Electronic Journal of Sociology (2003)

ISSN: 1198 3655

Masculinity and Crime: A Quantitative Exploration of Messerschmidt’s Hypothesis

Jessie L. Krienert
Illinois State University


Using a sample of newly incarcerated offenders in Nebraska, this study analyzes self-report data at the event level to examine the effects of masculine traits on violent and avoided violent situations1 . Different outlets that are available for the expression of masculinity as well as a traditional measure of masculinity are used to explore high-risk incidents. Results of logistic regression show that masculinity alone fails to significantly predict violent events. Men with high masculinity and few acceptable outlets to assert masculinity, however, are more likely to be in a violent incident. These findings suggest the need for better measures of both masculinity and appropriate outlets for masculine expression.

The United States has higher rates of violent crime than any other industrialized nation (Reiss and Roth, 1993; Oliver, 1994). An overwhelming majority of this crime (83 % in 2000) is committed by males (UCR, 2001). This is not a new trend, males have been grossly over-represented in all major violent crime categories since the beginning of the collection of official crime statistics (UCR, 1999). This preponderance of males occurs for arrest, self-report, and victimization data (see Steffensmeier, 1995). Therefore, the lack of attention paid to specifically what it is about being male that causes violence is surprising. While men have traditionally been used as subjects in social science inquiry, theories relating directly to why men commit more crime have routinely focused on inherent characteristics of women (Hagedorn, 1998). It is taken for granted that males are more criminogenic, theories of crime use this as an underlying assumption and rarely attempt to explain the phenomenon. Specific male traits, or male identity characteristics, have not been systematically examined in relation to violent crime. In fact, the National Research Council’s report “Understanding and Preventing Violence” failed to even mention gender as a social process that could lead to violence (Reiss and Roth, 1993).

Recent work, however, has suggested that masculinity is an important construct for understanding crime and violence (Messerschmidt, 1993; 1997; 2000; Bowker, 1998, Kimmel and Messner, 2001). Toughness, dominance, and the willingness to resort to violence to resolve interpersonal conflicts are central characteristics of masculine identity (Kersten, 1993; Messerschmidt, 1993; Anderson, 1994, Messner and Sabo, 1994). The belief that gender roles are socially constructed is widely agreed upon in social science research (Hagedorn, 1998). Based on this social construction, not all men have the same levels of masculine traits. Key differences in gender roles are highly likely to be important variables in the understanding of violence. Consequently, any examination of male-on-male violence is incomplete, without unfolding the meaning and influence of masculine gender.

Masculinity and Violence

The issue of masculinity and its link to criminal behavior in men has been identified in both past and present discussions of criminological thought. An early attempt to link masculinity with crime came from Parsons (1964). Parsons purported masculinity was internalized during adolescence, which led to boys engaging in more delinquent behavior than girls (Parsons, 1964). Sutherland also discussed elements of masculinity when stating that boys are taught to be “rough and tough,” which makes them more likely to become delinquent (Sutherland and Cressey, 1924). Cloward and Ohlin’s (1960) differential opportunity theory is also relevant to discussions of masculinity. They propose that younger gang members learn both legitimate and illegitimate behaviors from older male role models. Common masculine traits such as toughness and dominance, which are considered necessary in order to assert a strong masculine reputation, are learned through this contact with older males (Cloward and Ohlin, 1960).

Switching to a more current examination, Miedzian (1991) examines the socialization, peer pressure, media, and military influences that lead to violence being an acceptable behavior in men. Taking a similar route, in a qualitative examination of dating violence, Thompson (1991:275) found that men with more masculine gender orientations were more likely to be involved in violent exchanges. Additional research linking violence and masculinity includes an analysis of episodes of violence found in film and media, as well as in sporting events (Wilson and Daly, 1985; Gruneau and Whitson, 1993; Weinstein, et al., 1995). Common to these examinations, is the use of the term masculinity without an operationalized definition, and although the link between violence and masculinity is routinely proposed it is rarely empirically examined.

Examinations that do address the relationship between violence and masculinity often use samples of inner city adolescent boys (Campbell, 1986; Barker and Lowenstein, 1997; Fagan, 1998; Wilkinson, 1998;). An early study by Toby (1966:20-21) hypothesized that compulsive masculinity could explain violent behavior among this type of population. In his study, however, masculinity was never operationalized, and instead he shifted the focus to the occurrence of matriarchal homes and lack of a father figure to explain the affect of masculinity. There was no concrete definition or measurement of masculinity, or clear explanation of the link between masculinity and violence in his examination. During the same time frame, Wolfgang and Ferracuti, in their subculture of violence research (1967:259), listed masculinity as a main concern of this subculture. They went a step further than Toby, by defining masculinity as maleness with overt physical aggression. The actual relationship of masculinity and violent crime, however, was never empirically tested and thus not supported by their study.

More recently, masculinity in the form of “young male syndrome” has become a common focus of inner city youth examinations due to the masculine ideals of competition and risk (Pallone and Hennessy, 1993:134; Wilkinson and Fagan, 1996; Anderson, 1994). Recent exploration also delves into the presence of male peers which researchers have found intensifies the need to project a masculine image (Pallone and Hennessy, 1993; Wilkinson, 1998). Studies concerning peer relations and masculinity can also be seen in modern gang research. Gang activity has been seen as emphasizing masculine traits of achievement, aggressiveness, and daring behaviors (Spergel and Grossman, 1997; Lasley, 1998; Wilkinson, 1998). Although mentioned in research on inner city adolescents, masculinity has only played a minor role in the overall research. Typically, this research is qualitative, and since masculinity is not the major focus, it is not clearly identified nor defined through systematic inquiry.

