People’s gender role attitudes—i.e., their views on the different roles men and women should take on in private and public life (Davis & Greenstein, 2009)—are an essential driving force of their mindsets and behaviors as well as broader societal and institutional change. For women, egalitarian gender role attitudes are linked to greater participation in the labor market (Corrigall & Konrad, 2007; Cunningham et al., 2005; Lendon & Silverstein, 2012), more educational attainment (Cunningham et al., 2005; Tallichet & Willits, 1986), and higher income (Corrigall & Konrad, 2007; Stickney & Konrad, 2007). In couples and families, egalitarian gender role attitudes are associated with joint decision-making and a more uniform distribution of power (Xu & Lai, 2002). At the societal level, culturally established traditional gender role attitudes are a crucial source of women’s subpar political and public representation (Alexander & Welzel, 2011; Paxton & Kunovich, 2003).

Previous research suggests that gender role attitudes tend to be malleable in childhood and adolescence (Sánchez Guerrero et al., 2023; Ullrich et al., 2022; Wilhelm et al., 2023) but stabilize more in adulthood (Kågesten et al., 2016; Lendon & Silverstein, 2012). Accordingly, a substantial body of research has investigated how gender role attitudes develop among children and adolescents, with a particular focus on the influence of parents. Unequivocally, this literature finds egalitarian parents to usually have egalitarian children, while traditional parents have more traditional children (see Davis & Greenstein, 2009; Halimi et al., 2016 for reviews). Although parents are the most influential figures in the childhood development of gender role attitudes, the set of actors shaping attitudes expands in adolescence. In particular, friends grow increasingly influential as children age and enter adolescence (Halimi et al., 2016; Witt, 2000). While parental influence frequently results in intergenerational continuity in gender role attitudes, friends can also be drivers of attitudinal change.

A potential for change in gender role attitudes resides in cross-gender friendships. As cross-gender friends provide perspectives and information on the other gender and as boys and girls, on average, hold gender role attitudes that differ to some extent (Halimi et al., 2016; Kågesten et al., 2016), the influence of cross-gender friends is especially likely to result in a change in gender role attitudes. However, cross-gender friends’ potential influence on adolescents’ gender role attitudes has received minimal attention in past research. Moreover, the few studies that empirically investigated the association of cross-gender friendships and gender role attitudes had no access to longitudinal data. Therefore, these studies could not ascertain that an observed association between attitudes and cross-gender friends indeed results from the influence of cross-gender friends rather than potential alternative explanations, like the possibility of specific gender role attitudes promoting or hampering the formation of cross-gender friendships. This shortcoming also applies to research on the link between cross-gender friendships and gender-related attitudes more broadly, which is also largely cross-sectional. An exception is Halim and colleagues’ (2021) study of how children’s attitudes about the other gender change in reaction to cross-gender friendships. Halim et al. (2021) find that both boys and girls develop more positive and less negative attitudes toward the other gender when establishing cross-gender friendships. Longitudinal investigations of whether cross-gender friends influence gender role attitudes in a similar way has been missing so far. In this study, I provide a longitudinal analysis of cross-gender friends’ influence on adolescents’ gender role attitudes, avoiding the limitations of previous research.

How Cross-Gender Friends May Influence Adolescents’ Gender Role Attitudes

Exposure theory suggests that exposure to situations characterized by specific ideas, attitudes, and behaviors fosters the development of these ideas, attitudes, and behaviors among those exposed (Davis & Greenstein, 2009; Hanish et al., 2005). Applied to the domain of gender role attitudes, this suggests that exposure to situations and ideas that resonate with egalitarian ideals is likely to foster the development of egalitarian attitudes. By contrast, exposure to traditional ideas and situations promotes traditional attitudes (Bolzendahl & Myers, 2004; Davis & Greenstein, 2009). In childhood, the prime sources of exposure are parents, and indeed, parents’ gender-related ideas and activities are decisive for the gender role attitudes their children develop (see Davis & Greenstein, 2009; Halimi et al., 2016). As children grow older, however, they become increasingly exposed to other contexts than the parental home and to other interaction partners than their parents. Particularly in adolescence, youth increasingly spend time with their friends and become exposed to friends’ ideas, attitudes, and behaviors. This exposure has been shown to shape young people’s attitudes and behaviors in various domains, ranging from the development of children’s aggressive behavior in reaction to exposure to aggressive peers (e.g., Hanish et al., 2005) to improvements in academic performance when surrounded by high achievers (e.g., Kretschmer et al., 2018). Similarly, exposure to peers with specific conceptions of gender roles may shape adolescents’ gender role attitudes.

For the development of gender role attitudes, friendships with cross-gender peers are likely to be of particular importance, as gender role attitudes concern relations between the genders. Cross-gender friendships can provide perspectives and information on these relations. There are two main mechanisms through which cross-gender friends may influence gender role attitudes: an adaptation to their cross-gender friends’ attitudes and a change in gender stereotypes.

Adaptation to Cross-Gender Friends’ Gender Role Attitudes

In general, adolescents are highly susceptible to peer influence, adapting to friends’ attitudes and behaviors in a wide range of domains (e.g., Brechwald & Prinstein, 2011). Recent findings suggest that such adaptation processes are also relevant in the domain of gender role attitudes. Adolescents assimilate both the gender role attitudes prevailing in their larger school peer group (Halimi et al., 2020; Sánchez Guerrero & Schober, 2021) and among their close friends (Sánchez Guerrero et al., 2023).

Two main processes can drive this adaptation to cross-gender friends’ attitudes. The first is an exchange of ideas and opinions (Bussey & Bandura, 1999; Jenkins et al., 2023). Mismatched gender role attitudes among cross-gender friends can spark discussions on the reasons for diverging opinions and a revision of attitudes. And even without explicit discussion, adolescents obtain insights into their cross-gender friends’ diverging attitudes through friends’ behavior (Bussey & Bandura, 1999; Davis & Greenstein, 2009). This more subtle exposure can also lead to an assimilation of cross-gender friends’ attitudes.

