Rethinking Power: rapport de référence de l'évaluation d'impact des méthodologies combinées SASA! et Power to Girls dans le Sud-Est d'Haïti

Rethinking Power Program Evaluation in Southeast Haiti Baseline Report of the Impact Evaluation of the Combined Methodologies of SASA! and Power to Girls February 2021 Manuel Contreras-Urbina, Junior Ovince, Angela Bourassa, Elizabeth Rojas CONTENTS LIST OF TABLES AND FIGURES .......................................................................................3 ACKNOWLEDGEMENTS .................................................................................................5 ACRONYMS ....................................................................................................................5 GLOSSARY OF TERMS ....................................................................................................6 EXECUTIVE SUMMARY...................................................................................................7 BACKGROUNDS AND METHODS ...................................................................................7 RESULTS .......................................................................................................................................7 INTIMATE PARTNER VIOLENCE PREVALENCE: ...................................................................... 7 INTIMATE PARTNER VIOLENCE ASSOCIATED FACTORS AND TRIGGERS: ....................... 7 NON-PARTNER VIOLENCE PREVALENCE: .............................................................................. 8 REPORTING AND HELP SEEKING: ........................................................................................... 8 IMPLICATIONS FOR ACTIONS .................................................................................................. 8 BACKGROUND ...............................................................................................................9 STUDY DESIGN AND METHODS.................................................................................. 10 STUDY AIMS.............................................................................................................................. 10 STUDY AREA............................................................................................................................. 10 STUDY RESULTS ........................................................................................................... 13 INDIVIDUAL AND HOUSEHOLD CHARACTERISTICS OF RESPONDENTS ................... 13 GENDER DYNAMICS ................................................................................................... 13 ATTITUDES .................................................................................................................. 13 HOUSEHOLD GENDER ROLES AND DYNAMICS .............................................................. 16 INTIMATE PARTNER VIOLENCE................................................................................... 18 THE MAGNITUDE..................................................................................................................... 18 RISK FACTORS ASSOCIATED TO INTIMATE PARTNER VIOLENCE ................................. 20 SEXUAL VIOLENCE BY NON-PARTNERS...................................................................... 23 THE MAGNITUDE AND CHARACTERISTICS OF DIFFERENT TYPES OF SEXUAL VIOLENCE COMMITTED BY NON-PARTNERS............................................... 23 AFTER THE VIOLENCE.................................................................................................. 25 CONSEQUENCES OF SUFFERING VIOLENCE ................................................................... 25 SURVIVORS’ RESPONSE......................................................................................................... 25 CONCLUSIONS ............................................................................................................ 29 REFERENCES ............................................................................................................... 30 ANNEX 1: QUANTITATIVE AND QUALITIVE BASELINE SAMPLING DATA ................... 31 ANNEX 2: BASELINE DATA TABLES AND FIGURES...................................................... 32 GENERAL MALE AND FEMALE POPULATION DESCRIPTION ......................................... 32 IPV PREVALENCE..................................................................................................................... 35 RISK FACTORS.......................................................................................................................... 36 NON-PARTNER SEXUAL VIOLENCE ..................................................................................... 41 RESPONSES TO VIOLENCE BY SURVIVORS........................................................................ 43 ANNEX 3. MULTIVARIATE MODEL RESULTS................................................................ 45 ANNEX 4: RESEARCH DESIGN AND METHODS .......................................................... 54 Rethinking Power Program Evaluation in Southeast Haiti | 2 LIST OF TABLES AND FIGURES Table 1.1: Number of respondents surveyed .......................................................................................................................... 31 Table 1.2: Number of FGD and KII............................................................................................................................................. 31 Table 2.1: Percentage of women and men who agree with the different statements related to gender roles, norms, and dynamics ..................................................................................................................... 32 Table 2.2: Repartition of the chores for the households of ever partnered women........................................................... 33 Table 2.3: Final decision-making process for the households of ever partnered women ................................................. 33 Table 2.4: Couple discussion on sexual health and HIV test.................................................................................................. 33 Table 2.5: Couple sexual behavior protection ......................................................................................................................... 34 Table 2.6: Percentage of controlling behavior per commune ............................................................................................... 34 Table 3.1: Lifetime and 12-month IPV prevalence rate among ever-partnered women in Marigot and La Vallée ......... 35 Table 3.2: Frequency of current IPV among survivors per community ................................................................................. 35 Table 4.1: Significant individual risk factors for lifetime and 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée ............................................................................................. 36 Table 4.2: Partner’s characteristics as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée ..................................... 37 Table 4.3: Couple communication as significant risk factors for lifetime and 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée.................................. 37 Table 4.4: Confidence in relationship as significant risk factors for lifetime and past 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée.................................. 38 Table 4.5: Love and kindness as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée.................................. 38 Table 4.6: Controlling behaviors as significant risk factors for lifetime and 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée.................................. 39 Table 4.7: Intergenerational violence as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée.................................. 40 Table 4.8: Justification of violence as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée.................................. 40 Table 4.9: Triggers for IPV among ever-partnered women who have experienced IPV in Marigot and La Vallée ................................................................................................................................................................................. 41 Table 5.1: Lifetime and 12-month non-partner prevalence rate among all women in Marigot and La Vallée ................ 41 Table 5.2: Age at first non-partner sexual assault among all women in Marigot and La Vallée ........................................ 42 Table 5.3: Identification of perpetrators of non-partner sexual assault among all women in Marigot and La Vallée .... 42 Table 6.1: Disclosure of intimate partner violence, by group, among women who have experienced IPV in Marigot and La Vallée........................................................................................................................................................... 43 Table 6.2: Disclosure of forced intercourse, by group, among women who have experienced rape in Marigot and La Vallée ............................................................................................................................................................... 43 Table 6.3: Seeking help for intimate partner violence, by resource group, among women who have experienced IPV in Marigot and La Vallée.............................................................................................................................. 43 Table 6.4: Discussions regarding intimate partner violence, by initiator, among women who have experienced IPV in Marigot and La Vallée ............................................................................................................................. 43 Table 6.5: Individual support response received for forced intercourse, among women who have experienced NPV in Marigot and La Vallée ........................................................................................................................... 44 Table 6.6: Seeking help for forced intercourse, by resource group, among women who have experienced forced intercourse in Marigot and La Vallée................................................................................................... 44 Table 6.7: Awareness of a place to go in case of sexual violence among all women in Marigot and La Vallée ............. 44 Table 7.1: Logistic regression crude and adjusted odds ratios of the associations between lifetime IPV physical and/or sexual violence ever and explanatory variables – La Vallée.. ................................................................... 45 Table 7.2: Logistic regression crude and adjusted odds ratios of the associations between current IPV physical and/or sexual violence and explanatory variables – La Vallée.............................................................................. 47 Rethinking Power Program Evaluation in Southeast Haiti | 3 Table 7.3: Logistic regression crude and adjusted odds ratios of the associations between lifetime IPV physical and/or sexual violence and explanatory variables – Marigot................................................................................ 49 Table 7.4: Logistic regression crude and adjusted odds ratios of the associations between current IPV physical and/or sexual violence and explanatory variables – Marigot................................................................................ 52 Figure 1: Map of La Vallée and Marigot where the survey took place in 2017 ................................................................... 10 Figure 2: Percentage of women and men who agree with statements reflecting traditional gender roles in Marigot and La Vallée........................................................................................................................................................... 14 Figure 3: Percentage of men and women who agreed that certain scenarios justify violence by a husband against his wife in Marigot and La Vallée ............................................................................................................................... 15 Figure 4: Distribution of household labor, by gender in Marigot and La Vallée ................................................................. 16 Figure 5: Percentage of household decision-making authority in Marigot and La Vallée ................................................. 16 Figure 6: Past 12 months and lifetime prevalence of different types of IPV in Marigot and La Vallée ............................. 18 Figure 7: Frequency of physical and sexual IPV experienced by women who report IPV in Marigot and La Vallée, in the past 12 months ............................................................................................................................................................... 19 Figure 8: Percentage of women experiencing physical and/or sexual IPV (lifetime and last 12 months) by the number of controlling behaviors exhibited by their partners in Marigot and La Vallée ...................................... 21 Figure 9: Prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée, by variables related to the relationship dynamic .................................................................................................................. 22 Figure 10: Lifetime and past 12 months prevalence of non-partner sexual violence, in Marigot and La Vallée ............ 23 Figure 11: Age at first non-partner sexual assault among women reporting NPSV in Marigot and La Vallée ................ 24 Figure 12: Identification of perpetrators of non-partner sexual assault among all women in Marigot and La Vallée ... 25 Figure 13: Disclosure of intimate partner violence, by group, among women who have experienced IPV in Marigot and La Vallée ............................................................................................................................. 26 Figure 14: Seeking help for intimate partner violence, by resource group, among women who have experienced IPV in Marigot and La Vallée ........................................................................................................... 27 Figure 15: Disclosure of forced intercourse by women in Marigot and La Vallée .............................................................. 28 Figure 16: Seeking help for forced intercourse, by resource group, among women who have experienced rape in Marigot and La Vallée ................................................................................................................. 28 Rethinking Power Program Evaluation in Southeast Haiti | 4 ACKNOWLEDGEMENTS ACRONYMS This baseline report is part of the overall impact evaluation of AOR Adjusted Odd Ratio the Rethinking Power program which is being implemented BB Beyond Borders by Beyond Borders in Southeast Haiti. The draft report was EMMUS Mortality, Morbidity, and Service carried out by the Global Women’s Institute (GWI) of the Utilization Survey George Washington University, in partnership with Beyond FGD Focus Group Discussion Borders, Institut de Formation et de Service (IFOS), the Technical GBV Gender-Based Violence Advisory Group (TAG) members in Haiti, and the Inter-American GoH Government of Haiti Development Bank (IADB). GWI Global Women’s Institute HIV/STI Human Immunodeficiency Virus Manuel Contreras-Urbina is the principal investigator of the / Sexually Transmitted Infection overall impact evaluation, providing leadership and oversight for IADB Inter-American Development Bank the study design and sampling, training interviewers, collecting IFOS Institut de Formation et de Service data, conducting analysis, and drafting the baseline report. IPV Intimate Partner Violence KII Key Informant Interview IFOS is the national technical team in Haiti and has coordinated LAC Latin American and Caribbean the overall quantitative and qualitative baseline data collection NGO Non-Governmental Organization in the Southeast department of the country. A special thank you NPSV Non-Partner Sexual Violence for the field team hired by IFOS. Beyond Borders has also played NPV Non-Partner Violence a major role in the coordination of the data collection and the OR Odds Ratio training. TAG has reviewed and provided insightful comments to RCT Randomized Controlled Trial improve the overall study. RP Rethinking Power SRH Sexual Reproductive Health The research team for this baseline report is composed of TAG Technical Advisory Group Manuel Contreras-Urbina, Junior Ovince, Gladys Mayard, Pierre VAWG Violence Against Women and Girls Philippe Wilson Registe, Angela Bourassa, Maureen Murphy, WHO World Health Organization Elizabeth Rojas, Mary Ellsberg, and Ulrick Jean Claude. They have contributed at different levels for the overall impact evaluation design, implementation, and report. The research team benefited greatly from the thoughtful comments and revisions of: Laurence Telson from the IADB. Sara Siebert, Emanuela Paul, Jean Prosper Elie, and Coleen Hedglin from Beyond Borders. Myriam Narcisse, Lenz Jean Francois, Sylvain Evelyne, Lise-Marie Dejean, Gladys Mayard, and Ulrick Jean Claude from TAG. This research has financially benefited by the NoVo Foundation and the IADB. Rethinking Power Program Evaluation in Southeast Haiti | 5 GLOSSARY OF TERMS Physical violence: Any act of force intended to or which has the potential to harm another person. In this study, these acts included but were not limited to slapping, punching, or hitting Controlling behaviors: Acts by a partner designed to control a with a fist or object; throwing objects that could cause harm; woman’s interaction with others. pushing, shoving, hair pulling, kicking, dragging; threatening with Forced sex: Any sexual act which has been engaged or or using a weapon. attempted by use of physical force, coercion, or intimidation. Prevalence: The proportion of at-risk women in a community Economic violence: Acts that are committed in order to deny who have experienced the dimension of violence being or limit access to personal or shared monetary resources. In this discussed. In this study, all female respondents aged 15-64 are study, acts