Circular patterns of labor migration between Mexico and the United States play a significant role in Mexico's rural HIV epidemic.^1–6 Migrants' vulnerability to HIV partly reflects that migrants tend to be young men with little formal education and limited English skills. In addition, migrant men's vulnerability to HIV reflects the social characteristics of the communities to which they migrate, including generally more permissive norms about sexuality than are found in Mexico, the anonymity provided by being in a large urban context far from home, a lack of social support for migrants, exploitive working conditions and a lack of access to health care.^6,7

In response to increasing awareness of the migration-HIV nexus and of the importance of prevention work with Mexican migrants in the United States, several studies have focused on migrants' sexual risk behaviors and reviewed prevention approaches used among Mexican migrants.⁸ This work has described the proximate behavioral and individual correlates of sexual risk, and has stressed the need for more research on how environmental and contextual factors shape sexual risk practices among unaccompanied male migrants.^8–11 A 2007 review and meta-analysis by researchers at the Centers for Disease Control and Prevention, for example, concluded with a call for research that explores "Hispanic cultural features... along with structural factors to further disentangle the moderators of HIV risk behavior.’′¹²(p.42)

The current study addresses that gap in the literature by analyzing the associations of cultural and social factors with the sexual risk of unaccompanied Mexican migrant men in Atlanta. The primary cultural constructs explored are men's ideas about masculinity, emotional intimacy, sexuality and marriage. Prior work in this migrant community^{5, 13–15} described a generational shift in marital ideals. Older men and women, born in rural Mexico during the 1940s and 1950s, emphasized the fulfillment of obligations shaped by gender as a key characteristic of a successful marriage. Men and women born in the 1960s and 1970s, in contrast, shared a marital ideal characterized by a growing emphasis on companionship, pleasure, emotional intimacy and, in some cases, shared decision making. For younger women, this emerging ideal framed a new understanding of sexual fidelity, which became not only a demonstration of respect but proof of love.⁵ The study described here sought to explore whether Mexican men from a migrant community similar to one in which prior research was conducted saw this new marital ideal as requiring sexual fidelity, despite the fact that they were separated from their main partners by thousands of miles and a national border.

Cultural factors such as shared beliefs about masculinity and femininity certainly shape sexual behavior, but they hardly are the only, or even the most important, determinants of behavior.^16–18 Prior work suggests that loneliness might be a significant influence on migrant men's sexual risk behavior. Unaccompanied migrant men's sexual behavior is characterized by high levels of sexual risk, both in comparison with the level of risk found among men whose wives migrate with them¹⁰ and in comparison with the level of risk found among similar men in the Mexican communities of origin.¹¹ In addition, many of these men remarked on the relationship between loneliness, the alienation that characterizes migrant life and sexual risk behavior.¹⁹ One man, for example, noted that a common reason to seek out extramarital sex is that men "miss more than anything their normal life back home, the intimacy with a spouse." Therefore, in addition to exploring how cultural constructions of masculinity and sexuality shape Mexican migrants' sexual risk, we explored how the desire for companionship and the specific social options available may relate to men's motivations for seeking sex.

This study seeks to contribute to the body of work exploring how gender relates to sexual health. Our goal is to learn how masculinity, including both culturally specific measures of Mexican migrant men's notions about gender, marriage, and sexuality and the social ways in which they demonstrate their masculinity, is connected to sexual behavior. This work extends the approach to gender most common in public health, in which gender is conceptualized as the relationship-level inequality between men and women, by including the social activities men engage in as an element of gender.^20–28 Although scales of masculinity exist, most of them have been developed among white American college students.^29,30 Existing scales developed for U.S. Latinos ^31–33 do not capture the generational changes that we have observed among Mexican men both in Atlanta and in rural Mexican sending communities.¹⁵ Furthermore, the preponderance of public health research on masculinity and health focuses on roles, beliefs, ideologies and scripts. Our analyses, in contrast, explore additional domains, including social networks and participation in leisure-time activities.