Furthermore, peer pressure can contribute to the adaptation to friends’ gender role attitudes (Bussey & Bandura, 1999; Cook et al., 2019; Halimi et al., 2021). In exclusively same-gender friendship circles, youth risk being rejected if they stand out in terms of gender-related characteristics, so the pressure to conform to same-gender friends’ gender role attitudes is strong (Keener et al., 2013; Maccoby, 1998; McHale et al., 2004). Cross-gender friends can alleviate this pressure from same-gender friends to some degree, opening up a broader set of acceptable attitudes (Halimi et al., 2021). At the same time, cross-gender friends can become another source of peer pressure themselves, urging friends to adapt to their attitudes.

The gender role attitudes of boys and girls tend to differ, as girls hold more egalitarian gender role attitudes than boys on average (see Halimi et al., 2016 for a review). Accordingly, an adaptation to cross-gender friends’ attitudes is likely to change adolescents’ attitudes. Boys, who, on average, hold more traditional attitudes than girls, are likely to move towards their female friends’ more egalitarian attitudes. Conversely, girls may shift towards their male friends’ more traditional attitudes. However, as discussed later in more detail, this traditionalization among girls may be less likely due to girls’ greater interest in egalitarianism (Davis & Greenstein, 2009; Sánchez Guerrero et al., 2023).

Changes in Gender Stereotypes

In addition to an adaptation to cross-gender friends’ attitudes, friendships across gender boundaries can also impact gender role attitudes by dispelling gender stereotypes. Gender stereotypes are overgeneralized ascriptions of characteristics to men and women (Ellemers, 2018; Kite et al., 2008). For example, women are expected to be warmer and more caring about others than men, and men are expected to be more assertive and competent than women (e.g., Ellemers, 2018; Hentschel et al., 2019; Prentice & Carranza, 2002). Gender stereotypes, in turn, legitimize traditional gender role attitudes: Stereotyped as warmer and more caring, women are considered better suited for raising children; stereotyped as more assertive and competent, men are considered more suited for the breadwinner role (Croft et al., 2015; Eagly & Mladinic, 1994; Ellemers, 2018).

Gender stereotypes exaggerate differences between men and women and underestimate variation within each gender group (Ellemers, 2018; Kite et al., 2008). Dismantling these stereotypes can be facilitated by disproving evidence (Hilton & von Hippel, 1996). Stereotypes about the other gender are more likely to change in mixed-gender than in same-gender friendship circles, as cross-gender friends can provide direct evidence that disproves stereotypes about the other gender (Karpiak et al., 2007; Maccoby, 1998; Marmion & Lundberg-Love, 2004). As gender stereotypes justify traditional gender role attitudes (Croft et al., 2015; Eagly & Mladinic, 1994; Ellemers, 2018), this revision of stereotypes in cross-gender friendships may induce more egalitarian attitudes. Similar predictions can also be derived from contact-theoretical approaches to intergroup relations (Allport, 1954). Contact theory suggests improvements in outgroup attitudes in reaction to outgroup friendships (Pettigrew & Tropp, 2006), partially because of a reduction in cross-group  stereotypes. Contact theory has primarily been applied to contact and attitudes between ethnic and racial groups so far, but has recently been proposed as a framework to study cross-gender relations as well (Hanish et al., 2023; Martin et al., 2017).

Although these considerations suggest changes in gender stereotypes and gender role attitudes due to cross-gender friendships, other processes may limit or block such change. First, not all cross-gender friendships provide disconfirming evidence about gender stereotypes. Though stereotypes exaggerate gender differences, there is also actual variation in reported traits by gender (e.g., in warmth and caring behavior; Ellemers, 2018; Kite et al., 2008). Therefore, not all cross-gender friendships provide stereotype-inconsistent information. Second, even though many theoretical accounts of gender development suggest a link between gender stereotypes and gender-related attitudes (e.g., Liben & Bigler, 2002; Martin et al., 2002), this connection needs not always be present. Gender stereotypes and gender role attitudes do not necessarily only reflect youths’ experiences with cross-gender peers but also their individual interests (Liben & Bigler, 2002). For example, if adolescent boys have an interest in upholding traditional gender role attitudes, they may distort counterstereotypical evidence from cross-gender interaction to avoid adjusting their gender stereotypes (Kunda, 1990; Martin & Halverson, 1983). Third, in a process of subtyping, cross-gender friends and the disconfirmation of gender stereotypes they provide may be considered an exception to the rule, preventing broader stereotype change (Deutsch & Fazio, 2008; Kunda, 1990; Rothbart & John, 1985). For these reasons, the efficacy of cross-gender friendships for the reduction of gender stereotypes is likely to be limited in some cases.

Why the Influence of Cross-Gender Friends May Differ Between Boys and Girls

In principle, the influence of cross-gender friends can affect the gender role attitudes of both boys and girls. However, boys’ and girls’ susceptibility to influence on their gender role attitudes is likely to differ. As mentioned previously, girls and women hold more egalitarian attitudes than boys and men on average (for reviews, see Davis & Greenstein, 2009; Halimi et al., 2020), a difference that primarily originates from gender differences in the interest in egalitarianism (Bolzendahl & Myers, 2004; Davis & Greenstein, 2009; Sánchez Guerrero et al., 2023). Traditional gender role attitudes exclude women from activities associated with high status and material security, particularly from participation in the labor market in higher-level jobs requiring full-time commitment. At the same time, traditional attitudes push women towards activities (such as child-rearing or housework) that come with less status and security (Croft et al., 2015). Consequently, the material and status benefits women accrue from gender egalitarianism are straightforward. By contrast, men’s interest in egalitarianism tends to be not as clear-cut. While egalitarianism alleviates pressure on men’s labor market performance and allows them greater involvement in the family (Croft et al., 2015), it also confronts them with additional housework and childcare responsibilities (Davis & Greenstein, 2009). Although the balance of these considerations still results in many men supporting gender egalitarianism, support continues to be less unequivocal than among women (Davis & Greenstein, 2009; Halimi et al., 2016).

