Policy Research Working Paper 9825 The Risk That Travels with You Links between Forced Displacement, Conflict and Intimate Partner Violence in Colombia and Liberia Jocelyn TD Kelly Amalia Rubin Uche Ekhator-Mobayode Diana J. Arango Gender Global Theme October 2021 Policy Research Working Paper 9825 Abstract In 2020, the United Nations reported the highest number intimate partner violence. Displaced women in Colombia of displaced persons ever recorded; more than half of and Liberia have between 40 and 55 percent greater odds of this population was comprised of women and girls. Dis- experiencing past-year intimate partner violence compared placement and conflict substantially heighten the risk of with their nondisplaced counterparts. In each country, both gender-based violence, including intimate partner violence, conflict and displacement were independently and signifi- for women and girls. The current study aims to examine cantly associated with past-year intimate partner violence. the links between conflict, forced displacement, and inti- Recognizing the increased prevalence of intimate partner mate partner violence in two different conflict-affected violence for women who have been displaced is vital to settings: Colombia and Liberia. This paper draws on pop- providing effective assistance. As part of humanitarian, state, ulation-based data measuring intimate partner violence, and peacebuilding efforts, displaced and conflict-affected combined with political science data on political violence. women should be able to access a range of assistance services The findings show that forced displacement is highly and to help them heal from the impacts of the violence. significantly associated with increased lifetime and past-year 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 jkelly@hsph.harvard.edu, amalia.h.rubin@gmail.com, uekhator@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 The Risk That Travels with You: Links between Forced Displacement, Conflict and Intimate Partner Violence in Colombia and Liberia Jocelyn TD Kelly, [†] Amalia Rubin,[‡] Uche Ekhator-Mobayode, [‡] Diana J. Arango[‡] Key Words: Conflict, Forced Displacement, Health and Inequality, Health Behavior; Marriage; Violence against women and girls, Gender Based Violence, Intimate Partner Violence, Colombia, Liberia JEL: J14; J12; B45, I12 The authors of this paper conducted their research under Gender Dimensions of Forced Displacement project. The project is co-led by Lucia Hanmer and Diana Arango under the guidance of Hana Brixi, Global Director, Gender Unit, The World Bank Group. Funding: 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 Foreign, Commonwealth and Development Office (FCDO), 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 FCDO, the WBG or UNHCR. Introduction1 The United Nations High Commissioner for Refugees (UNHCR) reported 79.5 million people were forcibly displaced by the end of 2019– the highest number of displaced persons ever recorded (UNHCR, 2020a). Of these, more than half are internally displaced within their own country (UNHCR, 2020b). Multiple factors drive forced displacement: war, persecution, human trafficking, economic deprivation and, increasingly, natural disasters and climate-based disasters (Akram, 2013; ICFRC, 2012; United Nations, 2019). These factors often overlap, creating complex drivers for displacement that may vary from person to person and place to place. Recent scholarship advocates for further research to build our understanding of the intersection between various experiences of GBV and state and peacebuilding efforts. To facilitate the further intersection of the two disciplines Swaine et al. (2019) present an ecological model, often used to understand IPV, adapted to explore the drivers and risk factors for GBV during and after armed conflict. This paper uses this recently established framework to underpin analytical choices and the interpretation of results. The analysis presented contributes to closing the gap in understanding how IPV is impacted as a result of conflict, and how these experiences must be acknowledged and responded to during state and peacebuilding efforts. Although displacement is a global phenomenon, few studies undertake comparative country analyses to understand how displacement may be associated with other outcomes, such as violence. This lack of analysis is particularly pertinent for understanding women’s experiences in displacement. While gender-disaggregated data on displacement is still not systematically collected, UNHCR estimates that women represent 51% of the refugee and internally displaced population worldwide (UNHCR, 2020). Forced displacement impacts every aspect of women’s lives and brings unique stressors, vulnerabilities and threats. In 2019, the United Nations (UN) classified gender-based violence (GBV) as both a driver of forced displacement and a risk experienced by those who have been forcibly displaced (United Nations, 2019). While both women and men may experience GBV, women are disproportionately affected by multiple forms of violence – both during conflict and times of peace (WHO, 2013; IASC 2015, UNHCR, 2020). A number of studies underline how displacement and violent conflict substantially heighten the risk of sexual violence, physical violence and IPV (Ellsberg et al., 2020; Kelly, 2018; Kelly, 2019; Global Women’s Institute and International Rescue Committee, 2019; Wako et al., 2015; Stark and Ager, 2011)). One of the most common forms of GBV across the globe is intimate partner violence (IPV), with one out of every three women experiencing this abuse during her lifetime (WHO, 2021). Women and girls may be fleeing external dangers, but the risk of violence within the home can also intensify during crisis due to the breakdown in protective community and social structures, psychological stress associated with flight, rampant impunity and increased financial uncertainty (Ager et al., 2018; Annan & Brier, 2010; Ekhator-Mobayode, 2020; Gallegos & Gutierrez, 2010; Hynes et al., 2004; Janko, 2014; Kelly, 2018; Ostby, 2016; Wachter et al, 2018). [†] Harvard Humanitarian Initiative [‡] World Bank Group 2 The current study aims to build on this literature by comparing the links between conflict, forced displacement and IPV in two different conflict-affected settings: Colombia and Liberia. Both countries have faced long-running civil conflict and high levels of societal violence. This paper draws on the availability of Demographic and Health Survey (DHS) data, which provides population-based data on health outcomes in countries around the globe. Unique to Colombia and Liberia, however, is the fact that the DHS collected data on internal displacement in addition to information about exposure to IPV. The 2007 data from Liberia was collected four years post- conflict and can provide insight into the long-term impact of displacement on women. Similarly, 2010 data from Colombia gives insight into displacement and IPV during the ongoing Colombian conflict. The availability of this data for two conflict-affected countries allows for a unique comparative analysis. The analysis of experiences of IPV after conflict exposes the long- term impacts conflict has on the lives of women and calls us to include these experiences in our development programming to support peace and state building. In both countries, conflict has led to high levels of forced displacement. During Liberia’s two consecutive civil wars between 1998 and 2003, around 1.4 million people were forced to flee their homes (Global IDP Project, 2003). Colombia, a more populous country, has seen roughly 15 percent of its population displaced between 1995 and 2018 (IDMC, 2019). At the same time, both countries have also experienced some success in establishing peace agreements and undertaking subsequent social programs to aid the implementation of the agreements and demobilization efforts. Both Liberia and Colombia also present a unique opportunity to examine the links between IPV and displacement because each contains unique data from the DHS not readily available from other countries- namely, information about displacement as a result of conflict. Below, we examine the literature on the drivers of IPV in conflict and displacement settings, followed by a detailed examination of the drivers of IPV within each country context. Following this, a description of each country context is provided. Background A growing literature examines how violence can spread across time and space, as well as across the population and individual levels (Hudson, 2020; Catani, 2010; IOM, 2012; Kelly, 2017; Otsby, 2016; Saile, 2013). A number of studies have traced how political-level violence can “trickle down” to the community, family and individual level (Cummings et al., 2010; Cummings et al., 2011; Dubow et al., 2010; Gupta et al, 2009; Gupta et al, 2012; Mullins et al., 2004). In their examination of transmission of violence within war-affected families in northern Uganda, Saile et al. (2014) conclude that individual factors alone cannot account for the perpetuation of personal violence after war and hypothesize that social learning or diffusion may be at play. A vicious cycle may ensue, research from Hudson et al (2020) suggests that violence against women through socio-political structures and gender inequality can normalize intimate partner violence and this violence, in turn, at can normalize inequitable practices, socio-political structures and customs. This understanding builds upon the model produced by Patterson (2008) to show that violence against women reinforces the normalization of violence at different levels which can be seen through increased regular acts of aggression and violence. 3 The adapted ecological framework for addressing drivers of conflict and post-conflict VAWG allows us to understand a complex interplay between factors at the individual, interpersonal, institutional, community and societal levels that contribute to an individual’s experience of violence. (Swaine et al, 2019). At the societal level, conflict often triggers deterioration of important services (Murphy et al., 2020), including public health infrastructure (Annan & Brier, 2010), as well as the breakdown of the rule of law, governance, security, and justice mechanisms (Swaine et al., 2019). These systems can serve to deter interpersonal conflict and address its adverse effects. During the breakdown of legal systems, a culture of impunity can arise that emboldens perpetrators of violence and drives IPV (Swaine et al., 2019; Swaine, 2015). Economic systems often go into crisis leading to decreased employment and higher levels of poverty, which can be drivers of interpersonal stress and violence (Baranyi et al., 2011). Additionally, war can legitimize hyper-masculinity; exacerbating and normalizing violence when men seek or are expected to (re)assert masculinity in instances of community of familial vulnerability (Browne et al., 2019; Gibbs et al., 2020). This hyper-masculine aggression may become not only relatively more acceptable, but an adaptive behavior (Annan & Brier, 2010; Clark et al., 2010; Hudson et al., 2020; Saile et al., 2014; Vinck et al., 2007). War-related trauma can make men more likely to turn to violence as a way to exert power within the household or as a negative coping mechanism for trauma (Catani, Gupta et al, 2009; Gupta et al, 2012). Individuals in the household such a wives and children may also be targeted for violence because of their own manifestations of trauma or because they have experienced stigmatizing events during war (Albutt et al, 2016; Kelly et al., 2011). As Ostby notes, “The combination of exposure to childhood abuse and political violence may contribute to a ‘culture of violence’ in which violent response to conflict becomes both instrumental and normative in order to maintain male superiority” (Otsby, 2016, p. 6). In displacement settings specifically, studies from diverse areas have pointed to alcohol consumption, financial insecurity, a breakdown in social safety nets, and changing gender norms as potential drivers of IPV (Jewkes et al., 2017; Kohli et al., 2015; Mootz et al., 2018; Wirtz et al., 2014). In Uganda, Mootz et al. (2018) found that exposure to conflict and subsequent displacement increased alcohol consumption due to unemployment, stress, and depression among men. Financial insecurity can also drive IPV during displacement (Cardoso et al., 2016; Kohli et al., 2015; Wirtz et al., 2014). In a study on perceptions of displacement, Wirtz et al (2014) found that women reported increased IPV when their husbands faced unemployment, as well as when women sought employment or education to support their families. Women who have been forcibly displaced are often the heads of household and bear the responsibility for providing for family members, including children and elderly relatives (UNHCR, 2014). Particularly in refugee camps, some women have been found to be more successful than men in finding employment, partly because they are more willing to take on hard or low-paid jobs (Horn, 2010; Payne 1998). Shifting power within formal and informal camps mean that women may gain social and financial autonomy while men may feel a loss of authority and power and a need to fulfill the traditional ‘provider’ role (Carleson, 2004; Horn, 2010). While the global literature on IPV in the context of forced displacement has continued to grow, quantitative research on IPV in specifics displacement contexts is limited (Swaine et al, 2019). Literature from Liberia on refugees and refugee host populations has suggested that exposure to 4 conflict increased IPV (Vinck & Pham, 2013) and that the conflict normalized IPV as a response to frustration and stressors (Horn et al., 2014). A qualitative study of 110 women in Voinjama and Monrovia in Liberia and Freetown and Kailahun in Sierra Leone indicated that, for some women, IPV decreased when women took on more economic responsibilities due to decreased financial burdens on men (Horn et al., 2014). A much wider qualitative literature exists for Colombia and the drivers related to IPV among those experiencing forced displacement there (Browne et al., 2019; Wirtz, 2014; Alzate, 