spreading), verbal (e.g., teasing) or physical (e.g., pushing).

In order to improve preventive efforts to reduce such aggression, past research on school bullying has identified several individual factors that characterized involved as victims or bullies, such as gender. In fact, most research on this problem includes the analysis of gender differences, even when it is not the primary goal. Extensive research has revealed that males are most frequently involved in bullying as either a bully or victim (Cook et al. 2010; Slee 2006). Nevertheless, past research has tended to show that girls engage in far more relational aggression than boys (Crick and Grotpeter 1995; Wang et al. 2009). However, this claim has not been supported by recent research, which suggests that indirect aggression is a male, and not a female, issue (Artz et al. 2008; Artz et al. 2013).

Regardless of these results, we believe that is important to note that even though we know that these differences exist, it does not mean that we know why they occur or, conversely, why they do not occur. This implies that it is necessary to analyze how gender experiences contribute to differences between boys and girls in relation to their involvement in bullying more carefully. We understand gender as a set of beliefs learned through the socialization process, which relate specifically to either gender. They are marked differences as to how men and women are but, above all, as to how they should behave. The gender variable becomes important if consider, as Matud pointed out (Matud 2004), that gender may mediate social relations, thereby creating differences between men and women in a variety of fields, including involvement in aggressive behaviors. In this study, school bullying is our indicator of well-being and we test how victims, bullies and bully-victims roles are associated with gender experiences.

Among gender experiences, gender identity has been found to affect psychosocial well-being in different ways (Carver et al. 2003). Traditionally, gender identity has been defined as an individual’s knowledge of his or her membership in a gender category (Kohlberg 1966). However, according to Egan and Perry (2001), by the age 6 or 7, all children attain full gender constancy and this gender identity facet cannot account for within-sex individual differences in other variables, such as psychosocial adjustment. Another conceptualization of gender identity has described it as the degree to which one internalizes cultural gender-related attributes and sees themselves as masculine or feminine in relation to those attributes (Bem 1981). Following this conceptualization, many researchers have studied gender identity through children’s self-perceived gender typicality from their self-rating on sex-typed attributes as personality traits. Self-attributions of instrumental traits (e.g., independence, assertive, brave) have been used to infer self-perceived masculinity, and self-attributions of expressive traits (e.g., sensitive, kind, emotional) have been employed to infer self-perceived femininity (Spence 1993).

The most widely used instruments to assess the degree to which men and women identify with culturally prescribed traits expected for their sex (gender typing) are the Personal Attributes Questionnaire (PAQ, Spence et al. 1974) and the Bem Sex Role Inventory (BSRI, Bem, 1974). Both the BSRI and the PAQ were developed for use with adults. Gender typing with children and adolescents have been examined with the Children’s Personal Attributes Questionnaire (CAPQ, Hall and Halberstadt 1980). It was created by excluding three items from the adult PAQ because the content was judged to be inappropriate for children (i.e., interest in sex). Two forms of the tool were developed: the 51-item long form and the 21-item short form (Hall and Halberstadt 1980). The CAPQ assess two sets of gender-related traits (instrumental and expressive traits) and is comprised of the same three subscales found in the PAQ: masculine, feminine and feminine/masculine. Previous psychometric analysis have reported acceptable reliabilities on the scales and have found gender differences with boys scoring higher than girls on the masculine traits, whereas girls scoring higher than boys on the feminine traits (Absi-Semaan et al. 1993; Hall and Halberstadt 1980; Thomson and Zand 2005). In relation with aggressive behaviors, this perspective argues that those who build their self-identity through masculine traits like self-assertion, self-expansion and dominance may behave aggressively, whereas the construction of self-identity through feminine traits that emphasize self-sacrifice or concern others may be related with inhibition of aggression (Navarro et al. 2011). Different studies have explored the relations between these self-perceptions and bullying in children and teenagers. Research has found that self-attribution of masculine traits is positively related with being a bully among children and adolescents (Gianluca and Pozzoli 2006; Navarro et al. 2011; Young and Sweeting 2004), and also with positive attitudes to bullying (Carrera-Fernández et al. 2013). On the contrary, self-attribution of feminine traits has been positively related with negative attitudes to bullying (Carrera-Fernández et al. 2013), as well as less involvement as a bully, but with a higher risk of being a victim, at least among boys (Navarro et al. 2011). Other researchers have tested if internalization of feminine traits is related with the use of indirect or covert forms of aggression that are stereotypically associated more with girls. In line with this, Crothers et al. (2005) found that girls with higher self-attribution of feminine traits were more likely to be bullies through indirect forms of aggression. Nevertheless, other researchers have failed to find this relationship (Kolbert et al. 2010; Navarro et al. 2011).