This gender-specific interest in gender egalitarianism is also likely to affect how susceptible boys and girls are to the influence others can exert on their gender role attitudes. Given the material and status advantages egalitarian gender roles hold for women, many girls strongly embrace egalitarianism and reject traditional attitudes (Davis & Greenstein, 2009; Halimi et al., 2016; Kågesten et al., 2016). Under these conditions, an interaction with more traditional cross-gender friends is unlikely to convince them of a more traditional ideology. Similarly, the fact that traditional girls forgo these benefits by embracing more traditional roles suggests strong counterbalancing convictions among them (such as strong religious beliefs) that are likely to let them resist attitudinal change. This is different for boys, who may consider the benefits (and costs) of egalitarianism as more ambiguous and generally lower. Given these lower stakes, boys may be more open to changing their gender role attitudes. For the same reason, they may less consciously decide on their attitudes and more readily revise them in response to outside influence. This suggests an asymmetric adaptation to cross-gender friends’ gender role attitudes: Although boys may assimilate girls’ more egalitarian gender role attitudes in cross-gender friendships, girls are less likely to adapt to the more traditional gender role attitudes of their male friends.

Cross-gender stereotypes, however, prevail among both genders (Ellemers, 2018; Hentschel et al., 2019). In both boys and girls, stereotypes may thus diminish due to disconfirming evidence obtained in cross-gender friendships. This does not necessarily have the same consequences on gender role attitudes for both genders, though. Despite their gender stereotypes, many girls embrace highly egalitarian gender role attitudes, meaning that a further change in attitudes due to diminishing gender stereotypes is hardly possible for them in many instances. This is different among boys, who tend to hold more traditional attitudes, so changes in stereotypes are more likely to result in changes in their gender role attitudes.

Following these considerations, cross-gender friends are likely to shift the gender role attitudes of boys in a more egalitarian direction. In contrast, among girls, the influence of cross-gender friends is expected to be lower and may even be negligible. In line with these expectations on gender-specific influence effects, an early study by Bryant (2003) found more egalitarian gender role attitudes among young men with cross-gender friends but no variation in attitudes according to cross-gender friendships among young women. The few other studies investigating the link between cross-gender friends and gender role attitudes, however, reported different patterns. McHale et al. (2004) found no variation in gender role attitudes according to the time adolescent boys and girls spend with cross-gender peers. Halimi et al. (2021) documented more traditional gender role attitudes for girls with cross-gender friends but no variation for boys, and Perez-Brena et al. (2015) found more egalitarian attitudes among boys who spent more time with cross-gender friends but more traditional attitudes among girls. However, as the authors acknowledge, these studies are limited to demonstrating an association between cross-gender friends and gender role attitudes and cannot separate cross-gender friends’ influence from other processes that may induce this association.

A Threat to the Identification of Influence Effects: Selecting Cross-Gender Friends Based on Gender Role Attitudes

The most significant risk to mistakenly interpreting the association between cross-gender friendships and gender role attitudes as an influence of cross-gender friends is the fact that this empirical link can also emerge from a second process: the selection of cross- versus same-gender friends depending on gender role attitudes. Accordingly, empirically accounting for friendship-making processes based on gender role attitudes is essential to ensure that cross-gender influence is not confused with selection. A selection of cross-gender friends based on gender role attitudes can occur for various reasons (Kalmijn, 2002). For example, boys with more traditional gender role attitudes may shun cross-gender friends because they fear or experience incompatibilities in friendships with girls, who tend to be more egalitarian. Similarly, traditional girls may avoid friendships with boys because they believe in inherent gender differences that complicate close cross-gender interactions. Accounting for these selection processes is essential to avert the risk of obtaining inaccurate estimates of influence effects.

Identifying Cross-Gender Friends’ Influence with Stochastic Actor-Oriented Models

To estimate the influence of cross-gender friends on the development of adolescents’ gender role attitudes while accounting for selection effects, I employ stochastic actor-oriented models (SAOMs) for the coevolution of friendship networks and gender role attitudes (Ripley et al., 2023; Steglich et al., 2010). Using longitudinal data and agent-based simulation techniques, SAOMs simultaneously model two processes over time: the development of adolescents’ friendship networks and the development of their gender role attitudes. This allows SAOMs to assess both the driving forces of adolescents’ friendship-making and the mechanisms behind changes in their gender role attitudes. Crucially for the application at hand, this means that SAOMs can simultaneously model the effect of gender role attitudes on cross-gender friendship-making and the effect of cross-gender friends on gender role attitudes. Therefore, SAOMs can decompose the association between adolescents’ cross-gender friendships and gender role attitudes into the selection of cross-gender friends based on gender role attitudes and the impact of cross-gender friends on their gender role attitudes. As the latter effect is controlled for friendship selection, it represents the actual influence of cross-gender friends on gender role attitudes.

To accurately model the evolution of friendship networks and gender role attitudes and to provide unbiased estimates of cross-gender friends’ influence, the SAOMs further control for adolescents’ religion, religiosity, ethnic background, and socioeconomic status. In line with the observation that similarity breeds attraction (McPherson et al., 2001), past research has shown that sharing a religious affiliation (Leszczensky & Kretschmer, 2022; Leszczensky & Pink, 2017) or ethnic background (Leszczensky & Pink, 2015; Smith et al., 2014) as well as similar religiosity (Cheadle & Schwadel, 2012) and socioeconomic status (Malacarne, 2017) facilitates friendship-making. At the same time, some of these characteristics have been linked to adolescents’ gender role attitudes. Lower socioeconomic status (Halimi et al., 2016) and higher religiosity (Kretschmer, 2018) are associated with more traditional gender role attitudes. More traditional attitudes have also been reported in the ethnoreligious minority of Muslim youth (Kretschmer, 2018; la Roi & Mood, 2022). Accounting for variation in the development of gender role attitudes according to these characteristics is particularly important in cases where these characteristics are also related to cross-gender friendship-making. This has been suggested for Muslim youth, who sometimes face norms that complicate cross-gender interaction (Altinyelken, 2022; Basit, 1997; McGrath & McGarry, 2014). If these norms prevent cross-gender friendships and Muslim youth also hold more traditional gender role attitudes than non-Muslim youth, not accounting for these attitudinal differences can bias estimates of cross-gender influence effects. Links between cross-gender friendship-making and religiosity and socioeconomic background have not been demonstrated in previous research, but to avoid any risk of confounding, I still control for these characteristics in the analyses.