2018; Friedemann-Sánchez & Lovatón, 2012; Hynes, 2012; Hynes et al., 2016; Jones & Ferguson, 2009; Noe & Rieckmann, 2013). This literature includes a study that found over 50% of displaced women had suffered physical IPV versus 20% of non-displaced women in Colombia (Alzate, 2008). In multiple Colombian studies, IPV has been linked to women’s increased economic opportunities in displacement (Friedemann-Sánchez & Lovatón, 2012; Hynes et al., 2016). Browne et al. (2019) found that men who had been excluded from employment or politics due to displacement in Colombia were more likely to perpetrate IPV as a negative coping mechanism for stress and a way to and reassert their masculinity. This marks an important contextual difference between Colombia and Liberia, where, as noted above, women’s access to work was seen as a pathway to decrease household stress and, as a result, risk of IPV. While the literature on drivers of IPV in Colombia is larger, the research available in these two contexts has largely been qualitative and relies on relatively limited sample sizes, with few population-based studies on GBV available. Collecting systematic quantitative data on displaced populations can be challenging, particularly in countries currently or recently affected by conflict. In particular, few studies have undertaken multi-country comparisons. A notable exception is Otsby’s (2016) paper examining the link between sexual IPV and conflict experiences in 17 countries in Sub-Saharan Africa between 2006 and 2011. Otsby finds that conflict exposure heightens the risk of sexual intimate partner violence across the conflict-affected countries examined in the analysis. The paper calls for future studies to explore how conflict impacts other types of partner violence (Otsby, 2016). The current study builds on the literature above by drawing on nationally representative, population-based surveys to look at how displacement experience is associated with different forms of IPV (lifetime, past-year and injury-causing IPV) in Colombia and Liberia. Additionally, independent data on the level and location of conflict was merged with individual level data in each country. These unique data sets provide insight into how both types of adversity - conflict and displacement - may impact a woman’s risk of IPV. The goal of this work is to highlight the link between IPV and forced displacement and conflict, and to explore the policy implications for state and peace building efforts. This understanding will help direct scarce funding appropriately to create more effective and targeted programs. Colombia Background Colombia has endured civil war, violent conflict, and displacement for over 60 years. Levels and intensity of violence have fluctuated throughout this period between a multitude of groups, including guerrilla groups such as the Revolutionary Armed Forces of Colombia (FARC), the National Liberation Army (ELN) and the Popular Liberation Army (EPL), the government, and paramilitary forces (Palacios, 2006). The historical root cause of the conflict rests in the unequal 5 distribution of land and assets and lack of spaces for democratic transformative political participation coupled with weak institutions. These factors gave way to the use of violence and the rise of armed actors, who drove large numbers of rural, poorly educated peoples from agricultural areas into urban areas (Hynes et al, 2016). Hostility between political parties in Colombia ignited the period known as “La Violencia” (1946-1958). With La Violencia began a prolonged crisis of forced displacement, a phenomenon that has accompanied the 6 decades of conflict (Cabellero, 2018). The drivers of forced displacement in Colombia are a myriad of factors directly linked to violence such as fear of forced recruitment, fear of experiencing violence, appropriation of lands, and other violations of human rights. The conflict has been impacted by the subsequent rise in violence towards civilians, drug trafficking, terrorism associated to drug trafficking and the rise of paramilitary groups to protect the landlords targeted by guerrilla groups (Hynes et al., 2016). It is important to note that at beginning of the 2000s, women—adults, adolescents and girls—constitute 55 percent of all IDPs and that rural, Afro-Colombian, and Indigenous women were over-represented among them (Alzate, 2008). In 2010, the same year the DHS data which we will analyze was published, the government of Juan Manuel Santos took office in Colombia with over 3,672,000 internally displaced citizens (Meertens, 2010). Of these IDPs, 75 percent had moved from rural areas, 25 percent from urban areas, and 49 percent were women (UNHCR, 2010). While most IDPs have been displaced from rural to urban areas, violence in larger urban centers has led to substantial intra-urban displacement due to the conflict or violent disputes over territorial control, employment pressures, and gang-related violence (Internal Displacement Monitoring Center, 2013). As of July 2010, the Colombian NGO Instituto de Estudios para el Desarrollo y la Paz estimated 6,000 armed combatants were active in operations in 29 out of 32 departments. The first year of the Santos presidency experienced both increases in violence and a commitment to beginning peace negotiations (Human Rights Watch, 2011). The economic and social structures in Colombia improved despite the conflict. In 2010, Colombian average income had been steadily increasing since 2002. Between 2000-2010, the government increased investments in citizens as seen through spending on health and education. The Gender Development Index has consistently been above the global average in the last 15 years and the Gender Inequality Index has been on a steady decline since 2000 (UNDP, N.d) . However, Colombia data from 2010 indicates high numbers of women working in the informal sector, where more than 50 percent of women are protected by minimal regulations; have few or no benefits; lack voice, social security and decent work conditions; and are vulnerable to low salaries and possible job loss (UNDP, 2019). In addition, and in concert with issues of displacement, Colombia continued to report high levels of violence against women and gaps in gender equality in nationally representative surveys such as the Demographic Health Survey (DHS Statcompiler, 2010; Profamilia, 2010, UNDP, n.d.). In 2010, over one-third of women in Colombia reported ever having experienced physical violence by a partner, and 9% reported partner sexual violence. However, tolerance for abuse remained low. Only 2 percent of women believed that a husband is justified in beating his wife for at least one reason and 44 percent of ever-married women who experienced any physical or sexual 6 violence sought help to stop violence. Thirty-five percent of women surveyed indicated that they experienced some kind of physical violence from the age of 15 (DHS Statcompiler, 2010). In Colombia, Wirtz et al. (2014) documented women’s exposure to forced recruitment and labor, early and forced marriage, ongoing threats and multiple displacements perpetrated by armed actors in conflict affected communities prior to displacement. In November of 2016 the Colombian government signed a peace agreement with the FARC guerilla group that had been waging war against the state since 1964. The 2016 Peace Deal is largely considered one of the most inclusive of its kind, with women eventually making up 20 percent of the government negotiation team and over 40 percent of the FARC delegates (Olivieri & Muller, 2019). Nevertheless, in August of 2019 a small group of FARC members announced a return to armed activity. Colombian conflict continues, not just through the FARC but with several organized armed criminal groups, including paramilitaries and drug traffickers – exacerbated by the neighboring Venezuelan humanitarian crisis. The most recent information emerging from Colombia points to a more promising future for gender equality. For instance, a gender assessment of Colombia completed by the World Bank in 2019 noted that primary and lower secondary education was on the rise for Colombian women between 2007-2011. In addition, female labor force participation was high in 2010, with 73 percent of women ages 24-40 actively working. In the context of the conflict, it is important to note that there are large gaps between urban and rural women in Colombia -- particularly in regard to access to health services, educational attainment and performance, and labor force participation (Olivieri & Muller, 2019). Liberia Background Liberia, founded as a colony for returning American slaves, declared independence from the United States in 1847. After a century of independent rule, food riots in the late 1970s sparked unrest in the country that culminated in a 1980 coup, which marked the beginning of years of authoritarian rule. A rebellion launched by Charles Taylor incited the first Liberian war, which lasted over a decade from 1989-1997. A brief interlude of peace ensued but anti-government fighting broke out once again in 1999. Armed forces from neighboring countries joined the fight, inciting country-wide violence. The 13 years of conflict resulted in the death of over 150,000 people and the displacement of another 850,000 people (United Nations, 2013; BBC, 2018; CIA World Factbook, 2015). Liebling-Kalifani et al. (2011) detailed the consequences of displacement and its damaging effects on women’s lives, citing issues related to food insecurity, absence of community protection and support, disintegrated social support and kinship systems, early-forced marriages and disruption to educational programs. After four years of fighting, a peace agreement was signed in 2003. Both civil wars were characterized by horrific human rights abuses, including rape, torture, mutilation and murder (Johnson et al., 2008). The 13-year period spanning the two wars oversaw the greatest collapse of economic and social structures in Sub-Saharan African in modern history. At the time of the 2005 post-war elections, average income was 25% of its pre-war levels, and the government spent less on its citizens between 2000-2005 than any other country in the world (Radelet, 2007). Social services were nearly non-existent: with no health and educational services, the disintegration of economic 7 markets, destroyed infrastructure, and an absence of state-provided electricity and piped water until 2006 (International Crisis Group, 2003). Three years after the last peace agreement, only about half of those who had been forcibly displaced returned home. While many former combatants had entered reintegration programs, demobilized army officers and former security personnel continued to stage protests, some violent, throughout the years (Amnesty International, 2007). By the end of 2006, approximately 9% of Liberians still resided in other countries such as Guinea, Sierra Leone, Côte d’Ivoire, Ghana and Nigeria (Amnesty International, 2008). Violence against women, including rape and IPV, was pervasive in post-conflict Liberia, as reported in the DHS conducted five years after the war. In a 2007 survey, over one-third of women in Liberia reported ever having experienced physical violence by a partner, and 11% reported partner sexual violence (Liberia Institute of Statistics, 2008). Just under half of all women (49%) reported having experienced some kind of violence (physical, sexual or emotional) from a partner. Sixty percent of women believed that a husband is justified in beating his wife for at least one reason. Almost half of women surveyed (45%) indicated that they experienced some kind of physical violence from the age of 15, and one-third of women experienced this abuse in the past 12 months. During this time, other incidents of violent crime such as armed robbery and murder increased as well (Human Rights Watch, 2008). The Human Development Index (HDI) provides a rare insight into Liberia’s development pre-, during, and post-conflict. The scale aggregates information about lifespan, education and gross domestic product (GDP) to create a global ranking of countries (Anand & Sen, 1994). In 1970, Liberia scored in the lowest quartile of development. The country’s progress was reflected in its HDI scores, which climbed steadily in the 1970s, only to plunge to near the global bottom in 1989 (Klugman, 2010). In addition to high levels of intimate partner violence, data from Liberia indicates that large gaps in gender equality remained post conflict, despite Ellen Johnson Sirleaf’s election in 2006 and the appointment of several women to high-level positions in the administration . A UNESCO report written during the height of Liberia’s second civil war reported that Liberia had one of the lowest gender parity scores globally, and girls were disproportionately represented in the number of out-of-school children in Liberia. This disruption in education presaged lifelong consequences for this cohort of women (Kirk, 2003). In 2007, the literacy rate for adult men was 55% and women was 41% -- the number lowering to 26% for rural women (OECD et al., 2009). In 2001 an education law was enacted making primary education free and compulsory, however, as of 2006, primary school enrollment in urban areas was 63.7% for girls and 33.1% for girls in urban areas (Woodon, 2012). Liberia also scores in the bottom eight countries that reported women having fewer than half of the years of education as men (Klugman, 2010). Five years after the war ended, women’s labor force participant was high, with women accounting for 54% of the labor force. However, women were disproportionally represented in the informal sector (CWIQ, 2007). According to data from the 2017 Women, Peace, and Security Index, Liberia still has a number of gaps to fill particularly related to education, financial inclusion, legal discrimination and intimate partner violence (WPS, 2017). 