Correlates of Non-Partner Physical Violence among Ethiopian Women

: Intimate and non-partner violence against women are major societal and public health problems afﬂicting many women around the world. Unlike intimate partner violence (IPV), non-partner violence (NPV) has not been well researched in Ethiopia. Thus, individual, physical, social, and environmental factors associated with NPV are largely unknown. For the study, we used national data collected in the 2016 Ethiopia Demographic and Health Survey. The survey sample was selected with two-stage stratiﬁed cluster sampling. Multivariate logistic regression model was ﬁtted to understand the association between NPV and its correlates. Overall, 45% of the women reported having experienced physical NPV in Ethiopia. Women who owned assets, solely or jointly (land, house, or both), were 70% less likely to report NPV than women who had no such assets. Thus, structural interventions that integrate women’s asset ownership and provide special microﬁnancing services for women along with community-based gender education programs may have a potential to reduce violence against women in Ethiopia. Access to improved cooking fuel included electricity, natural gas, LPG, kerosene, biogas as opposed to fuel from wood, charcoal, crop residues, animal dung.


Introduction
Violence against women, a gross violation of fundamental principles of human rights, is a major public health and societal problem, afflicting many women around the world [1]. Worldwide, 35% of women have reported at some point experiencing sexual/physical violence by intimate partners or sexual violence by non-partners [2]. In 2017 alone, 87,000 women globally were intentionally murdered by their intimate partners or other family members. In the same year, 20,000 women were killed by family members other than their intimate partners [3]. Of all women killed by intimate partners or other family members in 2017; 20,000 of the deaths occurred in Asia, followed by 19,000 deaths in Africa. The lowest numbers were in Europe and Oceania, with 3000 and 300 homicides, respectively, of women by intimate partners or other family members [3].
Although violence against women can take many forms, interpersonal violence against women is generally categorized into intimate partner violence (IPV) or non-partner violence (NPV) [1]. The former is violence that occurs within the context of intimate relationships [4]; the latter is violence by other family members (parents, sons or daughters, siblings, other blood relatives, or affines) or by strangers or other persons familiar to the victims. Both forms of violence are linked to long-lasting and adverse health outcomes that can impact the health and wellbeing of women [1,5]. Owing to its pervasiveness across diverse human societies [3], IPV is the subject of the vast majority of research reports on violence against women from around the world [4,[6][7][8]. Even though NPV is equally pernicious to women's health [1], it is less researched, especially in Africa [9,10]. The prevalence of non-partner sexual violence in Africa is estimated to be 12%-the highest rate among low-and middle-income countries [1]. In Ethiopia, one in four (24.9%) women reported having experienced physical violence by non-partners [11]. For example, simply accessing common household essentials such as water, sanitation, and family energy needs could expose women to NPV in Africa [12,13], as many women in the region often travel long distances to fetch water and firewood for their households, risking sexual or physical violence along the way [12,14,15]. Ethiopia is predominately an agrarian society with 80 percent of its population residing in rural areas. Although women make significant labor contributions to agricultural production, access to family resources and community participation is often limited. Furthermore, women also experience physical, sexual, or emotional violence [16]. For example, in 2019, Ethiopia was ranked 125th out of 162 countries in the Gender Inequality Index (GII), and in the years, 2015-2019, only 11.5% of women had at least some secondary education, while the figure was 22.6% for men [17]. Ethiopia is a patriarchal society in which cultural values and practices undergird male dominance [18], and women still face significant inequities, including limited opportunities for development, education, and employment [19].
In general, violence against women stems from societal norms of acceptance of violence and unequal gender power relations [8] that are maintained by a system that is biased against women [20]. However, in Africa, including Ethiopia, factors associated with NPV are not fully understood, as research and information on NPV are limited [9,10].
Addressing gender-based violence may also contribute toward achieving sustainable development goals [21]. Ecological models are often used to explain multiple determinants of violence against women [22]. Thus, identifying exposing and preventive factors could help develop a thorough understanding of NPV correlates in the Ethiopian context. Therefore, the purpose of this study is to identify individual, social, and environmental factors associated with women's experiences of physical non-partner violence in Ethiopia. We found no prior studies that specifically examined non-partner violence among Ethiopian women. The study, beyond filling a knowledge gap in this respect, may also contribute towards understanding and preventing violence against women.

Sample Design
For the current study, we used data collected in the 2016 Ethiopia Demographic and Health Survey (EDHS-2016). A two-stage stratified cluster sampling was used to select the survey sample. In the first stage, a total of 645 enumeration areas were selected (202 from urban areas and 443 from rural areas) by using the proportional probability sampling method. In the second stage, an equal probabilistic systematic selection was applied [11]. For the study, we used data from women (n = 378) between 15 and 49 years of age who gave responses to the non-partner physical violence question. The survey procedure and protocol were reviewed and approved by the Federal Democratic Republic of Ethiopia's Ministry of Science and Technology and the Institutional Review Board of ICF [11]. In collecting data about violence against women, safety recommendations of the World Health Organization were followed [11].

