RESEARCH
Sexual violence against intimate partners in Cape Town: prevalence and risk factors reported by men
Violence sexuelle contre les femmes au Cap : prévalence et facteurs de risque rapportés par le partenaire
Violencia sexual contra la pareja en Ciudad del Cabo: prevalencia y factores de riesgo referidos por los hombres
Naeemah AbrahamsI,1; Rachel JewkesI; Margaret HoffmanII; Ria LaubsherIII
IGender and Health Research Group, South African Medical Research Council, PO Box 19070 Tygerberg 7505, South Africa
IIWomen's Health, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa
IIIBio-statistics Unit, South African Medical Research Council, Tygerberg, South Africa
ABSTRACT
OBJECTIVE: To describe the prevalence of and risk factors for the perpetration of sexual violence by men against female intimate partners.
METHODS: A cross-sectional study was conducted on 1368 randomly selected men working in three Cape Town municipalities. The men were interviewed with the aid of a questionnaire on current sexual partners in the preceding 10 years, personal and relationship characteristics and the use of violence against their partners.
RESULTS: The perpetration of sexual violence against intimate partners in the past 10 years was reported by 15.3% of the men. After adjustment for sociodemographic circumstances, the factors associated with such violence were involvement in physical conflict outside the home, problematic alcohol use, having more than one current partner and abusing partners verbally. While having frequent conflict with partners was important for the risk of sexual violence, only two types of conflict sources were significantly associated with this risk, namely conflict over sexual refusal and conflict when men perceived their authority to be undermined.
CONCLUSION: Sexual violence in intimate relations was common. The risk of being sexually violent was associated with the use of violence to solve problems in other settings, having more than one current partner, alcohol abuse and verbally abusing a partner. It was also associated with particular types of conflict stemming from ideas of male sexual entitlement and dominance. Prevention programmes that focus on gender relations and non-violent conflict resolution for men and youths may be useful in combating such sexual violence.
Keywords: Sexual partners; Spouse abuse; Rape; Gender identity; Men; Violence; Alcohol drinking/adverse effects; Risk factors; Cross-sectional studies; South Africa (source: MeSH, NLM).
RÉSUMÉ
OBJECTIF: Décrire la prévalence et les facteurs de risque de violence sexuelle exercée par les hommes sur leur épouse, compagne ou partenaire.
MÉTHODES: Une étude transversale a été réalisée après tirage au sort sur 1368 hommes travaillant dans trois municipalités du Cap, en Afrique du Sud. Les hommes ont été interrogés au moyen d'un questionnaire sur leurs partenaires des dix dernières années, sur leurs caractéristiques personnelles et celles de la relation, et sur l'usage de la violence à l'encontre de leurs partenaires.
RÉSULTATS: Des actes de violence sexuelle à l'encontre de leur épouse, compagne ou partenaire au cours des dix dernières années ont été rapportés par 15,3 % des hommes. Après ajustement sur les conditions socio-démographiques, les facteurs associés à la violence sexuelle étaient l'implication dans des bagarres hors du domicile, les problèmes d'alcool, le fait d'avoir plus d'une partenaire à la fois et la violence verbale contre les partenaires. Si les disputes fréquentes étaient un important facteur de risque de violence sexuelle, il n'existait d'association significative que pour deux types de sources de conflit, à savoir le refus de l'acte sexuel et la perte d'autorité ressentie par l'homme.
CONCLUSION: La violence sexuelle dans les relations intimes était fréquente. Le risque de violence sexuelle de la part de l'homme était associé au recours à la violence pour résoudre d'autres problèmes, à l'existence de plus d'une partenaire à la fois, à l'abus d'alcool et à la violence verbale contre les partenaires. Il était également associé à des types particuliers de conflits liés aux concepts de pouvoir et de domination sexuels masculins. Des programmes de prévention axés sur les relations homme-femme et sur la résolution non violente des conflits à l'intention des hommes et des adolescents pourraient être utiles pour combattre cette forme de violence.