8 Methods Data sources Demographic and Health Surveys (DHS): Individual-Level Data The DHS Program has been collecting data since 1984 in over 90 countries. The surveys examine fertility, family planning, maternal and child health, gender dynamics, HIV/AIDS, malaria, and nutrition. A core standard questionnaire is administered in all countries, with some variation to ensure that questions are culturally appropriate and relevant. This project uses DHS data from 2010 in Colombia and 2007 in Liberia. These surveys were chosen because they represent data collection that has occurred after or during a period of active conflict, and have Geographic Information System (GIS) information about the cluster where the women were sampled. 2 The DHS surveys use a two-stage cluster sampling design that first randomly selects clusters and then randomly selects households within the cluster. Only one woman per household is eligible to take the domestic violence module for privacy concerns. The DHS Women’s Questionnaire collects data on women aged 15 to 49 years. Data on interpersonal and partner violence is collected as part of the DHS Domestic Violence (DV) Module, now applied in conjunction with the Women’s Individual Questionnaire. 3 Dependent Variables This study examines three outcomes: lifetime IPV, past-year IPV and injury resulting from IPV, each is described below. Lifetime Intimate Partner Violence The DHS Domestic Violence module uses a modified Conflict Tactics Scale (CTS) to measure IPV, one of the most widely used and reliable measurement tools for IPV (Straus et al, 1990). Strengths of this assessment include the number of opportunities to disclose violent events; detailed information about a range of behaviors; and its widespread use (Hindin et al., 2008). The CTS provides comparable estimates of violence across different settings (Hindin, 2008). Ever-partnered women were asked about a list of eight specific behaviors they may have experienced that would classify as physical or sexual violence. Women answering “yes” to any of the items from a to g were classified as having experienced partner physical violence ever by 2 In the DHS data, Global Positioning System (GPS) data for clusters are randomly offset in order to safeguard respondent privacy and confidentiality (Burgert et al., 2013). In urban areas, clusters are displaced by 0 to 2 kilometers, and in rural areas locations are displaced between 0 to 5 kilometers (Perez-Heydrich et al., 2013). The displacement is checked to ensure that displaced clusters do not leave national or administrative boundaries (thus, a cluster in Colombia would not be displaced in a way that it would move to Peru, for example). Since this displacement does not move clusters across administrative boundaries, it does not impact the current analysis. 3 The number of women pre-selected to take the domestic violence module in each household is established using a matrix known as the Kish grid technique that matches the number of eligible women with a random number generated as part of the household identifier (Kish, 1965). When the interviewer arrives at the first question of the domestic violence module, he or she establishes whether the woman has been pre-selected; if so, the module is administered. Only one woman per household is eligible to take the domestic violence module in order to ensure that others in the house do not know what types of questions were asked. The module is administered only to individuals in a private setting, and an additional consent script is read to the respondent. If privacy cannot be ensured, the domestic violence module is not administered. 9 the DHS. Women answering “yes” to items h or i were classified as having experienced sexual partner violence ever (Table 1). Table 1. DHS questions assessing partner physical and sexual violence Does/Did your (last) husband/partner ever do any of the following things to you: a) Push you, shake you, or throw something at you? b) Slap you? Partner Partner physical c) Twist your arm or pull your hair? d) Punch you with his fist or with something that could hurt you? e) Kick you, drag you, or beat you up? sexual violence f) Try to choke you or burn you on purpose? g) Threaten or attack you with a knife, gun, or other weapon? h) Physically force you to have sexual intercourse with him even when you did violence not want to? i) Force you to do any sexual acts you did not want to? Intimate Partner Violence in the Past 12 Months Women who identified as having ever experienced IPV were asked about the frequency of the act in the 12 months preceding the survey. Past-year IPV is defined as having experienced any physical or sexual violence (items a to i) “often” or “sometimes” in the past 12 months. Injury from Intimate Partner Violence The 2010 Colombia DHS and the 2007 Liberia DHS asked women “Did the following ever happen as a result of what your (last) husband/partner did to you: 1) cuts, bruises, or aches; 2) burns, eye injuries, sprains, or dislocations; and 3) deep wounds, broken bones, broken teeth, or any other serious injury. Women who reported “yes” to any of the 3 categories were classified as having injury from IPV. Independent Conflict-Intensity Data: District-Level Data Within countries experiencing conflict, there is notable heterogeneity among districts that experience violence. This conflict-affectedness information at the district level can be combined with DHS data at the individual level to examine the links between interpersonal violence and conflict. Women’s experiences are nested in districts, which are classified according to whether or not the district has experienced conflict-related events. The multi-level modeling approach described below accounts for the natural clustering of women into these administrative units and acknowledges the hierarchical structure of the data (Kreft and Leeuw, 1998). The district-level covariate acts as a proxy for other important cluster-level characteristics that are not measured. Counting the number of fatalities in administrative units has been used successfully in similar efforts in the past, including in Peru (Gallegos and Gutierrez, 2011) and Liberia (Kelly et al, 2017; Kelly et al, 2018). Independent data sets exist to measure conflict-fatalities in both Colombia and Liberia. For Liberia, Armed Conflict Location and Event Data Project (ACLED) data provides a measure of the extent to which a community has been affected by conflict at the sub-national district level. While ACLED data is not available for Colombia in the years prior to the 2010 DHS survey, 10 there is another data source that provides independent information on conflict events. The Uppsala Conflict Data Program, Conflict Encyclopedia Database (UCDP) provides similar information that spans the time period of interest. In contrast, UCDP data is not available for Liberia - so each country draws on similar but distinct data set to incorporate independent information about conflict intensity. The study data set contains one observation for every woman in the DHS individual recode, i.e., the woman level data and variables showing the annual number of ACLED or UCDP events and fatalities beginning in the year the woman was interviewed up to 10 years preceding the survey. Providing conflict for the 10 years preceding the DHS survey is key to the methodology and helps establish temporality in how conflict may affect IPV outcomes. A 10-year period was chosen because this captured hostilities from both Liberian civil conflicts and allowed the data set to reflect the long-standing impact of the conflict in each country. Each conflict data set used is described below. Uppsala Conflict Data Program (UCDP) Data - Colombia The Uppsala Conflict Data Program (UCDP) Georeferenced Event Dataset is an event data set that disaggregates three types of organized violence: state-based conflict, non-state conflict, and one-sided violence. The database compiles articles containing information about individuals killed or injured, and triangulates information with reports from non-governmental organizations and the UN, as well as truth commission reports and other local sources of information. For Colombia, UCDP includes conflict data from as far back as 1946, the beginning of La Violencia period as noted above. The total number of conflict related deaths reported in Colombia since 1946 equals 27,743 as of January 2021. This includes deaths from state-based actors, non-state actors and one-sided violence. None of the 32 departments of Colombia has been spared from conflict related deaths, however the number of deaths varies across Colombia’s different regions. Armed Conflict Location and Event Data Project (ACLED) - Liberia ACLED data provide the dates and locations of all political events related to conflict and unrest in over 50 countries. ACLED data provide information on the implicated actors of political events that may occur in the course of civil and communal conflicts, violence against civilians, rioting and protesting. Armed actors may include governments, rebels, militia, organized political groups, ethnic groups, and civilians. ACLED geocodes event data at the first and second administrative boundary levels and provides latitude and longitude coordinates for each event. The database draws on three different types of sources in order to achieve comprehensive reporting: local, regional, national and continental media are reviewed daily; NGO reports are used to ensure reporting occurs in remote or hard-to-access locations; and Africa-focused news reports and analyses are used to supplement previous sources. ACLED states that this methodology achieves the most comprehensive source material currently available for digital conflict event coding (Raleigh et al., 2010). For this project, the ACLED Version 5 database was used to determine the numbers of fatal conflict events. Linking DHS and Conflict Data Using GPS points provided in ACLED and UCDP, and for each cluster of the Demographic and Health Survey (DHS), we temporally and spatially link information on conflict events to 11 determine the yearly number of fatalities in the district of residence for each woman in the DHS individual recode. First, the district for each DHS cluster is determined by superimposing GPS location in the DHS on spatial data from the Database of Global Administrative Areas (GADM), resulting in a data set where each observation is the DHS cluster and contains a variable with district identifiers from the GADM for each DHS cluster. Second, the UCDP or ACLED data is superimposed on the spatial data from the GADM to determine the annual number of conflict events and annual number of fatalities for each year beginning from the year of the DHS survey and up to 10 years preceding the survey. This results in a data set where each observation contains unique district identifiers with variables showing the number fatalities for various years. Third, the data sets in steps (1) and (2) are merged: the DHS cluster data containing unique district identifiers is linked with the district level data containing annual number fatalities for various years. The resulting data set is the DHS cluster data containing information on yearly number of ACLED or UCDEP events and fatalities beginning from the year of the DHS and for up to 10 years preceding the survey. Finally, the data set in step (3) is merged to the individual woman level data by the DHS cluster, resulting in the final study data set. Final Country-Specific Study Samples Given the unique context of each country and the widely varying sample sizes between each, as well as the fact the Colombia and Liberia each had different sources of information for conflict intensity, this paper conducts country-specific analyses for each context. In Colombia, 53,521 women took the domestic violence module and were also successfully merged with UCDP conflict data. Of these women, 34,681 (65% of the original sample) had ever had a partner and so were eligible to answer questions about IPV. All women within this group also answered questions about forced migration. In the final analysis, 92.5% of this sample (n=32,083) responded to all questions used in the model, and represented the final analytical sample for the Colombia models (Appendix Figure 1). In Liberia, 4,913 women took the domestic violence module and were also merged with ACLED conflict data. Of these women, 3,975 (81% of the original sample) had ever had a partner and so were eligible to answer questions about IPV. Of these eligible women, 98.5% (n=3,917) answered questions about IPV. All women within this group also answered questions about forced migration. Of these women, only 92.7% (n=3,631) had geographic information about their district of residence. In the final analysis, 84.5% of this sample (n=3,070) responded to all questions used in the model, and represented the final analytical sample for the Liberia models (Appendix Figure 2). Primary Exposures: Displacement and Conflict Forced Displacement The 2010 Colombia DHS asks respondents whether they have changed their place of residence. Those respondents answering “yes” were asked why this move occurred. Options for why the move occurred included: violence by paramilitary/guerrilla group; natural disaster; was too poor; working opportunities; education opportunities; health concerns; family reasons; seeking better conditions; other. Those respondents who stated they moved due to “violence by paramilitary/guerrilla group” or “natural disaster” were classified as forcibly displaced. 