S EX RATIO AND ITS EFFECTS ON SEXUAL BELIEFS AND PRACTICES: A COMPARISON OF WHITES AND BLACKS by BROOKE ALISON DORSEY B.A., North Carolina Central University, 2005 M.A., University of Denver, 2007 A thesis submitted to the Faculty of the Gradua te School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Health and Behavioral Sciences Program 2015
ii 2015 BROOKE ALISON DORSEY ALL RIGHTS RESERVED
iii This thesis for the Doctor of Philosophy degree by Brooke Dorsey Holliman has been approved for the Health and Behavioral Sciences Program by Ronica Rooks, Advisor Debbi Main, Chair Miriam Dickinson Sharon Devine Date April, 24, 2015
iv Dorsey Holliman, Brooke A. (Ph. D., Health and Behavioral Sciences) Sex Ratio and Its Effects on Sexual Beliefs and Practices: A Comparison of Whites and Blacks Thesis directed by Associate Professor Ronica Rooks. ABSTRACT Nationally representative data was used to examine whether sex ra tio influences individuals' sexual beliefs and practices to account for racial disparities in sexually transmitted infections (STI). Data were derived from wave II of the National Longitudinal Study of Adolescent Health (Add Health) Participants were age d 11 to 23 years old; analyses were limited to non Hispanic Whites and Blacks. Descriptive analyses statistical tests were used to explore and discover bivariate associations. Logistic regression analyses were used to capture the relationship between gende r, race, and sex ratio. Black participants were more likely than Whites to report having used a condom. Black participants were also more likely than White participants to report using a condom more frequently. Men were more likely than women to report usi ng a condom more frequently. Men were also more likely than women to report that they believed their chances of getting AIDS or other STIs were high. Participants in Census tract communities with higher female to male ratios were less likely to believe the ir chances of getting AIDS was high. Black young adults in the United States are at elevated STI risk. Racial differences in condom use, condom use frequency, and perceived STI risk could not be explained by racial differences in the availability of partne rs. Factors other than sexual beliefs and practices driven by sex ratio appear to account for racial disparities, indicating the need for individual level interventions.
v The form and content of this abstract are approved. I recommend its publication. Appr oved: Ronica Rooks.
vi ACKNOWLEDGEMENTS I am extremely honored, proud, and thankful to have had the support of family, colleagues, and friends throughout this process. Completing this degree was not easy but it was not impossible. Along the way there were many people who inspired me to keep going. In particular I would like to thank and acknowledge my advisor, Dr. Ronica Rooks, for her tireless optimism, guidance, and emotional/moral support throughout my graduate career. I would not have made it through this process without the love and support from my family. My husband, Olusanya, was very supportive and understanding throughout this whole process. This degree is as much his as it is mine. He encouraged me to push through very difficult times. My hu sband was as dedicated and committed as I was to completing this degree. He was willing to help me in anyway he could! Finishing this degree took on a whole new meaning when I had my daughter, Thistle, because I wanted her to dream big dreams. Special t hanks to my mom, Karen Talley, for her unwavering enthusiasm regarding my potential as a person, professional, and a woman ; as well as her constant willingness to assist at any hour in any way. I would also like to acknowledge my dad and step mom, Alfonzo and Rhobbie Dorsey, who lent an empathetic ear and boundless encouragement throughout my graduate pursuit. I am thankful for the consistent support from my close friend, Melinda Williamson. Finally, I would like to thank my dissertation committee for th eir valuable input and feedback. Additionally, I am thankful for the graduate process, as it has facilitated my personal, professional, and continuing development and understanding of the values and behaviors associated with a rewarding, balanced, and hea lthy lifestyle.
vii TABLE OF CONTENTS CHAPTER S I. 1 8 Purpose of St udy .. 8 11 Research Questions and Hypotheses 12 13 14 II. THEORETICAL FRAMEWO RK AND REVIEW OF THE LITERATURE 15 17 Sexual Risk . 17 18 Partner Pool .18 20 Dyadic Power . .20 21 Reason/Meaningfullness of Relationships 21 24 Sexual Networks and Reasons for Higher Concurrency among Blacks .. . .. 2 4 29 Sexual Attitudes and Beliefs .. 29 32 Destabilizes Existing Relationships . 32 36 Race and Gender Neutral Model .. 37 Race and Gender Specific Model 38 39 III. RESEARCH DESIGN AND METHODS 40 Purpose . 40 Statistical Analysis .. .. 40 43 Data .. . ... 43 44 Sample 44 45 Measure .. .. .. 45 48
viii TABLE OF CONTENTS cont d IV. RESULTS . 49 Descriptive Analyses .. . 49 59 Multivariate An alyses .. .. 59 68 Summary .. .. 68 69 V. DISCUSSION .. . .. 70 Purpose of Research .. . 70 71 Summary of Findings .. ... . 71 75 Interpretation of Finding s .. . 75 77 Context of Findings .. .. 77 78 Implications .. .. 78 80 Strengths and Limitations .. .. 80 82 Future Directions .. .. 83 87 REFERENCES .. ... . 8 8 94
ix LIST OF TABLES TABLE 4.1 . 4.2 51 52 4.3 54 Groups 4.4 . 4.5 ..56 getting other STIs by ever use a condom 4.6 Means and standard deviations for c ondom use frequency, perceived 57 chances of getting AIDS, and perceived chances of getting other STIs by race 4.7 Means and standard deviations for condom use frequency, perceived chances of getting AIDS, and perceived chances of getting o ther STIs by gender 4.8 .. 58 perceived chances of getting AIDS, perceived chances of getting other STIs and sex ratio 4.9 Cross .. .59 4.10 Logistic Regression analysis relating Ever Using a Condom to Gender .. .. .. 60 Race and Sex Ratio 4.11 Logistic Regression analysis relating Ever Using a Condom to Gender ......61 Race Sex Ratio, Gender*Race, Gender*Sex Ratio, Race*Sex R atio, and Gender*Race*Sex Ratio 4.12 Ordinal Logistic Regression analysis relating Condom Use Frequency to Gender, Race, and Sex Ratio 4.13 Ordinal Logistic Regression analysis relating Condom Use Frequency .. .. ..62 63 to Gender Race, Sex Ratio Gender*Race, Gender*Sex Ratio, Race* Sex Ratio, and Gender*Race*Sex Ratio
x LIST OF TABLES 4.14 Gender, Race, and Sex Ratio 4.15 Ordinal Logistic Regression relating C .. 4 Race, Sex Ratio, Gender*Race, Gender*Sex Ratio, Race*Sex Ratio, and Gender*Race*Sex Ratio 4.16 Stratified Ordinal Logistic Regression analysis relating Perceived . 6 5 Chances of Getting AIDS to Gender, Sex Ratio Gender*Sex Ratio (for the White Sample) 4.17 Stratified Ordinal Logistic Regression analysis relating Perceived 6 Chances of Getting AIDS to Gender, Sex Ratio, Gender*Sex Ratio (for the Black Sample) 4.18 Ordinal Logistic Regression analysis relating Perceived Chances 7 of Getting Other STIs to Gender, Race, and Sex Ratio 4.19 Ordinal Logistic Regression analysis relating Perceived Chances 68 of Getting Other STIs to Gender, Race, Sex Ratio, Gender*Race, Gender *Sex Ratio, R ace*Sex Ratio, and Gender*Race*Sex Ratio
xi LIST OF FIGURES FIGURES 1. Rates of Reported Gonorrhea Cases by Race/Ethnicity and Sex, U.S. 2013 . 2 2. Rates of Reported Chlamydia Cases by Race/Ethnicity and Sex, U.S., 2013 3. Rates of Reported Syphi lis Cases by Race/Ethnicity 4 4. Percentage Distribution of HIV by Race/Ethnicity, U.S., 2013 .. 5 5. Percentage Distribution of HIV among Females by Race/Ethnicity, U.S., 2 013 6 6. U.S. Ratio of Men to Women among Race and Ethn icity 8 7. Conceptual Model 14
1 CHAPTER I INTRODUCTION The increase in rates of sexually transmitted infections (STIs) among Black 1 men and women is alarming and the cause of much concern within the Black community. Finding s from the 2013 Centers for Disease Control and Prevention (CDC) sexually transmitted disease 2 (STD) surveillance report show persistent racial disparities in STD rates. Despite the fact that the national gonorrhea rate is at the lowest level ever recorde d, this rate is declining at a slower pace among minorities. The rate of gonorrhea among Blacks in 2013 was 12.4 times the rate among Whites (426.6 and 34.5 cases per 100,000 population, respectively) (see Figure 1) (CDC, 2014). In fact, Blacks accounted for 58.4 percent of reported gonorrhea cases with known race/ethnicity. As in previous years, the disparity in gonorrhea rates for Blacks in 2013 was larger in the Midwest and Northeast than in the West or South (CDC, 2014). Considering all racial/ethnic and age categories, the 2013 report indicates young Black women bear the heaviest gonorrhea prevalence (e.g., rate among those aged 15 19 is 1,768.5 per 100,000 women; rate among those aged 20 24 is 1,949.1 per 100,000 women) (CDC, 2014). Rates of gonorr hea are also alarmingly high among Black men. In 2013, the rate of gonorrhea among Black men aged 20 24 years was 13.0 times the rate among White men in the same age 1 race groups of Africa. The term Black is used in the text of this paper to refer to the U.S. born African American population. 2 Sexually transmitted diseases (STDs) are caused by infections tha t are passed from one person to another during sexual contact. These infections often do not cause any symptoms. Medically, infections are only called diseases when they cause symptoms. For this reason, STDs are also called sexually transmitted infections (STIs). The text of this paper will use the term sexually transmitted infection to refer to STDs and STIs.
2 group (CDC, 2014). Black men aged 25 29 years had a gonorrhea rate 10.4 times the rate o f White men in the same age group (CDC, 2014). Figure 1: Rates of Reported Gonorrhea Cases by Race/Ethnicity and Sex, U.S., 2013 Source: Centers for Disease Control and Prevention, 2014 Similar prevalence disparities are found with chlamydia and syphil is. In 2013, the rate of chlamydia among Blacks was more than six times higher than Whites (see Figure 2) (CDC, 2014). For Black women the rate of chlamydia was nearly six times higher than the rate among White women (CDC, 2014). The rate was highest amo ng young Black women aged 15 24. In this age group, there was one chlamydia case reported for every 14 Black women (14,250.3 per 100,000 women) (CDC, 2014). The same year, the chlamydia rate for Black men was almost 8 times as high as the rate among Whit e men.
3 Figure 2: Rates of Reported Chlamydia Cases by Race/Ethnicity and Sex, U.S., 2013 Source: Centers for Disease Control and Prevention, 2014 Blacks accounted for more than a third (37.3 percent) of all syphilis cases in 2013, which was 5.6 times the rate for Whites (see Figure 3) (CDC, 2014). The rate among Black women was 15 times the rate among White women; the rate of syphilis among Black men was 5.3 times the rate among White men. Rates among both Black men and women aged 20 29 years remaine d highest among Blacks (e.g., rate among those aged 20 24 is 17.0 cases and 96.4 cases per 100,000 population, respectively; rate among those 25 29 is 12.9 cases and 97.2 cases per 100,000 population, respectively) (CDC, 2014).
4 Figure 3: Rates of Reporte d Syphilis Cases by Race/Ethnicity and Sex, U.S., 2013 Source: Centers for Disease Control and Prevention, 2014 Blacks have the most severe burden of HIV of all racial/ethnic groups in the United States (see Figure 4) Compared with other races and ethni cities, Blacks account for a higher proportion of new HIV infections, those living with HIV, and those ever diagnosed with AIDS (CDC, 2013). Blac ks accounted for an estimated 46 % of all new HIV infections among adults and adolescents (aged 13 years or old er) in 2012 despite representing only 12% of the US population ( CDC, 2013 ). Considering the smaller size of the Black population in the United States, this represents a population rate that is 8 times that of Whites overall. In 2012, Black s had the larg est percentage (46 %) of the estimated 47,989 diagnoses of HIV infection in the United States (CDC, 2013). The same year, an estimated 14,102 Blacks were diagnosed with HIV infection ever classified as stage 3 (AIDS) in the United States (CDC, 2013).
5 Fig ure 4: Percentage D istribution of HIV by Race/E thnicity U.S., 2013 Source: Centers for Disease Control and Prevention, 2014 The disproportionate impact of HIV/AIDS on Black women is also seen in incidence and prevalence data (see Figure 5) When comparin g groups by race/ethnicity, gender, and transmission category, the fourth largest number of all new HIV infections in the United States in 2010 (5,300) occurred among Black women with heterosexual contact (CDC, 2012). Of the total number of estimated new HIV infections among women, 64% (6,100) were in Blacks, 18% (1,700) were in Whites, and 15% (1,400) were in Hispanic/Latino women (CDC, 2012). At some point in their lifetimes, an estimated 1 in 32 Black women will be diagnosed with HIV infection. While numerous factors contribute to the extreme racial disparity in STIs, reasons for its persistence remain poorly explained.
6 Figure 5: Percentage D istribution of HIV among Females by Race/E thnicity U.S., 2013 Source: Centers for Disease Control and Prevent ion, 2014 Research into the cause of STIs primarily focuses on individual level risk factors and calls attention to the importance of number of partners, behaviors, and condom use. Particular emphasis has been placed on the impact of demographic characte ristics (e.g., age, race, and sex), socioeconomic status, and risky sexual (e.g., numerous sexual partners and infrequent condom use) and nonsexual (e.g., drug use) behaviors (McQuillan Kruszon Moran, Kottiri, Kamimoto, Lam, Cowart, Hubbard & Spira 2006; Dunn, 2005; McNair & Prather, 2004). However, individual level risk factors alone have proven inadequate to explain the substantial heterogeneity in the diffusion patterns of STIs in the U.S. both between and within racial/ethnic groups. For example, al though, Black men tend to report more sex partners than White men, differences of individual high risk sexual behaviors have not been attributed to the disparity in STI rates. In addition, while socioeconomic status (SES) is a well known risk factor for S TIs, and the continuing racial disparity in income, wealth, education and other forms of economic
7 resources could be responsible for the marked difference in STI rates, the significant racial disparity in STI rates persists after controlling for SES and ot her potential confounders (Adimora & Schoenbach, 200 5 ). As a result, structural factors defined as barriers to, or facilitators of, an individual level and grou p Schoenbach, Boras, Martinson, Donaldson, Stancil, & Fullilove 2004). Health disparities are not exclusive to STIs; however, their nature, pattern, and distribution are uniquely complex wit h regard to Blacks. In Black communities, the high sex ratio seems to contribute to the racial disparity in STI rates (Senn Carey, Vanable, Urban & Sliwinski 2008; Adimora, Adaora & Schoenbach, 2005; Dunn, 2005), yet much about its influence remains unkn own. The sex ratio is a common measure used to describe the balance between males to females in the population. Many researchers and organizations measure sex ratio as the number of males per 100 females. A sex ratio of 100 indicates an equal number of males and females, with a sex ratio under 100 indicating a greater number of females. When there are more men than women, the sex ratio is low; when there are fewer men than women, the sex ratio is high. The Black population increased faster than the tot al population between 1990 and 2000 (U.S. Census Bureau, 200 1 ). In addition, between 2000 and 2010, among Blacks the male population grew at a slightly faster rate than the female population (U.S. Census Bureau, 20 11 ). Despite these trends, Blacks have t he lowest male to female sex ratio ( see Figure 6 ), with 90.5 males to every 100 females compared to other race and ethnicity groups in the U.S. (U.S. Census Bureau, 20 01 ).
8 Figure 6 : U.S. Ratio of Men to Women among Race and Ethnicity Source: Ce nsus 2000 Summary File 1  Census 2000 data is parallel to the data years analyzed in this study. The social environment created by an imbalance between men and women not only shapes opportunities to form heterosexual relationships that are sexual, it also interferes with existing partnerships and increases participation in concurrent sexual partnerships, which have been shown to increase popula tion transmission of STIs (Khan Behrend, Adimora, Weir, Tisdale & Wohl 2011; Adimora & Schoenbach, 200 5 ; Man hart Aral, Holmes & Foxman 2002). As a result, evidence concerning the impact of structural and societal factors on behaviors that affect the likelihood of STI transmission suggests that the sex ratio imbalance among Blacks may help create and maintain Purpose of the Study Research examining population specific issues that must be considered when evaluating the socio cultural context of sexual risk taking behavior in the Black community is limited. Within the li terature, several factors have been identified as contributing to the social context of risk taking for Black women (Dunn, 2005; McNair & Prather, 2004). These factors are attributed to the unique social and cultural dynamics found in the Black community, which often
9 in the literature are: the sex ratio imbalance (Pouget Kershas, Niccolai, Ickovics & Blankenship 2010; Adimora & Schoenbach, 2005; Dunn, 2005; McNair & Prather, 2004); low levels of condom use (Senn et al., ., 200 8 ; Payne, 2008; Cornelius, Okundaye & Manning, 2000); and high rates of STIs and risk behaviors among Black men (Aral Adimora & Fenton, 2008; Newman & Berman, 2008; Adimora & Schoenbach, 200 5 In many ways, the sex ratio imbalance provides an explanation and mechanism for each of the aforementioned aspects to transpire. Other than postwar male shortages experienced by various countries, the Black community has undergone the most persistent and severe shortage of men of any subculture since documentation of modern censuses ( Adimora, A., Schoenbach, V., Bonas, D., Martinson, F., Donaldson, K., & Stancil, T. 2002). The sex ratio among Blacks is among the lowest ever recorded due to higher mortali ty rates among Black male infants, children, and adults from disease and violence; and the disproportionate incarceration of Black men (McNair & Prather, 2004). The gender imbalance is even more prominent when considering the skewed Male Marriageable Pool Index (MMPI), which is the ratio of employed men to women of the same age and race (Adimora & Schoenbach, 2005). Specifically there is a shrinking pool of economically stable or marriageable men because Black males have high rates of premature mortality and involvement in the criminal justice system (Alexander, 2010). Therefore, marriageable Black partners, are less likely to enter into monogamous relationships, and ca n attract women without offering many incentives (Adimora & ratio imbalance in the Black community acts as a catalyst for generating and maintaining conditions in which sexual risk taking occurs at a higher rate.