The Current Study

The aim of this study is to investigate whether cross-gender friends influence the development of adolescents’ gender role attitudes. Cross-gender friends can offer unique perspectives, insights, and information about another gender. They can influence gender role attitudes by both an adaptation to cross-gender friends’ attitudes and a reduction of gender stereotypes.

A key question on cross-gender friends’ influence concerns potential gender differences. Given their greater interest in egalitarian gender roles, girls are less likely to adopt more traditional attitudes in their interaction with cross-gender friends than vice-versa. There may also be no room for further changes in girls’ gender role attitudes in response to a reduction in gender stereotypes. Consequently, the influence of cross-gender friends on gender role attitudes is expected to be limited or even negligible among girls, but more pronounced among boys.

A limited number of previous studies have investigated the association between cross-gender friends and gender role attitudes. However, data and methodological limitations prevented them from drawing definite conclusions on influence effects. This study avoids these limitations by investigating cross-gender friends’ influence with large-scale longitudinal network survey data and stochastic actor-oriented models (SAOMs) for the coevolution of friendship networks and gender role attitudes.

Method

Data

For the empirical analysis, I use German data from the first two waves of the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU; Kalter et al., 2016a, b, 2019). CILS4EU data collection was based on a random sample of schools, oversampling schools with a high proportion of ethnic minority students. The first wave of data was collected in 2010 and 2011; the second wave followed the next year. Within schools, all students from two randomly selected ninth-grade classrooms were surveyed. In the first wave, students were 14–15 years old on average; in the second wave, they were 15–16 years old.

In both waves, the student survey contained a sociometric questionnaire providing information on students’ friendships with classmates. In total, 5,013 students from 271 classrooms participated in the first wave. To ensure that selection and influence processes could be modeled appropriately, network data had to be available in both the first and second wave for classrooms to be included in the analysis. In 71 classrooms (mostly from lower secondary schools), students completed schooling between the first and second waves, so no longitudinal friendship network data is available. These classrooms were excluded from the analysis. I also excluded 51 classrooms with a student non-response of more than 50% in at least one of the waves. Though longitudinal social network models can accommodate missing data to some extent, a sufficiently large participation rate is necessary to capture friendship networks accurately (Huisman & Steglich, 2008). Results are also robust to stricter inclusion criteria for classroom networks (i.e., non-response rates of 70% and 80%). Applying these restrictions, 3,194 students from 149 classrooms remain in the analytical sample. I show that classroom demographic characteristics, gender role attitudes, cross-gender friendships, and the link between both are similar in the full and analytical sample (see Supplement A in the online supplement for detailed information on this analysis).

Measures

Friendship Networks

To capture adolescents’ friendship networks, I use information from the sociometric questionnaire. From a list of all students in their classroom, students could nominate up to five best friends. Because all students within a classroom were surveyed, individual-level information on all adolescents as well as their friends is available, including their gender and gender role attitudes. On average, students nominated 3.83 friends in the first and 3.71 in the second wave.

Gender

Students self-reported whether they consider themselves a boy or a girl. A more differentiated measure of gender and gender identity is not available, a limitation I return to in the discussion section. Information on students’ self-reported gender is complete. In total, 51% of the students identified as girls and 49% identified as boys.

Gender Role Attitudes

To measure gender role attitudes, I use four items on the preferred division of labor between men and women in a family. Respondents were asked, “In a family, who should do the following?” for four tasks: taking care of children, cooking, cleaning, and earning money. For each task, respondents could indicate whether the task should be done “mostly by the woman,” “mostly by the man,” or shared between both partners (“both about the same”). For childcare, cooking, and cleaning, I classified attitudes as traditional if students primarily allocated the task to the woman and as egalitarian when they indicated otherwise. For earning money, I classified attitudes as traditional if respondents mainly assigned the task to the man, and as egalitarian when they indicated otherwise. This also classifies cases as egalitarian in which respondents indicated that the man should primarily do the cooking, cleaning, or childcare, or the woman should primarily earn the money. The proportion of answers in this category is negligible, though (0.8–2.1% across the four tasks assessed). As a measure of egalitarian gender role attitudes, I use the number of egalitarian answers ranging from 0 to 4. The scale’s internal consistency is acceptable with a Cronbach’s α of 0.69, and a principal component analysis also suggests a single-component solution. This operationalization of gender role attitudes corresponds to that of previous studies using the CILS4EU data (e.g., Kretschmer, 2018; la Roi & Mood, 2022).

Control Variables: Religion, Religiosity, Ethnic Background, and Socioeconomic Status

Given previous findings on more traditional gender role attitudes among Muslim youth and norms that complicate their cross-gender friendship-making, I account for students’ religious affiliation by differentiating between students who identify as Muslim and non-Muslim. Students indicated their religious affiliation by selecting from a list of the most frequent affiliations or providing an open-ended answer. I capture adolescents’ religiosity with information on their frequency of prayer. Frequency of prayer was assessed on a 6-point scale, with students indicating that they pray “never” (0), “less than once a month” (1), “less than once a week” (2), “once a week or more” (3), “every day” (4), and “at least five times a day” (5).

