Partners’ controlling behaviors and intimate partner sexual violence among married women in Uganda

We used the 2011 Uganda Demographic and Health Survey (UDHS) data, accessed with permission from the DHS Program website [24]. This was a cross-sectional nationally representative survey that used a stratified two-stage cluster sampling design [8] based on the sampling frame from the 2002 population and housing census [25]. A detailed description of the sampling procedure was reported in the 2011 UDHS report [8].

The sample for the domestic violence (DV) module was 2,056 ever-married women. From this sample, we extracted a weighted sample of 1,307 women who reported being in a union, for further analysis [8]. We used the domestic violence weighting variable (d005) included in the UDHS data and the Stata survey (svy) command to apply the weights during the analyses. Survey weighting is necessary to account for the complex survey design [26].

In this paper, women in a union included those who were married or cohabiting with their partners. The domestic violence module was based on a shortened and modified version of the Conflict Tactics Scale (CTS) [27]. The survey was carried out based on the World Health Organization’s (WHO) ethical and safety recommendations for research on domestic violence [28].

In the DHS, intimate partner sexual violence (IPSV) was measured as having experienced any sexual violence from a current or former partner in the 12 months prior to the survey [29]. For IPSV, women in a union were asked the following questions: a) Have you ever been physically forced into unwanted sex by your husband/partner? b) Have you ever been forced into other unwanted sexual acts by your husband/partner? c) Have you ever been physically forced to perform sexual acts when you did not want to? The binary responses from these three questions were merged into an aggregate measure of intimate partner sexual violence (variable d108 in DHS), which was coded as 0 = no, and 1 = yes.

We classified independent variables into three categories: first, women’s empowerment indicators, namely participation in decision-making, attitudes justifying physical violence, and economic empowerment indicators. The second group of variables comprised husbands/partners’ controlling behavior, women’s fear of their partners, and their partners’ alcohol consumption. The third category of explanatory variables included women’s socio-demographic factors such as having witnessed parental violence themselves. This categorization was used in our earlier publication [20]. The term “partner” in this paper includes husbands as well as partners in cohabiting unions.

Five indicators in the DHS measured women’s participation in household decision-making. Earlier publications have used these indicators [20,29]. In this paper, we used all five measures of decision-making autonomy regarding who usually makes decisions about: a) how women’s earnings are used; b) women’s healthcare; c) large household purchases; d) visits to family or relatives; and e) what to do with the money the partner earns. Responses to these questions were recoded into two categories (1 = woman decides alone/jointly with partner, 0 = partner alone/others). The assumption was that women who made decisions either alone or jointly with their partners were more empowered than those in households where decisions were made by either their partners alone or other people [20,29].

Attitudes justifying physical violence were measured by questions concerning whether wife beating was justifiable if the wife: a) goes out without telling partner; b) neglects their children; c) argues with her partner and d) refuses to have sex with her partner. Responses to these variables were dichotomous (1 = “yes” or 0 = “no”). Although, some studies have aggregated these measures [29,30], we preferred individual measures of empowerment [29]; our aim was to investigate the contribution of specific empowerment measures in influencing IPSV in the Ugandan context.

As described elsewhere [20], “women’s economic empowerment included occupation and ownership of property (a house). Ownership of a house was recoded into two categories: woman alone/jointly with the partner as the empowered category and partner alone/others as the other. Ownership of a house is included because it is an important and contested asset for women in the Ugandan context” [20].

Partners’ behavior and controlling tendencies were analyzed using three variables: partners’ controlling behavior and alcohol consumption, and the women’s attitude towards their partners. To measure partners’ control, women were asked whether their present partners: a) were jealous if respondents talked with other men; b) accused them of unfaithfulness; c) did not permit them to meet female friends; d) tried to limit respondents’ contact with family and e) insisted on knowing where they were. All these variables had binary responses (0 = no and 1 = yes).

Partner’s alcohol consumption was measured by two questions: a) Does your partner drink alcohol? This was coded as binary outcome (0 = No, 1 = Yes). The second follow up question was asked to those who said yes to drinking alcohol: b) How often does (did) he get drunk: often, only sometimes, or never? The response categories were; 0 = never, 1 = often, 2 = sometimes). We generated a new measure of alcohol consumption from the second question and included those who do not drink alcohol. The new variable had four categories (0 = does not drink, 1 = drinks, but never gets drunk, 2 = drinks and sometimes gets drunk and 3 = drinks and often gets drunk). We also included women’s attitude towards their partners – whether they were afraid of their partners – in this category of variables. Women were asked if they were afraid of their partners. This was categorized as 0 = never, 1 = sometimes and 2 = most of the time.

Women’s socio-demographic characteristics included women’s age group, women’s education level, region, place of residence, wealth index, parity or number of children ever born and current marital status (married or cohabiting). Witnessing of parental intimate partner violence was measured by whether the respondent reported ever witnessing her father beating her mother (with a binary outcome of 0 = No, 1 = Yes). The analysis also included age at first sex and duration of the union.

We used frequency distributions to describe the characteristics of the women. Cross-tabulations with Pearson's chi-squared (χ²) tests were used to examine the associations between IPSV and women’s empowerment (economic empowerment, attitudes justifying physical violence, and decision-making autonomy), partners’ behaviors, and women’s socio-demographic factors. The level of statistical significance using p-values was set at p <0.05.