Explanatory Variables
In ecological models, multiple factors are assumed to be the determinants of human health [18]. It is thus anticipated that non-partner physical violence is to be associated with the following factors: (1) individual-level factors (women's age, religious affiliation, household wealth, and behavioral factors such as women's alcohol use); (2) environmental factors that could expose women to non-partner violence, including physical environments such as residence type, access to drinking water on the premise, access to improved toilet facility, and access to cooking fuel; (3) environments protective against non-partner physical violence, including women's empowerment variables (education levels, employment status, and women's asset ownership) and access to communication and awareness. The coding of variables is shown in Table 1. Not at all =0; often or sometimes = 1 † Categories based on women who had access to drinking water on the premise and spent no time in accessing drinking water were categorized in the "yes" group and those who spent any time in accessing drinking water grouped as the "no" group. ‡ Access to flush toilets as well as various forms of pit latrines and other forms of human waste disposal facilities was grouped as "yes", as opposed to open field/bush defecation, categorized as "no". * Access to improved cooking fuel included electricity, natural gas, LPG, kerosene, biogas as opposed to fuel from wood, charcoal, crop residues, animal dung.

Outcome Variable
The outcome variable reflecting women's experiences of non-partner physical violence was captured by assessing women's responses to this question: What is the frequency of being hit in the last 12 months by someone other than your husband/partner? Multiplechoice responses to the question were: (a) not at all, (b) often, or (c) sometimes. Women who responded "not at all" were coded "0", and those who answered "often" or "sometimes" were coded "1." See Table 1.

Data Analysis
Variables in the model were screened for multicollinearity. Variance inflation factors (VIF) showed no evidence of multicollinearity among variables (VIF range: 1.091-2.069). We analyzed the data by using the module for complex data analysis in SPSS v25.0. Domestic violence sampling weight was applied, and complex survey design effects arising from cluster and strata were accounted for. Bivariate associations were tested by chi-square statistics. In the multivariate analysis, we regressed the outcome variable on the explanatory variables using a logistic regression model. A p-value of less than 0.05 was used to determine statistical significance.

Results
Overall, 44.7% (95% CI: 36.3-53.4) of the women reported having experienced nonpartner physical violence in the prior 12 months. One in 10 of the women (11.4%; 95% CI 7.4-17.3) experienced non-partner physical violence "often", while one in three (33.2%; 95% CI 25.8-41.6) reported having experienced non-partner physical violence "sometimes" in the last 12 months. At the bivariate level, women's religious affiliations, women's access to improved cooking fuel, and women's asset ownership were significantly associated with women's experience of non-partner physical violence. Women's household wealth and alcohol use were marginally associated with non-partner physical violence at bivariate level ( Table 2). In multivariate analysis, however, only women's household wealth and asset ownership associated significantly with women's experience of non-partner physical violence in the previous 12 months. As compared to women from poor households, women from middle-income households were 81% (OR = 0.189; 95% CI: 0.052-0.692) less likely to report non-partner physical violence. However, there was no significant difference between women from poor and rich households. Women who reported owning assets solely or jointly (land, house, or both) were 70% (OR = 0.299, 95% CI: 0.149-0.600) less likely to have reported non-partner physical violence in the prior 12 months than women who had no such assets (Table 3). Table 3. Adjusted odd-ratios (OR) and 95% confidence (95% CI) for non-partner physical violence by hypothesized exposing and protective factors.

