Introduction

South Africa, with a population of 44 million, is home to more than five million HIV-infected people, more than any other country in the world.1 Among young people, HIV prevalence differs greatly by gender: 4.8% of men aged 15–24 are HIV-infected, compared to 15.5% of women.2 This gender gap is at its widest in the young adult years: 24.5% of young women aged 20–24 are HIV infected, compared to 7.6% of men.2 In the South African literature as elsewhere, much attention has been paid to the gendered pathways to HIV infection and the impact on women of relationship dynamics in the context of widespread gender inequality.36 In seeking to explain the gendered nature of the HIV epidemic, differences in male and female roles within sexual relationships have been examined.7 Less attention, however, has been paid to the specific factors that place men at risk for HIV and to the ways in which men's relationship behaviors may place them, as well as their female partners, at risk.8

It is widely accepted that gender-based inequalities in power, which are manifested within sexual relationships, place women at a disadvantage in terms of negotiating safer sex, or in refusing unwanted sex.3,4 In South Africa, levels of gender-based violence are unusually high, and experience of sexual coercion renders women more vulnerable to the acquisition of HIV.911 These vulnerabilities are enhanced by many young women having older male partners, a practice that has been demonstrated to amplify women's HIV infection levels considerably,12,13 and by other factors such as age, gender, biological susceptibility and relative inexperience.14

In contrast, the factors influencing men's pathways to HIV infection have received less research attention, particularly in the context of heterosexual interactions.8 Most knowledge of men's sexual behavior in Africa comes from large-scale surveys that document levels and trends in condom use, partnering practices and other risk and preventive behaviors.15 To the extent that men's risk behaviors, such as multiple partnering, have been examined, it is often in relation to their impact on women's risk and the dangers that young women, in particular, face from these practices. The few empirical studies available suggest that men's pathways to risk differ markedly from those of women: A multi-site analysis of African data suggests the importance of men's early sexual practices, which are likely to become entrenched risk behaviors in the adult years.16 Similarly, findings from our South African research setting have drawn attention to the consistent minority of men with early sexual debut and suggest that this may be an important factor associated with later risky behaviors.17 Ethnographic research on young men and sexuality underscores the complexity of these issues, as well as the dynamic and changing nature of masculinity over time.18,19 Most notably, a deeper understanding of the relationship context of men's risk is needed, particularly regarding the social factors that influence their interactions with women.

Some understanding of these factors is provided by a growing literature on masculinity, or the socially constructed set of values and beliefs that guide men's actions and behaviors.18,2022 The gendered differences between men's and women's risk behaviors that are reflected in HIV prevalence statistics reflect deep social differences in the construction of gender roles.7 In southern Africa, contemporary definitions of masculinity reflect long-term social and historical processes of change,21 rather than the more immediate or proximal forces that are often studied in relation to HIV/AIDS. The colonial and apartheid regimes permanently changed African household and family institutions, with a heavy impact on men through their participation in the migrant labor system.18,2123 At the same time, however, retention of rural homes and connections perpetuated patriarchal systems of authority, including the subordination of women.18,21 Thus, while African men were dispossessed in a larger, societal sense, male domination and control over gender relations was ensured locally. These social influences have yielded complex norms surrounding masculinity, which often reinforce patriarchal ideas of virility and manhood and are perpetuated at some level even among younger men.21 Further, these dominant social norms have made it difficult to negotiate more recent societal changes, such as movements toward greater gender equality.18,23 In sexual relationships, this masculinity may be negatively expressed, through the pursuit of multiple, risky partnerships or a tolerance of sexual coercion.22

There are few psychometric instruments developed in African settings to assess such gender-related constructs as relationship power, gender role norms and masculinity in a culturally specific manner.24 In addition, few studies have examined empirically aspects of masculinity within the context of men's interactions with women. The aim of the current study was to develop and test measures related to gender role and relationship norms, beliefs and attitudes in order to explore the interface of gender and HIV risk behavior in heterosexual interactions. In this paper, we focus primarily on the development of these measures in relation to changing masculinity in the context of the HIV/AIDS epidemic in rural KwaZulu/Natal, South Africa. To do this, we delineate the ways in which men's and women's views of gender roles in relationships, as well as their attitudes regarding interactions with their sexual partners, influence the extent to which they engage in behaviors placing them at risk for HIV infection.

