Policy Research Working Paper 9168 Effect of Armed Conflict on Intimate Partner Violence Evidence from the Boko Haram Insurgency in Nigeria Uche Eseosa Ekhator-Mobayode Lucia C. Hanmer Eliana Carolina Rubiano Matulevich Diana Jimena Arango Gender Global Theme March 2020 Policy Research Working Paper 9168 Abstract Intimate partner violence is the most common form of vio- behavior from a husband or partner, women’s autonomy lence against women in conflict and non-conflict settings, in household decision making, and their control over their but in conflict settings it often receives less attention than own earnings. The paper finds that the Boko Haram insur- other forms of gender-based violence, such as conflict-re- gency is associated with slower progress toward preventing lated sexual violence. Using data from the 2008 and 2013 and eliminating women’s experiences of physical and sexual Domestic Violence module of the Nigerian Demographic intimate partner violence. Controlling behaviors from hus- and Health Survey spatially linked to the Boko Haram actor bands/partners and reductions in women’s autonomy in file of the Armed Conflict Location and Events Database, household decision making are heightened in locations that this paper employs a kernel-based difference-in-difference are impacted by the Boko Haram insurgency, indicating model to examine the effect of the Boko Haram insurgency that the Boko Haram insurgency adversely affects women’s on women’s experience of physical and sexual intimate agency and exacerbates behaviors that are often precursors partner violence. It also examines the effect of the Boko to intimate partner violence. Haram insurgency on women’s experience of controlling This paper is a product of the Gender Global Theme. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at uekhator@worldbank. org, lhanmer@worldbank.org, erubiano@worldbank.org and darango@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Effect of Armed Conflict on Intimate Partner Violence: Evidence from the Boko Haram Insurgency in Nigeria +,* Uche Eseosa Ekhator-Mobayode a Lucia C. Hanmer b Eliana Carolina Rubiano Matulevich c Diana Jimena Arango d JEL: B54; J12; J16; O12 Keywords: Armed conflict, Forced Displacement, Violence against women and girls, Gender Based Violence, Intimate Partner Violence, Boko Haram Insurgency, Nigeria + This work is part of the program ``Building the Evidence on Protracted Forced Displacement: A Multi-Stakeholder Partnership''. The program is funded by UK aid from the United Kingdom's Department for International Development (DFID), it is managed by the World Bank Group (WBG) and was established in partnership with the United Nations High Commissioner for Refugees (UNHCR). The scope of the program is to expand the global knowledge on forced displacement by funding quality research and disseminating results for the use of practitioners and policy makers. This work does not necessarily reflect the views of DFID, the WBG or UNHCR. * This work benefitted from fruitful discussions from the WBG seminar series for young fellows. It also benefitted from insightful discussions from various teams at the WBG working on projects in Northeastern Nigeria as well as in FCV settings including Boko Haram. Special thanks to Alvaro Federico Barra, Kathleen Beegle, Jocelyn Kelly and Adanna Chukwuma for useful comments and suggestions. a Corresponding author, WBG-UNHCR Young Fellow, World Bank Group; Assistant Professor of Economics, University of Pittsburgh, Bradford. Email address: ucheekhator@gmail.com b Lead Economist, Gender Group, World Bank Group c Economist, Gender Group, World Bank Group d Senior Gender-based Violence and Development Specialist, Gender Group, World Bank Group 1. Introduction Sexual violence is often used as a weapon of war, to control or exert power over a population. Gender-based violence (GBV) experienced in conflict affected areas varies across settings in terms of prevalence and perpetrator; women and girls bear the brunt of this violence (UN Women 2015). Rape in war and other forms of conflict-related GBV, including child sexual abuse, and sexual exploitation and their devastating consequences have been documented extensively (Vu et al. 2014; United Nations 2017; United Nations; World Bank 2018). High rates of intimate partner violence (IPV) have been recorded in several conflict settings (The Global Women’s Institute and the International Rescue Committee 2016) but data are limited, and few studies are able to isolate the association between conflict and IPV risk from other IPV risk factors. This study adds to the literature that examines the association between exposure to conflict and IPV controlling for other risk factors using data gathered in Nigeria at the time of the Boko Haram (BH) insurgency. It adds to a small but growing evidence base that underscores the need for interventions that respond to IPV and seek to transform the norms that legitimize interpersonal violence in conflict-affected settings. The focus of this paper on IPV is important. IPV is the most common form of GBV 1 in both conflict and non-conflict settings (GIWPS and PRIO 2017). It is defined by the World Health Organization (WHO) as actions by an intimate partner or ex-partner that result in physical, sexual or psychological harm, including physical aggression, sexual coercion, psychological abuse and controlling behaviors (WHO 2014). Conflicts have been found to increase all forms of GBV (Clark et al. 2010), however IPV is under-reported and often receives less attention compared to sexual violence perpetrated by armed forces (Khawaja 2004; Mchale et al 2011; Pittaway 2005; Stark and Ager 2011; Vu et al 2014). Violence against women and girls is a violation of basic human rights that has adverse impacts on human development at the level of the individuals, family and society (Klugman et al 2014). A body of evidence implies that IPV will hinder recovery from conflict and is a barrier to increasing gender equality and moving people out of poverty. IPV causes physical and emotional traumas which are strongly correlated to poorer physical, sexual, reproductive and mental health outcomes throughout women’s lives (Champion 1998; Golding 1999; Campbell 2002; WHO 2013; Falb et al. 2013; Falb et al. 2014; García-Moreno et al. 2014). It is associated with increased drug and alcohol use (Krug et al. 2002), loss of income (Crowne et al. 2011; Lindhorst et al. 2007; Reeves & O’Leary-Kelly 2007) and negatively impacts child development (Krug et al. 2002; Asling-Monemi et al. 2003; Holt et al. 2008; Silverman et al. 2009; Duvvury et al. 2012; Duvvury et al. 2013; Klugman et al. 2014; Raghavendra et al. 2017). IPV also feeds the intergenerational transmission of violence, as witnessing IPV as a child or experiencing violence may increase the likelihood of becoming an aggressor or a victim during adulthood (Roberts et al. 2010; Wilkins et al. 2014). Between 2008 and 2013, rates of IPV in Nigeria decreased slightly from 18 to 16 percent. However, progress was uneven. Rates of IPV remained unchanged in urban areas, increased in the Southwest and Northeast regions of the country, and decreased in other regions (see Table 1). 1 The United Nations (UN) Inter Agency Standing Committee (IASC) defines Gender Based Violence as “an umbrella term for any harmful act that is perpetrated against a person’s will and that is based on socially ascribed differences (i.e. gender) between males and females” (IASC 2015). It includes physical, sexual or mental harm and encompasses Intimate Partner Violence (IPV) among others forms of violence. 2 Furthermore, progress in reducing IPV differs between Boko Haram affected areas and the rest of Nigeria. Controlling for differences in individual and household characteristics, we find that BH insurgency is associated with slower progress towards preventing and eliminating women’s experiences of physical and sexual IPV. This finding implies that conflict may not always increase the prevalence of IPV but may set back progress in lowering IPV that could otherwise have been achieved. Our findings contribute to two literatures. First, we extend the literature that quantifies the association between conflict and rates of IPV using nationally representative data. In addition, we analyze the association between conflict and indicators of behavior change that are associated with IPV risk; controlling behavior of male partners and women’s ability to influence household decision-making. We find controlling behaviors of husbands/partners and reductions in women’s autonomy in household decision-making are heightened in locations that are impacted by the BH insurgency, indicating that the BH insurgency adversely affects women’s agency and exacerbates behaviors that are often pre-cursors to IPV. Second, in analyzing IPV we expand the research literature examining the effect of conflict on human development in Nigeria. UNDP (2018) points to the destruction of social and economic infrastructure in the conflict affected areas. Measures of well-being – including food security, gender equality, access to health, education, clean water and sanitation - captured in the various reports and community surveys conducted in the region have all worsened. Research to date has focused on health outcomes and access to maternal health services. Ekhator-Mobayode and Asfaw (2019) and Howell et al. (2018) find the BH insurgency reduced anthropometric measurements of children under five. And Chukwuma and Ekhator- Mobayode (2019) find that the BH insurgency decreased the probability of any antenatal care visits and decreased the probability of delivery at a health center by a skilled health professional. We use a careful empirical strategy employing a quasi-experimental methodology to explore these conflict-related impacts. To estimate the effect of the BH insurgency on IPV we spatially link geo-referenced data on conflict events from the Armed Conflict Location and Event Database (ACLED) with survey data from two rounds of the Domestic Violence (DV) module of the Nigerian Demographic and Health Survey (NDHS) collected in the period before and during the BH insurgency, and apply a difference in difference approach. The remainder of the paper is structured as follows. Section 2 provides some background on the IPV prevalence in Nigeria and the Boko Haram insurgency. Section 3 discusses the conceptual framework. Section 4 presents the data and empirical model specification. Results are presented in section 5 and section 6 concludes. 2. Background: IPV prevalence in Nigeria and the Boko Haram insurgency The most recent estimate of IPV in Nigeria, based on data from the 2013 NDHS, suggests that 16 percent of women have ever experienced physical or sexual IPV (NPC 2014), a rate considerably lower than lifetime prevalence of IPV among ever-partnered women for Africa -37% (World Health Organization, 2013). However, IPV rates vary considerably across regions, reaching 28 percent in the South-South region (Table 1). Analysis of the NDHS data finds that there are significant ethnic and geographical differences in the likelihood of experiencing IPV (Lino et al. 2013; Nwabunike and Tenkorang 2015) and qualitative and smaller scale quantitative studies find 3 that attitudes to and experiences of IPV differ between religious and cultural groups 2 (Odimegwu and Okemgbo 2003; Antai and Antai 2008;). IPV is still widely condoned in Nigeria with 35 percent of women and 25 percent of men believing a husband is justified in beating his wife for any of the five reasons asked about in the survey 3 (NPC 2014). Table 1: Women who have ever experienced physical or sexual violence by a husband or partner in Nigeria, by location and region, percent. National Urban Rural North- North- North- South- South- South- Central East West East South West 2008 18 18 19 26 17 7 24 39 14 2013 16 18 15 21 21 6 20 28 20 Source: 2008 and 2013 Nigerian Demographic and Health Survey reports Between 2008 and 2013, a time period that corresponds to the onset of the BH insurgency, the national rate of IPV fell from 18 to 16 percent. However, this progress was uneven and in two regions, North East and South West, IPV prevalence rates increased. The BH insurgency was focused in northeast Nigeria and so the rise in IPV could be associated with conflict but equally could be driven by other factors that create variation in IPV prevalence rates in different locations in Nigeria. Our paper examines whether the conclusion that conflict has led to increased IPV can be supported by further analysis of the DHS data and we set out below some background to the conflict and identify the locations where communities were most impacted. Boko Haram (BH) carried out its first known terrorist activity in the Northeastern state of Borno, Nigeria in 2009. 4 BH’s activities were initially focused mostly in Northeastern Nigeria but from 2014 onwards its terrorist activities spread to the neighboring countries of Cameroon, Chad and Niger (UNHCR 2018). Between 2009 and 2017, the Armed Conflict Location and Events Database (ACLED) reports 2,378 conflict events 5 across Nigeria with BH as an actor. During this period, the Northeastern states of Borno, Yobe and Adamawa experienced the highest number of conflict events associated with the BH insurgency with 1,678, 219 and 179 events respectively. The insurgency is responsible for large-scale forced migration and an unprecedented regional and 2 Ethnicity is often used as a proxy for culture because it entails social norms that determine behaviors as well as the gender roles and power dynamics within the household and community as well as women’s contribution to household decision-making (see Oyediran and Isiugo-Abanihe 2005). 3 Reasons include: arguing with him, refusing sex, burning the food, neglecting the children, or going out without telling him. 4 Armed Conflict Location & Event Data Project (ACLED). https://www.acleddata.com/, accessed March 10, 2018. 