However, this view of gender identity may also represent only one dimension of the whole gender identity structure: gender typicality. Moreover, this conceptualization of gender typicality is based exclusively on gender typing in a single domain (personality traits), whereas gender typicality includes several domains: toy and activity preferences, academic pursuits and occupational preferences. For this reason, more recently, Egan and Perry (2001) proposed gender identity to be a multidimensional construct composed of five different components: a) the traditional view of gender identity: membership knowledge in a gender category; b) gender typicality, self-perceived similarity to other members of the same gender category that is more abstracted and synthesizes diverse information about one’s gender typing, and not only personality traits; c) gender contentedness, satisfaction with one’s gender assignment; d) felt pressure for gender conformity; and e) intergroup bias, the extent to which one believes one’s own sex is superior to the other. Egan and Perry (2001) developed a 34-item self-reported measure that assesses the last four of these components of gender identity in preadolescent children and found the components to be relatively independent and to relate to adjustment (i.e., self-esteem and peer acceptance) in different ways. Several studies have provided empirical support for Egan and Perry’s multidimensional gender identity model. Researchers have reported gender differences in three dimensions of gender identity. Boys scored higher than girls on gender typicality, gender contentment and felt pressure (Egan and Perry 2001; Carver et al. 2003). However, not all the studies have found gender differences for intergroup bias (Yu and Xie 2010). The model has been tested to know the relationship between these gender identity components and children’s adjustment. Previous research has found that gender typicality, gender contentedness and felt pressure, but not intergroup bias, are related to different psychosocial adjustment indices, such as self-esteem, peer-acceptance, sense of loneliness or social competence (Carver et al. 2003; Corby et al. 2007; Egan and Perry 2001; Yunger et al. 2004; Yu and Xie 2010). Very little research has been conducted on the relationship between these gender identity dimensions and bullying behaviors as indicators of well-being. Studies addressing these relationships have examined only the role of victim, and not all the studies have included all the gender identity dimensions of the model. One clearer result is that children who reported more gender typicality reported less victimization (Carver et al. 2003; Drury et al. 2013; Roberts et al. 2013), which indicates a relationship between gender atypical behavior and problems with peers.

Despite research in this area not being available, various relationships between the other three gender identity dimensions (gender contentedness, felt pressures and intergroup bias) and bullying behaviors have been theoretically suggested. Regarding gender contentedness, lack of one’s satisfaction with his or her gender can be related with victimization processes as a result of the negative social reactions to which they are exposed. That is, children who wish to be of the other sex or who desire to engage in cross-sex activities may be seen as easy prey to aggressive peers (Egan and Perry 1998). However, since these children can feel greater peer pressure to gender conformity, they could incur in aggressive behaviors as a way to confront this social pressure. Indeed, Carver et al. (2003) found that gender dysphoric girls were nominated by their peers as being more aggressive and disruptive than the other girls. Felt pressure may be associated with specific gender-typed social behaviors (Pollack 1998). According to this argument, pressure to gender conformity can predispose boys to antisocial tendencies (e.g., aggression) and predispose girls to subservience, which may be related to victimization. Finally, intergroup bias may be related with aggressive behaviors if we consider that negative attributes toward the other sex can increase hostility and dismiss cooperative and respectful peer interactions (Powlishta 1995).

1.1 The Present Study

In short, children’s involvement in bullying behaviors is assumed to be associated with gender identity components, such as gender typicality, and also with gender typing constructs, such as gender-typed personality traits. For example, high gender typicality is associated with less victimization, while high levels of masculine traits are related to perpetration.