Mots clés: Partenaire sexuel; Epouse maltraitée; Viol; Identité masculin féminin; Hommes; Violence; Consommation alcool/effets indésirables; Facteur risque; Etude section efficace; Afrique du Sud (source: MeSH, INSERM).
RESUMEN
OBJETIVO: Describir la prevalencia y los factores de riesgo de agresión sexual de los hombres hacia su pareja.
MÉTODOS: Se realizó un estudio transversal entre 1368 hombres seleccionados aleatoriamente que trabajaban en tres municipios de Ciudad del Cabo. Los hombres tuvieron que responder a un cuestionario sobre sus parejas en los diez años precedentes, las características personales y de la relación, y su posible recurso a la violencia contra las mujeres.
RESULTADOS: Un 15,3% de los hombres refirieron haber agredido sexualmente a su pareja en los diez años precedentes. Tras el ajuste en función de las circunstancias sociodemográficas, los factores asociados a la violencia fueron la participación en conflictos físicos fuera del hogar, los problemas de consumo de alcohol, el hecho de tener más de una pareja y el abuso verbal con la pareja. Aunque el hecho de tener conflictos frecuentes con la mujer era un factor de riesgo importante de violencia sexual, sólo había dos tipos de conflicto asociados significativamente a ese riesgo: los derivados del rechazo sexual, y las desavenencias que entrañaban menoscabo de la autoridad del hombre.
CONCLUSIÓN: La violencia sexual en las relaciones íntimas era un fenómeno común. El riesgo de perpetrar actos de violencia sexual estaba asociado al recurso a la violencia para resolver los problemas en otros entornos, al hecho de tener más de una pareja, al abuso de alcohol y al maltrato verbal de la pareja. Se asociaba también a tipos particulares de conflicto derivados de determinadas ideas sobre la dominancia y los derechos sexuales masculinos. Los programas de prevención dirigidos a los hombres y los jóvenes y centrados en las relaciones de género y en la resolución no violenta de los conflictos podrían ser muy útiles para combatir la violencia sexual.
Palabras clave: Parejas sexuales; Maltrato conyugal; Violación; Identidad sexual; Hombres; Violencia; Consumo de bebidas alcohólicas/efectos adversos; Factores de riesgo; Estudios transversales; Sudáfrica. (fuente: DeCS, BIREME).
Introduction
The World report on violence and health highlighted both the impact of sexual violence on public health and the scantiness of the research base on which to build an understanding of the problem and develop interventions (1). Whereas sexual violence is ultimately a product of male behaviour, most research on such violence has been concerned with the women who experience it. This has contributed little towards an understanding of the most important risk factors. Studies on women predict very little of their risk of sexual violence and point to the need for research on men (2).
Research on men's sexual violence towards women has mainly been conducted in developed countries on populations of mostly young college students, men in the military or men in treatment programmes (3-6). The relevance of the findings to the developing world are unclear. Studies have been conducted in India on the prevalence of and risk factors for sexual violence against intimate partners (7) and on the association between sexual behaviour and reproductive outcomes (8); 7% of husbands reported that at some time they had physically forced sex on their wives, and it emerged that this was more common among husbands who reported having extramarital sex and those who reported symptoms of sexually transmitted diseases than among other husbands.
Rape and sexual coercion in South Africa have been highlighted in several studies (9-12). Rape by non-partners has received considerable attention in the media but sexual violence perpetrated by intimate partners is believed to be more common (12). Most of the research on this topic in South Africa has focused on women as subjects and on physical intimate partner violence (13). The present study, while recognizing that there is substantial overlapping between the emotional, physical and sexual types of intimate partner violence, focuses on sexual violence perpetrated by men and describes the prevalence of and risk factors for such violence against intimate partners.