5 BH events in the ACLED are defined as any incidents including riots, clashes, protests, riots or battles where BH is an actor. 4 humanitarian crisis (Read 2017). 6 In January 2017 an estimated 1.8 million people, mostly women and girls, in Northeastern Nigeria, needed protection from GBV (UNOCHA 2017). Figure 1: Boko Haram events by Local Government Areas (LGAs) in Nigeria, 2009 – 2013 Source: Authors own calculation using data from ACLED BH actor file and Nigeria LGA shape files from the Database of Global Administrative Areas – GADM. ACLED records of the location of conflict events with BH actors for 2009–2013, the time period that coincides with the DHS data collection on IPV prevalence rates, are shown in Figures 1 and 2. 7 During this period, there were a total of 799 BH events (see Table A1 for number of BH events in each Local Government Area (LGA)). 8 By far the most affected area was Maiduguri LGA in the Northeastern state of Borno with 274 BH events. Other highly impacted areas were Nasarawa LGA in the Northwestern state of Kano with 55 attacks, Konduga LGA and Damaturu LGA in the Northeastern state of Borno with 48 and 40 BH events, respectively, and Nangere LGA in the 6 As of October 31, 2018, the UNHCR reports 231,474 Nigerian refugees in the region. Of these, 51% were in Niger while 44% and 5% of refugees were in Cameroon and Chad respectively. Most of the Nigerian refugees in Cameroon (98%) were in the far North and displaced by the BH insurgency (UNHCR 2018). It was also reported that 2,421,372 Internally displaced persons (IDPs) were in the region. Of these 80% were in Nigeria while 9% and 7% were in Cameroon and Chad respectively and 4% in Niger. All the IDPs in Chad and Niger and 94% of those in Nigeria were displaced by the BH insurgency. https://data2.unhcr.org/en/situations/nigeriasituation#_ga=2.136643888.1888885269.1543034189- 138741330.1543034189, accessed October 31, 2018. 7 In Nigeria, LGAs are the administrative subdivision of a state. There are 774 LGAs across 36 states and the Federal Capital Territory (FCT). 8 Based on data downloaded from www.acleddata.com on August 30, 2018. 5 Northeastern state of Yobe with 33 BH events. These five LGAs accounted for about 60% (479) of the total BH events for the period. Figure 2: BH events by state and year in Nigeria, 2009 – 2013 Source: Authors own calculation using data from ACLED BH actor file. 6 Conflict-induced displacement and violence associated with the BH insurgency have been widely reported to have exacerbated violence against women (UNOCHA 2017; Read 2017). The BH insurgency has been consistently associated with territorial control, forced recruitment, hostage taking and violence against women and girls as well as men and boys (Nagarajan 2017). BH is notorious for employing various tactics for terrorizing communities. 9 In February 2014, 59 boys were killed at a Federal Government College in Buni Yadi, Yobe State by BH and girls were ordered to leave school and get married (Hemba 2014). A few months later, in April 2014, over 200 school girls were abducted in Chibok, Borno state with BH insisting that the education of the girl child is taboo (National Consortium for the Study of Terrorism and Responses to Terrorism 2014). In February 2018, BH was also responsible for the abduction of 110 school girls in Dapchi, Yobe State. 10 Since 2013, at least 1,000 children have been abducted by BH. 11 Men and boys have been indiscriminately targeted and have fled their families because of the risk of being recruited by BH or killed by law enforcement (Dietrich 2015). Increases in IPV associated with the BH insurgency have become a matter of deep concern (IRC 2017a; IRC 2017b; UNOCHA 2018). Women in communities affected by the BH insurgency report increased arguments and violence within the home, and sometimes ultimately abandonment by their partners (Nagarajan 2017). They often attribute abandonment to the inability of a male partner to provide for his family and his frustration with being unable to do so (World Bank 2018). For men, living in conflict affected situations can create barriers to meeting traditional social norms surrounding masculinity, for example to be in control of property, land, income, their lives and the lives of their families in general and lead to increased stress and depression (IRC 2012; Narayan 2000; Slegh et al. 2014; Women’s Commission for Refugee Women and Children 2005) which are risk factors for the perpetration of IPV (Clark et al. 2010; Hanmer and Klugman 2016). The BH insurgency is also associated with changing marriage practices and incentivizing early marriage, the latter a risk factor for experiencing IPV. Women and girls have been coerced or persuaded into associating with BH with promises of religious knowledge, a higher social position than that accorded to them by preexisting patriarchal gender norms and/or reducing financial burdens through marriage to BH members (Nagarajan 2017). In some communities BH has eased the cost of marriage ceremonies by sponsoring them and ordered the payment of bride price to the bride instead of her family thereby incentivizing brides to marry early. Reports document that families in BH affected areas arrange early marriages for their daughters both to reduce pressure on household resources (Campagne and Begum 2017) and in exchange for protection (International Crisis Group 2016). 9 Report of the Secretary-General on the situation in Central Africa and the activities of the United Nations Regional Office for Central Africa, UN Doc. No. S/2016/996, 28 Nov 2016, para. 28; Amnesty International estimates that more than 2,000 women and girls were abducted by Boko Haram between 2009 and 2015, many of whom were forced into marriage with Boko Haram fighters. Amnesty International, Nigeria: ‘Our Job Is to Shoot, Slaughter and Kill’: Boko Haram’s Reign of Terror in North East Nigeria, April 14, 2015, p. 59. 10 Aljazeera, News (February 25, 2018). “110 Nigerian schoolgirls still missing after attack: Minister” Aljazeera Media network. Retrieved September 1, 2018. 11 https://www.unicef.org/press-releases/more-than-1000-children-northeastern-nigeria-abducted-boko-haram-since- 2013, accessed January 7, 2019. 7 3. Impact of armed conflict on IPV: Conceptual framework and evidence How does armed conflict translate into increased risk of violence perpetrated by an intimate partner? The ecological framework (Heise 1998) has been widely adopted to understand the origins of gender-based violence. This section applies the framework to the evidence on the impacts of BH insurgency on individuals and communities to identify potential causal pathways. The ecological framework conceptualizes violence as a multi-dimensional phenomenon, based on the complex interplay between factors related to violence against women at different levels of the social ecology. The social ecology consists of personal history, the microsystem, the exosystem and the macrosystem. Personal history refers to individual experiences that determine one’s response to microsystem and exosystem stressors. The microsystem explains the subjective meanings assigned to interactions in which a person directly engages with another. While the exosystem refers to both formal and informal social structures that affect one’s environment and influences what happens there. Conflict triggers changes in the social ecology at the individual level (microsystem) and in formal and informal social structures (exosystem). At the individual level, evidence suggests that men and women respond differently to the high levels of psychological stress experienced due to conflict. For example, a study on conflict-affected areas of the Democratic Republic of Congo (DRC) found that men tend to cope with extreme stress and trauma using strategies that seek to avoid and reduce feelings of vulnerability, including alcohol and substance abuse, behaviors commonly associated with increased perpetration of IPV, while women more frequently seek some form of help or turn to religion (Slegh et al. 2014). In the case of Boko Haram, conflict can also affect individual relationships through their influence on marriage practices. Increased rates of child and early marriage to Boko Haram insurgents result in some women and girls having little choice about marrying a potentially violent partner and often they have no option but to remain in the marriage (IRC 2012; Wirtz et al. 2014). Conflict can also alter individual behaviors. Personal violence outside the home may spill over into relationships with intimate partners, as disputes and disagreement often escalate into violence and abuse— potential triggers for increased rates of IPV (Kiss et al. 2015). The breakdown of law and order, or social support and safety systems such as the extended family or village groups (exosystem factors) may also reduce protective factors that, at the individual level, allow intervention and prevention of IPV (Clark et al. 2010; Horn 2010). The different consequences of stress related to unemployment and income loss for men and women is another potential driver of increased rates of male perpetration of IPV. Frequently conflict creates barriers to employment and so men may find it difficult to fulfill the traditional role of male breadwinner and protect the family. As a corollary, women’s labor market participation often increases in conflict settings and the income they generate becomes more important for family welfare (Buvinic et al. 2013; Justino 2018). In some conflict settings, men face higher barriers to employment than women, as evidenced by research on the Palestinian Israeli conflict which suggests that the localization of economic activities resulting from conflict is more likely to adversely impact the labor market opportunities for men than women (Institute for Women’s Studies 2008). These changes in the gender division of labor can threaten masculinity and are often associated with increased use of violence by men. For example, a study of DRC found that for men, work and income are central to their identities, and the loss of either seems to have serious 8 consequences in terms of mental health with both men and women reporting lack of work and income as a trigger for men’s use of violence (Slegh et al. 2014). Several studies in the IPV literature have examined the prevalence and key drivers of IPV using the ecological framework (see Heise 1998; Walker 2006; Capaldi et al. 2012; Breiding Chen, & Black 2014; Widom and Wilson 2015). The literature also encompasses analysis of the relationship between harmful social norms that condone IPV and women’s experience of IPV (see Linos et al. 2013; Haj-Yahia and Clark 2013; Antai 2011; Tenkorang 2018). Experience of controlling behavior and loss of autonomy have been categorized as forms of violence themselves but also, they may be precursors to or key drivers of more severe forms of IPV (Aizpurua et al. 2017). For example, Antai (2011) examines controlling behavior, power relations within intimate relationships and IPV against women in Nigeria and finds that controlling behavior by the husband/partner is associated with a higher likelihood of experiencing physical violence. Also, women who justified wife beating and earned more than their husband/partner were more likely to experience physical and sexual violence, while women who had decision-making autonomy had a lower likelihood of experiencing physical and sexual violence. Tenkorang (2018) examines women’s autonomy on three decisions- economic decision-making, 12 family planning decision- making 13 and sexual autonomy14 and women’s experiences of four types of IPV- physical, sexual, emotional and economic - in Ghana. He finds that at the individual level, family planning decision- making was negatively associated with all four types of violence while economic decision-making autonomy was positively associated with emotional and economic violence. Also, living in a community where women had higher levels of sexual autonomy was associated with a lower likelihood of physical and economic violence. Finally, in a systematic review of 23 studies, McCarthy et al. (2018) examine the association of 64 measures of harmful social norms and behaviors categorizing them into three main thematic areas—views on gender roles/norms, acceptance of violence against women, and gender-related inequities in relationship power and control—and their association with male perpetration of IPV. Overall, these measures were positively associated with IPV perpetration in 45 percent of the cases. Measures inclusive of acceptance of violence against women or beliefs about men’s sexual entitlement, followed by those that measured respondents’ views on gender roles/norms, were most consistently associated with IPV perpetration. Few studies have examined the effect of conflict on IPV or key drivers of IPV in fragile and conflict settings. The research conducted to date finds a positive correlation between conflict and increased rates of IPV. Kelly et al (2018) find that residing in a conflict fatality-affected district in 12 Economic decision-making is measured as a “weighted summative index derived from women’s responses to four questions as follows: Who usually decides how the money you earn will be used? Who usually decides how your husband’s/partner’s earnings will be used? Who usually makes decisions about health care for you? Who usually makes decisions about major household purchases?” (Tenkorang 2018) 13 Family planning decision-making is measured as a “weighted summative index derived from women’s responses to three questions: Who usually makes decisions about how many children to have? Who usually makes decisions about when to have sexual intercourse? Who usually decides whether or not to use contraceptives/ condoms?” (Tenkorang 2018) 14 Sexual autonomy is measured as “a weighted summative index derived from 11 questions asking if a married woman