included are prohibiting women from money-making at risk for non-partner sexual violence, while only the subset activities, taking her earnings against her will, and withholding of these women who have ever been partnered are at risk for available money for household expenses. intimate partner violence. Emotional violence: Acts that are committed in order to control Sexual Harassment: Any act of unwanted sexual advance, another person by damage to self-esteem and self-efficacy, or request for sexual favors, or exposure to media of a sexual nature fear. In this study, acts included are belittlement, humiliation in without consent. In this study, acts investigated included being front of others, and threats against the individual or someone groped or otherwise touched sexually without consent in a public they care about in order to gain control through fear or space (e.g., public transport); being asked to perform sexual acts intimidation. in exchange for employment or school advantages; being sent Ever-partnered: Any woman aged 15-64 who has ever had material of a sexual nature via physical or digital media without a male partner. Partnership includes marriage, unmarried consent and that is hurtful or upsetting to the recipient. cohabitation or relationships, and dating relationships. Sexual Violence: Any sexual act, completed or attempted Intimate partner violence (IPV): Any act of violence in the four without the consent of the individual. In this study, intercourse dimensions of gender-based violence included in this study or unwanted sexual touching that is engaged through physical committed against a woman or girl by her male partner. force, threat, coercion, or intimidation, regardless of the nature of the relationship of the perpetrator to the victim. Non-Partner sexual violence: Any sexual act engaged or attempted without consent. In this study, acts included unwanted Violence Against Women and Girls (VAWG): Any act of gender- intercourse completed or attempted by physical force, coercion based violence that results in or is likely to result in physical, or intimidation, or when unable to consent due to alcohol or sexual, or mental harm or suffering to women or girls, including drug intoxication or unwanted sexual touching or being forced, threats of such acts, coercion, or arbitrary deprivation of liberty, coerced, or intimidated into touching someone else sexually whether occurring in public or in private life. by someone other than an intimate partner. (Sexual violence by intimate partners is explored as one dimension of IPV). Perpetrator: One who commits an act of violence against another person. Rethinking Power Program Evaluation in Southeast Haiti | 6 EXECUTIVE SUMMARY Qualitative data was obtained through focus group discussions (6-12 participants) and key informant interviews conducted between July and August 2016 in each community. Translated BACKGROUND AND METHODS transcripts of these sessions were coded to identify overarching Violence against women and girls (VAWG) is highly prevalent all themes related to the research questions and to develop the over the world, including in the Latin American and Caribbean context in which the quantitative findings were experienced. region. It has been estimated that 36% of the region’s female Additional data about adolescent girls was collected through population has experienced violence from either a male partner a school-based survey and a survey with girls who attended (husband, common law union, dating, etc.) or from a non-partner the girl’s club (one of the activities of the Power to Girls (stranger, family member, friend, employer, etc.) during her methodology). The results of these surveys will be presented in a lifetime (WHO, 2012). In Haiti, previous household surveys that complementary report. included questions about intimate partner violence (IPV) and non-partner sexual violence (NPSV) demonstrated respective Results rates of 29% (among ever-partnered women) and 25% (prior to This report highlights the main findings at baseline for women the age of 18, among girls). (age 15-64) within the four dimensions of violence in both communities: prevalence of acts and types of violence as well as Previous programming to address the health repercussions of personal, interpersonal, and social factors that were associated VAWG in Haiti have focused on services to mitigate adverse with increased risk of experiencing intimate partner violence and health outcomes associated with the experience of IPV and NPSV non-partner sexual violence. among survivors, including access to health and legal services. However, there have been no comprehensive methodologies Intimate Partner Violence Prevalence: available to organizations endeavoring to address the root cause • 49% of ever-partnered women experienced some form of IPV of VAWG: patriarchal values that promote gender inequality at by a male partner in their lifetime; 37% reported experiencing every level of society. In 2010, Beyond Borders (a U.S.-based non- some form of IPV from a male partner in the 12 months prior to governmental organization) adapted the evidence-based SASA! completing the survey. methodology for community level social change addressing • Emotional violence was most commonly experienced in both gender power imbalances that drive VAWG. In addition, Beyond communities and time periods: last 12-month prevalence was Borders has developed a complementary methodology Power lower than lifetime in both communities (25% vs. 32% in La to Girls, which combines girl-centered programming and a Vallée; 27% vs. 35% in Marigot). comprehensive community mobilization process that highlights the roles of schools and caregivers. • Around one in three women has experienced lifetime physical and/or sexual IPV in both communities (30% La Vallée, 31% The two programs have been brought together as a single, dual- Marigot) and nearly one in four (23%) has experienced this model program Rethinking Power, in communities unfamiliar violence in the past 12 months. These findings are congruent with either stand-alone model. In 2016, evaluation and pilot with the WHO estimates for global prevalence rates. began of this dual model program using a quasi-experimental • The vast majority of women who experienced physical and/ design in two communal sections in Southeast Haiti. The quasi- or sexual IPV experienced it more than once (85% in La Vallée, experimental design was chosen to evaluate the effectiveness of 89% in Marigot). Beyond Borders’ Rethinking Power program in the intervention • Only economic violence prevalence was significantly different community (La Vallée de Jacmel - henceforth referred to as La between the communities: women in La Vallée reported nearly Vallée) when compared to the control community (Marigot) twice the rate of economic violence as women in Marigot, both in four areas: i) decrease social drivers of VAWG; ii) decrease for lifetime (27% vs. 15%) and past 12 months (17% vs. 10%). prevalence of VAWG; iii) decrease HIV/STI risk behaviors; and iv) increase girls’ sense of safety and agency. Intimate Partner Violence Associated Factors and Triggers: • Most of the women’s demographic characteristics can be Quantitative data was collected through 1,977 household surveys considered a risk or protective factor in IPV incidence within among a representative sample (see study design) of women the bivariate and multivariate analysis. Younger women, those aged 15-64 that were administered and completed in La Vallée who currently live with someone, and those who have a lower- (n=819) and Marigot (n=1158). In addition, 832 household level of education are more at risk of IPV according to the surveys with identical questions were administered to men aged multivariate model results. 15-64 in the two study areas (La Vallée n=306, Marigot n=526) in order to capture demographic data as well as men’s attitudes, • Partner characteristics, such as having extra-marital beliefs, and norms and compare them to the women’s answers. relationships and fights with other men increased risk of IPV, Men’s surveys did not include questions regarding violence respectively to 1.7 and 3.7 times. perpetration. Rethinking Power Program Evaluation in Southeast Haiti | 7 • The bivariate analysis also demonstrated that women who Implications for Actions reported love and kindness in their partnerships experienced VAWG and the current gender norms are concerning problems half the rate of lifetime physical and/or sexual IPV as those who that need to be addressed both in Marigot and La Vallée. The did not, and 25% less IPV during the past 12 months, while level of violence is high and women are not seeking help and/or controlling behavior was a consistent risk factor in all of the services. Moreover, gender norms support the acceptance of IPV models. and prevent women from disclosing cases of rape and IPV. The • Intergenerational violence and child abuse tended to double Rethinking Power program will play a major role in changing the the risk factors of IPV and were significant in all of the models. existing social norms at the community level. This intervention will work to increase helpful support among informal and formal Non-partner Violence Prevalence: service providers, given that there are major gaps in resources to • Women in Marigot and La Vallée experienced non-partner assist survivors of violence. sexual violence at high rates. Twenty-two percent of women in Marigot and 25% of women in La Vallée reported having experienced non-partner sexual assault in their lifetimes. More than half of these women reported that it had happened in the last 12 months. • Three out of five women in Marigot reported that sexual harassment or forced or attempted intercourse happened for the first time before the age of 20. This ratio is only one out of two women in La Vallée. Three out of five (56%) women both in Marigot and La Vallée experienced unwanted sexual touching for the first time before the age of 20. Reporting and Help-Seeking: • More than half of women who suffered from NPSV knew their attacker. • About half of women who suffered from NPSV never disclosed it. Likewise, three-quarters of women did not seek help from any formal service after experiencing non-partner sexual assault. • Among women who experienced IPV, the rate of non- disclosure was higher: 49% of women in La Vallée and 64% of women in Marigot remained silent. • Qualitative participants name ‘loss of status’ and ‘lack of resources’ as underlying causes that promote a woman’s silence. Rethinking Power Program Evaluation in Southeast Haiti | 8 BACKGROUND in intimate partner relationships and in broader community dynamics. While implementing SASA!, Beyond Borders received feedback from the community that more was needed to A limited but growing body of evidence suggests that violence specifically address violence against girls, and to engage youth against women and girls (VAWG) is highly prevalent in all parts as agents of change for interrupting the intergenerational cycle of the world, including the Latin American and Caribbean (LAC) of violence. For that reason, in 2013, Beyond Borders developed region. In 2013, the World Health Organization (WHO) estimated Power to Girls, a step-by-step methodology that combines that over one-third of women (36%) in this region have reported girl-centered programming with school- and community-wide experiencing either intimate partner violence (IPV) or non-partner social norm change. Power to Girls is designed to increase girls’ sexual violence (NPSV) at some point in their lifetime. Similar to safety and freedom by combining multiple violence-prevention many countries in the region, Haiti is facing high levels of VAWG. strategies.1 According to the 2012 Haiti Mortality, Morbidity, and Service Utilization Survey (EMMUS), 29% of ever-married women have In 2016, with the support of the NoVo Foundation, experienced violence perpetrated by their most recent husband Beyond Borders engaged a new cohort of communities or partner (GoH, IHE, and ICF, 2013). This rate was highest among in southeast Haiti, using the adapted and piloted SASA! women ages 15−19 (42%). Girls also experience high rates of methodology in tandem with Power to Girls. Through this non-partner sexual violence. One population-based survey that dual model, the Rethinking Power (RP) program expects to examined the prevalence of violence against children (aged reduce social acceptance of gender inequality and VAWG; 13-24), the Violence Against Children Survey in Haiti, conducted decrease experiences and perpetration of VAWG; increase by the Centers for Disease Control and Prevention and partners girls’ sense of safety and freedom to make decisions; and found that 1 out of 4 girls and 1 out of 5 boys had experienced at decrease HIV/SRH risk behaviors. To measure the effectiveness least one incidence of sexual abuse prior to 18 years of age. of the program, Beyond Borders and the Global Women’s VAWG is one of the clearest manifestations of patriarchal cultural Institute (GWI) at George Washington University—with the values, norms, and traditions that support gender inequality support of the NoVo Foundation and the Inter-American and encourage men to believe that they have the right to Development Bank— are carrying out a quasi-experimental control women. Social norms promoting power imbalance and impact evaluation of these two methodologies in two communal condoning VAWG are widespread in Haiti, as in most countries sections in the southeast of Haiti: La Vallée (intervention area) with a predominant patriarchal structure. A 2012 survey among and Marigot (comparison area).2 youth in Haiti revealed that nearly half of all girls and 2 out of 5 boys aged 13-17 years old believed that a man was justified in This report presents the main findings of the evaluation baseline beating a woman for one or more reasons (Government of Haiti data collected in 2016 for both the intervention and the et al, 2012). comparison areas. The robust quantitative and qualitative data obtained from this baseline provides an opportunity to develop This information demonstrates a significant need for an in-depth diagnosis of the situation of different types of VAWG programming to reduce this type of violence with specific in this part of Haiti. This attempts to be the most comprehensive attention to girls and youth. Changing patriarchal norms and diagnosis done on VAWG in this area. We expect that the curbing the cycle of VAWG in Haiti is an important step toward information will not only be useful to improve the RP program ensuring healthier, more productive, and safer communities for that is currently in place, but also to contribute to the evidence- women in Haiti. Past interventions to address VAWG in Haiti have base for future programs and policies in Haiti. focused mainly on response to survivors without incorporating a prevention focus, and those prevention efforts aimed at social Additional data about adolescent girls was collected through norms change in Haiti have largely been short-term, isolated, and a school-based survey and a survey with girls who attended disjointed, lacking long-term strategy and coordination. the girl’s club (one of the activities of the Power to Girls methodology). The results of these surveys will be presented in a In response to this gap, Beyond Borders, a United States non- complementary report. governmental organization (NGO) working in Haiti adapted SASA! in 2010. SASA! is a phased community mobilization approach to prevent VAWG and HIV, originally created by Raising Voices. A randomized control trial (RCT) published in 2014 demonstrated the methodology’s effectiveness in preventing both intimate partner violence and risk behaviors related to HIV transmission (Abramsky et al, 2014). SASA! aims to prevent violence against women by addressing the balance of power 1 For more information about SASA! and Power to Girls, see repansepouvwa.org/ and raisingvoices.org/sasa/ 2 For more information about the Impact Evaluation design, see Annex 4. Rethinking Power Program Evaluation in Southeast Haiti | 9 STUDY DESIGN Figure 1: Map of La Vallée and Marigot where the survey took place in 2017 AND METHODS STUDY AIMS The objective of this study is to evaluate the combined effectiveness of the SASA! and Power to Girls methodologies on preventing VAWG and sexual and reproductive health (SRH) risk behaviors, as implemented by Beyond Borders (BB) in Southern Haiti. This report presents the results of the baseline data for both Baseline data collection for La Vallee and Marigot monitored through the program intervention and comparison areas. The aim is to ONA (https://ona.io/). analyze in-depth the situation of different types of VAWG in this area of Haiti in order to inform policy and programs to prevent and respond to VAWG. To achieve this aim, the study explores STUDY METHODS the following research objectives: The baseline evaluation employs a mixed methods approach, utilizing quantitative and qualitative methodologies for the 1. Describe the survey participants, their household triangulation of findings by gathering data using multiple forms characteristics, the prevailing gender dynamics and societal of inquiry and by giving more depth and certainty to conclusions norms at the community level; made from the data. It includes different components that 2. Obtain a current estimate of the prevalence and characteristics capture data from women and adolescent girls, as well as men of physical, sexual, emotional, and economic violence by a and boys. This report focuses on the results of the women’s data current intimate partner as well as measure non-partner sexual collected through the following methods: violence; 3. Identify the main risk factors and consequences of VAWG Quantitative: along with the survivors’ coping strategies. Community-based Survey: A cross-sectional community population-based representative survey was conducted in STUDY AREA Marigot and La Vallée. The questionnaire was developed based on the tools used in the original SASA! Randomized Controlled The