Doing Gender

The idea of “doing gender” was introduced by West and Zimmerman in 1987. They conceptualize gender as a routine accomplishment that is created and maintained through everyday interaction. This argument identifies the way in which expressing masculinity is directly linked to criminal behavior, especially the use of violence (Walklate, 1995:172). Instead of seeing masculinity as something that just happens to men or is done to men, masculinity is seen as something that men do (Coleman, 1990:191). Specific patterns are learned through the socialization process that appropriately represent masculinity. Masculinity must be performed and presented recurrently in any situation--constant self-presentation occurs throughout every social interaction in which a man is involved. Ongoing re-creation is a defining feature of masculinity. This re-creation occurs in the family, at work, in school, and in all other social settings. The underlying goal of this performance is the assertion of power and dominance (Miedzian, 1991). The leap to violent activity is not difficult to make. Male violence can be used to support and maintain status in the male group and nurture a sense of male identity (Archer, 1994:121). The use of aggressive and violent acts can allow men to maintain status in their male group.

When “doing gender,” there are both traditional and alternative outlets that can be utilized to accomplish the same end result. To achieve manhood status, there are certain qualities that social scientists have labeled masculine and manly. These qualities and characteristics have been restructured throughout history; however, scientists in several disciplines list similar concepts that can be used to demonstrate masculinity. These include but are not limited to marriage, having dependents, providing for the family, and proclaiming an aura of physical and mental strength and dominance (Gilmore, 1990:48; Meidzian, 1991; Messerschmidt, 1993; Gutmann, 1997:403). Accepted by general society, these characteristics would allow a man to appropriately display his gender.

In any social interaction, both the actor and the audience have a shared understanding of ideal masculine gender. Appropriate presentation of masculine display is expected based on the shared understanding (Polk, 1997:130). To present a positive masculine image, a man will rely on learned cultural definitions of masculinity. In the United States, this includes physical strength, aggressiveness, and visible proof of achievement (Messerschmidt, 1993; Gutmann, 1997). Masculine identity is consistent with being a tough and courageous person. According to Tedeschi and Felson (1994:197), the preponderance of research indicates that men take more risks than women and this risk-taking behavior is also a feature of masculinity. Whether the behavior that asserts masculinity is criminal or legal, the important aspect is that it is fully asserted. The use of criminal masculine accomplishment is especially likely when a man’s masculinity is called into question or threatened (Polk, 1997). Masculine gender is not what one is, but something that one does--something one does at all times. Therefore, if traditional, non-criminal resources are not available, alternative resources, even criminal resources, will be used to accomplish masculine gender (Messerschmidt, 1993).

Messerschmidt’ s Masculinity Hypothesis

Toby (1966) argued that males who did not have power derived from family background, educational achievement, income, social and political connections or material goods would be more likely to exert a “compulsive” masculinity which could easily result in violent behavior. This can be seen as a precursor to Messerschmidt’s masculinity hypothesis.

Using West and Zimmerman’s concept of “doing gender,” in conjunction with the work of Robert Connell (See Connell 1987, 1995) Messerschmidt takes the idea a step further. He hypothesizes that criminal behavior can be used as a resource when other resources are not available for accomplishing masculinity. For example, if a person does not have a steady, reliable job, a stable family life, or other traditional indicators of successful masculinity, violent behavior may be considered an acceptable way to convey the “toughness” that is linked with masculine traits. Other traditional outlets of successful masculinity include success in school as well as both marriage and children. Perhaps the most well known indicator of masculine success is that of occupational achievement, usually derived from monetary success (Archer, 1994:135). Full-time work in the paid labor force is an acceptable outlet for accomplishing masculinity. A man can assert his masculinity by holding a steady job and bringing home reliable pay, therefore fulfilling the masculine role of “good provider” (Messerschmidt, 1993:70). These are all traditional, conventional, examples of how one can “do gender.” It is when these traditional means of demonstrating masculinity are stifled, or do not exist, that violent behavior is most likely to occur (Messerschmidt, 1993:81). If a man does poorly at school or at his job, or in his family life, he must seek out other, alternative, “masculine-validating resources” (Messerschmidt, 1993:83).

For men, performing well in violent situations enhances masculinity regardless of whether it is a traditional or alternative display (Messerschmidt, 1993; Cavender, 1999: 159). Men are considered masculine if they can defend themselves. The willingness to fight in any given situation is a measure of male self-worth (Toch, 1997, Kimmel, 1996). The question then becomes one of which situations will lead men to be more likely to use violence to “do” gender.

Messerschmidt (1993) hypothesizes an answer to this question by focusing on the use of criminal behavior as an acceptable, alternative, way to accomplish or project masculinity. Messerschmidt hypothesized that if traditional outlets of masculinity were unavailable in a given situation, alternative outlets, such as violence would be more likely to be used. Furthermore, people who have used violence as a way to assert masculinity in the past, come to accept violence as an acceptable route to display their manhood. In effect, they become predisposed to using certain masculine resources based on certain situational cues. Not only then, is gender systematically accomplished, but it is regulated, changed, and reproduced based on gender ideals in a socially structured setting (Messerschmidt, 2000).

Since Messerschmidt’s masculinity hypothesis in 1993, little empirical work has been conducted. Although notable qualitative studies have supported Messerschmidt’s basic ideas (see Bourgois, 1995; Perry, 2001, Martin & Jurik, 1996, Chesney-Lind & Hagedorn, 1999), the different routes men have to accomplish and project masculinity have not been empirically examined in relation to violent crime. Quantitative support for the interaction of masculinity and other factors that have been routinely linked to both masculinity and violent situations would enhance the existing qualitative research.

