Introduction

Men and women differ in aggressive behavior, especially in the most violent acts of homicide and aggravated assault. In the United States, men predominate national crime statistics as both perpetrators and victims of crime in almost every domain of violence and outnumber female murderers at a ratio of 10:1 (Federal Bureau of Investigation Uniform Crime Reports 2009). Evidence also suggests that men and women respond differentially to various types of provocation (Bettencourt and Miller 1996). The studies cited above, and the review of theoretical and experimental studies outlined below are based on U.S. samples unless otherwise noted. These data suggest the importance of including gender socialization and masculine-relevant processes into models for understanding and preventing men’s violence. However, this inclusion has been attenuated by the lack of consensus in the field over how masculinity may best be conceptualized, measured, and defined in relation to men’s behavior (e.g., Addis 2010; Smiler and Epstein 2010). The current investigation, therefore, sought to provide an empirical basis for evaluating the validity of common masculinity measures by assessing their unique predictive utility for men’s aggression in the United States.

During the last 30 years, psychologists, guided by gender role theory (Connell 1995), have primarily focused on operationalizing masculinity as an individual phenomenon. Importantly, within gender role theory, two distinct measurement approaches have emerged: gender orientation and gender ideology (Thompson and Pleck 1995). Despite a shared meta-theoretical assumption that gender can be reducible to the individual, these approaches conceptualize the individual’s relationship to gender in distinct ways. The gender orientation approach operates from a trait perspective by assuming that masculinity exists as a “configuration of fixed individual attributes” (Luyt 2015, p. 6). In contrast, the gender ideology approach emphasizes social norms in its assumption that gender is defined by sociocultural forces specifying a variety of norms (or multiple masculinities) that men may internalize to varying degrees based on their particular context. As such, measurement proceeding from the gender ideology approach has made genuine advances in acknowledging cultural context via utilization of multidimensional scales and items phrased as third person, prescriptive statements (Luyt 2015).

Despite laudable attempts to represent the multidimensionality and cultural specificity of masculinity in gender ideology measures, this approach assumes that social norms are ultimately internalized by individuals in a relatively static and unquestioned manner. Indeed, in a recent critical analysis of the challenges facing masculinity researchers, Addis points out that, “all measures of existing [masculinity] constructs…proceed from a trait-like framework because they ask respondents to describe themselves in general terms to generate a single score (or subscale scores) for each individual” (Addis 2010, p. 1). Although measures of gender produced within the gender ideology framework have been fruitful in their ability to associate masculinity with men’s violence (e.g., Garbarino 1999; Goodnight et al. 2014), the continued operationalization of masculinity from a positivistic, individual difference framework echoes longstanding concerns regarding the incremental validity of masculinity measures over and above trait variables, such as personality (e.g., Lippa 1991; Maruŝić and Bratko 1998).

Vigilance to the potential of masculinity measures to be overly saturated with static trait or individual-difference variance is particularly important for contemporary gender scholars who broadly conceptualize masculinity as a non-essentialist construct (e.g., Luyt 2015; O’Neil 2010) and are attuned to the broader consequences of disseminating research in which “masculinity” is discussed as shorthand for a host of negative effects on men’s lives (Sylvester and Hayes 2010). As such, the challenges of contemporary masculinity measurement exceed psychometric considerations to include fundamental theoretical, sociopolitical, and ontological questions. In an effort to address the “problem” of masculinity (Addis et al. 2010, p. 77) and resolve the apparent contradiction that masculinity is theorized as socially constructed but measured as individual difference, Luyt (2015) has argued for a more rigorous adoption of a socially constructionist approach to measuring masculinity termed gender (re)presentation. The latter view of masculinity promotes investigation of the construct as a set of “complex socially guided perceptual, interactional, and micropolitical activities” enacted in contexts in which it is required or expected (West and Zimmerman 1991, p.13-14). A gender (re)presentation approach assumes that individuals “do gender” as a situated action rather than “possess gender” to various extents. The current investigation shares the assumption that gender is a process variable (i.e., culturally shaped and maintained by contextual cues and structural resources).

Such a perspective dictates that masculinity should not reflect purely individual differences, a view also espoused by Smiler and Epstein (2010). Indeed, in a recent review of gender measurement, Smiler and Epstein (2010) express concern that “masculinity…as defined by some of the measures, includes phenomena that have their own literature” (p. 150). Commensurately, a distinct body of literature linking personality dimensions to antisocial behavior such as aggression has developed in parallel to the masculinity literature (e.g., see Bettencourt et al. 2006 for a meta-analytic review of the association between personality and aggression under laboratory conditions). In particular, agreeableness, a trait representing individual differences in level of prosocial, communal orientation has been shown to significantly and negatively relate to men’s physical aggression across a wealth of studies (e.g., see Jones et al. 2011 for a meta-analytic review of the association between Five-Factor Model of personality and anti-social/aggressive behavior). Therefore, elucidating gender-linked predictors of aggression may be complicated by the shared variance of individual differences in interpersonal style (i.e., trait agreeableness) and widely utilized masculinity measures (Parent et al. 2011).

