Article

Men and Masculinities

1-25
ª The Author(s) 2017
Hegemonic Masculinity, Reprints and permission:
sagepub.com/journalsPermissions.nav
Violence, and Gender DOI: 10.1177/1097184X17696171
journals.sagepub.com/home/jmm

Equality: Using Latent

Class Analysis to
Investigate the Origins
and Correlates of
Differences between Men

Rachel Jewkes1,2 and Robert Morrell3

Abstract
Messerschmidt and colleagues have pioneered work in criminology using masculi-
nities theory, yet many researchers in the field have not engaged with the possibility
that the different patterning of correlated violent, sexually risky, and antisocial
behaviors may reflect a disaggregation of the category of men into multiple mas-
culinities. This lens can help understand men’s violence and enable intervention
targeting. We analyzed household survey data and identified three classes of men
according to their use of violence and correlated behavior. Associations between
masculinity categories and other acts of violence (against women), gender attitudes,
and sexually transmitted diseases showed a dose–response relationship across the
masculinity categories. Structural equation modeling showed how the psychological
variables mediated pathways between exposure to trauma and teasing in childhood
and the more violent masculinity categories. Our analysis provides a bridge between

1
Gender & Health Research Unit, Medical Research Council, Pretoria, South Africa
2
School of Public Health, University of the Witwatersrand, Johannesburg, South Africa
3
University of Cape Town, Cape Town, South Africa

Corresponding Author:
Rachel Jewkes, Gender & Health Research Unit, Medical Research Council, Private Bag X385,
Pretoria 0001, South Africa.
Email: rjewkes@mrc.ac.za
2 Men and Masculinities XX(X)

gender analysis (with intersectionality) and the psychoanalytic in understanding

men’s violence. This is important for interventions to prevent men’s violence against
women and other men and support arguments for targeting violence prevention
interventions.

Keywords
masculinities, latent class analysis, gender-based violence, rape, transactional sex,
sexual risk-taking, crime, childhood trauma

There is a growing body of quantitative research on male ideas and behavior which
shows that many men have gender inequitable attitudes and engage in forms of risky
and exploitative sexual behavior, general acts of violence, and violence against
women. This research increasingly suggests that these ideas and behaviors track
together such that men who rape are also more likely, than those who do not, to
physically abuse and control women, have transactional sex, weapons, fight with
other men and be gender inequitable in their ideas, and the same applies for these
other behaviors (Jewkes and Morrell 2010; Jewkes et al. 2011; Dunkle et al. 2006,
2007; Jewkes et al. 2012b; Fulu et al. 2013; Jewkes et al. 2013; Barker et al. 2011).
Thus, among men, there seems to be a clustering of antisocial and violent ideas and
behaviors and gender inequitable attitudes.
The observation that behaviors cluster, and thus that some men engage in more
violent and antisocial behaviors of a range of types than others, has parallels which are
born out in observations from criminology. These are that a small proportion (1
percent to 10 percent in various studies) of men account for over half of all convic-
tions. For example, a Swedish study found that 1.0 percent of the total population
(overwhelmingly men) accounted for 63.2 percent of all convictions (Falk et al. 2014)
and the Cambridge Study in Delinquent Development found that 7 percent of their
sample of 411 men accounted for over half of the convictions (Farrington et al. 2006).
A notable feature of these papers, and the broader discipline, is that with the exception
of a few authors (notably Messerschmidt 1993, 1997, 2000, 2012; Newburn and
Stanko 1994), the gendered nature of male crime has been ignored or is a topic of
active disagreement (Collier 1998; Hood-Williams 2001; Mullins 2010). There has
usually been no attempt at theoretical linkage, even within feminist criminology, apart
from observations that violent and antisocial behavior is much more common among
males, between these types of contributions in the criminological literature to the
literature of critical men’s studies (Mullins 2010). Yet the analysis of clustering of
violent and sexually risky behaviors and their roots in gender inequity described above
suggests that there may be important connections and this was very comprehensively
argued by Messerschmidt (1993) in his classic text.
A reading of the clustering of violent and antisocial behavior is that it empiri-
cally supports ideas of multiple masculinities and the existence of differences
Jewkes and Morrell 3

among men expressed through positions on gender, use of violence toward and
control of women, sexual risk-taking, and use of violence and expressions of
toughness among men. Connell’s analysis of masculinities, within the broader
framework of gender and power, is an important theoretical reference point in
this regard (Connell 1987; Messerschmidt 1993). Connell argues that masculi-
nities are multiple in a given society, but one position, a particular configuration
of male attitudes and practices, is hegemonic and serves to perpetuate gender
inequality (men’s domination over women) and assumes power over other (often
minority groups of) men. The concept of hegemonic masculinity has been widely
used, as well as debated and refined (R. W. Connell and Messerschmidt 2005),
but the basic idea is that hegemonic masculinity is a culturally idealized form of
masculinity and “is both a personal and a collective project” (Donaldson 1993, p.
645). A core element of the construction of hegemonic masculinity is heterosexu-
ality, and to a greater or lesser extent, hegemonic masculinity is constructed as a
gender position which is as much “not gay” as it is “not female.” With these
characteristics, one can assume that essential features of the current configuration
of hegemonic masculinity include performance of heterosexuality, dominance and
control over women, and competition among men. Although masculinity is fluid,
and there has been a notable shift in northern masculinities and among the middle
class in the South (Anderson 2009), in many parts of the world, these constitutive
elements are still in place and have been empirically described, even if there is
evidence of gender change (Wood and Jewkes 2001).
Connell also argued that masculinities are fluid and dynamic and may also be
seen as positions that are occupied situationally, in that the position occupied,
practices, and values espoused in one context may be different from those of another.
Therein this poses a problem for quantitative research into masculine positions, as
the use of survey assessment and analysis “fixes” positions, in a snapshot, which
could be seen as anathema for a concept which is theorized as dynamic. A further
problem is that there are debates about hegemonic masculinity that have become
locked in an impasse of seeing the concept either as aspirational or as descriptive of
general, society-wide power (R. W. Connell and Messerschmidt 2005; Beasley
2008; Howson 2014). If hegemony is not expressed in action flowing from ideal
but only describes society-wide power, it would be very hard to investigate quanti-
tatively. This has been viewed as an obstacle to quantitative sociological research on
masculinities and men’s violence and antisocial behavior (Mullins 2010). It in turn
limits the analytical utility of hegemonic masculinity because it suggests that it can
only be described and never measured, that it is mercurial, and that it is hard to grasp.
This has implications for many gender organizations who work for gender equality
and who assume that a particular form of masculinity is hegemonic and then seek to
reform or transform this construction or configuration. If their assumption is wrong,
their interventions might be wrongly targeted (cf. Jewkes et al. 2015).
However, there are quantitative analytical methods that can be applied to a data
set that enables an empirical analysis of male ideals and practices. One example is
4 Men and Masculinities XX(X)

the derivation of latent variables and classes, which requires coherence of measured
practices and/or attitudes in order to be successful. This has not been previously
applied to these research questions and thus it has not been previously shown that it
is possible to identify particular clusters of men’s actions and attitudes and to create
a framework that distinguishes groups of men in relation to gender equality. The
successful use of these methods to identify and describe groups of men who are
either more or less violent in itself enables us to move beyond the previously
mentioned impasse, but also may have major implications for gender interventions,
potentially targeting and tracking impact. It can also enable us to see how multiple
masculinities operate and have effect in the social world and investigate how actions
and aspirations interact and trace their developmental origins.
The aim of this article is first to use statistical methods to identify different
masculine positions based on the probability of engagement in different male
practices in a large data set from a household survey conducted with adult men
in two provinces of South Africa. Second, to describe the statistical correlates of
these positions both through examining childhood and psychological factors asso-
ciated with each and through the use of structural equation modeling. The third aim
is to describe the correlations of these gender (masculinity) positions with other
male attitudes and practices and through this to discuss how these masculinity
positions affect current understandings of public health challenges and
interventions.