The transition from knowing males commit more crime than women to understanding the unique nature of masculinity in relation to this noted difference is an important part of future criminological research. In essence, this is the transition from the use of sex to the use of gender in criminological thought. Especially in areas of violence and aggression, masculinity can be used to garner a better understanding of male behavior. This study responds to the need for theoretical and methodological advances in research on masculinity and violence. Taking into consideration the work of West and Zimmerman (1987) and Messerschmidt (1993, 1997, 2000), the primary purpose of this examination is to provide an empirical evaluation of a traditional static measure of masculinity, as well as an in depth look at the traditionally acceptable outlets available to assert masculine gender in a violent event. Characteristics of masculinity will be identified and operationalized to gain a greater understanding of the male role and its relation to violent crime.

The Present Study

This research tests three hypotheses. The first is that incidents will be more likely to be violent when they contain men with masculine traits and behaviors. The second relates to the traditionally acceptable outlets available to assert masculinity, predicting that events are more likely to be violent when they contain individuals who have few of these appropriate outlets. The final hypothesis is that violent situations will be more likely to contain individuals with very masculine characteristics and few acceptable routes for asserting those characteristics. Although none of the hypotheses constitute a pure test of Messerschmidt’s masculinity hypothesis, they represent a preliminary attempt at a quantitative analysis of masculinity and violence.


Structured interviews were conducted with 704 newly incarcerated offenders at the Diagnostic and Evaluation Center in Lincoln, Nebraska. Inmates were randomly selected from all new intakes to the department by crossing off every third name on the official intake list and interviewing all remaining inmates. The data were collected with structured, computer-aided interviews, which lasted from 30 minutes to six hours. Length of interview was driven by amount of violent and non-violent incidents recalled by the inmate. Inmates were interviewed in private rooms where informed consent procedures were thoroughly explained: 90.37 percent of all inmates agreed to participate. This high participation rate is based on the entire sample of inmates who were selected to participate, even including those who did not want to come to the visiting room to have the study explained. Such a high response rate is common in prison-based research, especially research conducted at diagnostic and evaluation centers, due to several reasons, one of which being inmates have so few activities to participate in (see Junger-Tas and Marshall, 1999).

Once interview procedures were explained, information on the 36-month period prior to an inmate’s most recent arrest was gathered using a life-events calendar. A paper format of this calendar was placed in front of the respondent, who was encouraged to record helpful information directly onto the calendar. Holidays, birthdays, vacations, incarceration periods, and time with significant others were a few of the events inmates placed onto the paper calendar. If the respondent had trouble writing, the interviewer recorded whatever information the inmates reported onto the calendar. This calendar helped inmates recall details of their lives and sequences of events about residence, cohabitation, substance use, employment, school attendance, significant relationships, gang memberships, and routine daily activities (Horney, 2000).

The calendar approach also aided a respondent’s memory in order to gain detailed information on violent and non-violent events which occurred during the three-year period. The life event, or life history, approach has been found to be beneficial for retrospective data collection. Such an approach allows for a more detailed set of cues that may jog respondents’ memory (Axin, Pearce, and Ghimire, 1999). This not only expands the likelihood that the respondent will remember an elicited event, but also that the sequence of events will be more accurate. Freedman, et al. (1988) tested the reliability of this approach with a sample of 900 young adults who gave retrospective reports, gathered at two separate time periods with the calendar approach; reports were highly similar and contained very few instances of missing data.

The main study from which this investigation is derived contains 1,658 violent and 760 non-violent events. Of these inmates, the majority (54.3%) were 30 years of age or less. A fairly substantial number (40.8%) had either a high school diploma or G.E.D. Caucasians made up the largest percentage of respondents (58.9%) followed by African Americans (19.9%).

The sample for the current study is limited only to cases that involve male-on-male events. An event fit this category if all participants were males. Since this investigation is based on interpersonal violence, even in the avoided or non-event category, participants must be present together at the same location. These criteria reduce the overall sample to 1,707 events derived from 508 offenders. In addition, all incidents that occur in an incarcerative setting are excluded due to the thoroughly different dynamic that this setting creates (see Steinke, 1991). Demographic characteristics of this study are very similar to those for the larger sample. Difference of means tests were performed on the two samples, with no significant differences found between the two. A detailed comparison is conducted in Table 1. Of the inmates in the current sample, the majority (68.7%) were 30 years of age or less. A fairly substantial number (40.4%) had either a high school diploma or G.E.D. Caucasians made up the largest percentage of respondents (54.8%) followed by African Americans (21.3%).

Dependent Variable

This study examines interpersonal violence at the event level. The type of event serves as the dependent variable. All events fall into the two categories of this variable:

(1) violent event, consists of physical confrontations in which the respondent was an active participant. An incident was categorized as violent if there was attempted or actual hitting, slapping, use of weapon, throwing of objects, or choking or kicking. Each respondent averaged two violent incidents.
(2) avoided-violent event, includes situations in which the respondent did not hit anyone, nor was he hit by anyone else. Typically, this included events where there was minor violence such as pushing, shoving, grabbing, or threatened violent action by a participant in the event. In an avoided-violent incident, the fight or argument does not escalate into serious violence. Respondents averaged one non-violent incident.

Categories were created prior to data collection and separate collection instruments were used for each category. Although pushing, shoving, and grabbing have frequently been noted in literature as assaultive behavior, the original research from which data for the current study is drawn, defined an incident as non-violent if it stopped at this point. These categories fit well with the current examination, since both types of incident would be intimidating or high risk situations in which one might feel more of a need to affirm gender (Messerschmidt, 1997).

Independent Variables


Based on the weaknesses of past studies, two separate quantitative measures are used to gauge masculine traits and the outlets available to assert masculinity in the current examination. The first measure is an additive scale containing traditionally acceptable outlets of masculinity. The components of this scale were created using Messerschmidt’s masculinity hypothesis, as well as numerous researchers’ characterizations of what would be considered traditional ways to convey masculinity. The second measure is taken directly from scores on the Masculinity-Femininity Scale, MF, of the MMPI-2 test, which each inmate completed upon admission to the Diagnostic and Evaluation Center.