Previous research on school bullying has focused primarily on two categories of involvement: victims and bullies. Less research has examined youths who are both victims and perpetrators of bullying as a distinct group. Bully-victim peers have been described as easily angered and provoked (Veenstra et al. 2005), have more adjustment problems and are rejected more by peers (Haynie et al. 2001). Therefore, this group may be related differently to gender measures.

However, no study has to date examined the association among the three different roles in school bullying (victims, bullies and bully-victims) in terms of gender identity on the one hand and gender typing measures on the other. Consequently, the present study aims to explore the relative contribution of gender typicality, gender contentedness, felt pressure and intergroup bias (gender identity measures), and also the self-attributions of gender-typed personality traits (gender typing measures), to the prediction of victimization and perpetration status in school bullying. Moreover, since literature on this type of research in Spanish Children is scarce, the present research was designed to extend Egan and Perry’s (2001) multidimensional model of gender identity to the Spanish culture. Specifically, the following main hypotheses were examined.

1.1.1 Hypothesis of Gender Effects on Spanish Children’s Gender Identity and Gender-Typing Measures

  1. H1

    It was hypothesized that boys will feel more compatibility with their gender (in typicality and contentedness terms) and will feel more pressure to conform to gender stereotypes than girls.

Past research has shown that boys experience more pressure for gender typing than do girls (Carver et al. 2003; Yu and Xie 2010). In Spanish culture, research has found that boys hold more gender stereotypes than girls (Navarro et al. 2014). Therefore, it seems that Spanish boys should report higher levels of typicality and meanwhile be more satisfied with their gender than girls do.

  1. H2

    It was hypothesized that no gender differences will be found in children’s intergroup bias.

Research examining intergroup bias has shown that young children are especially given to same-sex favoritism (Ruble and Marin 1998; Powlishta 1995). However, previous studies have reported mixed results with researchers founding greater intergroup bias for girls than for boys, and other researchers showing that gender had no effects on intergroup bias (Yu and Xie 2010). Given the mixed results we hypothesized that intergroup bias may not show distinct gender differences.

  1. H3

    It was hypothesized that boys’ gender typing will be greater in terms of self-attribution of masculine traits than girls, and that girls’ gender typing will be greater in terms of self-attribution of feminine traits than boys.

Research examining the interiorization of masculine (instrumental traits) and feminine (expressive traits) has reported that boys self-attributed higher levels of masculine traits than girls do, whereas girls self-attributed higher levels of feminine traits than boys do (López-Sáez et al. 2008; Witt 2000). Based on this as well as on a previous study with Spanish adolescents indicating the same finding (Navarro et al. 2011) we expect to find this again with a sample of preadolescents.

1.1.2 Hypotheses of the Relationship Between Gender-Related Constructs and Bullying Behaviors in Spanish Children

  1. H4

    We hypothesized a negative relationship between typicality, contentedness and peer victimization in that more gender typical and contented children will be less victimized.

Past research has shown that there is an association between gender atypical behavior and problems with peers (Roberts et al. 2013). Based on this as well as on another study with preadolescents indicating that gender typical children are less victimized, we hypothesized a negative relationship between typicality and contentedness and bullying victimization in that more gender typical and content children would be less victimized.

  1. H5

    Intergroup bias and lack of gender contentedness will be positively related with bully and bully-victim roles.

We are not aware of past research examining gender identity dimensions and bullying perpetration; as such this portion of the study was exploratory. Based on research suggesting that intergroup bias can increase hostility and dismiss cooperative and respectful peer interactions (Powlishta 1995), we expect to find an association between intergroup bias and bully and bully-victim roles. In the same way, based on previous research suggesting that gender dysphoric children are described by their peers as more aggressive and disruptive (Carver et al. 2003), we expect that lack of gender contentedness would be associated with perpetration roles.

  1. H6

    Felt pressure will be positively related with bullies or bully-victim roles in boys and with the victim role in girls.

Due to the more stringent social rules imposed on boy’s gender typing over girls’ (Egan and Perry 2001) it is hypothesized that boys’ felt pressure would be positively related with boys’ perpetration. Moreover, since felt pressure may be associated with gender-typed social behaviors as subservience in girls (Pollack 1998), it is hypothesized that felt pressure will be positively related with girls’ victimization.