12 The 2007 Liberia DHS asks respondents “During the war, did you leave your house?” - a question intended to explore whether respondents moved to a camp, lived in the bush or faced another form of displacement because of conflict. Those respondents who answered “yes” to this question were classified as forcibly displaced. It is noteworthy that this question was followed by a question about where the respondent was displaced. Answers included: stayed with relatives or friends inside Liberia; went to a camp; living in the bush; went outside Liberia. However, respondents could choose multiple of these options. For this reason, the first question was chosen as the most simple and accurate measure of displacement. Conflict at the District-Level Previous studies have used number of conflict fatalities as an effective proxy for levels of political instability (Kelly, 2018), since fatalities are the most definitive and violent measure of armed conflict. In addition, since the UCDP and ACLED data sets code conflict events differently, fatality measures were the most comparable between the data sets. For both UCDP and ACLED, the conflict was coded as 1 if women lived in a district with conflict fatalities and 0 if she did not. Independent Variables The variables that are included in this analysis (Table 2) have been chosen based on the ecological framework for addressing drivers of conflict and post conflict VAWG, as well on those variables that have been found to be significantly associated with IPV in previous analyses (Heise, 2011; Feseha et al., 2012, Swaine et al., 2019). Data on religion is not available from the 2010 Colombia DHS and so this variable was not included in the analysis. Broadly, the covariates of interest include demographic information; household characteristics; risk factors associated with IPV; and women’s partners characteristics. Table 2.Model Covariates Women’s demographics Age married Education level Marital status (currently versus formerly in a union) Currently working Household Characteristics Wealth quintile Number of children under 5 Urban or rural Women is household head IPV risk factors Number of control issues husband exhibits Wife-beating justified Father beat respondent’s mother Woman has decision making autonomy on at least one major decision Partner characteristics Partner education level Partner drinks alcohol or uses drugs 13 Data Analysis All analyses were conducted with Stata/SE 14.0 (StataCorp LP, College Station, TX). A bivariate model was also used to examine the relationship between the main predictor (fatalities) and each outcome. For the final model, a multilevel approach was used to account for the nested structure of the data, with clustering of women within districts. Model Specification Multilevel logistic regression models were used to quantify the effect of district level conflict on the odds of IPV after sequentially adding blocks of independent variables as described in Table 2. The models included a random intercept for district, to account for the geographic clustering of the sample and systematic differences between districts that would not otherwise be captured in a simple logistic regression. This approach has been used in similar analysis in past research (Kelly et al, 2018; Kelly, 2019). Multilevel Model - Dichotomous Exposure: In the regression equation above, i indexes the district and j indexes the individual. is the indicator for whether a woman (j) in district (i) has reported experiencing violence in the last 12 months. β_0+b0i defines the district level odds of a woman experiencing violence in district i given no conflict holding the individual-level covariates fixed. This equation expresses the model in the case of a dichotomous measure of conflict. Here, it gives the odds ratio of IPV if the district experienced any conflict compared to the odds of IPV if the district had experienced no conflict. X_ij contains individual, household and partner characteristics summarized in Table 1. As noted before, the independent variables are added to the model in blocks to examine for possible confounding or effect modification with the main association. For all analyses, significance was assessed using an alpha of 0.05. For each country, a multilevel model assessed the association between IPV and forced displacement. Stepwise model fitting was undertaken to assess how the main association was affected by sequential blocks of variables. The final model results for each country are given in the tables below. Sample Weights In order to account for the complex survey design of the DHS, the survey weights for the DV module were included in all analyses, using the probability weight or pweight option within the gllamm survey command. The probability weight is defined as the inverse probability of the respondent’s being included in the sample. The pweight command assumes weights are specified at least two levels in the data. Since the data were not weighted at the district level, the level-1 weight within Stata was specified as 1. The level-2 weights were calculated using the 14 d005 variable for the DV module specified in the DHS. The d005 weight in the DHS was rescaled by dividing it by 100,000 to create a scale between 0 and 1. Results The following sections present country-specific models for the key outcomes and forced displacement. The outcome of past-year IPV is key to establishing temporality between IPV and previous displacement, and is a key outcome for this paper. This will be presented first, followed by lifetime IPV. While lifetime IPV may make it harder to establish whether displacement preceded this violence, or vice versa, it can establish important recognition of association between these key variables. Finally, this project also aimed to examine whether severe forms of IPV resulting in injury were associated with forced displacement. This can help answer a key question not only whether IPV occurs, but also whether it might become more violent or harmful in association with displacement. Since injury is also a lifetime measure, again, it may be challenging to establish temporality but still provides critical insight into the linkages between displacement and violence. In Colombia, the final sample consisted of 34,681 ever-married women who responded to questions about lifetime IPV experience. Of these women, 1.8% (n=624) reported being forcibly displaced. Fortunately, the relatively larger sample size in Colombia still allows for analysis of forced displacement in the adjusted model despite the relatively low occurrence of the experience. In Liberia, the final sample consisted of 3,917 ever-married women in Liberia who provided information about lifetime IPV. Overall, roughly 10% of this sample (n=412) reported experiencing forced migration. Table 3. Experiences of Forced Migration by Country Country Colombia Liberia No forced migration 34,057 3,505 98.2% 89.5% Forced Migration 624 412 1.8% 10.5% Total 34,681 3,917 100.00% 100.00% First row has frequencies and second row has column percentages Colombia Colombia Past-Year IPV Analysis In Colombia, one in five women who have experienced lifetime IPV also reported experiencing IPV in the past year – the rates among displaced women are even higher, approaching one in four. 15 Women who experienced past-year IPV were compared to women who have not experienced IPV. The analytical sample was further restricted to women who stated they have been in their place of residence for over a year, ensuring that migration would predate past-year violence. This analysis helps establish temporality between when migration occurred compared to recent violence. Table 4. Past-year IPV by Forced Migration Status in Colombia Forced Migration Colombia No Yes Total No IPV 27107 467 27574 79.59% 74.84% 79.51% Past-year IPV 6950 157 7107 20.41% 25.16% 20.49% Total 34057 624 34681 100.00% 100.00% 100.00% First row has frequencies and second row has column percentages In Colombia, women who experience forced migration face 40% greater odds of past-year IPV compared to their non-displaced counterparts after adjusting for other covariates (aOR 1.40, p=0.05). Younger age at marriage, being widowed or divorced, being employed in the past year, household headship, having a partner with control issues, having a father who beat the respondent’s mother and having a partner who uses alcohol or drugs were all risk factors. Protective factors were having children under the age of five in the home, and having a partner with higher versus no education. Table 5. Adjusted model of Past-year IPV and Displacement in Colombia aOR p-value SE Confidence Interval No forced migration (ref) Forced migration 1.396* 0.028 0.212 1.036 - 1.881 Age married 0.965** 0.000 0.004 0.957 - 0.973 No education (ref) Primary 1.026 0.837 0.129 0.802 - 1.312 Secondary 0.942 0.637 0.120 0.734 - 1.208 Higher 0.816 0.150 0.115 0.619 - 1.076 Currently partnered (ref) Formerly partnered 1.648*** 0.000 0.089 1.482 - 1.832 Not employed in past year (ref) Employed in past year 1.107** 0.008 0.042 1.027 - 1.194 Poorest wealth quintile (ref) Poorer wealth quintile 0.925 0.197 0.056 0.821 - 1.041 Middle wealth quintile 0.984 0.834 0.076 0.847 - 1.144 Richer wealth quintile 0.932 0.384 0.076 0.795 - 1.092 Richest wealth quintile 0.924 0.381 0.083 0.774 - 1.103 Number of children under 5 0.925** 0.001 0.021 0.884 - 0.968 Rural (ref) Urban 1.072 0.290 0.070 0.942 - 1.219 Not head of household (ref) Head of household 1.170** 0.002 0.058 1.061 - 1.290 16 Number of control issues 1.676*** 0.000 0.015 1.647 - 1.705 Wife beating not justified for any reason (ref) Wife beating justified 1.175 0.122 0.123 0.958 - 1.442 Father didn't beat mother (ref) Father beat mother 1.691*** 0.000 0.069 1.562 - 1.831 Woman's decision autonomy on no issues (ref) Woman's decision autonomy on at least 1 issue 1.084 0.507 0.131 0.855 - 1.373 Partner education- none (ref) Partner education- Primary 0.966 0.699 0.086 0.812 - 1.150 Partner education- Secondary 0.841 0.068 0.080 0.698 - 1.013 Partner education- Higher education 0.626*** 0.000 0.069 0.504 - 0.777 Partner doesn't use alcohol or drugs (ref) Partner uses alcohol or drugs 1.820*** 0.000 0.070 1.689 - 1.962 Constant 0.071*** 0.000 0.014 0.048 - 0.105 Observations 25,504 * p<0.05, **p<0.01, **p<0.001 Colombia Lifetime IPV Analysis More than two-thirds of women in Colombia (37%, n=12,747) reported having ever experienced IPV, with a notable difference between displaced (46%) and non-displaced women (37%). Table 6. Lifetime IPV by Forced Migration Status in Colombia Forced Migration No Yes Total No IPV 21597 337 21934 63.41% 54.01% 63.25% IPV 12460 287 12747 36.59% 45.99% 36.75% Total 34057 624 34681 100.00% 100.00% 100.00% In Colombia, women who reported forced displacement had 43% greater odds of experiencing lifetime IPV compared to women not reporting displacement in the unadjusted model (OR 1.43, p<0.001). This association remained relatively stable and consistently significant across the stepwise modeling procedure. In the final model with all covariates included, Colombian women who experienced any forced displacement had nearly 40% greater odds of experiencing lifetime IPV compared to non-displaced women (aOR 1.38, p<0.01). Other risk factors include being widowed or divorced versus being in a current union (aOR 1.54, p<0.001); being employed in the past year (aOR 1.16, p<0.001); and having an urban versus rural place of residence (aOR 1.20, p<0.05). The model also found protective factors for lifetime IPV in Colombia. For every year a woman waits to get married, she has a .03% lower odds of lifetime IPV (aOR 0.97, p<0.001). Women in the highest wealth quintile were also less likely to experience IPV (aOR 0.84, p<0.05), as were women with more children under the age of five in the home (0.93, p<0.001). Women with partners who completed higher education versus no education were 25% less likely to experience IPV (aOR 0.75, p=0.001). 17 Table 7. Final Adjusted Model of Lifetime IPV and Displacement in Colombia aOR p-value SE Confidence Interval No forced migration (ref) -- -- -- -- Forced migration 1.381** 0.002 0.142 1.129 - 1.688 Age married 0.972*** 0.000 0.003 0.966 - 0.979 No education (ref) -- -- -- -- Primary 1.048 0.626 0.101 0.868 - 1.265 Secondary 1.022 0.814 0.096 0.851 - 1.228 Higher 0.927 0.470 0.097 0.756 - 1.138 Currently partnered (ref) -- -- -- -- Formerly partnered 1.542*** 0.000 0.069 1.412 - 1.683 Not employed in past year (ref) -- -- -- -- Employed in past year 1.116*** 0.000 0.032 1.055 - 1.181 Poorest wealth quintile (ref) -- -- -- -- Poorer wealth quintile 0.956 0.309 0.042 0.877 - 1.043 Middle wealth quintile 0.918 0.150 0.054 0.818 - 1.031 Richer wealth quintile 0.922 0.210 0.060 0.812 - 1.047 Richest wealth quintile 0.842* 0.012 0.058 0.736 - 0.964 Number of children under 5 0.932*** 0.000 0.016 0.902 - 0.964 Rural (ref) -- -- -- -- Urban 1.120* 0.014 0.052 1.023 - 1.226 Not head of household (ref) -- -- -- -- Head of household 1.026 0.491 0.038 0.954 - 1.103 Number of control issues 1.574*** 0.000 0.012 1.551 - 1.597 Wife beating not justified for any reason (ref) -- -- -- -- Wife beating justified 1.185 0.059 0.106 0.994 - 1.413 Father didn't beat mother (ref) -- -- -- -- Father beat mother 1.619*** 0.000 0.044 1.535 - 1.708 Woman's decision autonomy on no issues (ref) -- -- -- -- Woman's decision autonomy on at least 1 issue 1.096 0.331 0.103 0.911 - 1.319 Partner education- none (ref) -- -- -- -- Partner education- Primary 1.013 0.857 0.072 0.880 - 1.165 Partner education- Secondary 0.890 0.116 0.066 0.770 - 1.029 Partner education- Higher education 0.745** 0.001 0.064 0.629 - 0.882 Partner doesn't use alcohol or drugs (ref) -- -- -- -- Partner uses alcohol or drugs 1.671*** 0.000 0.047 1.581 - 1.767 Constant 0.143*** 0.000 0.022 0.106 - 0.193 Fixed effects District Observations 32,083 * p<0.05, **p<0.01, **p<0.001 Colombia Lifetime Injury from IPV Analysis In Colombia, more than one in four women have experienced injury from IPV in their lifetimes. We see higher levels of this abuse among displaced (28%) compared to non-displaced women (21%). Table 8. Lifetime Injury from IPV by Forced Migration Status in Colombia Forced Migration No No No No injury 26776 450 27226 78.62% 72.12% 78.50% Injury 7281 174 7455 21.38% 27.88% 21.50% 18 Total 34057 624 34681 100.00% 100.00% 100.00% First row has frequencies and second row has column percentages In this analysis, women who experienced injury from IPV were compared to women who did not experience IPV. In Colombia, women who experienced forced displacement were 40% more likely to experience lifetime injury-causing IPV compared to non-displaced women in the adjusted model (aOR 1.40, p<0.03). Table 9. Final Adjusted Model of Lifetime Injury from IPV and Displacement in Colombia aOR p-value SE Confidence Interval No forced migration (ref) Forced migration 1.396* 0.028 0.212 1.036 - 1.881 Age married 0.965*** 0.000 0.004 0.957 - 0.973 No education (ref) -- -- -- -- Primary 1.026 0.837 0.129 0.802 - 1.312 Secondary 0.942 0.637 0.120 0.734 - 1.208 Higher 0.816 0.150 0.115 0.619 - 1.076 Currently partnered (ref) 1.648*** 0.000 0.089 1.482 - 1.832 Formerly partnered -- -- -- -- Not employed in past year (ref) 1.107** 0.008 0.042 1.027 - 1.194 Employed in past year -- -- -- -- Poorest wealth quintile (ref) 0.925 0.197 0.056 0.821 - 1.041 Poorer wealth quintile 0.984 0.834 0.076 0.847 - 1.144 Middle wealth quintile 0.932 0.384 0.076 0.795 - 1.092 Richer wealth quintile 0.924 0.381 0.083 0.774 - 1.103 Richest wealth quintile 0.925** 0.001 0.021 0.884 - 0.968 Number of children under 5 -- -- -- -- Rural (ref) 1.072 0.290 0.070 0.942 - 1.219 Urban -- -- -- -- Not head of household (ref) 1.170** 0.002 0.058 1.061 - 1.290 Head of household 1.676*** 0.000 0.015 1.647 - 1.705 Number of control issues -- -- -- -- Wife beating not justified for any reason 1.175 0.122 0.123 0.958 - 1.442 (ref) Wife beating justified -- -- -- -- Father didn't beat mother (ref) 1.691*** 0.000 0.069 1.562 - 1.831 Father beat mother Women's decision-making autonomy 1.084 0.507 0.131 0.855 - 1.373 Partner education- none (ref) -- -- -- -- Partner education- Primary 0.966 0.699 0.086 0.812 - 1.150 Partner education- Secondary 0.841 0.068 0.080 0.698 - 1.013 Partner education- Higher education 0.626*** 0.000 0.069 0.504 - 0.777 Partner doesn't use alcohol or drugs (ref) -- -- -- -- Partner uses alcohol or drugs 1.820*** 0.000 0.070 1.689 - 1.962 Constant 0.071*** 0.000 0.014 0.048 - 0.105 Fixed effects District Observations 25,504 * p<0.05, **p<0.01, **p<0.001 19 Colombia Conflict-Intensity, Forced Displacement and IPV This research also aims to explicitly examine how both displacement and residing in a conflict- affected place may both contribute to past-year IPV risk for women. To achieve this, a model was run that looked at the number of fatalities in a woman’s district of residence occurring in the 10 years previous to the DHS survey, whether she experienced forced displacement, and an interaction term to examine how these two experiences may impact one another. Because this measure looks at violence that occurred in the past 12 months, there is a high likelihood that the violence occurred in the place of current residence. In this model, we see that conflict fatalities are highly associated with past-year IPV. For every additional conflict death in a woman’s district of residence, she is 0.1% more likely to experience IPV (aOR 1.001, p=0.01). Women experiencing forced migration have 40% higher odds of past-year IPV compared to non-displaced women, although this relationship has lower significance than in previous models (aOR 1.40, p= 0.05). The interaction between fatalities and displacement did not emerge as significant in this model, suggesting that in Colombia, both displacement and conflict impact a woman’s risk of IPV, but these two variables do not interact with each other to affect the main relationship. Table 10. Association of Past-Year IPV with Forced Migration Status and Conflict Fatalities in Colombia aOR p-value SE Confidence Interval No forced migration (ref) -- -- -- -- Forced migration 1.398* 0.052 0.241 0.998 - 1.960 Number of fatalities in district over 10 years 1.001** 0.006 0.000 1.000 - 1.002 Interaction between fatalities and forced 1.000 0.891 0.002 0.995 - 1.005 migration Age married 0.970*** 0.000 0.004 0.961 - 0.979 No education (ref) -- -- -- -- Primary 1.092 0.526 0.152 0.832 - 1.433 Secondary 1.195 0.206 0.168 0.907 - 1.574 Higher 1.221 0.205 0.192 0.897 - 1.663 Currently partnered (ref) 1.540*** 0.000 0.129 1.307 - 1.814 Formerly partnered -- -- -- -- Not employed in past year (ref) 1.098* 0.028 0.047 1.010 - 1.193 Employed in past year -- -- -- -- Poorest wealth quintile (ref) 0.843** 0.010 0.056 0.741 - 0.959 Poorer wealth quintile 0.744*** 0.000 0.058 0.638 - 0.868 Middle wealth quintile 0.689*** 0.000 0.058 0.584 - 0.813 Richer wealth quintile 0.552*** 0.000 0.052 0.459 - 0.663 Richest wealth quintile -- -- -- -- Number of children under 5 1.096 0.199 0.078 0.953 - 1.260 Rural (ref) -- -- -- -- Urban 0.962 0.527 0.059 0.854 - 1.084 Not head of household (ref) -- -- -- -- Head of household 1.705*** 0.000 0.018 1.670 - 1.742 Number of control issues -- -- -- -- Wife beating not justified for any reason (ref) 1.338* 0.025 0.174 1.037 - 1.725 Wife beating justified -- -- -- -- Father didn't beat mother (ref) 1.692*** 0.000 0.068 1.564 - 1.831 Father beat mother -- -- -- -- Women's decision-making autonomy 1.000 0.997 0.102 0.818 - 1.221 Partner education- none (ref) 0.974 0.794 0.099 0.798 - 1.189 Partner education- Primary 0.791* 0.039 0.090 0.634 - 0.988 Partner education- Secondary -- -- -- -- 20 Partner doesn't use alcohol or drugs (ref) 2.011*** 0.000 0.111 1.805 - 2.241 Partner uses alcohol or drugs -- -- -- -- Constant 0.046*** 0.000 0.009 0.032 - 0.068 Fixed effects District Observations 21,007 * p<0.05, **p<0.01, **p<0.001 Liberia Liberia Past-Year IPV Analysis In Liberia, one-third of women who experienced lifetime IPV also experienced this violence in the past year, a number that rose to 44% among displaced women. As with the Colombia analysis, women who experienced past-year IPV were compared to women who have not experienced IPV. The analytical sample was further restricted to women who stated they have been in their place of residence for over a year, ensuring that migration would predate past-year violence. This analysis helps establish temporality between when migration occurred compared to recent violence. Table 11. Past-year IPV by Forced Migration Status in Liberia Forced Migration Liberia No Yes Total No IPV 2292 231 2523 65.39% 56.07% 64.41% Past-year IPV 1213 181 1394 34.61% 43.93% 35.59% Total 3505 412 3917 100.00% 100.00% 100.00% First row has frequencies and second row has column percentages In Liberia, women who have been displaced are 55% more likely to experience past-year IPV compared to non-displaced women (aOR 1.55, p<0.01). Younger age of marriage, no longer being in a union, receiving secondary versus no education, having a partner who used alcohol or has a great number of control issues, and having witnessed her father beat her mother were all risk factors. As before, having a partner who received secondary education was protective. Interestingly, many economic factors not significant in the lifetime IPV model do reach significance here. Household headship and being currently employed were both protective against past-year IPV, as is being in the highest wealth quintile. Table 12. Adjusted Model of Past-year IPV and Displacement in Liberia aOR p-value SE Confidence Interval No forced migration (ref) Forced migration 1.545** 0.005 0.238 1.143 - 2.089 Age married 0.988 0.335 0.012 0.964 - 1.013 No education (ref) Primary 1.207 0.084 0.131 0.975 - 1.494 Secondary 1.248 0.289 0.261 0.829 - 1.879 Higher 2.268** 0.001 0.563 1.395 - 3.689 Currently partnered (ref) Formerly partnered 1.575* 0.012 0.285 1.105 - 2.245 Not employed in past year (ref) Employed in past year 0.853 0.093 0.081 0.709 - 1.027 21 Poorest wealth quintile (ref) Poorer wealth quintile 1.055 0.791 0.213 0.710 - 1.568 Middle wealth quintile 1.075 0.698 0.199 0.747 - 1.546 Richer wealth quintile 1.110 0.581 0.210 0.766 - 1.608 Richest wealth quintile 1.023 0.922 0.240 0.646 - 1.620 Number of children under 5 1.081 0.118 0.054 0.981 - 1.191 Rural (ref) Urban 1.577** 0.002 0.229 1.187 - 2.096 Not head of household (ref) Head of household 0.679** 0.001 0.079 0.541 - 0.853 Number of control issues 1.441*** 0.000 0.059 1.329 - 1.562 Wife beating not justified for any reason (ref) Wife beating justified 1.258 0.116 0.183 0.945 - 1.674 Father didn't beat mother (ref) Father beat mother 1.903*** 0.000 0.215 1.525 - 2.373 Partner education- none (ref) Partner education- Primary 1.054 0.709 0.149 0.799 - 1.390 Partner education- Secondary 0.859 0.197 0.101 0.681 - 1.082 Partner education- Higher education 0.628 0.088 0.171 0.368 - 1.072 Partner doesn't use alcohol or drugs (ref) Partner uses alcohol or drugs 2.358*** 0.000 0.279 1.870 - 2.974 Constant 0.111*** 0.000 0.043 0.052 - 0.237 Fixed effects Observations 2,874 * p<0.05, **p<0.01, **p<0.001 Liberia Lifetime IPV Analysis Nearly one in four women in Liberia has experienced IPV during her lifetime (39%, n=1,545) – a number that rises to nearly 50% for displaced women. Table 13. Lifetime IPV by Forced Migration Status in Liberia Forced Migration No Yes Total No IPV 2154 218 2372 61.46% 52.91% 60.56% IPV 1351 194 1545 38.54% 47.09% 39.44% Total 3505 412 3917 100.00% 100.00% 100.00% In Liberia, forced displacement was significantly and consistently associated with lifetime IPV across the stepwise model fitting process. 4 In the unadjusted model, displaced women were 45% more likely to report lifetime IPV compared to their non-displaced counterparts (OR 1.45, P<0.05). In the final adjusted model, women who experienced forced displacement were 50% more likely to report lifetime IPV (aOR 1.49, p<0.01). Having received any education versus no education was a risk factor for lifetime IPV, with these associations reaching significance for primary education (aOR 1.23, p<0.05) and higher education (aOR 1.79, p<0.05). Women who were formerly in a union, compared to currently in a union, had nearly 60% greater odds of IPV 4 In Liberia, the variable related to women’s decision-making autonomy was dropped due to small sample size and lack of variation in this predictor. 22 compared to currently married women (aOR 1.60, p<0.01). Residing in an urban versus rural setting also increased the risk of IPV (aOR 1.57, p=0.001). A number of documented risk factors for IPV emerged as significant in the Liberia model. For each additional control issue women faced by her partner, she faced 41% greater odds of IPV (aOR 1.41, p<0.001). Women who had a father who beat their mother were nearly twice as likely to experience IPV (aOR 1.98, p<0.001). Women who had partners who used alcohol or drugs were more than twice as likely to experience IPV (aOR 2.33, p<0.001). Being the head of household was protective against lifetime IPV – women who identified in this role had 25% lower odds of lifetime IPV compared to their counterparts who do not serve as household head (0.75, p<0.05). Table 14. Final Adjusted Model of Lifetime IPV and Displacement in Liberia aOR P-value SE Confidence Interval No forced migration (ref) -- -- -- -- Forced migration 1.492** 0.006 0.217 1.121 - 1.985 Age married 0.988 0.338 0.013 0.964 - 1.013 No education (ref) -- -- -- -- Primary 1.231* 0.021 0.111 1.032 - 1.468 Secondary 1.279 0.154 0.220 0.912 - 1.792 Higher 1.788* 0.021 0.449 1.092 - 2.925 Currently partnered (ref) -- -- -- -- Formerly partnered 1.596** 0.002 0.243 1.185 - 2.150 Not employed in past year (ref) -- -- -- -- Employed in past year 0.887 0.192 0.082 0.740 - 1.062 Poorest wealth quintile (ref) -- -- -- -- Poorer wealth quintile 1.114 0.557 0.205 0.777 - 1.598 Middle wealth quintile 1.076 0.625 0.161 0.802 - 1.443 Richer wealth quintile 1.134 0.424 0.178 0.833 - 1.542 Richest wealth quintile 1.029 0.881 0.198 0.706 - 1.501 Number of children under 5 1.081 0.065 0.046 0.995 - 1.174 Rural (ref) -- -- -- -- Urban 1.574** 0.001 0.218 1.201 - 2.065 Not head of household (ref) -- -- -- -- Head of household 0.753* 0.014 0.087 0.601 - 0.944 Number of control issues 1.410*** 0.000 0.056 1.305 - 1.524 Wife beating not justified for any reason (ref) -- -- -- -- Wife beating justified 1.269 0.081 0.173 0.971 - 1.657 Father didn't beat mother (ref) -- -- -- -- Father beat mother 1.983*** 0.000 0.204 1.621 - 2.425 Partner education- none (ref) -- -- -- -- Partner education- Primary 1.045 0.744 0.141 0.802 - 1.361 Partner education- Secondary 0.848 0.125 0.091 0.686 - 1.047 Partner education- Higher education 0.737 0.230 0.188 0.447 - 1.213 Partner doesn't use alcohol or drugs (ref) -- -- -- -- Partner uses alcohol or drugs 2.327*** 0.000 0.260 1.869 - 2.898 Constant 0.126*** 0.000 0.049 0.059 - 0.269 Fixed effects District Observations 3,070 * p<0.05, **p<0.01, **p<0.001 23 Liberia Lifetime Injury from IPV Analysis In Liberia, 12% of women who ever experienced IPV also reported injury from IPV. As with other forms of IPV, this percentage is higher among displaced (14%) versus non-displaced (11%) women. Table 15. Lifetime Injury from IPV by Forced Migration Status in Liberia Injury from IPV Yes No Total No injury 3122 361 3483 88.64% 86.36% 88.40% Injury 400 57 457 11.36% 13.64% 11.60% Total 3522 418 3940 100.00% 100.00% 100.00% In the adjusted model, forced displacement does not rise to significance as a risk factor for injury from IPV, although the direction of the association remains the same (aOR 1.28, p=0.27). Table 16. Final Adjusted Model of Injury from IPV and Displacement in Liberia aOR P-value SE Confidence Interval No forced migration (ref) -- -- -- -- Forced migration 1.277 0.265 0.280 0.831 - 1.963 Age married 0.982 0.283 0.016 0.951 - 1.015 No education (ref) -- -- -- -- Primary 1.194 0.172 0.155 0.926 - 1.540 Secondary 1.298 0.298 0.326 0.794 - 2.124 Higher 1.141 0.644 0.327 0.651 - 2.000 Currently partnered (ref) -- -- -- -- Formerly partnered 2.582*** 0.000 0.485 1.787 - 3.732 Not employed in past year -- -- -- -- (ref) Employed in past year 0.760* 0.045 0.104 0.580 - 0.994 Poorest wealth quintile (ref) -- -- -- -- Poorer wealth quintile 1.105 0.642 0.238 0.725 - 1.686 Middle wealth quintile 1.109 0.664 0.264 0.696 - 1.767 Richer wealth quintile 0.629 0.113 0.184 0.354 - 1.116 Richest wealth quintile 0.664 0.256 0.239 0.328 - 1.345 Number of children under 5 1.041 0.536 0.068 0.916 - 1.183 Rural (ref) -- -- -- -- Urban 1.359 0.091 0.246 0.952 - 1.938 Not head of household (ref) -- -- -- -- Head of household 0.583** 0.002 0.103 0.412 - 0.826 Number of control issues 1.639*** 0.000 0.092 1.469 - 1.829 Wife beating not justified for -- -- -- -- any reason (ref) Wife beating justified 1.468* 0.015 0.232 1.077 - 2.001 Father didn't beat mother (ref) -- -- -- -- Father beat mother 2.182*** 0.000 0.292 1.677 - 2.837 Partner education- none (ref) -- -- -- -- Partner education- Primary 0.773 0.166 0.144 0.537 - 1.113 24 Partner education- 0.773 0.099 0.121 0.569 - 1.050 Secondary or higher Partner doesn't use alcohol or -- -- -- -- drugs (ref) Partner uses alcohol or 3.838*** 0.000 0.627 2.787 - 5.285 drugs Constant 0.021*** 0.000 0.013 0.007 - 0.071 Fixed effects District Observations 2,276 * p<0.05, **p<0.01, **p<0.001 