Social scientific researchers have provided significant insight into the determinants of family or intimate partner violence (IPV) and how it impacts the socioeconomic and health situation of women, especially those in the developing world (Stark and Flitcraft 1996; Jewkes 2002). While the existing research on this social and health problem suggests that a multiplicity of factors provide the context for the perpetration of IPV, one explanation that has gained some prominence in the literature deals with the role of resources and power in dyadic relationships (Anderson 1997; Dobash and Dobash 1992). According to the resource-based explanation, in situations where men lack the resources associated with their assumed dominant role of the male breadwinner, they are more likely to express their frustration through violence (Gelles 1974). In a similar fashion, it has been observed that if women have little access to resources, or if they have greater access to resources than their partners, thereby usurping the traditional position of men, they will be at a higher risk of being victimized by their male partners (Hindin 2003; Jewkes 2002; Brinkerhoff et al. 1992).

Even though the theoretical literature points to a strong causal link between the victimization of women and their access to resources (Koenig et al. 2003a; Kalmuss and Straus 1982), we know very little about how various types of resources help in shaping attitudes toward intimate partner violence. This is especially the case in sub-Saharan Africa where IPV is believed to be quite widespread (Johnson and Ferraro 2000; Heise et al. 1994b). Besides the reported high prevalence of IPV, many African societies are patriarchal in nature with women having limited resources and power within the home (Takyi and Dodoo 2005), thus making their situation more precarious when it comes to IPV. Given these considerations, this gap in knowledge reduces our understanding of the circumstances behind the practice in non-Western cultural settings, thus hindering the development of preventive strategies to reduce the practice. Not only that, Kishor and Johnson (2004) contend that the rate of violence against women is indicative of the degree of women’s status in a society. According to their thesis, where women routinely experience violence, they are likely to be devalued as well. Under such a scenario, a better understanding of the links between resources and attitudes toward IPV may reveal important new information about the socioeconomic circumstances of women in Africa, which then could help in the development of intervention programs to enhance their status.

Our main objective in this paper, then, is to assess how resources—which we define in several ways—shape attitudes about wife beating (battering), a common form of IPV in Africa. To answer this question we draw on data from the 2003 Ghana Demographic and Health Survey, a nationally-based representative sample that has measures of IPV to test models with resources, household decision making, and cultural processes on attitudes toward wife beating. We focus on Ghana because in this West African country, some earlier studies have reported that more than one third of all women have reported being abused in some form or another (Coker-Appiah and Cusack 1999). Despite this observation, there is a paucity of scholarly research on the topic, especially at the national level (for an exception, see Ofei-Aboagye 1994). Although we acknowledge that the use of attitudinal measures has its limitations (e.g., attitudes can change over time), we argue here that these measures do offer some insights into the links between resources, gendered dynamics, cultural processes, and intimate partner violence as pertains to the African context.

Background and Theoretical Framework

Social scientists have long noted that the African family is at the epicenter of its social institutions and organizations, and that the family is the main provider of economic and non-economic support to its members (Bledsoe 1990; Lesthaeghe 1989). Even though the family is considered the bedrock of African social life, and references have often been alluded to the high value Africans place on family and married life (e.g., Caldwell 1982), some researchers have also noted another side of the family, one that is considered a private matter and not often discussed in public: violence among intimate partners or family members (Fikree and Bhatti 1999; McWhirter 1999).

On the issue of intimate partner violence, a number of studies from the sub-Saharan African region have noted that violence against women by their partners or relatives (as well as other forms of family violence such as child abuse) is quite widespread (Coker-Appiah and Cusack 1999; Hindin 2003; Hindin and Adair 2002; Jewkes et al. 2002; Kim and Motsei 2002; Koenig et al. 2004; Maman et al. 2000; Nasir and Hyder 2003; Ofei-Aboagye 1994; Watts and Mayhew 2004). While it is true to suggest that the majority of studies on IPV in Africa are from one single country alone, South Africa (see for example, Abrahams et al. 2004; Jewkes et al. 2002). And as Bowman (2003) points out, most of the writings about the subject have been done by activists rather than academicians, scattered reports from other countries in the region suggest that the problem may be acute throughout the region.