can refuse to have sex with her husband/partner if she doesn’t want sex; he is drunk; she is sick; he mistreats her; she is menstruating; he does not want to use a condom/contraceptive; she finds out that he has other girlfriends/partners; he refuses to give her housekeeping money; he humiliates her; he refuses to pay the children’s school fees; or he has an STI, such as HIV.” (Tenkorang 2018) 9 Liberia was associated with a 50 percent increase in the risk of IPV compared to residing in a district where there were no conflict-related fatalities. Women living in a district that experienced 4-5 cumulative years of conflict were also more likely to experience IPV. Gallegos and Guitierrez (2016) examine the effect of women’s exposure to the internal conflict in Peru in childhood and teenage years on the likelihood that they will perpetrate or experience domestic violence. Findings suggest that female exposure to the internal conflict in Peru increases their risk of being both a perpetrator and a victim of IPV. The study also finds that women exposed to more conflict events are more likely to justify violence against women and stay in a violent relationship. Østby (2016) using subnational data on armed conflict events with individual-level data on partner abuse from DHS surveys in 17 Sub-Saharan African countries finds that the intensity of armed conflict in the home region has an independent and significant effect on women’s risk of experiencing sexual IPV. Falb et al. (2013) examine the effect of conflict on past-year IPV among refugee women affected by the conflict in Burma (Myanmar). They find that women who experienced conflict victimization 15 were 6 times more likely to report past-year IPV than women who had not experienced conflict victimization. Turning to drivers of IPV in conflict affected and fragile settings, results from a number of studies are consistent with the hypothesis that the impact of conflict outside the home spills over to affect individuals within the household. Bruck and Stojetz (2019) examine exposure to wartime sexual violence by armed groups and find exposure to sexual violence by armed groups against women makes male veterans about 30 percentage points more likely to commit physical – but not sexual – IPV 18 years later (on average). Saile et al. (2013) examine the prevalence and predictors of current partner violence experienced by women in post-conflict Uganda. The authors suggest that women’s prior exposure to war-related traumatic events as well as men’s level of alcohol-related problems were associated with higher levels of partner violence against women. And Gupta et al.’s (2012) study of South Africa during apartheid of men who were liberation supporters and men who were government supporters found an association between experiencing major human rights violation (HRV) and victimization of close friends or family members and with perpetration of physical IPV. For liberation supporters, experiences of custody related HRV were also linked with IPV perpetration. Clark et al. (2010) examine the effect of political violence on female experiences and male perpetuation of IPV in West Bank and Gaza and find that political violence is significantly related to higher risk of IPV. Specifically, the direct exposure of a husband to political violence was associated with 89 percent higher odds of reporting physical violence, and 123 percent higher odds of reporting sexual violence compared with women whose husbands were not directly exposed to political violence. Similarly, Gupta et al. (2009) find that exposure to political violence in their country of origin increased the likelihood of male immigrants to the United States perpetrating IPV. Qualitative studies have shed more light on the interactions between factors at different levels of the ecological framework, pointing to causal pathways and showing a more nuanced dynamic. For example, using focus groups, Horn et al. (2014) explore women’s perceptions of the causes of IPV in conflict settings in Sierra Leone and Liberia. The findings suggest that wars increased the use of violence by some men, which was driven by the normalization of violence as a way of 15 Falb (2013) describes conflict victimization as “targeted violence against civilian women, including sexual violence which is seen as a "tactic of war" and associated with a range of negative outcomes, including poor mental health stigma, isolation, and shame; and increased vulnerability to HIV and other sexually transmitted infections.” 10 responding to frustrations and challenges. However, the war also resulted in women becoming economically active, which in this case decreased IPV, as the pressure on men to provide for their families reduced. Also, economic independence and interventions provided women with the option of leaving a violent relationship. Thus, IPV risk is linked to both formal and informal social structures that create gender inequality, including financial dependence on men, traditional gender norms surrounding masculinity as well as the broader changes in social norms and behaviors related to interpersonal violence that occur during conflict. 4. Data and model specification 4.1 Data and sample construction Data are drawn from the 2008 and 2013 Nigerian Demographic and Health Survey (NDHS), which include Domestic Violence (DV) modules that sample 23,752 and 27,634 women respectively. 16 The NDHS includes information on the location of the interview and its GPS coordinates. Observations from the 2008 NDHS provide data for the period before the BH insurgency while observations from the 2013 NDHS provide data for the period during the BH insurgency. Exposure to BH conflict is measured using the Armed Conflict Location and Event Database (ACLED), which records events whether they generate fatalities or not. The data on events are reported by date, location, agent and type. Our sample is made up of data drawn from 664 Local Government Areas (LGAs) across all 36 states, including the Federal Capital Territory (FCT). Observations from the 2008 and 2013 NDHS are linked to the BH events recorded in ACLED using the GPS coordinates provided in both data sets. To match the timing of the NDHS surveys, we use geocoded BH events that occurred between 2009 and 2013. 17 During this period, 799 BH events in Nigeria were recorded in ACLED. Because of the intensity of the BH insurgency, interviewers during the field work of the 2013 NDHS could not reach some conflict-affected areas (NPC 2014). As we exploit variations in the timing and location of BH events, observations from the 2008 NDHS from LGAs not surveyed in the 2013 NDHS - Maidugiri in Borno State and Damaturu and Nangere in Yobe State - are excluded from our analysis. 4.2 Outcome variables Both the 2008 and 2013 NDHS DV modules follow international global standards for the measurement of intimate partner violence. Ever partnered women were asked questions about experiences of violence using a modified version of the Conflict Tactics Scale (Straus 1990), a commonly used instrument in both the DHS and stand-alone violence against women and girls surveys. 18 To measure experience of physical or sexual IPV, women are asked if they ever 16 The NDHS is a nationally representative cross-sectional household survey that provides data on population, health, and nutrition for women aged 15-49. 17 Because the ACLED often reviews specific periods of conflict to ensure accuracy, there may be changes to the data in selected countries or targeted conflicts throughout the year. The ACLED data used for the analysis in this study were downloaded from www.acleddata.com on August 30, 2018. 18 Sociologist Straus (1990) developed the original Conflict Tactics Scale (CTS), which consists of a series of individual acts of physical and sexual violence (Straus, 1979, 1990). The modified list used by the NDHS includes only fewer acts of physical and sexual violence (see appendix section A2). If a woman confirms that any one of the 11 experienced certain behaviors from their husband/partner including pushing, slapping, arm twisting, punching, kicking, choking, attacking with weapon, forceful sexual intercourse, forceful performance of other sexual acts. When an ever-partnered woman reports experiencing physical or sexual IPV, they are asked when the first episode occurred in years after marriage or having a partner. They are also asked if there were any episodes in the 12 months preceding the survey. 19 Because we exploit the timing and location of IPV in the BH affected area, our analysis focuses on women’s experiences of physical or sexual IPV in the year preceding the survey – commonly taken to mean that women are currently experiencing violence (Abramsky et al. 2011). Using this information, we create a dichotomous variable measuring past year physical or sexual IPV, which equals 1 if a woman answers yes to experiencing any behaviors about physical or sexual IPV from her husband/partner in the past 12 months and 0 otherwise. We use information on men’s controlling behavior and women’s autonomy in decision-making to examine key drivers of IPV. Both the 2008 and 2013 NDHS contain some information on women’s autonomy in household decision-making and individual behaviors that are related to gender norms that can be used to examine BH’s impact on these behaviors. We explore whether women’s experiences of controlling behavior from her husband/partner, women’s household decision- making autonomy and women’s decision-making autonomy over her own earnings are affected by the BH insurgency. To determine if a woman experiences controlling behavior from her husband/partner, the NDHS asks if a husband/partner gets jealous when she talks to other men, accuses her of unfaithfulness, does not permit her to meet female friends, limits her contact with family or insists on knowing where she is at all times. Using this information, we create the variable for controlling behavior, which equals 1 if a woman answers yes to any of these questions and 0 otherwise.Women’s household decision-making autonomy is measured by a woman’s involvement in three types of household decisions: her ability to decide to access health care services for herself; her ability to make decisions about major household purchases; and, her ability to decide when to visit her family or relatives. A woman has household decision-making autonomy if she makes decisions alone or jointly with her husband/partner. The variable capturing women’s household decision-making autonomy is an index that ranges from 0 to 3, corresponding to the number of household decisions in which a woman is solely or jointly involved with her husband/partner. A higher score indicates a higher level of empowerment (NPC 2014). To determine earnings decision-making autonomy, women who earned cash for work in the 12 months preceding the survey are asked who usually decides how their earnings are spent. The variable capturing women’s earnings decision-making autonomy is equals 1 if a woman makes decisions alone or jointly with her husband/partner and 0 otherwise. specified acts or outcomes has taken place, she is considered to have experienced physical or sexual violence. By asking separately about specific acts of violence, the NDHS measure of physical or sexual violence using the modified CTS approach is not affected by different perceptions of what constitutes violence by women (Sunita and Johnson, 2004). Sunita and Johnson (2004) state that “for example, a woman has to specify if she was ever “slapped,” not whether she has ever experienced violence. All women would probably agree what constitutes a slap, but what constitutes a violent act or what is understood as violence may vary among women and across cultures.” 19 Although the 2013 NDHS reports the woman’s location of residence at the time of the survey, information about any previous location and reasons for relocation is not collected, which means that information about migration due to the BH insurgency is not available from the survey. 12 4.3 Exposure to the BH insurgency Exposure to conflicts such as the BH insurgency is defined in different ways in the literature. It can be defined as residence in administrative areas with conflict events (see for example Akresh et al., 2012; Ekhator-Mobayode and Asfaw, 2018) or as residence within specified buffer zones (predefined radius) around conflict events. For example, in mapping the number of children affected by conflict in affected countries, Bahgat et al. (2018) use a buffer zone of 50 km to determine the number of children living in conflict zones, arguing that it is a reasonable distance within which conflict events could be expected to impact people’s lives. However, Chukwuma and Ekhator-Mobayode’s (2019) analysis of the impact of the BH insurgency on access to maternal health services in Nigeria uses much smaller buffer zones--between 3 km and 10 km radius--on the basis that 10 km is a reasonable maximum catchment area for access to health centers. Howell et al.’s (2018) examination of the impact of civil conflict on child nutrition and mortality in Nigeria uses buffer zones of between 2 km and 10 km. We define exposure to the BH insurgency as residence within a 10 km radius of any BH event during the study period (Figure 3). We chose a smaller exposure zone, rather than a Local Government Area or a larger zone, as recent research in BH affected states found that the experiences of women and girls were greatly dependent on BH activities in their localities (Nagarajan 2017). Applying the 10 km buffer zone definition of exposure, we find that 11 percent of the women in our study sample lived within 10 km of any BH event both before and during the BH insurgency. These are the women who make up the treatment group in our study. Figure 3: BH attacks between 2009-2013 occurring within 10km of communities surveyed by the Nigeria Demographic and Health Survey. 