The utility of masculinity measures in elucidating the uniquely gendered component of men’s aggression (i.e., aggression cued by masculine-relevant conditions rather than trait predispositions) is limited to the extent that they account for men’s aggression over and above that explained by individual differences in agreeableness. Therefore, the purpose of the current study was to provide an empirical replication and extension of recent critiques of social scientific masculinity constructs as they apply to men’s physical aggression. Specifically, this empirical investigation sought to clarify both the shared and unique contributions of two of the most widely studied masculinity constructs to men’s propensity for physical aggression, over and above trait agreeableness, in order to directly compare their relative predictive utility in accounting for the gendered component of men’s physical aggression.

Representations of Masculinity in Aggression Theory: Conformity to Masculine Norms and Gender Role Conflict

Understanding gender as a potential feature of an “aggressive situation” (i.e., a process variable) rather than, as a feature of the person in the situation, requires measurement that discriminates between masculinity and trait variables that it may resemble, such as personality. Two contemporary measures of masculinity, the Conformity to Masculine Norms Inventory, and the Gender Role Conflict Scale, have been increasingly utilized in an attempt to elucidate the link between contextually acquired and expressed masculine socialization processes and aggression (Levant and Pollack 1995).

The Conformity to Male Norms Inventory (CMNI: Mahalik et al. 2003) is a gender ideology measure that assesses men’s endorsement of a culture’s ideological beliefs about gender roles. The CMNI proceeds from the theoretical proposition that sociocultural forces that create pressure for men to conform to traditional gender norms influence men’s psychological functioning. Therefore, the CMNI is designed to assess the extent to which men adhere to a particular set of standards or expectations associated with being masculine in U.S. society. In developing the CMNI, Mahalik and colleagues proposed that the norms of masculinity constructed by those in positions of social dominance (i.e., those who are Caucasian, middle/upper class, and heterosexual) are adopted and incorporated into the affective, behavioral, and cognitive repertoire of all U.S. men, at varying levels across individuals (Mahalik et al. 2003).

A noted strength of the CMNI is its multifactorial structure (Smiler and Epstein 2010) that purportedly broadens the scope of masculinity measurement to include 11 distinct factors. These factors represent behaviors, thoughts, and emotions congruent with numerous male norms (e.g., Emotional Control, Risk-Taking, Violence). Of the CMNI norms, the Violence subscale has been shown to be the most robust correlate of clinically linked hostility as well as general aggression (Mahalik et al. 2003). Accordingly, aggression may arise as an expression of men’s enactment of this socially and personally constructed norm, particularly as this norm is posited to reflect prescriptive stereotypes for a way that men “ought” to act. Similarly, men’s aggression may also arise as the collective result of men’s enactment of the larger subset of masculine norms, given that these norms, in synthesis with one another, may activate a desire to appear dominant (Kilianski 2003). Indeed, assessment of convergent validity has demonstrated a strong positive correlation among the CMNI total score and social dominance (Mahalik et al. 2003), while more recent studies have shown CMNI total score to be associated with engagement in physical aggression in the laboratory (Berke et al. 2011; Reidy et al. 2009).

However, attempts to demonstrate a direct link between adherence to masculine norms and aggression have produced equivocal findings. For example, in a study of the relationships between adherence to masculine ideology and dating violence, measures of masculine ideology failed to explain a significant proportion of the variance of aggression and violence (Jakupcak et al. 2002). Moreover, Fitzpatrick et al. (2004) garnered a lack of support for a direct link between masculine ideology and intimate partner aggression. Inconsistencies in this literature suggest that shared relations between masculine socialization and other psychological processes operating at various levels (i.e., emotional, temperamental) inform associations between gender ideology and aggression.