Some studies have reported, for example, that about half of all ever-married women in Zambia, 46% in Uganda, 60% in Tanzanian, 42% in Kenya, and a high of 81% in Nigeria have experienced some form of violence in their lives (Heise et al. 1999; Kishor and Johnson 2004). In Sierra Leone, Coker et al. (2002) and Coker and Richter (1998) observed that about 67% of women in that country report being abused by a boyfriend or husband, while another 74% reported abuse by another family member. Not only is IPV prevalent in Africa, it has also been suggested that the practice afflicts all segments of the population and often transcends ethnic, religious, and social class boundaries.

Researchers point to an array of factors in explaining intimate partner violence in the developing world, and especially Africa. Among the causal mechanisms that have been identified as predictors of IPV in the region include the following: (1) women’s lack of resources thereby creating a dependency relations between them and their partners (Jewkes 2002; Koenig et al. 2003b; Oropesa 1997), (2) cultural ideologies that put women in subordinate positions (for a review, see Jewkes 2002; Olson et al. 2005), (3) gender inequities, female subordination, oppression, and empowerment (see Jejeebhoy 1998; Jewkes 2002; Koenig et al. 2003b; Mason 1987; Oropesa 1997).

A key argument behind the resource-based thesis, which forms the basis of our analysis, is that women’s financial independence and autonomy provide some “cushion” or a form of protection and leverage against abuse by their partners. Furthermore, those who subscribe to the resource-based argument contend that, the availability of resources (especially for women) reduces or alters the dependency relations that exist between men and women, and thus reduces men’s undue advantages over women in the domestic realm. This expectation, some have suggested, may only be true as long as women’s levels of resources do not exceed that of their partners (Brinkerhoff et al. 1992; Hindin 2003; Jewkes 2002). This observation has led to the suggestion that there is a curvilinear relationship, rather than a linear one, between female autonomy and abuse (Hindin and Adair 2002; Jewkes 2002).

Even though the resource-based theorists often focus on economic indicators-measured by such common indicators as employment, relative earnings, and education, the extant literature on women’s autonomy (status) also indicate that the resources that accrue to women in marital relationships can manifest themselves in other ways (Hindin 2003; Hindin and Adair 2002; Mason 1987; Oropesa 1997; Presser 1997). Indeed, as the research on gender and family relations have found, the decision-making process, and the type of decisions a partner makes is theoretically speaking, an important resource which can affect a host of family processes. Moreover, while the literature on women’s empowerment does not clarify the nature of women’s control over decisions and the type of decisions that could have an impact in altering attitudes toward IPV, some researchers have found male control of household decision making to be a good predictor of partner violence (Hindin and Adair 2002; Oropesa 1997).

Violence, some researchers have also noted may occur when women attempt to control some of the decisions (e.g., economic decisions) that are not normatively perceived to be in the realm of women’s control (Ezeh and Gage 2000). In a related note, social demographers, family and gender scholars have also argued that most of Africa operates in a patriarchal culture that upholds the secondary or subordinate status of women in male-female relationships (e.g., Ezeh and Gage 2000; Takyi and Dodoo 2005). This idea of submissiveness is believed to be very pervasive in the region to the extent that it affects the context and type of decisions women and men make within the home. Thus, Hindin and Adair (2002) report that the greater the number of decision-making domains that men dominate, the more likely they are to use intimate partner violence. However, the relationship between decision-making patterns and domestic violence is more complex than this. Indeed, the same study found that when women dominate household decisions, they also are more likely to experience intimate partner violence, thus lending support to the notion that violence may be a response to a man’s feeling of powerlessness. Other studies have suggested that in settings where men are expected to be dominant in the family, but lack educational and social resources or occupational prestige, they are more likely to choose violence as a means of maintaining their dominant position in the family (Jewkes and Abrahams 2002).

Jewkes (2002) and many others have also suggested that violence occurs when men are unable to live up to the socially presented ideas of what it means to be a “successful” man (e.g., Gelles 1974). In a context where men are unable to fulfill the instrumental roles assigned to them by society, they feel less secure, or feel that their position is threatened. Their marginality and insecurities, thus, increase men’s propensity to use abuse as a mechanism to maintain their masculinity and male identity. Not surprisingly, Oropesa (1997) found that wife abuse was higher in families with both spouses unemployed than in other types of families.

Besides resources, a considerable body of the literature on intimate partner violence in Africa emphasizes the close connections between violence and culture (Straus et al. 1981; Takyi and Mann 2006; Ucko 1994). Even though the cultural explanation takes many forms, a key aspect of this thesis holds that in male-dominated Africa, culture provides the social context for the perpetration of abuse; especially that committed against women and children. Rooted in the cultural argument is the notion that societal norms of socialization provide the justification for the oppression, exploitation, and abuse of African women and their children (Walby 1990). Thus, culture, some scholars have argued, permits men to treat women as their own property- submissive, and passive-thereby reducing their bargaining power within a marriage (Bowman 2003; Rude 1999).