Discussion
Nearly 45% of women (aged 15-49) reported having experienced non-partner physical violence in the previous 12 months with 11.4% reporting having been hit often by a person other than their partner in the prior year. Estimates of comparable national non-partner physical violence among other African women are unavailable. However, prevalence of non-partner physical or sexual violence in 20 African countries varied between 2.3% and 11.3% [9]. In the neighboring country of Somalia, 16.5% of women reported having experienced physical or sexual violence in their adult lifetime [17]. For example, while a decreasing trend of intimate partner violence against women was observed in Italy [23], violence against women has increased across Africa [24]. Our findings suggest that nonpartner violence among Ethiopian women is a widespread phenomenon. Regardless of the perpetrators, violence against women can have multiple adverse health outcomes, including fatal, physical, sexual, psychological, or behavioral consequences [25]. The risk of violence against women is greatest in patriarchal societies like Ethiopia [26] where violence against women and others is tolerated and gender norms and inequities put women in a subordinate position [8,27].
In general, violence against women does not happen in a vacuum, but rather occurs in community contexts because harmful social norms can sustain gender-based violence [28]. Thus, community level prevention programs targeted at altering risky environments can be effective in reducing violence against women [29]. Threats and violence against women in the community or at home could impose an aura of fear and insecurity that can limit women's participation in economic, social, and political activities [29]. In addition, violence and harassment, place significant constraints on women's freedom of movement and access to resources [30]. Beliefs, traditions, and norms about gender relations in societies operate within communities and therefore are considered significant barriers, hindering women access to justice and protection from violence. Addressing these barriers is vital for community-level interventions to have a successful impact [29]. Changing community norms that sustain violence requires the engagement of the whole community. The first step is to create enabling environments that empower women to resist and challenge violence [30]. Collaborative community programs that provide opportunities for women to come together and develop their self-confidence, increase awareness, leadership skills, and build social support networks are thus essential [30].
In multivariate analysis, household wealth was significantly associated with physical NPV. Although women living in middle-income households were less likely to experience physical NPV than women from poorer households, there was no difference between women from poor and wealthy households, suggesting the relationship was not linear. It may be that poverty is related to stress, which may induce violence [8], and better resources among women may cause a backlash that could lead to more abuse [31]. Similar results of non-linear relationships of violence and wealth have been reported among diverse African women populations [32,33]. For example, Bamiwuye and Odimegwu [32], who analyzed data from several African women populations observed that women from rich households in Zambia and Mozambique were more likely to experience violence than women from poor and middle-income households. In Nigeria and Cameroun, women from middle-income households were more likely to experience violence than women from poor or rich households [32]. Research reports on poverty-wealth differences and non-partner violence are scant as most literature reports are based on intimate partner violence [32,33] with inconsistent findings. For example, in India, women from richer households were reported to experience less physical and sexual violence [34]; however, a multi-country study showed that violence against women can transcend the economic divide and affect both poor and wealthy women alike [35].
Environmental factors that may expose women to NPV, such as access to drinking water, toilet facilities, and improved cooking fuel, were not associated with women experiencing non-partner physical violence. In other words, the odds of experiencing non-partner physical violence were not different between women who reported having those facilities on the premises and those who did not. In contrast, accessing common household essentials such as water, sanitation, and family energy needs in Africa has been reported to expose women to NPV [12,13].
However, women's asset ownership was associated with less experience of physical NPV among women. Women's asset ownership (land and/or house), solely or jointly, was associated with less likelihood of experiencing physical NPV. Grabe [36,37] showed that the mechanism through which women's asset ownership, particularly land ownership, could reduce violence against women. The author attributed female land ownership to the concomitant changes in cultural gender roles and power relationships in favor of women. For example, in India, women's land ownership was reported to have significantly reduced violence against women [38]. The authors ascribed the negative influence of women's land ownership to the associated and enhanced economic empowerment, social position, and awareness. In contrast, Pereira et al. [39], using a large sample of women from 28 countries, reported the relationship between asset ownership and intimate partner violence to be inconclusive. Ethiopia's new land reform, aimed at increasing land tenure security and women's rights to own land, has enabled the implementation and delivery of millions of land ownership certificates to women [40], although the impact of such structural change on violence against women is largely unknown. Nonetheless, in the Ethiopian context, our results may suggest that enhanced female asset ownership may have an important role to play in reducing violence against women and may have an implication in mitigating the impact of violence against women in Africa.
Women's access to assets and other resources, a reflection of women's autonomy and empowerment in society, often form part of integrated community intervention strategies to protect women against violence [30,41]. For example, in rural South Africa, microfinancing interventions that targeted the poorest women and provided participatory gender and HIV education in the community have been reported to reduce intimate partner violence by 55% [42]. In Ethiopia, the expansion and availability of microfinancing, especially to women, could play a vital role in creating a conducive environment for asset accumulation and improvement of the livelihoods of poor women. However, microfinancing institutions in Ethiopia currently have the capacity to meet only 20% of the financial demands of the poor [43]. Creating special microfinancing services for women [40] in tandem with community-based gender education programs may have the potential to reduce violence against women in Ethiopia. Girls' and women's education are instrumental to accessing remunerative jobs and financial resources, which, beyond improving women's status, could also contribute to overall economic development and gender equity in Ethiopia [44]; thus, community advocacy for increased investment in girls' and women's education may be warranted.

Conclusions
NPV, like IPV, is a widespread phenomenon in Ethiopia. Nearly 45% of women reported having experienced non-partner physical violence in the last 12 months. Women's asset ownership (land, house, or both), solely or jointly, was associated with less nonpartner physical violence. Thus, structural interventions that integrate women's asset ownership and provide special microfinancing services for women, along with communitybased gender education programs, may have the potential to reduce violence against women in Ethiopia.

Conflicts of Interest:
The author declares no conflict of interest.