For example, gender role conflict, which purportedly represents the “psychological state in which socialized gender roles have negative consequences for the person or others” (O’Neil et al. 1986, p. 336), has been postulated to drive male aggression through the experience of distress. Specifically, when men encounter actual or perceived challenges to masculine status, they may invoke stereotypical masculine behaviors in the form of aggression in order to maintain their sense of power and control (Moore and Stuart 2005). The Gender Role Conflict Scale (GRCS), developed by O’Neil et al. (1986), belongs to a class of measures that examines the degree to which internalization of traditional gender roles is likely to cause an individual stress (Smiler and Epstein 2010). As such, the GRCS shares core assumptions with the gender ideology approach (namely that socioculturally derived gender norms are internalized at the individual level) and extends this approach to account for individual differences in the adaptation to gender norms. In keeping with this theoretical distinction, the GRCS, which consists of four subscales that can be combined to yield a total score, has been shown to be psychometrically distinct from the CMNI (Mahalik et al. 2003), significantly related to hostility (Hayes and Mahalik 2000), and positively associated with dangerous interpersonal outcomes for men including sexual and dating violence (O’Neil 2008). Notably, men’s experience of stress concomitant to efforts to adhere to masculine norms has been associated with increased state anger, negative intent attributions, and verbal aggression following exposure to intimate conflict vignettes involving masculine gender-relevant contexts (Franchina et al. 2001; Moore and Stuart 2004). Gender role conflict has also been shown to predict physical aggression in the laboratory (Cohn et al. 2010).

Collectively, this evidence suggests that aggression may work as psychological salve to men’s experience of gender role conflict—a means of coping with the distress emanating from pressures to conform to a set of harsh prescriptive and proscriptive social norms. Indeed, Jakupcak et al. (2002) examined the role of masculine gender role stress in perpetration of intimate violence among men in college and found that this stress significantly predicted violence in dating relationships above and beyond the effects of adherence to masculine norms. Similarly, Jakcupcak and colleagues revealed a moderating effect whereby adherence to masculine ideology only emerged as a significant predictor of aggression in the context of high levels of gender role stress, suggesting the potential for men’s aggressive behavior to vary as a function of the affective strain of masculine norms rather than via masculine ideology alone. Moreover, in a 2006 study, Cohn and Zeichner found a significant moderating effect of gender role stress on the relation between adherence to masculine norms and laboratory aggression. These findings suggest that conformity to gender norms may only be salient if men also experience these norms as stressful. In other words, the tenuous link between masculine ideology and aggression reflected in mixed research findings is strengthened when men’s conformity to masculine norms is considered in tandem with gender role conflict.

However, it is also possible that the inconsistent findings on the link between masculine ideology and aggression detailed above may be driven by problematic overlap between interrelated masculinity measures (i.e., CMNI and GRCS) and other individual difference or trait variables (e.g., Lippa 1991; Maruŝić and Bratko 1998). That is, the variable link between masculine ideology and aggressive behavior may also be understood as a failure to consider the influence of a yet another relevant variable— trait agreeableness.

Trait Agreeableness and Aggression

To discern potential overlap between masculinity and personality, it is essential to understand how personality relates to aggression. Elucidation of traits linked to aggression has been greatly aided by psychometric advancement in the measurement of broad models of personality such as the Five-Factor Model, which posits five higher-order personality factors, including agreeableness, which reflects an individual’s interpersonal strategies and motivations (FFM; Costa and McCrae 1992a).

Researchers have utilized the FFM to examine the relations between personality and aggressive behavior across several populations including collegiate and community samples, and psychiatric patients (Gleason et al. 2004; Miller et al. 2003). In a meta-analytic review of antisocial/aggressive behavior manifested in real-world situations (Jones et al. 2011), FFM agreeableness emerged as the strongest FFM correlate (mean effect size r = −.37). Individuals low in agreeableness tended to be antagonistic, hostile, and irritable, and appeared to manifest aggressive tendencies via a hostile attribution bias whereby they interpret the behavior of others as intentionally hurtful (Miller et al. 2008). Accordingly, such individuals are more likely to evaluate aggressive behavior as an appropriate means of conflict resolution (Jensen-Campbell and Graziano 2001) and more likely to engage in both reactive and proactive forms of aggression (Miller and Lynam 2006). Moreover, recent research (e.g., Miller et al. 2009) has demonstrated an inverse relationship between agreeableness and trait aggression, which is purported to reflect one’s dispositional proclivity towards aggressive behavior (Buss and Perry 1992).

Personality research suggests that aggression may be understood in terms of relatively stable trait differences in interpersonal style (i.e., agreeableness), while masculinity theory points to the validity of understanding aggression in terms of the purportedly dynamic effects of gender socialization and conflict. Simultaneously examining these disparate perspectives within a single multivariate model of male aggression permits empirical assessment of the putative distinctiveness of masculinity constructs from trait differences.

Untested Elements of Gendered Aggression Theory: Discriminating Masculinity from Trait Agreeableness

Although the CMNI and GRCS were developed from a framework acknowledging the role of social norms in men’s adherence to and experience of gender expectations, studies utilizing these constructs proceed, in practice, from an individual differences framework given their reliance on self-reported endorsement of items that describe an individual in general terms (Addis 2010). This methodological practice speaks to psychologists’ dependence on traits as metaphors for dynamic gendered processes (Smiler 2004). While individuals indeed vary in terms of how they are impacted by hegemonic attitudes or the extent to which socially prescribed gender roles are internalized, it is important to recognize gender as a contextually enacted behavior that unfolds in social encounters and not purely as individual differences. Therefore, the validity of masculinity constructs hinges on their ability to predict gender-salient behavioral outcomes, over and above static variations in interpersonal style based in temperament.