Another manifestation of the cultural ideology, and how it may serve as a catalyst for spousal abuse, stems from the practice of bride wealth (dowry) that is often exchanged between the man’s family and that of his future wife prior to marriage. These payments are viewed in some African societies as legitimating the union between the two families (Assimeng 1981), but has the tendency to provide the “legitimation” to some men that they “own” their spouses after these transactions have occurred, thereby creating an environment of female subordination (Heise et al. 1994a). In addition, the payment of bride wealth may at times make it difficult for women to leave an abusive relationship, particularly among groups that have the obligation to return or refund the bride wealth to the husband’s family when divorce occurs (Takyi 2001). This may be the case in situations where these transactions involve the exchange of substantial resources from the part of the husband and his family, as is the case among some patrilineal groups in Ghana.

Another cultural explanation has to do with the nonperformance of marital duties and obligations. According to this paradigm, the incidence of intimate partner violence is intertwined with “traditional” gender roles and their performance. This occurs in many African countries (including Ghana), as traditional gender roles and socialization patterns implicitly or explicitly dictate what men and women do and how they behave (Shettima 1998). In most cases, men are socialized to be the main breadwinners with women providing supporting and nurturing roles (e.g., child care, food preparation, household labor). Thus, it has been argued that many aspects of married life in Ghana, including decision-making authority, reproductive behavior, responsibility for care giving for themselves, partners, and family members, as well as the control over economic resources, is gender stratified (Adomako Ampofo et al. 2004; Takyi and Dodoo 2005). Others point to the normative practices surrounding the operation of the family such as child care and child neglect, and the contradictory gender role expectations for the persistence of abuse in the region (Ucko 1994). It is, therefore, not surprising that Koenig et al. (2003b), reported in their study of rural Uganda that the reasons often cited by those studied in explaining marital violence included the neglect of household chores, disobeying husbands/elders, and withholding sex from partners.

In many ways, another cultural variable that could have some effect on IPV is religion, as it provides a belief system that helps to maintain the existing status quo. For example, in a context where some religious people interpret the scriptures (Bible or Koran or tradition) literally, women may be socialized into believing that “submission” to their male partners is the accepted order of the day. Not all researchers, however, agree that religion provides the context for the perpetuation of intimate partner violence. In fact, some studies have found religion to have a protective factor in reducing abuse. The argument here is that by providing the social support needed for family functioning, plus the fact that religion enhances social integration, limits couples’ isolation, and/or alcohol and substance use, it helps in reducing some of the cofactors that lead to abuse in the first place (Ellison and Anderson 2001). Despite these competing findings, it has been suggested elsewhere that the role of religion has largely been ignored in the empirical studies on IPV (e.g., Ellison et al. 1999; Nason-Clark 2004).

It is clear from our review of the literature that resource-based factors could influence attitudes on intimate partner violence. However, despite this possibility, to date few studies have specifically examined how different types of resources in conjunction with other cultural factors, help in shaping attitudes toward IPV in Africa. Thus, we test several hypotheses in an effort to fill this gap in knowledge and to deepen our understanding of attitudes toward intimate partner violence in Africa, with a focus on Ghana. Given our review of the literature we propose the following hypotheses for verification:

  1. 1.

    Couples who make joint decisions will be less likely to be supportive of abuse than those who do not make such decisions

  2. 2.

    Couples who make equal financial contributions to the household will be less likely to believe that abuse is justified than their counterparts who do not make equal contributions to the household budget.

  3. 3.

    Respondents who hold traditional religious beliefs will be more supportive of abuse than those who do not hold such beliefs.

Data and Methods

We tested our hypotheses on views towards intimate partner violence with data from the 2003 Ghana Demographic Health Survey (GDHS03). GDHS03 is a large, nationally representative dataset, which was originally collected by the Ghana Statistical Service, Macro-International, and the U.S. Agency for International Development to examine demographic and health issues in Ghana. GDHS03 is part of the Global Demographic and Health Surveys (DHS) program that has been conducted in several developing countries since the 1980s. GDHS03 employed a two-staged stratified sample frame and systematic sampling to identify 412 enumeration areas (EAs) and then households throughout the country. From the selected households, 5,691 randomly selected women aged 15–49, and 5,015 and men between the ages of 15–59 were interviewed, yielding a response rate of 95.7% for the women and 93.8% for the men. Our study uses the results from the 2,133 married couples who were interviewed in the 2003 survey. To protect the confidentiality and privacy of those questioned, the respondents were interviewed separately (without their partner present). The analyses we performed was done separately for men and women.