Method
A cross-sectional study was conducted in 2008 in three adjoining districts of the
Eastern Cape and KwaZulu-Natal provinces of South Africa, spanning rural areas,
commercial farmland, towns, and a major city. Detailed methods are described
elsewhere (Jewkes et al. 2009). A population-based sample of men aged eighteen
to forty-nine years was identified using a multistage proportionate sampling design.
Census enumeration areas (EAs) were the primary sampling units, and the sample
was drawn from those in the 2001 census by Statistics South Africa, stratified by
district and proportionate to population size. Within each EA, we mapped all house-
holds, randomly sampled twenty households, enumerated the eligible men who slept
there within the previous night, and randomly selected one. There was no replace-
ment of households without an eligible man.
Among 222 sampled EAs, 2 (0.9 percent) had no human dwellings. In the remain-
ing 220, 1 (0.45 percent) was excluded because gatekeepers refused access, and 4
(1.8 percent) were excluded because no homes with eligible men were identified
following multiple visits at different times of day. We sampled a total of 4,473
visiting points. Of these, 822 (18.4 percent) could not be rostered for eligibility after
a minimum of three attempts at contact. Among the remaining 3,651 visiting points,
1,353 (37.1 percent) were found to contain no eligible man, while 2,298 (62.9
percent) contained at least one eligible man. We thus estimated a total eligible
Jewkes and Morrell 5

population of 2,815 men in our sampling frame. Of this estimated population, 27

percent could not be contacted (estimated n ¼ 760), 10.5 percent (n ¼ 296) refused
to participate, 0.7 percent (n ¼ 21) agreed to complete interviews but then either
withdrew or failed to provide any usable data, and 61.7 percent (n ¼ 1,737) com-
pleted the questionnaire. Interviews were thus completed in 215 (97.7 percent) of the
220 eligible EAs and in 1,737 (75.6 percent) of the 2,298 enumerated and eligible
households. Of these, 1,220 (70.2 percent) also provided dried blood spots (DBS) for
HIV testing. These participation rates compare favorably to those obtained in other
population-based surveys in South Africa.
Participants self-completed a survey using audio-enhanced personal digital assis-
tants (PDA). The text of each question and associated answer choices were presented
on the PDA screen, while an accompanying voice recording read the question and
answers aloud. Participants listened through headphones and answered by tapping
their answer choice with a stylus. The questions could be read, or listened to, in
isiXhosa or isiZulu and English. This environment provided complete privacy for
respondents. Fieldworkers were nearby during questionnaire completion so they
could assist respondents if requested, but interviews were otherwise private. Ques-
tions included demographics, socioeconomic status (SES), childhood variables,
relationship and sexual histories, psychological measures, and detailed questions
on perpetration of violence.

Measures
The questionnaire included categorical variables measuring age and income. Ques-
tions on men’s childhoods included items on whether and how often their father was
at home. Scales measured men’s perceptions of the kindness of their mother (three
items, Cronbach’s a ¼ .75) and father (four items, Cronbach’s a ¼ .87). A typical
item was “I had a loving relationship with my mother/father while I was growing
up.” A four-point Likert-type response scale was used. Men were asked if they had
been teased or harassed as a child.
Data on adverse experiences before the age of eighteen were collected using a
locally modified version of the short form of the Childhood Trauma Questionnaire
(Jewkes et al. 2010; Bernstein et al. 2003). We assessed five dimensions of adver-
sity: emotional neglect, emotional abuse, physical neglect/hardship, physical abuse,
and sexual abuse using a four-point response scale (never, sometimes, often, and
very often; Cronbach’s a ¼ .79). A typical question was “before I reached eighteen
one or both of my parents were too drunk to take care of me.” Men were asked if
they had ever been raped by a man (persuaded or forced to have sex when you did
not want to).
Data were collected on two dimensions of psychopathy. Blame externalization
and Machiavellian egocentricity are two core affective and interpersonal deficits of
psychopathy (Lilienfield 2005). Blame externalization is a perception of the world
as hostile and others being at fault for one’s problems and Machiavellian
6 Men and Masculinities XX(X)

egocentricity is a measure of narcissism and ruthless attitudes toward others (Lilien-

field 2005). Thirteen questions on Machiavellian Egocentricity and Blame Externa-
lization subscales of the Psychopathic Personality Inventory–Revised (PPI-I) were
included. The Cronbach’s a for the scales together were .83. A typical item on the
Machiavellian Egocentricity subscale was “I get mad if I don’t receive special favors
I deserve” and on the Blame Externalization subscale was “I have often been
betrayed by people I trust.” Each has a four-level response option (false, mostly
false, mostly true, and true). We dichotomized the scales and present the proportion
scoring in the upper third of the scale versus the lower two-thirds. For blame
externalization, 28.4 percent were in the upper third and for Machiavellian egocen-
tricity, 18.5 percent were in the upper third. These were adapted and reproduced by
special permission of the Publisher Psychological Assessment Resources, Inc.,
16204 North Florida Avenue, Lutz, FL 33549, from the PPI-I by Scott O. Lilienfield,
PhD, Copyright 2005 by PAR, Inc. Further reproduction is prohibited without the
permission of PAR, Inc.
We asked four items to measure empathy, adapted from Abbey et al (2006;
Cronbach’s a ¼ .80). A typical item was “I am often touched by things that I see
happen.” These had a five-point response scale (doesn’t describe me well—
describes me well). Perceptions of life success were assessed with the following
question: “If you compare your life circumstances overall now with those of the
people you grew up with, would you say you have done much better for yourself,
somewhat better, the same, less well, much less well?”
Attitudes toward gender relations were measured using ten items from the Gender
Equitable Men (GEM) scale (Pulerwitz and Barker 2008; Cronbach’s a ¼ .78). A
typical item is “There are times when a woman deserves to be beaten.” A high score
denotes more equitable attitudes. The school bullying perpetration score was an eight-
item scale used to measure experiences with sexual bullying at school with four-level
response options (never, sometimes, often, and very often; Cronbach’s a ¼ .76). These
questions were developed for the study. A typical item was “My school friends and I
were a group and we would put pressure on a girl to date one of us until she agreed.”
We asked eleven items about lifetime experiences of participation in crime.
These were modified for the local context from Tremblay et al. (1995) who
developed them as a measure of delinquency in childhood. Eight of the items
related to theft (Cronbach’s a ¼ .81) and a typical item was “how often have
you stolen an animal from someone?” The response options were never, once,
two to three times, and more often. Men were also asked about weapons own-
ership, possession, and arrests.
Recent alcohol consumption in the past twelve months was assessed through a
question on frequency of having five or more drinks per drinking day. Drug use was
assessed through a question on how often the man had smoked dagga (cannabis) in
the past twelve months. This drug was selected, as it grows well locally and is cheap
and so is most widely used, and most men who use other drugs use this too. Men
were asked if they had ever been in a gang.
Jewkes and Morrell 7