Methods
A cross-sectional study of men working in three of the six municipalities in Cape Town was conducted between June 1998 and February 1999. The sampling frame was a list of male employees working in the Civil Engineering, Water and Cleansing, and Parks and Recreation Divisions. A computer-generated random sample of 600 names from each of the three municipalities was obtained, giving a total of 1800 names. Of these, 37 were women's names, 28 were of men who reported no female partners, 66 were of men who refused to participate, 283 were of men who were unavailable because of absenteeism or leave, and 18 were of men with whom incomplete interviews were conducted. Thus an overall response rate of 78.8% was achieved and the analysis was based on 1368 interviews. The data were collected by trained male personnel in face-to-face interviews conducted in Afrikaans, Xhosa or English, depending on which was preferred by the interviewees, with the aid of a structured questionnaire.
The choice of variables for the identification of risk factors was based on the integrated ecological model developed by Heise (14) and on previous work with men (15). Sociodemographic variables included age, ethnicity, education, occupation and type of housing. The following childhood variables were used: the presence of a father during childhood; childhood discipline, including physical punishment, classified as frequent (daily/ weekly) or infrequent; and witnessing abuse perpetrated against a mother. Perceptions on the acceptance of violence, and gender roles in relationships, were measured on two composite scales. An 18-item scale measured the acceptance of violence by representing various scenarios, and an 11-item scale, adapted from Rouse (16), measured views on gender roles: the corresponding values of Cronbach's alpha coefficient were 0.75 and 0.71. The interviewees were asked whether they considered it acceptable to hit a woman. Those who replied in the affirmative were asked to give reasons. Among the behavioural variables were the following: alcohol and drug use, including current use and use at any time in the past (for men who did not report current use), whether such use had ever created problems for the interviewees, whether they had ever been involved in gangs, whether they had been arrested by the police, reasons for arrest if this had occurred, whether they had ever been imprisoned, whether they had ever been involved in physical fights at work or in the community, and whether they had more than one current partner.
The respondents were asked to identify all the female partners with whom they had had meaningful relationships during the preceding 10 years. Such relationships were defined as ones in which the partners were married, lived together, had a child together or went out together for more than one month. The men could identify more than one partner, i.e. previous partners or concurrent partners. Most previous research on intimate partner violence has been based on lifetime and 12-month recall periods and lifetime and current partners. In order to improve recall, we considered a 10-year period and meaningful relationships. This was important because the study collected data on partner and relationship characteristics.
The 1368 men reported 2056 female partners, with 215 (15.7%) reporting more than one current partner, up to a maximum of four. Questions on each identified partner concerned the type and length of the relationship, the partner's age, work status and earnings in relation to those of the respondent, the number of children the man and partner had together, the partner's use of alcohol and drugs and whether such use created problems in the relationship. Conflict questions included the frequency of and reasons for conflict, adapted from Straus et al. (17). Verbal abuse, a strong predictor of intimate partner violence (13, 18, 19), was included as an explanatory variable. Data on only the current partners were used in the analysis, i.e. 1314 (54 men reported on previous partners only). Where more than one current partner was reported, data for the first partner mentioned were included.
For each partner, questions were asked on sexual violence following the questions on physical, emotional and verbal abuse. A respondent was identified as a perpetrator of sexual violence if he answered that he had ever tried to force or had forced one or more of his partners to have sex. WHO guidelines on ethics and safety in studies on violence against women (20) were adapted for male respondents, as described elsewhere (21).
The data were analysed using STATA version 7 software (22). Estimates of the prevalence of sexual violence were obtained. Unadjusted odds ratios with confidence intervals (CIs) were calculated for relevant risk factor variables. This process assisted in identifying risk factors for inclusion in the multiple logistic regression analysis. The possibility of interactions between risk factors and potential confounders was investigated by applying the log likelihood test, but no significant effects were found.