I capture ethnic background with data on the country of birth of students, their parents, and their grandparents. Students are considered to have German origin if they, their parents, and their grandparents have been born in Germany. Otherwise, I assign their ancestors’ country of birth, following standard procedures for the CILS4EU data (Dollmann et al., 2014). To capture socioeconomic status, I rely on information on parents’ occupational status measured on the ISEI scale. If available, I use data from the parental CILS4EU survey; otherwise, I use information provided by the adolescent respondent. I use the higher of both parents’ occupational status to measure socioeconomic status.

Analysis Strategy

To investigate the influence of cross-gender friends on adolescents’ gender role attitudes, I rely on stochastic actor-oriented models (SAOMs) for the coevolution of friendship networks and gender role attitudes (Snijders et al., 2010). Based on empirically observed friendship networks and gender role attitudes in two waves of data, these SAOMs simultaneously model the evolution of friendship networks and gender role attitudes over time. Through agent-based simulation, SAOMs decompose the change in networks and gender role attitudes observed between the time points into single changes in friendship ties and attitudes by individual students. These changes are modeled in so-called mini steps. In a network mini step, a (randomly selected) student can form a friendship tie, remove a tie, or leave their network unchanged. In an attitude mini step, a (randomly selected) student can increase their number of egalitarian gender role attitudes by one, decrease it by one, or leave attitudes unchanged.

These changes depend on behavioral tendencies for the development of friendship networks and gender role attitudes specified in the SAOM model specification, as well as the coefficients associated with these tendencies. Behavioral tendencies can represent various processes related to friendship-making and attitudinal development. Of greatest interest to this paper’s research question is cross-gender friends’ influence, i.e., the tendency to change gender role attitudes in reaction to having cross-gender friends. At the same time, it is essential to account for the selection of cross-gender friends, i.e., the tendency for cross-gender friendships to evolve or dissolve for students in conjunction with egalitarian attitudes, to ensure that influence effects are correctly estimated. Next to the selection and influence of cross-gender friends, I also include a variety of other behavioral tendencies to accurately capture the evolution of friendship networks and gender role attitudes over time, as discussed in more detail below.

Multilevel Random-Coefficients SAOM Analysis

To estimate the evolution of friendship networks and gender role attitudes across the 149 friendship networks in the sample, I rely on random-coefficient multilevel SAOMs (Koskinen & Snijders, 2023). Multilevel SAOMs estimate joint effects across all networks, so complex SAOMs can be estimated though each classroom network is small (Koskinen & Snijders, 2023; Ripley et al., 2023). Using random effects, variation in coefficients at the network level can be modeled as well. In all models, I treat structural network effects as well as the effects of control variables as random effects. I treat all effects that relate to gender role attitudes, the key theoretical interest of this study, as fixed effects to ensure sufficient precision in these estimates. This aligns with previous applications of multilevel random-coefficients SAOMs in the literature (e.g., Boda, 2018; Kretschmer & Leszczensky, 2022; Raabe et al., 2019). Multilevel random-effects SAOMs rely on a Bayesian estimation technique that models sequences of change in the friendship network and gender role attitudes between observation periods, implemented in the multiSiena extension of the RSiena package (Version 1-3.31) in R. SAOMs impute missing data internally with plausible values, ensuring that the impact of missing values on parameter estimation is minimized (Huisman & Steglich, 2008; Ripley et al., 2023). All models reported converge according to standard convergence criteria for multilevel random-coefficients SAOMs (Ripley et al., 2023). See Supplement B in the online supplement for more details on multilevel random-coefficients SAOMs, the choice of priors for Bayesian estimation, the specification of random effects, and convergence assessment.

Model Specification: The Evolution of Gender Role Attitudes

Regarding the evolution of gender role attitudes, the greatest substantive interest lies in whether cross-gender friends induce change in adolescents’ gender role attitudes over time. To capture the influence of cross-gender friends, I model how the evolution of gender role attitudes varies with adolescents’ proportion of cross-gender friends, which is captured with the avXalt effect in the SAOMs. I model this influence of cross-gender friends separately for boys and girls (with a corresponding avXalt effect for both genders). In line with recommendations from the literature on longitudinal social network models (Ripley et al., 2023), I further capture the general intertemporal development of gender role attitudes with a linear and quadratic shape effect. As discussed above, I also account for variations in the evolution of gender role attitudes between boys and girls, between Muslim and non-Muslim youth, and by adolescents’ religiosity and socioeconomic status (see Supplement B in the online supplement for details on these effects).

Model Specification: The Evolution of Friendship Networks

In this study, the most critical function of modeling the evolution of not only gender role attitudes but also friendship networks is to control whether students’ tendency to make same- or cross-gender friends depends on their gender role attitudes. I capture same- and cross-gender friendship-making among boys and girls with three effects: the ego girl, alter girl, and ego girl x alter girl effects. Jointly, these effects map the selection of same- and cross-gender friends and its variation between boys and girls (see Supplement B in the online supplement for a detailed derivation). To capture how selection differs by gender role attitudes, I interact all three effects with the ego gender role attitudes effect.

To accurately depict the evolution of friendship networks, I additionally model the selection of friends with the same religious affiliation, same ethnic background, similar socioeconomic status, and similar religiosity. I also account for a set of structural network-related processes well-known to shape the evolution of friendship networks (Ripley et al., 2023). This includes the tendency to reciprocate friendships (reciprocity effect), the tendency to become friends with one’s friends’ friends (transitive triplets effect), and the interaction of both (transitive reciprocated triplets effect). Finally, I model dispersion in the number of friendships sent and received, as well as the correlation of both, with the outdegree-activity, indegree-popularity, and indegree-activity effects, and consider the overall density of the friendship network with the outdegree effect. I provide more detailed information on these control effects in Supplement B in the online supplement.