  1. H7

    Self-attribution of masculine traits will be positively associated with bullies or bully-victim roles, whereas self-attribution of feminine traits will be positively related with less involvement as a bully or bully-victim. However, boys with self-attribution of feminine traits may be more victimized.

Past research has documented that self-attribution of masculine traits is related with perpetration whereas self-attribution of feminine traits is associated with less involvement in aggressive behaviors for both boys and girls (Gianluca and Pozzoli 2006; Young and Sweeting 2004). Therefore, it is hypothesized that masculine traits will be positively related with perpetration roles, and feminine traits will be negatively related with those same roles. Additionally, based on previous findings showing that peer victimization can be a way to encourage boys’ gender conformity to personality traits (Navarro et al. 2011), it is also hypothesized that self-attribution of feminine traits will be positively related with boy’s victimization.

2 Method

2.1 Participants and Procedure

In order to conduct the present exploratory study with a statistical power of .95, an a priori estimation of the minimum sample size required was performed. The error rates in statistical inference in the conventional limits (α = β = .05) and a medium-small effect size (f = .15, Cohen 1977) in the bullying involvement and not involvement groups were set. Results indicated that the minimum sample size required was 63 participants in each of the two groups (Faul et al. 2009).

Considering this result, we intentionally oversample in order to obtain enough children involved in bullying behaviors, but also to increase the statistical power by augmenting the number of participants in the control conditions. To do that, schoolchildren in grades 5 and 6 were sampled randomly from five primary schools. These primary schools were located in a city of central Spain with an approximate population of 60 000 (INE 2013).

With the Local Education Bureau’s assistance, four standard primary schools were randomly selected. All the children studying grades 5 and 6 were eligible to be included in the study. The children whose parents had agreed they could participate were recruited. Among them, 50 boys and 50 girls were selected randomly from each grade in each school. Completed data were obtained from 445 of the 500 distributed questionnaires. Participants included 208 girls (M(age): 10.78, SD = 0.74) and 237 boys (M(age) = 10.78, SD = 0.68). These children were asked to complete a survey about what they thought about themselves. It took approximately 35 min for them to complete the questionnaires. Data were collected at the end of academic year 2013/2014. The study was conducted in adherence to the legal requirements of the study country and in compliance with IRB approval.

2.2 Measures

Gender-related measures were originally developed in English. They were translated into Spanish and back-translated into English to ensure linguistic and conceptual equivalence.

2.2.1 Multidimensional Gender Identity Inventory

The Gender Identity Inventory is a 34-item self-report measure developed by Egan and Perry (2001). The inventory includes four subscales that assess gender typicality (item example: “I feel that my personality is similar to most girls/boys”), gender contentment (item example: “I like being a girl/boy”), felt pressure for gender conformity (item example: “I think that the girls/boys I know would be upset if I wanted to play with boys’/girls’ toys”), and intergroup bias (item example: I think that girls/boys are more creative than girls/boys”). For each item, children first decided which of the two kinds of children described in the item they were like more, and then indicated whether this choice was very true or sort of true for them. Scale scores were computed by averaging across items and ranged from 1 to 4, for which a higher score on each scale represented a higher level of gender typicality, gender contentment, felt pressure and intergroup bias, respectively. The reported Cronbach’s coefficients for the four scales in studies done in the United States were all around 0.80 (Carver et al. 2003). In this study, the internal consistencies of the four scales were gender typicality (0.71), gender contentment (0.60), felt pressure (0.81), and intergroup bias (0.80), which are similar values to those reported in other cultures (Yu and Xie 2010).

2.2.2 Gender-Typed Personality Traits

The short Children’s Personality Attributes Questionnaire (CAPQ; Hall and Halberstadt 1980) was used to assess gender-typed personality traits. The scale contains 21 items using a 4-point Likert scale ranging from 1 (not at all true of me) to 4 (very true of me). For the purpose of this study, a 16-item version was administered. In order to be able to compare our results with previous findings, the version used included only two subscales, the masculine scale (eight items; for example “I am often the leader among my friends) and the feminine scale (eight items; for example “I am a very considerate person”), and measured stereotypically for equally desirable masculine and feminine attributes. After reverse scoring was done as needed, scale scores were obtained by summing the ratings for the items of each scale; the higher the scores, the greater the endorsement of measured traits. Internal consistency (Cronbach’s Alpha) was .71 for the masculine scale, and .72 for the feminine scale, which are similar values to those reported by Absi-Semaan et al. (1993).