Liberia Conflict-Intensity, Forced Displacement and IPV In the following analysis, we incorporate independent data from ACLED about violence occurring in a woman’s district of residence in the 10 years preceding the survey. In this model, we see that conflict fatalities are highly associated with past-year IPV. For every additional conflict death in a woman’s district of residence, she is 0.2% more likely to experience IPV (aOR1.002, p=0.001). Women experiencing forced migration also have 60% higher odds of past- year IPV (aOR 1.60, p<0.01). Both conflict and displacement are independently and significantly associated with past-year IPV. Table 17. Association of Past-Year IPV with Forced Migration Status and Conflict Fatalities in Colombia aOR P-value SE Confidence Interval No forced migration (ref) -- -- -- -- Forced migration 1.601** 0.002 0.248 1.182 - 2.168 Number of fatalities in district 1.002** 0.002 0.001 1.001 - 1.004 over 10 years Age married 0.99 0.296 0.013 0.962 - 1.012 No education (ref) -- -- -- -- Primary 1.182 0.115 0.125 0.960 - 1.455 Secondary 1.103 0.621 0.218 0.748 - 1.626 Higher 2.062* 0.009 0.568 1.202 - 3.537 Currently partnered (ref) -- -- -- -- Formerly partnered 1.33 0.127 0.251 0.922 - 1.929 Not employed in past year -- -- -- -- (ref) Employed in past year 0.837* 0.046 0.075 0.703 - 0.997 Poorest wealth quintile (ref) -- -- -- -- Poorer wealth quintile 1.03 0.862 0.199 0.709 - 1.509 Middle wealth quintile 1.01 0.956 0.197 0.690 - 1.481 Richer wealth quintile 1.03 0.871 0.190 0.718 - 1.479 Richest wealth quintile 0.96 0.844 0.215 0.616 - 1.486 Rural (ref) -- -- -- -- Urban 1.523** 0.008 0.241 1.117 - 2.077 Not head of household (ref) -- -- -- -- Head of household 0.684*** 0.000 0.069 0.561 - 0.833 Number of control issues 1.431*** 0.000 0.054 1.328 - 1.542 Wife beating not justified for -- -- -- -- any reason (ref) Wife beating justified 1.23 0.119 0.164 0.948 - 1.599 Father didn't beat mother (ref) -- -- -- -- Father beat mother 1.695*** 0.000 0.192 1.357 - 2.117 Partner education- none (ref) -- -- -- -- 25 Partner education- Primary 1.07 0.622 0.146 0.818 - 1.399 Partner education- 0.86 0.223 0.109 0.667 - 1.099 Secondary or higher Partner doesn't use alcohol or -- -- -- -- drugs (ref) Partner uses alcohol or drugs 2.285*** 0.000 0.246 1.851 - 2.821 Constant 0.135*** 0.000 0.044 0.0713 - 0.257 Fixed effects District Observations 3,057 * p<0.05, **p<0.01, **p<0.001 Strengths and Limitations Both of the data sets used in this study are cross-sectional, making it difficult to establish causality between conflict and the outcome of IPV. However, the fact that this analysis looked at conflict events in the 10 years preceding the DHS helps establish temporality and suggests that previous conflict may influence IPV, particularly past-year IPV. Longitudinal studies with conflict-affected and mobile populations are particularly challenging - however tracing a woman’s experience over time through longitudinal surveys that ask about personal exposure to conflict, experiences with displacement and experiences with violence would be a more reliable way to understand the links between these abuses. Additionally, displacement was measured differently in Colombia and Liberia - making it potentially challenging to compare the two data sets. The Colombia measure is arguably more precise and might have resulted in better identification of displacement. The Liberia displacement question was relatively imprecise. However, this limitation in the Liberia DHS would likely have skewed the model results towards a null result rather than towards a spuriously strong association between IPV and displacement. Despite some imprecision in wording and lack of exact comparability, the questions identifying forced displacement seem to identify a meaningful construct. In the future, standardized questions about forced displacement in the DHS would allow for more comparable and precise measurement of this important outcome. Another limitation arises because the conflict analysis drew on different conflict data for each country (ACLED in Liberia and UCDP in Colombia). UCDP should be understood as a baseline estimate that used only injuries and fatalities that are reported in the media, which is very likely an under-estimate. ACLED data has similar limitations. For each data set, it is unlikely that there are fewer events than those reported in the database, so both provide a lower level estimate of fatalities and may not have captured all relevant events. Despite this, each data source is widely used in the academic literature and both measure fatalities in each country. Analyzing each country separately helped account for potential differences in the way conflict was measured, but unfortunately resulted in a smaller sample size as a result of disaggregating by country. Given the dearth of data at the sub-district level, this project does not account for district-level characteristics other than conflict experience. In this case, the district covariate captures unmeasured district level characteristics that are not explicitly included in the model. Looking at district-level effects may still be a relatively unsophisticated way of measuring exposure to conflict. In the future, looking at smaller administrative units, or looking at distance from the conflict event to the DHS sampling cluster could be informative. 26 A strength of this study is its use of data from the Demographic and Health Surveys (DHS), which are nationally-representative cross-sectional household surveys. A multi-level modeling approach accounted for the nested structure of the data, with clustering of individuals within districts. This temporal and geographic variation in conflict within the same country allowed for analyses that examine the impact of conflict on key IPV outcomes. Discussion Colombia and Liberia represent two very different contexts, but both are affected by a history of conflict, displacement and uneven recovery from this political violence. Both are the oldest republics in their continents; Colombia became an independent nation from colonial invaders in 1810 and Liberia declared independence in 1847. Both countries saw steady increases in GDP and GNI in post-conflict (or increasingly peaceful) times; Liberia had significant increases from 2003-2016 and Colombia from 2003-2014 (World Bank, n.d.). In past decades, both countries have attempted to usher in peace through talks, accords and treaties, each with varying levels of success. Colombia and Liberia both experience higher than the global average (27%) of intimate partner violence (WHO, 2021). While there are similarities between these two contexts, there are also differences – both in the political histories as well as the types of violence women are most likely to experience. In Liberia, women who had ever experienced IPV were more likely to report this abuse in the past 12 months compared to women in Colombia. In Colombia, women were twice as likely to report injury from IPV compared to Liberia. Yet, the findings from this study show the striking similarity in the heightened risk that political violence can have on women. Forced displacement is highly and significantly associated with increased lifetime and past-year IPV. Displaced women in Colombia and Liberia reported between 43% and 45% greater odds of lifetime IPV and between 40% and 55% greater odds of experiencing past-year IPV compared to their non-displaced counterparts. In Colombia, displaced women were also 30% more likely to experience lifetime injury-causing IPV. These results suggest that displacement is associated with increased risk of lifetime and past-year IPV, but may also impact the risk of experiencing the most violent forms of IPV that result in injury, as seen in Colombia. Additionally, this work is the first to our knowledge to look at both displacement and exposure to conflict, and to explore the relationship between these two risk factors as they relate to women’s experiences of violence. In each country, both conflict and displacement were independently and significantly associated with past-year IPV. In Liberia, a woman’s odds of past-year IPV increased by 0.2% for each additional conflict fatality in her district, and displacement increased the risk of the same outcome by 60%. Similar results were seen in Colombia. The odds of experiencing past-year IPV increased by 0.1% for each additional conflict fatality, and increased by 40% among displaced compared to non-displaced women. This research finds that some risk factors for IPV are context specific. For instance, in Liberia higher education was a risk factor for IPV, while employment was not generally associated with violence. In Colombia, education was not a significant predictor of the outcome, but 27 unemployment was a risk factor. Despite some context-specific differences, this work highlights that most risk factors highly associated with IPV are shared across both countries. For instance, childhood experiences of violence and partner alcohol use are significant and powerful risk factors for women’s experiences of violence (Abramsky et al.2011, Ellsberg et al 2020, Feseha et. al, 2012; Wachter et. al, 2017; Tlapek, 2015). While this research cannot speak directly to the drivers of increased violence among displaced women, an increasingly strong literature suggests some of the mechanisms that may be at play. (Friedemann-Sánchez & Lovatón, 2012; Hynes 2012; Wirtz et al, 2014). Risk factors for IPV are often discussed as part of an ecological framework – the interplay of complex factors at the societal, community, family and individual levels all play a role (Heise, 1998; Stark et. al, 2017; Usta & Singh, 2015). During conflict and displacement, a number of pre-existing factors may be exacerbated. The results here show an association between IPV and having a father who was violent toward a mother, and having more permissive attitudes towards violence oneself (Feseha et. al, 2012; Logie et. al, 2019). Being in a relationship based on control and unequal power is also a risk factor (Logie et. al, 2019). While existing risks may intensify, new vulnerabilities may also emerge as a result of political instability. Conflict may also lead to higher rates of violence which can, in turn, normalize abuse going forward (Ager et. al, 2018; Ellsberg et. al, 2020; Kelly et al, 2018; Kelly et al, 2019; Tlapek, 2015). In Colombia, qualitative evidence suggests that some men may place the blame for the decision to flee on women, increasing household tensions resulting in increased violence (Wirtz et al., 2014). In African contexts, women who have been directly victimized during war, including through militarized sexual violence, may then face stigmatization and retribution within the home (Albutt et al, 2016; Kelly, 2011; Kelly et al, 2012; Kelly at al 2017). Men who have experienced violence during conflict may then be more likely to perpetrate violence in the home post-conflict (Clark et al., 2010; Gupta et al., 2012; Vinck & Pham, 2013). As of 2019, Colombia had over five and a half million internally displaced people due to conflict (IDMC, 2019), the second largest in the world after the Syrian Arab Republic. While the number of displaced individuals in Liberia is currently much lower than during the conflict, lessons from this work are applicable to the multiple current crises in Africa, including in the Democratic Republic of Congo and South Sudan. The policy and programmatic implications of this research support previous findings and recommendations within the literature; building upon the limited but growing body of evidence on effective IPV prevention models in non-refugee settings. However, the challenge of adapting interventions to refugee and displacement contexts remains. Recent publications by the Global Women’s Institute at George Washington University and the International Rescue Committee (2016 and 2019) underscore the severe lack of available evidence on IPV interventions during conflict and humanitarian crisis. Interventions that have been evaluated include firewood distribution and alternative fuel programs, which have shown reductions of IPV in refugee camps (Stark & Ager, 2011). The evidence also suggests that economic interventions alone are not enough and can, sometimes, exacerbate or drive IPV increases in a displacement context (Global Women’s Institute and International Rescue Committee, 2016). To address this concern, in Burundi, the International Rescue Committee successfully combined economic empowerment efforts with communication 28 courses to reduce IPV, which also demonstrated the importance of addressing the underlying risks of violence (IRC, 2013). The IRC work and similar studies suggest inclusion of gender training within socio-economic programming can offer social and health benefits. Similar to Mootz et al. (2018), the evidence presented in this study finds that community, organizational, and policy-level interventions, in addition to psychological interventions, are necessary for displaced women and girls. Evidence from a community-based prevention program in Côte d’Ivoire points to involving men and women simultaneously in prevention programs (Tappis et al., 2016) and building interventions that engage community members from the start (Robbers & Morgan, 2017). Similarly, in Somalia the Communities of Care program demonstrated that harmful social norms that underlie the use of violence against women can be positively shifted, even in ongoing conflict (Glass et al. 2019). In the wake of armed conflict and forced displacement, policies must prioritize both prevention and response, acknowledging the drivers of violence during conflict and displacement. Space should be made to address increased IPV during displacement and reject the notion that reconciliation only includes “public” violence or that committed during the conflict by combatants. The research builds on the evidence across the globe to underscore that incidents of violence continue after a conflict has ended or peace negotiations have started (Aoláin et al., 2015; Kelly et al, 2018; Kelly et al 2019; Otsby 2016, Stark and Ager, 2011). Moreover, women displaced by conflict may suffer