communities La Vallée and Marigot are both located in the Trial (RCT) study undertaken by the London School of Tropical Southeast department of Haiti; the team has an office in Jacmel, Hygiene and Medicine. The survey measures community which is located in between. La Vallée is the intervention site, members’ (both men and women aged 15-64) knowledge, which includes three main communal sections: Musac, Ternier, attitudes, and behaviors on key indicators related to VAWG. and Morn à Brûler. Only the first two communal sections were Through a two-staged systematic random sampling, the clusters selected to be part of the study for implementation constraints. (Section d’Enumération) and the households were selected. Only La Vallée is located 800 meters (2600 feet) above sea level and one eligible woman or man was surveyed in each household. has around 37,000 inhabitants for its 33 square miles. La Vallée Survey questions were drawn from indicators related to both has an important diaspora in the US and Canada. SASA! and Power to Girls. The quantitative data from the study was analyzed using descriptive statistics as well as bivariate and Marigot is the comparison site of the study and has six communal multivariate statistical methods. sections. Only two communal sections (Corail Soult and Savanne Dubois) were selected, and they have the same demographic Qualitative: characteristics as Musac and Ternier. Marigot has around 75,000 inhabitants. There is an important flow of Haitian migration to the • Focus Group Discussions (FGD): General community focus Dominican Republic even for a short period of stay. Around 90 groups (targeting men and women separately) were held to percent of the people, both in Marigot and La Vallée, live in rural validate and contextualize the findings of the community- areas. based survey. These groups utilized interactive and participatory methods, such as free-listing, open-ended stories, The communal sections selected both in Marigot and La Vallée role-playing, community mapping, etc. Tools can be found in have similar characteristics in terms of geographical reliefs, Annex 2. farming production, economic activities, religious practices and • Key Informant Interviews (KII): Key informant interviews were beliefs, number of households, and population. These conditions held with women, community activists, school administrators, are necessary to establish a good comparison between the two teachers, community leaders, and service providers. This sites. data helped contextualize and triangulate the findings from the community-based survey and the focus groups. Semi- Rethinking Power Program Evaluation in Southeast Haiti | 10  structured interview guides were developed to give a general Marigot) and in the girl’s clubs in La Vallée in order to measure framework for the interviews and include opening questions the impact of the Power to Girls methodology.3 that helped guide the conversation towards answering the research questions, but still allow flexibility in the conversation. Qualitative: For all qualitative data collection, purposeful sampling was used to ensure a wide breath of knowledge and experience. Additional respondents were also found through snowball sampling if required. The FGDs included specific groups of women, men, girls, and boys. The KIIs focused on stakeholders, including community activists, teachers, girl’s club leaders, community leaders and service providers. A total of 25 KIIs (13 in La Vallée and 12 in Marigot) and 22 FGDs were conducted (12 in La Vallée and 10 in Marigot, with an average of 9 participants). Participatory and Gender Approach The evaluation process applied a participatory approach, Interviews and FGDs from the qualitative study were captured involving both programmatic and research teams in the by note takers, translated and transcribed. The data analysis was design and implementation of the research, as well as in the carried out by the research team using the software Atlas.ti. The interpretation and dissemination of findings. GWI, BB, and research team used a combination of a priori and grounded IFOS collaborated closely on the development of the research theory to develop and assign codes to the data. The information protocol and data collection tools, to ensure the design is gathered through the FGDs and KIIs allowed the researchers to relevant in the local context and to strengthen local capacity. further document the situation of women and girls in Southeast Local stakeholders, community-based organizations, and local Haiti, as well as the impact of the RP program. authorities were involved in the planning and implementation stages through the creation of a Technical Advisory Group (TAG). SAMPLE SIZE AND RESPONSE RATES Quantitative: The following principles were also adhered throughout the evaluation process: 1) engage women and girls who reflect the Community-based Survey: The goal of the study was to assess diversity of the primary beneficiaries; 2) engage community the effectiveness of the RP program in preventing VAWG. In members and researchers in a joint process in which each will order to achieve this, the sampling strategy was based on the contribute equally; 3) facilitate a co-learning process; 4) include following primary outcome: reduction of IPV in a 12-month systems development and local capacity building; 5) facilitate period. Based on available national statistics and knowledge of an empowering process that validates participants’ experiences, the area, the research team expected that the true prevalence ideas, and opinions and through which they can increase control of physical IPV in the program area would be 20% in the past 12 over their lives; and 6) achieve a balance between research and months. Over the course of the program, it was anticipated that action.4 This approach ensures that the data and findings are this rate would be reduced to at least 14% in the intervention relevant and useful not only for the program evaluation, but also community, La Vallée, while remaining unchanged in the for local stakeholders and program managers.5 comparison community, Marigot. Therefore, the sample size was calculated to ensure each arm of the study had sufficient power to detect this change. Based on this, a total sample of at least 615 completed surveys per arm was determined. Thus, during the baseline data collection more surveys were completed with regard to the sampling design. A total of 819 women and 317 men were interviewed in the intervention site, La Vallée, whereas 1,158 women and 547 men in Marigot were administered the household survey. These numbers represent a 94% household response rate and 97% individual response rate. Data was also collected in six schools (three schools in each, La Vallée and 3 Data from the school and girl’s club survey were analyzed in another report focusing on adolescent boys and girls. 4 Ellsberg, M; Heise, L. (2005). Researching Violence against Women: A Practical Guide for Researchers and Activists. Washington, D.C.: PATH, WHO. 5 An example of how we have put these principles into action is found in the following article: Ellsberg, M. et al. (2009). Using Participatory Methods for Researching Violence Against Women: An experience from Melanesia and East Timor. The article is attached to this proposal. Rethinking Power Program Evaluation in Southeast Haiti | 11 The evaluation also applied a gender approach to achieve • Field workers should be trained to refer to local services the ultimate goal of the program: to transform the underlying and sources of support that are available to women and roots of gender inequality. Following Patton6, this evaluation girls who request assistance. If there are few resources, it has five components that characterize a broadly defined may be necessary for the study to create short-term support gender approach. These include: 1) a central focus on gender mechanisms. inequalities; 2) the conceptualization of inequality, based • Researchers and donors have an ethical obligation to ensure on gender as systemic and structural; 3) the recognition that that their findings are interpreted appropriately and used to information and knowledge are powerful resources; 4) the promote regulations and development of interventions. acknowledgement that the evaluator is not “neutral” but brings specific experiences, sensitivities, awareness, and perspectives; The research protocol of this study was fully approved by the and 5) the acknowledgement that evaluation is not merely a Institutional Review Board of the George Washington University technical activity but also a political one. as well as the Comité National de Bioétique Haïtien. In addition, permission to conduct the research was secured with appropriate ETHICAL CONSIDERATIONS authorities at national and local levels.8 During the baseline data collection, particular attention was applied to ethical considerations through the use of the WHO eight recommendations for conducting safe and ethical research on VAWG. Seven of these recommendations were directly related to this investigation and were incorporated:7 • The safety of the respondents and the research team is fundamental and should guide all decisions about the project. • Prevalence studies should be methodologically sound and should be based on current research experience on how to minimize underreporting of violence. • The protection of confidentiality is essential to ensure the safety of women and girls, as well as the quality of the data. • All members of the research team should be carefully selected and receive specialized training and ongoing support. • The study design should include actions that seek to reduce any possible anguish caused by the study among its participants. 6 Patton, M. (2002). Qualitative Research and Evaluation Methods. Thousand Oaks, California: SAGE Publications. 7 The other ethical guideline refers to surveys that are designed for other purposes, such as reproductive health, citizen security, etc. 8 For more information on the ethical considerations, contact the research team for the research protocol. Rethinking Power Program Evaluation in Southeast Haiti | 12 STUDY RESULTS GENDER DYNAMICS Attitudes INDIVIDUAL AND HOUSEHOLD CHARACTERISTICS OF RESPONDENTS The household survey findings revealed patriarchal values around gender among both male and female participants. Overall, the respondents within Marigot and La Vallée were In particular, taking care of household chores stood out as mostly composed of youth, with more than half of respondents a role associated with women, especially among female aged between 15 to 39 years old. The female population was participants who overwhelmingly agreed that women bear younger than the male one; three out of five (60%) women were primary responsibility for these duties (see Table 2.1 in annex 2). in this age group (57% in Marigot and 63% in La Vallée) versus Participants in the focus group discussions expressed the belief slightly more than half of the men (57% in Marigot and 53% in La that these conventional roles were a woman’s obligation based Vallée). The median age for women was 31 in Marigot and 33 in on biblical principles and societal norms, particularly when they La Vallée, and among men the median age was 31 in Marigot and were married or in cohabitation. 36 in La Vallée. More than half of the female (60% in Marigot and 51% in La Vallée) and male (55% in Marigot and 53% in La Vallée) “Because the Bible clearly said that population is educated up to the primary level. In general, only all the household chores are the two out of five respondents have a secondary level of education. women’s main responsibility.” Very few of the respondents have attended —Focus group discussion with women in La Vallée or achieved higher education. Men are more likely to be educated and to achieve higher education levels than women are. “It is even the main reason a man decides to marry a woman so that Regarding income generation, 3 out of 4 men are working, she can take care of all the chores. It compared to only half of women. More women in Marigot work than in La Vallée. This difference could be a reflection of is her main duty.” the economic activity in Marigot, which seems to have more —Focus group discussion with male local leaders in La Vallée farming and fishing activities than in La Vallée. Women are more likely than men to have been partnered (more than 80% in both communities); however, there is a difference for the predominant union status between communities. In La Vallée, formal marriage (including religious or civil marriage) is the main form of partnership reported, whereas in Marigot, cohabitation represents the principal form of union. A total of three out of four women in both communities (73%) have been pregnant at some point in their life and seven out of ten women (around 72% in both communities) reported at least two children living in their households. Rethinking Power Program Evaluation in Southeast Haiti | 13 Figure 2: Proportion of women and men who agree with traditional gender roles in Marigot and La Vallée However, while women tended to accept traditional roles around “I completely disagree [with household duties more than men, they also believed in a more balanced power dynamic within the household in comparison to inequality] because women can be men. More men than women agreed that, “A man should have someone in the society as men can, the final word about decisions in his home” (see Table 2.1 in and they [women] can undertake annex 2). The focus group participants highlighted that men have the full responsibility of their homes and families, and more rights much more things [than men].” to decide over other household members’ lives, including their —Focus group discussion with male youth in La Vallée partner. Men were perceived as the chief of the house. These traditional views were stronger in Marigot than in La Vallée. “Because women can do more. For instance, there are some “People always said that men are the households in which the woman is head of the household, so the final the breadwinner and the man is the decisions are theirs.” househusband.” —Focus group discussion with girls from the Girl’s Club in La —Focus group discussion with men in Marigot Vallée Negative attitudes around gender equality were also reflected In this region of Haiti, most participants agreed with the in the justification of VAWG in certain circumstances from many stereotypical gender roles that define men and women under of the participants. For example, around 1 out of 4 participants patriarchal societies. However, it is important to highlight that agreed that a woman should accept violence to keep her family there is also a plurality of participants who did not agree with together and nearly 1 out of 5 believed that if a woman were these gender unequal attitudes. This was also reflected in the raped she had done something careless to put herself in that focus group discussions, in which several members of the situation (see Table 2.1 in annex 2). In the survey, participants community shared more thoughts around gender equality. For were also asked about different circumstances in which IPV could example, some expressed their belief that women have the right be justified. Almost half of female participants justified IPV in at to become what they want and household duties should not be least one scenario, compared to only 20% of male participants. their first responsibility. The circumstances with the highest justification of violence were if “she goes out without telling the husband” and “if he suspects that she is having an affair with another man.” This reveals that the worst perceived transgression of social norms by a woman is around the possibility that she interacts with other men. The patriarchal control of women’s bodies is normalized to the extent that society condones physical punishment for women who are suspected to be with other men besides their partners. Rethinking Power Program Evaluation in Southeast Haiti | 14 Figure 3: Proportion of women and men who agree that violence against a wife by a husband is justified in certain scenarios in Marigot and La Vallée Qualitative findings support this survey data. For example, many However, there were also participants who agreed that women participants shared the idea that sexual abuse against women is a having other relationships was an egregious offense against their consequence of women wearing provocative clothing. They also male partners and should be condemned, but they disagreed confirmed the idea that women deserve to be beaten if they have that men were justified in beating women in this or any other extra-marital relationships, a particularly extreme offense against scenario and instead there were non-violent ways to deal with men. such situations. “Some women attract men to rape “Any husband is not supposed to them. It is not good for a woman beat his wife because she is not his to wear something that can expose kids and they have the same right.” their bodies.” —Focus group discussion with girls from the Girl’s Club in La Vallée —Focus group discussion with female students in Marigot “Because when a woman is “There is no situation that can justify unfaithful, she dishonors her physical IPV on women.” —Focus group discussion with female local leaders in Marigot wedding vows and disrespects her husband. So, she deserves to be beaten.” “Men are not authorized to beat —Focus group discussion with male local leaders in La Vallée women in any situation.” —Focus group discussion with male students in Marigot “Because beating is violence, and violence is not good. Woman is our mother, and women and men have the same rights.” —Focus group discussion with male youth in Marigot Rethinking Power Program Evaluation in Southeast Haiti | 15 HOUSEHOLD GENDER ROLES Figure 5: Proportions of household decision AND DYNAMICS making power reported by women in Marigot Gender attitudes of individuals were also reflected in behaviors and La Vallée and dynamics within couples. In both the quantitative and the qualitative