For this study, masculinity will be defined as a socially constructed set of meanings, values, and practices that come together to form differing levels of behavior that men work to project. To create an empirical measure from this definition, the MMPI-2 MF score will be used to measure base masculine traits, in conjunction with positive outlets of masculinity that will be identified and placed into an additive scale reflecting the traditional, noncriminal ways males may have to assert their masculinity.

When arranging different types of masculinity on a continuum or hierarchy, hegemonic masculinity can be thought of as the top or ideal form of masculinity. Past research has identified this ideal masculinity through fulfilling certain masculine goals. Common ideal goals were marriage, having dependents, and providing for the family (Gilmore, 1990: 48). These are the three most common indicators of masculinity, regardless of time or culture. They are all uses of masculine accomplishment, and can be seen as ways to assert power and dominance (Miedzian, 1991). In addition to monetary and family success, Messerschmidt (1993) also proposes that success at education or school is another traditional source of masculine power. They are all noncriminal, traditional, means that men could use to show or prove their masculinity.

It should be noted that these are static measures of masculinity. Ideally, to test a theory of structured action, fluid measures of masculinity would be utilized. Static measures, such as the MMPI, however, still show importance. Examining the effects of a stable core masculine personality on a given situation is an important endeavor. The masculine traits people have, whether static or fluid, are brought into each and every social interaction. It is the differing levels of these traits that are being examined in the current investigation.

Traditional Outlets of Masculinity Scale. Identified through past research, the following elements outside of the violent event that traditionally show or prove a person’s manhood are included in an additive scale. Scores from each variable detailed below are added together to create a composite scale titled ‘Outlets’:

Education is measured by the highest grade of schooling the respondent has completed. Less than high school is coded as 0, high school or G.E.D. is coded as 1, more than high school is coded as 2.
Marital Status is measured by either married (1) or not married (0) at time of incident.
Children is measured as a dichotomous yes or no variable at time of incident.
Employment is measured as full-time (2), part-time (1) or unemployed (0) at time of incident.
Income is measured as amount of personal income through legal wages before taxes at time of incident. Less than $5,000 was coded as (0), $5,000-$14,999 as (1), and $15,000 or more as (2).

Scale reliability diagnostics were completed for this scale. As a measure of internal consistency, Cronbach’s Alpha was calculated. A value above .60 is commonly noted as adequate consistency (Mitchell and Jolley, 1999). The measure of appropriate traditional outlets of masculinity (Outlets) had an adequate alpha value of .62.

The outlet score means were significantly higher for non-violent outcomes than for violent outcomes. Incidents involving men scoring low on this scale were significantly more likely to be violent incidents. The small difference of means between the two, however, may point to a limited substantive significance due to the size of the sample.

In addition to the measurement of traditional reflections of masculinity, it is also important to gauge the base level of masculine personality traits that a man holds. The demonstration of physical strength, aggressiveness, competition, power, and control, are traditional facets of the masculine persona (Oakley, 1972; Miedzian, 1991; Manley, 1998). These are also typical characteristics that are measured through the MMPI-2 MF Scale.

MMPI-2 Masculinity-Femininity Scale MF. The Masculinity-Femininity (MF) scale is one of the most controversial scales used on the MMPI-2. The original purpose of this scale was the identification of gay men and individuals demonstrating role reversal (McGrath, Sapareto and Pogge, 1998). Men with high T-scores suggest a feminine orientation, while men who display low T-scores suggest a more masculine orientation (Duckworth and Anderson, 1995). The scale is made up of 56 yes/no items that were created to measure stereotypic masculine interests (Graham, 2000: 73). Many of the criticisms lodged against the use of this scale rely on continued use of stereotypical interpretations of masculinity that can be found in the questions that make up this scale. For the current investigation, however, stereotypical masculinity is precisely what is being examined. Those who show high levels of masculinity, based on common male stereotyped traits, would fit with the definition of masculinity used in this study. Several modern investigations utilize the MF scale in studies of delinquency and crime (Morton, Farris, and Brenowitz, 2002). Actual MMPI-2 T-scores will be used in the current analysis.

High Risk Masculinity (HRM). To better address Messerschmidt’s 1993 hypothesis, a dichotomous variable measuring high risk masculine men was created. Those who score high on masculinity as measured through the MMPI-2 MF T-score, and low on the traditional outlets of masculinity scale are coded as one. All other combinations are coded as zero. Initially, incidents that were one standard deviation away from the mean were included in each category. Due to lack of cases, however, the higher half of MF T-scores, and lower half of traditional outlet scores were used to create the high-risk masculinity (HRM) variable used in multivariate analysis.

T-tests showed statistical significance at the .05 level for this variable. Violent incidents are more likely than avoided violent incidents to involve men who are very masculine and have few traditional outlets (See Table 3).

Control Variables

As controls, standard demographic variables previously found to relate to masculinity and violence were included (Gutman, 1997; Miedzian, 1991). Age is measured through the question “what is your date of birth?” This date is subtracted from the year of data collection in order to assess relative age. Race/Ethnicity is measured through the question “what is your race/ethnicity”. Dummy variables are created for White, Black, Hispanic, Native American and Other categories for assessing race/ethnicity effects.


In order to test the likelihood that a violent or non-violent incident can be predicted by the independent variables, logistic regression was used. The current data contains events that are nested within persons (i.e. one person could give anywhere from 1 to 20 events). In many cases, Hierarchical Linear Modeling is an analysis strategy that is used with nested data. This strategy allows both the person level and event level variables to be examined simultaneously. Unfortunately, the current analysis is based on the use of a summed scale that combines variables from both the person and the event levels into a single measure. Placing this variable at either level in the HLM analysis could compromise the results (Osborne, 2000). Although logistic regression doesn’t account for the nested structure of the data, it is less problematic than the use of the summed scale in HLM. Using logistic regression may lead to a higher probability of rejection of the null hypothesis (Wooldredge, Griffin & Pratt, 2001), however a summed scale from both levels would reduce the ability of HLM to account for this problem. Due to the deletion of cases with missing values, 1507 cases were included in the regression analysis.