The usefulness of these data for our analyses is that, unlike the previous surveys that have been conducted in Ghana since the 1980s, it has detailed information on various aspects of gender relations and also questions about intimate partner violence. These data, plus other measures of age, marital status, parity, education, employment, information about how decisions are made in the household, as well as how much each of the partners contributes to the household budget or expenditure, present new opportunities for the analysis between different aspects of resources and attitudes toward intimate partner violence.

Even though the GDHS03 has detailed information on IPV, we must note that the measures included in the survey are not without their limitations. For the most part, the questions on domestic or intimate partner violence deal with attitudinal as opposed to actual behavioral measures. For this reason, it is not clear as to whether the respondents would actually react to what they believe in or not when it comes to actual behavior. Furthermore, people’s attitudes are not static but dynamic, and can change as one goes through the life course. Additionally, measures of alcohol usage, which have been shown to be a factor in the prevalence of IPV by several scholars (Busby 1999; Campbell 2002; Ellsberg et al. 2000; Hindin and Adair 2002; Jewkes et al. 2002; Nasir and Hyder 2003) are not included in our analysis since they were not measured in the survey.

Measures of Intimate Partner Violence

The dependent variable used in the study was constructed from five questionnaire items that capture different views and aspects of intimate partner violence in Ghana. The respondents were asked, “Sometimes a husband is annoyed or angered by things that his wife does. In your opinion, is a husband justified in hitting or beating his wife in the following situations?”

  1. 1.

    If she (wife) went out without telling the husband

  2. 2.

    If she (wife) neglected the children

  3. 3.

    If she (wife) argued with the husband

  4. 4.

    If she (wife) refused to have sex with him (husband)

  5. 5.

    If the wife burnt the food.

The original responses to these five items were coded into “yes”, “no”, “don’t know”, and “no answer.” There were a minimal number of “don’t know” respondents (34 men and 27 women) and none marking “no response” to any item. Because of this, the responses for these two categories (don’t know and non-response) were excluded from the study. This reduced the male sample size to 2,099 and the female sample size to 2,106.

The responses to these five items were dummy-coded (“yes”, “no”) and used to construct an index to tap the circumstances under which wife abuse was considered acceptable by our male and female respondents respectfully (alpha = .809, alpha = .828). The result of this transformation was subsequently recoded into a dummy variable with “1” implying acceptable in any of the five domains, and “0” unacceptable in any scenario. The dependent variable used in the multivariate models measures if abuse is acceptable under any of the five circumstances. Therefore, it does not distinguish between abuse scenarios.

Independent Variables

Several resource-based variables were included in the study. They include financial income contributions, relative household resources, and indicators of household wealth. We defined household contribution as the proportion of total household expenditure contributed by the respondent. The exact question here asked: “On average, how much of your household expenditures do your earnings pay for: almost none, less than half, about half, more than half or all?” This may reflect the proportion of resources the respondent feels he/she has access to, in relation to a partner. Although not a dollar amount, it does indicate resource differentials between men and women, which are relevant in the argument that a lack of resources promotes IPV (Jewkes 2002). We coded this variable into the following categories; less than half (includes those who responded as almost none), about half, more than half, and all. This was done due to the minimal amount of male respondents who chose almost none.

Household decision making (an aspect of gender and power dynamics) is a variable we created from responses to the GDHS03 question that asked: “Who do you think has the final say in the following five aspects of household life?”: (1) health care, (2) making large household purchases, (3) making household purchases for daily needs, (4) visits to see family or relatives, and (5) on the food to be cooked each day. We recoded the responses into the following: the wife alone, the husband alone, and jointly. A count was then generated to calculate the number of decisions made across these five domains (Cronbach alpha = .799 for women and .750 for men). Although this measure allows for a glimpse into the proportions of decisions made through egalitarian and non-egalitarian ways, it does not allow us to separate out responses to particular domains.