Practices of gender relations were measured through questions about number of

sexual partners, whether men had had makhwapheni in the past year or ever (a
usually secret concurrent partner), and about transactional sex with women, defined
as sex that was primarily motivated by a desire for material gain on the part of the
woman. This was defined as providing food, cosmetics, clothes, transportation,
items for children or family, school fees, somewhere to sleep, handyman work, or
cash (Jewkes et al. 2012c). Men were asked about lifetime perpetration of physical
intimate partner violence (IPV), using the modified World Health Organization
violence against women instrument (Garcia-Moreno et al. 2005). Specific acts of
violence were asked in five items ranging from slapping to threats with or use of a
weapon. A twelve-item scale assessed men’s control in their relationship, after
Pulerwitz, Gortmaker, and DeJong (2000), as adapted for South Africa by Dunkle
et al. (2004). These items were summed to derive a score (Cronbach’s a ¼ .78). A
typical question was “when I want sex I expect her to agree.”
Rape perpetration was assessed using seven questions developed for the study
and validated through cognitive interviewing, none of which actually used the word
“rape” (Sikweyiya, Jewkes, and Morrell 2007). They were modifications of those
used previously in the Eastern Cape (Jewkes et al. 2006). A typical item was “How
many times have you slept with a woman or girl when she didn’t consent to sex or
after you forced her?” The questions additionally asked about having forced a
(former) girlfriend or wife into sex, having forced a woman who was not a girlfriend
or wife into sex, and having sex with a woman who was too drunk to consent. Two
questions assessed multiple perpetrator rape perpetration. Men were also asked if
they had ever raped a man.

HIV
Participants were also asked for a finger-prick blood spot that was dried for HIV
testing. DBS were tested for HIV with a screening ELISA (Genscreen Bio-Rad,
Steenvorde, France) and positive results were confirmed with a second ELISA
(Vironostika, bioMérieux, Marcy d’Etoile, France). Men who declined to give blood
for HIV did not differ significantly from those who did in age, race, circumcision,
history of consensual men having sex with men (MSM) activity, or perpetration of
sexual violence. However, men who had completed secondary education were sig-
nificantly less likely to provide blood (66.6 percent vs. 75.1 percent, p ¼ .0003).

Ethics
Ethics approval was granted by the Medical Research Council’s Ethics Committee.
Participants signed informed consent separately for interviews and DBS. An incen-
tive of ZAR 25 (*US$3.5) was offered for each component, giving a total of ZAR
50 for both. The PDAs ensured that participant answers were entirely anonymous,
and we retained no identifying information on any participant. This was necessary to
8 Men and Masculinities XX(X)

Table 1. Measurement Probabilities of Variables Loading on the Masculinity Classes.

Measurement Probabilities

High Medium Lower Violence

Variable (24.7% of Men) (29.6% of Men) (45.7% of Men)

Has illegal gun 0.476 0 0

Has a weapon 1 .091 .093
>1 Episode of physical or sexual IPV 0.573 .406 .099
Has a khwapheni 0.8 .845 .268
Has had sex with a sex worker 0.4 .24 .015
Has had transactional sex 0.851 .86 .354
Drug use in the past year 0.631 .462 .157
Note: IPV¼ intimate partner violence.

protect participants reporting illegal activities from possible repercussions. It also

meant that HIV results could not be given. Free HIV testing is widely available from
government clinics in South Africa and all participants were advised to learn their
status.

Data Analysis
The sample was self-weighting. Questionnaire data were linked to HIV data using
anonymous codes. Most of the analyses were performed using Stata 12.0 and
accounted for the two-stage sampling structure, with stratification by district and
data clustered in EAs. No imputation methods were used to replace missing data.
The latent class analysis was conducted in MPlus, ver. 7 with R used to more
easily present the results. The initial variables used for class determination were
those which had been hypothesized from previous research to measure male beha-
viors and attitudes which were indicative of potential types of masculinities. We
tested four- and three-class models. We tested a range of variables which included
those shown in Table 1 as well as gang membership, problem drinking, not paying
maintenance for a child, gender attitudes, controlling behaviors, having raped, and
bullying. We also tested a range of positive male behaviors, but these did not load
coherently into classes. The best fitting classes were three in number and the fol-
lowing model fit were the statistics: AIC ¼ 12,466.77 and BIC ¼ 12,592.35, per-
centage of two-way marginal residuals > 4:1 and sum of two-way marginal residuals
large, out of twenty-one pairs:1. We have numbered the most violent #1 and most
moderate category #3.
The distributions of sociodemographic characteristics and childhood variables by
masculinity category were summarized as percentages (or means) using standard
methods for estimating confidence intervals from complex multistage sample sur-
veys (Taylor linearization). Pearson’s w2 was used to test two-way associations
Jewkes and Morrell 9

between categorical variables. A multinomial regression was used to model the

social and demographic characteristics associated with each masculinity category.
A second multinomial regression model was built to show associations between
childhood characteristics and the masculinity categories, adjusted for age, race, and
income. A third such model included psychological variables, adjusted for childhood
and social and demographic variables. For each of the variables in Table 5, a
separate multinomial regression model was built with the relationship variable,
violent practice, or sexual health problem as the dependent variable and the mascu-
linity categories, age, race, and income as independent variables.
The structural equation model (SEM) was built in Stata 13.0 to examine pathways
to the masculinity categories. The variables included were those used for the psy-
chological variables multinomial regression model and the other nonsignificant
childhood variables, with the exception that income was replaced by a measure of
SES which had greater variability and thus measurement properties for the model.
These variables included all those in the data set that may have been considered as
causal. The SES measure used three asset ownership variables, home wall building
materials, frequency of hunger, and not eating meat due to lack of money, and ease
of emergency resource mobilization (Cronbach’s a ¼ .643). Initially, a measurement
model was fitted and then a maximum likelihood estimation with missing values.
Nonsignificant pathways were deleted and modification indices reviewed and
adjusted to optimize model fit. The model presented is adjusted for race. The model
fit for the final was good as shown by the fit statistics (p ¼ .094, RMSEA ¼ .018,
CFI ¼ .995, and TLI ¼ .986).

Results
The best fit was found for a three-class model encompassing the set of seven violent
or sexually risky male behaviors as shown in Table 1. These included two weapons
possession measures, illegal gun possession and having a weapon; one intimate
partner violence measure, having a khwapheni (concurrent additional sexual part-
ner); two measures of providing material resources for sex, having had sex with a sex
worker and having had transactional sex and having used illegal drugs in the past
year. The model fit statistics showed that this fitted the data well.
The table shows that the greatest difference between the high and other two
categories was in weapons ownership. The middle category resembled the high one
in that most men (over 80 percent) had had transactional sex and had a khwapheni.
For having had sex with a sex worker, past year drug use, and intimate partner
violence, the middle category probability was roughly midway between that of the
high and the low categories. Some of the men in the low violence category had
previously used violence or were ready to do so. One in ten had a weapon and had
perpetrated more than one episode of physical or sexual violence against a woman
partner. However, they were much more law abiding, in that none had an illegal gun
10 Men and Masculinities XX(X)