13 4.4 Control Variables A body of research shows that factors affecting the risk of experiencing IPV such as education, socioeconomic status, formal marriage, alcohol abuse, cohabitation, age, attitudes and experiences of IPV appear to be common across several country settings (Abramsky et al. 2011; Heise 2011; Klugman et al. 2014, Hanmer and Klugman 2016). However, some risk factors may be culturally specific or specific to particular social groups within countries; thus, we include religion and ethnicity as Nigeria-specific control variables, as discussed in Section 2. We classify covariates as follows: individual characteristics, partner characteristics, household characteristics as well as attitudes and experiences of IPV and Nigeria-specific factors (Table 2). 20 Table 2: Variables used in the analysis A. IPV 1. Past year physical or sexual IPV Woman experienced physical or sexual IPV in the 12 months preceding the NDHS. B. Drivers of IPV 1. Controlling behavior by husband/ Woman’s husband/partner exhibits any control issue. ** partner 2. Women’s household decision- Number of household decisions made alone or jointly with making autonomy index husband/partner 3. Women’s earnings decision- Woman decides how her earnings are spent alone or jointly with making autonomy husband/partner. C. Covariates Individual characteristics Age in years*; Education*; Employment status*; Type of earnings; Total number of children born; Married as a child Partner characteristics Age in years*; Education*; Employment status*; Sometimes or often Drunk Household characteristics Polygamous household; Household size*; Location (rural/urban*); Wealth* Attitudes and experiences of IPV Father beat mother; Condones wife beating Nigeria specific factors Culture proxied by ethnicity, religion and state dummies capturing differences in socio-economic conditions between states Note: * denotes variables used in the matching stage. ** defined as answering yes to whether a husband/partner gets jealous when she talks to other men, accuses her of unfaithfulness, does not permit her to meet female friends, limits her contact with family or insists on knowing where she is at all times. 20 We do not include exosystem factors such as controlling behavior by husband/partner and indicators of women’s empowerment / agency as covariates because these factors are jointly determined with IPV and thus endogenous. 14 4.5 Model Specification We test the hypothesis that, in the absence of the BH insurgency, changes in the rates of IPV before and during the BH insurgency would have been the same in the BH affected areas and in areas unaffected by BH (hereafter non-BH areas). To estimate the effect of the BH insurgency on women’s experience of IPV, we specify the following difference-in-difference (DD) model. = 0 + 1 + 2 + 3 * + 4 + (1) The DD model exploits the variation in timing and location of the BH insurgency. This allows for the avoidance of biases from comparisons of IPV rates before and after 2009 that could be the results of factors other than BH attacks. The unit of observation for equation (1) is a woman. is the outcome variable indicating if a woman i, interviewed in an area l, and NDHS year y, experienced IPV. is a binary variable indicating the NDHS year the woman was interviewed. It is 1 for observations from the 2013 NDHS during the BH insurgency and 0 otherwise. indicates whether the woman resides in a BH affected area. It is a binary variable equal to 1 if the woman resides within a 10 km radius of any BH event between 2009 and 2013 and 0 otherwise. The covariates listed in Table 2 are represented by . Finally, is the stochastic error term. The parameter of interest is 3, which measures the average treatment on the treated. That is, the effect of the BH insurgency on women’s experiences of IPV, as measured by the change in IPV between BH affected areas and non-BH areas before and after the BH insurgency. In other words, it estimates the effect of the presence of the BH insurgency on IPV in the BH affected areas. 21 4.6 Estimation Bias In standard DD models, there are two common causes of estimator bias. First, selection bias across time and groups (Stuart et al. 2014). Second, using data from repeated cross sections as opposed to longitudinal data can cause estimates to be biased due to terms that must remain unchanged to ensure before-after comparability (Blundell and Dias 2009). For our model, selection bias across time could occur if the composition of the sample in the BH and non-BH affected areas changes in the period before and during the BH insurgency due to the BH insurgency. Alternatively, selection bias across groups could occur if the composition of the sample changes across the BH and non-BH affected areas. Population movements to and from the BH affected area because of the operations of law enforcement, forced returns and flight from the conflict could affect the composition of women in the BH affected area before and after the BH insurgency. This could cause selection biases to occur both across groups and time. We address these potential sources of bias as follows. First, as discussed in Section 4.1, we exclude from our analysis observations from the 2008 NDHS from areas not surveyed in the 2013 NDHS on the account of the BH insurgency to ensure before-after comparability. Second, we employ a 21 We also check if the intensity of exposure to BH may affect the likelihood of increased IPV beyond residing in an area where the BH insurgency is present, i.e., we determine the marginal effect of BH events in the BH affected area, by accounting for the number of BH events in the affected area within the study period. However, we find that each additional BH event has no significant effect on the outcome variables, except in the case of earnings decision-making autonomy where the impact is 0.1 percent. This is likely because we do not have observations from the areas of the country where the largest number of BH attacks took place. 15 kernel-based propensity score matching method to ensure that the profiles of women in BH areas and non-BH areas are similar based on observables. The kernel-based propensity score matching (PSM) method estimates each woman’s propensity for residing in the BH-affected area during the period of the BH insurgency using a kernel density function. Each of the three control groups (women who reside in the BH area in the period before the BH insurgency, women who reside in the non-BH area before the insurgency and women who reside in the non-BH areas after the BH insurgency) are then matched to the sample of women who reside in the BH area in the period during the BH insurgency. The common support is composed of the sample of women who reside in the BH area to whom counterfactuals are found in each of the three control samples. The method weights each of the individual observations in the three control groups based on propensity scores, creating a balanced control group that does not differ systematically from the treated group on observables. The balanced sample is then used to estimate the average treatment effect of residing in the BH area during the BH insurgency on IPV outcomes using the DD model specified in equation (1) and applying the weights from the PSM.22 One advantage of the kernel-based PSM is that it does not uniquely pair observations and as such minimizes the risk of data imbalance and model dependence (Jann, 2017). However, a key assumption is that only variables that are fixed over time (or pre-treatment variables) should be included in estimating the propensity scores. Caliendo et al. (2005) argue that previous research and information about the institutional settings should be employed in building the model. Hence, we rely on evidence from Ceriani and Verme (2018) who find that age; education; employment status: and, ownership of household assets are associated with the decision to stay or flee from BH affected areas. 23 The study sample is thus matched on individual characteristics of the women and their partners using the following variables: age; education; and employment status; household size, type of settlement (urban versus rural); and household wealth status 24 (see appendix section A3 for kernel-based matching results). Equation (1) is then estimated on the common support of the matched sample. 22 This is implemented with the “diff” command in Stata by Villa 2016. 23 Larger household size, being male, being older and being unemployed were associated with not migrating during conflict while more years of education was associated with migrating during conflict. 24 We use household wealth as a proxy for ownership of household assets. 16 5. Results 5.1 Difference-in-difference (DD) estimates IPV trends in BH affected areas from the matched sample are shown in Table 3. The matched sample contains data on 2,570 women living in the BH affected area and 21,300 women living in areas not affected by BH. In the BH affected area, the proportion of women experiencing physical or sexual IPV is unchanged at around 10 percent. In the non BH area however IPV rates fall from 17 to 12 percent between 2008 and 2013. Table 3: Rates of Intimate Partner Violence in Nigeria before Boko Haram (2009) and after Boko Haram (2013), percent BH Area Non BH Area Variable Before During Difference Before During Difference N BH BH BH BH (Total) Past year physical or sexual IPV 11 10 1 17 12 -5*** N 1126 1444 9585 11715 23870 Source: Authors own calculation using data from the 2008 and 2013 Nigerian Demographic and Health Survey Note: BH Area defined as NDHS clusters exposed to any BH Event within 10km between 2009 and 2013. *** Difference in means between before BH and during BH is significant at the 1% level. After controlling for a woman’s, partner and household characteristics, attitudes towards and experiences of IPV, ethnicity, religion and state fixed effects, the results of the DD model show that the presence of BH increases the probability of a woman’s experience of physical or sexual IPV by 3.7 percentage points (Table 4). The individual and household characteristics associated with increased risk of women’s experience of physical or sexual IPV are: having more children (0.3 percentage points for each additional child), marriage as a child (2.6 percentage points), having a sometimes or often drunk partner (14.6 percentage points) and marriage in a polygamous household (1.7 percentage points). Both indicators of attitudes towards and experiences of IPV are also associated with increased IPV risk; having a father who beat their wife/partner (12.4 percentage points) and condoning wife beating (4.4 percentage points). The finding that the highest IPV risk is associated with having a mother who was beat by their husband/partner and having a husband/partner who is sometimes or always drunk is consistent with other research about generational cycles of IPV (see Lipsky et al. 2005; Knight et al. 2013; Thornberry et al. 2011) and the association of IPV perpetration with drug and alcohol abuse (see Lipsky et al. 2005; Thompson and Kingree 2006; Tumwesigye et al. 2012; Devries et al. 2013; Hanmer and Klugman 2016). There is a weak association between receiving earnings in kind as opposed to no earnings and lower IPV risk (-4.7 percentage points) and membership of certain ethnic groups is associated with lower IPV risk. Our findings are consistent with other research using Nigeria’s DHS data; being Hausa and Fulani as opposed to being Yoruba is associated with lowering risk of physical or sexual IPV by -5.3 and -5.2 percentage points respectively. 25 25 Nwabunike and Tenkorang’s (2015) analysis of the 2008 NDHS finds that Hausa women are less likely to experience physical and sexual violence, compared with Yoruba women. 17 Table 4: Difference-in-differences estimate of the effect of the Boko Haram insurgency on Intimate Partner Violence in Nigeria Variables Past year physical or sexual IPV * 0.037** (0.015) Individual characteristics Type of earnings (reference category: no earnings) Cash only -0.017 (0.018) Cash and kind -0.031 (0.022) Kind only -0.047* (0.027) Total number of children born 0.003** (0.001) Married as a child (before 18 years of age) 0.026*** (0.008) Partner characteristics Sometimes or often drunk 0.146*** (0.013) Household characteristics Polygamous household 0.017** (0.008) Attitudes and experiences of IPV Father beat mother 0.124*** (0.017) Condones wife beating 0.044*** (0.009) Nigeria specific factors Ethnicity (reference category: Yoruba) Igbo 0.003 (0.020) Hausa -0.053** (0.021) Fulani -0.052* (0.030) Other ethnic group 0.006 (0.017) Religion (reference category: Christianity) Islam 0.004 (0.011) Traditional African religion -0.002 (0.033) Other religion -0.037 (0.163) Before BH (Non-BH Area) 9585 Before BH (BH Area) 1126 During BH (Non-BH Area) 11715 During BH (BH Area) 1444 Total 23870 R-squared 0.11 Note: Variables used to estimate the kernel propensity scores are dropped State fixed effects are included. The kernel density function is Epanechnikov with a bandwidth of 0.06. Probit estimation is used for the propensity score in the first stage. ***Significant at the 1% level, **Significant at the 5% level, *Significant at the 10% level. Robust standard errors in parentheses. 