Simulations Based on SAOM Results

The estimates from SAOMs are multinomial logit coefficients and, thus, are hard to interpret substantively beyond their sign and statistical significance. To provide a more intuitive assessment of effect sizes, I complement the SAOM results with a simulation analysis. For the simulation, I combine the SAOM estimates with the friendship networks and gender role attitudes observed in the first wave of data to generate simulations of the distribution of gender role attitudes in the second wave. In one set of simulations, I use the exact estimates from the SAOM model; in a second set, I set the estimates of cross-gender friends’ influence on gender role attitudes to zero. A comparison of the attitudes predicted for the second wave in both simulations thus shows how cross-gender friends contribute to the evolution of gender role attitudes. For all simulation analyses, I report averages across 1,000 separate simulations for each classroom to ensure the stochastic nature of the simulations averages out.

Results

Descriptive Results: Gender Role Attitudes, Cross-Gender Friendships, and Their Association

Figure 1 displays the distribution of gender role attitudes separately for boys and girls for the first wave of data. In line with previous findings, Fig. 1 shows that girls hold more egalitarian gender role attitudes than boys on average. Although boys, on average, have a score of 2.11 (of a maximum of 4) on egalitarian attitudes, girls have an average score of 2.61 on egalitarian attitudes, a difference that is statistically significant (p < .001).

Fig. 1
figure 1

Distribution of Gender Role Attitudes Among Boys and Girls

Full size image

Compared to their same-gender classmates, adolescents tend to rarely engage in friendships with their cross-gender peers. In the sample, only 12.1% of all friendships were cross-gender in the first wave. This proportion hardly differs between adolescent boys and girls, with girls nominating cross-gender peers in 12.0% of their friendships and boys nominating cross-gender friends in 12.3% of their friendships. 31.4% of both boys and girls nominate at least one cross-gender friend in their classroom.

Figure 2 provides a descriptive assessment of the link between cross-gender friendships and gender role attitudes based on data from the first wave. Figure 2 shows boys’ and girls’ predicted gender role attitudes at different proportions of cross-gender friends based on a linear regression model. This descriptive analysis suggests a gender-specific link between cross-gender friends and gender role attitudes. Among girls, gender role attitudes are independent of cross-gender friendships. By contrast, a higher proportion of cross-gender friends is associated with more egalitarian gender role attitudes among boys, a link that is statistically significant (p < .01). Predictions for gender role attitudes become increasingly imprecise at higher proportions of cross-gender friends, which is not surprising given that cross-gender friendships are rare. Substantive results are also very similar when only comparing adolescents with no cross-gender friends to adolescents with at least one cross-gender friend. These descriptive findings are in line with theoretical expectations of a stronger impact of cross-gender friends on boys’ compared to girls’ gender role attitudes. However, they only reflect cross-sectional associations and thus do not allow for the separation of influence from selection effects.

Fig. 2
figure 2

Gender Role Attitudes by the Proportion of Cross-Gender Friends Among Boys and Girls

Full size image

SAOM Results: Separating Selection and Influence Effects

To account for selection effects and provide actual estimates of cross-gender friends’ influence on adolescents’ gender role attitudes, I next turn to results from the SAOM analysis. In Fig. 3, I display estimated selection and influence effects separately for boys and girls. Complete model results, which show that estimates for control variables align with expectations, can be found in Table S1 in Supplement C in the online supplement.

Fig. 3
figure 3

Selection of Cross-Gender Friends Based on Gender Role Attitudes and Influence of Cross-Gender Friends on Gender Role Attitudes From Random-Effects Multilevel SAOMs

Full size image

The left part of Fig. 3 addresses selection effects. It provides estimates for the effect of egalitarian gender role attitudes on adolescents’ tendency to have cross-gender friends. The results suggest no systematic selection of cross-gender friends based on gender role attitudes, neither among boys nor girls (p > .2 for both estimates).

The right part of Fig. 3 displays estimates of cross-gender friends’ influence on gender role attitudes, net of selection. For girls, the results indicate no influence of cross-gender friends on gender role attitudes, with an estimate that is both close to zero and far from statistically significant (b = 0.07, p > .2). By contrast, there is strong evidence that cross-gender friends influence boys’ gender role attitudes. After controlling for selection, cross-gender friendships come with more egalitarian attitudes among adolescent boys (b = 0.71, p < .01). The gender difference in the influence of cross-gender friends is also statistically significant, with cross-gender friends more strongly influencing the gender role attitudes of boys than of girls (b = 0.64, p < .05). Figure S7 and Table S2 in Supplement C in the online supplement show that conclusions on gender-specific patterns of cross-gender friends’ influence are very similar when distinguishing between adolescents who have no cross-gender friends or at least one cross-gender friend. In these analyses, boys who have cross-gender friends develop more egalitarian gender role attitudes over time, while girls with cross-gender friends do not display any shifts in their attitudes.

Cross-Gender Friends and the Development of Egalitarian Gender Role Attitudes: A Simulation Analysis

To go beyond the multinomial logit estimates provided in Fig. 3 and to demonstrate the substantive size of cross-gender friends’ influence on adolescents’ gender role attitudes, I next turn to a simulation analysis based on the SAOM results. Combining the SAOM estimates with the observed friendship networks and gender role attitudes from the first wave of data, I simulate distributions of gender role attitudes in the second wave for two scenarios: a baseline influence scenario, in which simulations are based on the exact SAOM estimates from Fig. 3; and a no influence scenario, in which simulations assume all cross-gender friends’ influence to be absent.

For both scenarios and separately for boys and girls, Fig. 4 shows how the distribution of gender role attitudes changes between the first and second wave. Positive values indicate gender role attitudes that become more frequent from the first to the second wave; negative values indicate gender role attitudes that become less frequent over time. The only difference between the scenarios is the absence of cross-gender friends’ influence in the no influence scenario. Therefore, comparing both shows the contribution of cross-gender friends’ influence on the evolution of attitudes over time.

Fig. 4
figure 4

Predicted Change in Gender Role Attitudes From Simulations Based on Baseline Influence and No Influence Scenario Among Boys and Girls

Full size image

Figure 4 shows that, independent of the scenario, attitudes become more egalitarian over time. Among both boys and girls, the proportion of adolescents who indicate no or only a single egalitarian attitude decreases between the first and second wave. In contrast, the proportion with three or four gender-egalitarian attitudes increases. The intertemporal change is more substantial among girls.