data, male and female participants shared behaviors within their relationships that corresponded with traditional division of labor and decision-making within the household. For example, most ever-partnered women reported that women and girls bear primary responsibility for household chores, such as washing clothes (96% in Marigot and 88% in La Vallée), preparing food (96% in Marigot and 90% in La Vallée), and taking care of the children (91% in Marigot and 86% in La Vallée) (Figure 4). The inequality among men and women in Haitian couples is also reflected in sexual behavior. Both the female and male surveys found that more men than women were involved in extra-marital relationships, with around 30% reporting sexual relationships with other partners in the last 12 months compared to less than 3% of women. Of those participants who had extra-marital sexual relationships in the last 12 months, more men than women reported using condoms during their last sexual encounter, which speaks to women’s lack of capacity to negotiate the use of condoms with sexual partners compared to men. The qualitative findings showed that any resistance from women for not using a condom can be interpreted as disobedience. Figure 4: Proportion of women who report that women and girls are primarily responsible for “According to the Bible, men are the household labor in Marigot and La Vallée boss of women, so men’s decision should be the final one. Women need to obey their husbands in any situation and circumstances.” —Focus group discussion with men in Marigot “She is my wife and I am the man, so I need to decide if I want to use a condom or not. If she refuses to have sex with me without condom, she In terms of the final say on key decisions around the household must have another sexual partner.” economy and the leisure time, in many cases, the husband/ —Focus group discussion with men in La Vallée male partner is the one who makes the final decision. This clearly establishes how men are considered the heads in most households (Figure 5). In some cases, women even need to have “Because the men have the their husbands’ permission to go out. economic power so they decide “Only the husband can decide when when to use a condom.” —Focus group discussion with female youth in La Vallée his wife can go out.” —Focus group discussion with students in Marigot Rethinking Power Program Evaluation in Southeast Haiti | 16 In the household survey, most women also shared that they “If they …[women]… don’t respect were involved in relationships in which their male partners had controlling behaviors (see Table 2.6 in annex 2). Most of themselves, we need to control them these controlling behaviors were related to jealousy, control of and hit them.” women’s bodies, and control of women’s interactions with other —Focus group discussion with teachers (mixed group) in people, particularly other men. Qualitative findings confirm that Marigot many women experience these situations in this area of Haiti: In sum, findings based on gender roles, attitudes, and dynamics “Women don’t listen, and they will confirmed that these communities in Haiti have a patriarchal do things that can make the husband society structure that perpetuates gender inequalities, giving jealous or mad. It is necessary to more value and power to men and boys, who decide about women’s and girls’ lives. This patriarchal structure is reflected and control them and even hit them to spread through the actions of people and institutions and lead to make them obey.” the use of violence against women and girls. —Focus group discussion with men in La Vallée Rethinking Power Program Evaluation in Southeast Haiti | 17 INTIMATE PARTNER VIOLENCE In order to measure the magnitude of different types of IPV in these communities, women were asked about their experiences The Magnitude of specific acts of each type of violence by their partners, both IPV appears to be a frequent situation in both communities. In the lifetime experience and in the last 12 months prior to the survey. qualitative work, both female and male participants mentioned Through the responses, it was possible to obtain the prevalence that this type of violence was common in these communities of of each of these forms (Figure 6). In the community’s studies, Haiti. In both communities, all four forms of partner violence– 803 (49%) women who participated in the survey reported some physical, sexual, economic, and emotional–were identified. form of violence by male partners in their lifetime and 37% Informants also indicated that different forms of violence could reported violence in the 12 months prior to the survey. Similar be interconnected (for example, emotional violence leading to to most studies in the world (e.g., WHO Multi-Country surveys, physical violence) and there are many circumstances related to DHS surveys, among others), emotional violence was the form of controlling behaviors of men against women. violence with the highest prevalence. “Jealousy is one of the main reasons of physical IPV. If a woman is talking to another man, her partner can beat her just for that.” —Key informant interview with NGO representative in La Vallée Figure 6: Lifetime and current prevalence of different types of IPV in Marigot and La Vallée Rethinking Power Program Evaluation in Southeast Haiti | 18 Around 15% of ever-partnered women have suffered physical violence from their current partner and two-thirds of them (10% “She is not going to tell anyone of the total) experienced it in the 12 months prior to the survey. Sexual IPV was more prevalent than physical IPV among survey because she is afraid that her respondents: more than 1 in 5 women revealed sexual abuse husband might know, in which case, by their current partner and around 17% of the women had he will beat her more or even kill experienced sexual IPV in the past 12 months. A total of 3 in every 10 women had been sexually and/or physically harmed by her.” their current intimate partner during their lifetime, with at least 1 —Focus group discussion with teacher in Marigot in every 5 experiencing both physical and sexual violence in the past 12 months. “She is afraid of being teased by the Despite survey data showing higher levels of IPV than previous community if they find out.” surveys9, the levels are lower than expected, especially regarding —Focus group discussion with female local authority leaders physical violence. Based on previous studies, the levels of in La Vallée controlling behaviors from men against their female partners found in the survey and the qualitative results, we believe there is some underreporting of IPV in Haiti, including in this study. This Survey findings also showed that women who have suffered IPV may be due to a patriarchal culture in which IPV is normalized in these areas experienced this abuse frequently. This was a little and accepted but also hidden by shame, stigma, etc. Another higher in Marigot than La Vallée. In Marigot, 9 out of 10 women element to consider is the fear of women that their husbands or who reported violence in the past 12 months had this experience other acquaintances find out about their answers even though more than once (Figure 7). the enumerators ensured confidentiality and the interview was conducted in complete privacy.10 Figure 7: Frequency of physical and sexual IPV during the past 12 months among women who reported IPV, in Marigot and La Vallée Percentages may not total 100 due to rounding. 9 According to the EMMUS-VI results from 2016-2017, the national prevalence of the different types of IPV are: i) Emotional: 26.3 percent lifetime, 17.8 percent current, ii) Physical: 21.3 percent lifetime, 10.1 percent current, iii) Sexual: 14.0 percent lifetime, 7.2 percent current, iv) Physical and/or sexual: 26.0 percent lifetime, 13.9 percent current. The IPV prevalence from the current or most recent partner in the Southeast are estimated to be: 19.4 percent (emotional), 12.8 percent (physical), 5.7 percent (sexual), and 14.2 percent (physical and/or sexual). These estimates are for the overall department, however, baseline estimates for Marigot and La Vallée combined are calculated for women between the ages of 15 and 49 for a potential comparison with the Southeast estimates. Overall, the baseline estimates for the current or most recent are higher, with 35.8 percent (emotional), 15.6 percent (physical), 24.6 percent (sexual), and 32.6 percent (physical and/or sexual). 10 In this study, the focus is more on the IPV occurring by a current partner or the most recent one. The different prevalence rate does not include previous former partners as perpetrators. Rethinking Power Program Evaluation in Southeast Haiti | 19 Risk Factors Associated to Gender inequality as a root cause of IPV is reflected in Intimate Partner Violence relationship dynamics. Not surprisingly, those women who were in more patriarchal relationships were more at risk to suffer IPV Who are those women who are more at risk to suffer IPV? To than those who were in more equal relationships. For example, answer this question, a comprehensive analysis was carried out the analysis showed that half of women who reported 3 or more to find associations between different key variables and physical of the controlling behaviors exerted by their male partners had and/or sexual violence IPV for lifetime and present day. Findings also suffered physical and/or sexual IPV, compared to a quarter showed that gender inequality was the main root of IPV and was of women who reported only 1 or 2 controlling behaviors, and reflected within individual and relationship factors associated around 10% of those who did not report controlling behaviors. to IPV. At the individual level, women who justified violence in The controlling behaviors11 that are more associated with IPV certain circumstances had higher prevalence of physical and/or were related to women’s freedom to interact with other people. sexual IPV compared to those who did not. For example, almost 40% of women who thought that violence was justified at least by one circumstance had suffered physical and/or sexual IPV compared to around 21-25% who did not think that violence was justified in any circumstance. The normalization of violence was frequently mentioned during the qualitative work as women were seen as children who needed to be corrected if they failed to fulfill their home duty, show respect to their partner, or were suspected to be unfaithful. “If the woman violates the marriage contract, she is unfaithful, and she is being disrespectful to her husband. She deserved to be beaten.” —Focus group discussion with female local authority leaders in La Vallée “It is the man to decide when his partner can go out.” —Focus group discussion with male students in Marigot 11 Two-thirds (67%) of women in Marigot and four out of five (79%) women in La Vallée have suffered a least one act of controlling behavior in their lifetime from one of their intimate partners. See Table 4.6 in the annex for more information on the different acts of controlling behavior. Rethinking Power Program Evaluation in Southeast Haiti | 20 Figure 8: Proportion of women who have experienced physical and/or sexual IPV, ever and in the last 12 months, by partners’ controlling behaviors Lack of women’s agency, both social and economic, were also Even though social norm interventions were mentioned as one frequently mentioned as a problem associated with the use of of the strategies to prevent and reduce VAWG, participants also violence against women. For example, men take advantage agreed that economic support for women, such as interventions of the economic vulnerability of women and abuse them that address financial freedom and job opportunities were keys psychologically and physically. to the reduction of IPV. “Because of their lack of economic “It is important to create economic and financial resources, women opportunities for women in this tend to accept unfair treatment and community, so that they can any situation of violence from their experience economic freedom and partner.” so on.” —Key informant interview with NGO representative in La —Key informant interview with child protection sector Vallée representative in Marigot “[Women] don’t have any alternative, and they depend economically on their partner, and they are afraid to lose him.” —Key informant interview with civil society representative in La Vallée “When the man realized is the main provider economically, he can allow Other variables surrounding the dynamic of the relationship, himself to be violent to his partner.” such as communication, respect, affection, and trust were also important to understanding IPV. In all cases, the proportion —Key informant interview with education sector representative in Marigot of women who suffered IPV among those who said that they communicated with their partners, trusted them, and felt loved, valued and respected was lower than women who said the opposite. In addition, women who mentioned that their partners had extra-marital relationships presented higher prevalence of IPV compared to those who did not. Rethinking Power Program Evaluation in Southeast Haiti | 21 Figure 9: Proportion of women experiencing physical and/or sexual IPV, ever and in the past 12 months, by loving kindness in their relationship, in Marigot and La Vallée Finally, other variables at the individual level were also found years old or younger. In general, women who receive financial significantly associated to IPV. The most important ones were support from their partner or relatives are 1.6 times more at risk education level, age at marriage, partner involvement in physical than women who are the breadwinner. Finally, only women who fights against other men, and violence experienced during have a higher level of education have a reduced risk of IPV by childhood. Overall, women or their partners who had a lower at least twice, compared to women with a primary education educational level than high school, who got married under 18 level. However, only age, partnership status, and education were years old, whose partners were involved in physical fights against significant in the multivariate models. other men, and women or their partners who experienced or witnessed violence during childhood, had higher prevalence of For the partners’ characteristics, two main variables were IPV. These factors have been commonly found associated to IPV significant for the bivariate models. A woman with a partner that in many other studies, in different places around the world (see has another relationship will have an increased risk of IPV by Annex 3). 1.7 times. Moreover, the IPV risk for women with a partner who has been involved in fights with other men in the community Multivariate analysis of factors associated with intimate partner skyrockets to at least 3.7 times more. These variables were violence. significant for the multivariate models and have a lower rate of adjusted odd ratios. Multivariate logistic regression was utilized to further elucidate the strength of association between significant variables and The couple relationship and dynamics include several variables women’s experience of sexual and/or physical IPV over their that were significant in the bivariate models but not necessarily in lifetime and within the 12 months preceding the survey. As in the multivariate, with the exception of controlling behavior. The the bivariate analysis, the potential risks and protective factors quality of the relationship, the sense of safety and confidence, identified were grouped in five categories, such as the women’s and the level of communication were some key protective demographic characteristics, their partners’ characteristics, the factors that contributed to keeping IPV low. However, most of the couple relationship dynamics, the intergenerational violence, women experienced control by their partner, and their risk of IPV and the gender attitude towards justifying IPV, and non-pooling increased by 4 to 6 times within all models. bivariate and multivariate logistic models were produced for Marigot and La Vallée. Intergenerational violence and child abuse have been identified as crucial risk factors, and they are significant for the bivariate In terms of women’s demographic characteristics, variables and multivariate models. In general, the risk of IPV increases by such as age, age at first union, age at first pregnancy, education, at least 1.5 times for any woman who witnessed IPV in her home partnership status, and main source of income have shown a as a child. This risk increases by at least 2.6 times when it is her strong association with lifetime and current IPV for the bivariate partner who witnessed IPV. If a woman suffered child abuse from models. her parents or from others, her risk of IPV increases by at least 2 times. All of these variables are significant in the multivariate For instance, younger women are more at risk of IPV as IPV models. decreases 3% for every one additional year of age. Women in a current relationship are three times or more likely to be exposed to IPV. A woman can see her risk of IPV reduce by at least 25% if she gets pregnant at 20 years old or older compared to women who get pregnant at 19 years old or younger. Her risk can also increase by 1.5 times if she gets married or cohabitated at 19 Rethinking Power Program Evaluation in Southeast Haiti | 22 Finally, all the variables representing justification of IPV having an affair. This confirms how a patriarchal society, such in different situations were significant at the bivariate and as Haiti, condemns women who do not fulfill an expectation multivariate models, especially the justification that men should of a monogamous sexuality mainly focused on reproductive punish women when they are suspicious that their partners are purposes. SEXUAL VIOLENCE BY NON-PARTNERS of non-partner sexual violence: forced intercourse (rape), attempted forced intercourse (attempted rape), unwanted sexual The magnitude and characteristics of different touching, and sexual harassment. Women who indicated