and fewer than 2 percent had ever had sex with a sex worker, and only 16 percent had
used drugs (mainly cannabis) in the past year.
The procedures above resulted in the men falling into three categories: 24.7
percent were in the highly violent/antisocial category, 29.6 percent in the medium
violence category, and 45.7 percent in the lowest violence/most pro-social category.
Examination of the social and demographic characteristics of the men shows that the
categorization of men was associated with their age, race, income, education, and
marital status. In the sample overall, half the men interviewed were under twenty-
five years of age (see Table 2). Men in the high- and mid-violence categories were
younger than those in the low-violence category. There were racial differences, with
those in the high-violence category significantly more likely to be colored than those
in the low-violence category. There were also differences in income, and those in the
high-violence category had significantly higher income than those in the low-
violence category, with differences in income between the middle- and low-
violence categories only being visible for the low paid group (vs. unpaid). Men in
the low-violence category were less well educated than those in the higher category
and less likely to be cohabitating.
Table 3 presents data on some of the childhood circumstances and trauma expo-
sures of men in the different categories. Overall, the categorization of men was
associated with their experience of being teased as children, how kind their father
was and their exposure to trauma in childhood overall based on a score which
encompassed physical, sexual and emotional abuse, and physical and emotional
neglect. Men in the high- and medium-violence levels were both significantly more
likely to have been teased and also to have experienced childhood trauma than those
in the low-violence group. Men in the high-violence group were significantly less
likely to rate their father as kind than those in the low-violence group.
The table also shows that exhibiting high levels of Machiavellian egocentricity,
blame externalization, and empathy were all associated with the masculinity cate-
gories, after adjustment for social and demographic characteristics and childhood
trauma exposures. In the case of Machiavellian egocentricity and blame externaliza-
tion, a dose–response relationship is visible across the categories.
A structural equation model (SEM) of antecedents of the masculinity categories
is presented in Figure 1 and Table 4. The model shows a path linking SES and
masculinity categories mediated by empathy, such that men of higher SES are more
empathetic and demonstrate less violent and more pro-social behavior. Paths
between SES and masculinity categories were also mediated by childhood trauma
and teasing, such that men of lower SES would experience more teasing and child-
hood trauma. Both of these directly enhanced the likelihood of being in a more
violent masculinity category. However, the path between childhood trauma and the
masculinity categories was further mediated by Machiavellian egocentricity and
blame externalization, such that men experiencing more childhood trauma were
more likely to have higher scores on these and hence a more violent masculinity
categorization. A path between experience of teasing and more violent masculinity
 Table 2. Social and Demographic Factors Associated with the Masculinity Categories.

Multinomial Model of Social and Demographic Factors Associ-

ated with Masculinity Category

High versus Low Medium versus Low

High (24.7% Medium (29.6% Lower Violence
Variable of Men; %) of Men; %) (45.7% of Men; %) p Value RRR 95%CI p Value RRR 95%CI p Value

Age: eighteen to 54.7 48.8 51.6 .0002 1.00 1.00

twenty-four years
Twenty-five to 32.7 35.6 26.7 0.95 0.69 1.30 .73 1.35 1.01 1.81 .04
thirty-four years
Thirty-five years 12.6 15.6 21.7 0.48 0.31 0.73 .001 0.80 0.57 1.13 .21
and over
Race: African 77.8 88.3 86.8 <.0001 1.00 1.00
Colored 8.2 4.1 1.9 3.82 1.75 8.35 .001 1.87 0.93 3.75 .08
Indian 12.2 6.4 9.6 1.37 0.81 2.32 .23 0.75 0.45 1.24 .26
White 1.9 1.2 1.8 1.66 0.66 4.15 .28 0.98 0.32 3.03 .97
Income: None 39.0 47.6 55.4 <.0001 1.00 1.00
Up to ZAR 45.2 40.9 32.8 2.02 1.48 2.76 <.0001 1.46 1.11 1.93 .007
2000/month
Over 15.9 11.5 11.6 1.91 1.20 3.04 .007 1.13 0.72 1.78 .60
Education: Matric 43.0 41.6 38.6 .024 1.08 0.83 1.39 .56 1.15 0.89 1.48 .30
plus
Marital status: 18.7 19.7 26.2 .0008 1.00 1.00
Married
Cohabiting 13.4 14.3 8.5 2.22 1.45 3.39 <.0001 2.19 1.41 3.39 .001
Divorced/ 4.5 3.0 3.5 1.88 0.98 3.63 .06 1.28 0.63 2.57 .50
widowed
Single 63.5 63.1 61.9 1.35 0.98 1.85 .07 1.30 0.97 1.76 .08
Note: CI ¼ confidence interval; RRR ¼ relative risk ratio.

11
12
Table 3. Relationship practices, violent behaviour and outcomes and sexual health associated with the masculinity categories.

Multinomial’ Model of Childhood Factors Associated with Masculinity

Category, Adjusted for Age, Race, and Income

High versus Low Medium versus Low

Lower
Variable High Medium Violence p Value RRR 95%CI p Value RRR 95%CI p Value

Childhood characteristics
Mother rarely or never at home 41.43 42.35 38.89 .46 0.98 0.71 1.33 .88 1.17 0.89 1.54 .25
Father rarely or never at home 69.21 68.38 65.91 .47 0.95 0.68 1.34 .78 1.02 0.78 1.35 .87
Teased by others 66.82 57.53 45.84 <.0001 2.31 1.72 3.10 <.0001 1.53 1.17 1.99 .002
Father kindness score (mean) 11.11 12.12 12.2 <.0001 0.95 0.91 0.99 .02 1.02 0.98 1.07 .28
Mother kindness score (mean) 10.73 10.81 10.82 .401 1.03 0.94 1.13 .50 1.02 0.94 1.10 .68
Childhood trauma score 20.65 19.39 18.36 <.0001 1.07 1.04 1.10 <.0001 1.03 1.00 1.06 .03

Multinomial’ model of associations between psychological characteristics

and masculinity categories, adjusted for age, race, income, and
Psychological characteristics childhood characteristics

Machiavellian egocentricity: Cut 27.6 21.6 10.8 <.0001 2.47 1.72 3.52 <0.0001 2.03 1.42 2.91 <0.0001
Blame externalization: Cut 54.7 38.3 24.0 <.0001 3.17 2.28 4.05 <0.0001 1.89 1.41 2.53 <0.0001
Empathy 48.2 49.5 57.6 .004 0.97 0.94 1.00 0.046 0.97 0.95 1.00 0.031

Note: CI ¼ confidence interval.

Jewkes and Morrell 13

Figure 1. Structural equation model of antecedents of the masculinity categories (higher

categories are less violent).

categorization was also mediated through blame externalization. The model also
shows that having a kind father can provide protection against teasing, and this also
directly impacts (reduces) the risk of becoming a more violent man.
The more violent masculinity categories were seen in the multinomial models to
be significantly associated with both self-reporting more controlling practices in
men’s relationships and expressing less equitable gender attitudes (Table 5). They
were also associated with a range of violent behaviors (rape of a man or woman,
theft or robbery, and bullying others), experience of being raped by a man, and the
outcome of having ever been in prison. In all of these, there was a dose–response
relationship. Men in the more violent masculinity categories were more likely to
have ever had a genital ulcer and penile discharge. They were not more likely to have
HIV or always use a condom.

Discussion
In this household survey, it was possible to use latent class analysis to identify three
coherent categories of men through testing a candidate set of male behaviors and
attitudes. This technique is not often used in gender analysis. The set of variables
which provided the best fit very closely matches those which have been observed to
track together in previous South African research (Jewkes and Morrell 2010; Dunkle
et al. 2006, 2007; Jewkes et al. 2006). The analysis supports previous observations
and the derivation of a latent gender inequitable masculinity variable using structural
equation modeling (SEM) in a different data set from South Africa (Jewkes et al.
2016), which have shown that violence, sexually risky, and antisocial behaviors
often track together. We have shown here that although most South African men
commit violent acts (and fall in the mid- or high-violence categories), nearly half of
14
Table 4. Path Model: Direct Effects, Disturbance Variances, and Equation-level Goodness of Fit.