At the same time, there clearly are discrepancies between the baseline influence and the no influence scenario for boys, in line with the significant influence effects estimated in the SAOM analysis. In the no influence scenario, only about 1.1% of boys shift towards fully egalitarian attitudes between waves, while 3.5% do so in the baseline influence scenario. Correspondingly, boys’ shift from traditional attitudes is notably more substantial in the baseline than in the no influence scenario. Although a change of 2.4% points due to cross-gender friendships may appear moderate at first sight, it is essential to be aware that this shift is driven by a small number of friendships – only 12% of boys’ friendships are cross-gender. For girls, predicted change hardly differs between the baseline and no influence scenarios, in accordance with the null effect of cross-gender friends on attitudes estimated in the SAOM.

Discussion

In this study, I used large-scale network survey data and longitudinal network-analytical methods to assess the influence of cross-gender friends on adolescents’ gender role attitudes. In the empirical analyses, I found that boys with cross-gender friends develop more egalitarian gender role attitudes over time. By contrast, girls’ involvement in same- and cross-gender friendships proved independent of their gender role attitudes. At the same time, there was no evidence that either boys’ or girls’ selection of same- over cross-gender friends was driven by their gender role attitudes. Although cross-gender friendships were notably rarer than same-gender friendships in general, this pattern did not vary with adolescents’ gender role attitudes.

The influence of cross-gender friends on adolescent boys’ gender role attitudes is likely to reflect first-hand insights into the perspectives, characteristics, and attitudes of the other gender that these cross-gender friendships provide. Influence may be driven both by an adaptation to cross-gender friends’ gender role attitudes and by a reduction of gender stereotypes due to the disconfirming evidence cross-gender friendships provide. The gendered nature of these influence effects is likely to be due to gender differences in the conception of egalitarian gender role attitudes: As established in past research, girls and women hold more egalitarian gender role attitudes than boys and men on average, mostly because of their more steadfast interest in egalitarianism (Davis & Greenstein, 2009; Sánchez Guerrero et al., 2023). This interest is likely to also shield girls from adapting to boys’ more traditional gender role attitudes in cross-gender friendships. Furthermore, girls’ highly egalitarian gender role attitudes lower opportunities for further attitudinal change over time, even when gender stereotypes are dismantled in cross-gender friendships. Accordingly, it is not surprising that cross-gender friends’ influence on gender role attitudes affects adolescent boys but not girls. This gendered pattern is also mirrored in recent findings by Jenkins et al. (2023), who detected lower sexism among young men with cross-gender friends but no variation in sexism according to the gender composition of young women’s friendship networks.

Methodologically, this study avoided the limitations of previous research on the link between cross-gender friendships and gender role attitudes by applying stochastic actor-oriented models (SAOMs) to large-scale longitudinal data on adolescents’ friendship networks and attitudes. By simultaneously modeling the development of friendship networks and gender role attitudes, SAOMs allowed me to disentangle the selection of cross-gender friends based on gender role attitudes from cross-gender friends’ influence on these attitudes. Accordingly, estimates in this study represent cross-gender friends’ influence net of selection processes. Differentiating these mechanisms was not possible in the few previous studies on the association between cross-gender friendships and gender role attitudes, which may be one reason for their inconclusive findings (Bryant, 2003; Halimi et al., 2021; McHale et al., 2004; Perez-Brena et al., 2015).

Even beyond the domain of gender role attitudes, longitudinal studies of cross-gender friends’ influence on gender-related attitudes have been rare. A notable exception is a recent study by Halim et al. (2021), who used longitudinal data to identify the effect of cross-gender friends on changes in children’s positive and negative attitudes toward the other gender. Halim et al. (2021) also detected influence effects, finding cross-gender friends to induce more positive and less negative attitudes toward the other gender. Other than this study, however, Halim et al. (2021) found similar influence effects among boys and girls. One reason for this difference may be the general nature of the cross-gender attitudes that Halim et al. (2021) studied compared to the more specific gender role attitudes investigated here. As argued above, the gendered effects I found likely reflect the restrictive nature of traditional gender role attitudes for women and the steadfast egalitarianism among girls associated with it. Gender-specific conceptions may be less widespread for the more general cross-gender attitudes assessed by Halim et al. (2021), leading to similar influence effects among boys and girls. However, given differences in the samples, methods, and contexts investigated in these studies, other explanations for the different findings are also conceivable.

Limitations and Future Research Directions

Although this study provides important new insights into cross-gender friends’ influence, it also has limitations and raises questions for future research. First, although the findings clearly show that adolescent boys develop more egalitarian gender role attitudes in reaction to cross-gender friendships, they do not elucidate the mechanisms responsible for this change. I discussed two main candidates: an adaptation to (cross-gender) friends’ attitudes on the one hand; and a reduction of gender stereotypes due to the exposure to disconfirming evidence in cross-gender friendships on the other. The exact contribution of both mechanisms to cross-gender friends’ influence could not be assessed empirically, though. Therefore, determining their role in cross-gender friends’ impact on gender role attitudes and assessing further potential explanations is a key task for future research.

Second, limitations concern issues of measurement. On the one hand, the CILS4EU data only contained one question on adolescents’ gender, asking them whether they are a boy or a girl. Therefore, differentiating adolescents’ sex and their gender identity as well as capturing non-binary identities was not possible. This is a limitation in this study not only due to measurement error in general but also because gender identity may be linked both to adolescents’ gender role attitudes and to their openness to cross-gender friendships. In future research, adolescents’ gender identity should be captured in a more explicit and differentiated manner to provide a better understanding of how gender role attitudes and friendship-making vary with gender identity.