18 5.2 The BH insurgency and key drivers of IPV In both the BH affected and unaffected areas, the proportion of women experiencing controlling behavior from their husband/partner increases during the BH insurgency. However, the increase is greatest in the BH area, where the number of women reporting controlling behavior from their husband/partner increases from 60 to 70 percent during the BH insurgency. In contrast, there is only a small increase in the percentage of women experiencing controlling behaviors (from 60 to 62 percent) in the non-BH areas (Table 5). Table 5: Women’s experience of controlling behavior by husband/partner and women’s decision- making autonomy in Nigeria before and after Boko Haram, 2009 and 2013, percent Variables BH Area Non BH Area Before During Difference Before During Difference N BH BH BH BH (Total) Controlling behavior by husband/partner 60 70 10*** 60 62 2*** N 1127 1444 9600 11715 23886 Women’s household decision- making autonomy Autonomy over healthcare 50 44 -6*** 50 49 -1 Autonomy over major household purchases 39 41 2** 44 50 6*** Autonomy over visits to friends and family 52 49 -3* 62 58 -4*** N 1127 1443 9596 11711 23877 Women’s earnings decision-making 89 92 3*** 84 89 5*** autonomy N 1039 1383 7449 10823 20694 Source: Authors own calculation using data from the 2008 and 2013 Nigerian Demographic and Health Survey Note: *** Difference in means between before BH and during BH is significant at the 1% level. ** Difference in means between before BH and during BH is significant at the 5% level. * Difference in means between before BH and during BH is significant at the 10% level. The proportions of women with decision-making autonomy changes during the BH insurgency in both the BH affected areas and the non-BH areas. However, there is not a systematic pattern of decreased decision-making autonomy in the BH affected areas compared to the non BH areas. In both areas, autonomy over health care decisions and autonomy over decisions to visit friends and family decreases and autonomy over major household decisions and disposition of earnings increases. Reductions in autonomy over health care decisions are greater in the BH area than the non BH area (6 percentage points vs 1 percentage point) but there is little difference between the BH and non BH areas in the reductions in autonomy over visits to friends and family. The increase in the proportion of women who have autonomy over major household purchases is larger in the non BH areas than the BH areas (6 percentage points vs 2 percentage points). Finally, the increase in the proportion of women having earnings decision-making autonomy is lower in the BH area than in the non-BH area (3 percentage vs. 5 percentage points). . We examine the impact of the BH insurgency on the key drivers of IPV by estimating the DD model in equation (1) on the common support of the matched sample, using controlling behavior by husband/partner, an index of women’s household decision-making autonomy and women’s earnings decision-making autonomy as the outcome variables (Table 6). After controlling for individual and household characteristics, attitudes to and experiences of IPV, ethnicity, religion and state fixed effects, the BH insurgency increases the probability of women’s 19 experience of controlling behavior by husband/partner by 13.8 percentage points and reduces the probability of women’s household decision-making autonomy by 22.4 percentage points but has no statistically significant effect on women’s earnings decision-making autonomy. Having a sometimes or often drunk partner is associated with increased risk of women’s experience of controlling behavior from their husbands/partners (11.1 percentage points) while receiving earnings and having more children are associated with decreased risk of women’s experience of controlling behavior (5.2 percentage points for cash only earnings, 11.2 percentage points for earnings in cash and kind and 0.8 percentage points for each additional child). Also associated with increased risk of women’s experience of controlling behavior from husband/partner are having a father who beat his wife/partner, condoning wife beating and practicing Islam as opposed to Christianity (8.3, 13.1 and 3.4 percentage points respectively). Table 6: Difference-in-differences estimates of the effect of the Boko Haram insurgency on women’s experience of controlling behavior and decision-making autonomy in Nigeria (1) (2) (3) Variables Controlling Women’s household Women’s earnings behavior by decision- making decision-making husband/partner autonomy autonomy * 0.138*** -0.224** -0.013 (0.034) (0.101) (0.008) Covariates Individual characteristics Type of earnings (reference category: no earnings) Cash only -0.052** 0.076 - (0.022) (0.105) Cash and kind -0.112*** 0.169 - (0.031) (0.138) Kind only -0.062 -0.351 - (0.053) (0.097) Total number of children born -0.008*** 0.033*** 0.003*** (0.002) (0.001) (0.001) Married as a child 0.002 -0.147*** 0.001 (0.013) (0.061) (0.005) Partner characteristics Sometimes or often drunk 0.111*** 0.116 -0.010 (0.018) (0.079) (0.007) Household characteristics Polygamous household -0.005 -0.221*** 0.009* (0.013) (0.063) (0.005) Attitudes and experiences of IPV Father beat mother 0.083*** 0.073 -0.010 (0.019) (0.088) (0.007) Condones wife beating 0.131*** -0.318*** -0.011** (0.014) (0.006) (0.005) Continued. 20 Table 6 Continued. (1) (2) (3) Variables Controlling Women’s household Women’s earnings behavior by decision- making decision-making husband/partner autonomy autonomy Nigeria specific factors Ethnicity (reference category: Yoruba) Igbo -0.007 -0.004 -0.003 (0.030) (0.128) (0.010) Hausa 0.016 -0.071*** -0.033*** (0.038) (0.136) (0.011) Fulani 0.006 -0.813*** -0.094*** (0.042) (0.172) (0.014) Other ethnic groups -0.015 -0.484*** -0.004 (0.027) (0.107) (0.009) Religion (reference category: Christianity) Islam 0.034** -0.619*** 0.013** (0.017) (0.078) (0.006) Traditional African religions 0.022 -0.733*** -0.107*** (0.046) (0.300) (0.027) Other religion 0.075 -0.973** -0.137 (0.164) (0.135) (0.190) N Before BH (Non-BH Area) 9600 9596 7449 Before BH (BH Area) 1127 1127 1039 During BH (Non-BH Area) 11715 11711 10823 During BH (BH Area) 1444 1443 1383 Total 23886 23877 20694 R-squared 0.10 0.38 0.08 Note: Variables used to estimate the kernel propensity scores are dropped are dropped. We log-transform the index of women’s household decision-making autonomy before estimating the model to allow for the reporting of the estimates in terms of percentages. To take care of zero values, we add 0.0001 to the index before taking the log. Thus, the estimates reported calculated as ( − 1). The estimations account for state fixed effects to account for the clustering of IPV among women residing in the same state. The kernel density function is epanechnikov with a bandwidth of 0.06. Probit estimation is used for the propensity score in the first stage. Kernel-based DD estimates are on the common support. The results are robust to employing a logit estimation of the propensity scores in the first stage, a gaussian kernel density function as opposed to an epanechnikov kernel density function and varying the bandwidth of the kernel density function. ***Significant at the 1% level, **Significant at the 5% level, *Significant at the 10% level. Standard errors in parentheses are robust. 21 Turning to women’s household decision-making autonomy, there is a marked difference between the correlates of this variable and the correlates of controlling behavior of husbands/partners. The strongest correlates are related to marital status, child marriage, ethnicity and religion. Having more children is associated with increased likelihood of decision-making autonomy while marriage as a child, marriage in a polygamous household and condoning wife beating are associated with reduced likelihood of women’s household decision-making autonomy (14.7, 22.1, and 31.8 percentage points respectively). Ethnicity and religion are also associated with women’s household decision-making autonomy- being Hausa, Fulani and from other ethnic groups as opposed to being Yoruba decreases the probability of women’s household decision-making autonomy (7.1, 81.3 and 48.4 percentage points respectively) while practicing Islam, the traditional African religions and other religions as opposed to practicing Christianity is associated with decreased women’s household decision-making autonomy (61.9, 73.3 and 97.3 percentage points respectively). Women’s earnings decision-making autonomy is positively associated with having more children and marriage in a polygamous household (0.3 and 0.9 percentage points respectively), while condoning wife beating is associated with reduced likelihood of women’s earnings decision- making autonomy (1.1 percentage points). Again ethnicity and religion are strongly associated with decision-making autonomy- being Hausa, or Fulani as opposed to being Yoruba and practicing the traditional African religion as opposed to practicing Christianity decreases the probability of women’s earnings decision-making autonomy (3.3, 9.4 and 10.7 percentage points respectively), while practicing Islam as opposed to Christianity increases the probability of women’s earnings decision-making autonomy by 1.3 percentage points. Our finding of the association between being an ethnic group other than Yoruba and practicing other religions as opposed to Christianity with lower household and earnings decision-making autonomy underscore the significance of ethnic and religious social variation as factors determining women’s empowerment are consistent with other research on Nigeria (Fuseini and Kalule-Sabiti 2015; Kritz and Makinwa-Adebusoye 1999). To sum up, the presence of BH is associated with large increases in recognized drivers of IPV, controlling behavior of husband/ partners and women’s lack of agency as proxied by autonomy in household decision-making. It increases the probability of women’s experience of controlling behavior by husband/partner by 13.8 percentage points and reduces the probability of women’s household decision-making autonomy by 22.4 percentage points. However, the correlates of these drivers differ. While controlling behavior is mitigated by women’s earnings of any kind and exacerbated by women’s attitudes to and experience of IPV and her husband/ partner’s use of alcohol, cultural factors such as religion and ethnicity are strongly associated with women’s household decision-making autonomy and being in a polygamous marriage, being married as a child and condoning IPV are associated with less decision-making autonomy for women. 5.3. Evidence in support of the DD estimation method A key identifying assumption in the DD model specified in equation (1) is that in the absence of the BH insurgency, the differences in the trends of IPV between the BH affected and unaffected areas would be the same. To reinforce the plausibility of our DD model, we proceed to a placebo test. While we cannot explicitly test this assumption, we present evidence in support of parallel trends. First, in Figure 4, we present graphs of the mean levels of IPV and its key drivers before and during the BH insurgency. The graphs suggest that in the period before the BH insurgency, 22 the trends in outcomes are parallel between the BH affected area for physical or sexual IPV, controlling behavior by husband/partner and women’s household decision-making autonomy as well as women’s earnings decision-making autonomy, suggesting that the key identifying assumption holds. Figure 4: Before and during BH trends Second, we present results from a placebo analysis. For this analysis, we examine only observations from the DV module of the 2008 NDHS- the unaffected cohort- and assume women interviewed in the first half of the survey period (June, July and August 2008) make up the placebo sample before BH insurgency while those interviewed in the second half of the survey period (September, October and November 2008) make up the placebo sample for the period during the BH insurgency. Since IPV outcomes were less favorable in the BH area, if we find significant coefficients, our conclusion is that factors other than BH are responsible for the outcomes and the association between BH and less favorable outcomes is spurious. The results from this estimation are presented in Table 7. We find no spurious significant coefficient. Although the coefficient of women’s household decision-making autonomy is significant, this seems unrelated to the BH insurgency, as the direction of the effect is not the same as we find in our main analysis. While this result is not proof of parallel trends in outcomes between the BH and non BH areas, it supports the conclusion that the significant effects of the BH insurgency on women’s experiences of IPV and its drivers are not spurious. 23 Table 7: Difference-in-differences estimates of the effect of the Boko Haram insurgency on Intimate partner violence, women’s experience of controlling behavior by husband/partner and women’s decision-making autonomy in Nigeria- Placebo Analysis. (1) (2) (3) (4) Variables Past year Controlling Women’s household Women’s earnings physical or behavior by decision- making decision-making sexual IPV husband/partner autonomy autonomy * -0.008 -0.018 0.236*** 0.017 (0.021) (0.043) (0.049) (0.016) Covariates Individual characteristics Yes Yes Yes Yes Partner characteristics Yes Yes Yes Yes Household characteristics Yes Yes Yes Yes Before BH placebo (Non-BH Area) 5985 5997 5994 4611 Before BH placebo (BH Area) 664 664 664 604 During BH placebo (Non-BH Area) 3136 3138 3138 2663 During BH placebo (BH Area) 460 461 461 433 Total 10245 10260 10257 8311 R-squared 0.14 0.15 0.31 0.12 Note: Variables used to estimate the kernel propensity scores are dropped are dropped. Note: The estimations account for state fixed effects to account for the clustering of IPV among women residing in the same state. The kernel density function is epanechnikov with a bandwidth of 0.06. Probit estimation is used for the propensity score in the first stage. Kernel-based DD estimates are on the common support. The results are robust to employing a logit estimation of the propensity scores in the first stage, a gaussian kernel density function as opposed to an epanechnikov kernel density function and varying the bandwidth of the kernel density function.***Significant at the 1% level, **Significant at the 5% level, *Significant at the 10% level. Standard errors in parentheses are robust. 