they types of sexual violence committed by non- had experienced one of these acts were asked whether this had partners happened to them in the 12 months preceding the survey. Based As with IPV, NPSV appeared to occur frequently in the on survey responses, prevalence of each type of violence was communities surveyed in Haiti. In the focus group discussions calculated for lifetime and current (last 12 months) experience. and key informant interviews, participants identified gender inequality and women’s lack of power as key drivers of NPSV. In the study communities, non-partner sexual assault was This was also the primary obstacle to reporting, even though reported at more than three times the global average. Lifetime the perpetrator was usually known to the victim. Often, the prevalence among survey respondents was 22% in Marigot perpetrator was older than the victim, had more stature in the and 24.7% in La Vallée, and half of women in both communities community, and had the ability to control the victim and other who have ever experienced non-partner sexual assault reported community members through force or social pressure. This is experiencing it in the last 12 months (12% Marigot, 14% La particularly true for restavèk girls12 living as unpaid domestic Vallée). Both communities had similar prevalence of each type of servants in households of higher status than their families of NPSV, with slightly higher rates in La Vallée for each. Non-partner origin. sexual harassment was nearly as high as half of all non-partner sexual assault at 12% in La Vallée vs. 11% in Marigot. Forced In order to measure the magnitude of NPSV in Marigot and La intercourse, attempted forced intercourse, and unwanted sexual Vallée, all women who participated in the survey were asked touching were less prevalent. about their experiences of specific acts within four dimensions Figure 10: Proportion of women who have experienced different types of NPSV, ever and in the past 12 months, in Marigot and La Vallée In Marigot, three out of 5 women (58.5%) who have ever Marigot mostly at a young age, compared to women in La Vallée experienced forced or attempted forced intercourse had done so (30.0%). However, the trend is the same in both communities in the first time before the age of 20, while this ratio was nearly one terms of unwanted sexual touching, as 56% of women reported out of two in La Vallée. The same trend holds in terms of sexual that their first experience happened before the age of 20. harassment. It happened for the first time to women (43.5%) in 12 Restavek girls or boys are children given away by their parents because of their economic incapacity to take care of the kids. These children will serve the host family as a domestic in the hope that they can have access to basic needs, such as food, home, and school. Rethinking Power Program Evaluation in Southeast Haiti | 23 Figure 11: Age at time of first sexual assault by persons other than partners among women who reported NPSV in Marigot and La Vallée Identification of the perpetrators In both communities, most women who have experienced non- “Most of the time, the perpetrators partner sexual assault indicated that the perpetrator was known are known and they are men in to them. More than half of the women knew their rapists or the community who have a sort of potential ones (60% in La Vallée, 64% in Marigot). Almost no one identified a community figure as their attacker, though the low influence over the community.” response rate may be due, in part, to fear of retribution by a more —Key informant interview with civil society representative in powerful man, regardless of safeguards to protect participant La Vallée anonymity. Qualitative data can provide context to better understand this. As mentioned above, participants discussed the ongoing practice of restavèk as a risk factor for young girls, as they live under complete control of the families they serve. Men in these households would be known to them yet not be family members. Further, men from the Haitian diaspora who returned to visit family were identified as perpetrators of coerced or transactional sexual contact due to their relative wealth and status. Again, these men would likely fall under the broad category of persons known to the participants but not family members or community figures. Lastly, the discussion group and interview participants discussed a practice of drugging women to assault them in remote areas. This may account for the more similar ratio of unknown to known perpetrators of forced intercourse. Rethinking Power Program Evaluation in Southeast Haiti | 24 Figure 12: Proportion of known and unknown perpetrators of different types of NPSV reported by women who have experienced NPSV in Marigot and La Vallée AFTER THE VIOLENCE Consequences of suffering violence “Her life will never be the same; she VAWG can have serious consequences on the health of will be thinking all the time about her women, as globally demonstrated in many studies on this situation.” topic. In this research, consequences of both IPV and NPSV —Focus group discussion with female local authority leaders were only captured through qualitative techniques. Participants in Marigot identified many health consequences that affect survivors of IPV. In particular, they highlighted the mental and emotional consequences of suffering violence, such as depression, trauma, Physical consequences were also identified. In particular, suicidal ideation, and psychological troubles. While community survivors frequently mentioned negative sexual and reproductive shame and dishonor can be among the emotional consequences health consequences. Participants discussed outcomes, such of NPSV. as being infected with sexual transmitted infections (STIs) and unwanted or early pregnancies. Other physical aspects were high blood pressure, diabetes, stomach cramps, tuberculosis, and “Women are victim more than once physical injuries to the body. …[when she is raped]…, she will Survivors’ responses feel worthless and she knows she As previously mentioned, women in Haiti face many challenges will lose her esteem and value in when trying to disclose violence. This was confirmed in both the community. She will want to die quantitative and qualitative findings of this study. According or even try to kill the perpetrator if to the household survey, 64% of women in Marigot and 49% of women in La Vallée who have suffered IPV have never told possible.” anyone about the violence experienced (Figure 13). Maintaining —Focus group discussion with male youth in La Vallée silence is common among women survivors of violence. In both La Vallée and Marigot, women who have disclosed the violence have done so mainly with a friend first, followed by a family Regarding IPV consequences, depression can be an immediate member. consequence for the survivor because she is constantly thinking about her situation, including the inability to leave the violent relationship because of her economic dependence on her partner. Most of these emotional consequences were identified as life-long consequences. Rethinking Power Program Evaluation in Southeast Haiti | 25 Figure 13: Proportion of women who have the family to break up. Participants said that in many cases if disclosed IPV to friends, family, or others in violence was reported to formal services, the relationship would end. This was usually because the abuse was quite severe and, in Marigot and La Vallée rare cases, the woman was ready to break up with the partner. “They don’t want to disclose because they don’t have any alternatives if they lose the husband. If they look for legal services, that means the relationship is over.” —Key informant interview with mayor in Marigot Qualitative findings confirmed what was found in the survey. During the interviews and focus groups, participants shared that “Even if a woman is looking for women who suffered violence generally did not disclose the help, she does not want to leave situation with anyone, and if they did disclose, they would go the relationship. Her ultimate goal to a friend or a family member, hoping that this person would maintain confidentiality. is to stop the violence and keep her family. Sometimes, she will not want us to help her by any means to stop “[Women] will report the IPV case the violence and will refuse to do any to the best man or the maid of legal or police follow-up.” honor who most of the time is their —Key informant interview with women’s organization best friend. They will try to fix the representative in Marigot situation, which can get better or worse.” Figure 14 demonstrates that there are only a few survivors who —Focus group discussion with teachers in Marigot seek help, and they mostly go to health centers or the police. This corroborates the assumption that most women that disclose IPV to formal actors do so because the severity of the violence has The occurrence of IPV is so normalized that society places reached dangerous levels that could cause severe injuries. While the responsibility on women to fix the situation and look for the qualitative data suggests that women’s organizations are reconciliation by any means. These survivors of violence tend considered to be the most helpful institutions providing support to not seek a formal legal response and expect partners to to women, the quantitative data demonstrates there are very eventually stop exerting control over them. They focus on few women who actually access them. Qualitative participants keeping the family together. Most of the time, they will keep agreed that there were few institutions that provided support and silent and, if they disclose violence, they will consult a community usually they lacked resources to provide a good response for the leader (local authorities), members of their church, or their survivor. parents in the hopes that they will intercede, so that the husbands can change their behavior. “We are also conscious that we don’t have all the means to support. We “Most of the time, [women] will keep don’t have a car right now available silence or try to talk to a local leader to go the field and proceed to any in order to bring the family together arrest. We don’t have a budget for and eventually stop the violence.” fuel. And when the IPV case happens —Key informant interview with police inspector in La Vallée in the rural areas, we won’t be able to intervene.” In many cases, disclosing violence could increase the risk for women to suffer more violence. In others, disclosure could cause —Key informant interview with police inspector in La Vallée Rethinking Power Program Evaluation in Southeast Haiti | 26 Figure 14: Proportion of women who sought help for intimate partner violence, by resource group, among women who have experienced IPV in Marigot and La Vallée The key elements preventing women from reporting intimate Similar to IPV, about half of women who suffer NPSV never partner violence in this part of Haiti were social stigma regarding disclose the situation to anyone (Figure 15). In case of disclosure, violence, social norms around marriage, fear of escalation of they do so mainly to their families. According to qualitative violence from the husbands, fear of losing husbands which could participants, the main reason for not disclosing non-partner result in a vulnerable economic situation, and the expectations sexual violence is social stigma; a survivor would be ashamed for women to be “good wives” who keep the family together. because she would lose her reputation as a “good woman” and it would be harder for her to find a partner. She would be judged and blamed by the community, as illustrated in the attitudes “Most of the times women… accept surrounding women’s culpability in sexual assault (see Table 2.1 in annex 2). The confidentiality of service providers was the violence because they are afraid not reliable and, in locations such as clinics or police stations, of losing the partner and most qualitative participants indicated it was very likely that they importantly his economic support.” would be retraumatized and blamed. Most women, especially adolescents, who were most at risk to suffer this type of abuse, do —Key informant interview with Peredo CASEC in Marigot not know their rights and do not seek services. “They …[women]… keep silence “They … [service providers] … will because they don’t want to be the ask her if she is not the one to put main subject of gossiping in the herself in that situation.” community and being finger-pointed —Focus group discussion with female youth in La Vallée all the times.” —Key informant interview with health sector representative, La Vallée. “They … [service providers] … will blame her behavior and question her disclosure of rape.” —Focus group discussion with female authority leaders in La Vallée Rethinking Power Program Evaluation in Southeast Haiti | 27 Figure 15: Proportion of women who have Of those who sought support for NPSV, police and health centers disclosed rape to friends, family, or others in were mentioned in both quantitative and qualitative surveys as the main services that survivors approached. During qualitative Marigot and La Vallée research, participants stated that most service providers would question and blame survivors about the incident and request proof of rape. “At the police station, the police officers will ask for proof, such as the clothes with blood on them, and sometimes she can even be retraumatized with the questions they will ask her.” In the narratives of participants, there was also an indication —Focus group discussion with female youth in La Vallée that several cases of rape were related to an economic power disparity between the perpetrator and the survivor, in which the perpetrator had more money and power than his victim. For that reason, the survivor was afraid of retaliation and preferred to “The police officers can say she is handle the case silently. responsible for that because she attracted the perpetrators by wearing “The reality in this community is if short skirt.” —Focus group discussion with female authority leaders in the perpetrator has a good economic Marigot background and well connected with some political leaders, police officers, or judges, they [women and her “They [health providers] … will ask family members] … are not going to her if she was not responsible for that disclose the rape.” by walking late in the community.” —Focus group discussion with parents in La Vallée —Focus group discussion with female authority leaders in La Vallée Figure 16: Proportion of women who sought help for rape, by resource group, among women who have experienced rape in Marigot and La Vallée Rethinking Power Program Evaluation in Southeast Haiti | 28 CONCLUSIONS Men were considered to be the head of households and should have power over women’s lives. Controlling behavior was The SASA! methodology has proved successful in preventing incredibly high in these communities, especially in La Vallée, violence against women in Uganda (Abramsky et al, 2014), and and there was a difference between Marigot and La Vallée in the is being used in many countries around the world, including overall gender dynamics. Haiti. The first adaptation and implementation in the country by Beyond Borders’ Rethinking Power program in 2010, followed Regarding the main outcomes of this study, La Vallée and Marigot by a process evaluation a process evaluation, led to the design tend to be similar in terms of IPV and NPV prevalence. We found and implementation of a combined intervention of SASA! and a relatively high prevalence of current physical and/or sexual IPV Power to Girls. The latter one is the result of the lessons learned in the past 12 months as one out of five women had experienced from the summative evaluation and uses the same community- IPV in the past 12 months from the current or most recent partner. led methodology as SASA! but centered on adolescent girls’ This rate is similar for NPSV, which is very high compared with safety and freedom. In order to create solid evidence of the other sites worldwide. effectiveness of SASA! and Power to Girls in Haiti, GWI has partnered with BB to lead the impact evaluation. Many IPV risk factors were identified through bivariate and multivariate analysis and the strongest ones, such as controlling GWI decided to use a quasi-experimental design as the best behavior, acknowledge the existing patriarchal structure of approach to evaluate the effectiveness and the impact of this these communities. It is worth mentioning the contribution of new combined methodology. A randomized controlled trial was intergenerational violence and child abuse in predicting IPV. In considered, but it was determined not to be appropriate for two addition, the level of disclosure of IPV and NPV and the potential main reasons. First, it would be difficult in a mountainous, rural risks associated were among other factors that revealed the area to implement a randomization process without incurring strong power imbalance in these communities, and between men high transportation costs and having a large staff to potentially and women. cover a very large implementation area. Second, SASA! and Power to Girls are community mobilization interventions and Economic disparities between men and women were mentioned use of media and other parts of the methodology, combined in the qualitative surveys as one of the contributing factors with close community connectedness and relationships would that creates power imbalance and acceptance of violence mean high risk of contamination and spillover to control sites. at the community level. While SASA! and Power to Girls do Intervention and comparison communities were selected in not have a livelihoods component, they work specifically on close collaboration with Beyond Borders, looking at factors like financial decision-making in couples, and girl’s groups include proximity to their office, proximity to original implementation a component on economic empowerment of girls. It will be communities, and reasonable distance and similarities