Materials and Methods

Setting

The study was conducted in the province of KwaZulu/Natal, South Africa, an area along the eastern coast that includes the country's third largest city of Durban. KwaZulu/Natal is South Africa's largest province, with a population of nearly ten million out of a total population of 52 million.25 It is also among the poorest of the country's nine provinces, with an annual per capita income in many rural areas of about U.S. $1,000, generally derived from migrant labor remittances, government pensions or subsistence trading.25 The study was conducted in one subdistrict in rural northern KwaZulu/Natal, an area 3 h from Durban. This area is typical of many rural areas and is characterized by few employment opportunities, high levels of out-migration for labor, and relative social isolation. The study area is relatively homogeneous; almost all residents describe their ethnic and linguistic background as Zulu.

Method Development

The data reported here were collected as part of a larger, three-phase study addressing the role of gender and relationship norms in relation to HIV risk behaviors among young South African adults. The aim of the study was to develop culturally specific measures of relationship and gender role norms, beliefs and attitudes. Few such measures have been developed in African settings, and while some measures of gender role norms and ideologies, masculinity and relationship power do exist,24,26 we felt it imperative to inform the study of HIV risk and the development of interventions with culturally appropriate measures that could more easily be adapted to similar settings. To this end, formative work focused on method development was conducted prior to the survey. In Phase 1, 47 participants (N = 25 women, N = 22 men) completed a structured daily diary of relationship events and sexual activity over a 3-week period.11 Following that, in-depth qualitative interviews were conducted with the same 47 participants (Phase 2). These interviews were intended to elicit a pool of items to develop the gender and relationship measures. The interviews, conducted in isiZulu by trained qualitative researchers, covered a range of topics, including the dynamics of relationships, sexual interactions, attitudes and beliefs, and preventive practices and are valuable because they offer the perspective of young people currently in relationships, providing a detailed look at the relationship context of HIV risk in this population. Qualitative data analysis through coding of transcripts yielded seven main categories: power, traditional/non-traditional sex roles, meaning of relationships, vision of the future, condoms and contraception, the body and the spatial context of relationships. The survey instrument was developed from the qualitative data, using representative items in each of the key domains. More broadly, the qualitative findings provided insight into the dynamic and changing nature of gender role and relationship norms among young adults in this setting.27 Preliminary versions of the questionnaire were developed, refined, and pilot tested prior to administration of the survey. To ensure the items constituted useful and meaningful scales, we established the construct validity using factor analyses and internal consistency via Cronbach's alpha to establish the psychometric properties of the measures. We also established test–retest reliability by having a random subsample of 30 participants complete the survey for a second time. Data for the analyses reported here were collected during the third phase of research, a structured survey of young adults aged 18–24 that included the scales developed during the formative phases of the study. The outcome of this work was a battery of six psychometrically tested measures relevant to the study population (see Appendix), addressing key aspects of relationship dynamics and sexual interactions. The study was conducted between March 2002 and October 2004.

Participants

Participants were recruited from two secondary schools selected because of prior participation in research and an established relationship between researchers and the schools in that area. The participants included men (n = 101) and women (n = 199) aged 18–24 who were currently enrolled in school and who had been in a current heterosexual relationship for at least 3 months. In South Africa, education tends to be prolonged for many young people, with typical age distributions within secondary schools ranging from age 12–25, particularly in rural areas, and as much as 20% of the school population comprised of students above age 18.28

Procedures

Participants were recruited through general announcements about the study in all classrooms of Grades 11 and 12 in the school. Interested students were then directed to meet with a member of the research team, who administered a brief eligibility questionnaire. Eligibility criteria, which were known only to the research team, were age (18–24 years), participation in a heterosexual relationship of at least 3-months duration, and sexual activity within the prior 3 months. Approximately 320 students were screened in order to recruit the final sample of 300, with women more often ineligible than men due to lower levels of sexual activity. The higher proportion of women in the final sample reflects the gender distribution of the school population, where males are under-represented, particularly in higher grade levels. Data for the survey were collected using an interviewer-administered individual questionnaire. The questionnaire was translated from English into isiZulu and then back-translated to ensure accuracy and comprehension. All interviews were conducted in isiZulu.

Measures

The survey included questions on socio-demographic background, sexual risk behavior, partnership characteristics, and the six measures of gender attitudes and gender role norms. The measures of sexual behavior had been previously validated in this setting.29 In the survey, as in the in-depth interviews that preceded it, a primary relationship was defined as a boyfriend or girlfriend with whom an ongoing relationship had been established for at least 3 months. Secondary partners were defined as any additional partners, who could be long-term or short-term, serious or casual. Participants were asked to specify how many secondary partners they had and to elaborate on differences in sexual and preventive behaviors in their primary and secondary partnerships. These definitions were chosen so as to include the widest possible range of partners and related behaviors in participants' reports.

Sociodemographic Measures

These included age, gender, ethnic/linguistic background, religious affiliation, level of schooling, parents' education, and household wealth.