2.2.3 School Bullying

We assessed school bullying behaviors using the items from the Spanish Self-reported Bullying-Victimization Questionnaire (Cava et al. 2007). Each scale was a 12-item self-reported measure on which participants indicated how often they had been victims of school bullying or had been a school bully in the last 8 weeks. Items scored on a 5-point scale (1 = Never, 2 = Once a month, 3 = Once a week, 4 = Once a day, 5 = Several times a day). Participants first read the school bullying definition provided by Roland and Idsøe (2001). After reading the definition, participants rated each behavior. One example item measuring school bullying victimization was “one student or some has/have despised or humiliated me”. The equivalent item measuring direct school bullying perpetration was “I have despised or humiliated a student or other students”. Internal consistency in this sample (Cronbach’s Alpha) was .92 for the victimization scale and .90 for the perpetration scale.

2.3 Data Analysis

We first detailed the general descriptive of the independent variables. Then we examined the descriptive data related to participants’ involvement in bullying behaviors. Third, we performed Pearson correlations between online and traditional bullying involvement and all the study variables. Finally, we used Multinomial Logistic Regression to examine the relationship between the online and traditional bullying categories and the independent variables, while simultaneously controlling for how other variables may influence them all. The reference group in this analysis was the students uninvolved in bullying behaviors. We utilized the SPSS 19.0 statistical software for all the analyses.

3 Results

The results are shown in four parts. We first presented the prevalence data of bullying behaviors. Second, we summarized gender differences in the gender identity dimensions and gender-typed personality traits. Subsequently, we presented the intercorrelations among all the variables, followed by the relationship among the different components of gender identity and bullying roles as indicators of psychosocial adjustment.

3.1 School Bullying Involvement by Gender and Grade

Participants’ categorization as victims, bullies or bully-victims was done by following a highly restrictive criterion, similar to that which other bullying researchers have used (Scheithauer et al. 2006). Students who indicated suffering, but not perpetrating, for at least one of the behaviors included in the questionnaire several times a week were classified as victims. Students who reported perpetrating, but not suffering, at least one of the behaviors several times a week were classified as bullies. Students who indicated suffering and perpetrating at least one of the behaviors several times a week were classified as bully-victims. The remaining students were considered not involved in bullying.

This procedure resulted in 36 (8.1 %) children categorized as victims, 22 (4.9 %) as bullies, 22 (4.9 %) were classified as bully-victim, and 366 were uninvolved. The prevalence rates were similar than those reported in previous studies from different countries (Copeland et al. 2013; Guzzo et al. 2014). Table 1 shows the distribution of bullying involvement according to gender and grade.

Table 1 Frequency of cyberbullying and traditional bullying victims for the entire sample, and according to gender and grade
Full size table

3.2 Testing H1, H2, and H3: Gender Differences in Gender Identity Measures

T-tests were done to examine the differences in the gender identity measures between girls and boys. The results are presented in Table 2. In comparison to girls, boys reported higher gender contentment, felt pressure and gender-typing through masculine personality traits. Girls reported more gender-typing through feminine personality traits than boys. No differences were found in gender typicality and intergroup bias.

Table 2 Gender differences in gender identity dimensions and gender-typed personality traits
Full size table

These results suggest that male students experienced more satisfaction with their own gender, and they self-attributed more masculine traits, but felt more pressure for gender conformity than female students. Female students self-attributed more feminine traits than boys.