Standardized [95% Confidence

Parameter Coefficients SE z p > |z| Interval]

Direct effects
SES ! empathy 0.466 0.046 10.48 <.0001 0.379 0.554
SES ! childhood trauma 0.227 0.042 5.44 <.0001 0.309 0.145
SES ! teased 0.008 0.004 2.09 .036 0.016 0.001
Kind father ! teased 0.017 0.004 4.91 <.0001 0.024 0.010
SES ! race 0.054 0.005 10.93 <.0001 0.045 0.064
Childhood trauma ! Machiavellian egocentricity 0.226 0.028 8.06 <.0001 0.171 0.280
Race ! Machiavellian egocentricity 0.907 0.220 4.13 <.0001 1.34 0.477
Empathy ! masculinities categories 0.007 0.004 2.03 .043 0.000 0.014
Childhood trauma ! masculinities categories 0.010 0.004 2.50 .012 0.018 0.002
Teased ! masculinities categories 0.119 0.039 3.03 .002 0.195 0.042
Machiavellian egocentricity ! masculinities categories 0.017 0.004 4.30 <.0001 0.024 0.009
Blame externalization ! masculinities categories 0.031 0.004 8.35 <.0001 0.039 0.024
Kind father ! masculinities categories 0.015 0.006 0.264 .008 0.004 0.026
Race ! masculinities categories 0.093 0.029 3.180 .001 0.150 0.036
Race ! Kind father 0.556 0.132 4.210 .000 0.297 0.815
Childhood trauma ! Blame externalization 0.191 0.030 6.47 <.0001 0.133 0.249
Race ! Blame externalization 0.886 0.221 4.000 .000 0.452 1.319
Teased ! Blame externalization 2.51 0.294 8.53 <.0001 1.93 3.09

Disturbance variances Estimate SE [95% confidence

interval]

Empathy 29.17 1.06 27.16 31.32

Childhood trauma 25.19 0.882 23.52 26.98
Teased 0.244 0.008 0.228 0.261
(continued)
 Table 4. (continued)

Disturbance variances Estimate SE [95% confidence

interval]
Machiavellian egocentricity 30.39 1.09 28.32 32.6
Masculinities categories 0.567 0.019 0.531 0.607
Blame externalization 31.18 1.11 29.07 33.44
Kind father 11.87 0.415 11.09 12.72
Race 0.380 0.0130322 0.3549992 0.406
SES 10.04 0.366 9.35 10.79
Equation-level goodness of fit R2
Empathy .0697
Childhood trauma .0249
Teased .0177
Machiavellian egocentricity .0547
Masculinities categories .1383
Blame externalization .0980
Kind father .0105
Race .0727
Note: mc2 is the Bentler–Raykov squared multiple correlation coefficient. SE ¼ standard error; SES ¼ socioeconomic status.

15
16
Table 5. Social and Demographic Factors Associated with the Masculinity Categories.

RRR/coefficient Calculated in multinomial Models with the Relationship Prac-

tice, Violent Behavior, or Sexual Health Problem as the Independent Variable
Column Percentages Showing % of Men and Masculinity Categories as an Exposure, Adjusted for Age, Race, and
with the Variable in Each Category Income

High versus Low Medium versus Low

Lower
Variable High Medium violence p Value RRR/coeff 95%CI p Value RRR/coeff 95%CI p Value

Relationship practices
Relationship control scale 13.4 13.8 12.8 <.0001 0.60 0.22 0.98 .002 0.95 0.60 1.31 <.0001
(mean)
GEM scale 11.6 11.4 12.0 .016 0.39 0.74 0.03 .034 0.50 0.83 0.17 .003
(mean)
Violent behavior and outcomes
Any rape 51.9 29.8 12.1 <.0001 8.44 6.14 11.62 <.0001 3.03 2.22 4.14 <.0001
Theft or robbery three 65.1 29.1 17.5 <.0001 10.35 7.44 14.40 <.0001 2.05 1.49 2.81 <.0001
plus times
Rape of a man 6.6 2.3 1.2 <.0001 6.39 2.32 17.59 <.0001 2.73 0.94 7.94 .065
Raped by a man 14.1 12.1 5.3 <.0001 3.19 2.02 5.04 <.0001 2.45 1.57 3.81 <.0001
Bullied others many times 63.4 39.8 22.7 <.0001 6.04 4.57 7.97 <.0001 2.29 1.76 2.97 <.0001
Ever been in prison 15.0 6.3 3.9 <.0001 6.41 3.74 11.01 <.0001 2.20 1.23 3.94 .008
Sexual health problems
Ever genital ulcer 35.0 33.5 22.6 <.0001 2.42 1.81 3.25 <.0001 1.75 1.33 2.31 <.0001
HIV seropositive 17.4 20.1 17.5 .53 1.31 0.84 2.05 .24 1.12 0.74 1.69 .59
Never had a penile 57.4 67.5 76.9 <.0001 0.36 0.27 0.48 <.0001 0.63 0.47 0.82 .001
discharge
Always uses a condom 35.0 38.5 39.3 .31 0.78 0.58 1.05 .10 0.97 0.74 1.28 .85

Note: CI ¼ confidence interval; GEM ¼ gender equitable men.

Jewkes and Morrell 17

all men lean toward more respectful and less violent behavior toward women and are
more law-abiding. However, among the 55 percent of men who are sometimes
violent, more than half of this group are very violent, criminal, and sexually prolific.
Objections to the tautological nature of the concept of masculinity and its relation-
ship to gender-specific or patterned behaviors such as violence (Collier 1998; Hood-
Williams 2001) have called into question the value of gender analysis. In this article,
we seek to renew the claim for the importance of the collective gendered analysis of
male behaviors and attitudes by showing how it can shed light on such behaviors and
prevent a descent either into aggregated and undifferentiated or highly individua-
lized analyses of violent behavior.
We have explored the antecedents of this categorization. Our findings strongly
support a developmental psychological argument that men’s adult practice is very
importantly influenced by their childhoods (Fonagy and Target 2003; Contreras
et al. 2011; Walker et al. 2011; Stouthamer-Loeber et al. 2001). We have also shown
the influence of structural factors (poverty), which strongly supports the well-
developed arguments about the importance of intersectionality of gender and
resource access (mostly called class in the literature) (Messerschmidt 1993; Bour-
gois 1996). While we have only a measure of current SES, given that half of the men
where under twenty-five, it is unlikely that current and childhood SES for most men
would have been very different. We have shown how their positioning in the more
violent categories have been influenced by their experience of childhood trauma,
being teased, and paternal kindness (which was protective). The SEM has confirmed
theoretical perspectives on the processes through which this is likely to have
occurred (e.g., Fonagy and Target 2003; Mathews, Jewkes, and Abrahams 2011;
Frosh, Phoenix, and Pattman 2002). It has shown that exposure to trauma impacts on
personality development as seen here in scores on measures of Machiavellian ego-
centricity and blame externalization, both of which are dimensions of Lilienfeld’s
psychopathy scale (Lilienfield 2005), and thence on positioning in more violent
masculinity categories. We have also shown how being positioned in a more violent
and antisocial category predicts being more gender inequitable, more controlling
with their relationship with their main partner, and prior engagement in a range of
violent behaviors including rape, repeated theft or robbery, bullying, and having
been in prison. It also predicts having had a sexually transmitted infection. This
analysis is important, as it provides an analysis bridge between the psychoanalytic
and gender analyses of men’s involvement in crime and violence and thus has the
potential for advancing understanding of men and violence more broadly.
The male positions described through the latent class analysis process are not
based on a random selection of male behaviors and nor are they based solely on male
violent-related behaviors. Research on transactional sex and the male provider role,
particularly from South Africa, has shown a strong correlation between these prac-
tices and male use of violence and engagement in crime (Dunkle et al. 2007; Jewkes
et al. 2012a, 2012b). Similarly, analysis of rape perpetration in South Africa and in
the United Nations multicountry study on men and violence in Asia and the Pacific
18 Men and Masculinities XX(X)