10 Qualitative data suggest that some Black women perceive that male shortage facilitates female partner acquisition for men (Ferguson Quinn, Eng & Sadelowski 2006; Adimora et al., ., 2002, Adimora and Schoenbach, 200 5 Adimora S choenbach & M artinson, 2001). Several qualitative studies have examined the sex ratio among Blacks as a potential determinant of concurrent sexual relationships (Adimora & Schoenbach, 200 5 Adimora et al., ., 2002). These studies suggest that male shortage supports part nership concurrency and leads to more dense sexual networks. Quantitative studies of sex ratio, sexual behavior, and STI rates in the U.S. have had mixed results. Contradicting the previously mentioned qualitative results, a recent clinical study found t hat sex ratios calculated at the Census tract level were not related to the number of partners among men and were positively related to those among women (Senn et al., ., 2008). Further analysis indicated that this relationship was due to an increase in exch anging sex for money or drugs among women in tracts with higher sex ratios. The relationship between sex ratios and sexual behavior has also been studied quantitatively in international settings, with results generally supporting a negative association (B enefo, 2008; Smith & Subramanian, 2006; Schmitt, 2005). Conflicting findings from these studies highlight the importance of addressing if and how sex ratio is associated with STI risk among Blacks. Addressing this disagreement in findings within the liter ature is necessary to develop prevention strategies that speak to determinants of STIs in Black communities. Historically, STI prevention and control strategies have included surveillance, clinical services, partner management and behavioral interventions which are often viewed and implemented independently of each other (Aral 2004; Aral, et al., 2008; Barrow Berkel, Brooks, Groseclose, Johnson & Valentine 2008). Although gender roles and power relations are taken into account by STI prevention programs
11 ratio imbalances, making training in communication and negotiation skills difficult behaviors to sustain. In order for traditional strategies to meet the needs of Black communities and improv e dynamics driving disparities, such as how the effects of sex ratio could vary across race and gender. Therefore, the impact of sex ratio imbalance must be uncove red in order for the public health agenda to effectively address STI prevention within the Black population. To date, much about the influence of the sex ratio imbalance on the sexual beliefs and practices of Black women remains unknown. Research explori ng the impact of imbalanced sex ratios is limited and the literature is inconsistent regarding the influence of mate availability on dimensions of sexual activity. Moreover, the disproportionate incarceration of Black men in recent decades, as well as the remarkable lack of scholarly research in this arena, make this project both timely and significant. The goals of this research were to: 1) investigate the interactive effects of race, gender, and sex ratio related to STI risk; 2) contribute significantly to the small but growing body of literature on imbalanced sex ratios related to STI risk; 3) position the effects of variations in the sex ratio within larger discussions of STIs; and 4) contribute to prevention and intervention efforts that address the t ransmission of STIs among Blacks.
12 Research Questions and Hypotheses In order to assess the impact that sex ratio imbalance is having on STI risk among Blacks, the following research questions and hypotheses guided this project. 1. Is there a relationship be tween sex ratios and sexual risk taking? Does this relationship vary by race and gender? a. Hypothesis 1 Sex ratios will be related to condom use. Women who live in high sex ratio areas will be less likely to report using condoms. This relationship will be significant with a larger coefficient for Blacks vs. Whites specifically Black women because Black women have less negotiating power in their intimate relationships than White women at similar sex ratios. b. Hypothesis 2 Men who live in low sex ratio a reas will report more frequent use of condoms. This relationship will be significant with a larger coefficient for Blacks vs. Whites specifically Black women because Black women have less negotiating power in their in timate relationships than White women at similar sex ratios. 2. Is there a relationship between sex ratios and STI risk perception? Does this relationship vary by race and gender? a. Hypothesis 1 Sex ratios will be related to AIDS and other STI risk perception. Men who live in low sex ratio are as and women who live in high sex ratio areas will report less STI risk perception. This relationship will be significant with a larger coefficient for Blacks vs. Whites specifically Black women because the availability of partners has a greater influenc e on Black s STI risk perception than White at similar sex ratios.
13 Scope of the Study This research hypothesized that characteristics of the population such as the sex ratio affect beliefs and practices regarding sexual risk taking. This r esearch explored the magnitude of these aspects, in relationship to low sex ratio among Blacks, to examine their effect on STI transmission among Black women. Though, specific interest lies in the impact sex ratio has on Black women, this research utilize d Black men, White men, and White women as comparison groups. I expected that sex ratio influences STI risk perceptions and sexual risk taking, but that the effects may differ by gender and by race. The effects would all have a greater impact on Blacks, particularly Black women. The conce ptual model (Figure 7 ) summarizes the hypothesized paths between sex ratio and STI risk perceptions and behaviors. Single arrows represent direct effects on the outcome. Double headed arrows represent conditional or in teractive effects on the outcome. This research investigated whether the effects of race, gender, and sex ratio are dependent upon one another, in addition to or in place of having direct effects. This research provides an analysis of the potential direc t and/or conditional effects of sex ratio, gender, and race on beliefs about STI risk and risky sexual practices.
14 STI RISK PERCEPTION AND RISKY BEHAVIOR Figure 7 : Conceptual Model This research utilized multiple logistic regression and ordinal logistic regression to examine the relationship between sex ratio, and perceptions of sexual beliefs and practices. Particular attention was paid to variables regarding sexual risk taking, such as the frequency of condom use and STI risk perception. The items illustrated in th e proposed conceptual model were significant, though sensitive to model specification, providing evidence that sex ratio impacts the Black community beyond the prevalence of infection in the partner pool, which is currently propositioned as driving STI tra nsmission among Blacks. Therefore, investigating the influence of sex ratio provided a unique pathway towards understanding the importance of social determinants of disease and health within the Black population. SEX RATIO GENDER SEX
15 CHAPTER II THEORETICAL FRAMEWORK AND REVI EW OF THE LITERATURE The guiding theoretical framework for this project is a sex ratio perspective that examined the relationship between variations in sex ratio and beliefs regarding STI risk perception and sexual practices. The most prominent theoretica l attempt to relate the sex ratio to & Secord 1983; South & Trent, 2010; Uecker & Regnerus, 2010). This work (Guttentag & Secord 1983) combines demographic data with elements of social exchange theory t o illustrate the relationships between sex ratio and closely allied with their numerical representation. Sex ratio theory primarily considers how imbalanced sex ratios interact with gender inequality to affect marriage, divorce, and the growth of single parent families (South & Trent, 2010; Uecker & Regnerus, 2010). Demograph ic opportunity is the principal foundation of sex ratio theory; the likelihood of forming cross sex relationships such as marriage and other romantic associations is shaped largely by the number of available opposite sex members with whom such associations can be formed. In general, prior research has confirmed the propositions implied by Guttentag and marriage market will attempt to secure the best match possible given their own resources and mate preferences (England & Farkas 198 7 ; Lichter, Anderson, & Hayward, 1995; South & Ll o y d, 199 2 ). According to sex ratio theory, mate preferences and the criteria used for selection are shaped by social norms as well as fin ancial support, child rearing, and domestic services (Albrecht Fossett, Cready & Kiecolt, 1 997; Buss & Barnes 1986; Mare & Winship 1991). Existing work has focused to date primarily on outcomes such as marriage prevalence and mate
16 quality (South & Tren t, 2010; Uecker & Regnerus, 2010; Albrecht & Albrecht, 2001). South & Trent (1987) found statistically significant relationships between the sex ratio and rates of female marriage, fertility, divorce, illegitimacy, and literacy. Albrecht & Albrecht (2001 ) used sex ratio theory to explain variations in marriage prevalence, mate quality, and types of family structures in non metropolitan counties. Although sex ratio theory is based on heterosexual assumptions its failure to include non heterosexual relat ionships, i.e., homosexuality, bisexuality, etc., limit s the validity of the data analysis for currently acknowledged sexual behaviors that impact the prevalence of STIs. Subsequent to the Add Health data collection, people may have become more open about the ir sexual orientation, and therefore, more people are admittedly homosexual or bisexual And certainly bisexuality impacts the topic of concurrent relationships, unknown participation in a relationship that is presumed to be based on heter osexual only activity, and the exposure to risk taking in sexual partners. Sex ratio research on Blacks has focused on the stability of Black families and found similar associations between sex ratio and illegitimate births, marriage, and single parent f amilies. Guttentag and Secord argue that social consequences of a high sex ratio do not seem to be directly produced by the historical circumstances that Blacks have experienced (Guttentag & Secord, 1983) Additionally, they conclude a high sex ratio inf about having children, their economic role as mothers, and increased rates of depression. Furthermore, Guttentag and Secord ascertain high sex ratio is correlated with Black men expecting to have the upper hand in their rel ationships with women, and expecting women to compromise their own needs in order to meet male demands (Guttentag & Secord, 1983) Overall, current literature using sex ratio theory has shown that the implications of a high sex
17 ratio include increases in t he number of female headed households, lower marriage rates, and lower proportions of children being raised by married couples (South & Trent, 2010; Uecker & Regnerus, 2010). Sexual Risk Taking The current study extends sex ratio theory to the outcomes o f sexual beliefs and practices, such as sexual risk taking. Research examining the sex lives of Black women has been confined largely to the study of sexual activity and contraceptive use of teenage girls. This rather narrowly focused research effort has established that: Black heterosexual teenagers initiate sexual intercourse at an earlier age than White teenagers; they are less likely to know about and use contraception than White teenagers; and they wait for a longer period of time between initiating sex and initiating contraception (Manhart et al., ., 2002; Harvey, Bird, Galavotti, Duncan, & Greenberg, 2002; Upchurch Aneshensel, Sucoff & Levy Storms 199 9 ). Though Blacks often engage in risk behavior at rates equal to, or less than, Whites, they often f ace consequences disproportionate to their actions ( Barrow, Berker, Brooks, Gloseclose Johnson & Valentine 2008). Some literature suggests Blacks are more likely than any other racial or ethnic groups to suffer the negative consequences of sexual behavi or. For example, Black women have among the highest STI prevalence but do not have the highest levels of risk behaviors (Aral, 200 4 ). Data from the National Survey of Family Growth (2002) revealed that fewer Black women than White women reported having f our or more partners in the past year and 15 or more in their lifetime. One explanation is that a high prevalence of infection in the pool of potential partners can be more important than the number of sexual partners in determining individual infection r isk (Aral et al., ., 2008; Barrow et al., ., 2008; McNair & Prather, 2002). Partner pool is also referred to as the relationship market (Banks, 2011). Partner
18 pool represents individuals available for relationships, whereas sex ratio delineates the actual numb er of males to females in the population. Partner Pool It appears that partner pool may have a moderating effect. In regards to the influences of sex ratio, partner pool changes the degree of association between sex ratio and the dependent variables and i mplies when or under what conditions particular outcomes can be expected. For instance, while Black women typically limit themselves to dating only Black men (Banks, 2011; Johnson & Raphael, 2006; Dunn, 2005; McNair & Prather, 2004), and therefore are pri marily affected by race specific sex ratio, White women are generally not as apprehensive about dating outside of their race/ethnicity, and therefore overall sex ratio should be taken into consideration to examine the influence of the partner pool. Having a larger pool of potential partners decreases the probability of sex ratio having the same influences on White women that it has on Black women. Not only will Black women have stronger associations between sex ratio and the items illustrated in the propo sed conceptual model, they will also have a greater risk of contracting an STI since they are exposed to high risk partners. Several studies have demonstrated the n & Catania, 2000; Ickovics & Rodin, 1992; Kalichman, Rompa, Luke & Austin, 2002). Whereas White Americans acquire STIs predominately when they engage in high risk behaviors, Blacks acquire them through low risk behaviors because prevalence infection in the population is high (Morris Morris & Ferguson 200 9 ; Aral, 200 4 ) The fact that even with only one lifetime partner, Black women are at greater risk for STIs (Barrow et al., ., 2008 ) raises concerns about how much individual behavioral interventions based on the findings of current
19 research can accomplish. Currently, Black women represent the fastest growing group of individuals infected with HIV in the United States. Furthermore, Blacks carry the largest disease burden for gonorrhea, chlamydia, and syphilis in the U.S. (CDC, 2014) ; therefore, uncovering more detail about important aspects of Black sexual relationships is needed to promote sexual health. At the individual level, the most important risk factor for STIs is sex with an infected partner (Aral et al., ., 2008). Therefore, the high rates of STIs among Black men have important other rac ial/ethnic groups (Banks, 2011), it is probable that the majority of heterosexual Black women have sexual contact with Black men. The rate of STIs among Black males is currently greater than the rates of any other male racial/ethnic group (CDC, 2013). Ac cording to the CDC, Black men had an STI incidence rate that was six times that of White men in 2010. As a result, the higher STI prevalence among Blacks perpetuates itself, because infection risk is partly determined by the prevalence of infection in the population from which one selects sex partners (Aral Holmes, Padian & Cates 199 6 ; Koopman & Longini, 1994 ; Allard, 1990 ). Beyond the effects of STI transmission generated by the partner pool, a shortage of Black males can be viewed as increasing the bar gaining power of Black men and reducing the bargaining power of Black women in sexual relationships. Specifically, this occurs by reducing available alternative relationships for Black women and increasing available alternative relationships for Black men Research suggests sex ratio imbalance creates an environment where heterosexual Black men have access to numerous sexual partners (Dunn, 2005; McNair & Prather 2004). Concurrently, Black women may feel silenced by the virtual unavailability of male sex ual partners given the fact that males have more options for sexual partners should a
20 relationship dissolve. In many ways, the additional competition among single Black women further increases the possibility for Black men to have sexual partnerships with many women. As a result, the sex ratio imbalance in Black communities has the potential to expose Black women to social forces driven by opportunity, which significantly increases their STI risk. Dyadic Power The shortage of Black men can serve to furth which has been recognized as a central issue in STI prevention. Lack of power in sexual relationships is described as a major obstacle to the practice of risk reduction behaviors (Senn et al., ., 20 08 ). Empirical rese arch indicates that reliance on use of condoms requires the active cooperation of the male partner, equal power distribution between sex partners, and the ability of the woman to persuade the male to use them (Senn et al., ., 20 08 ; Harvey & Bird 2004; Pulerw itz Amaro, DeJong, Gortmaker & Rudd 2002). Studies that have explored the effects of power is, the higher her reported rates of condom use, birth control, and the more control she has over sexual decision making (Senn 200 8 ; Harvey et al., 2002; Soet, Dudley, & Dolorio, 1999). Black men may exert more dyadic power in heterosexual relationships, therefore making it more difficult for Black women to negotiate safer s ex practices including condom use. An individual who lacks dyadic power in a relationship may be unable to convince a partner to engage in safer sexual behavior, despite having the information, motivation, and skills to do so. Lower levels of dyadic po about condom use due to concerns that the topic can lead to conflict and threaten the future of the relationship (Logan Cole & Lukefeld 2002). Therefore, it is reasonable to hypothesize that a risk sexual
21 behavior, including unprotected sex, to increase the chances of securing a partner. The more difficult it is to find a partner, the less likely women will be to make demands, such as insisting their partner use a condom, which could result in the loss of the partnership altogether. Data on condom use rates among Black women are consistent with this perspective. According to Cornelius et al., (2000) only one third of B lack women aged 14 to 44 years reported that their partners always use condoms. Black women who assume that finding an alternative romantic partner will be difficult may be less likely to mandate sexual behavior consistent with sexual health. Additionall y, due to the high sex ratio among Blacks these women may be involved in relationships in which they feel powerless in terms of sexual negotiation and as a result do not communicate with sexual partners about possible STI risks. A major problem in these r elationships is the lack of effective communication about sexual practices, particularly the use of condoms when partners are not mutually monogamous ( Cornelius Okundaye & Manning, 2000). Communication difficulties are aggravated by sex ratio imbalance b etween men and women in the Black community, however few studies have focused on how Black women manage the emotional, social, and cultural demands of a high sex ratio. Reasons/Meaningfulness of Relationships Existing research supports the hypothesis that the emotional, social, and cultural meanings attached to sex and sexual behavior contribute to the higher prevalence and increasing incidence rates of STIs among Black women. For instance, Cummings Battle, Barker & Krasnovsky (1999) found after interview ing 142 Black women that sexual risk taking is associated with a variety of factors. First, being in a relationship is a major source of self validation and esteem for women, particularly Black women. Second, having unprotected sex helps women maintain t heir beliefs that their partners are faithful. Third, Black women are often in denial regarding
22 their vulnerability to STIs. According to Foreman (2003), this denial stems from relationship ideals and status considerations. Foreman (2003) conducted in d epth interviews with 15 Black women aged 19 to 33 attending a 4 year university in Southeast, Texas. Trends from these studies suggest that contraception use and sexual risk taking behaviors may be related to normative attitudes and beliefs shaped by the high sex ratio for members of this population. Not only does the high sex ratio create an atmosphere in which Black men have access to a number resulting in rel atively more partnering between women with low risk behaviors and men with high risk behaviors. Although many women can clearly articulate some of the risks associated with sex, many Black women struggle with identifying themselves personally at risk. Al so, it is important to note that knowledge of the risks associated with unprotected sex does not mean that these women perceive themselves as at risk. In fact, according to Foreman (2003), many of these women had trouble conceptualizing themselves as bein g at risk. According to Leigh (1995), women are unlikely to alter their behavior unless they recognize their personal risk for acquiring STIs. Robinson Scheltema & Cherry (2005) confirmed these findings concluding that women frequently denied their vuln erability to STIs for extended periods of time, choosing sex and a man over self protection. In general, Black women are well informed regarding ways to make sex safer and the potential risk associated with their behaviors, however, they continue to engage in sexual behaviors that might put them at risk for acquiring STIs. The majority of Black women identify condoms as the safest sex measure, after abstinence, with no prompting (Foreman, 2003; Cornelius, Okundaye & Manning, 2000). However other safe sex behaviors, such as limiting
23 partners, participation in monogamous relationships, and engaging in non penetrative sex are rarely mentioned (Foreman, 2003). Rather than invoke their knowledge of risks, Black women are often likely to acquiesce and give in the belief that sexual intercourse creates a bond between the people engaged often ser ves as a barrier to condom use. Foreman (2003) found that as Black women discussed their willingness to take s exual risks and engage in behaviors, two significant themes emerged; both the expectation or longing for intimacy and the desire for a long term relationship seem to overshadow risk reduction decision to engage in sexual Black women were more likely to participate in unprotected sex or these behaviors if they were seeking a closer bond or more intim ate relationship with their partner. These findings suggest be achieved through sexual intercourse. The desire to feel more connected with a partner often ou relationship, beyond dating, may influence risk taking behavior. In addition to the imbalance of power in heterosexual relationships is the desire or self impos ed cultural requirement that women must be mated to be legitimized (Banks, 2011; Foreman, 2003). This norm suggests that a woman is incomplete without a man and thus may act as a barrier to consistent self protecting behaviors. Consequently, increased ri sk taking has also been expressed as a strategy for securing a long term relationship. Women seeking a long term relationship are likely to use condoms as a bargaining tool for leverage to secure that type of relationship. Many Black women confided that if they hope for a long term or serious
24 relationship with their partners, they will engage in more risky sexual behaviors, such as using unprotected sex as a tactic to secure a more substantial relationship with their partner (Foreman, sire to please her partner in the hopes that he will make her his one and only allows her to deny her own susceptibility to STIs. In contrast, some women are more likely to use condoms with men to whom they seek no long term attachment and employ condoms as a dating strategy. For these women, condom use seems to be a negative behavior useful for sending the message that nothing more than sex was expected. Sexual Networks and Reasons for Higher Concurrency among Blacks Another social factor that influence concurrent partnerships among Blacks in the U.S. Concurrent sexual partnerships are sexual relationships that overlap in time. Causes of concurrency patterns are complex, and social scientists have prop osed a variety of explanations. Most recent theories emphasize demographic and macroeconomic constructs that focus on mate availability and marital feasibility (Banks, 2011; Senn et al., ., 2008; Dunn 2005). Aspects of the social context appear to have the biggest influence on concurrency (Adimora & Schoenbach, 200 5 environment (Adimora & Schoenbach, 2005). Demographic features and other structural aspects individual behaviors, including sexual behaviors, transmission of STIs, and other health outcomes. Because social conte xt has an influence on sexual practices, variations in context could result in differences between Black and White populations in ways that foster more rapid dissemination of STIs among Blacks. A key demographic factor in these explanations is the sex
25 rat io. Community attributes such as sex ratio can affect norms for sexual practices by increasing the frequency and risk associated with sexual practices and impeding the ability of individuals to adopt preventive behaviors. Social science data suggest that the social and economic environment can dramatically term monogamous relationships (Adimora & Schoenbach, 200 5 ) For example, while Black women remain unmarried becau se they think they have too few options, Black men stay single because they think they have so many. Black men may see little incentive to settle down and enter a committed relationship when there are so many available women. Following this logic, when f aced with a relative shortage of Black men, Black women will more likely be exposed to frequent short term partnerships and engage in sexual relationships with men who have concurrent relationships with other women. Epidemiological studies indicate sexual networks and concurrent partnerships, which are both prevalent in Black communities, play a critical role in the transmission of STIs (Aral et al., 2008 ; Newman & Berman 2008). Compared with sequential monogamous partnerships, concurrent relationships may permit more rapid spread of STIs through a group of people who are connected by a sexual network (Adimora & Schoenbach, 2005). A sexual network refers to the chain of individuals Individuals at the periphery of a sexual network have one partner, while individuals who have more partners form a network core. The pattern least likely to spread infection is a population composed of individuals in long term monogamous relationships (Ar al et al., 2008; Newman & Berman 2008). However, the presence of a small number of individuals who change partners frequently has dramatic implications for transmission and persistence of STIs in a population. With regard
26 to disease transmission, the im portant characteristic of a network is the mixing between individuals at the core and periphery of sexual networks (Aral et al., 2008; Newman & Berman 2008). Therefore, the increase of STIs among Black males may be attributed to the fact that an abundan ce of women increases their chances of finding a willing sexual partner, the opportunity to have extramarital sexual intercourse and frequent sexual intercourse with multiple partners. As a result, sexual networks shaped in response to the high sex ratio have the potential to strongly influence the rate of STI transmission in the Black community. If Black women choose sexual partnerships within sexual networks of Black men who have relatively high risk of STIs, then they too will have a greater risk of in fection. Consequently, sex ratio imbalance serves as a potential explanation to the current disparity in STI rates. The high sex ratio in the Black community appears to establish core and periphery networks. Due to the sheer unavailability of Black men Black women tend to make up the periphery of sexual networks, while Black men serve as the core. Individuals at the core are important for STI transmission, if they are infected, because they contact more partners than those at the periphery (Adimora & Schoenbach, 2005; Laumann & Youm, 1999). Sex ratio imbalance not only makes it initially more difficult for Black women to find a partner, it also increases their likelihood of being exposed to high risk sexual networks when they do. Once a concurrent pa rtner acquires infection, transmission to a third person can occur without the delay involved in completing the first partnership and beginning the next. Since relationships overlap in time, early concurrent partners are not more protected from infection than partners acquired later in the sequence (Adimora & Schoenbach, 2005). The degree of connectivity of sexual networks also affects the likelihood of transmission across networks throughout the population.