3.3 Bivariate Correlations

The intercorrelations among child grade, gender identity measures and bullying involvement were calculated to preliminarily provide evidence for the construct validity of the multidimensional model of gender identity and the relationship between gender identity measures and bullying behaviors. Table 3 presents the correlations for both boys and girls. First, grade correlated significantly with felt pressure and feminine traits in girls, but correlated with intergroup bias and bullying perpetration for boys. Second for the correlations among gender identity dimensions, gender typicality correlated positively with gender contentment and intergroup bias for both boys and girls. Felt pressure related positively to the intergroup bias for both sexes. Third for gender-typed personality traits, masculine traits related negatively with gender contentment in girls and correlated positively with gender typicality in boys. Feminine traits related positively with gender typicality, gender contentment and intergroup bias in girls, but correlated negatively with gender typicality and positively with felt pressure in boys.

Table 3 Correlation matrix per gender among grade, gender identity dimensions, gender-typed personality traits and bullying behaviors
Full size table

Table 3 also provides a picture of the relationship between gender identity measures and bullying involvement. Bullying victimization related negatively to gender typicality and gender contentment for both boys and girls. For girls, but not for boys, victimization also correlated with felt pressure. Bullying perpetration related positively with intergroup bias and masculine traits for both boys and girls. Perpetration also related positively to felt pressure only for boys, and correlated negatively with feminine traits for girls Table 4.

Table 4 Multinomial logistic regression analysis of reports of bullying involvement for the whole sample
Full size table

3.4 Testing H4, H5, H6, and H7: Relationships Between Gender Identity Measures and Bullying Roles

Multinomial logistic regressions were performed to determine the associations between the psychosocial factors included in the study. In these analyses, the model for each bullying category (victims, bullies and bully-victims) was estimated in relation to the model of students who were not bullied and did not bully others, which was considered the reference group.

Multinomial regressions were first computed for the whole sample (see Table 4). The results indicated that neither gender nor grade predicted any bullying role. Among the gender identity measures, the results showed that the children who reported more gender typicality reported less victimization. Bullies were more likely to experience felt pressure to gender conformity and bully-victims were more likely to experience less contentment with their gender assignment. Among the gender typing measures, schoolchildren with a higher self-attribution score of masculine traits were more likely to be a bully or bully-victim. However, those who self-attributed feminine traits were more likely to be victims of bullying.

Multinomial regression analyses were conducted in the same way as above for boys and girls separately. The results (Tables 5 and 6) indicated that grade did not predict any bullying role for boys or girls. Regarding the gender identity measures, gender typicality was a protective factor for victimization for both boys and girls. The children of each gender who felt they were typical boys or typical girls were more likely to report less victimization. Other factors were found to be specific for one gender. In particular for boys, those who were bullies reported feeling significantly more pressure to gender conformity. Moreover, the boys experiencing less gender contentment reported greater involvement as bully-victims. For girls, those reporting higher intergroup bias were bully-victims.

Table 5 Multinomial logistic regression analysis of reports of bullying involvement for girls
Full size table
Table 6 Multinomial logistic regression analysis of reports of bullying involvement for boys
Full size table

Among the gender typing measures, feminine traits were a risk factor for victimization for boys, but not for girls. Boys reporting self-attribution of feminine traits indicated suffering more victimization. On the contrary, adherence to masculine traits was found to be a risk factor for both boys and girls. Children who self-attributed masculine traits were more likely to be bullies and bully-victims than those no involved.

4 Discussion

This paper presents school bullying data from a sample of 445 schoolchildren aged 10–12 years old. The aim of this study was to examine the gender effects on gender identity and gender typing measures, and to analyze how these measures contribute to involvement in each bullying category (victim, bullies, and bully-victims) in comparison to those students not involved in school bullying. By considering bullying an indicator of well-being, this study suggests both cross-cultural similarities and differences compared to those studies that analyzed multiple components of gender identity and their impact on psychological adjustment.

4.1 Gender differences in gender identity measures and gender typing measures in Spanish Children

Regarding measures from multidimensional gender identity, and in line with our hypothesis H1, Spanish boys tended to be more content about their gender ad felt more pressure to gender conformity than Spanish girls. These results confirmed that boys still experience more pressure for gender typing than girls, and this is a common phenomenon across different cultures (Egan and Perry 2001; Yu and Xie 2010). No gender differences were found in the children’s intergroup bias, which confirmed our hypothesis H2 and is consistent with previous research (Yu and Xie 2010). This finding can be related with the participant’s age since young children are especially prone to same-sex favoritism (Maccoby, 1998). No gender differences were found in gender typicality between boys and girls, a result which indicates that not only boys are gender typed. This result is not rare if we considered that it is important to be a typical girl/boy in middle childhood if children are to be accepted more by peers (Smith and Leaper 2005).