multiple traumas that need support and physiological care. They may have been victims of violence not only at the hands of their partners but also other conflict related actors. This reality often limits their capacity to take advantage of opportunities for economic independence, a change that may protect them from continued experiences of IPV. As part of humanitarian, state and peacebuilding efforts, women in conflict-affected locations and displaced women in host communities should be able to easily access physical and mental health care services and economic autonomy programming to help them heal from the impacts of the violence and seize opportunities to leave violent relationships. Recognizing the increased burden of IPV for conflict-affected women should permeate every interaction to assist women in this situation. Quality and confidential multisectoral services for displaced IPV survivors should be prioritized for funding at the onset of humanitarian assistance interventions, including in all protection and response efforts. GBV, including IPV, should also remain prioritized issues in peacebuilding and state-building efforts, with clear recognition of IPV as a pressing, long-term consequence of conflict. By including IPV and sexual violence in peace accords, women who experienced violence can also access justice and reparations guaranteed by peace processes. Throughout the peace and state- building process, acknowledging the experience of IPV and sexual violence of both combatants and civilians will be important to usher in stable peace. The investment in both prevention and response will not only provide women opportunities to live a life free from violence, but can have positive impacts for children in the same household (Guedes et al 2016). This study shows that women who have been forcibly displaced have unique experiences of IPV meriting recognition and inclusion in state and peacebuilding programs and policies. In both 29 Colombia and Liberia, women were impactful actors in the peace processes. In Colombia, the participation of women resulted in a broader peace agenda, the successful negotiation of local cease-fires, increased accountability, and increased public support for the peace deal (Bouvier, 2016). Colombia’s Constitutional Court and peace process recognized that violence against women and girls was used by all armed actors, paving the way for its treatment and investigation and accountability in the implementation of the peace accords. This recognition did not extend to IPV. Similarly, in Liberia, women’s groups are credited with bringing then President Charles Taylor to the negotiation table and working toward disarmament, demobilization, and reintegration (Shilue and Fagen, 2014). However, displaced women are notably missing from these peace negotiations and post-conflict transition processes (Gururaja, 2020). While discussion of conflict-related sexual violence has been widely recognized by the UN Security Council as a threat to peace and security, there is relatively little recognition that other forms of violence -including violence within the home – also has implications for peace and security. This study reinforces previous scholarship that calls for transitional reforms and reformers to acknowledge and include forms of GBV that occur during the conflict period as well as in its aftermath, both in public and private spheres. Conclusion These results provide some of the first population-based, multi-country evidence that both forced displacement as well as living in a conflict affected location independently and significantly heighten the risk of IPV for women in fragile environments. These analyses point to the fact that both displacement and political conflict can independently heighten a woman’s risk of IPV post- conflict – a finding that points to a need for more robust policies to identify, address and prevent IPV post-conflict and in displacement settings. Women in conflict-affected settings experience intimate partner violence before, during and after displacement. As conflict intensifies, so do the experiences of IPV and the physical and mental health consequences of the violence. Because we know that conflict is not only impacting violence in public spaces, but also in private spaces, peace and reconciliation processes should acknowledge and address all forms of violence that were impacted by conflict. Acknowledging IPV and sexual violence as a form of conflict- related violence means this important issue must be incorporated in state and peace building efforts. This inclusion will allow for formulation of policies and programs built to give women a chance to access services needed to begin a path to healing from the violence they have experienced. This in turn will enable women to take advantage of opportunities to develop freely and contribute to a peaceful state. Acknowledging that IPV is happening at a large scale in conflict allows for significant funding to be mobilized directly for IPV prevention programs in conflict affected populations and peace building programming. Women bear the brunt of conflict – not only during active hostilities, but in the years after hostilities formally end. We cannot achieve justice and security for all women without acknowledging how conflict impacts their lives in both visible and hidden ways. 30 References Abramsky T, Watts CH, Garcia-Moreno C, Devries K, Kiss L, Ellsberg M, Jansen HA, Heise L. 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. 2011 Feb 16;11:109. doi: 10.1186/1471-2458-11-109. PMID: 21324186; PMCID: PMC3049145. Adjei, S. B., & Mpiani, A. (2018). Bride price, cultural and gender identity, and husband-to-wife abuse in Ghana. Victims & Offenders, 13(7), 921-937. Ager, A., Bancroft, C., Berger, E., & Stark, L. (2018). Local constructions of gender-based violence amongst IDPs in northern Uganda: Analysis of archival data collected using a gender- and age- segmented participatory ranking methodology. Conflict and Health, 12(1), 10. Akram, S. M. (2013). Millennium development goals and the protection of displaced and refugee women and girls. Laws, 2(3), 283-313. Albutt, K., Kelly, J., Kabanga, J., & VanRooyen, M. (2016). Stigmatisation and rejection of survivors of sexual violence in eastern Democratic Republic of the Congo. Disasters Alzate, M. M. (2008). The sexual and reproductive rights of internally displaced women: the embodiment of Colombia's crisis. Disasters, 32(1), 131-148. Amnesty International. (2007). Amnesty International Report 2007 - Liberia. Amnesty International. https://www.refworld.org/docid/46558ed316.html Amnesty International. (2008). Amnesty International Report 2008 - Liberia. Amnesty International. https://www.refworld.org/docid/483e279b4b.html Annan, J., & Brier, M. (2010). The risk of return: Intimate partner violence in Northern Uganda's armed conflict. Social Science & Medicine, 70(1), 152-159. Anand, S. & Sen, A. (1994). Human development index: Methodology and Measurement. Human Development Occasional Papers (1992-2007). HDOCPA-1994-02, Human Development Report Office (HDRO), UNDP. Aoláin, F. N., O'Rourke, C., & Swaine, A. (2015). Transforming reparations for conflict-related sexual violence: Principles and practice. Harv. Hum. Rts. J., 28, 97. Bandura, A. (1973). Aggression: A social learning analysis. Oxford, England: Prentice-Hall. Baranyi, S., Beaudet, P., & Locher, U. (2011). World development report 2011: Conflict, security, and development. Canadian Journal of Development Studies/Revue Canadienne d'Études du Développement, 32(3), 342-349. Barnett, O. W., Miller-Perrin, C. L., & Perrin, R. D. (2010). Family violence across the lifespan: An introduction. Thousand Oaks, California: Sage. 31 BBC News. (2018). Liberia country profile. Retrieved at http://www.bbc.com/news/world-africa- 13729504 Bohstedt, J. (1994). The dynamics of riots: Escalation and diffusion/contagion. The dynamics of aggression: Biological and social processes in dyads and groups, 257-306. Bouvier, V. M. (2016). Gender and the role of women in Colombia's peace process. United States Institute of Peace. Browne, A., Bennouna, C., Asghar, K., Correa, C., Harker-Roa, A., & Stark, L. (2019). Risk and refuge: Adolescent boys’ experiences of violence in “post-conflict” Colombia. Journal of interpersonal violence, 0886260519867150. Burgert, C. R., J. Colston, T. Roy, and B. Zachary (2013). Geographic displacement procedure and georeferenced data release policy for the Demographic and Health Surveys. DHS Spatial Analysis Report No. 7. Calverton, Maryland, USA: ICF International. Cabellero, Antonio. 2018 historia de Colombia y sus oligarquías. Editorial Planeta. Bogota, Colombia Cardoso, L. F et al., 2016. What Factors Contribute to Intimate Partner Violence Against Women in Urban, Conflict-Affected Settings? Qualitative Findings from Abidjan, Côte d’Ivoire. Journal of urban health, 93(2), pp.364–378. Carlson, S. (2004). Contesting and reinforcing patriarchy: An analysis of domestic violence in the Dzaleka refugee camp. CIA World Factbook. (2015). Liberia Country Profile. [Central Intelligence Agency website] Retrieved at https://www.cia.gov/library/publications/the-world-factbook/geos/print_li.html. Clark, C. J., Everson-Rose, S. A., Suglia, S. F., Btoush, R., Alonso, A., & Haj-Yahia, M. M. (2010). Association between exposure to political violence and intimate-partner violence in the occupied Palestinian territory: A cross-sectional study. The Lancet, 375(9711), 310-316. Cummings, E. M., Schermerhorn, A. C., Merrilees, C. E., Goeke-Morey, M. C., Shirlow, P., & Cairns, E. (2010). Political violence and child adjustment in Northern Ireland: Testing pathways in a social-ecological model including single-and two-parent families. Developmental psychology, 46(4), 827. Cummings, E. M., Merrilees, C. E., Schermerhorn, A. C., Goeke-Morey, M. C., Shirlow, P., & Cairns, E. (2011). Longitudinal pathways between political violence and child adjustment: The role of emotional security about the community in Northern Ireland. Journal of abnormal child psychology, 39(2), 213-224. Dubow, E. F., Boxer, P., Huesmann, L. R., Shikaki, K., Landau, S., Gvirsman, S. D., & Ginges, J. (2009). Exposure to conflict and violence across contexts: Relations to adjustment among Palestinian children. Journal of Clinical Child & Adolescent Psychology, 39(1), 103-116. 32 Ekhator-Mobayode, Uche Eseosa, Hanmer, Lucia C, Rubiano Matulevich, Eliana Carolina, & Arango, Diana Jimena. (2020). Effect of Armed Conflict on Intimate Partner Violence : Evidence from the Boko Haram Insurgency in Nigeria. Policy File, Policy File, 2020-03-02. Ellsberg M, Ovince J, Murphy M, Blackwell A, Reddy D, Stennes J, et al. (2020) No safe place: Prevalence and correlates of violence against conflict-affected women and girls in South Sudan. PLoS ONE 15(10): e0237965. https://doi.org/10.1371/journal.pone.0237965 Eves, R. (2019). ‘Full price, full body’: norms, brideprice and intimate partner violence in highlands Papua New Guinea. Culture, health & sexuality, 21(12), 1367-1380. Fagan, J., Wilkinson, D. L., & Davies, G. (2007). Social contagion of violence. In The Cambridge handbook of violent behavior and aggression (pp 688–727). New York, NY: Cambridge University Press. Feseha, G, G/mariam, A., & Gerbaba, M. (2012). Intimate partner physical violence among women in Shimelba refugee camp, northern Ethiopia. BMC Public Health, 12(1), 125. Fox, J. (2004). Is ethnoreligious conflict a contagious disease? Studies in Conflict & Terrorism, 27(2), 89-106. Gallegos, J.V, & Gutierrez, I. A. (2011). The effect of civil conflict on domestic violence: The case of Peru. [Unpublished manuscript]. Retrieved from http://jvgalleg.mysite.syr.edu/default_files/Research_files/Gallegos- Gutierrez%20%20Civil%20Conflict%20and%20Domestic%20Violence%20JDE.pdf Georgetown Institute for Women, Peace and Security and Peace Research Institute Oslo. 2017. Women, Peace and Security Index 2017/18: Tracking Sustainable Peace through Inclusion, Justice, and Security for Women. Washington, DC: GIWPS and PRIO Glass, N. et al,. (2019). Effectiveness of the Communities Care programme on change in social norms associated with gender-based violence (GBV) with residents in intervention compared with control districts in Mogadishu, Somalia. BMJ open, 9(3), e023819. https://doi.org/10.1136/bmjopen- 2018-023819 Global IDP Project. (2003). Profile of internal displacement in Liberia. Retrieved from https://reliefweb.int/report/liberia/profile-internal-displacement-liberia Global Women’s Institute and International Rescue Committee. (2019). What works to prevent violence against women and girls in conflict and humanitarian crisis: Synthesis Brief. Washington DC and London: The George Washington University and the International Rescue Committee. Available at https://reliefweb.int/report/world/what-works-prevent-violence-against-women-and- girls-conflict-and-humanitarian-crisis 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. Available at https://www.whatworks.co.za/resources/evidence-reviews/item/215- evidence-brief-what-works-to-prevent-respond-to-violence-against-women-and-girls-in-conflict-and- humanitarian-settings 33 Global Women’s Institute and International Rescue Committee. (2019). Synthesis brief: What works to prevent violence against women and girls in conflict and humanitarian crisis? Washington DC and London: The George Washington University and the International Rescue Committee. Available at https://globalwomensinstitute.gwu.edu/sites/g/files/zaxdzs1356/f/downloads/P868%20IRC%20Synth esis%20brief%20report_LR.PDF Guedes A, Bott S, Garcia-Moreno C, Colombini M. Bridging the gaps: a global review of intersections of violence against women and violence against children. Glob Health Action. 