27 This pattern of linkages can dramatically infl uence the speed of STI transmission and the number of individuals who are infected in a short time period in comparison to STI transmission among monogamous sequential sexual relationship networks (Aral et al., 2008; Newman & Berman 2008; Adimora & Schoen bach, 2005). Thus, the epidemiological study of STIs involves research into the extent to which infected persons have additional partners. There is a significant difference among race in the extent of participation in concurrent sexual partnerships. Blac ks and Whites appear to have separate sexual networks and have different numbers of concurrent sexual partnerships. Descriptive evidence from 4 nationally representative U.S. surveys found that reported rates of concurrency among Black men (aged 20 38 ye ars) ranged from 5% to 10% (Morris Kurth, Hamilton, Moody, Wakefield & Handcock 2009). On average, these rates were 3.5 times higher than those among White men. Rates were similar among young Black men (aged 19 25 years), with estimates ranging from 5% to 13%, but the disparities were slightly smaller (2.9 times higher than rates among Whites) (Morris et al., ., 2009). Among women, levels of concurrency were lower, but the race specific differential was still large. Approximately 4% of Black women report ed concurrent partners (Morris et al., ., 2009). Rates among adult Black women were 2.1 times higher than rates among Whites and rates among young Black women were 1.4 times higher than rates among Whites (Morris et al., ., 2009). This can partly be attributed to the fact that sex ratio does not have the same effect on sexual networks among Whites. Assortative sexual mixing is the formation of partnerships between people with similar characteristics (e.g., racial background or risk for infection) (Khan et al., ., 2011). Conversely, disassortative sexual mixing occurs when individuals with differential risk form partnerships. Black men and women in the U.S. are more likely to mix assortatively by race than are White
28 Americans, but more likely to mix disassortative ly by risk. Therefore, more frequent sexual contact occurs between those with many partners and those with fewer partners, a sexual mixing pattern that disseminates and maintains infection within the Black community (Khan et al., ., 2011; Barrow et al., ., 2008 ; Aral et al., ., 2008; D oherty, Padian, Marlow & Aral 2005; Laumann & Youm, 1999). Due to racially assortative mixing, the likelihood among Blacks that individuals in the periphery of a network will have partners in the core of a network is five times that of Whites (Laumann & Youm, 1999). Furthermore, the preference of same race partnerships among Blacks exacerbates the effects of a high sex ratio by increasing the number of concurrent sexual partnerships. Findings from the 1994 National Health and Socia l Life Survey, as reported in 2002, indicated that concurrent partnerships were more common among Blacks than Whites and more common among men than women (Laumann & Youm 1999 ). Nearly one third (31 percent) of Black men reported having been involved in c oncurrent partnerships during the preceding year, compared with 23 percent of Hispanic men, 13 percent of White men, 12 percent of Black women, nine percent of Hispanic women, and six percent of White women (Laumann & Youm, 1999). In turn, data from the 1 995 National Survey of Family Growth indicate concurrent partnerships are nearly twice as prevalent among Black women 15 to 44 years of age (21 percent in the preceding five years) than among White women of the same age (11 percent in the preceding five ye ars) (Adimora & Schoenbach, 2005; Laumann & Youm, 1999). In 2002, Blacks reported a substantially higher prevalence of concurrent partnerships (21 percent) than the rest of the population (12 percent) (Laumann & Youm, 1999 ). Recent research indicates th at Black men report having more lifetime partners than any other racial or ethnic group. A recent study examining concurrent sexual partnerships at the
29 population level (2002 National Survey of Family Growth) found that during the preceding year Black men (28 percent) were more likely to have had multiple sex partners than were White men (13 percent) (A dimora et al., ., 2004 ). The prevalence of concurrent partnerships influences both the speed of STIs spread during its initial phase and the number of individ uals who are infected at a later time period. Infection is much less likely to propagate in a population composed of individuals in unconnected triads (each individual with 2 partners) than in a population composed of individuals with 2 partners within a completely connected network (Koopman & Lynch 1999). These trends suggest sex ratio is likely a key determinant of the structure of sexual networks and concurrent partnerships. These factors warrant examination because they affect the availability of sex partners and influence partnership choices, thus STI transmission. Sexual Attitudes and Beliefs In contrast to the wealth of literature relating sexual networks and concurrent partnerships, there are very few relevant studies exploring sexual attitudes an d beliefs among Blacks Interspersed through this literature, however, is a body of useful information that describes significant gender differences in attitudes toward sexual behavior and sexual activity (Senn et al., ., 200 8 Valentine, 2008; Adimora & Schoe nbach, 2005). For instance, Black boys and men have attitudes toward sexuality that differ sharply from those of Black girls and women. In general, Black men endorse premarital sex more frequently, have more sexual partners, and are more likely to have s ex outside of m arriage than Black women (Senn et al., ., 200 8 ; Adimora & Schoenbach, 2005; Harvey et al., ., 2002). These findings suggest that traditional sexual roles, which permit men to have sexual freedom but censure women for the same activities, though not exclusive to, are operating in the Black community.
30 In regards to attitudes about infidelity, research has focused on three domains: the relationship. Being Blac k, male, and well educated are all associated with permissive sexual values and extramarital sex (Treas & Giesen, 2000). Opportunities, namely potential partners and circumstances assuring secrecy, facilitate extramarital sex. Some Americans admit they w ould have extramarital sex if their mate would not find out (Treas & Giesen, 2000). Couples who do not live together, for example, have more opportunities and are more likely to have secondary sex partners. Furthermore, people who perceive alternative pa rtners to be available are not as likely to be as sexually exclusive. The lack of eligible Black men relative to the number of single Black women influences the likelihood that Black men will have the opportunity to have multiple sexual partners. The com petitive environment created by a high sex ratio may reframe the way in which an individual conceives of sexuality, in part by changing interest. As a result, Black women who fear remaining single may feel a need to take adva ntage of sexually permissive attitudes, perhaps despite traditional sexual roles as part of the culture in the Black community. Black men have more options for sexual partners, and therefore more opportunities for infidelity while in committed relationship s, as well as more options should a relationship dissolve (Banks, 2011). Because of their relative scarcity, Black men have the option of either moving sequentially from woman to woman or maintaining simultaneous relationships with different women. Black men recognizing their sexual options may be more prone to infidelity as & Prather 2004). The scarcity of eligible sex partners can lead Black women to place fewer demands on their male sex partners for fear of losing them and an increased tolerance for male
31 sexual infidelity (CDC, 2006; Thomas, 2006). The relative shortage of Black men improves their bargaining position in negotiating sexual relationships. In a rec ent study, Whyte (2006) examined the sexual assertiveness of low income Black women, and found that study participants often engaged in sex out of fear of being left, rather than out of desire, and concluded that these women were less likely to practice ri sk reduction. In fidelity or know they are having sex outside their relationship, they may accept this as long as they maintain main partner status and all of th e perceived emotional benefits of the relationship (Eyre Auerswald & Hoffman, 1998). Research also suggests that it is rare for women to talk partner status outweighs the understood STI risk. These findings reflect a high tolerance of infidelity within sexual activities, such as cultural norms and practices. One principle of sex rati o theory provides additional support regarding the impact of cultural norms and practices. Guttentag and Secord define structural power as control over the (G u ttentag & Secord, 1983 p. 26) While dyadic power derives directly from imbalanced sex ratios, structural power resides universally with men; therefore s sharply limited. Men existing relationships by prizing virginity in w omen until marriage, by emphasizing monogamous relationships, and by praising the roles of mother and homemaker (Guttentag and
32 Secord, 198 3 their life options by activating m characterized by fewer men and women marrying, low marital fertility, and frequent divorce (Guttentag & Secord, 198 3 ) Guttentag and Secord claim that the association between sex ratios and heterosexual relationships is conditioned by the possession of dyadic and structural power (Guttentag and Secord, 198 3) Destabilizes Existing Relationships Sociologists have theorized that a high sex ratio may impact ideals regarding romantic relationshi ps and discourage long term monogamy (Andrinopoulos K errigan & Ellen 2006; Schoenbach Adimora, Donaldson & Stancil 2000). In a study aimed at improving understanding of how Black adults select their sex partners, the investigators found that female pa rticipants described their sex partners as romantic partners, meaning partners who are respectful, honest, kind, and monogamous. The male participants, however, described 2 distinct types of sex partners: romantic partners and sex only partners. Whereas the men expected their sex only partners to have multiple sex partners, they expected their romantic partners to be monogamous (Andrinopoulos et al., ., 2006). In this study, women reported having sex both for intimacy and emotional support, while men have s ex to feel wanted, and for them, sex provides both physical and emotional benefits. The sex ratio imbalance places Black women at a disadvantage in negotiating and maintaining mutually monogamous relationships, because Black men can easily find another r elationship if they perceive their primary relationship to be problematic. Men who maintain multiple simultaneous partnerships may be confident that their primary partner will not end the relationship, because primary relationships, especially marriage, a re relatively difficult for Black
33 women to attain. Because marriage provides a constraint on sexual interactions, data concerning marital status and patterns have some relevance to studies of relationship stability or commitment and sexual behavior. Comp ared with those who are unmarried, married people are far less likely to have concurrent partnerships (Schoenbach et al., ., 2000; Laumann Cagnon & Michael 1994). A shortage of marriageable Black men and economic difficulties of Black women have been ident ified as major causes of the widespread concurrency among unmarried Blacks in their communities. Those who are unmarried are much more likely to be involved in concurrent sexual partnerships (Adimora et al., ., 2002; Laumann et al., ., 1994; Schoenbach et al., ., 2000). According to the 1995 National Survey of Family Growth, only four percent of married women (of all races) reported simultaneous partnerships; Whites were more than twice as likely (54 percent) as Blacks (25 percent) to be married (Adimora et al., ., 2 002). This difference exists despite the apparent similarity between Blacks and Whites in values concerning marriage and the family (Banks, 2011; Andrinopoulos, 2006). The dramatic difference in marriage rates for Blacks and Whites is a relatively recen t development. Black men were less often never married compared with white men until 1960, whereas Black women were less likely to be never married than white women up until 1970 (Norton & Mormon, 1987). The turning point comes in 1980, when both Black m en and women begin a sharp increase in the proportion never married by age 35 and age 45. During the past 30 years, Black men and women have begun marrying at later ages than ever before and now marry later than Whites (Cherlin, 1992). Between 1975 and 1985, the percentage of Black women age 20 to 54 who had never married rose from 20 percent to 35 percent; the proportion divorced rose from 22 percent to 31 percent; and the proportion of these who remarried decreased from 55
34 percent to 46 percent (Norton & Mormon, 1987). By 1993, 58 percent of Black men and 61 percent of Black women were currently unmarried, compared with 38 percent of White men and 41 percent of White women (Norton & Morman, 1987). In 1998, proportions never married were: 46 percent fo r Black men, 41 percent for Black women, 29 percent for White men, and 22 percent for White women (U.S. Bureau of the Census, 1998). Recent trends illustrate similar marital patterns among Blacks. Data from the 2004 American Community Survey (ACS) examine d the marital status of the population and found that only 40 percent of Black men were currently married, whereas 60 percent of White men were married ( Krieder, 2007 ). In fact, at every income level, Black men are less likely to marry than their White co unterparts (Banks, 2011). Similar patterns exist among women, with 29 percent of Black women currently married compared to 55 percent of White women (Krieder, 200 7 ). By 2010, the median age at first marriage for black women was 30 years in contrast with 26.4 years for white women. The same year, the median age at first marriage for black men was 30.7 years, compared with 27.8 years for white men. The causes of these marital patterns are complex and social scientists have proposed a variety of explanation s. Most recent theories emphasize demographic constructs that focus on mate availability and marital feasibility. A key demographic factor in these explanations is the high sex ratio of men to women among Blacks. The higher prevalence of unmarried Black women could also be attributed to the fact that Black women are less than half as likely as Black men to wed across racial lines. Only about one in 20 Black women are interracially married (Banks, 2011). Sex ratio theory compliments theories that call a ttention to demographic constructs. For example, a key premise of sex ratio theory is that members of the gender in short supply have an
35 advantage over members of the other gender because more alternative marriage partners are available to them, and this insight should apply to the outcome of sexual beliefs and practices. dyadic power, or lack thereof, are expected to apply to both sexes. Because of gender related d ifferences in mate selection criteria and incentives for marriage and family formation, hypotheses based on sex ratio predict that sex ratio variations will have different outcomes depending on which gender has the dyadic power. For example, Guttentag & S ecord (1983) argue that since men have weaker incentives to marry for economic reasons, in marriage markets where men are relatively scarce, they will use their dyadic power to obtain companionship without necessarily entering into a formal marriage arrang ement. In contrast, when women are in short supply, it is expected that they will convert their dyadic power into higher rates of more economically stable relationships such as marriage. According to sex ratio theory these effects will be linear and pos itive for women. A lower on marriage for financial support (Banks, 2011), they are predicted to use their bargaining advantage when potential mates are plentiful t o marry, and to marry higher status mates than of marrying (Guttentag & Secord, 1983). When the sex ratio is high (men are scarce), the relationship is expected to be positive. When the sex ratio is low (men are abundant), the relationship is expected to be negative. The rationale for this prediction is that the further the sex ratio is above 100, the less likely men are to marry because they lack poten tial mates. The further the sex ratio is below 100, the less likely men are to marry because they can obtain companionship and sexual relationships outside marriage.