5.4 Limitations of the study It is likely that the study underestimates the effect of the BH insurgency, as areas with the largest number of conflict events are not captured in our analysis. Also, the BH insurgency is a regional crisis and we do not consider BH activities in the neighboring countries of Cameroun, Chad and Niger, which affect Nigerian communities located in areas bordering these countries. Therefore, our estimates represent a lower bound for the effect of the BH insurgency. Another limitation is caused by the study reliance on GPS coordinates provided in the NDHS as well as the ACLED to identify the BH area. The geo-coordinates of the NDHS clusters are randomly displaced to protect the privacy of the respondents (Elkies et al. 2015). Urban clusters are displaced up to 2km while rural clusters are displaced up to 5km (Perez-Heydrichet al. 2013). The measurement error introduced might lead to attenuation bias in the regressions (Skiles et al. 2013), although robustness analysis does not suggest this is a problem in our estimations. 26 Furthermore, although the NDHS allows us to examine the effect of conflict on some drivers of IPV, it does not provide enough information to examine important pathways through which the BH insurgency may translate to IPV, such as extreme stress and trauma, changing marriage practices and other microsystem and exosystem factors suggested by the ecological framework. 26 We check if our results are likely to be affected by this bias by restricting the analysis in tables 4 and 6 to only urban areas where the offset is smaller. We find that our results are robust to this restriction (see appendix table A4a and A4b). 24 6. Conclusion Our findings add to a growing body of research that establishes a link between exposure to violent political conflict and IPV. We find that the presence of BH increases the probability that women experience physical or sexual IPV by about 4 percentage points after controlling for known correlates of IPV; partner’s alcohol use, previous exposure to IPV and condoning IPV as a social norm. Finding an association between conflict and higher IPV risk is noteworthy, as it is not revealed by descriptive statistics on changes in IPV prevalence rates in locations exposed to the BH insurgency. Controlling for differences in individual and household characteristics, descriptive statistics show that IPV rates in the BH affected areas remained unchanged from the period before the BH insurgency to the period during the BH insurgency. However, IPV declined by 5 percentage points in non-BH affected areas. Therefore, the association of an increased probability of women’s experience of physical or sexual IPV by about 4 percentage points with the BH insurgency implies that conflict has stopped progress towards the reduction of IPV that could otherwise have been achieved. We interpret our results as a lower bound of the average causal effect of the BH insurgency on physical or sexual IP, as there are no observations during the BH phase for many of the most severely impacted areas. Previous research has established the association between increased male perpetration of IPV with conflict-related trauma, exposure to political violence and behaviors that are associated with increased IPV risk, such as increased alcohol abuse (Bruck and Stojetz 2019; Clark et al. 2010; Gupta et al. 2009, 2012; Saile et al. 2013). Our results extend this literature through establishing a link between exposure to conflict and other behaviors that are known to increase IPV risk. We find that exposure to BH increases the probability of women’s experience of controlling behavior from husbands/partners by about 14 percentage points and decreases the probability of women’s autonomy in household decision-making by around 22 percentage points. Access to incomes for women is associated with reduced risk of experiencing controlling behavior from a male partner, which suggests that promoting women’s economic opportunities in conflict settings might have positive spillover effects on behavioral norms. These results add to the thin evidence base on how exposure to conflict may change interpersonal behaviors. Further research is required to understand the pathways through which conflict exacerbates women’s experience of IPV. IPV is a violation of women’s and girl’s human rights, a manifestation of gender inequality, and constrains and compromises human development. IPV response and prevention is therefore integral to the success of humanitarian and development policies and programs in conflict settings. Programs that respond to and prevent IPV include strengthening health and other services and facilitating easy and affordable or free access for survivors of IPV. Health services need to respond to post-traumatic psychological and mental health issues and substance abuse, due to their potential to be both a trigger and consequence of IPV. Programs to address and change harmful gender norms and behaviors at the individual and community levels are needed. Promising interventions include community, school and faith-based initiatives that can be used to affect wider community- level change (Glass et al., 2019; Palm, 2019; Siddiq, Hemat & Corboz, 2018). Maintaining an ‘all- of-community’ approach and ensuring programming is centered on women and girls and designed in a way to ensure accountability to them are critical elements to the success of these types of interventions (What Works to Prevent VAWG Programme, 2016). 25 References Adejimi, A. A., Fawole, O. I., Sekoni, O. O. and Kyriacou, D. N. 2014. “Prevalence and Correlates of Intimate Partner Violence among Male Civil Servants in Ibadan, Nigeria.” African Journal of Medicine and Medical Sciences, 43 (Suppl 1): 51–60. Akresh R., Bhalotra S., Leone M., and Osili O.U., 2012 War and Stature: Growing Up During the Nigerian Civil War. American Economic Review: Papers and Proceedings 102 (3): Abramsky T, Watts CH, Garcia-Moreno C, Devries K, Kiss L, Ellsberg M, et al. 2011 “What factors are associated with recent intimate partner violence? Findings from the WHO multi-country study on women's health and domestic violence.” BMC Public Health 111:1. Aizpurua E., Copp J., Ricarte J.J. and Vazquez 2017 “Controlling behaviors and intimate partner violence among women in Spain: An examination of individual, partner and relationship risk factors for physical and psychological abuse.” Journal of Interpersonal Violence. https://doi.org/10.1177/0886260517723744 Antai D.E. and Antai J.B. 2008. “Attitudes of women toward intimate partner violence: a study of rural women in Nigeria” The International Electronic Journal of Rural and Remote Health Research, Education, Practice and Policy. Antai D. 2011 “Controlling behavior, power relations within intimate relationships and intimate partner physical and sexual violence against women in Nigeria.” BMC Public Health 11: 511. doi: 10.1186/1471-2458-11-511 Asling-Monemi K et al. 2003. “Violence against women increases the risk of infant and child mortality: a case-referent study in Nicaragua.” Bulletin of the World Health Organization, 81(1):10–6 Babaro Nicole, September 2017. “How can evolutionary psychology help explain intimate partner violence?” Accessed 4th June 2019. https://behavioralscientist.org/can-evolutionary-psychology-help-explain-intimate- partner-violence/ Bahgat K., Dupuy K., Østby G., Rustad S.S., Strand H. and Wig T. 2018 “Children Affected by Armed Conflict, 1990-2016” Peace Research Institute Oslo (PRIO). Conflict Trends Project www.prio.org/ConflictTrends Berkowitz L. 1993. Aggression: its causes, consequences, and control. Philadelphia: Temple University Press. Bloom, S.S., Wypij, D. & Das Gupta, M. 2001 “Dimensions of women’s autonomy and the influence on maternal health care utilization in a north Indian city.” Demography 38: 67. https://doi.org/10.1353/dem.2001.0001 Blundell, R., Dias, M. 2009. "Alternative Approaches to Evaluation in Empirical Microeconomics". Journal of Human Resources, Vol. 44, No. 3. pp. 565-640. Breiding, M. J., Chen, J., & Black, M. C. 2014. “Intimate partner violence in the United States—2010.” Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention Bruck, T. & Stojetz 2019. “Intimate Partner Violence and the Perpetrator’s Past: Causal Long-run Evidence from Angolan Families.” Unpublished Manuscript Buvinic M., Gupta M.D., Casabonne U., & Verwimp P. 2013 “Violent Conflict and Gender Inequality: An overview” The World Bank Group Policy Research Working Paper 6371 Caliendo M., Kopeinig S., 2005. “Some practical guidance for the implementation of propensity 26 score matching.” IZA Discussion Paper No. 1588 Campagne A. Van Lookeren and Begum S 2017.” Red Gold and Fishing in the Lake Chad Basin: Restoring Destroyed Livelihoods and Protecting People in Niger’s Diffa Region”, (Oxfam, February). Campbell J.C. 2002. “Health consequences of intimate partner violence.” The Lancet, 3599314: 1331–1336. Capaldi, D. M., Knoble, N. B., Shortt, J. W., & Kim, H. K. 2012. “A systematic review of risk factors for intimate partner violence.” Partner Abuse, 3, 231-280. doi:10.1891/1946- 6560.3.2.231 Centre for Population and Reproductive Health, 2016, “Sexual and Gender Based Violence Assessment in North East Nigeria,” September. Ceriani L., Verme P., 2018 “Risk preferences and the decision to flee conflict.” The World Bank Group Policy Research Working Paper 8376 Champion J., & Shain R., 1998. “The context of sexually transmitted disease: life histories of woman abuse”. Issues in Mental Health Nursing, 19(5):463–79. Chukwuma A., Ekhator-Mobayode U.E., 2019. “Armed conflict and maternal health care utilization: evidence from the Boko Haram Insurgency in Nigeria.” Social Science & Medicine 226, 104-112 https://doi.org/10.1016/j.socscimed.2019.02.055. Clark, C.J., S.A. Everson-Rose, S. Suglia, R. Btoush, A. Alonso, and M.M. Haj-Yahia. 2010. “Association between Exposure to Political Violence and Intimate-Partner Violence in the Occupied Palestinian Territory: A Cross-Sectional Study.” The Lancet 3759711: 310–316. Crowne, S. Joon, H., et al. 2011. “Concurrent and Long-Term Impact of Intimate Partner Violence on Employment Stability.” Journal of Interpersonal Violence 26: 1282-1304 Devries, Karen & Mak, Joelle & García-Moreno, C & Petzold, Max & C Child, J & Falder, G & Lim, S & Bacchus, Loraine & Engell, Rebecca & Rosenfeld, L & Pallitto, C & Vos, Theo & Abrahams, Naeemah & Watts, Charlotte. 2013. “The Global Prevalence of Intimate Partner Violence Against Women.” Science (New York, N.Y.). 340. 10.1126/science.1240937. Dietrich K. 2015 “When We Can’t See the Enemy, Civilians Become the Enemy: Living Through Nigeria’s Six Year Insurgency.” Center for Civilians in Conflict. Dobash R.E, Dobash R. 1979 Violence against wives: a case against the patriarchy. New York: The Free Press Duvvury Nata, Aoife Callan, Patrick Carney, and Srinivas Raghavendra. 2013 “Intimate Partner Violence: Economic Costs and Implications for Growth and Development”. Women’s Voice and Agency, and Participation Research Paper 3, World Bank, Washington, DC. Duvvury Nata, Huu Minh Nguyen and Patricia Carney. 2012. “Estimating the Cost of Domestic Violence against Women in Vietnam.” Hanoi: UN Women. Ekhator-Mobayode, U.E., Abebe Asfaw, A., 2019. “The child health effects of terrorism: evidence from the Boko Haram Insurgency in Nigeria.” Appl. Econ. 51 (6), 624–638. Elkies, Noam, Fink, Gunther, Barnighausen, Till, 2015. “Scrambling’ geo-referenced data to protect privacy induces bias in distance estimation.” Popul. Environ. 37 (1), 83–98. Ellsberg M., Arango D.J., Morton M., Gennari F., Kiplesund S., Contreras M., Watts C. 2014. “Prevention of violence against women and girls: what does the evidence say?” The Lancet. doi: 10.1016/S0140-67361461703-7 Falb K.L., McCormick M.C., Hemenway D., Anfinson K., Silverman J.G. 2014. “Symptoms 27 Associated with Pregnancy Complications Along the Thai-Burma Border: The Role of Conflict Violence and Intimate Partner Violence.” Maternal and Child Health Journal. Falb K.L., McCormick M.C., Hemenway D., Anfinson K., & Silverman J.G.2013. “Violence against refugee women along the Thai-Burma border.” Int J Gynaecol Obstet (120):279– 83. Fuseini K. and Kalule-Sabiti S. 2015. “Women’s autonomy in Ghana: does religion matter?” African Population Studies. Vol. 29, No. 2, 2015 Gallegos, J., and I. Gutierrez. 2016. “The Effect of Civil Conflict on Domestic Violence: The Case of Peru.” RAND Working Paper Series WR-1168. RAND, Santa Monica, CA. García-Moreno C., Zimmerman C., Morris-Gehring A., Heise L., Amin A., Abrahams N., Montoya O., Bhate-Deosthali P., Kilonzo N., Watts C. 2014. “Addressing violence against women: a call to action.” The Lancet. doi: 10.1016/S0140-67361461830-4. Georgetown Institute for Women, Peace and Security GIWPS and Peace Research Institute Oslo (PRIO) 2017. “Women, Peace and Security Index 2017/18: Tracking Sustainable Peace through Inclusion, Justice, and Security for Women”. Washington, DC: GIWPS and PRIO. Georgetown Institute for Women, Peace and Security (GIWPS) and Peace Research Institute Oslo 2019. “Women, Peace and Security Index 2019/20: Tracking sustainable peace through inclusion, justice, and security for women.” Washington, DC: GIWPS and PRIO. Georgetown Institute for Women, Peace and Security (GIWPS) 2018. “Nigeria Country Profile.” https://giwps.georgetown.edu/country/nigeria/ Glass, N., Perrin, N., Marsh, M., Clough, A., Desgroppes, A., Kaburu, F., Ross, B., Read- Hamilton, S. 2019. Effectiveness of the Communities Care programme on change in social norms associated with gender-based violence (GBV) with residents in intervention compared to control districts in Mogadishu, Somalia. BMJ Open, 9. Global Women’s Institute and International Rescue Committee. 