With regard to gender typing measures, boys identified themselves more with masculine traits, and girls did with feminine traits, which confirmed our hypothesis H3. These results, which are in line with previous findings, confirm that Spanish children share gender stereotypes and use these stereotypes as self-descriptors (Hoffman 2001; Navarro et al. 2011). This result might also be linked to lack of differences in gender typicality in the sense that during preadolescence, the boys and girls interacting in the same-sex group tend to exhibit traits associated stereotypically with each gender (Maccoby, 1998), and present themselves to others peers as typical members of their gender category.

4.2 Gender identity measures and bullying roles

The present study found significant relationships between Spanish children’s gender identity dimensions and bullying behaviors, which is consistent with previous findings that have shown associations between gender identity and psychological adjustment (Carver et al. 2003; Egan and Perry 2001; Corby et al. 2007). Consistently with our hypothesis H4, those children who did not report feeling different form same-sex peers (gender typicality) experienced less victimization. This relationship seemed stronger for girls than for boys. In any case, this result indicates that self-perception as a typical member of the same-sex peer group is important for the psychological well-being of preadolescents in terms of suffering less victimization, which is consistent with previous research (Drury et al. 2013). Nevertheless, although gender content was expected to be associated with less victimization, no relationship was found. This result supports the idea that although the four dimensions in Egan and Perry’s (2001) model are related, they are mostly independent of each other, at least in their associations with different indices of psychological adjustment.

Despite the lack of direct relationships between gender contentedness and victimization, we found an indirect relationship since the children who were bully-victims were less content with their gender. This is an interesting result that supports our hypothesis H5, and is in line with previous research which indicated that gender-dysphoric girls were termed by peers as being more aggressive, disruptive and argumentative than other girls (Carver et al. 2003). However, when we conducted the analyses separately for each gender, the association between gender content and the bully-victim role was significant only among boys. It can be argued that the bully-victim pattern of victimization and perpetration can be a response to dissatisfaction with own gender, and also a way to confront social pressure (e.g., victimization) to gender conformity. Nevertheless, our data are cross-sectional and the direction of casualty in this relationship remains unclear. Future research should address this question more carefully.

Felt pressure was found to be a risk factor for being a bully. Children who felt compelled to engage in gender-congruent behaviors were more aggressive with peers. We hypothesized with H6 that felt pressure is associated with specific gender-typed social behaviors: aggression for boys, victimization for girls. However, when we conducted the analyses separately for gender, the association between felt pressure and bullying remained significant only for boys. This partially supports our hypothesis and indicates that felt pressure may cause children to seek ways to fit in, and perpetration among boys may be a way to adhere more stringently to gender roles and to conform to social expectations (Eagly et al. 2004).

Surprisingly, a relationship between intergroup bias and the bully-victim role appeared among girls when we conducted the regression analyses separately for gender. This result supports our hypothesis H5 regarding a relationship between intergroup bias and bully or bully-victim roles. It is unclear, however, why this association appeared among girls. Intergroup bias has been theoretically related with aggressive behaviors after considering that negative attributes toward the other sex can increase hostility and dismiss respectful interaction with peers. Following this argument, intergroup bias could be related with cross-sex aggression. However, our data did not allow us to discover whether the intergroup effect bias on the bully-victim role was explained by a possible cross-sex conflict. This result calls for further examination as to how intergroup bias can be related with intergroup conflict as part of school bullying.

4.3 Gender typing measures and bullying roles

Hypothesis H7, which states that self-attribution of masculine traits is related with the role of bully or bully-victim, was supported. Boys and girls who are bullies or bully-victims perceived themselves with masculine traits. This is in line with previous findings which have shown that internalization of masculine stereotypical traits is a significant risk factor for getting involved in bullying perpetration (Gianluca and Pozzoli 2006; Young and Sweeting 2004). However, unlike other studies documented (Navarro et al. 2011), we did not find that self-attribution of feminine traits was a protective factor against involvement as a bully or bully-victim.