36 Guttentag & Secord (1983) claim the effects of variations in the sex ratio are largely in dependent of other socioeconomic or cultural factors, and therefore the predictions should be accurate regardless of the population being studied; meaning the same predictions should hold true for Whites with similar sex ratios. Supporting this claim, Sch oen and Kluegel found only a small proportion of the pronounced Black White difference in female marriage rates was attributable to racial differences in the sex ratio (Schoen & Kluegel, 198 8). However, a recent analysis shows that Black women currently represent the most unmarried group in the U.S. because of the current shortage of Black men (Banks, 2011). Not only do Black women confront the worst relationship market of any group, research has also shown Black women needlessly worsen their situation b y limiting themselves to Black men (Banks, 2011; Johnson & Raphael, 2006; Dunn, 2005; McNair & Prather, 2004). Primarily engaging in relationships with Black men, has resulted in nearly 70 percent of Black women being unmarried (Banks, 2011). Instead of c onsidering relationships with men of other races, many Black women report trying to come to terms with the fact they may never get married (Banks, 2011). The implications of high sex ratio suggest the scarcity of Black men profoundly influences partner se lection, the sexual availability of Black women, the type of male sexual behavior that Black women tolerate, and participation in high risk sexual behavior. Prior research has focused on more objective family related behaviors such as marriage, childbeari ng, and divorce and ignored more subjective outcomes such as attitudes and beliefs. In response, influence sexual risk taking. Two models were examined: a race and gender neutral model and a race and gender specific model.
37 Race and Gender Neutral Model A race and gender neutral model would predict similar outcomes of the effects of sex ratio for me n and women regardless of race. If a race and gender neutral model holds true, beliefs regarding STI risk perception would be associated with sex ratios and condom use, however these associations will be similar by race and gender. Prior applications of sex ratio theory predict Blacks will endure social consequences given a high sex ratio and support from cultural factors. However, I dispute this claim and arg ue that sex ratio imbalance has given rise to a new set of social norms and mechanisms that people use to find sex partners, which are largely shaped by cultural factors. Engaging in concurrent partnerships in the Black community, for instance, has evolve d into an acceptable form of dating for many, which is a clear change in social norms (Ferguson et al., ., 2006). As a result, I hypothesized a race and gender neutral model would not apply to any of the contextual factors in the proposed conceptual model. Although previous research has identified that a high sex ratio sustains marital patterns among Blacks (Banks, 2011), sex ratio theory has not been used to examine Black White differences beyond marriage rates (South & Trent 2010; Senn et al., ., 2008; Albre cht & Albrecht, 2001), illegitimate births (South & Trent ., 2010, Uecker & Regnerus, 2010), and sin gle parent families (Albrecht & Albrecht, 2001 ; Buss & Barnes 1986; Mare & Winship, 1991). In light of recent STI trends among Blacks, this research address ed this gap in the literature by extending previous work to examine Black White differences between sex ratio and sexual beliefs and practices.
38 Race and Gender Specific Model A race and gender specific model suggests different dynamics for men and women as well as race. These differences are expected because of the overall greater power that men yield both within and outside the relationship dyad. I hypothesized a race and gender specific model would apply to all of the contextual factors in the propo sed conceptual model. According to this be less eager for, and desirous of, a mutual commitment when women are abundant and will have lower levels of commitment to a given relationship. Although women will be more eager and desirous of committed relationships when men are scarce, they will also report lower levels of satisfaction in relationships. When men are scarce women will have lowered expectations of commitm ent and be more likely to have relationships that they would otherwise not contemplate. Consequently, a high sex ratio will more likely result in women being involved in concurrent partnerships simply to have a companion. Ideal romantic relationships wil l not mirror actual satisfaction in current relationships and women will be more tolerant of infidelity. Additionally, a relative shortage of men gives women less dyadic power to leverage negotiation in their relationships. In an attempt to secure partne rships, women will have stronger motivations to engage in sexual risk taking, putting them at greater risk for STIs. However, when women are in short supply, men will not suffer the same consequences, at least not for the same reasons. They may actually report less sexual risk taking and report fewer concurrent partnerships simply because there is far less opportunity for relationships, sexual or otherwise. If a race and gender specific model holds true, beliefs regarding sexual risk taking would be rela ted to sex ratios and condom use, however these relationships would vary by race and gender. In particular, the fact
39 that Black women restrict themselves to dating only Black men, lead s to the following race and gender specific predictions. This research i s one of the first to evaluate sex ratio alongside a two sided race and gender model. To my knowledge, the only other paper to examine sex ratio and gender models is the study of spousal alternatives and marital relations by Trent and South (2003). In Tr ent and specific association is found between sex ratio and in happiness if divorced. In the model presented here, I predic ted the interaction between the sex ratio and beliefs would play an important role in explaining the differences in sexual behavior and perceived STI risk across Blacks and Whites. No prior study has as yet attempted to determine whether racial difference s in perceived STI risk can be explained by racial differences in the availability of alternative partners. This research is the first study to empirically examine the impact of an objective measure of sex ratio on the previously mentioned subjective meas ures to examine their impact on sexual behaviors, particularly in regards to STI risk perception.
CHAPTER III RESEARCH DESIGN AND METHODS Purpose The purpose of this study was to evaluate how the sex ratio is associated with socio cultural factors amon g Black women during their engagement in heterosexual relationships by testing whether the above proposed conceptual model is supported by the data. The research sought to determine if sex ratio imbalance affects sexual beliefs and risk taking behavior of Black adolescents and women, aged 11 23, either through direct effects or indirectly (conditional effects) via individual or group perceptions of STI risk and condom use. Black White differences were examined to develop an accurate understanding of what individual and population factors are associated with sexual practices and whether a conditional relationship exists between sex ratio and STI risk. Statistical Analysis The analysis section will begin with descriptive analyses. The purpose of the descr iptive analyses statistical tests is to explore and discover bivariate associations, as well as preliminary detection of multicollinearity. Descriptive analyses included frequencies, percentages, means, and standard deviations to provide a description of the sample. The statistical tests used in the descriptive analyses provided a deeper understanding of the associations between the variables and the data, including: crosstabs with chi square, independent sample t correlations, and M ann Whitney U tests. Full consideration was given to the measurements of the variables and the assumptions of the methods when choosing test statistics. Cross tabulation V tests were conducted to examine t he associations among categorical variables. Independent sample t tests were conducted to compare
41 the means of continuous variables, like sex ratio, across dichotomous groups such as gender and race. Nonparametric Spearman correlations were conducted to examine the associations among ordinal variables, such as perceived chances of getting other STDs, condom use frequency, and these variables more closely resemble o rdinal, rather than continuous variables. Nonparametric Mann Whitney U tests were conducted to test for differences in perceived chances of getting AIDS and perceived chances of getting other STDs between participants who never use a condom and participan ts who reported that they do use a condom. The Mann Whitney U test is a non parametric test used to assess whether two measures from independent samples are statistically different. It is the nonparametric alternative to the independent sample t test, wh ich is utilized in cases on non normally distributed continuous and ordinal variables. Alphas of .05 were used to determine significance levels; although alphas of less than .10 in the primary analyses with interactions were investigated further using st ratified analyses to clarify the complexity of the conditional effects. STATA Version 13 was used for most analyses and SPSS Version 19 was used for the Mann Whitney U analyses. For the multivariate analyses, logistic regression was used to capture the re lationship between the independent variables, control variables, and the dichotomous condom use dependent variable to evaluate research question 1. Logistic regression is a type of analysis that tests for relationships between a single dichotomous depende nt variable and a set of categorical or continuous independent variables. Because logistic regression coefficients do not accurately reveal effect sizes, all coefficient results will be translated into odds ratios for more substantively meaningful interpr etation. For each dependent variable, two models will be specified, including: one without interaction terms and one with interaction terms. Interaction terms are 2 or more
42 independent variables multiplied together into a new variable. This allows resea rchers to determine if there are conditional effects. In other words, direct relationships (no interaction terms) determine if the change in probability of experiencing an outcome, Y, is associated with a direct change in an independent variable X. Conve rsely, conditional relationships (models with interaction terms) determine if the change in probability of Y is associated with a change in X, but that the rate or direction of change in the probability of Y is different for some groups (male versus female ) as opposed to others. Given the large sample size of over 10,000 observations, special attention should be paid to effect sizes. Small effect sizes coupled with statistical significance is considered evidence of observed correlations simply existing as an artifact of a large sample. Interpretation of results under these circumstances will be highlighted in the discussion. The equation for the baseline logistic regression model investigating the direct effects is: ln = 0+ 1+ 2+ 3+ 4+ 5+ 6+ 7 and for the subsequent inclusion of the interaction terms in the logistic regression model investigating the conditional effects, the equation is: ln = 0+ 1+ 2+ 3+ 4+ 5+ 6+ 7+ 8+ 9+ 10+ 11 Ordinal logistic regression was used for analysis on the remaining three ordinal dependent variables for condom use frequency (research question 1 ) perceived STI risk ( research question 2), and perceived AIDS risk (research question 2). Ordinal logistic regression is a special case of logistic regression, appropriate for ordered outcome variables with more than
43 two categories. Each model, for both logistic regression an d ordinal logistic regression, will be initially run only with the independent variables and the control variables (Model 1). Next, the model will be run with all independent variables, control variables, and all necessary interaction terms (Model 2). Th e equation for the baseline ordinal logistic regression model investigating the direct effects is: ln = 0+ 1+ 2+ 3+ 4+ 5+ 6+ 7 an d for the subsequent inclusion of the interaction terms in the ordinal logistic regression model investigating the conditional effects, the equation is: ln = 0+ 1+ 2+ 3+ 4+ 5+ 6+ 7+ 8+ 9+ 10+ 11 Data This study used a non experimental, retrospective co hort research design using quantitative methodologies to examine secondary, cross sectional survey data. De identified data from Wave II of the National Longitudinal Study of Adolescent Health (Add Health) was obtained through a contractual agreement with the University of North Carolina. Add Health is a nationally representative longitudinal survey designed to study adolescent health in the United States. Add Health was designed to study how social environments and behaviors in adolescence are linked to health and achievement outcomes in young adulthood. The Add Health cohort was followed into early adulthood with four in home interviews, the most recent in 2008, when the sample was aged 24 32. The survey sample was a clustered and multistage, stratifi ed sample of adolescents attending schools across the country; specific minority groups
44 were oversampled. Groups oversampled include: Blacks from well educated families, Chinese, Cuban, Puerto Rican, and adolescents residing together. Oversamples were se lected by Wave I collaborators who had research interests in the specific populations. Add Health began with a stratified, random sample of 132 U.S. high schools. Among the 90,000 students attending these schools and selected feeder schools, 2,745 studen ts in 7 th through 12 th grade were identified for Wave I in home interviews (September 1994 December 1995). The total number of participants in Wave I was 20,745. During Wave II (n=14,739), which took place one year later (May 1996 August 1996), all Wave I participants completed follow up in home interviews, with the exception of high school seniors who had since graduated. Due to a lack of relevant data to answer the proposed research questions, only data from Wave II were utilized in this study. Add He alth fundamentally altered or deleted these questions in Waves III and IV. Additionally, Wave IV was not yet available to the public and only contextual data from Wave III was available at the start of this research. Though the intention at the outset of this research was to generalize to the population of 18 to 32 year olds, the sample used here only consists of 11 to 23 year olds. Despite the younger age of this cohort, this data was chosen due to the large sample size and the inclusion of questions re garding sexual practices. Sample Although Add Health is a longitudinal study, unfortunately many of the interview questions were asked in only one or two waves of interviews, making it difficult to examine change over time. For these reasons, the study s ample was not restricted to participants who completed all four waves, resulting in a strictly cross sectional data set, rather than cross sectional time series. Inclusionary criteria consisted of respondents whose reported race or ethnicity was Black or Non Hispanic White, henceforth White, and participated in Wave II.
45 Since Black White was the focal comparison for this study, all other ethnicities/races were excluded from the study. Future research may benefit from investigating comparisons across othe r ethnicities/races. Wave III did not contain the necessary information to construct a sex ratio and the geocodes were not yet available for Wave IV; therefore, it was not possible to include these Waves. As a result, the ultimate number in the sample va ries among different outcomes. Although these restrictions reduced the sample size, every effort was made to use as much information as possible. Wave II encompassed all data collection in 1996. At Wave II respondents lived in 2,096 different census trac ts; however, 3 respondents were removed from the dataset due to having extreme outlier values for sex ratio. The total size of the final sample of Black and White participants was 12,597, but most models had fewer observations due to missing responses or legitimate skips. The questions regarding condom use had a particularly large amount of legitimate skips due to sexual inactivity among a large proportion of younger egitimate skips, imputation was not considered. Imputing this legitimately missing data would likely result in invalid results for the behavioral data, therefore imputation was not used. Measures Dependent Variables Variable Question Responses Condom Us e Have you/has a partner of yours ever used a condom during sexual intercourse? 0 No (skip condom frequency) 1 Yes Condom Use Frequency About what proportion of the time have you/has a partner of yours used a condom? 2 none or some of the time 3 half of the time 4 most of the time 5 all of the time Perceived Chances of Getting AIDS What do you think your chances are of getting AIDS? 1 none 2 very low 3 low 4 high 5 very high Perceived Chances of Getting Other STIs What do you think are your chances of getting another sexually transmitted disease, such a s gonorrhea or genital herpes? 1 no chance 2 very low 3 low 4 high 5 very high
46 Each dependent variable utilized in this study was chosen because it mostly closely related to the research questions proposed in this study. The outcome of interest in research question 1 was sexual risk taking. The variables in the Add Health data that most closely measured this construct were Condom Use and Condom Use Frequency. Condom use was a dichotomo us variable measuring whether (1) or not (0) the respondent had ever used a condom before. The variable Condom Use Frequency was recoded from the original coding of an ordinal variable ranging from 1 (none of the time) to 5 (all of the time). The first tw o categories were very low frequency. The outcome of interest in research question 2 was STI risk perception. The 2 variables in the data that most closely measu red this construct were Perceived Chances of Getting AIDS and Perceived Chances of Getting Other STIs. Both variables were ordinal ranging from 1 (none/no chance) to 5 (very high). Independent Variables Variable Question Responses Gender Interviewer, female. (Ask if necessary) 0 Male 1 Female Race Which one category best describes your racial background? 0 Non Hispanic White 1 Black/African American Sex Ratio Computed using census tract level populations by ge nder (no question for participant); total women divided by total population. n/a Gender and Race were both measured as dichotomous variables, categorizing male female and White African American, respectively. For these variables, male and White are the reference categories for all multivariate analyses. Sex ratio is calculated as the proportion of females in the census tract. Specifically, the formula is the number of females divided by the total population, therefore, sex ratios above 0 .5 represent ce nsus tracts with a disproportionate
47 number of females to males, and sex ratios below 0 .5 represent a disproportionate number of males to females. Control Variables Variable Question Responses Median Household Income Census tract level median household i ncome (no question for participant). $4,999 to $125,053 Age Calculated age from Wave I 11 to 23 Perceived Likelihood of College On a scale of 1 to 5, where 1 is low and 5 is high, how likely is it that you will go to college? 1 Low 2 3 4 5 High Resist S ex Without Birth Control How sure are you that you could resist sexual intercourse if your partner did not want to use some form of birth control? 1 Very Sure 2 Moderately Sure 3 Neither Sure nor Unsure 4 Moderately Unsure 5 Very Unsure 6 I never want to use birth control. Several control variables were chosen from the Add Health data on the basis of proposed conceptual or theoretical relationships with the outcomes of interest in the study, sexual risk taking and perceived chances of getting STIs. Two categorical variables were used as controls for the multivariate analyses. The first variable, Perceived Likelihood of College, was meant to capture empowerment and education level. The second categorical contr ol variable, Resist Sex Without Birth Contro l, was included as a control in the multivariate analyses as a measure of risky sexual behavior and dyadic power Recoded variables included Perceived Likelihood of College and Resist Sex Without Birth Control. Perceived Likelihood of College was collaps ed from the original 5 level ordinal scale into a 4 level scale because there were too few responses for the High (5) category for analysis to be valid; therefore, responses of 4 and 5 were collapsed into a single High category. Similarly, too few partici