CONTEXT

In Atlanta, as throughout the southeastern United States, individuals of Mexican origin account for a growing share of both the foreign-born population and the growing Latino population. Statewide, 108,922 Latinos of any race were enumerated in Georgia in 1990, accounting for 2% of the state's overall population; in 2000, these figures were 435,227 and 5%, respectively.^34,35 In Georgia's Dekalb County (where this research took place), the growing enrollment of children for whom Spanish is a primary language³⁶ suggests that this population includes not just individual migrant laborers such as those who are the focus of this study, but also families who have migrated to the area and settled down. Many migrants, however, are male laborers who travel alone, drawn by opportunities for employment in residential construction and landscaping, agriculture and food processing, and light industry (such as the carpet mills of northern Georgia).^37–39

The 10-county Atlanta metropolitan area, which was home to some 3.4 million people when these data were collected, reflects these regional trends. Atlanta includes some of the fastest growing counties (in terms of population size) in the United States; much of this growth is a result of increases in the Latino population.⁴⁰

METHODS

Study Design and Sample

The data for this analysis were collected for a mixed- methods study that consisted of two phases. In the first phase, ethnographic observations and 31 semistructured interviews were conducted among men from Maravatio, in the state of Michoacán, between May and September of 1999. The interviews covered men's demographic attributes, migration, work experience, ideas about masculinity, sexuality, marriage and extramarital sexual relationships. Findings from this portion of the research are presented elsewhere.¹⁹ In the second phase, a structured survey that included items about sexuality, masculinity and marriage that were based on findings from the first phase as well as the first author's prior research with Mexican migrant families in Atlanta and in two sending communities in rural Mexico^13–15 was used to explore the associations between cultural and social aspects of masculinity and the sexual behavior of Mexican migrant men. Inclusion criteria were having been born in Michoacán, Mexico; having been in the United States for a minimum of a month; and having a wife who was in Mexico or having una relación de pareja (a couple relationship) with someone in Mexico. In both phases of the study, all data collection took place in Spanish.

The lack of a sampling frame for this population, combined with the men's mobility and largely undocumented status, made probability sampling difficult, if not impossible.⁹ Therefore, we selected a community-based convenience sample that was based on extensive ethnographic knowledge of the migrant community in Atlanta. The first phase of the study involved work with local community networks to choose a soccer team in Atlanta's northern suburbs that represented a particular Mexican sending community. This team represented the universe of ethnographic research inquiry and the starting point for recruiting individuals to participate in the study. The sample for the survey consisted of 200 men who either lived in the apartment complex that was adjacent to the field where this team practiced or played in the soccer league that practiced there. The six Latino male interviewers had attended soccer games throughout the prior season, during which time they became well known to residents of the apartment complex and members of the league.

This study was submitted for review to the institutional review board at Emory University. Study goals, instruments and procedures were also reviewed by a community advisory board, which included representatives from the Mexican consulate, a prominent local priest whose parish was composed largely of Mexican immigrants, and a social worker who directed a number of service programs at a faith-based community organization. In addition to seeking consent from the study participants, we sought permission from the managers of the apartment complex and the directors of the soccer league. After this permission was granted, interviewers approached spectators during weekend soccer matches, and players after the matches, to invite them to participate. The interviewers recruited additional participants by going door-to-door through the apartment complex, seeking participation from one resident per unit. Given the legal vulnerability of this population, Emory's institutional review board permitted the use of oral rather than written informed consent, so that participants would not be required to sign their name on study documents or even to tell us their names.

Measures

The survey instrument had seven parts: demographic characteristics; migration experience; masculinity ideologies; social network and social support; sexual risk behavior; substance use (which was later combined with sexual risk behavior); and leisure-time activities. Our conceptual model examines the association of these domains with the number of partners the respondents had had in Atlanta.

•Demographic characteristics.

We used multiple variables to capture men's demographic attributes: their place of birth (i.e., city, town or ranch), their age at time of interview, highest level of education they completed (i.e., less than primary, primary, or secondary or higher), whether they owned a house in Mexico, whether they had any children and what types of jobs they had had since moving to Atlanta (e.g., construction, cleaning or janitorial, business owner, gardener).

•Migration experience.