Finally, we conducted multivariable logistic regression analyses to examine the association between IPSV and the selected explanatory variables whose p-values were less than 0.05 during the chi-square tests. We presented the results in the form of Odds Ratios (OR) and their 95% confidence intervals. The level of statistical significance using p-values was set at p < 0.05. We estimated three models during the multivariable analysis. In the first step, IPSV was modeled with decision-making indicators and attitudes justifying wife beating. In the second model, we added partners’ controlling behaviors and women’s fear of their partners. In the final model, we adjusted for socio-demographic factors. We weighted all of the analyses to take into consideration the survey design, clustering and stratification [26].

From Table 1, 41% of the women were aged 25–34 years. There was nearly an even distribution of women by region, although more than a quarter (28%) were from the central region. Most (84%) of the women were rural residents. Most (70%) engaged in sex before they were 18 years. More than half (55%) were married and had been in a union for over 10 years (55%). Half (50%) of the women reported giving birth to between 1 and 4 children in their lifetime. More than half (60%) had primary education and 41% were from the richer and richest wealth quintiles. Close to half (46%) had witnessed their fathers beat their mothers during childhood.

Concerning measures of women’s empowerment, just over half (53%) of the women were engaged in agriculture, more than half (57%) reported ownership of a house either alone or jointly. Participation in household decision-making among women in a union was as follows: how women’s earnings were spent (44%); women’s own healthcare (59%); large household purchases (56%); visits to women’s family (58%); and how their partners’ earnings were used (46%).

The prevalence of attitudes justifying physical violence varied. Wife beating was justified if the woman: went out without telling her partner (41%); neglected their children (48%); argued with her partner (30%); refused sex with her partner (24%); or burnt food (15%). Just over one in four (27%) women in a union reported having experienced IPSV in the 12 months preceding the survey.

Results of the cross tabulations in Table 1 show that region, age at first sex, women’s level of education, and whether the woman’s father beat her mother were significantly associated with IPSV. Among the measures of empowerment, IPSV was only significantly associated with decision-making regarding the partner’s income, and with attitudes towards wife beating only if the woman neglected their children, argued with her partner, or burnt food. However, some of the key measures of economic empowerment, such as property ownership and decision-making concerning large purchases were not significantly associated with IPSV (see Table 1).

Table 2 presents male partners’ background characteristics, their behavior (controlling tendencies and alcohol consumption) and women’s fear of their partners. Over half (55%) of the partners were aged 35 years and above. Over a third (36%) had secondary or higher education. Concerning partners’ controlling behavior, over half of the women reported that their partners were jealous if they talked with men (56%), and insisted on knowing where they were (55%). Less than one third reported that their partners accused them of unfaithfulness, did not permit them to meet female friends, and limited their contact with family (32%, 26% and 18% respectively). Four in ten (40%) of the women had partners who got drunk and 46% of the women were afraid of their partners.

In Table 2, all measures of male partners’ controlling behaviors, alcohol consumption, and women’s attitudes towards their partners were strongly associated with reporting IPSV. IPSV was higher among women whose partners were jealous when they talked to other men (35%), accused them of unfaithfulness (41%), denied them permission to meet female friends (42%), limited contact with family (46%), and insisted on knowing their whereabouts (33%). IPSV was higher among women whose partners often got drunk (44%) and among women who were often afraid of their partners (39%). Partner’s age and level of education were not significantly associated with IPSV.

Table 3 shows the results of logistic regression of IPSV controlling for empowerment indicators and selected explanatory variables. In the first model, we adjusted for decision-making and attitudes justifying wife beating. Woman’s participation in decision-making concerning the partner’s income was the only significant predictor of IPSV. Women who participated in decision-making concerning their partners’ incomes had reduced odds (OR = 0.63; 95% CI: 0.47-0.86) of experiencing IPSV in comparison to those who did not.

In the second model, we added partners’ behaviors and women’s fear of their partners. Partners’ behaviors and women’s fear of their partner weakened women’s participation in decision-making concerning their partners’ income. Among partner’s behaviors, significant predictors of IPSV were if the partner was jealous if the woman talked with other men, did not permit the respondent to meet female friends, tried to limit the woman’s contact with family, and the frequency of the partner’s excessive alcohol consumption. Whether the woman was afraid of her partner was also significantly associated with IPSV. The odds of reporting IPSV were higher among women whose partners were jealous if they talked to other men (OR = 1.93; 95% CI: 1.28-2.90), did not permit the respondent to meet female friends (OR = 1.53; 95% CI: 1.03-2.27), and tried to limit her contact with family (OR = 1.82; 95% CI: 1.16-2.83). Women whose partners often got drunk had increased odds (OR = 1.80; 95% CI: 1.18-2.77) of reporting IPSV. Likewise, woman who were sometimes or often afraid of their partners were more likely to report IPSV (OR = 1.82; 95% CI: 1.25-2.65 for women who were sometimes afraid, and OR = 1.59; 95% CI: 1.04-2.44 for those who were often afraid).

In the final model, we adjusted for socio-demographic factors. While most of the partners’ behaviors and women’s attitude towards their partners retained their significance, women’s empowerment and socio-demographic variables were not significant. The odds of reporting IPSV were higher among women whose partners were jealous if they talked with other men (OR = 1.81; 95% CI: 1.22-2.68), accused them of unfaithfulness (OR = 1.50; 95% CI: 1.03-2.19) and did not permit them to meet with their female friends (OR = 1.63; 95% CI: 1.11-2.39). The odds of IPSV were also higher among women whose partners tried to limit their contact with family (OR = 1.73; 95% CI: 1.11-2.67) and often got drunk (OR = 1.80; 95% CI: 1.15-2.81). Women who were sometimes or often afraid of their partners had increased odds of reporting IPSV (OR = 1.78; 95% CI: 1.21-2.60 and OR = 1.58; 95% CI: 1.04-2.40 respectively).