48 Very Unsure.
49 CHAPTER IV RESULTS Descriptive Analyses The results chapter follows with the sample description, bivari ate descriptive analyses, and multivariate analyses for research questions 1 and 2 The sample description provides an overview of the characteristics of the sample, the bivariate descriptive analyses provide a rich picture of relationships within the dat a, and the multivariate analyses directly investigate the research questions in this study. As stated in Chapter III, the dependent variables Condom Use and Condom Use Frequency were chosen as the outcomes of interest for research question 1, therefore, t he multivariate analysis section begins with the logistic regression relationships with Condom Use, followed by the ordinal logistic regression relationships with Condom Use Frequency. Next in the multivariate analysis section, research question 2 is inve stigated using ordinal logistic regression to examine Perceived Chances of Getting AIDS and then Perceived Chances of Getting Other STIs. These two variables, as described in Chapter III, are the outcomes of inte rest for research question 2. The sample s ize for the multivariate analyses varies due to legitimate skips and stratification. Specifically, participants who were not sexually active do not have valid responses for condom use frequency and stratified models split the samples by race, therefore, t hey have lower sample sizes. Table 4 .1 shows the frequencies and percentages for categorical variables. Half of the sample was female (51.3%), and the majority of pre teens and teenagers sampled were White (73.7%). The majority of the sample who responde d to the condom use question used a condom (59.3%), and the greatest percentage reported condom use none or some of the time (46.7%). More participants perceived that they had no chance (34.8%) or had a very low chance (33.1%) of getting AIDS. In additio n, more participants perceived that they had no chance (39.7%) or
50 had a very low chance (31.8%) of getting other STIs. Just over half of the participants perceived their chances of attending college as high (55.1%). A majority of participants who answere d the question about resisting sex without birth control felt very sure that they could resist having sexual intercourse (60.0%). Table 4.1 : Frequencies and Percentages for Categorical Variables n Valid % % Gender Male 7181 48.7 48.73 F emale 7555 51.3 51.27 Missing 0 0.00 Total 14736 100.0 Race White 9289 73.7 63.04 Black 3308 26.3 22.45 Missing 1779 12.07 Total 12957 100.0 Ever Use a Condom No 745 40.7 5.06 Yes 1084 59.3 7.36 Missi ng 12907 87.59 Total 1829 100.0 How Often Use a Condom None of some of the Time 505 46.7 3.43 Half of the Time 158 14.6 1.07 Most of the Time 247 22.8 1.68 All of the Time 172 15.9 1.17 Missing 13654 92.66 Total 108 2 100.0 Perceived Chances of Getting AIDS None 5103 34.8 34.63 Very Low 4854 33.1 32.94 Low 3448 23.5 23.40 High 904 6.2 6.13 Very High 349 2.4 2.37 Missing 78 0.53 Total 14658 100.0 Perceived Chances of Getting Ot her STIs No Chance 5813 39.7 39.45 Very Low 4662 31.8 31.64 Low 3102 21.2 21.05 High 819 5.6 5.56 Very High 247 1.7 1.68 Missing 93 0.63 Total 14643 100.0 Likelihood of College Low (1 & 2) 1754 12.0 11.90 3 2 084 14.2 14.14 4 2750 18.8 18.66 High (5) 8072 55.1 54.78 Missing 76 0.52 Total 14660 100.0 Resist Sex Without Birth Control Very Unsure/Never 719 6.0 4.88 Moderately Unsure 545 4.6 3.70 Neither Sure nor Unsure 1774 14.8 12.04 Moderately Sure 2227 18.6 15.11 Very Sure 6699 60.0 45.46 Missing 2772 18.81 Total 11964 100.0 Note: Frequencies not summing to N = 14,736 and percentages not summing to 100 reflect
51 missing data. Rows labeled as missing inc lude legitimate skips. For all variables, the overwhelming majority of missingness is due to legitimate skips. Table 4 .2 displays the frequencies and percentages of the categorical variables separated by race. The sample was similarly split by gender ac ross White and Black participants. Among both Black (8.2%) and White (7.2%) participants, a larger percentage of participants reported using a condom at least once; however, a larger percentage of Black participants (2.2%) reported using a condom all of t he time, compared with White participants (0.9%). Both Black and White participants had decreasing percentages from none or no chance to very high in response to perceived chances of getting AIDS and perceived chances of getting other STIs. Conversely, b oth Black and White participants had increasing percentages of low to high in response to likelihood of attending college and increasing percentages of very unsure/never to very sure in response to resisting sex without birth control. Table 4 .2 : Frequenc ies and Percentages for Categorical Variables by Race ___________________________________________________________________________ __ White Black n % n % Gender Male 4573 49.2 1526 46.1 Female 4716 50.8 1782 53.9 Missing 0 0 0 0 Total 9289 100.0 3308 100.0 Ever Use a Condom No 485 5.2 138 4.2 Yes 666 7.2 271 8.2 Missing 8138 87.6 2899 87.6 Total 9289 100.0 3308 100.0 How Often Use a Condom None or Some of the Time 344 3.7 87 2.6 Half of the Time 85 .9 48 1.5 Most of the Time 155 1.7 62 1.9 All of the Time 80 .9 74 2.2 Missing 8625 87.6 3037 91.8 Total 9289 100.0 3308 100.0 Perceived Chances of Getting AIDS None 3117 33.6 1196 36.2 Very Low 3178 34.2 1004 30.4 Low 2275 24.5 730 22.1 High 514 5.5 241 7.3 Very High 166 1.8 111 3.4 Missing 39 .4 26 .8 Total 9289 100.0 3308 100.0
52 Table 4 .2 : Frequencies and Percentages for C ategorical Variables by Race White Black n % n % Perceived Chances of Getting Other STIs No Chance 3547 38.2 1375 41.6 Very Low 3035 32.7 991 30.0 Low 2039 22.0 642 19.4 High 495 5.3 196 5.9 Very H igh 123 1.3 75 2.3 Missing 50 .5 29 .9 Total 9289 100.0 3308 100.0 Likelihood of College Low (1 & 2) 1212 13.1 315 9.5 3 1233 13.3 494 14.9 4 1708 18.4 629 19.0 High (5) 5098 54.9 1841 55.7 Missing 38 .4 29 .9 Total 9289 100.0 3308 100.0 Resist Sex Without Birth Control Very Unsure/Never 390 4.2 192 5.8 Moderately Unsure 336 3.6 109 3.3 Neither Sure nor Unsure 1082 11.7 311 9.4 Moderately Sure 1484 16.0 361 10.9 Very Sure 4172 45.0 1694 51.2 Missing 1825 19.7 641 19.4 Total 9289 100.0 3308 100.0 ______________________________________________________________________________ Note: Percentages not summing to 100 represent rounding error. Table 4 .3 d isplays the frequencies and percentages of the categorical variables stratified across age groups 11 to 14, 15 to 17, and 18 to 23. A slight majority of the participants were female in the youngest two age groups and a slight majority of the participants were male in the oldest age group. Across all age groups, a larger percentage of participants reported having used a condom at least once and a smaller percentage of participants reported using a condom half of the time. Participants in all age groups re ported none or no chance for perceived chances of getting AIDS and perceived chances of getting other STIs at a higher rate than very high. Participants in all age groups had increasing percentages of low to high in response to likelihood of attending col lege. None of the participants in the youngest age group responded to likelihood of resisting sex without birth control and participants in both older age groups had increasing percentages of very unsure/never to very sure in response to resisting sex wit hout birth control.
53 Table 4 .3: Frequencies and Percentages for Categorical Variables by Age Groups _____________________________________________________________________________________________________________________ Ages 11 14 Ages 15 17 Ages 18 23 n % n % n % Gender Male 1193 45.7 4167 48.3 1821 52.1 Female 1416 54.3 4463 51.7 1676 47.9 Missing 0 0 0 0.0 0 0.0 Total 2609 100.0 8630 100.0 3497 100.0 Ever Use a Co ndom No 38 1.5 410 4.8 297 8.5 Yes 49 1.9 578 6.7 457 13.1 Missing 2522 96.7 7642 88.6 2743 78.4 Total 2609 100.0 8630 100.0 3497 100.0 How Often Use a Condom None or Some of the Time 14 .5 253 2.9 238 6.8 Half of the Time 3 .1 84 1.0 71 2.0 Most of the Time 7 .3 138 1.6 102 2.9 All of the Time 25 1.0 102 1.2 45 1.3 Missing 2560 98.1 8053 93.3 3041 87.0 Total 2609 100.0 8630 100.0 3497 100.0 Perceived C hances of Getting AIDS None 1046 40.1 2958 34.3 1099 31.4 Very Low 823 31.5 2819 32.7 1212 34.7 Low 552 21.2 2037 23.6 859 24.6 High 125 4.8 573 6.6 206 5.9 Very High 42 1.6 202 2.3 105 3.0 Missing 21 .8 41 .5 16 .5 Total 2609 100.0 8630 100.0 3497 100.0 Perceived Chances of Getting Other STIs No Chance 1205 46.2 3357 38.9 1251 35.8 Very Low 774 29.7 2700 31.3 1188 34.0 Low 469 18.0 1854 21.5 779 22.3 High 106 4.1 527 6.1 186 5.3 Very High 28 1.1 151 1.8 68 1.9 Missing 27 1.0 41 .5 25 .7 Total 2609 100.0 8630 100.0 3497 100.0 Likelihood of College Low (1 & 2) 166 6.4 1018 11.8 570 16.3 3 319 12.2 12 62 14.6 503 14.4 4 604 23.2 1696 19.7 450 12.9 High (5) 1511 57.9 4619 53.5 1942 55.6 Missing 9 .3 35 .4 32 .9 Total 2609 100.0 8630 100.0 3497 100.0 Resist Sex Without Birth Control Very Unsure/Never 0 0 .0 499 5.8 220 6.3 Moderately Unsure 0 0.0 379 4.4 166 4.8 Ages 11 14 Ages 15 17 Ages 18 23 n % n % n % Neither Sure nor Unsure 0 0.0 1246 14.4 528 15.1 Moderately Sure 0 0.0 1560 18.0 667 19.1 Very Sure 0 0.0 4832 56.0 1867 53.4 Missing 2609 100.0 114 1.3 49 1.4 Total 2609 100.0 8630 100.0 3497 100.0 ______________________________________________________________________________ Note: Percentages not summing to 100 repres ent rounding error. Next, Table 4.4 shows the mean and standard deviation of the continuous independent variable sex ratio (the ratio of women to men within Census tracts), and two additional continuous control variables: median household income (by Census tract) and the age of the
54 participant. Median household incomes ranged between approximately $5,000 and $125,000 per year, where the national median household income in 1996 was about $35,000. The average age of participants was approximately 16 years o ld. Sex ratios of Census tracts ranged from .27 to .65 with a mean of .51 ( SD = .02). Table 4.4 : Means and Standard Deviations for Continuous Variables. ______________________________________________________________________________ N M SD Min Max S ex Ratio 14568 .51 .02 .27 .65 Median Household Income 14557 30736.17 13213.75 4999 125053 Age 14736 16.22 1.65 11 23 _____________________________________________________________________________________________________________________ Note N not e qual to 14,736 reflects missing data. The following preliminary analyses are intended to describe the complex bivariate relationships among the independent variables, among the dependent variables, and between the independent and dependent variables. All bivariate associations between and among all dependent and independent variables were explored, but only those relationships that were statistically and substantively meaningful were reported here. Tables 3 through 7 display the results of these analyses Though these analyses are not directly related to the research questions, the preliminary analyses do provide insight into the interconnectedness of the predictor and outcome variables. 2 (1) = 9.38, p = .002, V = .027 (not shown). A greater proportion of White participants were male (49.23%) compared to Black participants who were male (46.13%). Independent sample t tests did not reveal a statistically significant difference in mean s ex ratios by gender. Results did, however, show that Black participants ( M = .53, SD = .03) more often lived in Census tracts with higher ratios of women to men than did White participants ( M = .51, SD = .02), t (4276) = 28.61, p < .001. However, the sta tistical significance for this result is likely due to the large sample size
55 and does not necessarily represent a meaningful difference in means. This result should be interpreted with caution. Table 4.5 shows means and standard deviations for chances of getting AIDS and chances of getting other STIs by ever using a condom. Results of nonparametric Mann Whitney U tests revealed a significant difference between condom use groups for perceived chances of getting AIDS, U = 367,706.00, p = .001. Mean ranks f or participants who reported that they use condoms were significantly greater ( MR = 944.47) than were mean ranks for participants who reported that they do not use condoms ( MR = 867.06). Mean ranks are used in place of means for Mann Whitney U tests becau se the variables are closer to ordinal measures, rather than continuous. This means that the test compares the average of the ranks across the two groups. An interpretation of this result is comparable to the interpretation of a t test, where participant s who used a condom typically thought their chances of getting AIDS were higher than the perceptions of participants who did not use a condom. Results also revealed a significant difference between condom use groups for perceived chances of getting other STIs, U = 644,242.00, p = .002. Mean ranks for participants who reported that they use condoms were significantly greater ( MR = 943.66) than were mean ranks for participants who reported that they do not use condoms ( MR = 868.25). Alternatively stated, t he perceived chances of getting other STIs were typically higher among those who had used condoms than those who had not used condoms.
56 Table 4.5 : Means and standard deviations for chances of getting AIDS and chances of getting other STIs by ever use a c ondom ______________________________________________________________________________ n M SD MR U p Perceived Chances of Getting AIDS 367706.00 .001 No 742 2.21 1.03 867.06 Yes 1083 2.37 1.05 944.47 Perceived Chances of Getting Other STIs 644242.00 .002 No 742 2.11 1.00 868.25 Yes 1083 2.26 1.03 943.66 _____________________________________________________________________________________________________________________ As shown in Table 4 .6 nonparametric Mann Whitney U tests were conducted to test for differences among condom use frequency, perceived chances of getting AIDS, and perceived chances of getting other STIs between White and Black participants. Results revealed a significant difference between race groups fo r condom use frequency, U = 68,849, p < .001. Mean ranks for Black participants were significantly greater ( MR = 545.94) than were mean ranks for White participants ( MR = 436.19). Alternatively stated, condom use frequency was typically higher among Blac k participants. Results also revealed a significant difference between race groups for perceived chances of getting other STIs, U = 14,764,183.00, p = .023. Mean ranks for White participants were significantly greater ( MR = 6,300.97) than were mean ranks for Black participants ( MR = 6,142.65). White participants typically perceived their chances of getting other STIs as higher than Black participants. Results revealed no significant difference between race groups for perceived chances of getting AIDS.
57 Table 4 .6 : Means and standard deviations for condom use frequency, perceived chances of getting AIDS, and perceived chances of getting other STIs by race ______________________________________________________________________________________________________ ____________ n M SD MR U p How Often Use a Condom 68849.00 < .001 White 664 1.96 1.11 436.19 Black 271 2.45 1.20 545.94 Perceived Chances of Getting AIDS 15159359.00 .907 White 9250 2.07 .98 6,268.65 Black 3282 2.11 1.08 6,260.44 Perceived Chances of Getting Other STIs 14764183.00 .023 White 9239 1.98 .97 6,300.97 Black 3279 1.96 1.03 6,142.65 _____________________________________________________________________________________________________________ _______ Table 4.7 shows means and standard deviations for condom use frequency, perceived chances of getting AIDS, and perceived chances of getting other STIs by gender. Results revealed a moderately significant difference between gender groups for condo m use frequency, U = 134,281.00, p = .053. Mean ranks for male participants were greater ( MR = 561.34) than were those for female participants ( MR = 526.49). Reported condom use frequency was slightly higher among male compared to female participants. R esults revealed a significant difference between gender groups for perceived chances of getting AIDS, U = 23,078,767.50, p < .001. Mean ranks for male participants were significantly greater ( MR = 7,856.32) than those for female participants ( MR = 6,829.5 8). The perceived chances of getting AIDS were typically higher among males compared to females. Results also revealed a significant difference between gender groups for perceived chances of getting other STIs, U = 22,504,804.00, p < .001. Mean ranks fo r male participants were significantly greater ( MR = 7,922.03) than the mean ranks for female participants ( MR = 6,752.25). The perceived chances of getting other STIs were also typically higher among males compared to females.
58 Table 4.7 : Means and stand ard deviations for condom use frequency, perceived chances of getting AIDS, and perceived chances of getting other STIs by gender ____________________________________________________________________________________________________________________ n M S D Mdn U p How Often Use a Condom 134281.00 .053 Male 466 2.16 1.18 561.34 Female 616 2.02 1.13 526.49 Perceived Chances of Getting AIDS 23078767.50 < .001 Male 7137 2.20 1.02 7,856.32 Female 7521 1.97 1.01 6,829.58 Perce ived Chances of Getting Other STIs 22504804.00 < .001 Male 7132 2.11 .99 7,922.03 Female 7511 1.85 .97 6,752.25 ____________________________________________________________________________________________________________________ Table 4.8 chances of getting AIDS, perceived chances of getting other STIs and sex ratio. Results revealed a significant positive correlation between sex ratio and condom use frequency ( = .0 89, p < .01). This finding suggests that participants who lived in Census tracts with a higher proportion of females compared to males also have higher condom use frequencies. The results also revealed a significant positive correlation between perceived chances of getting AIDS and perceived chances of getting other STIs ( = .729, p < .01). This finding suggests that participants who believe their chances of getting AIDS are high also believe their chances of getting other STIs are high. Table 4.8 of getting AIDS, perceived chances of getting other STIs and sex ratio _____________________________________________________________________________________________________________________ Sex Ratio How Often Use a Condom Perceived Chances of Getting AIDS Perceived Chances of Getting Other STIs Sex Ratio 1.000 How Often Use a Condom .089 (1076) ** 1.000 Perceived Chances of Getting AIDS .008 (14493) .006 (1081) 1.000 Perceived Chances of Getting Other STIs .018 (14477) .055 (10 81) .729 (14629) ** 1.000 _____________________________________________________________________________________________________________________ Note p < .05, ** p < .01. Sample size in parentheses. Table 4.9 shows cross tabulations for gender and r ace by ever using a condom. The relationship betwee 2 (1) = 7.18, p = .007, V = .06; however, this is considered a small effect size. A greater proportion of participants who reported using condoms were female (56.8%) compared to males (43.2%). The
59 re 2 (1) = 8.87, p = .003, V = .075, with a small effect size. A greater proportion of participants who reported using condoms were White (71.1%) compared to Black (28.9%) participant s. Table 4.9 : Cross tabulations for gender and race by ever using a condom ____________________________________________________________________________________________________________________ Ever Use a Condom Yes No n % n % p V Sex 7.18 .007 .063 Male 468 43.2 275 36.9 Female 616 56.8 470 63.1 Race 8.87 .003 .075 White 666 71.1 485 77.9 Black 271 28.9 138 22.2 ________________________________________________________________________ _____________________________________________ Note: Totals may sum to more than 100% due to rounding error. Independent sample t tests were conducted to compare the mean of the sex ratio to ever using a condom (not shown). Independent sample t tests rev ealed no statistically significant difference in mean sex ratio based on condom use. The mean sex ratio of those who used condoms was .51 ( SD = .03) and the mean sex ratio of those who did not use condoms was .51 ( SD = .02), t (1675) = .94, p = .348. Mul tivariate Analyses Research Question 1 : Is there a relationship between sex ratios and sexual risk taking? Does this relationship vary by race and gender? Hypothesis 1 : Sex ratios will be related to condom use. Women who live in high sex ratio areas wi ll be less likely to report using condoms. This relationship will be significant with a larger coefficient for Blacks vs. Whites, specifically Black women because Black women have less negotiating power in their intimate relationships, than Whites at simi lar sex ratios. Tables 4.10 and 4.11 display the results of multivariates models investigating research question 1, hypothesis1, for sexual risk taking using condom use as the dependent variable. Table 8 shows logistic regression analysis relating partici pants ever using a condom to gender, 2 (7) = 22.13, p = .002, pseudo R 2 =
60 .011. Both race, p = .015, and gender, p = .004, were significantly related to ever using a condom. Race had an odds rat io of 1.387, indicating that Black participants had approximately 1.4 times greater odds of using a condom than White participants, and females had approximately 28% lower odds of using a condom than males. Table 4.10 : Logistic Regression analysis relati ng Ever Using a Condom to Gender, Race, and Sex Ratio ____________________________________________________________________________________________________________________ B SE Odds Ratio p Race .327 .134 1.387 [1.07, 1.80] .015 Gender .327 .114 .721 [.58, .90] .004 Sex Ratio .946 2.338 2.576 [.03, 252.01] .686 _____________________________________________________________________________________________________________________ Note 2 (7) = 22.13, p = .002, Pseudo R 2 = .011, N = 1,448. Ref erence category for Race is White and reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported are median household income, likelihood of college, resist sex without birth control, and age. Table 4.11 shows logistic regression analysis relating participants ever using a condom to gender, race, sex ratio, and the following interaction terms: gender and race; gender and sex ratio; race and sex ratio; and gende 2 (11) = 24.16, p = .012, pseudo R 2 = .012. The increase in pseudo R 2 reveals that the model including the interaction terms performs better than the baseline model at predicting the likelihood of having used a condom. However, none of the independent variables were statistically significant. Race and gender, which were significant predictors in the baseline model, were no longer significant. The insignificant findings for the interaction ter ms suggest that there are no conditional effects of gender, race, and sex ratio on condom use, therefore, the findings are inconsistent with hypothesis a in research question 1. In other terms, the effects of gender and race on the outcome of ever using a condom are only direct relationships, which do not vary dependent upon one another or sex ratios.