On the other hand, the gender role attitudes measured in the CILS4EU data refer to the division of labor in families, which usually affects adult family members more than adolescents. A limited salience of concerns about the division of household labor may be one reason for the low reliability of the gender role attitudes scale (Cronbach’s α = 0.69), which could not be traced back to any of the specific items. For adolescents, the co-occurrence of certain household tasks may be less obvious than for adults, which may have led to lower consistency in the responses across items. This may have been further exacerbated by the fact that each item only allowed for a rough differentiation into three answer categories. If the household division of labor indeed is less salient for adolescents, cross-gender friends’ influence on these gender role attitudes may also be limited in adolescence. In particular, gender stereotypes may rarely focus on gendered household roles, so reductions in stereotypes due to cross-gender friendships may hardly affect corresponding attitudes. Cross-gender friends’ influence may differ for gender role attitudes that more directly apply to adolescents, such as attitudes toward gendered access to higher education or gendered fields of study. Influence effects may also differ when other gender-related attitudes (beyond gender role attitudes) are investigated, as exemplified by the similar effects of cross-gender friends among boys and girls on the more general positive and negative cross-gender attitudes documented by Halim et al. (2021).

Third, the CILS4EU data only capture adolescents’ friendships within the classroom context. Although adolescents in Germany spend most of their in-school time within these fixed classroom contexts, they also make friends with schoolmates from other classrooms and with peers outside of school. Accordingly, classroom friendship networks are bound to miss some of their cross-gender friendships. Therefore, the gender composition of classroom friendship networks is not necessarily representative of the gender distribution among the totality of adolescents’ friends. Future research should obtain a more comprehensive understanding of adolescents’ cross-gender friends beyond the classroom and their influence on gender role attitudes.

Fourth, measurement of cross-gender friendship-making in the CILS4EU data is not only constrained in terms of context, but also in terms of time. Specifically, this study has provided an assessment of how adolescents’ cross-gender friendships and gender role attitudes develop over one year rather than a long-term perspective. Due to resistance to stereotype and attitude change (Kunda, 1990; Martin & Halverson, 1983), changes frequently require a variety of experiences that challenge preconceptions, thus demanding a considerable investment of time. Therefore, long-term cross-gender friendships may be even more effective in changing gender role attitudes. Furthermore, the short time frame of this study also implies that only a snapshot of adolescents’ friendship networks is captured. Over the course of adolescence, friendship networks change considerably, and cross-gender friendships become increasingly common as adolescence progresses (e.g., Strough & Covatto, 2002). At the same time, adolescents’ cross-gender friendships may be more fragile than their same-gender friendships, and cross-gender friends’ influence on gender role attitudes may also vanish if these relationships dissolve over time. Comprehensively determining cross-gender friends’ influence on gender role attitudes in the face of friendship networks’ changing composition thus calls for a longer-term perspective in future research.

Finally, the data analyzed in this study were collected in the years 2010–2012. Since then, societal discourse on gender egalitarianism has shifted further and, at the same time, the rise of social media has created room for new types of social relationships among adolescents, including cross-gender relationships. Still, there is less reason to expect that the link between cross-gender friendships and gender role attitudes has changed substantively, as it is shaped by basic social influence processes like communication, peer pressure, and stereotype reduction. It is conceivable that cross-gender friends’ influence may be less directly reflected in actual attitudinal change in newer data if increasingly egalitarian attitudes induce a ceiling to influence effects. From that perspective, older data from populations with more varied attitudes may even have an advantage in identifying the presence of and mechanisms behind influence effects. Still, to fully uncover how both gender role attitudes and the social processes generating them have evolved, studying cross-gender friends’ influence on gender role attitudes with newer data in future research is desirable.

Practice Implications

The empirical results from this study suggest that, for boys, cross-gender friendships hold the potential to shift gender role attitudes in a more egalitarian direction. By contrast, girls’ gender role attitudes remain unchanged in cross-gender friendships, so the development of more egalitarian attitudes among boys is not accompanied by a simultaneous re-traditionalization of girls. At the same time, cross-gender friendships generally prove rare among young people. Whenever establishing more egalitarian gender role attitudes among boys is a priority, supporting the development of cross-gender friendships in school thus appears as a promising strategy. Further benefits of cross-gender friends are also suggested by other recent studies that found young men with more cross-gender friends to hold less sexist attitudes (Jenkins et al., 2023) and children to develop more positive and less negative attitudes toward the other gender in reaction to cross-gender friendships (Halim et al., 2021).

At the same time, the school context provides educators with opportunities for fostering positive cross-gender interaction and cross-gender friendships. For example, tasks assigned to mixed-gender groups can provide a structured setting supporting cooperation and more positive attitudes across gender boundaries (Hanish et al., 2023). General social and communication skills training may also increase boys’ and girls’ awareness of each other’s preferences and prevent cross-gender conflicts that can emerge from gendered communication or interaction styles (Mehta & Strough, 2009). Furthermore, curricula focused on equality, diversity, and inclusion may not only directly affect gender stereotypes and gender role attitudes but also have indirect effects by supporting the formation of cross-gender friendships and their influence effects. Finally, the findings suggest potential drawbacks of single-gender schools, which curb adolescents’ opportunities to develop cross-gender friendships (Keener et al., 2013). Accordingly, parents of adolescent boys in single-gender schools may encourage their children to engage in cross-gender friendships outside of school to support the development of their gender-related attitudes. Whether out-of-school cross-gender friends influence gender role attitudes similarly to the in-school friends studied here has to be further ascertained in future research, though.

Conclusion

In this study, I provided the first longitudinal network-analytical assessment of how cross-gender friends influence adolescents’ gender role attitudes. Applied to large-scale survey data, this analytical approach allowed me to account for the most important threats to accurately estimating influence effects. The empirical results showed that girls’ gender role attitudes did not change in reaction to cross-gender friendships, although boys developed more egalitarian attitudes if they had cross-gender friends. These findings highlight that, at least for boys, cross-gender friendships are an avenue for combatting traditional and supporting more egalitarian gender role attitudes.