Participants were asked to report the month and year in which they migrated to the United States for the first time; we computed participants' average length of time since they first came to the United States from this variable. Men were asked whether they had migrated alone, the number of people with whom they had migrated and the number of times they had returned to Mexico. In addition, using an open-ended question, we asked participants to provide their reasons for having migrated to the United States. Answers were coded into the following reasons: to save money to marry, to open up a business, to bring family to the United States, to address immediate financial needs, to support family and to have an adventure. The analyses were based on their primary reason.

•Masculinity ideologies.

A central contribution of recent research on men and gender is the development of a pluralistic notion of masculinities.^41–43 We developed a set of contextually appropriate items, using a slightly modified version of Mirandé’s scale³¹ and including additional items about intimacy and pleasure suggested by our prior research in this community (see box).^5,15 Using a response card, interviewers asked men to indicate their level of agreement (NO!, no, yes or YES!) with 19 statements about marriage and sexuality as elements of men's masculinity ideologies; responses were coded on a scale from 0 for (Yes!) to 3 for (No!).

We used exploratory principal-axis factor analysis with Varimax rotation to ascertain the latent factors behind this scale and assess its psychometric properties. We observed two underlying factors: emotional intimacy and power, which accounted for 31% of the total variance (eigenvalue, 6.3), and sexual intimacy and pleasure, which explained 11% of the variance (eigenvalue, 2.6). We created a composite score for each subscale. A higher score on the sexual intimacy and pleasure scale reflects participants' beliefs that sexual intimacy and mutual pleasure strengthen a relationship (seven items; range, 1–21; Cronbach's alpha, 0.86); a higher score on the emotional intimacy and power scale indicates participants' endorsement of gender equity with regard to love, sex and marital life (four items; range, 3–12; Cronbach's alpha, 0.86). These two factors reflect and underscore the finding from other research that men from rural western Mexico may articulate an intimacy- oriented marital ideology without necessarily expressing a concomitant desire for a gender-equitable relationship.^15,44

•Social network and social support.

We measured the size of the current social network with the item "Thinking about the people on whom you can count, how many of them live in Atlanta or the surrounding area?" To determine the frequency of contact with social network members, participants were asked "How often do you see one of them?"; response options, coded as 0–3, were "almost never,""1–2 times per month,""1–2 times per week" and " every day."

We also measured men's access to four types of social support. Informational support was measured by men's use of social network members to find work in Atlanta (coded as 0 for "found work on my own or through a job agency" and 1 for "family/friends helped me find a job"). Tangible support was measured by two dichotomous items: whether the participant has someone to lend him money in an emergency and whether he has a place to stay if needed. Emotional support was measured by several items, including a four-point scale asking participants how much they like the people they live with (1 for "not at all" to 4 for "like a lot") and a dichotomous question on whether participants have someone to talk to when they are homesick. Finally, we developed an appraisal support scale to measure the extent to which social network support helps to increase a person's self-worth. The scale was composed of the score of four items: "You feel very close to your friends and/or relatives here"; "You have friends and/or relatives here that are always open to talking with you about your problems"; "Your friends and/or relatives here make you feel like you are a valuable person"; and "When you are with your friends and/or relatives here, you feel calm and at ease." These items were scored 0–2, signifying "no,""more or less," or "yes." This scale had a strong internal consistency (range, 0–8; Cronbach's alpha, 0.82).

•Sexual risk behavior.

The number of sexual partners that the men reported having had since their arrival in Atlanta was the outcome of interest. We included in the analysis risk factors that could confound the association with number of partners. These risk factors included participants' age (in years) at first intercourse, partner type at first intercourse and partner type for all partners in Atlanta. Partner type was categorized as girlfriend or female friend, prostitute or another man; we created a dummy variable for each partner type. Only one participant reported same-sex behavior, so we did not include "another man" as a partner type in subsequent analyses.

•Substance use.

We included measures on substance use, given their association with HIV risk behaviors. We used dichotomous variables to measure alcohol use in the past month ("almost never" vs. "1–3 times per week") and drug use in the past month (any vs. none).

•Leisure-time activities.