To also support our hypothesis H7, the results showed that self-adherence to feminine traits was a risk factor for victimization, which seemed clearer for boys. Previous research also found this same effect (Navarro et al. 2011) and explained this relationship as a result of stronger gender typification for males. According to this explanation, boys who fail to attach traditional views of masculinity, or are involved in cross-sex behavior or activities, can be rejected more.

In short, masculine traits have been seen to relate more strongly with bully perpetration for both genders, while feminine traits are related with victimization among boys. Although this approach of studying gender typing has been criticized because it assesses only gender typicality in terms of self-perceived personality traits and ignores that gender typing includes multiple domains, our results show that they are related with indicators of adjustment as bullying behaviors. Indeed, if we bear in mind the results of the correlational analyses, both masculine and feminine traits correlated positively with Egan and Perry’s gender typicality. So these two measures of gender identity can be complementary and are important for analyzing children’s well-being.

5 Limitations

This study has several limitations that should be noted. First, the analyses in this study are correlational and cross-sectional, and longitudinal studies should be conducted to know whether the risk or protective function of the gender measures tested holds in the long-term. Second, only self-reported data were collected in this study, and the relationships found in the current study might be influenced by response bias. Third, the sample was homogeneous in terms of age, which limits the extent to which these findings can be generalized to other age groups. Fourth, given the small number of participants in the bullying groups if compared to the not involved children, significant differences should be cautiously examined. However, even though this was an exploratory study, it contributes to the understanding of how gender development leads to the involvement in bullying behaviors. Conducting future research with larger samples that are more representatives of different ages is necessary to extend the present findings. Fifth, the construct validity of the gender measures was not systematically examined in this study. Consequently, cross-cultural equivalence cannot be assumed without further investigation of the factor-structure based on Spanish children. Future research that uses factor analyses to examine the construct of these models in Spanish children is required.

6 Theoretical and Practical Implications

The above findings have important practical and theoretical implications. Theoretically, the present study support the multidimensional model of gender identity proposed by Egan and Perry (2001) and is applicability to other cultures. Our results show that the four dimensions of gender identity were mostly independent and were differentially associated in the theoretically expected ways not only with victimization but also with perpetration within the peer group. It draws attention to the fact that gender identity development does not reduce to the unfolding of a single entity but rather involves the development of several components. However, in our study, despite the profound differences in conceptual assumptions and research methodologies, both the multidimensional gender identity and the gender typing perspective offers support to the theoretical hypothesis that gender typicality is associated with aggressive behaviors, especially for boys. In this regard, the present research shows the importance that studies will be designed to test competing hypothesis about explanations from theories on gender development involving social influences or cognitive development.

Practically, the present study suggests, as previous studies have done, that being a typical member of the same-sex peer group is important for psychosocial well-being in both boys and girls, at least in terms of suffering less victimization. In parallel, our findings indicate that, especially for boys, felt pressure for gender conformity may make them confront social expectations through ways that are harmful for their self-concepts (e.g., adopting an aggressive role) and for those who suffer their victimization. This implies that it is important for parents, educators and other professionals to show them ways to establish a sense of compatibility with one’s gender category.

Regarding self-attribution of gender-typed personality traits, the results reveal that internalization of gender cues is associated with risk behaviors, such as bullying perpetration and victimization. This implies that it is still important to raise awareness of influence of gender on identity formation in boys and girls. Socialization agents should provide children with other forms to confront peer pressures while offering spaces that are free of social expectation to explore cross-sex activities.

7 Conclusion

This study is unique in that it examines associations of multiple gender measures with different ways of involvement in bullying behaviors (victims, bullies and bully-victims). It is worth noting that gender measures relate differently to each bullying role. The findings highlight the importance of moving beyond the analysis of gender differences to analyze how gender variables are associated with the youths involved in school bullying through different roles, and not only as victims or perpetrators. The present study extends the bullying research database by opening up new directions to study antecedents of traditional forms of peer aggression.