has shown that this practice is strongly associated with transactional sex, as well as a
range of other male violent practices against women and with other men, including
gang membership and having weapons (Jewkes et al. 2011, 2013). Connell’s (1987)
concept of hegemonic masculinity serves as an analytical instrument to identify
those attitudes and practices among men who perpetuate gender inequality, involv-
ing both men’s domination over women and the power of some men over other
(often minority groups of) men.
Masculinities are multiple, fluid, and dynamic, and hegemonic positions are not
the only ones available in a given society. Our presentation of three classes of
masculinity, or masculinity positions, provides an illustration of this. Although this
model best fitted out data, the data set was not large enough, nor analysis sophisti-
cated enough to exclude gradations within these classes, nor other masculinities
which may be less common within the general population. We would also not like
the analysis to be interpreted as indicating that masculinities are fixed, although for
these men, drawing on data provided in interviews at a life snapshot point in 2008,
we have fixed them in categories based on their reported practices. Several of the
variables used in so doing may change over time and so individual men might move
between masculinity categories.
Previous authors have identified forms of destructive and exaggerated masculi-
nities (or hypermasculinity) (Herek 1987; Torbenfeldt Bengtsson 2015), among
socially marginalized men which emphasize power and force (Totten 2003; Bour-
gois 1996). These are akin to the most violent category of masculinity and might be
understood as a form of protest masculinity that rejects societal proscriptions against
conformity to laws and the use of violence. We have shown in the SEM that indeed
these are a product of poverty, combined with harshness of childhood experiences
which themselves are much more commonly experienced by boys growing up in
poverty. Other authors have argued that they are not entirely separate from hege-
monic masculinity to the extent that they emerge out of the relationship between
hegemonic ideals and (some) men’s ability to meet them. Their origins lie in adver-
sity, including in violence experiences in childhood which have enduring psycho-
logical impact manifesting in a lack of empathy and remorse which enable acts of
violence while positioning the male actors as themselves victims (Mathews, Jewkes,
and Abrahams 2011; Fulu et al. 2013; Jewkes et al. 2011, 2013; Bourgois 1996).
This is strongly supported by our data.
An obvious question is whether any of the masculinity positions we have
described would be considered as “hegemonic”—and whether we can use any of
the quantitative attributes to identify it. The lowest violence category was the
numerically most dominant single category, but we caution against assuming that
this would then reflect South African hegemonic masculinity. The uneven relation-
ship between hegemony and numerical dominance has been well worn in the liter-
ature (Beasley 2008; R. W. Connell and Messerschmidt 2005; Donaldson 1993).
Furthermore, many authors regard hypermasculinity as an intensification of hege-
monic masculinity rather than as completely distinct. Our analysis suggests that the
Jewkes and Morrell 19

most violent group can be distinguished from the middle group, but there are many
behavioral overlaps. Further, this analysis would lead us to conclude in itself that the
category which is somewhat less violent (the middle one) would thus by definition
reflect hegemonic masculinity. We are strongly drawn to this analysis in South
Africa, there is a considerable body of research on hegemonic masculinity which
elaborates on the situated “contents” of hegemonic masculinity and argues that
demonstrating strength, toughness, and the capacity to use and often actual use of
violence are very much part of hegemonic masculinity in South Africa (Morrell et al.
2013; Wood and Jewkes 2001).
The analysis presented here very clearly demonstrates the challenge of work with
men to build less violent, more pro-social, and gender equitable masculinities, which
is vital to promoting women and men’s health (Jewkes and Morrell 2010; Barker,
Ricardo, and Nascimento 2007). Crudely put, the challenge is to shift men from
being positioned in the high or middle categories toward the lower. Our analysis has
shown that structural interventions to reduce poverty within families and exposure to
trauma in childhood are essentially part of this process. None of the men interviewed
were young enough to have benefited from South Africa’s child support grant which
is payable as a cash transfer for every child in a low-income household. This has
made a contribution to reducing dire poverty in families (Makiwane and Udjo 2006;
Lund 2008). Understanding that gender-based and other criminal violence has a life-
course development trajectory is critical for public sector policy that seeks to build a
safer and more law-abiding society. We recognize that this is not a novel observa-
tion, but our findings are important to the extent that they have demonstrated through
statistical analysis the roots of adult male violent and antisocial practices in men’s
gender identities and linkages to all forms of violence against women and girls. Our
analysis has also provided a link between these and sexually risk-taking behaviors,
and related sexually transmitted infection acquisition, and provides a powerful fur-
ther illustration that one part of the efforts to promote safer sexual practices should
involve changing masculinities (Jewkes and Morrell 2010).
This study has limitations. The process of determining masculinity categories was
limited by the variables in the data set and the number of classes was limited by its
size. We would not suggest that these categories are the only one possible to find
among South African men or even in the data set, but merely that they provided a
good fit for the data given these constraints. Similarly with the SEM, there may have
been other variables which mediate the positioning of men within masculinity cate-
gories, including measures related to social norms, but we did not have them in the
data set and so are constrained by the individual-level measures available. Psycho-
analytic literature on personality disorders suggests that psychopathy is likely to
develop from early childhood experiences and may be genetically influenced
(Johansson et al. 2008; Caspi et al. 2002). Again we did not have genetic information
for the model. The study was cross sectional and so it is impossible to be sure of the
temporal sequence of many of the experiences and behaviors measured. Although
men’s engagement in crime, violence, and sexually risky behavior often starts in the
20 Men and Masculinities XX(X)

teenage years, the childhood factors overall are likely to have preceded these and
preceded the development of personality attributes. The main strength of this study
is that it involved a large randomly selected sample of adult men from the general
population and the survey had a good response rate. The findings should be
generalizable.

Conclusion
This article shows that it is possible to use quantitative sociological techniques to
identify particular clusters of men’s actions and attitudes and to create a framework
that distinguishes groups of men in relation to gender equality. This enables disag-
gregation of the gendered category of “men” on the basis of expressed views and
reported actions. This is critically important if we are to move away from essentialist
research where traits are attributed on grounds of biological belonging and to begin
to use quantitative methods. We have identified and described groups of men who
are either more or less violent which has major implications for gender interventions.
It also gives empirical support to the idea of multiple masculinities and shows how
these operate and have effect in the social world and makes a contribution to debates
about hegemonic masculinity that seek to move beyond the impasse of seeing the
concept either as aspirational or as descriptive of general, society-wide power. We
have shown that hegemony is expressed both in action and in ideal, and in South
Africa, it is a threat to gender equality, and in so doing, we respond to the justifiable
objection that “what hegemonic masculinity actually is” is never illuminated
(Whitehead 2002). The approach that we have used shows how actions and aspira-
tions interact and how these are causally related to SES and early childhood trauma.
Finally, our article provides an important bridge between the highly polarized crim-
inological camps of masculinities and intersectionality and the psychoanalytic. It
shows that multiple masculinities can be identified and they have a varied relation-
ship with perpetration of violent and antisocial behavior and that these have origins
in poverty, childhood trauma, and the enduring impact of this on men’s psyches.

Authors’ Note
The views expressed and information contained in it are not necessarily those of or
endorsed by DFID, which can accept no responsibility for such views or information
or for any reliance placed on them.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research,
authorship, and/or publication of this article: UK Department for International
Jewkes and Morrell 21

Development (DFID), and grant was managed by their local partner Human Life
Sciences Partnership (HLSP). Robert Morrell was assisted to participate in the
project with funding from the National Research Foundation (NRF). This document
is an output from the What Works to Prevent Violence: a Global Programme which
is funded by the UK Aid from the DFID for the benefit of developing countries.