61 Table 4.11 : Logistic Regression analysis relating Ever Using a Condom to Gender, Race, Sex Ratio, Gender*Race, Gender*Sex Ratio, Race*Sex Ratio, and Gender *Race*Sex Ratio ___________________________________________________________________________________________________________________ B SE Odds Ratio p Race 2.650 4.662 14.156 [<.01, 131687] .570 Gender 2.135 3.214 8.453 [.02, 4603.36] .507 Sex Rati o 4.076 4.975 58.938 [<.01, 1011917] .413 Race*Gender 3.094 5.494 .045 [<.01, 2153.53] .573 Race*Sex Ratio 4.139 8.932 .016 [<.01, 638604] .643 Gender*Sex Ratio 4.680 6.296 .009 [<.01, 2,122.64] .457 Race*Gender*Sex Ratio 5.423 10.535 226.578 [< .01, >210000000000] .607 ____________________________________________________________________________________________________________________ Note 2 (11) = 24.16, p = .012, Pseudo R 2 = .012, N = 1,448. Reference category for Race is White and reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported a re median household income, likelihood of college, resist sex without birth control, and age. Research Question 1 : Is there a relationship between sex ratios and sexual risk taking? Does this relationship vary by race and gender? Hypothesis 2 : Men who live in low sex ratio areas will report more frequent use of condoms. This relationship will be significant with a larger coefficient for Blacks vs. Whites, specifically Black women because Black women have less negotiating power in their intimate relatio nships than Whites at similar sex ratios. Tables 4.12 and 4.13 display the results of multivariate models investigating research question 1, hypothesis 2, for sexual risk taking using condom use frequency as the dependent variable. Table 4.12 shows an ord inal l ogistic regression analysis relating condom use frequency 2 (7) = 63.92, p < .001, pseudo R 2 = .029. Of the main independent variables, race was highly significant, p < .001, and had an odds ratio of 1.993. This finding indicates that Black participants had nearly 2 times higher odds of more frequent condom use than White participants. Gender was also significant, p = .017, and had an odds ratio of .722. This finding indicates that th e odds of using condoms more frequently was approximately 28% lower for females than were the odds for males.
62 Table 4.12 : Ordinal Logistic Regression analysis relating Condom Use Frequency to Gender, Race, and Sex Ratio ________________________________ ____________________________________________________________________________________ B SE Odds Ratio p Race .689 .148 1.993 [1.50, 2.67] < .001 Gender .3.26 .137 .722 [.55, .94] .017 Sex Ratio .809 2.681 2.248 [.01, 430.10] .736 _______________ ______________________________________________________________________________________________________ Note 2 (7) = 63.92, p < .001, Pseudo R 2 = .029, N = 870. Reference category for Race is White and reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported are median household income, likelihood of college, resist sex without birth control, and age. Table 4.13 shows ordinal logistic regression analysis relating condom use frequency to gender, race, sex ratio, and the following interaction terms: gender and rac e; gender and sex ratio; race and sex ra 2 (11) = 65.28, p < .001, pseudo R 2 = .030. The increase in pseudo R 2 reveals that the model including the interaction terms performs better than the baseline model at predicting c ondom use frequency; however, as with the previous models predicting condom use, the increase in pseudo R 2 is quite small, therefore, it should not be considered a substantively meaningful increase in model fit. Similar to the results in Table 10, none of the key independent variables or interaction terms were statistically significant; therefore, the findings are inconsistent with research question1, hypothesis 2. Table 4.13 : Ordinal Logistic Regression analysis relating Condom Use Frequency to Gender, Ra ce, Sex Ratio, Gender*Race, Gender*Sex Ratio, Race*Sex Ratio, and Gender*Race*Sex Ratio ____________________________________________________________________________________________________________________ B SE Odds Ratio p Race .474 4.726 .622 [<.01 6558.27] .920 Gender 3.229 3.939 .040 [<.01, 89.15] .412 Sex Ratio 1.017 5.932 .362 [<.01, 40549.47] .864 Race*Gender 4.295 5.863 73.307 [<.01, 7176093] .464 Race*Sex Ratio 2.073 9.082 7.945 [<.01, > 427000000] .819
63 Table 4.13 : Ordina l Logistic Regression analysis relating Condom Use Frequency to Gender, Race, Sex Ratio, Gender*Race, Gender*Sex Ratio, Race*Sex Ratio, and Gender*Race*Sex Ratio ______________________________________________________________________________________________ ______________________ B SE Odds Ratio p Gender*Sex Ratio 5.570 7.709 262.230 [<.01, > 956000000] .470 Race*Gender*Sex Ratio 7.971 11.275 < .001 [<.01, 1366429] .480 _______________________________________________________________________________ ______________________________________ Note 2 (11) = 65.28, p < .001, Pseudo R 2 = .030, N= 870. Reference category for Race is White and reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported are median household income, likelihood of college, resist sex without birth control, and age. Research Question 2 : Is there a relationship between sex ratios and STI risk perception? Does this relationship vary by race and gender? Hypothesis 1 : Hypothesis 1 Sex ratios will be related to AIDS and other STI risk perception. Men who live in low sex ratio areas and women who live in high sex ratio areas will report less AIDS and other STI risk perception. This rel ationship will be significant with a larger coefficient for Blacks vs. Whites, specifically Black women because the availability of partners sex ratios. Tables 4 .14 through 4.17 display the results of multivariate tests for research question 2, hypothesis 1, using perceived chances of getting AIDS as the dependent variable. Table 4.14 shows ordinal logistic regression analysis relating the perceived chances of ge tting AIDS to 2 (7) = 309.90, p < .001, pseudo R 2 = .012. Race was significant, p = .024 and had an odds ratio of 1.106. This finding indicates that Black vs. White participants had about an 11.0% higher odds of perceiving their chances of getting AIDS as high. Of all the key independent variables, gender was highly significant, p < .001, and had an odds ratio of .715. This finding indicates that females had about 28.0% lower odds of hav ing higher perceived chances of getting AIDS than males.
64 Table 4.14 : Ordinal Logistic Regression analysis relating Chances of Getting AIDS to Gender, Race, and Sex Ratio __________________________________________________________________________________ __________________________________ B SE Odds Ratio p Race .100 .044 1.106 [1.01, 1.21] .024 Gender .335 .038 .715 [.66, .77] < .001 Sex Ratio .627 .803 1.872 [.39, 9.03] .435 ____________________________________________________________________ _________________________________________________ Note 2 (7) = 309.90, p < .001, Pseudo R 2 = .012, N = 9,962. Reference category for Race is White and reference category for Gender is Male. Sex ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported a re median household income, likelihood of college, resist sex without birth control, and age. Table 4.15 shows ordinal logistic regression analysis relating the perceived chances of getting AIDS to gender, race, sex ratio, and the following interaction te rms: gender and race; gender a nd sex ratio; race and sex ratio; and gender, race, and sex ratio. Overall, the model was 2 (11) = 316.37, p < .001, pseudo R 2 = .012. The pseudo R 2 did not increase after including the interaction terms for predicting perceived chances of getting AIDS; therefore, there was no increase in overall model fit. More importantly, the 3 way interaction term and two other interaction terms, race by gender and race by sex ratio, revealed significant relationships with perceived chances of getting A IDS. Table 4.15 : Ordinal Logistic Regression relating Chances of Getting AIDS to Gender, Race, Sex Ratio, Gender*Race, Gender*Sex Ratio, Race*Sex Ratio, and Gender*Race*Sex Ratio ___________________________________________________________________________ ___ B SE Odds Ratio p Race 2.347 1.207 .096 [.01, 1.02] .052 Gender 1.589 1.102 .204 [.02, 1.77] .149 Sex Ratio 1.026 1.536 .358 [.02, 7.28] .504 Race*Gender 3.735 1.659 41.904 [1.62, 1082.93] .024 Race*Sex Ratio 4.767 2.321 117.605 [1.24 11124.73] .040 Gender*Sex Ratio 2.487 2.145 12.030 [.18, 820.97] .248 Race*Gender*Sex Ratio 7.297 3.190 < .001 [<.01, .35] .022 ____________________________________________________________________________________________________________________ N ote 2 (11) = 316.37, p < .001, Pseudo R 2 = .012, N = 9,962. Reference category for Race is White and reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not report ed are median household income, likelihood of college, resist sex without birth control, and age.
65 To further investigate these complex interactive relationships and being consistent with research question 2, hypothesis 1, the data were stratified by race, in which all Black participants were pooled into one subset of data and all White participants were pooled into another subset of data. Tables 14 and 15 show models run on each of these subsets separately to achieve a more lucid interpretation of the comp lex 3 way interaction. 3 Table 4.16 shows a stratified ordinal logistic regression analysis for the subset sample of White participants relating perceived chances of getting AIDS to gender, sex ratio, and the interaction term gender by sex ratio (see Table 14). 2 (7) = 250, p < .001, pseudo R 2 = .013. However, none of the key predictor variables were significant. Table 4.16 : Stratified Ordinal Logistic Regression analysis relating Perceived Chances of Getting AIDS to Gender Sex Ratio, Gender*Sex Ratio (for the White Sample) ___________________________________________________________________________________________________________________ B SE Odds Ratio p Gender 1.640 1.110 .194 [.02, 1.71] .139 Sex Ratio 1.068 1. 549 .344 [.02, 7.16] .490 Gender*Sex Ratio 2.590 2.171 13.335 [.19, 938.90] .233 ____________________________________________________________________________________________________________________ Note 2 (7) = 250.76, p < .001, Pseudo R 2 = .013, N = 7,348. Reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported are median household income, likelihood of college, resist sex without birth control, and age. Table 4.17 shows a stratified ordinal logistic regression analysis for the subset sample of Black participants relating perceived chances of getting AIDS to gender, sex ratio, and the interaction term g ender by sex ratio. Overall, the model w 2 (7) = 81.82, p < .001, pseudo R 2 = .011. Now, the interaction term for gender and sex ratio is marginally significant, p = .051, and has an odds ratio of .011. Also, sex ratio is marginally significant, p = .056, with an odds ratio of 28.366. The interaction term results indicate that among Black participants, in a 3 It is important to be mindful of the fact that when performing stratified analysis the researcher is essentially interacting every variable in the model (even control variables) with the categorical variable upon which the data are being stratified. In this case, it is race.
66 hypothetical Census tract area composed entirely of females, females have 98.9% lower odds than males of believing their chances of getting AIDS are high. This is consisten t with research question 2, hypothesis 1, at least for the specific STI, AIDS. Among Black participants, as the sex ratio increases their odds of thinking their chances of getting AIDS are high increases by about 28 times. Table 4.17 : Stratified Ordinal Logistic Regression analysis relating Perceived Chances of Getting AIDS to Gender, Sex Ratio, Gender*Sex Ratio (for the Black Sample) ___________________________________________________________________________________________________________________ B S E Odds Ratio p Gender 1.981 1.219 7.253 [.67, 79.02] .104 Sex Ratio 3.345 1.748 28.366 [.92, 873.15] .056 Gender*Sex Ratio 4.504 2.309 .011 [<.01, 1.02] .051 _______________________________________________________________________________________ ______________________________ Note 2 (7) = 81.82, p < .001, Pseudo R 2 = .011, N = 2,614. Reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported are medi an household income, likelihood of college, resist sex without birth control, and age. Research Question 2 : Is there a relationship between sex ratios and STI risk perception? Does this relationship vary by race and gender? Hypothesis 1 : Hypothesis 1 Sex ratios will be related to AIDS and other STI risk perception. Men who live in low sex ratio areas and women who live in high sex ratio areas will report less AIDS and other STI risk perception. This relationship will be significant with a larger coe fficient for Blacks vs. Whites, specifically Black women because the availability of partners sex ratios. Tables 4.18 and 4.19 display the results of additional multivariate tests for research question 2 using perceived chances of getting other STIs as the dependent variable, rather than perceived chances of getting AIDS. Table 4.18 shows ordinal logistic regression analysis relating perceived chances of getting other STIs to gender, race, and sex ratio. Overall, the
67 2 (7) = 431.23, p < .001, pseudo R 2 = .017. Of all the independent variables, gender was highly significant, p < .001, and had an odds ratio of .645. These results indicate that females vs. males have about 35% lower odds of perceiving their own chances of getting other STIs as high. Table 4.18 : Ordinal Logistic Regression analysis relating Perceived Chances of Getting Other STIs to Gender, Race, and Sex Ratio _____________ ________________________________________________________________________________________________________ B SE Odds Ratio p Race .022 .045 1.022 [.94, 1.12] .623 Gender .438 .038 .645 [.60, .70] < .001 Sex Ratio .105 .807 1.110 [.23, 5.40] .897 ____________________________________________________________________________________________________________________ Note 2 (7) = 431.23, p < .001, Pseudo R 2 = .017, N = 9,963. Reference category for Race is White and reference category for Gender is M ale. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported are median household income, likelihood of college, resist sex without birth control, and age. Table 4.19 shows ordinal logistic regre ssion analysis relating perceived chances of getting other STIs to gender, race, sex ratio, and the following interaction terms: gender and race; gender and sex ratio; race and sex ratio; and gender, race and sex ratio. Overall, the model was significant 2 (11) = 438.97, p < .001, pseudo R 2 = .017, though there was no increase in pseudo R 2 to change the model fit However, similar to previously mentioned results, none of the independent variables were significant. Therefore, the findings are inconsiste nt with research question 1, hypothesis 1.