Other resource-based measures considered include household wealth; a variable based on the GDHS03 constructed Wealth Index, which was created from responses to questions on the type of materials used in constructing their house, source of drinking water, main cooking fuel, toilet facilities, and access to electricity, radio, TV, refrigerator, telephone, and means of transportation. Responses to these items were then used to delineate the socioeconomic status of the household as follows: poorest, poorer, middle, richer, and richest. In addition, we also considered the respondents’ education (no schooling, elementary/primary, secondary or above); and, occupation (professional, managerial, and technical (PMT), agricultural, manual skilled/unskilled labor, and the unemployed).

Based on previous findings we included controls that consider the role of culture and basic socio-demographic characteristics of the respondents in our models. The cultural variables used in the models are ethnic identity and religion. We defined ethnicity in terms of affiliation to the five major groups in Ghana: Akan, Ga-Adangbe, Ewe, Mole-Dagbani, and Other. Of these groups, the Ga-Adangbe, Ewe, and Mole-Dagbani are patriarchal, while the Akan are matriarchal. Religion was coded into the following categories: Catholic, Protestant, Other Christian, Muslim, Traditional, and no declared affiliation/no religion. Additionally, household contribution and household decision making do have cultural components and were included in the analysis from both a resource as well as cultural perspectives. The socio-demographic characteristic variables included the respondents’ age, children ever born, and place of residence (urban vs. rural).

For our multivariate analyses, we used the logistic regression models in estimating our equations. We used this approach because our outcome variable is measured as a dummy variable. Three main models were estimated to evaluate our hypotheses. The first evaluated the role of economic resources alone, the second included decision making context, an aspect of resource in the original model, while the third (full model) include cultural and other demographic factors in our equations. The estimated coefficients from the logistic models were exponentiated to provide the more readily interpretable odds ratios, which we report in our final tables.

Results

Table 1 provides descriptive statistics for the variables used in the study. As can be seen, there are some differences between the men and women sampled. For example, the women tended to be younger, had fewer children, and lower levels of education than the men counterparts. The sample also included more rural than urban respondents. In addition, a sizable proportion of those interviewed were from the Akan and Mole-Dagbani ethnic groups. A particularly interesting difference between the two groups has to do with their perceptions. Since the group we studied consisted of married men and women, we expected that if the man contributed all or more than half of the money used for household expenses, by default the woman should have also contributed less than half of the household budget. If that is the case, then the two combined should not equal more than all, if the answers given were factual. Yet what we observed from our data is that these numbers do not add up, meaning that men and women have different perceptions of what they contribute financially to the household budget. For example, while 22% of the women stated that they contributed over half of the household budget (resources), only 11% of the men stated that they did so. Thus, it appears that the men, women, or both, were over-reporting what they actually contributed to the household budget. A similar pattern is discernible with regard to the decision-making variables where women tended to believe that egalitarian decisions occurred more often than reported among the men. What may be the most startling finding from Table 1 is that women were more supportive of abuse than their men counterparts.

Table 1 Overall distribution and percentages of couples in Ghana who believe that intimate partner violence is justified by contribution, decision-making, demographic and control variables: GDHS2003
Full size table

In examining the views or attitudes of our respondents toward the various forms of abuse identified in the survey (Table 2), we find a larger percentage of women to believe that abuse is justifiable in all of the five situations we examined. For example, while 12.5% of the men felt it was justified to beat their wife if she refused sex, the comparable figure for the women was almost twice that of the men: 26.6%. Interestingly enough, about two-thirds of the men (64.2%) felt that abuse was never justified. In contrast, less than half (44.6%) of the women thought abuse was unjustified in any situation whatsoever. Indeed, in comparison to the men, the women respondents were more likely to believe that abuse was justified; a pattern that is true for the most part in every scenario we examined. This observation suggests to us that that married women in Ghana tend to believe that abuse is justifiable far more often than their husbands.

Table 2 Beliefs and attitudes towards domestic violence among couples in Ghana: GDHS2003
Full size table

Table 3 reports bivariate associations between one of our key resource variables-decision making-and attitudes toward wife beating. When men feel their wives have the final say, the less likely they are to feel that abuse is justified. Indeed, the percentage that believes abuse is justified drops substantially from 38.4% when the wife makes no decisions, to 12.5% when she makes all of them. The same trend is not apparent with the women when they perceive themselves as having the final say over decisions. Overall, both men and women tend to believe that abuse is justified; with 47.7% of the men and 60.6% of the women believing it is allowable, when the husband always has the final say. When household decisions are made in an egaliatarian way, men and women were less likely to believe that abuse was justified (Panel III).