Current Study

Although we agree with social psychological theories of gender that assert performances of masculinity depend on situational determinants (e.g., Deaux and Major 1987), we believe that an empirical assessment of the consistency of the CMNI and GRCS with this stance is essential. Failure to isolate trait/personality components of masculinity obfuscates discernment of information about the contingency of hegemonic masculinity (i.e., the situational context driving expression of aggressive norms). As such, widely utilized measures of masculinity merit further theoretical and empirical examination in order to determine how they may operate in relation to trait agreeableness. Such exploration may yield support for the discriminant validity of masculine ideology from gender role conflict and aid in the creation of a more comprehensive account of the gendered component of men’s aggression.

The current study aimed to provide an empirical evaluation of the question, “can contemporary masculinity measures adequately predict the uniquely gender-linked component of men’s physical aggression?” We attempted to provide the answer via development of two separate, but analogously structured, path-analytic models of physical aggression based on endorsement of masculine conformity (i.e., CMNI) and conflict (i.e., GRCS), trait agreeableness, and physical aggression self-reported by a sample of U.S. men. Structural consistency of between models was maintained to allow for direct comparisons of the incremental validity of the CMNI and GRCS over and above trait agreeableness.

Structural Models

The proposed model first included the CMNI and GRCS as interrelated factors, each predictive of physical aggression, as evidence suggests that adherence to masculine norms and gender role conflict predict men’s physical aggression (e.g., Cohn and Zeichner 2006; Jakupcak et al. 2002). Second, consistent with the wealth of literature linking trait agreeableness to antisocial behavior (e.g., Jones et al. 2011), we specified a path linking agreeableness to physical aggression. Third, consistent with the theory that masculine norms prescribe antagonistic social conduct (Mahalik et al. 2003), we also specified paths linking the CMNI and GRCS to agreeableness (see Figs. 1 and 2). With agreeableness positioned as endogenous to intercorrelated CMNI and GRCS total scores, the structural model permitted tests of the direct effects of CMNI and GRCS total scores on physical aggression, controlling for the effects of agreeableness as well as the indirect effect of these variables via agreeableness. Moreover, as the ability of the CMNI to assess specific masculine norms is cited as a particular strength of this measure (e.g., Smiler and Epstein 2010), a structural model analogous to that described above was also constructed using the subscale most associated with men’s physical aggression (i.e., CMNI-Violence) in place of the CMNI-total score.

Fig. 1
figure 1

Predicted model comparing mediated effect of CMNI and GRCS on aggression through agreeableness

Full size image
Fig. 2
figure 2

Predicted model comparing mediated effect of CMNI violence norms and GRCS on aggression through agreeableness

Full size image

Hypotheses

  1. 1)

    Consistent with the theoretical contention that all masculinity measures are reducible to individual difference/trait-like variance (Addis 2010; Luyt 2015), we hypothesized a significant indirect effect of CMNI and GRCS on physical aggression through trait agreeableness.

  2. 2)

    However, given that the CMNI is purported to assess an individual’s general adherence to masculine norms (Mahalik et al. 2003), while the GRCS assesses conflict produced by such adherence (O’Neil et al. 1986), we hypothesized that the CMNI would be more saturated by trait variance than by the GRCS such that:

    1. a.

      Agreeableness would fully mediate the relationship between CMNI and physical aggression so that no direct effect of CMNI on aggression would remain when controlling for agreeableness

    2. b.

      Agreeableness would only partially mediate the effect of GRCS on physical aggression such that a significant direct effect of GRCS on aggression would remain even when controlling for agreeableness.

  3. 3)

    Because the Violence norm similarly captures a participant’s general beliefs, behaviors, and actions as they relate to violence, just as the CMNI total score captures individual differences in these domains across a broader set of norms, we hypothesized that the pattern of findings expected above would hold for both predicted models (i.e., for both the model including CMNI-total score and for that where CMNI-Violence is entered in its stead).

Method

Participants and Experimental Design

A sample of 181 men was recruited from the psychology department’s research participation pool at a large university in the southeastern United States. All men enrolled in introductory psychology courses at the university were eligible for the study and were recruited via an advertisement posted on the departmental website. The sample was demographically representative of the university community from which it was derived. See Table 1 for a summary of the sample demographic characteristics.