Following the general approach to logistic regression diagnostics set forth by Menard in 1995, several procedures were conducted as precautionary data checks. A thorough examination of studentized residuals, the leverage statistic, and the DBETA warranted no cause for concern. As an additional test, the issue of multicollinearity was examined. No problematic coefficients were found in this examination when looking at the correlation matrix and tolerance statistics.

Using the .05 level of statistical significance, three separate regression models were examined. In each statistical model race and age were included as control variables. In all models, white was the omitted category for race variables. Age was found to be significant in all models. Race, however, was never significant in any model that was run.

In the first model, the dependent variable was regressed on traditional MMPI-2 indicators of masculinity. In this analysis, the most common way to measure masculinity and its effect upon violence was isolated. The second model contains the addition of the traditional outlets of masculinity scale. The final model contains the created measurement of high risk masculinity containing low traditional outlets and high MMPI-2 scores. An additional model was run that created an interaction term between high masculinity and low traditional outlets, however, it was not significant. Table 5 reports the results of a series of logistic regression analyses. As shown in the table, traditional indicators of masculinity and appropriate outlets of masculinity were not significant indicators of a violent outcome. No significant differences were found between violent and avoided violence incidents based on masculinity as indicated by the MMPI-2 MF scale. Similarly, the addition of traditional outlets to assert masculinity did not significantly enhance the understanding of violent incidents. In the third model, however, the created High Risk Masculinity variable was significant. As predicted, violent incidents were more likely to include highly masculine men who had few traditional outlets to assert their masculinity.


Although many studies make mention of masculinity being intertwined with violent behavior, few have put the idea to an empirical test. This paper builds on the ideas of Messerschmidt and West and Zimmerman by attempting to empirically test both the masculinity violent crime relationship and the hypothesis that those with few outlets to assert masculinity will be more likely to resort to violence. Through the creation of a traditional outlets of masculinity scale and the use of a traditional measure of masculinity three hypotheses were examined.

Results show that traditional masculinity and acceptable outlets alone are not significant indicators of a violent event. The MMPI-2 measure used to assess masculine attitudes and beliefs failed to show significance in the current analysis. This does not necessarily, however, lead to the conclusion that masculinity has no effect on the likelihood of violent incidents.

Mmpi-2 (Mf) Scale

There are several plausible reasons the masculinity and violence hypotheses failed. First, using traditional MMPI-2 measures for masculinity may not tap into true masculine traits and values. Although the MMPI-2 masculinity scales are commonly used, they may not be the best measures of masculinity. Long and Graham (1991) found that the MF scale was highly correlated with education. Educated men were consistently found to have more “feminine” characteristics than less educated men (Long and Graham, 1991: 47). This study is one of many that attacks the validity of the MF scale for explaining personality and behavior of normal men (Johnson, Jones, and Brems, 1996; McGrath, Sapareto, and Pogge, 1998).

Empirical testing of the MF scale notes that it has low internal consistency and low concurrent validity with established sex role measures such as the BEM sex role inventory. The BEM inventory measures cultural definitions of gender in order to determine how influential gender roles are on the respondent (Lenney, 1991). When comparing the MF scales to other sex role measures such as the BEM inventory, or the sex role behavior scale, the MMPI-2 scales show a stronger relationship to personality traits of interpersonal strength and sensitivity than masculinity (Johnson, Jones, and Brems, 1996). This suggests that MMPI-2 measures may not be directly measuring the construct of masculinity.

Further examination into the specific items that make up the MF scale lead to additional questions about the adequacy of measurement. For example, whether or not one likes the work of a librarian, nurse, or florist, may not be as important as one’s physical attributes, which are not measured in the MMPI-2 scales (Butcher, 1990). Anthropological and sociological examinations into masculinity have routinely examined issues of aggression, success, occupational achievement, honor, status, dominance, toughness, loyalty, respect, courage, and risk-taking (Herek, 1986; Miedzian, 1991; Messerschmidt, 1993; Gutmann, 1997; Wilkinson, 1998). When examining the items that make up the MMPI-2 subscales, these elements are hard to detect.

Finally, the MMPI-2 scales are based on stereotypical traits of masculinity (Duckworth and Anderson, 1995). The MF scale was created in 1943 and has sustained only minor changes in the past 57 years (Colligan and Offord, 1992). It is quite likely that stereotypical masculinity in 1943 is different than stereotypical masculinity in 2000.

The need for a more appropriate measure of masculinity is linked directly to the need for an operational definition. Current investigations of masculinity are commonly qualitative in nature and even then, rarely have masculinity as the main focus (Hearn and Morgan, 1990). A better and agreed upon definition of masculinity is necessary in order to create a successful measure.

Traditional Outlets Of Masculinity

Lack of support for the second hypothesis could fall in the creation of the traditional outlets scale. Messerschmidt (1993) does not specify whether the outlets for traditional masculinity are important only in a dichotomous sense, or if the quality of each outlet plays an important role. The operationalization of what are traditionally considered appropriate masculine outlets, therefore, may be poorly constructed.

Although the idea of appropriate traditional outlets addresses the issue of perceived status, personal satisfaction may play a large role in how one believes others perceive them. For instance, being married at the time of the offense may not be as important as being happily married at time of offense. Although the status of being married may be important, the quality of the marriage may play an equally important role in masculine achievement. If success in a marital relationship is a coveted position, men with this status can promote their coveted success.