Table 3 Decision making and attitudes towards domestic violence among couples in Ghana: GDHS2003
Full size table

Table 4 reports the multivariate findings from our logistic regression models on resources and non-resource based factors on attitudes toward wife beating or battering. Findings from Model 1 that examine the direct effects of perceived household contribution and attitude toward IPV suggests that the relationship between the two processes are curvilinear for the wife’s financial household contribution and her acceptance of abuse. In situations where the couples contribute equally to the household budget, women are less likely to believe that abuse is justified (OR = .422 p < .001). This association between resources and attitudes is not that strong among men, with only those contributing less than half of the household budget reporting any significant change in their attitudes. Such an observation lends support to the idea that when a man does not believe that he is living up to his role as the breadwinner, he is more likely to find IPV acceptable.

Table 4 Logistic regression estimates (odds ratios) of husband’s and wives’ attitudes about IPV on contribution, decision making, demographic and control variables within Ghana: GDHS2003
Full size table

In Model 2, we introduced the household decision-making variables and observed that when decisions are made equally (egalitarian decision making) within the home, both men and women were less likely to believe that abuse was justified (Men OR = .788, p < .01; Women OR = .840, p < .05). Such a finding is consistent with our observation in the bivariate analysis reported in Table 3. Also, when the wife is seen by the man as making sole decisions within the home, he is significantly less likely to believe that abuse is justified (OR = .761, p < .001).

With the inclusion of the socio-demographic variables (Model 3), a number of key findings emerge. First, we find relative resources to be significant in shaping the respondents attitudes towards IPV. This is more so with women than men. Women who contribute half or less of the household budget tend to believe that abuse is justified less often than those who contribute everything. This is most noticeable among the women who contribute equally to the household budget, with their odds of believing abuse is justified being 40% less than that of those women who contribute everything (OR = .582 p < .001). The opposite is true among men, where the odds of believing abuse is justified among those men who contribute less then half were almost 50% higher than those who contribute everything to the household (OR = 1.490, p < .05).

Second, decision making has a significant effect in shaping men’s attitudes toward abuse. When men perceive that decisions were made jointly or solely by their wives, they were less likely to be supportive of abuse (OR = .910 p < .001; OR = .846 p < .05). This suggests that when men perceive their wives as being capable equals, or have some autonomy, they are more likely to believe that abusing them is not a justifiable form of action. Surprisingly, this observation was not seen among the women, with household decision making not being a statistically significant predictor of their views toward abuse.

Education appears to be a major shaper of attitudes toward gender equality and worth, with both the men and women who had completed higher education less likely to believe that abuse was justifiable in any situation (OR = .283 p < .001; OR = .328 p < .05, respectively). Household wealth also had an effect on views toward abuse as the wealthier an individual was, the less likely he/she believed abuse was justifiable (Richest versus Poorest; OR = .446 p < .01; OR = .295 p < .001, respectively).

Aside from the resource measures, we also found ethnicity to be associated with attitudes toward IPV. Compared to the matrilineal Akan, the patrilineal groups were more likely to be supportive of abuse (e.g., Mole-Dagbani Men: OR = 1.540, p < .05, Women: OR = 1.941, p < .001). This finding seems to support Morrow’s (1986) observation that the presence of matrilineal ties helps to improve women’s power in marriage; based in large part on the social networks and ties these women derive from their kin or matriclan. Surprisingly enough, neither religion nor occupation significantly influenced men’s or women’s attitudes towards wife battering. Likewise, place of residence did not seem to play a significant role in shaping attitudes about IPV.

Discussion

Data from the 2003 Ghana Demographic and Health Survey, a nationally representative sample, have been used to explore the forces that help in shaping attitudes toward intimate partner violence, a major social and health problem facing many sub-Saharan African nations. In particular, we assessed whether access to resources affect the respondents attitudes toward wife battering (beating), a common form of intimate partner violence in the African region.

Findings from the study support our first hypothesis that egalitarian household decision making reduces abuse-supporting ideologies. This seems to be the case among husbands more so than their wives. We found somewhat limited and mixed results for our second hypothesis on the level of household contribution and attitudes towards abuse. The effect of perceived household contributions differed in how they impacted male and female ideologies toward abuse. These findings suggest that, to some extent, male and female notions about intimate partner violence may be shaped by different mechanisms. The findings on our other hypothesis, that traditional religious believers would be more supportive of abuse than their non-traditional counterparts, was also not supported.