68 Table 4.19 : Ordinal Logistic Regression analysis relating Perceived Chances of Getting Other STIs to Gender, Race, Sex Ratio, Gender*Race, Gender*Sex Ratio, Race*Sex Ratio, and Gender*Race*Sex Ratio ___________ __________________________________________________________________________________________________________ B SE Odds Ratio p Race .092 1.217 1.096 [.10, 11.90] .940 Gender .929 1.111 .395 [04, 3.49] .403 Sex Ratio .617 1.549 1.853 [.09, 38.57] 691 Race*Gender 2.031 1.666 7.620 [.29, 199.62] .223 Race*Sex Ratio .008 2.340 1.008 [.01, 98.82] .997 Gender*Sex Ratio 1.043 2.174 2.838 [.04, 200.99] .631 Race*Gender*Sex Ratio 4.198 3.204 .015 [<.01, 8.02] .190 _______________________________ ______________________________________________________________________________________ Note 2 (11) = 438.97, p < .001, Pseudo R 2 = .017, N = 9,963. Reference category for Race is White and reference category for Gender is Male. Sex Ratio is measured as the proportion of Women to Men. Control variables included in the model but not reported are median household income, likelihood of college, resist sex without birth control, and age. Summary This chapter included descriptive statistics, as preliminary analyses, and multivariate statistics, as primary analyses, for the data in this study. I n the investigation of the research questions, the interactive term of gender, race, and sex ratio was only statistically significant in the ordinal logistic regression model predicting perceived chances of getting AIDS, which provides insight into Researc h Question 2. Stratified analyses by race for this model revealed that among Black participants, being female in a Census tract community with a higher ratio of as high; however, this relationship was not significant among White participants. These results conditional on the sex ratio of the community; evidence of this effect only exists within the sub chances of contracting AIDS is conditional on gender among Black participants. These conditional effects for gende r, race, and sex ratio were not observed for the other measures of the
69 dependent variables: whether the participant ever uses a condom, condom use frequency, or f contracting AIDS were subject to these conditional effects. However, both research questions contain a direct and conditional effect component. were direct, not conditional. Results for research question 1 revealed that Black participants were more likely than White participants to report having used a condom. Black participants were also more likely than White participants to report using a condom more freq uently. Men were more likely than women to report using a condom more frequently. Regarding research question 2, men were more likely than were women to report that they believed their chances of getting AIDS or other STIs were high. Finally, participan ts in Census tract communities with higher female to male ratios were less likely to believe their chances of getting AIDS was high. A deeper discussion of these findings will be the focus of Chapter V the Discussion
70 CHAPTER V DISCUSSION This study sou ght to analyze the direct and conditional effects of sex ratio, gender, and race on sexual risk taking and STI risk perception. Specifically, the study evaluated the d practices among Blacks and Whites. The goals of the research were to: 1) investigate the interactive effects of race, gender, and sex ratio related to STI risk; 2) contribute significantly to the small but growing body of literature on imbalanced sex ra tios related to STI risk; 3) position the effects of variations in the sex ratio within larger discussions of STIs; and 4) contribute to prevention and intervention efforts that address the transmission of STIs among Blacks. In this chapter, I will presen t an interpretation of my findings as well as provide the context of the findings. After addressing the research questions driving this inquiry, I will discuss the implications and potential ideas for future research. Purpose of Research While adolescent sexual behaviors are not necessarily solidified, experiences during this period of life often shape adulthood. Understanding young people has indications for interpreting important life events that influence subsequent sexual behavior as adults. For exa mple, Simpson (2007) found that preferences in adult romantic relationships are not just relative to present environmental conditions. Actions as adults are a direct result of past relationships and personal attachment extending through adolescence all th e way back to childhood (Simpson, 2007). Adolescence is a high risk period regarding sexual behavior. Thus, sexual beliefs and practices that carry on into adult hood. I believe early intimacy experiences
71 gradually give rise to a system of thoughts, memories, beliefs, expectations, emotions and adult behaviors. function of d emographic opportunity, with opportunity determined by the number of available opposite sex members with whom such relationships can be formed. Previous researchers in this area have evaluated this model in relation to marriage prevalence and mate quality among a variety of populations with significant findings (South & Trent 2010; Uecker & Regnerus, 2010; Albrecht and Albrecht, 2001). Given that marriage rates are only one of a number of outcomes impacted by sex ratio imbalance, it seems relevant to bro aden the scope of sex ratio theory to include other behaviors, such as sexual beliefs and risk taking practices. The current study sought to expand on the evaluation of sex ratio by considering a race and gender neutral model and a race and gender specifi c model. Summary of Findings I predicted the interaction between the sex ratio and beliefs and practices would play an important role in explaining the differences in sexual behavior and perceived STI risk across Blacks and Whites. I expected that sex rat io influences STI risk perceptions and sexual risk taking, but that the effects would have a greater impact on Blacks, particularly Black women. However, a comparison of Blacks and Whites revealed that Blacks used condoms more frequently than Whites. In addition, Blacks were significantly more likely to consider themselves at risk for AIDS in comparison to Whites, yet females reported lower levels of perceived STI risk in comparison to males. The control variables were median household income, age, likel ihood of college, and resist sex without birth control. Though median income and likelihood of college were often significant covariates in the model, their effect sizes were
72 extremely low, which indicates that the significant relationship was likely due to a large sample, rather than a true statistical relationship. The overall results suggest that although Black adults are at very high risk for STIs, racial differences in condom use, condom use frequency, and perceived STI risk could not be explained by racial differences in the availability of alternative partners. The conceptual model hypothesizing the relationship between sex ratio and STI risk perceptions and behaviors was significant, though sensitive to model specification. For most of the model sp ecifications the effects of race, gender, and sex ratio were not dependent upon one another. For perceived chances of AIDS, there was some evidence of dependent relationships found. Therefore, findings from the current research do not support the race an d gender specific model nor the race and gender specific model. The initial hypothesis (research question 1, hypothesis 1a) in this study sought to examine the extent to which the principles of sex ratio theory were supported in relation to condom use am ong Black women. Black versus White participants were more likely to report using a condom. Blacks were more likely to have used a condom at least once, and were also more likely to use them more frequently. These findings corroborated the fact that Bla cks are aware of their risk of contracting AIDS and are therefore taking active roles in an effort to prevent transmission. However, females versus males were less likely to report using a condom before, and use them less frequently than males. This sugg ests that men take on a higher burden of responsibility when it comes to condom use. The following speculations were made as to why findings did not support the first ed by social desirability bias. In other words, Blacks may have been more likely to report socially desirable
73 behaviors. Due to perceived negative stereotypes Blacks may be more acutely aware of social norms and therefore try to portray themselves in ali gnment with what they believe is socially acceptable. Stereotypically, it is perceived that Blacks participate in more risky sexual activity than Whites; however, research has established contrary findings (Khan et al., ., 2011; Adimora et al., ., 2008; Foreman 2003). Second, contradictory to the literature (Foreman, 2003; Cummings et al., ., 1999), Blacks may be increasingly more aware of the significantly higher rate of STIs among their race and therefore take greater precautions, such as the use of condoms. T hird, another explanation to findings for hypothesis 1a, could be that condoms are designed for men and seen somewhat as the social responsibility of men, therefore, women may not take it upon themselves to provide condoms and therefore are less likely to report using condoms. In addition to whether participants ever used a condom, their frequency of condom use was investigated in research question 1, hypothesis 1b. In the baseline model (without interaction terms) for condom use frequency, race and gend er were significant. Blacks were twice as likely as White participants to report higher frequencies of condom use. Women were less likely than men to report frequently using condoms. Sex ratio was not significant. Race, gender, and sex ratio were not s ignificant in the fully specified ordinal logistic regression model with interaction terms. The interaction terms also were not significant, which means that though there were direct effects of race and gender, they were not conditional on each other or s ex ratio. These findings suggest varying evidence of the relationship between race and gender on risky behavior but no evidence of significant conditional or unconditional sex ratio effects of risky behavior. It seems that sex ratio alone is not a predic tor of risky sexual behavior. Condoms are used out of necessity not out of desirability. Therefore, as intimacy becomes more frequent with a partner, the use of condoms may diminish significantly for a
74 variety of reasons. Individuals may use a condom the first time they are intimate with a partner, but in an attempt to build a stronger connection within the relationship, over time the requirement for condom use diminishes. For example, Cummings et al., (1999) found that having unprotected sex helps wo men maintain their beliefs that their partners are fa ithful. Another possibility is if a condom is not used during the first intimate encounter without negative consequences, then it may be falsely assumed that the need to use a condom as a preventative m easure is unnecessary. The above factors are not exclusive to race, gender, or sex ratio. The instinctive desire to achieve a more pleasurable and intimate experience can outweigh the rational and long term need to regularly use condoms. Consistent wit h the premise of sex ratio theory, research question 2, hypothesis 1a, sought to explore the relationship between sex ratio and risk perception. Gender was significant in the baseline ordinal logistic regression model without interaction terms for perceiv ed chances of getting AIDS. Females were less likely to believe their chances of getting AIDS was high versus males. In all the other unconditional models it is only gender and race that are occasionally significant. The findings revealed that race had an impact on whether or not people believed their chances of getting AIDS was high. Black participants were more likely to report believing their chances of getting AIDS was high versus White participants. Sex ratio was the only variable that was not sig nificant. Evidence of conditional effects were discovered for perceived chances of getting AIDS and gender, race, and sex ratio. The data were split by race to simplify the interpretation of the findings for perceived chances of getting AIDS. The model for White participants revealed no significant conditional relationships, though the model run on the sample of Black participants did reveal some conditional relationships. Sex ratio and the gender sex ratio interaction were
75 marginally significant. For the Black sample, females in census tract neighborhoods with an increasingly high proportion of females to males were less likely than males to believe their chances of getting AIDS was high. Among the Black female population, as the proportion of females to males increases, participants were less likely to think their chances of getting AIDS was high. Among the Black male population, as the proportion of females to males increases, participants were more likely to believe their chances of getting AIDS wa s high. A baseline ordinal logistic regression model, without interaction terms, was used to further analyze the second hypothesis in order to predict perceived chances of getting other STIs. There were no conditional relationships between other STI ris k perceptions and race, gender, or sex ratio. Females compared to males, are less likely to believe their chances of getting other STIs is high. Race, gender, and sex ratio were not significant in the ordinal logistic regression model with interaction te rms. Interpretation of Findings Guttentag and Secord (1983) claimed that the effects of variations in the sex ratio are independent of socioeconomic and cultural factors, and therefore predictions should hold true for Whites with similar ratios (race and gender neutral model). Supporting this claim, almost all of my models showed no significant racial differences in sex ratio in relation to sexual beliefs and practices. Only on model showed racial differences in sex ratio (race*sex ratio interaction) in relation to sexual beliefs and practices, particularly for perceived chances of getting AIDS. While sex ratio does matter, the influence is conditional upon race and gender. Specifically, sex ratio has a greater influence on Blacks, and Black females in particular. For example, Black women were less likely to think their chances of getting AIDS was high, while Black men seem to be acutely aware of their chances of contracting AIDS. This could be because Black men are
76 aware of the sex ratio imbalance and as a result, participate in concurrent relationships. On the other hand, Black women may be aware of the lack of Black men, however, may not initially recognize they are in concurrent relationships. Moreover, women may be unaware of their risk due to mi sleading perceptions. The literature suggests Black women have trouble conceptualizing themselves at risk, although they can clearly identify the risks associated with sex (Robinson et al., ., 2005; Foreman, 2003; Cornelius Okundaye and Manning, 2000). Sp mates in comparison to men; however, they remain at risk due to choices their partners make regarding concurrent relationships. Because Black women are less likely to date outside their race (Banks, 2011; Johnson & Raphael, 2006; Dunn, 2005; McNair & Prather, 2004) they should be more aware of the significant role the limited partner pool of Black men plays into their STI risk. The above factors could be explanations o f the conditional influence of sex ratio. Findings from this research highlight the validity of prior research into the cause of STIs that primarily focused on individual level risk factors such as number of partners, behaviors, and condom use (McQuillan et al., ., 2006; Dunn, 2005; McNair & Prather, 2004). Research efforts have been typically guided by the premise that the best way to prevent STIs is by targeting risk behaviors at the individual level. While the literature seemed to suggest that this strat egy may be appropriate for Whites but not for Blacks (because their STI risk increases even when their behavior is not risky), the current findings, derived from a large, nationally representative sample of young adults, indicated otherwise. Though individ ual level risk factors alone may not be adequate to explain the racial disparity in STIs, the lack of significant findings in this study, calls into question the impact of population level risk factors. Further analysis into how individual
77 level and popul ation level variables interact in shaping health outcomes could elucidate the marked difference in STI rates among Blacks and Whites. Context of Findings & 3 ) sex ratio theory are outdated. For example, the principal foundation of sex ratio theory, demographic opportunity, only takes into account the number of available opposite sex members with whom partnership can be formed. In regard to Black women, the literature clearly indicates that Black women typically limit themselves to same race/ethnic partnerships (Banks, 2011; Johnson & Raphael, 2006; Dunn, 2005; McNair & Prather, 2004), and therefore are primarily affected by race specific sex ratio. However, this premise does not take into a ccount same sex partnerships, bisexuality, or the affect of gender specific sex ratio. Moreover, according to sex ratio theory, mate preferences are somewhat shaped by financial support. However, the means and ways to acquire financial support have chang be more reliance on government assistance as well as child support, neither of which necessitate having a partner in the traditional sense. Another primary consideration of sex ratio theory is ho w imbalanced sex ratios interact with gender inequality. While Guttentag and Secord argue that structural power resides current study suggest that sex ratio imbalance does not limit the possession of dyadic power Guttentag & Secord 1983) have diminished. In the last 2 3 decades there has been an increase in the number of women in the labor force, a growing number of women serving as chief executive officers (CEOs), and a
78 structural power today, could change the fundamental argument made by Guttentag and S ecord. Measuring individual agency or relationship power was beyond the scope of this research, however, the closest measure I had to examine these issues was the resist sex without birth control variable which served as my proxy for dyadic power For e xample, the fact that 51.2% of Blacks reported being very sure they would be able to resist sex without birth control, illustrates that in this case, members of the gender in short supply may not have an advantage over members of the other gender. Anothe r important consideration is the fact that s exual attitudes and beliefs have changed since the Add Health data was collected in 1996. Unfortunately the age of this data does not address the current environment concerning prominent issues such as the soci al responsibility of condom use. Along the same lines, during this time period it is possible that HIV was still viewed as primarily a disease transmitted by and among homosexual males, therefore heterosexuals may not have considered themselves at risk. If there is a temporal/historical effect, Implications The implication is that factors in addition to or other than sex ratio, such as high infection among potential partners, are influencing high infection rates among Black women. Racial disparities in STIs probably reflect environmental, institutional, and contextual differences between Blacks and Whites. For example, sexual mating patterns are largely segregated according to race (Khan et al., ., 2011), and Blacks are more likely than Whites to cross high and low risk behavior groupings when choosing sexual partners (Khan et al., ., 2011; Barrow et al., ., 2008; Aral et al., ., 2008; Doherty et al., ., 2005; Laumann & Youm, 1999). This combination of ass ortative (like with like) mixing by race and disassortative mixing by risk group may create a
79 important determinant of STI acquisition is the infection status high and low risk groups are more likely to encounter an infected partner given this combination of assortative and disassortative mixing. Furthermore, it may be necessary for future studies to control for prior STIs amon sexual behaviors and perceptions of risk. Given the consistent evidence that sexual mating patterns are segregated according to race, I suggest policy strategies that will reduce infection in the burdened population. Most STIs can be cured, and the health and life span of individuals infected with HIV can be greatly increased by current therapies. In addition, the spread of HIV may be reduced through prompt diagnosis and treatment, given that antiretroviral treatment reduces viral load and thus, risk of transmission ( Quinn Wawer, Sewankambo Serwadda Li Wabwire Mangen Meehan Lutalo & Gray 2000). A good policy strategy to prevent STIs might be to provide more funding to community based clinics where young people might be more likely to go for screenings, treat ment, and prevention strategies. Many STIs increase the likelihood of both infectivity of and susceptibility to HIV infection, so reducing prevalence will also help to prev ent the spread of HIV (CDC, 2006 ). The burden of HIV morbidity and mortality is hi gh. There are existing tools to greatly lower disease rates, such as prevention programs aimed at reducing sexual health risk behaviors, and appropriate efforts should be directed toward reaching this important public health goal. Addressing racial dispa rities in STI prevalence is extremely sensitive, and policy makers must proceed carefully, with dialogue and consensus among all community groups. Nevertheless, it is critical that the problem be addressed with due speed and guided by scientific
80 findings. Given the documented higher prevalence of STIs among Black than White young adults, a more aggressive strategy to reach all Black young adults is needed, such as offering free universal screening to this population. I recommend a media campaign informin g Blacks of the high prevalence of HIV and other STIs across behavior groups, reducing the stigma associated with testing and an STI diagnosis, and encouraging young adults to be tested on an annual basis. I also recommend that information, testing, and t reatment services be offered through nontraditional venues such as churches and beauty salons, as well as through colleges, prison and jail facilities, and health care venues. In addition, I recommend that surveillance be conducted to monitor whether prev alence is reduced over time and whether subgroups (such as men) are routinely being tested. Finally, I recommend further research into structural factors that may account for some of the racial STI disparity. Strengths and Limitations The relative numbe rs of women and men in a population are likely to shape opportunities to engage in various forms of sexual behavior, but studies of the sexual risk behaviors that contribute to STIs tend to overlook the sex ratio as a cause factor. This research examines this connection using a national longitudinal data set, which oversampled Blacks. The case of Blacks in the U.S is well suited to such an investigation, both because the U.S. has been experiencing a growing imbalance between the numbers of Black men and w omen in its population and because there is widespread speculation that the increasing deficit of males has profound implications for the spread of STIs (Khan et al., ., 2011; Adimora et al., 2002; Manhart et al., ., 2002). The major strength of this project is i ts timeliness and relevance to the persistent racial disparities in rates of STIs in the U.S. Public health practitioners and researchers have devoted increasing attention to the role of sexual networks and partnership concurrency in STI
81 epidemiology, yet there are few studies that address the nature and root of socio cultural factors that help shape STI risk (Aral et al., ., 2008; Newman et al., ., 2008; Adimora et al., 2002). Beginning to uncover the nature of these constructs is a contribution to the literatu re, as well as STI prevention strategies. The vast bulk of prior research on sexual risk taking and STIs focuses on their proximate, individual level determinants. My analysis directs attention to salient factors farther up the causal chain, underscorin g how more distal characteristics of the social environment also influence these behaviors. In this way, my findings help to inform the macro micro linkage in the study of individual health behavior and health status. While these outcomes certainly lend directions for future research, there are limitations to the current study. Several limitations may affect the strength and external validity of certain findings within this study. First, Add Health data is self reported, an information gathering techniqu e susceptible to inaccurate self monitoring as well as the reporting of socially desirable behavior. Sexual risk behaviors were measured via self reports and could have been underestimated. Questions about sexual risk behaviors focus on very specific and potentially sensitive types of sexual activities, therefore the possibility exists that some participants who may be less comfortable discussing these sorts of issues may underreport instances of risky sex. For instance, valid measurement of condom use h as posed challenges to researchers generally, given that study participants may not recall condom use behavior accurately or may be influenced by social desirability issues (Zenilman, Weisman, and Rompalo, 1995). However, computer assisted self interviewi ng technology was used in Add Health; this methodology has been found to increase reporting of sensitive behaviors, and there are indications that it is effective across racial/ethnic groups (Turner, Ku, Rogers, Lindberg, Pleck, & Sonenstein. 1998).
82 A sec ond limitation is the limited age of the sample population, given the lack of relevant data in the later waves. While the overall sample may be sufficient to yield main effects, a larger age continuum would provide greater variance on measures assessing m oderation effects and be more representative of the general population. The inclusion of data from the younger end of the sample of 11 and 12 year olds in the Add Health study does not address factors of actual experience or maturity of these preteens. T he possibility exists that these preteens were more nave concerning sexual practices from their own experiences and relationships, which contributes to the ambiguity of the data. The inclusion of pre teens suggests that some participants who took part in the study may not have had the knowledge to evaluate sexual relationships due to lack of maturity. The exclusion of 24 to 32 year olds from the sample affected the inferences derived from this study; therefore, a data set with an older age range would st rengthen findings. Moreover, while Add Health is a longitudinal data set, I was unable to look at causal relationships. Future research would benefit from a true longitudinal analysis of an older age range. Last, economic adversity (e.g. high rates of un employment) and racial segregation may well play a role in the association of sex ratios and the outcomes of this study, but unfortunately current work in the field provides insufficient guidance to develop specific hypothesis about magnitude or direction of effects. Although it is possible to examine these issues in another dataset that examines economic adversity and racial segregation, it is outside of the scope of this e beneficial to examine, include the following: the number of concurrent sexual relationships (past or present), religion, relationship fidelity beliefs, and individual agency.