2016. Evidence brief: What works to prevent and respond to violence against women and girls in conflict and humanitarian settings? Washington DC and London: The George Washington University and the International Rescue Committee. Goode WJ 1971. “Force and violence in the family.” J Marriage Fam 33: 624–36. Golding JM 1999. “Intimate partner violence as a risk factor for mental disorders: a meta- analysis.” Journal of Family Violence, 14(2):99–132. Gupta J, Acevedo-Garcia D, Hemenway D, et al. 2009 “Premigration exposure to political violence and perpetration of intimate partner violence among immigrant men in Boston.” Am J Public Health 99:462–9. Gupta J, Reed E, Kelly J, et al 2012 “Men's exposure to human rights violations and relations with perpetration of intimate partner violence in South Africa.” J Epidemiol Community Health 66:e2. Haj-Yahia M.M and Clark C.J. 2013. “Intimate partner violence in the occupied Palestinian Territory: Prevalence and Risk Factors.” Journal of Family Violence 28:797-809 Hanmer, L., and J. Klugman. 2016. “Exploring women’s agency and empowerment in developing countries: Where do we stand?” Feminist Economics 22(1): 237–263. Heise, L. 1998. “Violence against women: An integrated, ecological framework.” Violence Against Women, 4, 262-290. Heise, L. 2011. “What works to Prevent Partner Violence? An Evidence Overview.” Working Paper 2.0., United Kingdom Department for International Development. 28 Hemba J 2014. “Nigerian Islamists kill 59 pupils in boarding school attack,” Reuters, 26 February. Holt S., Buckley H. & Whelan S. 2008. “The impact of exposure to domestic violence on Children and young people: a review of the literature.” Child Abuse & Neglect, 32(8):797–810 Horn, R. 2010. “Responses to Intimate Partner Violence in Kakuma Refugee Camp: Refugee Interactions with Agency Systems.” Social Science & Medicine 70(1): 160–168. Horn R, Puffer Eve S., Roesch E., Lehmann H., 2014 “Women’s perception of the effects of war on intimate partner violence and gender roles in two post conflict West African Countries: consequences and unexpected opportunities”. Conflict and Health Howell E., Waidmann T., Holla N., Birdsall N., and Jiang K. 2018. “The Impact of Civil Conflict on Child Malnutrition and Mortality, Nigeria, 2002-2013.” CGD Working Paper 494. Washington, DC: Center for Global Development. https://www.cgdev.org/publication/impact-civil-conflict-child-malnutrition-and- mortality-nigeria IASC (Inter-Agency Standing Committee). 2015. Guidelines for Integrating Gender-Based Violence Interventions in Humanitarian Action: Reducing risk, promoting resilience and aiding recovery. Institute for Women’s Studies 2008. “The impacts of Israeli mobility restrictions and violence on gender relations in Palestinian society: 2000–2007.” Background report prepared for the World Bank. Palestine: Institute for Women’s Studies, Birzeit University. International Crisis Group, “Nigeria: Women and the Boko Haram Insurgency,” December 2016. IRC (International Rescue Committee) 2012. Let Me Not Die Before My Time: Domestic Violence in West Africa. New York. IRC (International Rescue Committee) 2015 “Private violence, Public concern: Intimate Partner Violence in humanitarian settings.” Practice brief, January 2015. http://educationcluster.net/wp-content/uploads/sites/3/2015/04/Private-Violence-Public- Concern-2015.pdf accessed on 2nd Dec 2019. IRC (International Rescue Committee 2017a, “Protection and Empowerment Baseline Survey”, March 2017 IRC (International Rescue Committee) 2017b, “Lake Chad Basin Crisis: An Analysis of Violence Against Women, Children and Displaced Populations in the Region.” February. https://www.rescue-uk.org/sites/default/files/document/1328/reportoslo- analysisofviolenceagainstwomenchildrenanddisplacedpopulations-17march2017.pdf , accessed October 17, 2019 Jann B. 2017. "kmatch: Kernel matching with automatic bandwidth selection," United Kingdom Stata Users' Group Meetings 2017.11, Stata Users Group. Joint Shadow Report CEDAW Committee, 67th Session July 2017. “Women, Peace and Security in Nigeria.” Women’s International League for Peace and Freedom. Justino P. 2018 “Violent Conflict and Changes in Gender Economic Roles: Implications for Post-conflict Economic Recovery.” The Oxford Handbook of Gender and Conflict Kelly J.T.D., Colantuoni E., Robinson C., & Decker M.R. 2018 “From the battlefield to the bedroom: a multilevel analysis of the links between political conflict and intimate partner violence in Liberia.” BMJ Glob Health 3:e000668. doi:10.1136/bmjgh-2017-000668 29 Kethineni, S., Srinivasan, M. & Kakar, S. 2016. “Combating Violence against Women in India: Nari Adalats and Gender-Based Justice.” Women and Criminal Justice 26 (4): 281– 300. Kiss, L., Schraiber, L. B., Hossain, M., Watts, C., & Zimmerman, C. 2015. “The Link Between Community-Based Violence and Intimate Partner Violence: the Effect of Crime and Male Aggression on Intimate Partner Violence Against Women.” Prevention Science, 16(6) 881–889. doi:10.1007/s11121-015-0567-6 Khawaja M. 2004. “Domestic violence in refugee camps in Jordan.” International Journal of Gynecology & Obstetrics, 86(1), 67–69. Klugman, Jeni; Hanmer, Lucia; Twigg, Sarah; Hasan, Tazeen; McCleary-Sills, Jennifer; Santamaria, Julieth 2014.Voice and Agency: Empowering Women and Girls for Shared Prosperity. Washington, DC: World Bank Group. © World Bank. https://openknowledge.worldbank.org/handle/10986/19036 License: CC BY 3.0 IGO Krug EG et al., 2002. World Report on Violence and Health. Geneva, World Health Organization 87– 121 Knight K.E., Menard S., Simmons S., Bouffard L.A. and Orsi R. 2013. “Generational cycles of Intimate partner violence in the US: A research brief.” Crimes Victims’ Institute, College of Criminal Justice, Sam Houston State University. Kritz, M.M. & Makinwa-Adebusoye, P. “Determinants of Women’s Decision-Making Authority in Nigeria: the Ethnic Dimension.” Sociological Forum 1999 14: 399. https://doi.org/10.1023/A:1021495418633 Lindhorst, T., Oxford, M., & Gillmore, M. R. 2007. “Longitudinal effects of domestic violence on employment and welfare outcomes.” Journal of Interpersonal Violence 22:812-828. Linos N, Slopen N. Subramanian S.V., Berkman L. and Kawachi I 2013. Influence of community social norms on spousal violence: “A population-based multilevel study of Nigerian Women” American Journal of Public Health. Vol 103, No 1 Lipsky S. Caetano R., Field C.A. and Larkin G.L. 2005. “Psychosocial and Substance-use Risk Factors for Intimate Partner Violence” Drug and Alcohol Dependence 78;1 pg 39-47 McCarthy KJ, Mehta R, Haberland NA 2018. “Gender, power, and violence: A systematic review of measures and their association with male perpetration of IPV.” PLOS ONE 13(11):e0207091. https://doi.org/10.1371/journal.pone.0207091 Mchale T., Hadley M., Thomson D. 2011. “Every home has its secrets”: A mixed-methods Study on intimate partner violence, women’s empowerment and justice on Ijdwi island, Democratic Republic of the Congo.” Harvard Humanitarian Initiative Student Working Paper Series. Nagarajan Chitra., 2017. “Gender Assessment of Northeast Nigeria.” Narayan, D., 2000. Voices of the Poor: Can anybody hear us? Washington D.C.: World Bank. National Consortium for the Study of Terrorism and Responses to Terrorism (START). 2014. “Boko Haram Recent Attacks. Background Report”. Accessed May 14 2018. https://www.start.umd.edu/pubs/STARTBackgroundReport_BokoHaramRecentAttacks_ May2014_0.pdf National Consortium for The Study of Terrorism and Responses to Terrorism (START). 2016. Global Terrorism Database [Data file]. Accessed 14th May 2018. URL: https://www.start.umd.edu/gtd National Population Commission (NPC) [Nigeria] and ICF Macro. 2009. Nigeria Demographic 30 and Health Survey 2008. Abuja, Nigeria: National Population Commission and ICF Macro National Population Commission (NPC) [Nigeria] and ICF International. 2014. Nigeria Demographic and Health Survey 2013. Abuja, Nigeria, and Rockville, Maryland, USA: NPC and ICF International. Nwabunike Collins and Tenkorang Eric Y. 2015 “Domestic and marital violence among three ethnic groups in Nigeria” Journal of Interpersonal Violence. https://doi.org/10.1177/0886260515596147 Odimegwu, C., Bamiwuye, O., & Adedini, S. 2015. “Gender-based violence as a new proximate determinant of fertility in sub-Saharan Africa.” Southern African Journal of Demography,16 (1), 87-121. Retrieved from http://www.jstor.org/stable/soutafrijourdemo.16.1.87 Odimegwu C. & Okemgbo C.N 2003 “Gender Role Ideologies and Prevalence of Violence Against Women in Imo State, Nigeria.” The Anthropologist, 5:4,225- 236, DOI: 10.1080/09720073.2003.11890814 Ofie-Aboagye RO 1994. “Domestic violence in Ghana: an initial step.”Colum J Gender Law ; 4(1): 1- 25. Okemgbo C.N., Omideyi A.K., and Odimegwu C.O. 2002. “Prevalence, patterns and correlates of domestic violence in selected Igbo communities of Imo State, Nigeria.” African Journal of Reproductive Health;6(2): 101- 114. Østby G. 2016. “Violence Begets Violence: Armed Conflict and Domestic Sexual Violence in Sub-Saharan Africa.” HiCN Working Paper 233 Oyediran, K., & Isiugo-Abanihe, U. 2005. P”erceptions of Nigerian Women on Domestic Violence: Evidence from 2003 Nigeria Demographic and Health Survey.” African Journal of Reproductive Health 9(2), 38-53. doi:10.2307/3583461 Palermo T, Bleck J, Peterman A. “Tip of the Iceberg: Reporting and Gender-Based Violence in Developing Countries.” American Journal of Epidemiology. 2014;179(5):602-612. doi:10.1093/aje/kwt295. Palm, S., Le Roux, E., Bezzolato, E., Deepan, P., Corboz, J. Lele, U., O’Sullivan, V & Jewkes, R. 2018. Rethinking Relationships: Moving from Violence to Equality. What works to prevent violence against women and girls in the DRC. Perez-Heydrich, Carolina, Warren, Joshua L., Emch, Michael E., 2013. “Guidelines on the Use of DHS GPS Data. Spatial Analysis Reports.”www.measuredhs.com, Accessed date: 4 January 2019. Pittaway E. 2005. “The Ultimate Betrayal: The experience of domestic and family violence in refugee communities.” The Centre for Refugee Research, The University of New South Wales Promundo. 2016. “Living Peace: Scaling Up Healing After Conflict in Democratic Republic of the Congo.” Rio de Janeiro. http:/ /promundoglobal.org/resources/living-peace-scaling - up-healing-after-conflict-in-democratic-republic-of-the -congo/ Read Joe 2017. “Sexual Violence and the Boko Haram Crisis in North-East Nigeria” Humanitarian Practice Network. Accessed May 15th, 2018. https://odihpn.org/magazine/sexual-violence-and-the-boko-haram-crisis-in-north-east- nigeria/ Reeves, C., & O'Leary-Kelly, A. 2007. “The effects and costs of intimate partner violence for work organizations.” Journal of Interpersonal Violence 22(3): 327-44. 31 Raghavendra S., Duvvury N. and Ashe S. 2017 “The Macroeconomic Loss due to Violence Against Women: The Case of Vietnam.” Feminist Economics. https://doi.org/10.1080/13545701.2017.1330546 Roberts L.A, Gilman S.E., Fitzmaurice G., Decker M.R, Koenen C.K. 2011 “Witness of Intimate Partner Violence in Childhood and Perpetration of Intimate Partner Violence in Adulthood.” Epidemiology. 21(6): 809–818 doi:10.1097/EDE.0b013e3181f39f03 Saile, R., E.V. Ertl, F. Neuner, and C. Catani. 2013. “Prevalence and Predictors of Partner Violence against Women in the Aftermath of War.” Social Science & Medicine 86: 17–25. Shalhoub-Kevorkian N. 2009. Militarization and violence against women in conflict zones in the Middle East: a Palestinian case-study. Cambridge: Cambridge University Press. Silverman JG et al. 2009. “Maternal experiences of intimate partner violence and child morbidity in Bangladesh: evidence from a national Bangladeshi sample.” Archives of Pediatrics & Adolescent Medicine, 163(8):700–05 Siddiq, W., Hemat, O., & Corboz, J. (2018). “What Works to Prevent Violence Against Children in Afghanistan? Findings from an evaluation of a school-based peace education and community social norms intervention.” South African Medical Research Council & Help the Afghan Children. Skiles, Martha Priedeman., et al. “Geographically linking population and facility surveys: methodological considerations.” Population Health Metrics. 2013; 11(14). doi:10.1186/1478-7954-11-14 Slegh, H., G. Barker, B. and Levtov, R. 2014. “Gender Relations, Sexual Violence and the Effects of Conflict on Women and Men in North Kivu, Eastern Democratic Republic of Congo: Results from the International Men and Gender Equality Survey (IMAGES).” Washington, DC and Cape Town South Africa: Promundo US and Sonke Gender Justice Network. Stark L., Ager A. 2011. “A Systematic Review of Prevalence Studies on Gender-Based Violence in Complex Emergencies. Trauma, Violence and Abuse.” 12(3) 127–134. Straus, M.A. 1979. “Measuring intrafamily conflict and violence: The conflict tactics (CT) scales.” Journal of Marriage and the Family 41(1): 75-88. Straus, M.A. 1990. “Measuring intrafamily conflict and violence: The Conflict Tactics (CT) Scales.” In M.A. Straus and R.J. Gelles (eds.), Physical violence in American families: Risk factors and adaptations to violence in 8,145 families. New Brunswick, New Jersey, USA: Transaction Publishers. Stuart, E. A., Huskamp, H. A., Duckworth, K., Simmons, J., Song, Z., Chernew, M. E., & Barry, C. L. 2014. “Using propensity scores in difference-in-differences models to estimate the effects of a policy change.” Health Services and Outcomes Research Methodology, 14(4), 166-182. DOI: 10.1007/s10742-014-0123-z Sunita K. and Johnson K. 2004. Profiling Domestic Violence – A Multi-Country Study. Calverton, Maryland: ORC Macro. Tenkorang E. 2018 “Women’s autonomy and intimate partner violence in Ghana”. International Perspectives on Sexual and Reproductive Health. DOI: https://doi.org/10.1363/44e6118 Thompson M.P. & Kingree J.B. 2006. “The Rles of Victim and Perpetrator Alcohol Use in Intimate Partner Violence Outcomes”. Journal of Interpersonal Violence 21:2 https://doi.org/10.1177/0886260505282283 Thornberry, T. P., Knight, K. E., & Lovegrove, P. J. 2012. “Does maltreatment beget maltreatment? A systematic review of the intergenerational literature.” Trauma, Violence, 32 Abuse, 13(3), 131–112 Tumwesigye N.M., Kyomuhendo G.B., Greenfield K.T., and Wanyenze K.R 2012. “Problem drinking and physical intimate partner violence against women: evidence from a national survey in Uganda.” BMC Public Health. 