Overall, the multivariate findings provide some support for the resource-based arguments in shaping attitudes towards wife battering in the context of Ghana, a finding that is consistent with some recent studies in other parts of the developing world as well (e.g., Yount and Carrera 2006). Indeed, for the most part, most of our key resources variables we considered (household decision-making, household contribution, and household wealth) were significant in predicting abuse shaping ideologies in Ghana. In contrast to the economic indicators, several of the cultural variables included in our models produced little or mixed results. For example, we did not observe any significant difference between those with traditional religious beliefs from those without. Despite these observations, our bivariate results lead us to suggest that there are several interesting aspects of the cultural arguments that should be further investigated. For example, even though the logistic regressions showed no significant relationship between traditional religion and abuse, net of the other variables, the bivariate analysis showed that an exceedingly high number of them (75.6% of women, 54.4% men) supported abuse. While this may appear to be statistically non-significant, public health workers in the field would still be well advised to focus on this cultural enclave.

We find some evidence in support of Gelles’ (1974) theory about identity which suggests that when the man’s identity is challenged, he is more likely to lash out at his significant other in order to maintain a semblance of power. Indeed we observed that when men were not the primary breadwinner, they were much more likely to support abuse. On another level, the notion that men believe IPV is justified because of societal or cultural circumstances, which place men above women and therefore create a “culture of abuse,” may not be the case. Our study calls the belief that abuse is socially acceptable by men into question. In some cases, the man’s behavior may actually be viewed as unacceptable by a majority of men within society. This perspective changes the issue from one where abuse by a man is painted broadly as being acceptable within a society, to one where such actions might latently be considered deviant. This raises a number of key questions such as the following: why are men, if a majority of them are opposed to it, not vocalizing their dissent? What are inhibiting men from transitioning from passive roles to active ones within the discourse on abused women?

In addition to men’s perceptions of self-worth and identity, women’s ideas should also be considered in the theoretical discourse on resources and abuse. While it is difficult for a married woman without resources to leave the home and stand on her own if abuse is occurring, any actions or inabilities to act are first predicated on her understanding and belief that what is occurring is not appropriate. If a woman does not believe that abuse is wrong, then her ability to do something about it matters little. This study shows that a majority of women, more than 55%, believe that abuse is justified in at least one of the situations examined. Moreover, more than 60% of those women believe it is acceptable in a majority of the described situations. Women were also more likely than men in every single situation we investigated to support abuse. The underlying assumption present in many of the theoretical arguments is that if women had power, status, or financial equality with men, they will experience less abuse. It appears that such a generalization may not be entirely true in some cases as our data from Ghana indicates. To that extent, there is the need to change women’s views also as one way to lessen or reduce the incidence of abuse.

At first glance, it appears as though the male and female perceptions of abuse are shaped by different mechanisms. For example, while men’s attitudes were often shaped by decision making, that of women tended to take the form of household contributions. When the man perceives his wife as being equal or capable to make decisions for the family on her own, he is also less likely to believe abusing her is justified. When the wife believes that she is a contributing member of the household, she is less likely to believe that abuse practices are acceptable. This is most notable when she views her financial contribution to the home as being equal to her husband’s contribution. This analysis supports a much larger role for perceptions of value, self-worth, and identity than has been previously suggested. Measures of self-worth and perceived spousal worth would be a useful addition for future surveys in order to definitively support or reject this idea.

Finally, we argue that education may be an important variable that needs to be emphasized in campaigns against violence in Ghana. This is especially the case since our data show that education, especially higher education, was associated with reductions in abuse supporting ideologies. This is especially the case among men. Not only that, education in the form of programs aimed at influencing household decision-making processes, or of establishing the true value and worth of women within the minds of men and women alike might be useful in fighting abuse in the context of Ghana, and perhaps other African societies.

Future research is needed to tease out what is occurring within the traditional structures of African society and how they impact attitudes toward abuse. For example, untangling the intersection between traditional religious beliefs and values, and also ethnicity, and resources would be helpful to the discourse on IPV in Africa. Research also needs to be conducted to assess the link between attitudes toward abuse and the actual prevalence of abuse. This study shows that male and female perspectives are intrinsically linked to their belief structures about IPV. If these structures are to be influenced, then critical evaluations need to be made about the underlying factors, which are shaping these perspectives.