83 Future Directions My analysis leads to several recommendations for future rese arch. For example, I believe that an important area for future research is the possible link between the disproportionate incarceration of Blacks (relative to Whites) and STI disparities. Unfortunately, I was unable to look at incarceration rates in my m odel. The incarceration disparity is enormous: the U.S. Bureau of Justice Statistics estimates that, if current trends continue, 1 in 3 Black males will spend time in prison during their lives, compared with 1 in 17 White males (Khan et al., ., 2011). The i ncrease in the incarceration of Black males is a factor that exacerbates the sex ratio imbalance for Black females. Questions concerning the history of incarceration of past sexual partners could provide investigation into an area that could certainly be impactful to the analysis of the sex ratio imbalance and provide data concerning lack of power in sexual relationships. The literature notes a relationship between STI infections and incarceration (Alexander, 2010; Khan et al., ., 2011), but a better unders tanding of causal mechanisms and opportunities for prevention is needed. Future studies examining the influence of sex ratio would be more relevant if the study data reflected current sexual attitudes. Additional factors that should be taken into consider ation include: the impact of the social characteristics of a demographic area inter and the diverse range in education among Blacks. Characteristics that may be helpful in understanding sex ratio include: population density, median age, educational attai nment, median and/or per capita income and percent persons below poverty. For instance, the implications of oversampling well educated Blacks was not necessarily beneficial to the results, as well educated Blacks were not the target of the study and may h ave more knowledge concerning the subject. An additional factor is that assortative mating between the races seems to be more prominent in relationships
84 now than it most likely was in 1996 when the Add Health data was collected. A larger age continuum of participants and the ability to address causal pathways via change over time are also recommended. Also, worth noting, qualitative data (Ferguson et al., ., 2006; Adimora et al., 2002,; Adimora et al., ., 2001) generally supports the hypothesis that sex ratio am ong Blacks is a potential determinant of STI risk and transmission. However, quantitative studies (Benefo, 2008; Smith & Subramanian, 2006; Schmitt, 2005), like the one presented here, have had more mixed results that generally support a negative relation ship between sex ratios and sexual behavior. While this study attempted to address this disagreement in findings within the literature, the lack of significant findings, speaks to the overall difference in methodology between qualitative and quantitative studies and illustrates the importance of future research supplementing qualitative data with quantitative results. Given the nature of quantitative data, the analyses did not provide rich detail of experiences or capture the distinct and interconnected n ature of sexual beliefs and practices. I would like to focus on this aspect more in future studies through the use of qualitative methods, but believe the method used in this current research had the potential to provide an alternative way of telling coll ective stories by first uncovering the broader importance of sex ratio as a basis for future work. would be complete without also considering how the growing oversuppl y of Black females partner sexual intercourse and of contracting an STI. Consequently, any e
85 and STI risk may be counterbalanced by a parallel but opposite effect of a female surplus on Finally, i t is important to note certain issues concerning sex ratio theor y as well as the Add Health data. Sex ratio theory and the data utilized for the study only address the issue of heterosexual relationships and do not include references to homosexuality or bisexuality, which are now more openly recognized. Basing the in formation on only one category of sexual orientation demonstrates assumption and implication that risk behaviors are the same for all types of sexual orientation. The present research illustrates the importance of beliefs as well as the differential import ance of referent norms in predicting perception to engage in risky sexual behavior. These findings implicate the notion that risk reducing interventions might benefit from not only ceptions of their relationship between social determinants and sexual health outcomes for Blacks (Pouget et al., ., 2010; Senn et al., ., 200 8 ; Payne, 2008; Dunn, 2005) can support integrated public health prevention approaches that focus on upstream interventions such as education, housing, and health care access. The academic and governmental sectors can help by increasing funding for research and supporting the transl ation of the resulting knowledge into prevention program leaders trained to do culturally relevant research within their communities and to understand, value, and prioritize structural interventions. The multifaceted determinants of the STI epidemic among Blacks require preventive responses with legal, human rights, and health equity dimensions. Particularly, essential steps should include establishing mandatory screening and treatment in prisons; the possibility of
86 expansion to health care insurance cover age through the Affordable Care Act; improved targeting and adequate implementation of behavioral and biomedical interventions, including fully funded prevention and treatment clinics to provide evidence based services in underserved neighborhoods; as well as education regarding the risks of partner concurrency and high risk partners. In addition to these prevention efforts, accelerated progress is needed in four key intervention domains: research and evaluation, community mobilization, interagency collabo ration, and leadership (Aral, et al., 2008). Efforts in these areas could target Blacks through community based participatory research led by grass roots organizations and community stakeholders to create a suite of wrap around services guided by research findings. For example, services could include education and support groups regarding sex ratio imbalance and its effects on sexual beliefs and practices among Black women. Additionally, this research suggests that assisting women in planning to use condom s may not translate into actual use of these items. As such, the aim of future research in this area health promoting behavior. Potential determinants explored in the literature are perceived self efficacy and predicted outcomes regarding STIs (Senn et al., 200 8 ; Harvey et al., 2002); however, these have not been explored with this population. These factors must be further examined in relation to Black women, given t heir disproportionate risk of contracting HIV/AIDS and other STIs, if we as social scientists are to make significant strides toward decreasing risk within this demographic. Developing and implementing strategies to provide support and hope is a significa nt challenge facing mental health professionals working with Black women with STIs. However, with increased knowledge, sensitivity, and awareness or the interlocking nature of social and
87 contextual risks confronting the lives of Black women, we can develo p more effective prevention programs that will significantly decrease the rates of STIs for this group. In addition, there is a pressing need for developing support programs and services that comprehensively address social and contextual influences in the lives of Black women who are affected by STIs (Aral, et al., 2008). Doing so will have positive psychological, physical, and social effects that can cascade through the lives of Black women, their children and families, and their communities. Moreover, t he above suggestions could help reduce sexual health disparities by race. Diversity is more than a political catch phrase. No group is monolithic; there are unique factors that may affect gender, racial and ethnic groups differently. Although, the findin gs from this study show that sex ratio is at best a marginal indicator of sexual beliefs and practices among Whites, sex ratio has stronger implications for Blacks. Therefore, future studies should look at specific elements of sex ratio that may only affe ct diverse populations.
88 REFERENCES Adimora, A., Schoenbach,V., & Martinson, F. (2001). Social context of sexual relationships among rural African Americans. Sexually Transmitted Diseases 28 69 76. Adimora, A., Schoenbach, V., Bonas, D., Martinson, F. Donaldson, K., & Stancil, T. (2002). Concurrent sexual partnerships among women in the United States. Epidemiology 13, 320 327. Adimora, A., Schoenbach, V., Bonas, D., Martinson, F., Donaldson, K., Stancil, T., & Fullilove, R. (2004). Concurrent sexual partnerships among African Americans in the rural South. Epidemiology 14, 155 160. Adimora, A., Adaora A. & Schoenbach. V.J. (2005). Social context, sexual networks, and racial disparities in rates of sexually transmitted infections, The Journal of Infe ctious Diseases 191(S1): 115 122. Albrecht, C., Fosset, M., Cready, C. & Kiecolt, K. ( 1997 ) Journal of Family Issues, 18, 429 452. Albrecht, C. & Albrecht, D. (2001). Sex ratio an d family structure in the nonmetropolitan United States. Sociological Inquiry 71: 67 84. Alexander, M. (2010). The new Jim Crow: Mass incarceration in the age of colorblindness. New York: The New Press. Allard, R. (1990). A family of mathematical models to describe the risk of infection by a sexually transmitted agent. Epidemiology 1, 30 33. Andrinopoulos, K., Kerrigan, D. & Ellen, J. (2006). Understanding sex partner selection from the perspective of inner city black adolescents. Perspectives on Sexua l and Reproductive Health 38: 132 138. Aral, S O., Holmes, K K., Padian, N S. & Cates, W. (1996). Individual and population approaches to the epidemiology and prevention of sexually transmitted diseases and human immunodeficiency virus infection. Journal of Infectious Diseases 174, S127 133. Aral S.O. (2004). Sexual risk behaviour and infection: epidemiological considerations. Sex ually Transm itted Infect ions, 80 (Suppl 2):ii8 12. Aral, S.O., Adimora, A.A., & Fenton, K.A. (2008). Understanding and respond ing to disparities in HIV and other sexually transmitted infections in African Americans. Lancet 372: 337 340. Banks, R. (2011). Is Marriage for White People? How the African American Marriage Decline Affects Everyone. New York: Dutton Press.
89 Barrow, R. Berkel, C., Brooks, L., Groseclose, S., Johnson, D., & Valentine, J. (2008). Traditional Sexually Transmitted Disease Prevention and Control Strategies: Tailoring for African American Communities, Sexually Transmitted Diseases 35(12): S30 S39. Benefo K marital sex: a multi level analysis. Archives of Sexual Behavior 37: 517 29. Buss, D. M. & Barnes, M. ( 1986 ). Preferences in human mate selection. Journal of Personality and Social Psychology 50(3), 559 570. Centers for Disease Control and Prevention. (2006). Sexually Transmitted Disease Surveillance, US Department of Health and Human Services. Centers for Disease Control and Prevention. (2013). HIV surveillance r eport, 25. Centers for Disease Cont rol and Prevention. (2014). Sexually transmitted disease surveillance 2013. Atlanta: U.S. Department of Health and Human Services. http://www.cdc.gov/hiv/library/reports/surveillance/. Published February 2015. Accessed April 7, 2015. Cherlin, A. (1992). M arriage, Divorce, Remarriage Cambridge, MA: Harvard University Press. Cornelius, L.J Okundaye, J.N & Manning M.C ( 2000 ) Human immunodeficiency virus rel ated risk behavior among African American females. J Natl Med Assoc 2000 Apr. 92(4):183 95. Cummings, G., Battle. R., Barker, J. & Krasnovsky, F. (1999). AIDS Education and Prevention 11(4), 331 342. Doherty, I., Padian, N., Marlow, C., & Aral, S. (2005). Determinants and consequences of sexual networks as they affect the spread of sexually transmitted infections The Journal of Infectious Disease 191 (Suppl 1): S42 54. Dolcini, M.M., & Cat ania, J.A. (2000). Psychosocial profiles of women with risky sexual partners: The national behavioral surveys (NABS). AIDS and Behavior 4, 297 308. Dunn, J (2005). Examining the theory of reasoned action in relation to sexual risk taking behavior among African American Women England, P. & Farkas, G. (1987). Households, employment and gender: a social, economic and demographic view. by the sexual division of work: conceptual revolutions in the social sciences. American Journal of Sociology 92, ( 4), 10 42 1044. Eyre, S.L., Auerswald, C, Hoffman, V. (1998). Fidelity management: African American Journal of Health Psychology 3:393 406.
90 Ferguson, Y, Quinn S.C, Eng, E, & Sadelowski / M. (2006). The gender ratio imbalance and its relationship to risk of HIV/AIDS among African American women at historically black colleges and universities. AIDS Care 18; 323 31. Foreman, F. (2003). Intimate risk: sexual risk behavior among African American college women, Journal of Black Studies 33(5), 637 653. Guttentag, M. & Secord, P. (1983). Too Many Women: The Sex Ratio Question Beverly Hills: Sage. Harvey, S. M., Bird, S. T., Galavotti, C., Duncan, E. A., & Greenberg, D. (2002). Relationship power sexual decision making, and condom use among women at risk for HIV/STDs. Women & Health 36, 69 84. Harvey, S.J. & Bird, S.T. (2004). What makes women feel powerful? An exploratory study of relationship power and sexual decision making with African Amer icans at risk for HIV/STDs. Women and Health 39, 1 18. Ickovics, J.R. & Rodin, J. (1992). Women and AIDS in the United States: Epidemiology, natural history, and mediating mechanisms. Health Psychology 11, 1 16. Johnson, R. & Raphael, S. (2006). The e ffects of male incarceration dynamics on AIDS in fection rates among African American women and men U npublished manuscript. Goldman Scchool of Public Policy. University of California Berkeley. Kalichman, S.C., Rompa, D., Luke, W., & Austin, J. (2002). HIV transmission risk behaviors among HIV positive persons in serodiscordant relationships. International Journal of STD and AIDS 13, 677 682. Khan, M., Behrend, L., Ad imora, A., Weir, S., Tisdale, C., & Wohl, D. (2011). Dissolution of primary intimate rela tionships during incarceration and associations with post release STI/HIV Risk behavior in a southeastern city, Sexually Transmitted Disease 38(1) 43 47. Koopman, JS. & Longini, IM. Jr. (1994). The ecological effects of individual exposures and nonlinear disease dynamics in populations. American Journal of Public Health 84, 836 842. Koopman J.S & Lynch J.W. (1999) Individual causal models and population system models in epidemiology. American Journal of Public Health 89: 1170 74. Kreider, F. (2007). Living Arrangements of Children Current Population Reports, P70 114. U.S. Census Bureau, Washington, DC. Laumann, EO., Cagnon, JH. & Michael, S. (1994). The Social Organization of Sexuality Chicago: The University of Chicago Press.
91 Laumann, E.O. & Youm Y. (1999). Racial/ethnic group differences in the prevalence of sexually transmitted diseases in the United States: a network explanation. Sexually Transmitted Diseases 26, 250 261. Leigh, W. (1995). The health of African American women Health issues for women of color: A cultural diversity perspective (112 132) Thousand Oaks, CA: Sage. Lichter, D.T, Anderson, R.N. & Hayward, M.D. ( 1995 ). Marriage markets and marital choice. Journal of Family Issues ; 16 :412 431. Logan, T.K., Cole,J., & Lukefeld, C. (2002). Women, sex, and HIV: Social and contextual factors, meta analysis of published interventions, and implications for practice and research. Psychological Bulletin 128 851 885. Manhart, L.E., Aral, S.O., Holmes, K.K., Foxman, B. (2002). Sex partner concurrency: Measurement, prevalence, and correlates among urban 18 39 year olds. Sexually Transmitted Diseases 29, 133 143. Mare, R., & Winship, C. (1991). Socioeconomic change and the decline of marriage in blacks and whites. In C. Jencks & P. Peters on (eds.), The urban underclass (pp. 175 20 2). Washington, DC: The Brookings Institution. McNair, L.D & Prather, C.M. (2004). African American women and AIDS: Factors influencing risk and reaction to HIV disease. Journal of Black Psychology 30, 106 123. McQuillan, G. M Kruszon Moran, D., Kottiri, B. J Kamimoto L A Lam L. Cowart F. Hubbard M & Spira T J. ( 2006 ). Prevalence of HIV in the U.S. household population The National Health and Nutrition Examination Surveys, 1988 to 2002. Journal of Acquired Immune Deficiency Syndrome 41 :651 656. Morris, C., Morris, S. & Ferguson, A. (2009). Sexual behavior of female sex workers and access to condoms in Kenya and Uganda on the trans Africa H ighway, AIDS and Behavior 13 (5), 860 865. Morris, M., Kurth, A., Hamilton, D., Moody, J., Wakefield, S., & Handcock, M. (2009). Concurrent partnerships and HIV prevalence disparities by race: linking science and public health practice. American Journal of Public Health 99(6): 1023 31. Newman, L. & Berman, S. (2008). Epidemiology of STD disparities in African American communities, Sexually Transmitted Diseases 35(12): S4 S12. Norton, A., & Mooman, J. (1987). Current trends in marriage and divorce amon g American women. Journal of Marriage Family 49, 3 14.
92 Payne, J line). Available from http://health.usnews.com/articles/health/2008/09/12/black womens burden an epidemic of hiv.html (accessed Oc tober 21, 2009). Pouget E.R., Kershaw T.S., Niccolai L.M., Ickovics J.R. & Blankenship K.M. ( 201 0 ). Associations of sex ratios and male incarceration rates with multiple opposite sex partners: potential social determinants of HIV/STI transmission Public Health Rep .; 125(Suppl 4): 70 80. PMCID: PMC2882977 Pulerwitz, J., Amaro, H., DeJong, W., Gortma ker, SL., & Rudd, R. (2002). Relationship power, condom use and HIV risk among women in the USA. AIDS Care 14, 789 800. Quinn, T.C., Wawer, M.J., Sewankambo, N., Serwadda, D ., Li, C ., Wabwire Mangen, F ., Meehan, M.O ., Lutalo, T & Gray, R.H (2000).Viral load and heterosexual transmission of human immunodeficiency virus type 1. N ew England Journal Me dicine. 342: 921 929. Robinson, B., Scheltema, K., & Cherry, T. (2005). Risky Sexual Behavior in Low Income African American Women: The Impact of Sexual Health Variables, The Journal of Sex Research 42(3), 224 237. Schmitt D.P. (200 5). Sociosexuality from Argentina to Zimbabwe: a 48 nation study of sex, culture, and strategies of human mating. Behavioral and Brain Sciences 28:247 311. Schoen, R. & Kluegel, J. (1988). The widening gap in Black and white marriage rates: the impact of population composition and differential marriage propensities. American Sociological Review 53:408 411. Schoenbach, V., Adimora, A., Donaldson, & K., Stancil, T. (2000). Ethnic differences in concurrent sexual partnerships for women in the 1995 national survey of family growth. Presented at: American Public Health Association, 128 th Meeting: November 12 16, Boston: MA. Senn, T.E, Carey, M.P, Vanable, P.A, Urban, M.A, & Sliwinski M.J. (2008). The male to female ratio and multiple sexual partners: multile vel analysis with patients from an STD clinic. AIDS Behavior [cited on May 2011]. Available from: http://www.springerlink.com/content/n2871320t541735m Simpson, J. (2007). Interpersonal foundations of trust. In A. Kruglanski and E.T. Higgins (Eds.), Socia l psychology: Handbook of basic principles (2d ed) pp. 587 607. New York: Guilford Press. Smith, A.M. & Subramanian, S.V. (2006). Population contextual associations with heterosexual partner numbers: a multilevel analysis. Sexually Transmitted Infections, 82: 250 4.
93 Soet, J. E., Dudley, W. N., & Dilorio, C. (1999). The effects of ethnicity and perceived power on Psychology of Women Quarterly 23, 707 723. South, S. & Lloyd, K. M. ( 1992 ). Marriage markets and nonmarital fertility in the United States. demography. ; 29 :247 64. transmitted infection in China, Journal of Health and Social Behavior 51: 376 391. Thomas, J.C. (2006). From sl avery to incarceration: Social forces affecting the epidemiology of sexually transmitted diseases in the rural south. Sexually Transmitted Diseases, 33S6 S10. Treas, J. & Giesen, Deirdre. (2000). Sexual infidelity among married and cohabiting Americans. J ournal of Marriage and Family 62(1), 48 60. Trent, K. & South, S. (2003). Spousal alternatives and marital relations. Journal of Family Issues 24(6): 787 810. Turner, C.F., Ku, L, Rogers, S.M., Lindberg, L.D., Pleck, J.H., Sonenstein, F.L. (1998). Adol escent sexual behavior, drug use, and violence: increased reporting with computer survey technology. Science 280:867 873. U.S. Bureau of the Census. (1998). Current Population Survey. Marital Status and Living Arrangements Washington DC: US Bureau of th e Census. U.S. Bureau of the Census. ( 2000 ). Current Population Survey. Census 2000 Summary File 1[United States]/prepared by the U.S. Census Bureau, 200 0. Washington DC U.S. Census Bureau (2001)Population by age, sex, race, and Hispanic or Latino origin for the United States: 2000 (PHC T 9) [cited 2011 June 11]. Available from: http://www.census.gov/population/www/cen2000/briefs/phc t9/index.html U.S. Bureau of the Census. ( 2010 ). Census 2010 Summary File 1 United States/prepared by the U.S. Census Bur eau, 2011 Washington DC The Sociological Quarterly 51:408 35. Upchurch, D.M., Aneshensel, D.M., Sucoff, C.A. & Levy Storms, L. ( 199 9) Neighborhood and family contexts of adolescent sexual activity. Journal of Marriage and Famil. 61, (4) 920 933 Valentine, J. (2008). Impact of attitudes and beliefs regarding African American sexual behavior on STD prevention and control in African A merican communities: Unintended consequences, Sexually Transmitted Diseases 35(12): S23 S29.
94 Whyte, J. (2006). Sexual assertiveness in low income African American women: Unwanted sex, survival and HIV risk, Journal of Community Health Nursing 23: 235 244 Zenilman, J., Weisman, C., Rompalo, A., Ellish, N., Upchurch D., Hood, E., & Celentano. (1995). Condom use to prevent incident STDs: The validity of self reported condom use, Sexually Transmitted Diseases, 22(1): 1 67.