12:399 UNDP 2018. National Human Development Report: Nigeria. http://hdr.undp.org/en/content/national-human-development-report-2018-nigeria UNHCR and Caux H., October 2018, “Participatory protection assessment report: Adamawa State, Nigeria.” Accessed 4th June 2019. URL: https://data2.unhcr.org/en/documents/download/67391 UNHCR-United Nations High Commissioner for Refugees:2018. “Every day we Pray for Protection and Peace” UNHCR Accessed 14th May 2018. URL: http://www.unhcr.org/en-us/nigeria-emergency.html?query=boko%20haram United Nations 2017. “Report of the Secretary-General on Conflict-Related Sexual Violence.” Report S/2017/249, United Nations, New York United Nations 2019. “Report of the Secretary-General on Conflict-Related Sexual Violence.” Report S/2019/280, United Nations, New York United Nations Security Council. Resolution 1325. New York, NY, USA: United Nations Security Council, 2000. United Nations; World Bank 2018.Pathways for Peace: Inclusive Approaches to Preventing Violent Conflict. Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/28337 License: CC BY 3.0 IGO.” UNOCHA- United Nations Office of Coordination of Humanitarian Affairs 2017. “Nigeria- Northeast: Humanitarian Emergency Situation Report No 4 as of 31 January 2017.” March UNOCHA 2018: “Humanitarian Needs Overview- Nigeria” Villa, J.M., 2016. “Simplifying the estimation of difference-in-differences treatment effects.” Stata Journal 16, pp. 52-71 Vu A., Adam A., Wirtz A., Pham K., Rubenstein L., Glass N., Beyrer C., Singh S. 2014. “The Prevalence of Sexual Violence among Female Refugees in Complex Humanitarian Emergencies: A Systematic Review and Meta-Analysis.” PLOS Current Disasters, 18(1). doi: 10.1371/currents.dis.835f10778fd80ae031aac12d3b533ca7. Walker, L. A. 2006. “Battered woman syndrome: Empirical findings.” Annals of New York Academy of Sciences, 1087, 142-157. doi:10.1196/annals.1385.023 Ward, J. and Brewer, J. 2004. “Gender-based violence in conflict-affected settings: overview of a multi-country research project” Forced Migration Review. Vol. 19, 26-28. What Works to Prevent VAWG Programme 2016. “What Works to Prevent Violence against Women and Girls in Conflict and Humanitarian Crisis: Synthesis Brief” https://www.whatworks.co.za/documents/publications/355-p868-irc-synthesis-brief-report-lr- 26092019/file Widom, C. S., & Wilson, H. W. 2015. “Intergenerational transmission of violence.” In J. Lindert & I. Levav (Eds.), Violence and mental health: Its manifold faces 27-45. New York, NY: Springer Wilkins, N., Tsao, B., Hertz, M., Davis, R., Klevens, J. 2014. “Connecting the Dots: An Overview of the Links Among Multiple Forms of Violence.” Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention Oakland, CA: Prevention Institute. Women’s Commission for Refugee Women and Children 2005 “Masculinities: Male Roles and 33 Male Involvement in the Promotion of Gender Equality. – A Resource Packet”. ISBN: 1-58030-040-5 Accessed 19th May 2019 https://www.unicef.org/emerg/files/male_roles.pdf World Bank 2018 “GBV Risk Assessment Portfolio Review, Nigeria” World Health Organization 2013. Global and regional estimates of violence against women: prevalence and health effects of intimate partner violence and non-partner sexual violence. Switzerland, World Health Organization World Health Organization (WHO) 2014 updated. “Fact Sheet No. 239: Violence against Women”, www.who.int/mediacentre/factsheets/fs239/en World Health Organization (WHO) 2017 (updated). “Fact Sheet: Violence against Women”, https://www.who.int/news-room/fact-sheets/detail/violence-against-women Accessed 1st September 2019 Wirtz, A., K. Pham, N. Glass, S. Loochkartt, T. Kidane, D. Cuspoca, L.S. Rubenstein, S. Singh, and A. Vu. 2014. “Gender-based violence in conflict and displacement: Qualitative findings from displaced women in Colombia.” Conflict and Health 8(1): 10. Yoshikawa, K., Shakya, T. M., Poudel, K. C., & Jimba, M. 2014. “Acceptance of wife beating and its association with physical violence towards women in Nepal: a cross-sectional study using couple's data.” PlOS ONE, 9(4), e95829. doi:10.1371/journal.pone.0095829 34 Appendix A1. BH Events in Nigeria Table A1: Distribution of BH events in Nigeria (2009 to 2013) Geo-political State LGA 2009 2010 2011 2012 2013 Total zone North Central Benue Gboko 0 0 0 0 1 1 Oturkpo 0 0 0 1 0 1 Federal Capital Territory (FCT) Bwari 0 0 4 3 1 8 Kogi Adavi 0 0 0 2 0 2 Kotonkar 0 0 0 1 0 1 Lokoja 0 0 0 1 0 1 Okehi 0 0 0 1 0 1 Nassarawa Lafia 0 0 0 1 1 2 Niger Chanchaga 0 0 1 3 0 4 Lapai 0 0 2 0 0 2 Suleja 0 0 1 2 0 3 Plateau Jos North 0 2 5 6 0 13 Total 0 2 13 21 3 39 North East Adamawa Demsa 0 0 1 0 0 1 Ganye 0 0 0 0 2 2 Girie 0 0 2 0 1 3 Hong 0 0 0 5 2 7 Madagali 0 0 0 1 2 3 Maiha 0 0 0 1 0 1 Michika 0 0 0 0 1 1 Yola South 0 0 0 3 1 4 Bauchi Bauchi 2 3 5 7 0 17 Darazo 0 0 0 0 1 1 Gamjuwa 0 0 0 2 0 2 Itas/Gad 0 0 0 1 0 1 Jama're 0 0 0 1 0 1 Katagum 0 0 1 0 0 1 Misau 0 0 1 0 0 1 Ningi 0 0 0 1 1 2 Borno Abadam 0 0 0 0 2 2 Askira/U 0 0 0 2 4 6 Bama 0 0 0 3 4 7 Biu 0 0 3 7 2 12 Chibok 0 0 0 6 0 6 Damboa 0 0 2 4 23 29 Gubio 0 0 0 2 0 2 Guzamala 0 0 0 2 4 6 Gwoza 0 0 0 0 22 22 Hawul 0 0 0 0 3 3 Jere 0 0 0 1 2 3 Kaga 0 0 0 0 6 6 Kala/Balge 0 0 0 1 0 1 Konduga 0 1 0 14 33 48 Kukawa 0 0 0 0 10 10 Mafa 0 0 1 1 4 6 Magumeri 0 0 0 11 1 12 Maidugur 10 27 73 104 60 274 Marte 0 0 0 1 3 4 Mobbar 0 0 0 2 2 4 Ngala 1 0 0 4 7 12 Nganzai 0 0 0 1 6 7 Shani 0 0 1 0 0 1 Table A1 continued. 35 Table A1 continued. Geo-political zone State LGA 2009 2010 2011 2012 2013 Total Gombe Akko 1 0 0 7 2 10 Funakaye 0 0 0 1 0 1 Kwami 0 0 1 0 0 1 Nafada 0 0 1 0 1 2 Taraba Ibi 0 0 0 1 0 1 Takum 0 0 0 1 1 2 Yobe Damaturu 1 1 7 24 7 40 Fika 0 0 0 1 0 1 Fune 0 0 0 1 2 3 Geidam 0 0 1 1 2 4 Gujba 0 0 0 2 6 8 Gulani 0 1 0 0 2 3 Karasuwa 0 1 0 0 0 1 Nangere 4 0 0 26 3 33 Potiskum 1 0 0 0 2 3 Tarmuwa 0 0 0 0 1 1 Yusufari 0 0 0 1 1 2 Total 20 34 100 254 239 647 North West Jigawa Biriniwa 0 0 0 1 0 1 Guri 0 0 0 0 1 1 Kaduna Birnin-G 0 0 0 0 1 1 Chikun 0 0 0 1 0 1 Igabi 0 0 0 4 0 4 Ikara 0 0 0 1 0 1 Kaduna North 0 0 0 1 0 1 Kaduna South 0 0 1 11 3 15 Kagarko 0 0 0 0 1 1 Kajuru 0 0 0 1 0 1 Kudan 0 0 0 1 0 1 Sabon-Ga 0 0 0 7 0 7 Kano Bunkure 0 0 0 0 1 1 Gezawa 0 0 0 0 2 2 Gwale 0 0 0 0 1 1 Minjibir 0 0 0 0 1 1 Nassaraw 0 0 2 39 14 55 Tarauni 0 0 0 1 0 1 Ungogo 0 0 0 0 2 2 Warawa 0 0 0 1 0 1 Wudil 1 0 0 0 0 1 Katsina Bakori 0 1 0 0 0 1 Kankara 0 0 1 0 0 1 Kebbi Gwandu 0 0 2 0 0 2 Sokoto Silame 0 0 0 1 0 1 Wamakko 0 0 0 2 1 3 Yabo 0 0 0 1 0 1 Total 1 1 6 73 28 109 36 Table A1 Continued. Geo-political State LGA 2009 2010 2011 2012 2013 Total zone South East Enugu Enugu North 0 0 0 1 0 1 Total 0 0 0 1 0 1 South South Edo Etsako We 0 0 0 1 0 1 Total 0 0 0 1 0 1 South West Lagos Ikeja 0 0 0 1 0 1 Ikorodu 0 0 0 1 0 1 Total 0 0 0 2 0 2 Grand Total 21 37 119 352 270 799 Source: Authors own calculation from ACLED BH Actor file 37 A2. Definition of Physical and Sexual IPV in the NDHS Table A2. Questions asked to women in the NDHS to determine experience of physical or sexual violence. Physical violence Sexual violence Does (did) your (last) husband/partner ever do Does (did) your (last) husband/partner ever do any of the following things to you? any of the following things to you? (a) Push you, shake you, or throw something at you? (a) Physically force you to have sexual intercourse (b) Slap you? with him even when you did not want to? (c) Twist your arm or pull your hair? (b) Physically force you to perform any other sexual (d) Punch you with his fist or with something that acts you did not want to? could hurt you? (c) Force you with threats or in any other way to (e) Kick you, drag you, or beat you up? perform sexual acts you did not want to? (f) Try to choke you or burn you on purpose? (g) Threaten or attack you with a knife, gun, or any other weapon Source: National Population Commission (NPC) [Nigeria] and ICF International. 2014. Nigeria Demographic and Health Survey 2013. Abuja, Nigeria, and Rockville, Maryland, USA: NPC and ICF International. 38 A3. Matching Table A3 presents the summary statistics in the period before the BH insurgency for the matched and unmatched sample in the BH and non-BH areas, to examine if the composition of the sample is reasonably balanced. Figure A3 also presents the kernel densities of the propensity scores before and after matching. Over 99 percent of the sample in the period before the insurgency is on common support. The differences in means of the key covariates in the BH affected and unaffected areas but for women’s age are significantly different. However, after matching, the differences in means of all the key covariates are not significantly different. This provides evidence that the composition of the sample is reasonably balanced on the selected set of observable characteristics between women residing in the BH affected area as well as women residing in the BH unaffected area after matching. Table A3: Balancing Test for Before-BH insurgency sample Unmatched Matched (Common support) Non-BH Area BH Area P value Non-BH Area BH Area P value Variables Individual characteristics Age (years) 31.74 32.05 0.1194 32.22 31.92 0.4984 Primary educ 0.26 0.18 <0.0001 0.19 0.17 0.3929 Secondary educ 0.25 0.30 0.0005 0.30 0.30 0.9406 Higher educ 0.07 0.14 <0.0001 0.15 0.14 0.7720 Partner characteristics Age (years) 41.47 42.10 0.0369 41.99 41.90 0.8539 Primary educ 0.24 0.18 <0.0001 0.19 0.19 0.4766 Secondary educ 0.30 0.30 0.490 0.32 0.29 0.4799 Higher educ 0.12 0.22 <0.0001 0.22 0.23 0.7378 Household characteristics Household size 5.64 5.92 0.0009 5.64 5.81 0.3667 Urban 0.26 0.65 <0.0001 0.65 0.65 0.9698 Poor 0.22 0.17 0.0003 0.15 0.18 0.5389 Middle 0.19 0.15 0.0008 0.13 0.16 0.3767 Rich 0.18 0.21 0.0119 0.18 0.19 0.5918 Richest 0.16 0.40 <0.0001 0.43 0.41 0.7213 Number of observations 9590 1126 10716 9585 1126 10711 Note: Data are % or as indicated. The p-values are for the two-sample t-test with equal variances (Ho: Difference in means =0 and Ha: Difference in means ≠0). Variables capturing employment status are omitted because there is not enough variation. ***Significant at the 1% level, **Significant at the 5% level, *Significant at the 10% level. 39 Figure A3: Kernel density graph before and after matching 40 A4. Restricting the analysis to only urban areas Table A4a: The effect of the Boko Haram insurgency on Intimate Partner Violence in Nigeria- Urban sample Variables Past year physical or sexual IPV * 0.035** (0.018) Individual characteristics Type of earnings (reference category: no earnings) Cash only -0.027 (0.022) Cash and kind -0.062** (0.027) Kind only -0.044* (0.044) Total number of children born 0.004** (0.002) Married as a child 0.018* (0.010) Partner characteristics Sometimes or often drunk 0.141*** (0.015) Household characteristics Polygamous household 0.025** (0.011) Attitudes and experiences of IPV Father beat mother 0.109*** (0.021) Condones wife beating 0.048*** (0.012) Nigeria specific factors Ethnicity (reference category: Yoruba) Igbo -0.001 (0.021) Hausa -0.058** (0.024) Fulani -0.028 (0.050) Other ethnic group 0.001 (0.019) Religion (reference category: Christianity) Islam -0.007 (0.012) Traditional African religion -0.040 (0.060) Other religion 0.351 (0.375) Before BH (Non-BH Area) 2482 Before BH (BH Area) 732 During BH (Non-BH Area) 4123 During BH (BH Area) 1126 Total 8463 R-squared 0.10 Note: Variables used to estimate the kernel propensity scores are dropped are dropped. The estimations account for state fixed effects. The kernel density function is epanechnikov with a bandwidth of 0.06. Probit estimation is used for the propensity score in the first stage. ***Significant at the 1% level, **Significant at the 5% level, *Significant at the 10% level. Robust standard errors in parentheses 41 Table A4b: The effect of the Boko Haram insurgency on women’s experience of controlling behavior by husband/partner and women’s decision-making autonomy in Nigeria – urban sample (1) (2) (3) Variables Controlling behavior by Women’s household Women’s earnings decision- husband/partner decision- making making autonomy autonomy * 0.116*** -0.295* -0.001 (0.003) (0.167) (0.014) Covariates Individual characteristics Type of earnings (reference category: no earnings) Cash only -0.063** 0.251 - (0.029) (0.198) Cash and kind -0.124*** 0.841 - (0.038) (0.260) Kind only -0.044 -0.594 - (0.084) (0.667) Total number of children born -0.009*** 0.037*** 0.005*** (0.003) (0.018) (0.001) Married as a child 0.013 -0.258** 0.007 (0.016) (0.010) (0.008) Partner characteristics Sometimes or often drunk 0.109*** 0.195 -0.010 (0.021) (0.119) (0.010) Household characteristics Polygamous household 0.011 -0.137*** 0.011 (0.018) (0.110) (0.009) Attitudes and experiences of IPV Father beat mother 0.089*** -0.043 -0.002 (0.022) (0.137) (0.011) Condones wife beating 0.137*** -0.352*** -0.014* (0.016) (0.100) (0.008) Nigeria specific factors Ethnicity (reference category: Yoruba) Igbo -0.009 0.046 -0.002 (0.032) (0.183) (0.015) Hausa 0.014 -1.484*** -0.019 (0.047) (0.214) (0.017) Fulani -0.007 -1.033*** -0.083*** (0.060) (0.334) (0.027) Other ethnic group -0.016 -0.341*** 0.011 (0.030) (0.159) (0.013) Religion (reference category: Christianity) Islam 0.019 -0.885*** 0.013 (0.019) (0.121) (0.010) Traditional African religion 0.062 -1.519** -0.097* (0.079) (0.587) (0.051) Other religion -0.090 -2.142** 0.101 (0.285) (3.563) (0.374) N Before BH (Non-BH Area) 2485 2484 2112 Before BH (BH Area) 732 732 694 During BH (Non-BH Area) 4122 4121 3951 During BH (BH Area) 1126 1125 1092 Total 8465 8462 7849 R squared 0.10 0.37 0.07 Note: Variables used to estimate the kernel propensity scores are dropped are dropped. We log-transform the index of women’s household decision-making autonomy before estimating the model to allow for the reporting of the estimates in terms of percentages. To take care of zero values, we add 0.0001 to the index before taking the log. Thus, the estimates reported calculated as ( − 1). 42