Along the same line, knowing whether or not the man has children is somewhat different than knowing if the man is living with his children. Fathering a child is very different than being a father to a child; successful parenting may be a better traditional outlet, than just having a child. The quality of parent-child relations may afford a sense of pride which in turn can create a positive outlet for masculine display. Any position in which other males may envy would be a positive arena for masculinity.

Similarly, in the area of work and income, job satisfaction may be a more important measure than just having a job. If a man has a job that he enjoys, that may be seen as a better way to assert masculinity than just having a job. Currie (1985) finds that the quality of employment is much more important than general employment when looking at a crime-employment relationship. A man may be able to better assert masculinity by having enjoyable employment that others may covet, rather than a job that no one else would want. Additionally, if a man feels he’s making a “good salary”, regardless of the actual amount of income, that may be a better way to assess the amount of money necessary to assert his masculinity. If he sees himself as doing well, he has proven his masculinity.

A different problem with traditional outlets is the use of only traditional outlets to assert masculinity. Alternative outlets may also play a very important role. Messerschmidt (1993) discusses the use of violence as an alternative when legitimate means for masculine display are unavailable. What is missing, however, is a discussion of other not so legitimate, alternative, means of masculine display other than violence. For instance, the number of past sexual partners could be another way to tap into masculine display. Showing ones manhood through sexual encounters has been theorized as an important aspect of masculinity by several researchers (Gilmore, 1990; Gutmann, 1997; Cavender, 1999). As an extension of this idea, anthropological literature suggests the number of children a man fathers may play an important role. The more children a man has, the easier it is to prove his virility. Past violent display is an additional area that may be important to the study of masculinity. If a man has proven his masculinity several times in the past, he can draw from those experiences to assert his masculinity in any given incident. The existence of several less appropriate outlets for a man to prove masculinity is an important area to explore in the future.

To summarize, Messerschmidt (1993) was not entirely clear on how he would operationalize the idea of traditional outlets for masculine display. The idea of traditional outlets may need to be re-conceptualized and broken down into two separate categories. First, a category that contains positive social outlets, such as successful parenting and job satisfaction would be a way to assess successful masculinity. Second, additional less appropriate outlets such as number of past sexual partners, number of children, or number of past successful violent encounters may tap into a different way that a man could assert masculinity without the need for violence in the current situation.

Other shortcomings of Messerschmidt’s hypothesis should also be noted. The interaction of race and class with gender adds a different dynamic to negotiated masculinity. Although “doing gender” may be one way to look at masculine accomplishment, imposed characteristics may play an equally important role in gender display (Hood-Williams, 2001). Examinations of masculinity as it is constructed for men by political, class, and culture may lead to different results, than examinations of masculinity believed to be constructed by men (Carver, 1996).

It is also important to examine factors outside the realm of masculinity. Other factors, such as situational variables may be more important when discussing violence, than the personal-level trait of masculinity. There could be important contextual characteristics that would change the outcome of a given event. Instead of focusing on what types of masculine traits seem to be inherent to violent men, it may be better to focus on what types of situational variables, such as location, presence of others, and time of day are more likely to occur in male-male violent situations. Messerschmidt (2000) discusses the occurrence of a “masculinity challenge” that may enhance the use of violence in a situation. The current investigation focused on high-risk events, future research may wish to further limit the events to separate events in which a “masculinity challenge” occurs. Future research should explore these possibilities and control for appropriate contextual situational variables in each event.

Hegemonic Masculinity

Preliminary support was found for the third hypothesis. Males who exhibit more masculine traits and have fewer acceptable outlets to assert masculinity were more likely to be involved in violent events. Thus support was found for Messerschmidt’s general theoretical notion that men who lack masculine resources are more likely to turn to violence. Although this relationship was found to be significant, several additional areas need to be considered. First, this analysis does not control for contextual or situational characteristics that may be important. Traditional situational variables including third party presence, drug or alcohol use, and physical location could greatly impact the outcome of an event. Similarly, the true unfolding an event, through the use of identity attacks and sequencing of actions was not available for use in this research. The addition of contextual factors could easily strengthen or change the impact of masculinity on violent and nonviolent events.

Next, it may be better to measure differing levels of masculinity, such as hegemonic masculinity, instead of relying on a scale of general masculinity. The MMPI-2 scales do not specifically measure the different levels or categories of masculinity. Although high scores on the MMPPI-2 scales may indicate higher overall masculine traits, the scales were not formed to identify ultramasculine men. Scoring high on a traditional scale measuring general masculinity is different than scoring high on a scale measuring hegemonic masculinity. A scale created to measure hegemonic masculinity instead of simply general masculinity may hold greater explanatory power in the area of violence. In essence, Messerschmidt was predicting that very masculine traits would be linked with violence, therefore, a measure of specifically hegemonic, or ultramasculine characteristics may better tap the underlying theme of his hypothesis. If the MMPI-2 scales are not tapping into true masculine characteristics, that could account for the failure of masculinity as a significant variable in this examination. Perhaps, a better measure of masculine gender is necessary in order to truly test this hypothesis.


Future examinations of masculinity and violence should move towards a better operationalization of the concept of masculinity. Past research, especially in the areas of crime and violence rarely use an empirical measure of masculinity. Without a strong operational definition, interactions between masculinity and other variables are impossible. Second, the use of routes to assert masculinity is an area that has not been frequently examined. More elaboration on both positive and negative outlets that exist to assert masculinity is necessary. Third, future efforts need to include both men and women in the sample. Masculinity is too often only examined in relation to men. Masculinity is a characteristic of gender, not sex, therefore the addition of women to the sample may greatly increase the current knowledge base. Although Messerschmidt (1993) only discussed men in his masculinity hypothesis, the appropriate and less appropriate outlets that women have to assert their gender may lead to a better conceptualization of Messerschmidt’s original idea.