Using a backward stepwise elimination regression analysis, we first tested whether potential confounders made significant contributions to the prediction of sexual abuse; none was found. The model was adjusted for race, age and occupation and was built in stages: sociodemographic variables, sociocultural and childhood variables, variables relating to individual behaviours, and female partner variables including conflict variables were added in succession. The final model had only the remaining explanatory variables with odds ratios (ORs) significant at or below the 0.05 level. In addition the final model was tested for significant interviewer effects. The xtlogit command in STATA was used to adjust for these random effects in the final model (22).
Results
The average age of the men was 38.9 years (standard deviation (SD) = 10.3), and their ethnic profile was similar to the general ethnic profile of Western Cape Province; 6% (82) reported not having had any formal education, 28.1% (384) reported having completed grades 10, 11 or 12, and 18% (247) reported that they had received further education after leaving school; 54.3% (743) were unskilled workers, while only 4.6% (63) belonged to the professional category; 77.3% (1016) described relationships with wives and 22.6% (298) reported relationships with girlfriends, 62.4% (186) of whom were non-cohabiting girlfriends.
Reports of forcing sex on an intimate partner and of attempting to do so during the preceding 10 years were given by 7.1% (97) and 8.2% (112) of the men, respectively; of the 209 men who reported sexual violence, 80.9% (169) also reported perpetrating physical and emotional abuse.
In Table 1, crude ORs are used to compare men who reported sexual violence with those who did not, in respect of demographic characteristics, childhood experiences, behavioural variables and views on gender roles. There was a significant difference (t test, P <0.0001) between the mean age of men reporting sexual abuse (35.6 years, SD = 9.2) and that of men not reporting such violence (39.5 years, SD = 10.4). Men who had not received post-school training were three times more likely to report perpetrating sexual abuse than those who had had such training. The childhood experiences that were predictors of sexual violence included being physically punished frequently, i.e. daily or weekly, and witnessing abuse perpetrated on a mother. Significant positive associations were found for all the behavioural variables. A strong positive association was found with the perception that hitting a partner was justified. The mean score for the scale measuring men's views on gender relations and gender roles was significantly lower for men reporting sexual abuse than for other men, indicating reduced support for gender equity; in contrast, the median score for accepting use of violence in various scenarios was significantly higher for men reporting sexual violence.
The associations between sexual violence and partner characteristics are shown in Table 2. Reporting sexual violence was not associated with the female partner's age, education, employment or use of alcohol. However, an increase in risk was found if the partner was a cohabiting girlfriend and if her alcohol use was reported as problematic. Strong positive associations were found in respect of reports of frequent conflicts, i.e. weekly or more often, and for most of the reasons given for conflicts.
The multiple logistic regression models for factors associated with the reporting of sexual violence after adjusting for sociodemographic variables and for interviewers in the final model are shown in Table 3. Before conflict variables were included in the model, sexual violence was significantly positively associated with involvement in fights in communities, problematic alcohol use, having more than one current partner and the perception that hitting a partner was justified. After conflict variables were added, the perception that hitting a partner was justified did not remain a significant predictor, while verbal abuse of the partner, more frequent conflicts, conflict related to a partner's refusal to have sex and to a perception by a man that his authority was being undermined all became significant. The change in log likelihood indicated the inclusion of the conflict variables to be important. However, involvement in physical fights in the community and conflict frequency only reached borderline significance.
The conflict variables in the model were tested by means of a scale that included all the reasons for conflict and a summary variable that encompassed the number of different reasons for conflict and conflict frequency. Neither of these improved the model significantly.
Discussion
Comparisons of prevalence estimates of intimate partner violence studies are problematic (23). The present study concerned a sample of working men who reported on significant partners during the preceding10 years, and the results were very difficult to compare with those obtained in other studies. Nevertheless, the prevalence of forced sex (7.1%) was similar to that reported from a study in India (8). In our study, 80.9% of the men who reported sexual violence also reported both emotional and physical violence against partners. This substantial overlap in types of abuse suggests that they may be part of the same underlying construct. However, some men in our study were exclusively sexually violent, a phenomenon that has been reported to varying degrees in other settings (24).