Sexual Risk Behaviors and Partnership Characteristics

These measures included age at first sex, number of partners, secondary relationship, length of relationship, frequency of condom use, use of other contraception, HIV risk perception and perceived behavior of partners.

Gender Attitudes and Norms within Heterosexual Relationships

These measures included the six scales below. Each item was evaluated on a four point scale ranging from 1 = strongly disagree to 4 = strongly agree. Individual items are presented in the Appendix.

Violence in Relationships

Separate male and female versions of this scale were developed, including parallel items. Examples from this six item scale are “If I felt that my partner was with someone else, I would beat her” (men) and “If he felt that I was with someone else, my partner would beat me” (women) [alphas = 0.82 (men) and 0.69 (women)]. Higher scores indicate stronger endorsement of violence as a norm for sexual relationships. The test–retest reliability coefficient for this scale was 0.32, p < 0.05.

Power in Relationships

This scale has four items, including, “We come to agreements together about what we want to do” [alphas = 0.76 (men) and 0.70 (women)]. Higher scores indicate greater endorsement of shared power in the relationship with one's partner. The test–retest coefficient was 0.77, p < 0.001.

Egalitarian Sexual Scripts

This scale includes four items, such as “We are both free to say if we don't want sex” [alpha = 0.69 (men) and 0.68 (women)]. Higher scores reflect stronger beliefs in value of shared sexual scripts for men and women. The test–retest coefficient for this scale was 0.67, p < 0.001.

Male Dominance Scripts

This scale includes nine items, such as the following examples: “Men cannot live with just one girlfriend” and “Only a bad girl shows that she likes sex a lot” [alphas = 0.67 (men) and 0.67 (women)]. Higher scores reflect endorsement of sexual standards that reflect normative views of men's patriarchal role, particularly in relation to the control of women, and reinforce views of women's need to follow strict rules of social and sexual comportment. The test–retest coefficient was 0.89, p < 0.001.

Mutuality in Relationships

This scale includes five items that assess investment in shared intimate experiences in one's relationship. An example of the items is “We are taking care of each other well” [alphas = 0.76 (men) and 0.81 (women)]. Higher scores reveal greater care and consideration of one's partner. The test–retest coefficient was 0.81, p < 0.001.

Hyper-Romanticism

This is a six-item scale assessing strongly expressed beliefs in romance and the power of an intimate relationship. Examples of the items include, “I cannot live without my partner for even 1 day” [alpha = 0.77 (men) and 0.72 (women)]. Higher scores indicate stronger romantic beliefs for one's relationship. The test–retest was 0.91, p < 0.001.

All data analyses were performed using SPSS Version 13.0. In the first stage of data analysis, means and proportions were compared by gender via t-tests. In the second stage of data analysis, correlations were used to examine associations between scale scores and four sexual risk outcomes: 1) frequency of condom use with the primary partner (from 1 = never to 4 = every time); 2) frequency of condom use with secondary partners (N = 52 men; N = 62 women) (from 1 = never to 4 = every time); 3) number of partners in prior 3 months; and 4) number of lifetime partners. Logistic regression was then used to examine how scale scores predicted consistency of condom use (every time vs. not every time) with the primary and secondary partners, for men and women. All analyses were performed separately for men and women.

We designed the analytical component of this work based on assumptions drawn from the broader literature, gender script theory and our own preliminary findings.11,27 Script theory posits that normative beliefs about gender roles guide heterosexual interactions at the individual, interpersonal and cultural levels.3032 We hypothesized that higher endorsement of violence, power and male dominance would be negatively correlated with safer sexual behaviors, measured here as condom use and fewer sexual partners, among both men and women. Specifically, if gender-inequitable cultural scripts were exerting influence, men would be more likely to take risks in the face of threats, such as HIV infection, and pressure or force female partners to comply. In contrast, we thought that higher endorsement of egalitarian sex roles, mutuality and hyper-romanticism would correlate with safer sexual behaviors: Such relationships would rely more on gender equitable cultural scripts that permit both partners to introduce or insist on condom use. We conducted analyses separately with the expectation that different predictors of sexual risk might emerge in the multivariate analysis. In this paper, which focuses primarily on the development of measures related to masculinity and related beliefs, we present results for both genders, using the data on women for comparison.

Results

Young men and women in the sample were drawn from a relatively homogeneous background: 97% were born in the province of KwaZulu/Natal, and all described themselves as Zulu. Most (89%) attended church regularly. Levels of parental education were low, with only 12% of mothers and 15% of fathers reporting high school completion.