Drawing on our familiarity with the range of options of primary leisure-time activities in this sample, we asked men how they regularly spent their free time on Saturday nights and Sunday afternoons: going dancing, drinking, hanging out on the street, playing pool, visiting a friend, going to a strip club, staying home, doing errands, working or going to mass. Participants answered yes or no to each of these activities for both Saturday and Sunday. Given the number of potential leisure-time activities, we explored Spearman correlations between participation in each weekend activity and number of partners; those with statistically significant correlations (p<.05) were included in the regression.

Analytic Strategy

First, we described the sample by the study variables. After ensuring normal distribution of continuous variables, we excluded 13 men who were missing data on the outcome (number of partners). To ensure sufficient statistical power for multivariate regression analyses and avoid multicollinearity across domains, we created separate regression models to explore the interrelationship between number of partners and each of our theoretical domains. In each model, we controlled for demographic characteristics that were associated with the outcome and for time lived in the United States.

RESULTS

Sample Characteristics

The final sample of 187 men was generally diverse (Table 1). The average age of the men in the study was 28 (standard deviation, 7.4). Overall, the men had a low level of formal education; only 13% had completed high school or a higher degree. Most respondents were born in a city or town (81%).

Fifty-six percent of the sample had migrated alone to the United States. The majority had migrated to save money to get married (50%), to open a business (41%) or to support their families (42%). Only 11% said that friends had helped them find a job.

The mean age at first sexual intercourse was 17. As is increasingly common in Mexico,^45,46 most men in the study had had their first sexual encounter with a girlfriend (46%) or a female friend (31%; not shown). Ten percent reportedly had had their first sexual experience with a female sex worker, and only about 9% reported having had sex for the first time with their wife.

Several participants reportedly used drugs (14%), but the main substance used in the sample was alcohol (50%). Almost half of the study population stayed at home on Saturdays and Sundays (46% and 48%, respectively). Substantial proportions reported drinking (24%), eating out (17%), playing sports (28%), playing pool (23%), dancing (29%) or working (38%) on Saturdays. On Sundays, half of the respondents played sports (51%) and more than half (55%) reported attending mass.

Multivariate Findings

•Demographic characteristics.

After other demographic covariates were adjusted for, the older men were and the higher their level of education, the fewer partners they reported (coefficients, –0.9 and –1.0, respectively; Table 2). On the other hand, men who owned a home in Mexico reported a greater number of partners than others (1.2), even after other demographic predictors were adjusted for. No association was observed between number of partners and any other demographic variable.

•Migration experience.

The number of partners increased as men reported a greater number of trips back to Mexico (coefficient, 0.7). We found no association between number of partners and any other measure of migration experience.

•Masculinity ideologies.

Emotional intimacy with one's spouse was negatively associated with men's number of extramarital partners (coefficient, –0.4). A marginally significant positive association was observed between number of partners and sexual intimacy and pleasure.

•Social network and social support.

Men's numbers of partners increased as their social network size grew (coefficient, 0.3). Interestingly, the greater participants' frequency of contact with friends and family in their social network, the fewer their partners (–0.6). We found no associations between men's number of partners and the other social support indicators.

•Sexual risk behavior.

Men who had engaged in sexual intercourse with a sex worker in Atlanta and those who had used drugs reported a greater number of partners than other men (coefficients, 4.1 and 1.1, respectively). We found no association between number of partners and men's age at first intercourse or alcohol use.

•Leisure-time activities.

After demographic factors and other weekend activities were accounted for, men who reported greater number of partners also reported going out dancing or working on Saturdays (coefficients, 0.8 and 0.9, respectively), or attending a strip club or dancing on Sundays (3.9 and 2.6, respectively). We found a marginal positive association between men's number of partner and playing pool on Saturdays.

DISCUSSION

The goal of this study was to explore social and cultural factors shaping Mexican migrant men's sexual risk behavior. In addition, we sought to develop quantitative measures of masculinities that reflect current social science theorizations of masculinity as multidimensional and historically variable, and to assess the relationship between these measures and sexual risk behavior. Overall, the major contribution of this study is our finding that multiple aspects of masculinity—what men think, where they go and whom they can rely on—are associated with sexual risk.