2016 Jun 20;9:31516. doi: 10.3402/gha.v9.31516. PMID: 27329936; PMCID: PMC4916258. Gupta, J., Acevedo-Garcia, D., Hemenway, D., Decker, M. R., Raj, A., & Silverman, J. G. (2009). Premigration exposure to political violence and perpetration of intimate partner violence among immigrant men in Boston. American Journal of Public Health, 99(3), 462-469. Gupta, J., Reed, E., Kelly, J., Stein, D. J., & Williams, D. R. (2012). Men's exposure to human rights violations and relations with perpetration of intimate partner violence in South Africa. Journal of epidemiology and Community Health, 66(6), e2-e2. Gupta, J., Falb, K. L., Lehmann, H., Kpebo, D., Xuan, Z., Hossain, M., ... & Annan, J. (2013). Gender norms and economic empowerment intervention to reduce intimate partner violence against women in rural Côte d’Ivoire: a randomized controlled pilot study. BMC international health and human rights, 13(1), 1-12. Gururaja, S. (2000). Gender dimensions of displacement. Forced Migration Review, 9, 13-16. Retrieved at https://www.peacewomen.org/assets/file/Resources/Academic/disp_genderdimensionsdisp_forcedmig ration_2000.pdf Heise, L. L. (1998). Violence against women: An integrated, ecological framework. Violence against Women, 4(3), 262-290. Heise, L. (2011) What Works to Prevent Partner Violence? An Evidence Overview. Working Paper. STRIVE Research Consortium, London School of Hygiene and Tropical Medicine, London. Retrieved from: http://researchonline.lshtm.ac.uk/21062/ Hindin, M. J., Kishor, S., & Ansara, D. L. (2008). Intimate partner violence among couples in 10 DHS countries: Predictors and health outcomes. DHS Analytical Studies No. 18. Calverton, Maryland, USA: Macro International Inc. Horn, R. (2010). Exploring the impact of displacement and encampment on domestic violence in Kakuma refugee camp. Journal of refugee studies, 23(3), 356-376. Hudson, V.M.; Bowden, D.L.; Nielsen, P.L. (2020) The First Political Order: How Sex Shapes Governance and National Security Worldwide. New York: Columbia University Press Human Rights Watch. (2008). Liberia: Events of 2007. Human Rights Watch World Report. Retrieved 03 02, 2021, from https://www.hrw.org/world-report/2008/country-chapters/liberia#fb527b 34 Hynes, M. E., Sterk, C. E., Hennink, M., Patel, S., DePadilla, L., & Yount, K. M. (2016). Exploring gender norms, agency and intimate partner violence among displaced Colombian women: A qualitative assessment. Global public health, 11(1-2), 17-33. Hynes, M., Ward, J., Robertson, K., & Crouse, C. (2004). A determination of the prevalence of gender‐based violence among conflict‐affected populations in East Timor. Disasters, 28(3), 294-321. IASC. (2015). Guidelines for Integrating Gender-Based Violence Interventions in Humanitarian Action, 2015. Retrieved at https://interagencystandingcommittee.org/system/files/2015-iasc-gender- based-violence-guidelines_lo-res.pdf IDMC. (2019). Global Internal Displacement Database. Retrieved at https://www.internal- displacement.org/database/displacement-data IFRC. (2012). World Disasters Report 2012: Focus on Forced Migration and Displacement. Edited by Roger Zetter. Geneva: Switzerland; pg 117 Retrieved from https://www.ifrc.org/Global/Documents/Secretariat/2012_WDR_Full_Report.pdf IRC (2013). Getting Down to Business: Women’s Economic and Social Empowerment in Burundi, New York: International Rescue Committee International Crisis Group (2003). Liberia: Security Challenges,” Africa Report no. 71, Nov. 3, 2003, 8, Accessed at: www.crisisgroup.org/en/regions/africa/west-africa/liberia/071-liberia-security- challenges.aspx. Institute of Medicine (IOM) and NRC (National Research Council). 2012. Contagion of violence: Workshop summary. Washington, DC: The National Academies Press. Johnson, K., Asher, J., Rosborough, S., et al. (2008). Association of combatant status and sexual violence with health and mental health outcomes in post-conflict Liberia. Journal of the American Medical Association, 300(6), 676-690. Janko, M., Bloom, S. & Spencer, J. (2014, May). Community exposure to violent conflict increases the risk of intimate partner violence in Rwanda: Paper presented at the annual meeting of the Population Association of America. Boston, MA. Abstract. Retrieved from: http://paa2014.princeton.edu/abstracts/141125 Jewkes, R. (2002). Intimate partner violence: Causes and prevention. The Lancet, 359(9315), 1423- 1429. Jewkes, R., Jama-Shai, N., & Sikweyiya, Y. (2017). Enduring impact of conflict on mental health and gender-based violence perpetration in Bougainville, Papua New Guinea: a cross-sectional study. Plos one, 12(10), e0186062. Kathman, J. D. (2011). Civil war diffusion and regional motivations for intervention. Journal of Conflict Resolution, 55(6), 847-876. 35 Kelly, J. T., Betancourt, T. S., Mukwege, D., Lipton, R., & VanRooyen, M. J. (2011). Experiences of female survivors of sexual violence in eastern Democratic Republic of the Congo: A mixed-methods study. Conflict and Health, 5(1), 25. Kelly J, Kabanga J, Cragin W, Alcayna-Stevens L, Haider S, Vanrooyen M. “If your husband doesn’t humiliate you, other people won’t”: Gendered attitudes towards sexual violence in eastern Democratic Republic of Congo. Global Public Health. 2012; 7(3): 285-298. Doi:10.1080/17441692.2011.585344 Kelly J, Albutt K, Kabanga J, Anderson K, VanRooyen M. Rejection, acceptance and the spectrum between: understanding male attitudes and experiences towards conflict-related sexual violence in eastern Democratic Republic of Congo. BMC Women’s Health. 2017;17(1). Doi: 10.1186/s12905- 017-0479-7. Kelly, J. T., 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 global health, 3(2). Kelly J, Colantuoni E, Robinson C, Decker M. From political to personal violence: Links between conflict and non-partner physical violence in post-conflict Liberia. Global Public Health. 2019;14(12):1639-1652. Doi:10.1080/17441692.2019.1650949 Kim, J., Ferrari, G., Abramsky, T., Watts, C., Hargreaves, J., Morison, L., ... & Pronyk, P. (2009). Assessing the incremental effects of combining economic and health interventions: the IMAGE study in South Africa. Bulletin of the World Health Organization, 87, 824-832. Kirk, J. (2003). Women in contexts of crisis: Gender and conflict. Commissioned paper for the EFA Monitoring Report. Retrieved at http://unesdoc.unesco.org/images/0014/001467/146794e.pdf Kish, L. (1965). Survey Sampling. John Wiley and Sons, Inc., New York. Xvi-643. Kitzmann, K. M., Gaylord, N. K., Holt, A. R., & Kenny, E. D. (2003). Child witnesses to domestic violence: A meta-analytic review. Journal of Consulting and Clinical Psychology, 71(2), 339. Klugman, J., (2010). Human Development Report 2010. The Real Wealth of Nations: Pathways to Human Development. New York, NY: UNDP. Retrieved from http://hdr.undp.org/sites/default/files/reports/270/hdr_2010_en_complete_reprint.pdf. Koch, G. (1982). Intraclass correlation coefficient. In Samuel Kotz and Norman L. Johnson. Encyclopedia of Statistical Sciences. (pp. 213–217). New York, New York: John Wiley & Sons. Kohli, A., Perrin, N., Mpanano, R. M., Banywesize, L., Mirindi, A. B., Banywesize, J. H., ... & Glass, N. (2015). Family and community driven response to intimate partner violence in post-conflict settings. Social Science & Medicine, 146, 276-284. Liberia Institute of Statistics and Geo-Information Services (LISGIS) [Liberia], Ministry of Health and Social Welfare [Liberia], National AIDS Control Program [Liberia], and Macro International Inc. (2008). Liberia Demographic and Health Survey 2007. Monrovia, Liberia: Liberia Institute of Statistics and Geo-Information Services (LISGIS) and Macro International Inc. 36 Liebling-Kalifani, H et al,. (2011). Women war survivors of the 1989-2003 conflict in Liberia: The impact of sexual and gender-based violence. Journal of International Women’s Studies, 12(1), 1. Mootz, Jennifer J et al., (2018). Armed conflict, alcohol misuse, decision-making, and intimate partner violence among women in Northeastern Uganda: a population level study. Conflict and health, 12(1), p.37. Mullins, C. W., Wright, R., & Jacobs, B. A. (2004). Gender, street life and criminal retaliation. Criminology, 42(4), 911-940. Murphy, M., Ellsberg, M., & Contreras-Urbina, M. (2020). Nowhere to go: disclosure and help- seeking behaviors for survivors of violence against women and girls in South Sudan. Conflict and health, 14(1), 6. Murphy, M., Bingenheimer, J. B., Ovince, J., Ellsberg, M., & Contreras-Urbina, M. (2019). The effects of conflict and displacement on violence against adolescent girls in South Sudan: the case of adolescent girls in the Protection of Civilian sites in Juba. Sexual and reproductive health matters, 27(1), 181-191. Nacos, B. L. (2010). Revisiting the contagion hypothesis: Terrorism, news coverage, and copycat attacks. Perspectives on Terrorism, 3(3). OECD/African Development Bank, United Nations Economic Commission for Africa. (2009). African Economic Outlook 2009. OECD Publishing, 2009. p 360 Østby, G. (2016). Violence begets violence: Armed conflict and domestic sexual violence in sub- Saharan Africa. In SVAC (Sexual Violence and Armed Conflict) Workshop, Harvard University, Cambridge (pp. 2-3). Patterson, G. R. (2008). A comparison of models for interstate wars and for individual violence. Perspectives on psychological science, 3(3), 203-223. Perez-Heydrich, C et al,. ( 2013). Guidelines on the Use of DHS GPS Data. Spatial Analysis Reports No. 8. Calverton, Maryland, USA: ICF International. Profamilia (2011). Encuesta Nacional de Demografía y Salud 2010. Asociación Probienestar de la Familia Colombiana (Profamilia). Bogota, Colombia. Raleigh, C. et al., (2010). Introducing ACLED: An Armed Conflict Location and Event Dataset. Journal of Peace Research 47, no. 5: 651-60. Radelet, S. (2007). Reviving Economic Growth in Liberia. Center for Global Development. Working Paper No. 133. Retrieved at https://www.cgdev.org/sites/default/files/14912_file_Liberia_Growth.pdf Rohwerder, B. (2016). Women and girls in forced and protracted displacement (Governance and Social Development Resource Centre (GSDRC) Helpdesk Research Report 1364). Birmingham, UK: University of Birmingham. 37 Saile, R., Neuner, F., Ertl, V., & Catani, C. (2013). Prevalence and predictors of partner violence against women in the aftermath of war: A survey among couples in northern Uganda. Social Science & Medicine, 86:17-25. Sedgwick, M. (2007). Inspiration and the origins of global waves of terrorism. Studies in Conflict & Terrorism, 30(2), 97-112. Shilue, J. S., & Fagen, P. (2014). Liberia: Links between peacebuilding, conflict prevention and durable solutions to displacement. Washington, DC: Brookings Institution, 20. Stith, S. M., Rosen, K. H., Middleton, K. A., Busch, A. L., Lundeberg, K., & Carlton, R. P. (2000). The intergenerational transmission of spouse abuse: A meta‐analysis. Journal of Marriage and Family, 62(3), 640-654. Straus, M. A., Gelles, R. J., & Smith, C. (1990). Physical violence in American families: Risk factors and adaptations to violence in 8,145 families (pp. 49-73). New Brunswick, NJ: Transaction Publishers Swaine, A. (2015). Beyond strategic rape and between the public and private: Violence against women in armed conflict. Hum. Rts. Q., 37, 755. Swaine, A., Spearing, M., Murphy, M., & Contreras-Urbina, M. (2019). Exploring the intersection of violence against women and girls with post-conflict statebuilding and peacebuilding processes: A new analytical framework. Journal of Peacebuilding & Development, 14(1), 3-21. UNDP. (n.d.) Human Development Indicators Dashboard: Colombia. Accessible at http://hdr.undp.org/en/countries/profiles/COL UNDP. (2019). Human Development Report 2019. Beyond income, beyond averages, beyond today: Inequalities in human development in the 21st century. New York. http://hdr.undp.org/en/content/human-development-report-2019 UNHCR (2014). Woman Alone: The fight for survival by Syria's refugee women. Available at: https://www.refworld.org/docid/53be84aa4.html UNHCR (2020a). Global Trends: Forced Displacement 2019. Copenhagen: UNHCR Retrieved from https://www.unhcr.org/globaltrends2019/ UNHCR (2020b) Mid-year Trends: Global Trends for Forced Displacement 2020. Copenhagen: UNHCR. Retrieved from https://www.unhcr.org/en-us/statistics/unhcrstats/5fc504d44/mid-year- trends-2020.html United Nations (2019) Conflict related sexual violence: report of the United Nations Secretary- General. (Report No. S/2019/280). New York: United Nations United Nations. (2013). Security Council extends UN peace mission in Liberia for another year. Retrieved at http://www.un.org/apps/news/story.asp?NewsID=45888#.V-0vgvkrLIU. 38 Vinck, P., & Pham, P. N. (2013). Association of exposure to intimate-partner physical violence and potentially traumatic war-related events with mental health in Liberia. Social Science & Medicine, 77, 41–49. Wachter, K. et al. (2018). Drivers of Intimate Partner Violence Against Women in Three Refugee Camps. Violence against Women, 24(3), 286-306. Wako, Etobssie, Elliott, Leah, De Jesus, Stacy, Zotti, Marianne E, Swahn, Monica H, & Beltrami, John. (2015). Conflict, Displacement, and IPV. Violence against Women, 21(9), 1087-1101. Wirtz, AL, Pham, Kiemanh, Glass N, et al. (2014). Gender-based violence in conflict and displacement: Qualitative findings from displaced women in Colombia. Conflict and Health, 8(1), 10. World Bank (n.d.) Country Data Portal. World Development Indicators. The World Bank Group. Accessible at https://data.worldbank.org/ 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. Geneva, Switzerland: WHO World Health Organization. (2016). Ethical and safety recommendations for intervention research on violence against women. Building on lessons from the WHO publication Putting women first: ethical and safety recommendations for research on domestic violence against women. Geneva, Switzerland: WHO. February 2016 World Health Organization. (2021). Violence against women prevalence estimates, 2018. Global, regional and national prevalence estimates for intimate partner violence against women and global and regional prevalence estimates for non-partner sexual violence against women. Geneva, Switzerland: World Health Organization, on behalf of the United Nations Inter-Agency Working Group on Violence Against Women Estimation and Data (UNICEF, UNFPA, UNODC, UNSD, UNWomen). Wodon, Q. (2012). Poverty and the Policy Response to the Economic Crisis in Liberia. World Bank Study; Washington, DC: World Bank. © World Bank. https://openknowledge.worldbank.org/handle/10986/13080 License: CC BY 3.0 IGO.” 39 Appendix Figure 2. Analytic Sample from Colombia 2010 DHS Figure 1. Analytic Sample from Colombia 2010 DHS 40