References
Abbey, A., M. R. Parkhill, R. BeShears, A. M. Clinton-Sherrod, and T. Zawacki. 2006.
“Cross-sectional Predictors of Sexual Assault Perpetration in a Community Sample of
Single African American and Caucasian Men.” Aggressive Behaviour 32:54–67.
Anderson, E. 2009. Inclusive Masculinity. The Changing Nature of Masculinities. London,
UK: Routledge.
Barker, G., J. M. Contreras, B. Heilman, A. K. Singh, R. K. Verma, and M. Nascimento.
2011. Evolving Men: Initial Results from the International Men and Gender Equality
Survey (IMAGES). Washington, DC: International Center for Research on Women.
Barker, G., C. Ricardo, and M. Nascimento. 2007. Engaging Men and Boys to Transform
Gender-based Health Inequities: Is There Evidence of Impact? Geneva/Rio de Janiero:
World Health Organization/Institute Promundo.
Beasley, C. 2008. “Rethinking Hegemonic Masculinity in a Globalizing World.” Men and
Masculinities 11:86–103.
Bernstein, D., J. A. Stein, M. D. Newcomb, E. Walker, D. Pogge, T. Ahluvalia, J. Stokes,
L. Handelsman, M. Medrano, D. Desmond, and W. Zule. 2003. “Development and
Validation of a Brief Screening Version of the Childhood Trauma Questionnaire.” Child
Abuse & Neglect 27:169–90.
Bourgois, P. 1996. “In Search of Masculinity: Violence, Respect and Sexuality among Puerto
Rican Crack Dealers in East Harlem.” British Journal of Criminology 36:412–27.
Caspi, A., J. McClay, T. E. Moffitt, J. Mill, J. Martin, I. W. Craig, A. Taylor, and R. Poulton.
2002. “Role of Genotype in the Cycle of Violence in Maltreated Children.” Science 297:
851–54.
Collier, R. 1998. Masculinities, Crime and Criminology. Thousand Oaks, CA: Sage.
Connell, R. 1987. Gender and Power: Society, the Person and Sexual Politics. Palo Alta:
University of California Press.
Connell, R. W., and J. Messerschmidt. 2005. “Hegemonic Masculinity: Rethinking the
Concept.” Gender & Society 19:829–59.
Contreras, M., A. Singh, B. Heilman, B. Barker, and R. Verma. 2011. “Connections between
Early Childhood Experiences of Violence and Intimate Partner Violence.” Early Child-
hood Matters 116:26–32.
Donaldson, M. 1993. “What Is Hegemonic Masculinity?” Theory and Society 22:643–57.
Dunkle, K. L., R. Jewkes, M. Nduna, P. N. Jama, J. B. Levin, Y. Sikweyiya, and M. Koss.
2007. “Transactional Sex and Economic Exchange with Partners among Young South
African Men in the Rural Eastern Cape: Prevalence, Predictors, and Associations with
Gender-based Violence.” Social Science & Medicine 65:1235–48.
22 Men and Masculinities XX(X)

Dunkle, K. L., R. K. Jewkes, H. C. Brown, G. E. Gray, J. A. McIntryre, and S. D. Harlow.

2004. “Gender-based Violence, Relationship Power, and Risk of HIV Infection in Women
Attending Antenatal Clinics in South Africa.” Lancet 363:1415–21. doi:10.1016/S0140-
6736(04)16098-4.
Dunkle, K. L., R. K. Jewkes, M. Nduna, J. Levin, N. Jama, N. Khuzwayo, M. P. Koss, and N.
Duvvury. 2006. “Perpetration of Partner Violence and HIV Risk Behaviour among Young
Men in the Rural Eastern Cape, South Africa.” AIDS 20:2107–14. doi:10.1097/01.aids.
0000247582.00826.52.
Falk, O., M. Wallinius, S. Lundström, T. Frisell, H. Anckarsäter, and N. Kerekes. 2014.
“The 1% of the Population Accountable for 63% of All Violent Crime Convictions.”
Social Psychiatry and Psychiatric Epidemiology 49:559–71.
Farrington, D. P., J. W. Coid, L. M. Harnett, D. Jolliffe, N. Soteriou, R. E. Turner, and
D. West. 2006. Criminal Careers up to Age 50 and Life Success up to Age 48:
New Findings from the Cambridge Study in Delinquent Development. London, UK:
Home Office Research.
Fonagy, P., and M. Target. 2003. Psychoanalytic Theories: Perspectives from Developmental
Psychopathology. New York: Routledge.
Frosh, S., A. Phoenix, and R. Pattman. 2002. “Taking a Stand: Using Psychoanalysis to
Explore the Positioning of Subjects in Discourse.” British Journal of Social Psychology
42:39–53.
Fulu, E., R. Jewkes, T. Roselli, and C. Garcia-Moreno. 2013. “Prevalence and Risk Factors
for Male Perpetration of Intimate Partner Violence: Findings from the UN Multi-country
Study on Men and Violence in Asia and the Pacific.” The Lancet Global Health 1:
e187–207.
Garcia-Moreno, C., H. A. Hansen, M. Ellsberg, L. Heise, and C. Watts. 2005. WHO Multi-
country Study on Women’s Health and Domestic Violence against Women. Geneva, Swit-
zerland: World Health Organization.
Herek, G. M. 1987. “On Heterosexual Masculinity: Some Psychological Consequences of the
Social Construction of Gender and Sexuality.” In Changing Men: New Directions in
Research on Men and Masculinity, edited by M. S. Kimmel, 365–74. Beverly Hills,
CA: Sage.
Hood-Williams, J. 2001. “Gender, Masculinities and Crime: From Structures to Psyches.”
Theoretical Criminology 5:37–60.
Howson, R. 2014. “Re-thinking Aspiration and Hegemonic Masculinity in Transnational.”
Masculinities and Social Change 3:18–35.
Jewkes, R., K. Dunkle, M. P. Koss, J. B. Levin, M. Nduna, N. Jama, and Y. Sikweyiya. 2006.
“Rape Perpetration by Young, Rural South African Men: Prevalence, Patterns and Risk
Factors.” Social Science & Medicine 63:2949–61. doi:10.1016/j.socscimed.2006.07.027.
Jewkes, R., K. Dunkle, M. Nduna, N. Jama, and A. Puren. 2010. “Associations between
Childhood Adversity and Depression, Substance Abuse & HIV & HSV2 in Rural South
African Youth.” Child Abuse and Neglect 34:833–41.
Jewkes, R., E. Fulu, T. Roselli, and C. Garcia-Moreno. 2013. “Prevalence and Risk Factors
for Non-partner Rape Perpetration: Findings from the UN Multi-country Cross-sectional
Jewkes and Morrell 23