The first key finding is that men's marital ideals are associated—in some cases—with their sexual behavior outside of marriage. High scores on the sexual intimacy and pleasure scale, which emphasizes the pursuit of mutual pleasure, was not statistically related to the number of sexual partners men had had in Atlanta. However, men who had higher scores on the emotional intimacy and power factor, which emphasizes emotional connectedness and a rejection of physical violence as a means to settle disagreements, were likely to report relatively few extramarital sexual partners while in Atlanta. That only one of these factors is significant underscores the need to examine multiple dimensions of modern masculinity in cultural context rather than to assess men on a single, presumably universally applicable scale distinguishing between traditional and modern masculinity ideologies. Each of these visions of what it means to be a married man diverges from more traditional constructions of Mexican marriage and masculinity, and both potentially imply a reorganization of sexual behavior within marriage. However, only men who emphasize emotional companionship and equity, rather than shared sexual pleasure and companionship, seem to have relatively few extramarital sexual partners. Future research should continue to explore the multidimensionality of modern masculinities, rather than focusing on a universalistic notion of "gender-equitable men.’′³³

Overemphasizing the impact of culture and ignoring the effects of social inequality may lead to a distorted picture of sexual risk in Latino communities in the United States. We respond here to that concern by presenting findings on social dimensions of migrant men's lives: The second major finding is that men's sexual risk behavior is associated with both masculinity ideals and leisure-time activities. These findings underline the importance of considering masculinity to be characterized not just by what men say, but also by what they do—and, particularly, by the spaces in which they spend their limited leisure time.

Strikingly, the number of individuals men can count on is positively associated with their number of sexual partners. This finding may reflect that the larger men's social networks, the more likely men are to spend time going to strip clubs and going out dancing. Participation in these activities, however, may facilitate both sexual risk behavior and the development of social ties with other men. Notably, the positive association between the size of a man's social network and his number of partners is mediated by increased frequency of interactions with family members and friends within social networks. This finding suggests that men without close contact with family members and friends may seek to create relationships with others to offset the absence of these interactions.

Multiple aspects of masculinity—what men think, where they go and whom they can rely on—are associated with sexual risk.

As we have found elsewhere,^4,47 extramarital sex among Mexican men is frequently a social rather than individual pursuit—something that men do to develop relationships with other men as well as for the pursuit of pleasure for pleasure's sake. For Mexican migrant men in Atlanta, the desire to buffer the loneliness of life far from one's family intersects with a social landscape in which the predominant choices are either to pray, to drink or to play soccer. At home in Mexico, many of these men might prefer a quiet game of dominoes with a compadre, or an ice cream on a bench in the plaza with their family. Particularly in contexts in which physical mobility is a challenge (because undocumented immigrants are ineligible for legal driver's licenses), men who find neither the church nor the soccer leagues appealing have few options other than going to strip clubs, dollar dance halls and pool halls. Research on the associations between masculinity and sexual health practices should focus on not only what men say regarding their ideals about masculinity, but also how their actions as men during work and leisure time are associated with sexual risk. The social context of HIV risk practices includes both migrants' beliefs and the social environments of the communities in which they reside during their time in the United States.

The findings highlight another, broader way in which masculinity relates to HIV risk; the act of migrating is a strategy through which men seek to succeed as men in rural Mexico—that is, the migration regime is itself a part of the Mexican social organization of gender.⁴⁸ The vast majority of these men saw migration to the United States as related to some aspect of their economic obligations to their families. Although many of these men may have engaged in extramarital sex had they remained in Mexico,⁴⁹ they would have done so in a context in which the overall prevalence of HIV is much lower, and so the relative riskiness of the same behavior is quite different.