Table 1 Mean and standard deviation of participant age, and percentages for race/ethnicity and relationship status
Full size table

Measures

Demographic Form

The participants completed a brief demographic form to provide age, race, marital status, income, and level of education.

Revised NEO Five-Factor Inventory (NEO-FFI; Costa and McCrae 1992a)

The NEO-FFI is a 60-item, self-report measure of the FFM of personality. Although it assesses the five broad personality factors of the FFM (i.e., Neuroticism, Extraversion, Openness, Agreeableness, Conscientiousness), each rated on a 5-point scale ranging from 0 (strongly disagree) to 4 (strongly agree), the Agreeableness factor alone was utilized in the current study (sample item: “I would rather cooperate with others than compete with them”). In a validation study, the NEO-FFI provided reliable assessment of the FFM domains, with Cronbach alpha coefficients reported as .86, .77, .73, .68, and .81 for Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness, respectively (Costa and McCrae 1992b). The internal consistency for Agreeableness in the current study was acceptable, α = .76.

Conformity to Masculine Norms Inventory (CMNI; Mahalik et al. 2003)

The CMNI was used in the current study to assess men’s meeting of societal expectations for what constitutes masculinity. Respondents indicated the extent of agreement with behaviors, thoughts, and emotions congruent with male norms using a scale ranging from 1 (strongly disagree) to 4 (strongly agree) on items reflecting enactment of 11 distinct themes (Parent and Moradi 2009). A CMNI total score was derived for use in the current study by adding scores for all items across the 11 subscales. This total score has shown good convergent validity, as it has been positively related to gender role conflict via the GRCS and masculine gender role stress on the masculine gender role stress (MGRS: Eisler and Skidmore 1987) scale, as well as a strong positive correlation with social dominance and general aggression (Mahalik et al. 2003). Furthermore, the CMNI total score has been shown to have good test-retest reliability (Mahalik et al. 2003). In the current sample, Cronbach alpha for total CMNI was high (α = .92). In terms of subscale scores, the Violence subscale (example item, “Sometimes violent action is necessary.”) was exclusively selected for inclusion in the current study as construct validation studies have shown this subscale to be most strongly correlated with hostility and aggression (Mahalik et al. 2003). Cronbach alpha for CMNI-Violence was high (α = .84).

Gender Role Conflict Scale (GRCS; O’Neil 1981)

The GRCS is purported to measure “gender role conflict,” a construct developed and defined in a formal model as “a psychological state in which gender roles have negative consequences or impact on the person or others” (p. 336). The scale comprises 37 items that use a 6-point scale ranging from 1 (strongly disagree) to 6 (strongly agree) to indicate the degree participants agree with the given statement (sample item: “Making money is part of my idea of being a successful man”). Tallying up all GRCS items yields a total score that has been utilized to identify correlates and consequences of gender role conflict across 25 years of empirical literature (O’Neil 2008). For example, positive correlations have been demonstrated between GRCS total scores and other measures of masculinity supporting its construct validity (Good et al. 1995). In the current sample, Cronbach alpha for total GRCS was, α = .93.

Buss-Perry Aggression Questionnaire (BAQ; Buss and Perry 1992)

This 29-item questionnaire was used in the current study to assess trait aggression. Participants indicated their endorsement of items as characteristic of them on a 1 (not at all characteristic of me) to 7 (extremely characteristic of me) rating scale. The questionnaire comprises four subscales: Physical Aggression, Anger, Hostility, and Verbal Aggression. The Physical Aggression subscale (BAQ-PA), consisting of nine items (e.g., “If somebody hits me, I hit back.”), is conceptualized as self-reported aggressive behavior and was used as the dependent variable in the current study. The use of this subscale in isolation from anger, hostility, and verbal aggression promotes the goal of the current study—to elucidate potential mechanism driving gender differences in physical aggression in particular. The BAQ has been demonstrated to comprise stable, empirically confirmed components (Bernstein and Gesn 1997). In the current sample, Cronbach alpha for the Physical Aggression subscale of the BAQ was, α = .82.

Procedure

Male participants responded to a study approved by the university’s Institutional Review Board, including a questionnaire session that yielded the data analyzed for the purpose of this study. Participants were directed to contact the experimenter via email to gain access to an online survey posted on the Qualtrics website. A link to the survey was sent to the personal email addresses participants used to contact the experimenter. Contacting participants directly permitted extraction of an IP address that could be matched to their Qualtrics survey response to assure that participants completed the surveys themselves. Importantly, the IP address was removed prior to data analysis to protect confidentiality. Once logged on, participants were presented with an informational letter describing the survey and the option to skip over questions or terminate their participation at any time without penalty. As such, participants received credit even in the event that they chose to terminate their involvement. After providing informed consent, all participants completed a questionnaire battery comprising demographic information, NEO-FFI, BAQ, CMNI, and GRCS. All surveys were presented in the stated order so that any unintended order effects would be held constant across all participants. Finally, participants were presented with an electronic debriefing and given research participation credit.