between interesting to see how such intervention impacts the community. intervention and comparison communities. As traditional gender norms and attitudes prevail in both Additionally, since Beyond Borders intends to implement its communities, these methodologies can contribute to a reduction program in comparison communities after research is completed, of risk of violence against women and girls in the community there had to be reasonable possibility for this to occur. Through and increase helpful support to women experiencing violence. In a multi-clustering, cross-sectional survey, the primary outcomes, general, the community is also expected to increase activism to such as prevalence of IPV and gender norms have been prevent violence and demonstrate other indicators that represent measured at the household level for women and men. Other a greater balance of power. survey components were also implemented, including girl’s club- based, school-based, and qualitative surveys in order to track changes in traditional social norms and reduction of risks related to sexual health behaviors. In this baseline report, individual demographic characteristics, gender attitudes, prevalence and circumstances around IPV and NPV were explored. We found some differences between women in La Vallée and those in Marigot in their demographic characteristics. In La Vallée, women were more educated and were less likely to be employed. They also were more likely to marry than women in Marigot. However, women who did marry in Marigot tended to do so at younder ages than those in La Vallée. In terms of gender attitudes, roles, and dynamics, the findings revealed that these communities were rooted in gender inequalities and the predominance of the patriarchal norms. Rethinking Power Program Evaluation in Southeast Haiti | 29 REFERENCES Jansen, H. A. F. M., Watts, C., Ellsberg, M., Heise, L., & Garcia-Moreno, C. (2004). Interviewer training in the WHO multi-country study on women’s Abramsky, T., Devries, K., Kiss, L., Nakuti, J., Kyegombe, N., Starmann, E., health and domestic violence. Violence Against Women, 10(7), 831-849. Cundill, B., Francisco, L., Kaye, D., Musuya, T., Michau, L., Watts, C. (2014). Findings from the SASA! Study: A cluster randomised controlled trial to Johnson, L. E., & Benight, C. C. (2003). Effects of trauma-focused research assess the impact of a community mobilisation intervention to prevent on recent domestic violence survivors. Journal of Traumatic Stress, 16(6), violence against women and reduce HIV risk in Kampala, Uganda. BMC 567-571. doi:10.1023/B:JOTS.0000004080.50361.f3 [doi] Medicine, 122(12). Jorm, A. F., Kelly, C. M., & Morgan, A. J. (2007). Participant distress in Cromer, L. D., & Newman, E. (2011). Research ethics in victimization psychiatric research: A systematic review. Psychological Medicine, 37(7), studies: Widening the lens. Violence Against Women, 17(12), 1536-48; 917-926. doi:S0033291706009779 [pii] discussion 1549-58. doi:10.1177/1077801211436365 [doi] Kyegombe, N., Abramsky, T., Devries, K. M., Starmann, E., Michau, L., Ellsberg, M., Arango, D. J., Morton, M., Gennari, F., Kiplesund, S., Nakuti, J., Musuya, T., Watts, C. (2014). “SASA! is the Medicine That Contreras, M., Watts, C. (2014). “Prevention of Violence against Women Treats Violence. Qualitative Findings on How a Community Mobilisation and Girls: What Does the Evidence Say?” The Lancet, A Special Series on Intervention to Prevent Violence against Women Created Change in Violence against Women and Girls. Kampala, Uganda.” Global Health Action 7. Ellsberg, M., & Heise, L. (2002). Bearing witness: Ethics in domestic National Disability Authority. (2009). Ethical Guidance with People with violence research. Lancet, 359(9317), 1599-604. Disabilities. Dublin, Ireland: NDA. Ellsberg, M., & Heise, L. (2005). Researching violence against women: Salazar, M., Valladares, E., Ohman, A., & Hogberg, U. (2009). Ending A practical guide for researchers and activists. Geneva, Switzerland: intimate partner violence after pregnancy: Findings from a community- Program for Appropriate Technology in Health, World Health based longitudinal study in Nicaragua. BMC Public Health, 9, 350-2458-9- Organization. 350. doi:10.1186/1471-2458-9-350 [doi] Ellsberg, M., Heise, L., Peña, R., Agurto, S., & Winkvist, A. (2001). Schenk, K. & Williamson, J. (2005). Ethical Approaches to Gathering Researching domestic violence against women: Methodological and Information from Children and Adolescents in International Settings: ethical considerations. Studies in Family Planning, 32(1), 1-16. Guidelines and Resources. Washington, DC: Population Council. Fontes, L. A. (2004). Ethics in violence against women research: The Sikweyiya, Y., & Jewkes, R. (2012). Perceptions and experiences of sensitive, the dangerous, and the overlooked. Ethics & Behavior, 14(2), research participants on gender-based violence community based 141-174. doi:10.1207/s15327019eb1402_4 [doi] survey: Implications for ethical guidelines. PloS One, 7(4), e35495. doi:10.1371/journal.pone.0035495 [doi] Garcia Moreno, C., Watts, C., Jansen, H., Ellsberg, M., & Heise, L. (2003). Responding to violence against women: WHO’s multi-country study on Sikweyiya, Y., & Jewkes, R. (2013). Potential motivations for and perceived women’s health and domestic violence. Health and Human Rights, 6(2), risks in research participation: Ethics in health research. Qualitative Health 112-127. Research, 23(7), 999-1009. doi:10.1177/1049732313490076 [doi] Griffin, M. G., Resick, P. A., Waldrop, A. E., & Mechanic, M. B. (2003). Wagman, J., Francisco, L., Glass, N., Sharps, P. W., & Campbell, J. C. Participation in trauma research: Is there evidence of harm? Journal of (2008). Ethical challenges of research on and care for victims of intimate Traumatic Stress, 16(3), 221-227. doi:10.1023/A:1023735821900 [doi] partner violence. The Journal of Clinical Ethics, 19(4), 371-380. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=cmedm& Heise, L. (1998). Violence against women: An integrated ecological AN=19189769&site=ehost-live framework. Violence Against Women, 262-290. World Health Organization. (1999). Putting women’s safety first: Ethical Hlavka, H. R., Kruttschnitt, C., & Carbone-Lopez, K. C. (2007). Revictimizing and safety recommendations for research on domestic violence against the victims? Interviewing women about interpersonal violence. Journal of women. ( No. WHO/EIP/GPE/99.2). Geneva, Switzerland: Global Interpersonal Violence, 22(7), 894-920. doi:22/7/894 [pii] Programme on Evidence for Health Policy, World Health Organization. Institut Haïtien de l’Enfance (IHE) et ICF. 2018. Enquête Mortalité, Morbidité et Utilisation des Services (EMMUS-VI 2016-2017) Pétion-Ville, Haïti, et Rockville, Maryland, USA : IHE et ICF. Jansen, H. (2010). Swimming against the tide: Lessons learned from field research on violence against women in the Solomon Islands and Kiribati. Suva, Fiji: UN populations fund. Rethinking Power Program Evaluation in Southeast Haiti | 30 ANNEX 1: QUANTITATIVE AND QUALITIVE BASELINE SAMPLING DATA Table 1.1: Number of respondents surveyed Community La Vallée Marigot Women’s household survey concluded 819 1158 Household response rate 94% 97% Individual response rate 97% 99% Men’s household survey concluded 317 547 Household response rate 91% 98% Individual response rate 96% 99% Schools 363 368 Girl’s Clubs 267 Table 1.2: Number of FGD and KII Focus Groups / Participation In-depth Interviews La Vallée 12 focus groups 2 girl’s club (12-18) 1 male student (14-22 years old) (Musac and Ternier) 1 female student (14-22 years old) 1 male adult (25 or more) 1 female adult (25 or more) 1 male youth (18-24) 1 female youth (18-24) 1 male local authorities and leaders 1 female local authorities and leaders 1 school teachers (mixed) 1 school parents (mixed) 13 in-depth interviews 2 local authorities (Mayors, CASEC) 1 civil society 1 school director 1 police inspector 3 local leaders (pastor, women’s local NGO, international NGO) 1 court 2 girl’s club mentors 2 health sector representatives Marigot 10 focus groups 1 female student (14-22 years old) 1 male student (14-22 years old) (Corail Soult, Marigot downtown, Peredo) 1 female adult (25 or more) 1 male adult (25 or more) 1 female youth (18-24) 1 male youth (18-24) 1 female local authorities and leaders 1 male local authorities and leaders 1 school teacher (mixed) 1 school parent (mixed) 12 in-depth interviews 1 Rethinking Power Program Coordinator 2 local authorities (mayors, ASEC, city delegate) 1 civil society representative 1 education sector representative 3 local leaders (women’s local NGO, voodoo priests, international NGO) 1 private sector representative 1 school director 1 health sector representative 1 civil protection representative Rethinking Power Program Evaluation in Southeast Haiti | 31 ANNEX 2: BASELINE DATA TABLES AND FIGURES These symbols, when not defined for a table, are used to indicate the p-value level for a specific tests or models. 0.05 (*), 0.01 (**), 0.001(***): Pearson Chi-Square p-value significant difference for women in Marigot and La Vallée. 0.05 (+), 0.01 (++), 0.001(+++): Pearson Chi-Square p-value significant difference for men in Marigot and La Vallée. 0.05 (^), 0.01 (^^), 0.001(^^^): Pearson Chi-Square p-value significant difference for lifetime IPV between Marigot and La Vallée. 0.05 (#), 0.01 (##), 0.001(###): Pearson Chi-Square p-value significant difference for current IPV between Marigot and La Vallée. 0.05 (■), 0.01 (■■), 0.001(■■■): Pearson Chi-Square p-value significant difference for lifetime IPV in Marigot and other variables. 0.05 (□), 0.01 (□□), 0.001(□□□): Pearson Chi-Square p-value significant difference for current IPV in Marigot and other variables. 0.05(▲), 0.01 (▲▲), 0.001(▲▲▲): Pearson Chi-Square p-value significant difference for lifetime IPV in La Vallée and other variables. 0.05 (◊), 0.01 (◊◊), 0.001(◊◊◊): Pearson Chi-Square p-value significant difference for current IPV in La Vallée and other variables. GENERAL MALE AND FEMALE POPULATION DESCRIPTION Table 2.1. Percentage of women and men who agree with the different statements related to gender roles, norms, and dynamics Marigot La Vallée Female Male Female Male (n = 1158) (n = 547) (n = 819) (n = 317) Gender Roles Changing diapers, giving a bath, and feeding kids is mainly the mother’s 92.3 58.0 87.5 56.2 responsibility*** A woman’s role is taking care of her home and family*** 93.6 73.7 88.6 71.9 A man should have the final word about decisions in his home* ++ 65.9 75.7 60.2 66.6 Women and men should share authority in the family+ 84.6 87.4 85.5 82.0 Boys and Girls Treatment Boys should spend as much time as girls doing household chores++ 61.2 53.4 60.9 43.2 Girls should be allowed to socialize outside the home just as boys do** 19.8 23.2 15.1 17.7 Boys have more of a need to go to school than girls*** 10.5 13.0 4.9 11.4 Girls should be allowed to choose when to get married*** 54.8 56.5 46.9 52.7 VAWG Violence between husband and wife is a private matter++ 36.5 52.3 36.4 41.6 It is the entire community’s responsibility to prevent men from beating their 73.8 68.4 77.4 76.0 wives+ A woman should accept violence to keep her family together* 23.2 28.0 19.4 23.3 If a woman is raped, she has done something careless to put herself in that 20.4 16.5 18.2 15.5 situation Justification of IPV If she goes out without telling him* 22.3 7.9 18.4 8.8 If she neglects the children 17.0 5.9 14.2 5.4 If she argues with him 15.7 5.1 14.8 4.4 If she refuses to have sex with him*** 11.3 3.8 6.8 3.5 If he suspects she is having an affair with another man+ 34.9 13.3 34.4 18.6 At least one act 45.3 20.1 44.7 23.7 Rethinking Power Program Evaluation in Southeast Haiti | 32 Table 2.2: Repartition of the chores for the households of ever-partnered women Marigot La Vallée Community (n =968) (n =673) Washing clothes*** Women/Girls 95.9 88.1 Men/Boys/both/Other 4.1 11.9 Preparing food*** Women/Girls 95.9 90.2 Men/Boys/both/Other 4.1 9.8 Caring for child/children** Women/Girls 90.6 86.2 Men/Boys/both/Other 9.4 13.8 Spending more time on household chores*** Women/Girls 97.7 92.9 Men/Boys/both/Other 2.3 7.1 Table 2.3: Final decision-making process for the households of ever-partnered women Marigot La Vallée Commune (n =968) (n =673) Food and clothing yourself 18.2 16.5 partner/husband 30.2 28.2 Both 34.3 37.1 Someone else 14.4 12.8 other/no answer 3.0 5.3 Time with family, friends, or relatives yourself 13.5 13.5 partner/husband 25.7 25.4 Both 25.9 30.5 Someone else 10.4 10.5 other/no answer 24.4 20.1 Table 2.4: Couple discussion on sexual health and HIV test Marigot La Vallée Female Male Female Male (n= 968) (n= 361) (n= 673) (n= 237) Couple communication on sexual health discussed condom use 43.7 46.5 43.5 44.3 discussed sexual relationships with other partners 34.9 34.6 34.9 30.8 discussed HIV/AIDS 53.5 56.8 57.9 58.6 discussed woman’s (man’s) risk for HIV* 52.9 56.8 59.3 59.5 discussed how to protect family from HIV** 50.7 58.7 58.2 62.9 discussed getting tested for HIV* 57.5 50.4 62.4 44.7 Couple HIV test Ever tested* 61.1 43.8 66.9 40.5 Currently tested 30.3 22.7 33.4 21.9 Partner ever tested* 39.2 36.0 45.3 41.8 Partner currently tested** 16.0 23.0 22.1 25.3 Rethinking Power Program Evaluation in Southeast Haiti | 33 Table 2.5: Couple sexual behavior protection Marigot La Vallée Female Male Female Male (n= 968) (n= 361) (n= 673) (n= 237) Use of condom with partners Used condom, last 12 months 22.3 38.0 23.8 38.8 Reason given for not using condom n 550 183 343 119 8.5 10.9 8.2 7.6 wanted to have a child 1.8 1.6 3.5 1.7 did not have a condom man decided not to 11.3 9.3 11.1 5.9 woman decided not to 17.8 16.9 19.5 16.0 was in a monogamous relationship 60.5 61.2 57.7 68.9 Couple HIV test Partner has had other sexual partners, last 12 months*** 25.9 5.5 12.5 5.9 Respondent has had other sexual partners, last 12 months 2.9 31.6 3.1 29.1 Used condom last x sex*** n 258 124 97 76 22.5 72.6 45.4 78.9 Table 2.6: Percentage of controlling behavior per community Community Marigot La Vallée does not permit you to meet your female friends 21.7 19.5 tries to limit your contact with your family of birth* 8.1 11.0 insists on knowing where you are at all times*** 42.7 54.1 gets jealous or angry if you talk with another man 42.1 46.4 frequently accuses you of being unfaithful 19.9 16.8 expects you to ask his permission before seeking health care for yourself* 15.5 20.4 does not trust you with any money** 30.3 37.6 checks your cellphone to see who you have called / who has called you 25.6 21.5 At least one controlling behavior*** 67.4 79.6 Rethinking Power Program Evaluation in Southeast Haiti | 34 IPV PREVALENCE Table 3.1: Lifetime and 12-month IPV prevalence rate among ever-partnered women in Marigot and La Vallée Marigot (n=978) La Vallée (n=673) Community Lifetime Current Lifetime Current Economic^^^### 15.2 10.3 26.6 16.9 Emotional 35.3 26.9 32.2 24.5 Physical 14.9 10.2 14.7 10.0 Sexual 22.9 17.5 22.3 17.8 Physical and/or sexual 30.6 22.7 29.6 22.6 Table 3.2: Frequency of current IPV among survivors per community Marigot (n=220) La Vallée (n=152) Community once few many once few many Physical 18.2 50.5 31.3 22.4 44.8 32.8 Sexual 7.7 62.7 29.6 11.7 49.2 39.2 Physical and/or sexual 10.5 57.3 32.3 15.1 45.4 39.5 Rethinking Power Program Evaluation in Southeast Haiti | 35 RISK FACTORS Table 4.1: Significant individual risk factors for lifetime and 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Age■□□□, ▲◊◊◊ 15-24 37.8 34.0 28.0 25.6 25-34 32.7 27.3 30.9 25.4 35-44 31.1 19.9 39.2 31.5 45-54 22.9 12.1 24.0 17.4 55-64 23.0 12.2 24.1 10.3 Education Completed■■□□, ◊ no school/not stated 25.5 17.4 30.9 16.4 primary 37.3 27.0 34.4 28.9 secondary 32.4 26.8 26.5 22.6 higher 14.8 7.4 15.2 12.1 Cohabitating Relationships□□, ▲▲▲◊◊◊ currently married 27.5 20.2 26.9 20.4 living with man, not married 29.5 25.2 40.2 35.2 have married or lived with a man 35.6 12.6 31.4 11.4 have or have had a partner, living apart 33.3 28.0 18.8 14.6 Early Marriage■□, ▲▲ 19 or younger 36.7 28.1 43.6 31.7 20 or older 27.2 19.4 29.2 22.3 Main Source of Income■■□□□, ◊◊◊ Income from own work 25.4 15.9 28.8 21.2 Support from partner/husband 35.4 30.5 34.6 30.9 Support from parents/other relatives 35.7 30.8 28.6 20.6 No income/pension/social services/other/not stated 35.2 21.1 18.6 5.7 Have ever been pregnant□, ▲▲◊ no 30.8 30.8 18.0 14.6 yes 30.5 21.5 31.3 23.8 Rethinking Power Program Evaluation in Southeast Haiti | 36 Table 4.2: Partner’s characteristics as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Partner’s age■□□□ 15-24 38.2 35.5 27.8 25.9 25-34 34.3 31.8 29.6 24.4 35-44 30.2 22.7 33.6 25.6 45-54 30.1 17.3 22.6 17.9 55-64 22.0 13.2 31.7 23.8 65 + 16.1 7.1 29.7 14.1 Partner’s education□, ◊ no school/no answer 30.8 19.1 27.4 18.1 primary 32.7 24.0 38.7 30.7 secondary 31.0 26.9 26.9 22.3 higher 19.6 16.1 25.7 20.0 Partner has had another relationship■■■□□□, ▲▲▲◊◊◊ yes 39.2 29.4 45.7 34.9 no 25.8 19.0 25.7 19.7 Partner has been involved in a physical fight with another man■■■□□□, ▲▲▲◊◊◊ yes 61.0 51.2 71.0 58.1 no 29.2 21.5 27.6 20.9 Table 4.3: Couple communication as significant risk factors for lifetime and 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Couple communication Things that have happened to him in the day■■■□□, ▲▲▲ yes 27.9 21.2 27.4 22.1 no 45.9 31.5 48.6 27.1 Things that have happened to you during the day■■■□□□, ▲▲▲◊◊◊ yes 27.9 21.3 27.5 21.8 no 48.1 31.8 52.7 30.9 Her worries or feelings■■□, ◊◊◊ yes 27.7 21.0 27.3 21.6 no 48.1 33.3 55.6 33.3 His worries or feelings□□□, ◊◊◊ yes 28.4 21.3 26.4 21.9 no 36.1 26.4 41.7 25.2 Rethinking Power Program Evaluation in Southeast Haiti | 37 Table 4.4: Confidence in relationship as significant risk factors for lifetime and past 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Confidence in relationship Safe from violence in your relationship■■■□□□, ▲▲▲◊◊◊ often 20.4 14.0 20.9 15.9 sometimes 33.7 27.9 37.2 33.1 rarely 44.9 40.2 58.5 45.3 never/no answer 56.5 36.3 44.9 26.9 Confident in your ability to discuss issues of equality with your