Finally, the current study is only a preliminary test of masculinity and violence. An integrated theoretical model of masculinity and violence, resulting in more research would be beneficial for understanding behavior in both violent and avoided violent incidents.


Anderson, Elijah. 1994. “The code of the streets.” The Atlantic Monthly.

Archer, John. 1994. Male Violence. London and New York: Routledge.

Axin, William G., Lisa D. Pearce, and Dirgha Ghimire. 1999. “Innovations in life history calendar applications.” Social Science Research 28:243-264.

Barker, G. and I. Lowenstein. 1997. “Where the boys are.” Youth & Society 29:166-196.

Bourgois, Philippe I. 1995. In search of respect: selling crack in El Barrio. New York: Cambridge University Press.

Bowker, Lee. 1998. Masculinities and Violence. Thousand Oaks, CA: Sage Publications, Inc.

Butcher, James Neal. 1990. Development and Use of the MMPI-2 Content Scales. Minneapolis: University of Minneapolis Press.

Campbell, Anne. 1986. “The streets and violence.” in Violent transactions: the limits of personality, edited by A. Campbell and J. Gibbs. Oxford: Basil Blackwell.

Carver, Terrell. 1996. “Public Man” and the critique of masculinities.” Political Theory. 24: 673-686.

Cavender, Gary. 1999. “Detecting masculinity.” in Making Trouble: cultural constructions of crime, deviance and control, edited by J. Ferrell and N. Websdale: Aldine de Gruyter.

Chesney-Lind, Meda and John M. Hagedorn. 1999. “Female Gangs in America: Essays on Girls, Gangs, and Gender.” Chicago: Lake View Press.

Cloward, Richard A. and Lloyd E. Ohlin. 1960. Delinquency and Opportunity: a theory of delinquent gangs. Glencoe, IL: Free Press.

Coleman, Wil. 1990. “Doing masculinity/doing theory.” in Men, Masculinities and Social Theory, edited by J. Hearn and D. H. Morgan. London: Unwin Hyman Ltd.

Colligan, Robert C. and Kenneth P. Offord. 1992. The MMPI: A contemporary normative study of adolescents. Norwood, New Jersey: Ablex Publishing Corporation.

Connell, Robert. W. 1987. Gender and power : society, the person and sexual politics. Stanford, Calif. : Stanford University Press.

___. 1995. Masculinities. Berkeley : University of California Press.

Currie, Eliot. 1985. Confronting Crime: an American Challenge. New York: Pantheon.

Duckworth, Jane C. and Wayne P. Anderson. 1995. MMPI & MMPI-2: interpretation manual for counselors and clinicians. Bristol, PA: Accelerated Development, Inc.

Fagan, Jeffrey. 1998. “Adolescent violence: a view from the street.” Pp. 167-193 in National Institute of Jutice: Research Preview, vol. 22, edited by J. Travis: US Department of Justice.

Federal Bureau of Investigation. 2001. “Crime in the United States 2000: Uniform Crime Reports.”

Federal Bureau of Investigation. 1999. “Crime in the United States 1998: Uniform Crime Reports.”

Freedman, Deborah, Arland Thornton, Donald Camburn, Duane Alwin, and Linda Young Demarco. 1988. “The life history calendar: a technique for collecting retrospective data.” Sociological Methodology 18:37-68.

Gilmore, David. 1990. Manhood in the making: cultural concepts of masculinity. New Haven: Yale University Press.

Graham, John R. 2000. MMPI-2: assessing personality and psychopathology. New York: Oxford University Press.

Gruneau, R. and D. Witson. 1993. Hockey night in Canada. Toronto, Ontario: Garamond Press.

Gutmann, Matthew C. 1997. “Trafficking in men: the anthropology of masculinity.” Annual Review of Anthropology 26:385-409.

Hagedorn, John M. 1998. “Gang violence in the post industrial era.” in Youth Violence, edited by M. Tonry and M. H. Moore. Chicago, IL: University of Chicago Press.

Hearn, Jeff and David H. Morgan. 1990. “Men, masculinities and social theory.” in Men, Masculinities and Social Theory, edited by J. Hearn and D. H. Morgan. London: Unwin Hyman Ltd.

Herek, G. 1986. “On heterosexual masculinity: some psychical consequences of the social construction of gender and sexuality.” American Behavioral Scientist 29:563-577.

Hood-Williams, John. 2001. “Gender, masculinities and crime: From structures to psyches.” Theoretical Criminology 5(1): 37-60.

Horney, Julie. 2000. “Criminal Events and Criminal Careers.” in The process and structure of crime: criminal events and crime analysis, edited by R. Meier, L. Kennedy, and V. Sacco. New Brunswick, N.J.: Transaction Publishers.

Johnson, Mark E., Georgia Jones, and Christiane Brems. 1996. “Concurrent validity of the MMPI-2 Feminine Gender Role (GF) and Masculine Gender Role (GM) scales.” Journal of Personality Assessment 66:153-168.

Junger-Tas, Josine and Ineke Haen Marshall. 1999. “The self-report methodology in crime researchers.” in Crime and Justice: a review of the research, vol. 25, edited by M. Tonry. Chicago, IL: University of Chicago Press.

Kersten, Joachim. 1993. “Crime and Masculinities in Australia, Germany and Japan.” International Sociology 8:461-478.

Kimmel, Michael S. 1996. Manhood in America : a cultural history. New York : Free Press.

Kimmel, Michael S and Michael A. Messner (eds). 2001. Men’sLlives. Boston : Allyn and Bacon.

Lasley, James. 1998. “Designing Out” Gang Homicides and Street Assaults. Washington, D.C.: US Department of Justice.