Results

Preliminary Analyses

A multivariate analysis of variance was performed to rule out significant differences in key variables of interest (i.e., CMNI, GRCS, BAQ-PA, and FFM Agreeableness) across racial groups and categories of relationship status (e.g., single, committed/long-time partnership). The analysis failed to reveal significant differences on key variables based on group membership. Therefore, homogeneity of variance of masculinity measures, FFM Agreeableness, and BAQ-PA was supported across pertinent demographic variables. As such, it is unlikely that statistical associations among variables of interest were confounded by either race or relationship status.

Correlations, Means, and Standard Deviations

Pearson product–moment correlations, means, and standard deviations were computed among FFM agreeableness, self-report physical aggression, the total score of continuous predictor variables (i.e., CMNI and GRCS), and the CMNI-Violence subscale (Table 2). The magnitude of intercorrelations among these variables implies potential multicollinearity. Multicollinearity is unavoidable with mediation analyses (Baron and Kenny 1986), but the problems of multicollinearity tend to occur at higher correlations (Tabachnick et al. 2001). To determine the effect of multicollinearity in the present study, a regression analysis was performed in SPSS 21. This analysis showed that the variance-inflation factors of FFM agreeableness (VIF = 1.26), GRCS (VIF = 1.44), CMNI-Total (VIF = 1.59) and CMNI-Violence (VIF = 1.25) were all below suggested cutoffs (Cohen et al. 2003).

Table 2 Correlations, means, and standard deviations for all variables
Full size table

Path Analyses

To construct estimates for modeled paths among masculinity measures (CMNI-Total/Violence; GRCS), trait agreeableness, and physical aggression, two separate path analyses were performed, one including the CMNI total score and one including CMNI-Violence (see Figs. 1 and 2). These analyses utilized ML estimation performed in Mplus version 6.1 (Muthén and Muthén 2010). Models were considered to have adequate fit on the basis of several indices. These included the Comparative Fit Index (CFI) and the Root Mean Square Error Approximation (RMSEA). CFI values approximating or exceeding .95 and RMSEA values of less than .05 were considered as criteria for acceptable fit (Kline 2011). Based on these criteria, the fit of both Model 1 (CFI = 1; RMSEA = 0) and Model 2 (CFI = 1; RMSEA = 0) were excellent and reached near perfect fit given that each model specified more parameters to be estimated than variables included.

Hypothesis Testing

Estimates of the indirect effects for the hypothesized association between CMNI and physical aggression through agreeableness and GRCS and physical aggression through agreeableness were obtained using bootstrapping with 1000 samples (Preacher and Hayes 2004, 2008) using Mplus STDYX standardization (see Table 3). Hypothesis 1, which predicted a significant indirect effect of CMNI and GRCS on physical aggression through trait agreeableness, was partially supported. A significant indirect effect emerged through agreeableness for CMNI, but the indirect effect for GRCS only achieved marginal significance. These results partially supported the prediction that the relationship between these masculinity variables and men’s physical aggression would be mediated, at least to some extent, by the shared association between CMNI and low trait agreeableness.

Table 3 Estimate of indirect and direct effects of predictor total scores on physical aggression
Full size table

Hypothesis 2 pertained to predicted comparisons between the mediated effect of CMNI and that of GRCS through aggression. Hypothesis 2a predicted that the relationship between CMNI and aggression would be fully mediated by agreeableness, while hypothesis 2b predicted that a direct effect of GRCS on aggression would remain while controlling for trait agreeableness. These hypotheses were not supported, as a direct effect emerged for both CMNI and GRCS when controlling for effects of trait agreeableness and the other independent variable (see Table 3). As such, full mediation was not supported for either variable.

Hypothesis 3 pertained to the second path analytic model (see Fig. 2) which was analogous to the first with the exception that only the Violence subscale of the CMNI was utilized as the relevant measure of masculine norm adherence rather than the total score. For this model, we hypothesized a pattern of mediation effects identical to that emerging from the first model. This hypothesis was partially supported, as there were both similarities and marked difference between the two models. As in the first model, an indirect effect emerged through agreeableness for CMNI –Violence. In contrast to the first model, an indirect effect emerged through agreeableness for GRCS in the second model. However, a direct effect on physical aggression after controlling for trait agreeableness only remained for CMNI-Violence (Direct effect = 0.483, p = .00) but not GRCS, which failed to directly predict men’s physical aggression when simultaneously controlling for CMNI-Violence and Agreeableness. Therefore, full mediation was supported for the GRCS in this model (see Table 4).