Demographic and childhood variables
None of the demographic and childhood variables were independently associated with reported sexual violence. An analysis, not presented here, points to a close interrelationship between childhood variables: witnessing abuse perpetrated against a mother is associated with being physically punished in childhood and with being involved in fights in the community in adulthood (25). This suggests that the exposure of these men to violence during their childhood was associated with a greater risk of being involved in other violent behaviour and that this behaviour was more strongly associated with the risk of sexual intimate partner violence than would otherwise have been the case.
Problematic alcohol use
The association between reported sexual violence and problematic alcohol use is consistent with previous studies on women. There have been reports of strong associations between abuse and conflict created by men's use of alcohol (13), their drunkenness (18) and their problem drinking (26). A study of men in Thailand found that there was a similar relationship between physical violence and problematic alcohol use (19) and that alcohol use had an indirect effect on the abuse of wives by weakening marital companionship. The precipitating effect of alcohol use on the perpetration of interpersonal violence in general as well as sexual violence is more complex than the explanation based on disinhibition which has commonly been given. More recent anthropological data suggest that social learning may be involved (27).
Number of current partners and HIV infection status
The finding that sexual violence was associated with having more than one current partner reflects a report from India that sexually violent husbands were more likely than others to have extramarital affairs (8). It has been suggested that such men find it comparatively difficult to engage in intimate and loving sexual relationships (28). Moreover, there is a possible connection between sexual violence and potential infection with human immunodeficiency virus (HIV). Dunkle (29) suggested that an important dimension of the links between overall intimate partner violence and HIV infection may be that there is an increased likelihood of violent men being infected with HIV because of other associated behaviour. This could be an example of such behaviour.
Verbal abuse
There was a positive relationship between sexual violence and verbally abusive tactics used during conflict. Verbal marital conflict has been reported to be associated with physical violence against intimate partners and to be one of the three main predictors of husbands' use of physical violence against wives (13). In our study the frequency of conflict only reached borderline significance. Nevertheless, this corroborated a previous finding that frequent verbal and non-verbal conflicts were one of the strongest predictors of women's experiences of physical violence (13).
Gender relations
Our study demonstrates the important links between sexual violence and ideas about gender relations, notably in the finding that only two types of conflict were associated with sexual violence, namely sexual refusal, the strongest predictor, and perceived challenges to male control in relationships. There have been reports on the prevalence of ideas among South African men about entitlement to sexual access to partners whenever they want it (30-33), and on patriarchal notions of masculinity involving distinctly hierarchical gender positions and definitions of male success in terms of controlling women (32, 34). In this context it is not surprising to find that the perceived challenges to male authority are associated with sexual violence: the act of forcing sex on a partner demonstrates superior strength and ultimately symbolizes the gender dimension in inequalities of power in intimate relationships. This finding is also an indication of how conflict in relationships is influenced by concepts within the broader culture and social environment (32). In India, such factors include dowry disputes, not having a male child, and female sterilization (2). This highlights the importance of understanding that sexual violence occurs within different contexts and of developing interventions that are culture-specific.
Limitations on the study
Our sample of working men was not representative of all men in Cape Town. A review of risk factors has shown that unemployed men and men in low socioeconomic categories report higher rates of intimate partner violence than other men (19, 35). Consequently, perpetrators of such violence were possibly underrepresented in this sample. Together with recall, this suggests that the prevalence of violence could have been underestimated. Our results, based on a 10-year period, are not easily comparable with other studies, and we did not investigate violence in less important relationships or violence directed against non-intimate partners. Moreover, our study may not have collected data on all the important risk factors for the perpetration of sexual violence. Finally, it should be pointed out that although the cross-sectional design does not allow causation to be demonstrated, the factors identified are of great value in relation to intervention strategies.