partner■■■□□□, ▲▲▲◊◊◊ often 21.4 14.1 17.1 13.6 sometimes 25.9 21.8 36.3 32.2 rarely 26.7 20.9 34.2 27.8 never/no answer 45.0 32.3 43.5 27.3 Table 4.5: Love and kindness as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Love and Kindness in Relationship Emotionally close to your partner■■■□□□, ▲▲▲◊◊ yes 24.9 19.5 25.7 20.0 no 45.8 31.4 44.1 32.2 Valued by your partner■■■□□□, ▲▲▲◊◊◊ yes 25.3 19.9 25.7 20.0 no 47.4 31.7 51.0 37.3 Respected by your partner■■■□□□, ▲▲▲◊◊◊ yes 24.9 19.5 24.9 19.1 no 48.5 33.0 54.2 41.1 Rethinking Power Program Evaluation in Southeast Haiti | 38 Table 4.6: Controlling behaviors as significant risk factors for lifetime and 12-month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Controlling Behaviors Does not permit you to meet your female friends■■■□□□, ▲▲▲◊◊ yes 50.5 41.0 45.0 33.6 no 25.1 17.7 25.8 19.9 Tries to limit your contact with your family of birth■■■□□□, ▲▲▲◊◊◊ yes 53.8 39.7 54.1 44.6 no 28.5 21.2 26.5 19.9 Insists on knowing where you are at all times■■■□□□, ▲▲▲◊◊◊ yes 44.1 35.8 39.3 32.1 no 20.5 13.0 18.1 11.3 Gets jealous or angry if you talk with another man■■■□□□, ▲▲▲◊◊◊ yes 45.3 36.5 40.1 33.0 no 19.8 12.7 20.5 13.6 Frequently accuses you of being unfaithful■■■□□□, ▲▲▲◊◊◊ yes 61.1 50.3 63.7 49.6 no 23.0 15.9 22.7 17.1 Expects you to ask his permission before seeking health care for yourself■■■□□□, ▲▲▲◊◊◊ yes 54.7 46.7 44.5 35.0 no 26.2 18.3 25.7 19.4 Does not trust you with any money■■■□□, ▲ yes 39.2 28.3 34.8 25.3 no 26.8 20.3 26.4 21.0 Checks your cellphone to see who you have called / who has called you■■■□□, ▲ yes 46.0 38.7 42.8 40.0 no 25.3 17.2 25.9 17.8 At least one controlling behavior■■■□□□, ▲▲▲◊◊◊ yes 38.3 29.1 34.7 27.2 no 14.6 9.5 9.5 4.4 Number of controlling behaviors■■■□□□, ▲▲▲◊◊◊ 0 14.6 9.5 9.5 4.4 1-2 23.4 15.8 24.7 18.9 3+ 51.4 40.8 44.0 35.0 Rethinking Power Program Evaluation in Southeast Haiti | 39 Table 4.7: Intergenerational violence as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Intergenerational Violence Woman witnessed violence as a child■, ▲◊◊ yes 39.1 29.6 41.8 37.3 no 29.4 21.8 28.2 21.0 Partner witnessed violence as a child■■■□□□, ▲▲▲◊◊ yes 60.0 50.0 58.5 41.5 no 29.3 21.6 27.7 21.4 Experienced violence as a child from her parents■■■□□□, ▲▲▲◊◊◊ yes 54.2 43.0 44.1 36.2 no 26.5 19.2 26.2 19.4 Experienced violence as a child from others■■□□□, ▲▲▲◊◊◊ yes 45.3 40.7 46.5 38.4 no 29.1 21.0 27.1 20.3 Table 4.8: Justification of violence as significant risk factors for lifetime and 12 month prevalence of physical and sexual IPV among ever-partnered women in Marigot and La Vallée Marigot (n=968) La Vallée (n=673) Lifetime Current Lifetime Current Justification of Violence If she goes out without telling him▲▲◊ yes 35.3 26.1 40.5 29.8 no 29.3 21.8 27.2 21.0 If she neglects the children■■■□□, ▲ yes 42.4 31.6 39.5 25.6 no 28.3 21.0 28.1 22.1 If she argues with him■■□□, ▲▲ yes 41.5 31.7 42.9 28.6 no 28.7 21.2 27.3 21.6 If she refuses to have sex with him■■■, ▲ yes 45.1 29.2 44.7 25.5 no 28.7 21.9 28.4 22.4 If he suspects she is having an affair with another man■■, ▲▲▲◊◊◊ yes 36.7 26.0 41.7 31.1 no 27.3 21.0 23.1 18.0 at least one act■■■□□□, ▲▲▲◊◊◊ yes 37.4 27.9 39.9 29.4 no 25.0 18.5 21.0 16.9 Rethinking Power Program Evaluation in Southeast Haiti | 40 Table 4.9: Triggers for IPV among ever-partnered women who have experienced IPV in Marigot and La Vallée Community Marigot La Vallée No reason* 37.8 29.6 When drunk* 4.4 9.5 Money problems*** 6.8 17.6 Difficulties at work** 0.3 3.5 When he is unemployed 2.0 4.5 No food at home* 3.4 7.5 Problems with his or her family 4.4 7.0 She is pregnant 2.0 2.0 He is jealous* 18.2 25.6 She refuses sex*** 8.4 19.6 She is disobedient 8.4 10.1 He wants to teach her a lesson 4.4 5.5 He wants to show her he is the boss 8.4 8.5 Other 9.1 13.6 NON-PARTNER SEXUAL VIOLENCE Table 5.1: Lifetime and 12-month non-partner prevalence rate among all women in Marigot and La Vallée Marigot (n=1158) La Vallée (n=819) Community Lifetime Current Lifetime Current Rape or Attempted rape 11.7 5.1 13.6 6.0 Sexual touching 7.3 4.3 8.7 4.8 Sexual harassment 11.3 7.3 11.8 7.3 Any Non-Partner Sexual Abuse 21.8 12.0 24.7 14.0 Rethinking Power Program Evaluation in Southeast Haiti | 41 Table 5.2: Age at first non-partner sexual assault among all women in Marigot and La Vallée Marigot La Vallée Rape or Attempted Rape n=135 n=111 <15 25.9 17.1 15-19 32.6 29.7 20-24 16.3 18.0 25 + 25.2 35.1 Unwanted Sexual Touching* n=85 n=71 <15 12.9 26.8 15-19 43.5 29.6 20-24 27.1 18.3 25 + 16.5 25.4 Sexual Harassment** n=131 n=97 <15 3.1 9.3 15-19 42.0 22.7 20-24 26.0 27.8 25 + 29.0 40.2 Sexual Harassment** n=131 n=97 <15 12.6 14.9 15-19 37.5 27.7 20-24 22.1 19.8 25 + 27.7 37.6 Table 5.3: Identification of perpetrators of non-partner sexual assault among all women in Marigot and La Vallée Marigot La Vallée Rape or Attempted Rape n=135 n=111 Family member or friend 8.1 4.5 Known perpetrator 64.4 59.5 Community figure 0.7 0.0 Unknown perpetrator 43.7 45.0 Unwanted Sexual Touching n=85 n=71 Family member 9.4 5.6 Known perpetrator 64.7 60.6 Community figure 0.0 0.0 Unknown perpetrator 40.0 42.3 Sexual Harassment n=131 n=97 Family member 1.5 1.0 Known perpetrator 26.0 25.8 Community figure 2.3 3.1 Unknown perpetrator 12.2 13.4 Rethinking Power Program Evaluation in Southeast Haiti | 42 RESPONSES TO VIOLENCE BY SURVIVORS Table 6.1: Disclosure of intimate partner violence, by group, among women who have experienced IPV in Marigot and La Vallée Marigot (n=296) La Vallée (n=199) No one** 63.5 49.2 Her family 15.5 19.6 His family 6.4 7.5 Friend** 16.2 27.1 Leader/NGO/Other 4.7 4.0 Table 6.2: Disclosure of forced intercourse, by group, among women who have experienced rape in Marigot and La Vallée Marigot (n=85) La Vallée (n=63) No one 44.7 54.0 Her family 38.8 23.8 In-laws 1.2 4.8 Leader/NGO/Other 7.1 12.7 Friend 14.1 25.4 Table 6.3: Seeking help for intimate partner violence, by resource group, among women who have experienced IPV in Marigot and La Vallée Marigot (n=296) La Vallée (n=199) Police 5.4 8.5 Hospital/Health center* 4.4 10.6 Legal advice center/lawyer/court 3.4 7.0 Community leader 2.0 3.0 Women’s organization 3.0 3.0 Priest/religious leader* 2.0 5.5 Elsewhere 1.4 4.0 Nowhere* 85.8 77.4 Table 6.4: Discussions regarding intimate partner violence, by initiator, among women who have experienced IPV in Marigot and La Vallée Marigot (n=296) La Vallée (n=199) Partner initiated a discussion about violence in the last 12 months 24.3 27.1 Someone initiated a discussion about dealing with partner's violence*** 18.9 32.7 Rethinking Power Program Evaluation in Southeast Haiti | 43 Table 6.5: Individual support response received for forced intercourse, among women who have experienced NPV in Marigot and La Vallée Marigot (n=85) La Vallée (n=63) Blamed me 4.7 9.5 Supported me 30.6 22.2 Were indifferent 2.4 3.2 Told me to keep it quiet 5.9 11.1 Advised me to report it to the police 5.9 3.2 Other response 9.4 11.1 Table 6.6: Seeking help for forced intercourse, by resource group, among women who have experienced forced intercourse in Marigot and La Vallée Marigot (n=85) La Vallée (n=63) Police 10.6 6.3 Hospital/Health Center 9.4 9.5 Legal Advice Center/Lawyer/Court 4.7 3.2 Community Leader 1.2 6.3 Women's Organization 2.4 4.8 Priest/Religious Leader 0.0 9.5 Elsewhere 0.0 3.2 Nowhere 82.4 73.0 Table 6.7: Awareness of a place to go in case of sexual violence among all women in Marigot and La Vallée Marigot La Vallée Community (n=1158) (n=819) Know of a place a woman/girl could go for 62.6 59.8 help if she is victim of sexual violence (n=725) (n=490) Family 1.7 2 Police 53.5 55.1 Health Services** 6.9 11.6 Community Leader 9.1 7.1 Religious Leader 1 1.6 Women’s group 59.6 54.1 School 0 0 Friends 0.3 0.6 Other*** 4.8 10.8 Rethinking Power Program Evaluation in Southeast Haiti | 44 ANNEX 3: MULTIVARIATE MODEL RESULTS Table 7.1: Logistic regression crude and adjusted odds ratios of the associations between lifetime IPV physical and/or sexual violence ever and explanatory variables – La Vallée.13 Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper Age at first union (cohabited or married) (20 or older) 19 or younger** 1.87 1.19 2.93 1.15 0.68 1.97 0.88 0.46 1.69 never married/cohabitated 0.70 0.47 1.06 1.04 0.50 2.18 0.97 0.41 2.30 Age at first pregnancy (19 or younger) 20-24**+ 0.55 0.37 0.84 0.62 0.39 0.98 0.68 0.39 1.18 25 or older***+ 0.43 0.27 0.67 0.51 0.30 0.86 0.58 0.31 1.07 never pregnant*** 0.34 0.19 0.61 0.59 0.29 1.22 0.78 0.33 1.83 Partnership status (detailed) (have or have had a partner - dating) currently married% 1.60 0.97 2.63 1.61 0.64 4.07 3.04 1.03 8.98 living with man, not married***+% 2.91 1.76 4.83 2.44 1.04 5.77 2.95 1.09 8.00 have married or lived with a man* 1.99 1.03 3.83 1.85 0.71 4.80 2.12 0.70 6.43 Partner has had another relationship***+++ 2.43 1.64 3.61 2.38 1.59 3.57 1.24 0.75 2.06 Partner has been involved in a physical fight with another man***+++%% 6.42 2.90 14.22 6.22 2.78 13.91 4.46 1.76 11.29 Things that have happened to him in the day*** 0.40 0.24 0.66 1.14 0.35 3.70 1.15 0.30 4.32 Things that have happened to you during the day*** 0.34 0.19 0.59 1.32 0.25 7.03 0.74 0.10 5.37 Your worries or feelings*** 0.30 0.17 0.53 0.79 0.18 3.49 1.19 0.21 6.67 His worries or feelings*** 0.50 0.34 0.74 0.89 0.51 1.56 0.79 0.43 1.47 Safe from violence in your relationship (rarely) never/no answer 0.58 0.29 1.17 0.54 0.24 1.21 0.45 0.18 1.12 often***+% 0.19 0.10 0.34 0.39 0.19 0.81 0.38 0.17 0.83 sometimes** 0.42 0.22 0.81 0.61 0.29 1.33 0.57 0.25 1.31 13 *(0.05), **(0.01), ***(0.001): model coefficients p-value significance for univariate models. +(0.05), ++(0.01), +++(0.001): model coefficients p-value significance for multivariate models 1. %(0.05), %%(0.01), %%%(0.001): model coefficients p-value significance for multivariate models 2. (continued on next page...) Rethinking Power Program Evaluation in Southeast Haiti | 45  Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper Confident in your ability to discuss issues of equality with your partner (rarely) never/no answer 1.48 0.85 2.59 1.88 0.97 3.64 1.72 0.84 3.52 often** 0.40 0.23 0.69 0.71 0.37 1.37 0.72 0.36 1.46 sometimes 1.10 0.62 1.95 1.61 0.83 3.13 1.52 0.74 3.12 Emotionally close to your partner*** 0.44 0.30 0.64 0.95 0.48 1.88 1.08 0.51 2.27 Valued by your partner*** 0.33 0.22 0.51 2.50 0.64 9.77 1.92 0.44 8.36 Respected by your partner***+% 0.28 0.18 0.43 0.21 0.06 0.73 0.25 0.06 0.96 Number of acts of controlling behavior (none) 1-2'***+++%%% 3.13 1.66 5.92 3.64 1.87 7.09 3.38 1.67 6.81 3 +'***+++%%% 7.51 4.04 13.94 7.43 3.89 14.19 6.44 3.19 12.98 Partner witnessed violence as a child***+++ 3.69 1.93 7.03 3.34 1.72 6.48 1.94 0.89 4.26 Woman witnessed violence as a child* 1.83 1.09 3.06 1.45 0.84 2.51 1.46 0.77 2.74 Experienced violence as a child from her parents***++ 2.22 1.49 3.31 1.94 1.28 2.94 1.31 0.81 2.13 Experienced violence as a child from others***++% 2.34 1.48 3.71 2.11 1.31 3.40 1.95 1.12 3.38 If she goes out without telling him** 1.82 1.21 2.74 1.15 0.71 1.89 1.07 0.60 1.91 If he suspects she is having an affair with another man***+++% 2.39 1.70 3.36 2.07 1.41 3.03 1.72 1.10 2.69 If she neglects the children* 1.67 1.05 2.67 1.04 0.60 1.82 1.00 0.51 1.96 If she argues with him** 2.00 1.29 3.10 1.25 0.74 2.12 1.39 0.75 2.57 If she refuses to have sex with him* 2.03 1.11 3.71 1.31 0.68 2.52 1.09 0.49 2.40 Rethinking Power Program Evaluation in Southeast Haiti | 46 Table 7. 2: Logistic regression crude and adjusted odds ratios of the associations between current IPV physical and/or sexual violence and explanatory variables – La Vallée. Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper Age***++% 0.97 0.96 0.99 0.97 0.95 0.99 0.97 0.95 1.00 Education of respondent (higher) no school/NA 1.42 0.46 4.36 1.74 0.51 5.93 0.80 0.20 3.21 primary* 2.95 1.00 8.73 2.49 0.79 7.84 1.44 0.40 5.14 secondary 2.11 0.71 6.26 1.64 0.53 5.05 0.88 0.26 3.04 Main source of Income (income from own work) Support from partner/husband* 1.66 1.10 2.52 1.07 0.68 1.70 1.21 0.70 2.09 Support from parents/other 0.97 0.57 1.64 1.39 0.74 2.61 1.09 0.54 2.21 relatives No income/pension/social 0.23 0.08 0.65 0.22 0.07 0.64 0.28 0.09 0.88 services/other/NA**++% Age at first pregnancy (19 or younger) 20-24 0.75 0.48 1.17 0.93 0.57 1.51 1.19 0.68 2.08 25 or older* 0.54 0.33 0.88 0.91 0.51 1.64 1.08 0.54 2.12 never pregnant* 0.43 0.23 0.82 0.68 0.29 1.62 1.10 0.42 2.90 Partnership status (detailed) (have or have had a partner - dating) currently married+%% 1.50 0.86 2.61 2.53 1.16 5.50 4.19 1.77 9.92 living with man, not 3.18 1.84 5.49 3.39 1.60 7.19 3.37 1.47 7.73 married***++%% have married or lived with a man 0.76 0.32 1.80 1.24 0.45 3.40 1.22 0.40 3.68 Partner has had another relationship***+++ 2.19 1.44 3.33 2.13 1.39 3.26 1.29 0.74 2.24 Partner has been involved in a physical fight with another man***+++%% 5.25 2.51 10.98 5.04 2.38 10.64 3.53 1.41 8.83 Safe from violence in your relationship (rarely) never/no answer* 0.45 0.21 0.93 0.48 0.21 1.11 0.63 0.25 1.62 often***+% 0.23 0.13 0.42 0.45 0.22 0.95 0.43 0.18 1.00 sometimes 0.60 0.31 1.15 0.83 0.38 1.79 0.75 0.32 1.78 (continued on next page...) Rethinking Power Program Evaluation in Southeast Haiti | 47  Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper Confident in your ability to discuss issues of equality with your partner (rarely) never/no answer 0.97 0.53 1.78 1.23 0.62 2.46 1.23 0.56 2.66 often** 0.41 0.22 0.74 0.70 0.35 1.40 0.79 0.37 1.67 sometimes 1.23 0.67 2.25 1.71 0.86 3.40 1.63 0.77 3.46 Emotionally close to your partner** 0.53 0.35 0.79 1.01 0.53 1.90 0.94 0.46 1.92 Valued by your partner*** 0.42 0.27 0.66 2.51 0.65 9.68 1.67 0.39 7.16 Respected by your partner***+% 0.34 0.22 0.52 0.20 0.06 0.70 0.27 0.07 1.05 Number of acts of controlling behavior (none) 1-2'***+++%% 5.09 2.12 12.23 5.49 2.26 13.34 4.55 1.79 11.56 3 +'***+++%%% 11.77 5.01 27.66 10.59 4.45 25.22 7.70 3.04 19.53 Partner witnessed violence as a child**+ 2.61 1.36 4.99 2.27 1.16 4.46 1.28 0.56 2.94 Woman witnessed violence as a child**+ 2.25 1.32 3.82 1.78 1.02 3.13 1.68 0.87 3.26 Experienced violence as a child from her parents***++ 2.36 1.55 3.59 2.00 1.29 3.09 1.56 0.93 2.64 Experienced violence as a child from others***++%% 2.45 1.52 3.95 2.19 1.34 3.59 2.21 1.23 3.98 If she goes out without telling him*% 1.59 1.02 2.47 1.18 0.73 1.90 1.03 0.57 1.85 If he suspects she is having an affair with another man***+++% 2.05 1.42 2.96 1.95 1.31 2.90 1.65 1.02 2.66 0.09 Rethinking Power Program Evaluation in Southeast Haiti | 48 Table 7. 3: Logistic regression crude and adjusted odds ratios of the associations between lifetime IPV physical and/or sexual violence and explanatory variables – Marigot. Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper Age***+ 0.98 0.97 0.99 0.98 0.97 1.00 0.99 0.97 1.01 Education of respondent (higher) no school/NA 1.97 0.67 5.85 2.47 0.80 7.62 2.05 0.58 7.23 primary*+ 3.42 1.15 10.20 3.23 1.06 9.83 2.61 0.76 8.96 secondary 2.76 0.93 8.17 2.50 0.84 7.47 1.82 0.55 6.10 Age at first union (cohabited or married) (20 or older) 19 or younger* 1.55 1.10 2.19 1.06 0.69 1.63 1.08 0.65 1.77 never married/cohabitated 1.34 0.96 1.87 1.10 0.70 1.73 0.77 0.45 1.30 Main source of Income (income from own work) Support from partner/husband** 1.61 1.15 2.25 1.40 0.99 1.99 1.31 0.88 1.96 Support from parents/other 1.63 1.13 2.35 1.73 1.11 2.69 1.11 0.65 1.87 relatives**+ No income/pension/social 1.59 0.94 2.71 1.61 0.94 2.75 1.81 0.97 3.39 services/other/NA Age at first pregnancy (19 or younger) 20-24* 0.69 0.49 0.98 0.80 0.53 1.19 0.88 0.55 1.40 25 or older** 0.54 0.37 0.80 0.76 0.47 1.22 1.13 0.65 1.96 never pregnant*+ 0.64 0.41 1.00 0.51 0.29 0.90 0.79 0.41 1.54 Partner has had another relationship***+++ 1.86 1.40 2.46 1.76 1.33 2.34 1.20 0.85 1.70 Partner has been involved in a physical fight with another man***+++%% 3.78 1.99 7.20 3.33 1.74 6.39 2.06 0.95 4.44 Things that have happened to him in the day*** 0.46 0.32 0.65 1.07 0.40 2.86 1.60 0.57 4.49 Things that have happened to you during the day*** 0.42 0.29 0.61 1.45 0.37 5.69 0.51 0.12 2.15 Your worries or feelings*** 0.41 0.29 0.60 0.58 0.15 2.30 1.49 0.37 5.90 (continued on next page...) Rethinking Power Program Evaluation in Southeast Haiti | 49  Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper His worries or feelings* 0.70 0.52 0.94 0.96 0.58 1.61 0.98 0.60 1.61 Safe from violence in your relationship (rarely) never/no answer 1.59 0.95 2.68 0.57 0.30 1.08 1.26 0.65 2.44 often***+++% 0.32 0.20 0.49 0.33 0.19 0.59 0.54 0.30 0.95 sometimes 0.62 0.39 1.00 0.71 0.40 1.23 0.82 0.46 1.46 Confident in your ability to discuss issues of equality with your partner (rarely) never/no answer***+% 2.24 1.49 3.38 1.77 1.05 2.97 1.79 1.07 3.00 often 0.74 0.48 1.17 0.97 0.52 1.79 1.03 0.58 1.85 sometimes 0.96 0.62 1.49 1.19 0.69 2.05 1.08 0.64 1.85 Emotionally close to your partner*** 0.39 0.29 0.53 0.83 0.43 1.59 0.60 0.31 1.16 Valued by your partner***+ 0.38 0.28 0.51 5.93 1.06 33.29 5.16 0.91 29.21 Respected by your partner*** 0.35 0.26 0.48 0.19 0.04 1.02 0.20 0.04 1.08 Number of acts of controlling behavior (none) 1-2'***+% 1.79 1.19 2.70 1.90 1.15 3.15 1.76 1.11 2.80 3 +'***+++%%% 6.22 4.27 9.06 6.46 4.11 10.16 4.93 3.06 7.94 Partner witnessed violence as a child***+++% 3.62 1.89 6.92 3.43 1.75 6.76 2.65 1.21 5.78 Woman witnessed violence as a child* 1.54 1.03 2.31 1.00 0.64 1.55 1.11 0.67 1.84 Experienced violence as a child from her parents***+++% 3.28 2.28 4.73 3.06 2.09 4.49 1.76 1.13 2.73 Experienced violence as a child from others** 2.02 1.29 3.16 1.50 0.93 2.42 1.17 0.69 1.99 If she neglects the children*** 1.87 1.32 2.65 1.42 0.93 2.18 1.20 0.73 1.96 If she argues with him** 1.77 1.22 2.55 1.13 0.71 1.80 1.18 0.70 2.00 (continued on next page...) Rethinking Power Program Evaluation in Southeast Haiti | 50  Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper If she refuses to have sex with him*** 2.05 1.37 3.05 1.55 0.98 2.45 1.45 0.85 2.48 If he suspects she is having an affair with another man**+++% 1.54 1.16 2.05 1.19 0.85 1.65 1.03 0.70 1.51 0.12 Rethinking Power Program Evaluation in Southeast Haiti | 51 Table 7. 