Attitudes toward masculinity have been theorized as contributing to social hierarchy formation among males (Connell, 1995). Personal and perceived attitudes toward masculinity were measured through the Masculine Attitudes Index (MAI; Lusher, 2008). This inventory has both adolescent and adult versions, with major subscales including antifemininity, gay male homophobia, and violence. For adolescents, an antiacademic subscale is also included, and for adults, a subscale measuring masculinity as playboy attitudes (i.e., manliness pertaining to sexual success with women). Scale scores are averaged across all items of the four subscales (minimum = 1, maximum = 7). Higher scores more strongly endorse antifeminine, homophobic, and violence attitudes (as well as antiacademic and playboy attitudes), which we refer to as indicative of a more “dominative masculinity.” That is, high scores on the MAI imply that males hold antifeminine, homophobic, and violent attitudes, whereas low scores on the scale imply that they do not. The MAI is derived from psychological characteristics deemed pertinent to masculinity, empirical research, and theoretical work by Connell (1995), assessing some core components of masculinity in Western English-speaking countries. The inclusion of the “violence as manly” subscale is especially pertinent for the football team context (Lusher, 2008).
To measure both personal and perceived attitudes, participants were asked to answer the MAI items twice: first, with respect to their personal attitudes, and second, with regard to their perceptions of the attitudes of others around them (i.e., their “reference group”). Prentice and Miller (1993) found that paired-sample t-tests showed significant differences between personal and perceived attitudes measured in this way, with the perceived attitudes always more extreme. For school students, the reference group comprised the students’ close friends, indicative of a peer group that has been shown to be important to young people (e.g., Poteat, Espelage, & Green, 2007). For the football team, the reference group was the team itself, given the strong identification players (are encouraged to) develop for the team in which they play. We refer to an individual’s endorsement as “personal MAI” and to the individual’s perceived attitudes of their reference group as “perceived MAI.” For the schoolboys, personal MAI had a mean of 3.77 (SD = 0.92) and perceived MAI a mean of 4.15 (SD = 0.89); for the footballers, personal MAI had a mean of 2.88 (SD = 0.63), and perceived MAI a mean of 3.15 (SD = 0.69).
Other individual-level measures acted as control variables and were introduced as other actor-relation effects in our model. For schools, these were ethnocultural background and socioeconomic status (SES). Ethnocultural background was a binary variable. Dominant culture students were those who identified as Australian or Anglo-Australian, or of British or Irish heritage, and spoke only English at home. All other students were considered to be of nondominant cultural status. SES was calculated using the Socio-Economic Index For Areas (SEIFA) Index of Relative Socio-Economic Advantage/Disadvantage (ABS, 2001), derived from household postal codes, with higher scores indicative of greater socioeconomic advantage.
For the footballers, control variables were playing ability, experience, and relationship status. A measure of playing ability was included as an exogenous covariate network, which constituted binary responses to “Who are the best players in the club?” Experience was measured by the number of games played. A binary variable indicating whether players were in an ongoing romantic relationship was also included.
-  As measured by the network question, “Who is in your close group of friends?”