Conclusion
Sexual violence in intimate relationships was a common occurrence in the group of men studied. Our investigation was one of only a few that have described men's factors associated with such violence. The risk of being sexually violent was associated with using violence to solve problems in other settings, having more than one current partner, alcohol abuse and verbally abusing a partner. It was also associated with particular types of conflict stemming from ideas of male sexual entitlement and dominance. It is particularly important to have prevention programmes that focus on gender issues and non-violent conflict resolution among men and youths.
The study confirms a critical need for further research among men so that a better understanding of this important public health problem can be gained. This was recognized by the Global Forum for Health Research in 2000 in its support for a Sexual Violence Research Initiative to be launched in 2004 and in its current development of a comprehensive international research agenda on sexual violence. 
Acknowledgements
The study was supported financially by the South African Department of Art, Culture, Science and Technology. The support of municipal management and the unions is gratefully acknowledged. Thanks are extended to the field workers and to Miriam Hossain, Desiree Pieterse and Charles Maisel for support during fieldwork training. The men who participated in the study are thanked for sharing their life experiences with the researchers.
Conflicts of interest: none declared.
References
1. Jewkes R, Sen P, Garcia-Moreno C. Sexual violence. In: Krug EG, Dahlberg LL, Mercy JA, Zwi AB, Lozano R, editors. World report on violence and health. Geneva: World Health Organization; 2002. p.149-81.
2. Heise L, Garcia-Moreno C. Violence by intimate partners. In: Krug EG, Dahlberg LL, Mercy JA, Zwi AB, Lozano R, editors. World report on violence and health. Geneva: World Health Organization; 2002. p.89-121.
3. Leonard KE, Blane HT. Alcohol and marital aggression in a national sample of young men. Journal of Intentional Violence 1992;7:19-30.
4. Murphy CM, Meyer SL, O'Leary KD. Family of origin violence and MCMI-II psychopathology among partner assaultive men. Violence and Victims 1993;8:165-76.
5. Neidig PH, Friedman DH, Collins BS. Attitudinal characteristics of males who have engaged in spouse abuse. Journal of Family Violence 1986;1:223-33.
6. Pan HS, Neidig PH, O'Leary KD. Predicting mild and severe husband-to-wife physical aggression. Journal of Consulting Clinical Psychology 1994;62:975-81.
7. Martin SL, Tsui AO, Maitra K, Marinshaw R. Domestic violence in northern India. American Journal of Epidemiology 1999;150:415-26.
8. Martin SL, Kilgallen B, Tsui AO, Maitra K, Singh KK, Kupper LL. Sexual behaviors and reproductive health outcomes - associations with wife abuse in India. Journal of the American Medical Association 1999;282:1967-72.
9. Human Rights Watch. Scared at school: sexual violence against girls in South African schools. Johannesburg: Human Rights Watch; 2001.
10. Human Rights Watch. Violence against women in South Africa: state response to domestic violence and rape. New York/Washington: Human Rights Watch; 1995.
11. Human Rights Watch. South African violence against women and the medico-legal system. New York/Washington: Human Rights Watch; 1997.
12. Jewkes R, Abrahams N. The epidemiology of rape and sexual coercion in South Africa: an overview. Social Science and Medicine 2002;55:1231-44.
13. Jewkes R, Levin J, Penn-Kekana L. Risk factors for domestic violence: findings from a South African cross-sectional study. Social Science and Medicine 2002;55:1603-17.
14. Heise LL. Violence against women: An integrated ecological framework. Violence Against Women 1999;4:262-90.
15. Abrahams N, Jewkes R. Men on violence against women. Psychological Bulletin 1998;8(2):30-4.
16. Rouse LP. Conflict tactics used by men in marital disputes. In: Hotaling GT, Finkelhon D, Kirkpatrick JT, Straus MA, editors. Family abuse and its consequences. Newbury Park: Sage Publications; 1988. p.176-91.