Table 4 Estimate of indirect and direct effects of CMNI violence norm and GRCS total score on physical aggression
Full size table

Discussion

The purpose of the current study was to provide a rigorous test of the critique that masculinity measures assess individual difference/trait-like constructs, despite the aspirations of the CMNI and GRCS to capture the socially constituted nature of gender. Specifically, this empirical investigation utilized structural equation modeling to clarify both the shared and unique contributions of two of the most widely studied masculinity construct, conformity to masculine norms and gender role conflict, to men’s propensity for physical aggression over and above trait agreeableness. The use of a structural equation modeling approach permitted mediation analyses that accounted for the structural association linking conformity to masculine norms to gender role conflict. Additionally, the examination of multiple models allowed for an evaluation of the incremental validity of masculine norms over and above trait agreeableness in terms of a composite score and in terms of one masculine norm in particular, violence.

Results of the meditational path models revealed a set of novel findings which both confirmed the presence of direct associations between masculinity measures and physical aggression and revealed conditions under which these associations are explained by trait differences in agreeableness. In the first model, consistent with previous literature, total scores of both the CMNI score and GRCS emerged as predictors of men’s self-report of physical aggression (Berke et al. 2011; O’Neil 2008; Reidy et al. 2009). As predicted, an indirect effect emerged such that men’s trait agreeableness scores accounted, in part, for the association between the conformity to masculine norms and aggression. These effects support the presence of partial or inconsistent mediation (MacKinnon et al. 2007) and, therefore, provide partial support for critiques of modern masculinity measures as problematically proceeding from an individual difference/ trait-like framework (Addis 2010; Luyt 2015).

Importantly, however, direct effects remained for both the CMNI and GRCS while no indirect effects emerged for GRCS. These results suggest that, when considering its association with men’s adherence to the total subset of masculinity norms, the GRCS accounts for unique variance in physical aggression scores above and beyond that accounted for by trait agreeableness. Furthermore, the emergence of a direct effect of CMNI on aggression in addition to its effect via agreeableness, suggest that both conformity to masculine norms and gender role conflict appear to explain men’s aggressive responding in the current study above and beyond trait differences in interpersonal strategies (i.e., agreeableness), suggesting that these are not purely trait constructs but, rather, capture a phenomenon less static than an antagonistic personality style.

Mahalik et al. (2003) original construal of conformity to masculine norms posits the construct to be distinct from measures of gender role ideology (e.g., Male Role Norms Inventory; Levant et al. 1992) that purportedly reflect proscriptive and prescriptive beliefs about men’s behavior by virtue of their gender. In contrast, Mahalik et al. (2003) suggest that conformity to masculine norms assess personal conformity, and may, therefore, have positive, neutral, or negative implications contingent on the individual or the social context. That this construct predicted variance in men’s self-report physical aggression over and above a relatively static personality variable (i.e., agreeableness), lends support for the argument that conformity to masculine norms may account for a particularly gendered aspect of aggressive outcomes.

Similarly, for gender role conflict, results of the first model suggest that men who experience conflict as a result of the constraints of the masculine gender role may be physically aggressive regardless of whether or not they possess an antagonistic personality, as gender role conflict uniquely accounted for a proportion of men’s aggression in the first model. Within the framework of the Levant’s Gender Role Strain Paradigm, gender role conflict characterizes an internal dilemma associated with behavior that is not in accordance with the “ideals of manhood” (Levant 1996, p. 260). This may, in turn, predispose men high on this variable to engage in (potentially aggressive) actions designed to resolve internal conflict.

In the second path model, men’s adherence to the violence norm, as opposed to conformity to a broader set of masculine norms, was investigated. As the multidimensional structure of the CMNI is cited as a strength of this measure, in that it allows for more targeted focus on those norms most likely to be associated with a gendered outcome of interest (Smiler and Epstein 2010), we believe that the presentation of this second path model provided a particularly rigorous test of the scholarly critique of contemporary masculinity measures (Addis 2010; Luyt 2015). In this model, conformity to the violence norm predicted men’s aggression both directly as well as through its shared relationship with trait agreeableness, a pattern analogous to that which emerged for adherence to all masculine norms in sum in the first model. This result suggests that when only considering adherence to those norms of masculinity that explicitly prescribe violence as a means of resolving conflict and asserting status, masculine adherence to this norm accounts for a subset of men’s violence beyond that explained by broader temperamental tendencies towards a competitive interpersonal style.