Study on Men and Violence in Asia and the Pacific.” The Lancet Global Health 1:
e208–18. doi:10.1016/S2214-109X(13)70069-X e1.
Jewkes, R., and R. Morrell. 2010. “Gender and Sexuality: Emerging Perspectives from the
Heterosexual Epidemic in South Africa and Implications for HIV Risk and Prevention.”
Journal of the International AIDS Society 13:1–10.
Jewkes, R., R. Morrell, J. Hearn, E. Lundqvist, D. Blackbeard, G. Lindegger, M. Quayle, Y.
Sikweyiya, and L. Gottzen. 2015. “Hegemonic Masculinity: Combining Theory and
Practice in Gender Interventions.” Culture Health & Sexuality 17:112–27.
Jewkes, R., R. Morrell, Y. Sikweyiya, K. Dunkle, and L. Penn-Kekana. 2012a. “Men,
Prostitution and the Provider Role: Understanding the Intersections of Economic
Exchange, Sex, Crime and Violence in South Africa.” PLoS One 7:e40821. doi:10.
1371/journal.pone.0040821.
Jewkes, R., R. Morrell, Y. Sikweyiya, K. Dunkle, and L. Penn-Kekana. 2012b. “Men, Sex and
the Provider Role: Crime, Violence, Correlated Psychological Attributes Associated with
South African Men’s Engagement in Prostitution and Transactional Sex in South Africa.”
PLoS One 7:e40821. doi:10.1371/journal.pone.0040821.
Jewkes, R., R. Morrell, Y. Sikweyiya, K. Dunkle, and L. Penn-Kekana. 2012c. “Transactional
Relationships and Sex with a Woman in Prostitution: Prevalence and Patterns in a Rep-
resentative Sample of South African Men.” BMC Public Health 12:325. doi:10.1186/
1471-2458-12-325.
Jewkes, R., M. Nduna, N. Shai, E. Chirwa, and K. Dunkle. “Understanding the Relationships
between Gender Inequitable Behaviours, Childhood Trauma and Socio-economic Status in
Single and Multiple Perpetrator Rape in Rural South Africa: Structural Equation Model-
ling.” PLoS One.
Jewkes, R., Y. Sikweyiya, R. Morrell, and K. Dunkle. 2009. Understanding Men’s Health
and Use of Violence: Interface of Rape and HIV in South Africa. Technical Report,
Medical Research Council, Pretoria.
Jewkes, R., Y. Sikweyiya, R. Morrell, and K. Dunkle. 2011. “Gender Inequitable Masculinity
and Sexual Entitlement in Rape Perpetration South Africa: Findings of a Cross-sectional
Study.” PLoS One 6:e29590. doi:10.1371/journal.pone.0029590.
Johansson, A., P. Santtila, N. Harlaar, B. von der Pahlen, K. Witting, M. Algars, K. Alanko, P.
Jern, M. Varjonen, and N. K. Sandnabba. 2008. “Genetic Effects on Male Sexual Coer-
cion.” Aggressive Behaviour 34:190–202.
Lilienfield, S. O. 2005. Psychopathic Personality Inventory—Revised. Lutz: Florida Psycho-
logical Assessment Resources.
Lund, F. 2008. Changing Social Policy: The Child Support Grant in South Africa. Cape Town,
South Africa: HSRC Press.
Makiwane, M., and E. Udjo. 2006. Is the Child Support Grant Associated with an Increase in
Teenage Fertility in South Africa? Pretoria, South Africa: Human Sciences Research
Council.
Mathews, S., R. Jewkes, and N. Abrahams. 2011. “‘I had a Hard Life’’: Exploring Childhood
Adversity in the Shaping of Masculinities among Men Who Killed an Intimate Partner in
South Africa.” British Journal of Criminology 51:960–77.
24 Men and Masculinities XX(X)

Messerschmidt, J. W. 1993. Masculinities and Crime: Critique and Reconceptualization of

Theory. Lanham, MD: Rowman & Littlefield.
Messerschmidt, J. W. 1997. Crime as Structured Action: Gender, Race, Class, and Crime in
the Making. Thousand Oaks, CA: Sage.
Messerschmidt, J. W. 2000. “Becoming ‘Real Men’: Adolescent Masculinity Challenges and
Sexual Violence.” Men and Masculinities 2:286–307.
Messerschmidt, J. W. 2012. “Engendering Gendered Knowledge: Assessing the Academic
Appropriation of Hegemonic Masculinity.” Men and Masculinities 15:56–76.
Morrell, R., R. Jewkes, G. Lindegger, and V. Hamlall. 2013. “Hegemonic Masculinity:
Reviewing the Gendered Analysis of Men’s Power in South Africa.” South African Review
of Sociology 44:3–21.
Mullins, C. W. 2010. “Messerschmidt, James W: Masculinities and Crime.” In Encyclopedia
of Criminological Theory, edited by F. T. Cullen and P. Wilcox, 620–624. Thousand Oaks,
CA: Sage.
Newburn, T., and E. A. Stanko, Eds. 1994. Just Boys Doing Business?: Men, Masculinities
and Crime. London, UK: Routledge.
Pulerwitz, J., and G. Barker. 2008. “Measuring Attitudes toward Gender Norms among
Young Men in Brazil: Development and Psychometric Evaluation of the GEM Scale.”
Men and Masculinities 10:322–38.
Pulerwitz, J., S. Gortmaker, and W. DeJong. 2000. “Measuring Sexual Relationship Power in
HIV/STD Research.” Sex Roles 42:637–60.
Sikweyiya, Y., R. Jewkes, and R. Morrell. 2007. “Talking about Rape: Men’s Responses to
Questions about Rape in a Research Environment in South Africa.” Agenda 74:48–57.
Stouthamer-Loeber, M., R. Loeber, D. L. Homish, and E. Wei. 2001. “Maltreatment of Boys
and the Development of Disruptive and Delinquent Behavior.” Development and Psycho-
pathology 13:941–55.
Torbenfeldt Bengtsson, T. 2015. “Performing Hypermasculinity: Experiences with Confined
Young Offenders.” Men and Masculinities, 19: 410–428.
Totten, M. 2003. “Girlfriend Abuse as a Form of Masculinity Construction among Violent,
Marginal Male Youth.” Men and Masculinities 6:70–92.
Tremblay, R., L. Pagani-Kurtz, L. Massc, F. Vitaro, and R. O. Pihl. 1995. “A Bi-model
Prevention Intervention for Disruptive Kindergarten Boys: Its Impact through Mid-
adolescence.” Journal of Consulting and Clinical Psychology 634:560–68.
Walker, S. P., T. D. Wachs, S. Grantham-McGregor, M. M. Black, C. A. Nelson, S. L. Huff
man, H. Baker-Henningham, S. M. Chang, J. D. Hamadani, B. Lozoff, J. M. Meeks
Gardner, C. A. Powell, A. Rahman, and L. Richter. 2011. “Inequality in Early Child-
hood: Risk and Protective Factors for Early Child Development.” The Lancet 378:
1325–38.
Whitehead, S. 2002. Men and Masculinities: Key Themes and New Directions. Cambridge,
MA: Polity.
Wood, K., and R. Jewkes. 2001. “‘Dangerous’ Love: Reflections on Violence among Xhosa
Township Youth.” In Changing Men in South Africa, edited by R. Morrell, 317–36.
Pietermaritzburg, South Africa: University of Natal Press.
Jewkes and Morrell 25

Author Biographies
Rachel Jewkes is the director of the South African Medical Research Council’s Gender and
Health Research Unit, a former vice-president of the SAMRC and an honorary professor at the
University of the Witwatersrand. She is the secretary of the Sexual Violence Research Ini-
tiative and the director of the DFID-funded What Works to Prevent Violence? Global Pro-
gramme, working to advance knowledge in 14 countries.

Robert Morrell is author of From Boys to Gentleman (2001) and (with Debbie Epstein,
Elaine Unterhalter, Deevia Bhana and Relebohile Moletsane) Towards Gender Equality?
South African schools during the HIV and AIDS pandemic (2009). Amongst his edited works
are Changing Men in Southern Africa (2001) and (with Linda Richter) Baba: Men and
Fatherhood in South Africa (2006).