Sexual and Relationship Histories

Men reported a younger age of first intercourse than did women, t(155) = −3.22, p < 0.01 (Table 1). Men reported more sexual partners in the preceding 3 months, t(297) = 2.37, p < 0.05, and lifetime, t(126) = 9.25, p < 0.001. Men had an average of almost two partners (1.9) in the previous 3-month period, compared to 1.6 for women, and 5.3 partners lifetime, compared to 2.2 among women. Over half of the men (51.5%) reported a current secondary partner, compared to only one-third of women [(34.3%), (χ 2(1) = 7.48, p < 0.01)]. Primary sexual relationships were of longer duration for women than men, t(113) = −2.05, p < 0.001, as were their secondary relationships, t(113) = −2.05, p < 0.05. In addition, men reported fewer occasions of intercourse with primary partners during the past 3 months, t(285) = −4.85, p < 0.001, but no gender differences were found for secondary relationships. Rates of reported condom use were relatively high among both men and women; no gender differences were found across relationship types. Use of other contraceptives was low among both men and women. Almost 12% of women and 6.9% of men reported having at least one child (not shown). No differences were found in the proportions of men and women who perceived that their partners' behaviors were placing them at risk for HIV (Table 1). Women were more likely than were men to believe that their partners had other sexual partners, both with regard to primary [t(296) = −5.68, p < 0.001] and secondary partners [t(118) = −3.34, p < 0.001].

Table 1 Sexual behavior and relationship history among young adult men and women

Gender and Sexual Risk

Mean scale scores were calculated for men and women (Table 2). No differences by gender were found. We then examined associations between the scale scores and the four sexual risk outcomes: condom use (frequency with regular and secondary partners) and number of partners (prior 3 months and lifetime). Correlations between the scale scores and these continuous variables can be found in Table 3. For men, both endorsement of relationship violence and hyper-romanticism were associated with more frequent condom use in their primary relationships. Further, men who were more dominant used condoms more frequently with partners outside of their primary relationship. Men with lower scores related to power in relationships had fewer sexual partners in the prior 3 months. In contrast, women with more egalitarian sexual scripts and higher hyper-romanticism scores reported more sexual partners in the prior 3 months, and those with higher egalitarian scores reported more lifetime sexual partners compared to their counterparts with lower scores. More frequent condom use was associated with lower levels of violence in their primary relationships for women.

Table 2 Gender scale scores for men and women
Table 3 Pearson correlations between scale scores and sexual risk variables for men and women

Results of the logistic regression analyses showed that predictors of condom use differed for men and women, with some important gender differences observed. Four logistic regression analyses were conducted to predict reports of consistent condom use (every time versus not every time) in primary and secondary partnerships, using the six scale scores separately by gender. For secondary partnerships, scale scores did not predict consistent condom use for men or women nor did they predict condom use in women's primary relationships. However, the analysis for men's reports regarding their primary relationship was significant, χ 2(6) = 14.09, p < 0.02. Specifically, men who reported more egalitarian sex norms were less likely to report consistent condom use [OR = 0.11, p < 0.05; CI(95%) = 0.015–0.792]. Regression analyses predicting reports of two or more partners were unrelated to scale scores for men. However, the analysis for women was significant, χ 2(6) = 13.25, p < 0.05. Women with higher relationship power scores reported less consistent condom use [OR = 0.29, p < 0.05; CI(95%) = 0.093–0.874], whereas those with higher mutuality scores reported more consistent condom use [OR = 6.61, p < 0.01; CI(95%) = 1.616–27.058] although the wide confidence interval indicates that this latter finding may be unstable.

Discussion

The findings from this study provide empirical evidence for the ways that gender-related attitudes and relationship norms among young South African men and women relate to sexual risk behaviors. We developed a set of six scales designed to capture important dimensions of the relationship context of risk. For example, the male dominance scale succinctly expresses many of the conflicts between men and women in the sexual realm surrounding patriarchal norms and challenges to them. The violence measures captured men's behavior within relationships if a female partner acted in some way that displeased him, for instance “hitting a partner” or “forcing her to do something.” Although no differences were found between the scales by gender, interesting associations between the scales and sexual risk outcomes did emerge, and these differ according to gender. Four of the six scales focus primarily on men's normative behaviors—and women's reactions to them—and thus our findings offer particular insight into masculinity and its association with sexual risk.