Perhaps more importantly, when considering adherence to violent masculinity norms in the second model, no direct relationship between men’s gender role conflict and aggression remained when controlling for trait agreeableness. In other words, when men’s adherence to violent masculinity norms is taken into account, the conflict men experience as a consequence of adhering to masculine norms (i.e., GRCS) is only associated with physical aggression to the extent that those men are also high in trait antagonism. Results from this model suggest a more complex, temperament-linked, relationship between gender role conflict and aggression than that revealed in the first, implicating gender role conflict as more akin to trait measurement than conformity to masculine norms in its linking of “societal norms scripting traditional masculinities to individual’s adaptation” (Thompson and Pleck 1995, p. 151).

Rather than understanding gender role conflict as a singular risk factor for problematic externalizing behavior such as aggression, results from the current study suggest that such outcomes may be more comprehensively explained by the static temperamental tendencies reflected in the extent to which adherence to violent masculinity norms results in personal conflict. Low trait agreeableness is purported to reflect a tendency towards competitive (versus cooperative) interpersonal strategies (Costa and McCrae 1992a). Thus, it stands to reason that the indirect effect linking gender role conformity to self-report physical aggression through agreeableness reflects the overlapping variance of low cooperation shared by these measures. Indeed, men’s temperamental propensity towards competition and greater gender role conflict (i.e., men with low flexibility in their enactment of this component of hegemonic masculinity) were both positively correlated with physical aggression in our sample. Our findings demonstrate that gender role conflict may not be a necessary condition for gendered aggression if a man is sufficiently adherent to violent norms of masculinity and sufficiently disagreeable.

Study Limitations

Several limitations of this study merit comment. Two representative measures of contemporary masculinity measurement were selected for comparison in the current study; however, research of the psychology of men has given rise to a host empirically investigated, reliable measures purported to assess masculine socialization processes. In future studies, the growing universe of these constructs and their respective subscales should be subjected to a well-powered multifactorial path analysis such as the one undertaken in the current study. Likewise, we chose to focus on aggressive outcomes due to their obvious social implications. However, numerous other variables are related to masculinity and personality (e.g., help-seeking, dominance, work place success, etc.), and such variables may better shed light unto the overlap between trait masculinity and personality.

Additionally, the current study is limited by its cross-sectional, self-report design. In this investigation, self-report assessment of physical aggression was employed as a proxy for a behavioral outcome purportedly linked to gendered provocation, producing results of a correlational nature. Given the study design, it is not possible to make causal statements regarding men’s aggressive behavior. Moreover, participants in this study were a relatively homogenous sample of men whose identification with racial and class categories were treated as “control” factors. Thus, caution should be exercised in generalizing findings beyond the population of Caucasian, high-school graduates, enrolled in a Southeastern U.S. university.

Conclusion and Future Directions

Despite these limitations, results from the current study indicate that men’s aggressive behavior contains features of gender socialization as well as trait differences. Our results provide greater clarity regarding the distinct, yet interrelated constructs of conformity to masculine norms, gender role conflict, and trait agreeableness as they relate to male perpetrated aggression. Findings from the current study suggest that the unique association between gender role conflict and physical aggression over and above trait agreeableness may vary as a function of men’s adherence to violent masculine norms, in particular, rather than a broader subset of these norms. Importantly, this study also builds on the theoretical and epistemological debate in the psychology of men and masculinity literature in which commonly used measures of gender are critiqued as overly reductionistic in their focus on individual difference (Addis 2010; Luyt 2015). We were able to subject this critique to empirical scrutiny to reveal a more informed and nuanced assessment of the incremental validity of masculinity measures over and above trait agreeableness.

The current study reveals adherence to masculinity norms to be a robust direct predictor of men’s aggressive behavior. However, when considering men’s adherence to the violence norm, gender role conflict appears to be more fully steeped in trait antagonism given that its association with physical aggression is explained by trait agreeableness. As such, this investigation highlights the relative merits of extant masculinity measures while simultaneously revealing a goal for future research: to account more fully for unique variability within gendered identification and expression as it relates to men’s aggressive behavior, particularly the ways in which this variability may be driven by systematic, rather than trait difference. Elucidating unique components of this dynamic is necessary given the socializing function of gender systems to organize all men within a set of power relations pairing masculinity with dominance (Connell 1995).

A greater focus on the systematic nature of masculinity, as unique from and in interaction with trait factors, is vital for researchers interested in developing a more complete account of the gendered aspects of men’s aggressive behavior. For example, the gender (re)presentation approach, proposed by Luyt (2015), calls for the stringent adoption of social constructionist methodology of gender measurement by exclusively assessing masculinity in groups rather than as individuals. Results from the current investigation lend support for the value of pursuing such radical departures from current measurement practices in continuing efforts to develop the study of masculinity. Indeed, advancing measurement to match theoretical progress in the psychology of men and masculinity is important for clarifying how gender may come to be linked to aggression in specific contexts—a goal requisite to both violence reduction and to the promotion of gender justice.