Lenney, Ellen. 1991. “Sex roles: the measurement of masculinity.” in Measures of Personality and Social Psychological Attitudes, edited by J. P. Robinson and P. R. Shaver. San Diego, Ca: Academic Press, Inc.

Long, Kathleen A. and John R. Graham. 1991. “The Masculinity-Femininity Scale of MMPI-2: is it useful with normal men?” Journal of Personality Assessment 57:46-51.

Manley, James C. 1998. “Masculinity and Moral development as contributors to violent crime among Mexican American male offenders.” Ph.D. dissertation Thesis, New State University.

Martin, Susan Ehrlich and Nancy C. Jurik. 1996. Doing Justice, Doing Gender: Women in law and criminal justice occupations. Thousand Oaks CA: Sage.

McGrath, Robert E., Elizabeth Sapareto, and David L. Pogge. 1998. “A new perspective on gender orientation measurement with the MMPI-2: development of the Masculine-Feminine pathology scale.” Journal of Personality Assessment 70:551-563.

Menard, Scott W. 1995. Applied Logistic Regression Analysis. Thousand Oaks CA: Sage Publications, Inc.

Messerschmidt, James W. 1993. Masculinities and Crime: Critique and Reconceptuatlization of Theory: Rowan & Littlefield Publishers, Inc.

—. 1997. Crime as structured action: gender, race, class and crime in the making: Sage Publications.

—. 2000. Nine Lives: Adolescent Masculinities, the Body, and Violence: Westview Press.

Messner, Michael A and Donald F. Sabo. 1994. Sex, violence & power in sports : rethinking masculinity. Freedom, CA : Crossing Press.

Miedzian, Myriam. 1991. Boys will be boys: breaking the link between masculinity and violence. New York: Doubleday.

Mitchell, Mark and Janina Jolley. 1999. Research Design Explained. New York: Holt, Rinehart and Winston.

Morton, Todd L., Kelly L. Farris and Lisa H. Brenowitz. 2002. MMPI-A Scores and High Points of Male Juvenile Delinquents: Scales 4, 5, and 6 as Markers of Juvenile Delinquency.” Psychological Assessment 14: 311-319.

Oakley, Ray. 1972. Drugs, Society, and Human Behavior. St. Louis, MO: McGraw Hill.

Oliver, William. 1994. The Violent Social World of Black Men. New York: Lexington Books.

Osborne, Jason. 2000. “Advantages of Hierarchical Linear Modeling.” Practical Assessment, Research & Evaluation 7.

Pallone, Nathanial and James Hennessy. 1993. Criminal Behavior: a process psychology analysis. New Brunswick: Transaction Publishers.

Parsons, Talcott. 1964. Social Structure and Personality. New York: Free Press of Glencoe.

Perry, Barbara. 2001. In the Name of hate: understanding hate crimes. New York: Routledge.

Polk, Kenneth. 1997. “Men, Honour Contests and Violence.” in American Society of Criminology. Chicago.

Reiss, Albert J. and Jeffrey A. Roth. 1993. Understanding and Preventing Violence. Washington, D.C.: National Academy Press.

Spergel, Irving and Susan Grossman. 1997. “The Little Village Project: a community approach to the gang problem.” Social Work 42:456-470.

Steffensmeier, Darrell. 1995. “Trends in female crime: It’s still a man’s world.” in The Criminal Justice System and Women: Offenders, Victims, and Workers, edited by B. Price and N. R. Sokoloff: McGraw-Hill.

Steinke, Pamela. 1991. “Using situational factors to predict types of prison violence.” Journal of Offender Rehabilitation 17, 1-2, 119-132.

Sutherland, Edwin H. and Donald R. Cressey. 1924. Criminology. Philadelphia: J.B. Lipencott.

Tedeschi, James and Richard B. Felson. 1994. Violence, Aggression, and Coercive Actions. Washington, D.C.: American Psychological Association.

Thompson, Edward H., Jr. 1991. “The Maleness of Violence in Dating Relationships: An Appraisal of Stereotypes”. Sex Roles 24, 5-6, 261-278.

Toby, Jackson. 1966. “Violence and the masculine ideal: some qualitative data.” Annals of the American Academy of Political and Social Science 364:19-27.

Toch, Hans. 1997. Corrections: A humanistic approach. Guilderland, NY: Harrow and Heston.

Walklate, Sandra. 1995. Gender and Crime: Prentice Hall.

Weinstein, Marc D., Michael D. Smith, and David L. Wiesenthal. 1995. “Masculinity and hockey violence.” Sex Roles 33:831-847.

West, Candace and Don H. Zimmerman. 1987. “Doing Gender.” Gender and Society 1:125-181.

Wilkinson, Deanna L. 1998. “The Social and symbolic construction of Violent Events among inner city adolescent males.” Ph.D. dissertation Thesis, Rutgers University.

Wilkinson, Deanna L. and Jeffrey Fagan. 1996. “Understanding the role of firearms in violence ’scripts’: the dynamics of gun events among adolescent males.” Law and Contemporary Problems 59:55-89.

Wilson, Margo and Martin Daly. 1985. “Competitiveness, risk taking, and violence: the young male syndrome.” Ethology and Sociobiology 6:59-73.

Wolfgang, Marvin and Franco Ferracuti. 1967. The Subculture of violence: towards an integrated theory in criminology. New York: Travistock Publications.

Wooldredge, John, Timothy Griffin, and Travis Pratt. 2001. “Considering Hierarchical models for research on inmate behavior: predicting misconduct with multilevel data.” Justice Quarterly 18:203-232.


1. This research was supported by grant 96-IJ-CX-0015 awarded by the National Institute of Justice. Points of view in this document are those of the author and do not necessarily represent the official position or policies of the granting agency. Special thanks to Dr. Julie Horney, PI, for her invaluable help during this process.