Limitations

A few limitations of this research warrant discussion. First, the single-site study design precludes an exploration of how migrants' sexual risk behavior might vary across neighborhoods or cities. For example, in the late 1990s, when these data were collected, Atlanta's vast physical dispersion, poor public transportation, considerable economic opportunity, and limited bilingual health and social services presented a distinct urban landscape among major migrant-receiving communities; the combined effects of these social structures on migrant men's sexual risk behavior are largely unknown. Second, a lack of parallel data collection in the Mexican sending community precludes a comparison of the sexual risk behaviors and ideals of masculinity of otherwise similar migrating and nonmigrating men. Third, the survey failed to distinguish the duration of each respondent's current stay in the United States. Although some men may travel back and forth to Mexico, we were able to account only for the time elapsed since men first migrated to the United States. All other things being equal, the longer a man is away from home, the more partners he may have. Future research should measure whether number of partners is associated with men's migration patterns. Fourth, given our recruitment procedures, we were unable to compute the participation rate among eligible men who were invited to enroll. Fifth, because of concerns related to multicollinearity and statistical power, we were unable to test a model that included all the domains jointly. Consequently, some associations may be spurious because of confounding. Finally, our cross-sectional design prevents us from making causal statements regarding the findings. Future research should explore these questions with a larger sample of Mexican migrant men followed over time.

Policy Implications

A growing body of work has called attention to the risk of marital HIV transmission.^{4, 49–52} In this study, men's endorsement of specific masculine ideologies was associated with their having relatively few sexual partners. However, this does not mean that interventions should aim to change men's perceptions about marriage and masculinity. These ideologies are not individually held beliefs. Rather, they are the products of a complex intersection of shared cultural experience, individual life histories, and exposure to national and international political and economic structures, as well as to multiple ideological conduits, including educational systems and the media. Myriad social forces are altering ideas about masculinity in Mexico and the United States; a public health intervention alone is unlikely to transform men's ideas about manhood, and thereby effect sustainable behavioral change. Prevention interventions designed to change individually held beliefs, and to influence behavior through these changes, have failed to make significant contributions to health on the population level.⁵³ Moreover, associations between number of partners and practices of masculinity, such as patterns of socializing, underline the inadequacy of approaching masculinity purely as a cognitive construct that could be transformed through consciousness-raising strategies.

Our findings do, however, reveal promising avenues through which environmental determinants of HIV risk can be addressed. The development of health-enhancing social spaces could have a powerful impact on sexual risk behavior in communities of migrants living in the United States. From a migrant's perspective, however, the risk of HIV may pale in comparison with the more immediate risks of discovery and deportation. There are few safe spaces, for example, where migrants can enjoy leisure-time activities. Such safe spaces might include community centers, public libraries and athletic leagues with evening and weekend hours. The creation of Internet cafes in predominantly migrant communities may enable migrant men to communicate with their families in Mexico, and thereby reduce the loneliness that leads to risky sex. The Catholic Church, which provides an important social resource for men in both the sending and the receiving communities, might explore opportunities to mitigate the sexual risk associated with labor migration. Community-level interventions, moreover, would better integrate migrants into receiving communities by enabling the development of spaces where men may interact frequently with social network members without the presence of environmental risk factors, such as drugs.

The association between number of partners and the characteristics of men's social networks suggests the potential utility of further research on how integration into social movements might shape individual risk practices. Parallel work on gay rights activism, for example, has suggested that the community organizing that occurred in the early response to the AIDS epidemic among urban gay communities in the United States played a critical role in shaping individual risk practices by making safer sex "a community practice.’′⁵⁴ Similarly, it may be useful to explore the extent to which participation in social movements focused on the protection of migrants' social, economic and human rights (such as the organizations that mobilized vast numbers of immigrants to march in support of immigration reform during the 2006–2007 U.S. congressional session) might relate to sexual risk practices, via either social or psychological mechanisms. At the broadest policy level, the current immigration regime, in which labor migration from Mexico is officially restricted while the U.S. economy is heavily dependent on low-wage Mexican workers, is a critical aspect of the HIV risk context.

Acknowledgments

The data collection for this study was supported by a grant from the Developmental Core of the Emory Center for AIDS Research. Secondary data analysis was supported by a minority supplement grant from the National Institute of Child Health and Human Development (parent grant R01 HD 041724). José A. Bauermeister was supported by training grant T32 MH19139 from the National Institute of Mental Health. The authors gratefully acknowledge the support of Carlos del Rio and Kimberly Sessions, the assistance of Rachel Albalak with the development of the quantitative instrument and of Hrishikesh Chakraborty with the preliminary quantitative analyses, and institutional support from the Department of Global Health at Emory University.