17. Straus MA, Hamby SL, Boney-McCoy S, Sugarman DB. The revised Conflict Tactics Scales (CTS2). Journal of Family Issues 1996;17:283-316.
18. Ahuja RC, Bangdiwala S, Bhambal SS, Jain D, Jeyaseelan L, Kumar S, et al. Domestic violence in India a summary report of a multi-site household survey. Washington DC: International Center for Research on Women; 2000.
19. Hoffman KL, Demo DH, Edwards NE. Physical wife abuse in a non-Western society: an integrated theoretical approach. Journal of Marriage and Family 1994;56:131-46.
20. World Health Organization. Putting women first: Ethical and safety recommendations for research on domestic violence against women. Geneva: World Health Organization; 2001.
21. Jewkes R, Watts C, Abrahams N, Penn-Kekana L, Garcia-Moreno C. Ethical and methodological issues in conducting research on gender-based violence in southern Africa. Reproductive Health Matters 2000;8:93-103.
22. STATA. Stata Reference Manual. 6th ed. College Station (TX): Stata Press; 1999.
23. Heise L, Ellsberg M, Gottemoeller M. Ending violence against women. Baltimore: The Johns Hopkins University School of Public Health; 1999 (Population Reports, Volume XXVII, Number 4; Population Information Program Center for Communication Programs).
24. World Health Organization. The WHO multi-country study on violence against women: preliminary findings. Second South African Gender-Based Violence Health Initiative Conference, Johannesburg, 7-9 May 2003 (keynote address by the Department of Gender and Women's Health, World Health Organization).
25. Abrahams N, Jewkes R. Impact of childhood exposure to violence. South African-United States Consultation on Child Sexual Abuse Research Meeting, Johannesburg, 4-7 May 2003.
26. Sharps PW, Campbell D, Gary F, Webster D, Campbell JC. The role of alcohol use in intimate partner femicide. American Journal of Addictions 2001;10:122-35.
27. McDonald M. Gender, drink and drugs. Oxford: Berg Publisher; 1994.
28. Malamuth NM, Linz D, Heavy CL, Barnes G, Acker M. Using the confluence model of sexual aggression to predict men's conflict with women: a 10-year follow-up study. Journal of Personality and Social Psychology 1995;69:353-69.
29. Dunkle KL. Gender-based violence as a risk factor for HIV infection among women attending antenatal clinics in Soweto, South Africa [DPhil dissertation]. University of Michigan; 2003.
30. Buga GAB, Amoko DHA, Ncayiyana DJ. Sexual behaviour, contraceptive practice and reproductive health among school adolescents in rural Transkei. South African Medical Journal 1996; 86(5):523-6.
31. Varga CA. Sexual decision-making and negotiation in the midst of AIDS: youth in KwaZulu/Natal, South Africa. Health Transition Review 1997;7:45-67.
32. Wood K, Jewkes R. 'Dangerous' love: reflections on violence among Xhosa Township youth. In: Morrell R, editor. Changing men in southern Africa. New York: Zed Books; 2001. p.317-36.
33. Wood K, Maforah F, Jewkes R. "He forced me to love him": Putting violence on adolescent sexual health agendas. Social Science and Medicine 1998;47:233-42.
34. Guy J. Gender oppression in southern Africa's pre-capitalist societies. In: Walker C, editor. Women and gender in southern Africa to 1945. Cape Town and London: David Phillip Publisher; 1999. p.33-47.
35. Black DA, Schumacher JA, Smith Slep AM, Heyman RE. risk factors for partner abuse and child maltreatment: a review of the literature. National Network of Family Resilience. Available from: URL: http://www.nnh.org/risk/
(Submitted: 25 February 03 Final revised version received: 14 October 03 Accepted: 15 October 03)
1 Correspondence should be sent to Dr Abrahams at this address (email: naeema.abrahams@mrc.ac.za).