Our findings capture differences between men and women in gender role norms and attitudes as predictors of sexual risk and preventive behaviors in established relationships. Few of the findings, however, reflected expected associations between gender and relationship norms and sexual risk behaviors. The finding that men with lower relationship power scores had fewer sexual partners in the preceding 3 months appears to fit with stereotyped views of masculine relationship norms in this setting, whereby multiple partnering is associated with greater power and sense of masculine achievement.18 However, when considering that the power scale measured relative support for gender equality in relationship decision-making, one might expect a higher, rather than lower score, to be associated with fewer partners. The association of relationship violence and greater dominance with condom use may reflect recognition of greater risk among those men or continued negative associations with the need for HIV prevention, such that it is relegated to contexts of sexual violence or secondary partnerships often viewed as more casual. However, more frequent condom use was also associated with higher levels of partner attachment, or hyper-romanticism, which could reflect an increasing association of condoms with love and respect for a partner in this younger generation of men.33 Further, the association of more egalitarian relationship norms with less consistent condom use in men's primary partnerships is possibly a reflection of perceptions surrounding lower risk in established relationships or of the fact that their female partners did not want to use condoms.

In contrast to the men, more consistent condom use was correlated with a lower endorsement of relationship violence and higher mutuality scores among women, findings suggestive of an association between greater empowerment and safer sex behaviors. The negative association between higher power and condom use appeared to contradict this, although it is possible that more empowered women select safer relationships and, therefore, use condoms less often in those relationships. Our findings about women and partnerships perhaps offered the most insight into changing gender role and relationship norms for both men and women in this setting. The fact that women with more egalitarian relationship norms had more sexual partners is interesting in a setting where sexual activity for young women is highly stigmatized.29 Given this, it seems plausible that women with multiple partners would have characteristics that reflect a belief in their greater equality with men. However, women who more strongly endorsed the relationship norms reflected in the hyper-romanticism scale were also more likely to have had multiple partners in the past 3 months. Although this result appears more difficult to explain in a setting where women widely perceive men's practice of multiple partnering to be a factor that increases their HIV risk,29 it may relate to the high value placed on having a boyfriend at all times, captured in the sense of desperation regarding emotional connection reflected in the items of the hyper-romanticism scale. Regardless of HIV risk, multiple partnering among women who perceive themselves as more equal may also reflect women's reaction against restrictive gender norms, allowing women to assert their equality by “behaving like men.”22 Findings from our qualitative data shed some additional light here: In interviews, some young women referred to the pursuit of multiple partnerships as an insurance strategy for when a primary relationship broke up.27 The often fluid nature of relationships in this setting further explains some of the strategies that protect against this, such as secondary partnerships of long duration for both men and women.

It is helpful to interpret these findings in the context of recent scholarship on gender and sexual relationships, particularly masculinity, in South Africa.18,21,23 Our findings about men and women offer insight into the nature of intimate relationships among young South Africans and into the ways that gender roles are—and are not—changing. These findings support our earlier qualitative research, which suggested that both men's and women's ideas of normative gender roles in sexual relationships are changing but that enactment of new roles is a slow and dynamic process.27 In the current study, young women's endorsement of egalitarian gender roles in sexual relationships appears to impact positively on levels of condom use. However, gender role norms related to partnering practices and heterosexual relationships are still governed by dominant masculinities and prevailing patriarchal norms. For many South African men, identities of manhood remain tied to control over gender relations and dominance in sexual relationships. While there are clearly challenges to these normative views, they remain entrenched, particularly in rural settings. However, the patriarchal norms that guide men's sense of self may be shifting, and younger men in particular continue to seek ways to shape new masculine behaviors.18

The main limitation of the current study includes the reliance on self-reported data on sexual behavior in a sub-population of youth attending school in one relatively homogeneous area of South Africa. With a small, non-random sample, these results are not broadly generalizable. Even so, our measures appear to have captured normative beliefs and attitudes about gender roles, particularly for men, which we have found to be associated in plausible ways with sexual risk behaviors. Further studies will need to test these measures in different populations.

The development of psychometrically established measures to elucidate the relationship context of risk for men and women in the setting of South Africa's severe HIV epidemic provides an important tool for HIV prevention research. Overall, these findings reinforce the idea that the context of sexual interaction is key to understanding HIV risk behavior and that gender-based attitudes and beliefs help to explain the dynamics surrounding sexual risk and barriers to prevention. The empirical findings related to men are particularly useful because, in the African context, the study of masculinity has generally been undertaken with qualitative methodologies. These measures can be useful in the design and measurement of interventions, particularly interventions focused on men as a way of improving prevention outcomes for both men and women through an emphasis on fostering positive masculinity. Perhaps most importantly, these findings suggest the need to look more deeply at prevailing assumptions regarding gender role norms, gender equality and safer sexual behavior, and to understand change surrounding gender roles as a dynamic and ever-changing, rather than linear, process.