4: Logistic regression crude and adjusted odds ratios of the associations between current IPV physical and/or sexual violence and explanatory variables – Marigot. Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper Age***+++%%% 0.96 0.95 0.97 0.96 0.94 0.98 0.96 0.94 0.98 Education of respondent (higher) no school/NA 2.63 0.61 11.39 5.36 1.18 24.34 5.31 1.00 28.34 primary*+% 4.62 1.07 20.03 5.60 1.25 25.08 5.23 1.01 27.18 secondary*+ 4.57 1.06 19.72 4.20 0.96 18.40 3.64 0.72 18.31 Age at first union (cohabited or married) (20 or older) 19 or younger* 1.62 1.11 2.37 1.00 0.61 1.63 1.03 0.60 1.80 never married/cohabitated* 1.48 1.03 2.13 0.67 0.25 1.78 0.75 0.27 2.11 Main source of income (income from own work) Support from partner/ 2.32 1.60 3.35 1.78 1.21 2.62 1.55 1.00 2.41 husband***++ Support from parents/other 2.35 1.58 3.51 2.47 1.46 4.19 1.91 1.04 3.49 relatives***++% No income/pension/social 1.42 0.76 2.64 1.56 0.82 2.96 1.89 0.91 3.91 services/other/NA Age at first pregnancy (19 or younger) 20-24 0.76 0.52 1.11 0.87 0.55 1.39 1.00 0.59 1.70 25 or older** 0.51 0.33 0.80 0.80 0.45 1.39 1.15 0.61 2.17 never pregnant 1.07 0.68 1.68 0.87 0.47 1.61 1.52 0.73 3.16 Partnership status (detailed) (have or have had a partner - dating) currently married 0.65 0.42 1.02 1.16 0.37 3.61 3.13 0.89 10.99 living with man, not married 0.86 0.58 1.28 0.98 0.34 2.81 2.01 0.64 6.29 have married or lived with a 0.37 0.20 0.67 0.51 0.16 1.61 0.52 0.15 1.81 man** Partner has had another relationship***++ 1.77 1.31 2.41 1.67 1.23 2.28 1.41 0.96 2.08 Partner has been involved in a physical fight with another man***+++ 3.84 2.04 7.23 3.42 1.80 6.48 1.99 0.92 4.31 Things that have happened to him in the day** 0.58 0.40 0.86 1.06 0.40 2.80 1.50 0.50 4.48 (continued on next page...) Rethinking Power Program Evaluation in Southeast Haiti | 52  Univariate Models Multivariate Model 1 Multivariate Model 2 Crude Odd 95% C.I. Adjusted 95% C.I. Adjusted 95% C.I. Ratio Lower Upper Odd Ratio Lower Upper Odd Ratio Lower Upper Things that have happened to you during the day** 0.58 0.39 0.87 1.46 0.37 5.71 1.46 0.30 6.97 Your worries or feelings** 0.53 0.36 0.79 0.56 0.15 2.09 0.66 0.15 2.91 Safe from violence in your relationship (rarely) never/no answer 0.85 0.50 1.44 0.57 0.30 1.07 0.98 0.49 1.99 often***+++%% 0.24 0.15 0.38 0.33 0.19 0.59 0.41 0.22 0.76 sometimes* 0.58 0.35 0.94 0.71 0.40 1.23 0.75 0.41 1.38 Confident in your ability to discuss issues of equality with your partner (rarely) never/no answer**% 1.80 1.16 2.80 1.77 1.05 2.98 1.96 1.12 3.44 often 0.62 0.37 1.03 0.96 0.52 1.78 0.97 0.49 1.89 sometimes 1.05 0.65 1.70 1.19 0.69 2.04 1.27 0.70 2.29 Emotionally close to your partner*** 0.53 0.38 0.73 0.83 0.43 1.60 0.60 0.30 1.21 Valued by your partner***+% 0.53 0.38 0.75 5.92 1.06 33.24 9.32 1.32 65.90 Respected by your partner***% 0.49 0.35 0.68 0.20 0.04 1.03 0.10 0.02 0.67 Number of acts of controlling behavior (none) 1-2'**+ 1.79 1.10 2.91 1.90 1.15 3.13 1.46 0.85 2.50 3 +'***+++%%% 6.57 4.26 10.13 6.44 4.11 10.10 3.94 2.33 6.68 Partner witnessed violence as a child***+++% 3.64 1.92 6.90 3.39 1.74 6.59 2.59 1.18 5.70 Experienced violence as a child from her parents***+++% 3.16 2.17 4.59 2.82 1.91 4.16 1.81 1.15 2.85 Experienced violence as a child from others***++ 2.59 1.63 4.09 1.95 1.20 3.18 1.34 0.77 2.33 If she neglects the children** 1.74 1.20 2.54 1.47 0.95 2.30 1.19 0.70 2.01 If she argues with him** 1.73 1.17 2.55 1.40 0.88 2.22 1.13 0.66 1.95 0.05 Rethinking Power Program Evaluation in Southeast Haiti | 53 ANNEX 4: Community mobilization interventions are successful in transforming harmful gender norms because they guide RESEARCH DESIGN community members at all levels of the ecological framework through gradual and sustainable change. This approach is AND METHODS founded in the transtheoretical model of behavior change. According to this model, there are six key stages through which Theory of Change behavior changes occurs: precontemplation, contemplation, preparation, action, maintenance, and termination (Prochaska, The original SASA! methodology and Power to Girls were 1997). Community mobilization efforts informed by this developed based on the perspective that IPV and other forms framework assist individuals and communities as they move of VAWG are the result of multiple causal layers. The framework, through these six stages in an organic and empowering manner. originally developed by Urle Bronfrenbrenner (1994) and since The RP intervention is based on this model. SASA!’s four steps adapted by Lori Heise for VAWG (1998), presents the causes of take stakeholders from all levels of the community through each IPV at the macro social, community, interpersonal, and individual step of the intervention, changing over time their knowledge, levels. At the community level, imbalance of power is manifested attitudes, and behaviors that perpetuate harmful gender norms in inequitable norms and harmful practices. These norms are and creating an enabling and empowering environment that often perpetuated by religious or cultural justification, which supports safety, non-violence and the dignity of women, men, deters action by stakeholders who are crucial to preventing girls and boys. violence. Lori Michau and colleagues argue that inequitable norms can be addressed through education and capacity Rethinking Power Model building, which fosters collective action (Michau et al., 2014). This approach allows participants to feel empowered to contribute In order to conceptualize the theory of change of the Rethinking to an enabling environment in which a significant part of the Power project, the evaluation and project teams reviewed the population practices and advocates for gender equality. Using original SASA! logic model and the complementary Power a similar approach, norms that perpetuate gender inequality to Girls to identify which areas were still applicable for the and drive violence can be addressed in the healthcare sector expanded project and where additions needed to be made to by carrying out educational and behavior change interventions incorporate the new focus on girls. These are detailed in the among health care professionals and other stakeholders model below. (Gennari, 2014). Rethinking Power Program Evaluation in Southeast Haiti | 54  Levels of SASA! Initial Intermediate Longer terms Context Activities reaching each Impact outcomes outcomes outcomes circle of influence Socio-demographic factors Societal Knowledge Skills Individual & Reduced social acceptance Sex National Policy • Recognizing VAWG • Response to Collective Capacity of gender inequality, IPV age Makers, Media as a problem women and girls • Supportive and sexual abuse of girls income • Types of VAWG experiencing environment Education • Consequences of violence • Enhanced ability Decrease in experience& Employment VAWG • Hold boys & men to prevent and perpetration of IPV and Religion • VAWG/HIV linkage accountable respond to VAWG sexual abuse of girls Place of residence/levels • Promote balanced of mobility power Increase in girl’s freedom • Support activists / and feelings of safety Couples Socio-cultural factors Decrease in HIV& SRH risk Family Characteristics Action behaviors Social Support Institutional Awareness Behaviors Alcohol Us School • Root case as • Intentional act • Balancing power imbalance of power • Personal change: • Communication between women balancing power with partner and men, girls and • Public change: • Decreased risk boys sanctions against behaviors • Change can VAWG • Community activism happen • Acceptability of expanded gender roles Community Critical Thinking Acceptance Sustained action Police, Local Leaders, and Dialogue and influence • Sustained action Health Services, Girl´s • Public debate and • Public debate and • Changed policies Clubs discussion discussion • Organized groups • Personal reflection • Personal reflection • Changed practice in relationships, community, Relationship Participation Participation institutions Relatives, elders, • Activists at • Activists / leaders friends, neighbors grassroots, in / professionals leadership, in increasingly institutions connected and Individual active Women, men, Youth, community activities Rethinking Power Program Evaluation in Southeast Haiti | 55 The Rethinking Power model details the logical progression that of power) and build consensus that it is possible to achieve the SASA!/Power to Girls approach employs to reduce violence. change within a community. It also hopes to promote critical It begins first by detailing some of the most important factors thinking, dialogue, and the development of community activists that need to be considered when designing a program aimed who can facilitate further community level change among men, at reducing violence and providing an overview of how the women, girls and boys. approach seeks to influence these factors at multiple levels. It then describes the initial, intermediate, and long term outcomes The initial outcomes give way to intermediate outcomes that take of the program approach, as well as the final impact RP seeks to longer to develop. This includes building skills on appropriately achieve. responding to women and girls who experience violence, holding men and boys accountable for their actions, and The model begins by noting some of the most important risk promoting a balance of power. In addition, changes in individual factors for VAWG that should be considered when implementing and community behaviors will begin to occur, along with changes the program. These include socio-demographic factors (such as in attitudes on concepts such as power, gender, and human sex, age, income, education, employment, religion, residence, rights. Parallel to these achievements, the program will continue etc.) and socio-cultural factors (such as family characteristics, to strengthen and build its community activist network that drives social support, and alcohol use). For example, education levels these changes within the community. correlate with the likelihood of experiencing VAWG. As such, These intermediate outcomes lead to long-term outcomes, RP may work to increase the safety of girls on school grounds, including increased capacity that support longer-term behavior as well as to advocate for family/community acceptance of change and sustained action to reduce VAWG. At this level, there girls’ education. Each of these interventions may increase girls’ will be improved individual and collective capacity to prevent educational attainment and subsequently reduce their risk of and respond to incidents of violence within the community. experiencing VAWG. Conversely, these factors may provide Behavior change will also continue at the community (e.g., contextual knowledge that help improve program delivery. An increased community activism) and individual levels (reduced risk example of this would be to determine the primary employment behaviors, balanced power, improved partner communications). or income generating activities are used in an area may change In addition, larger societal and community level transformations how the program approaches its activism (such as: where and will take place to provide a more supportive environment for when to hold activities and which key business leaders or other women and girls (e.g., improved policies, transformed institutions informal sector employees to engage). and community groups, etc.) The model then presents a visual of the ecological model Together, these initial, intermediate and longer-term outcomes approach that informs the RP program, describing the different will facilitate a number of expected program impacts. As detailed levels that the methodology seeks to impact. As can be seen in in the above model, these include: 1) reducing the social the model, the RP methodology addresses each of these inter- acceptance of gender inequality, IPV and the sexual abuse of linked drivers of violence by working with community activists girls; 2) decreasing experiences of/perpetration of IPV and the to change gender norms at multiple levels with a variety of sexual abuse of girls; 3) increasing girls’ freedom and feelings of groups. The initial outcomes expected from the program include safety; and 4) decreasing SRH/HIV risk behaviors. increased knowledge of community members (male and female) of the types and consequences of VAWG and the linkages This model will guide program and evaluation design for the RP between violence and HIV. The program also expects to increase program. community awareness of the root cause of VAWG (an imbalance Rethinking Power Program Evaluation in Southeast Haiti | 56 Evaluation Design Data Collection Methodologies GWI will be conducting an impact evaluation of the RP program The evaluation design will employ a mixed method approach in Southeast Haiti. As part of the design process, the research that utilizes both quantitative and qualitative methodologies to team considered both a randomized control trial and quasi- allow for the triangulation of findings by gathering data using experimental design for the evaluation. Because the RP program multiple forms of inquiry, giving more depth and certainty to is a multi-level community based program (including program conclusions made from the data. It will also provide opportunities implementation through media campaigns, schools, churches to collect information on and explore complex issues that are and other services points that draw from multiple communities), not easy to quantify. Finally, a mixed-methods approach will there is strong potential for contamination of control provide additional insights that will increase understanding communities in a RCT design. To address this, the BB program about the ways in which impact was achieved and the factors team considered a program model where large buffer zones and conditions that influenced them. The evaluation will employ between communities were established to minimize potential three complementary components that together will measure contamination. However, the program team did not have the the impact of the RP program. In order to measure overall logistic or financial resources to expand their implementation community-level impact of the RP program – including both zone to a wide enough area that would allow them to the SASA! and Power to Girls methodologies -- a community efficiently implement the RP while simultaneously minimizing population-based household survey will be conducted. See contamination between control and intervention sites. Therefore, below in Component 1 for details on this approach. In order the research team decided that the evaluation will employ a to document the contribution of the Power to Girls program quasi-experimental design that will include both intervention and component of the wider RP program, two smaller data collection comparison communities. However, the comparison communities activities will take place. See details below, under Component 2 will be selected from a different commune (approximately and 3. equivalent to an American county), from the same department (similar to a US state), than the intervention communities – Component 1- Measuring Overall Community Impact: thereby minimizing the potential for contamination. • Quantitative: A repeated cross-sectional population-based household survey with both women and men aged 15-64 will For the quasi-experimental design, BB has selected eight be undertaken at three time points – baseline, midterm, and program implementation areas in LaValle Commune in the endline – in both intervention and comparison communities. In Southeast Department that will receive the full RP program. addition, qualitative data will be collected to complement the Eight comparison communities will be selected from another data collected via these surveys. commune that has similar characteristics, such as size, population • Qualitative: Participatory focus group discussions (FGDs) density, access to nearby cities, etc. to the intervention with women and men in the community and semi-structured communities. Data will be collected in all sites – implementation key informant interviews with stakeholders, such as service and comparison – before, during and after the program (baseline, providers, police, community leaders, representatives of midterm, and endline) to track progress. Any changes over women’s organizations, community activists, etc. time among participants in the RP program communities will be compared against changes within the comparison group to Component 2 – Understanding the effect of RP on Girl’s Club establish the impact of the RP program. Participants: • Quantitative: In order to better understand the program’s effect The evaluation of the disability component is part of the whole on girl’s participating in RP’s girl’s groups, a cross-sectional Rethinking Power program impact evaluation. In this case, the survey will be employed with participants in the clubs. This baseline of the disabilitycomponent will be part of the midline survey will be implemented at both baseline and endline of the whole RP program and the end line will be part of the (repeated cross-sectional surveys). endline of the entire RP program. Further, no midline is going to be conducted. Changes, at least in attitudes and norms around • Qualitative: Participatory FGDs will be undertaken with girl’s women and girls with disabilities, are expected to happen in club participants to better understand the effect of this shorter time due to all the previous work of sensitization that BB intervention on their lives. Key informant interviews will also be has implemented with the communities around issues about undertaken with Girl’s Club leaders. equality. Rethinking Power Program Evaluation in Southeast Haiti | 57 Component 3 – Understanding the effect of RP on School- Research Tools based Participants: Specifically, the data collection tools that the evaluation will utilize • Quantitative: The second component specifically designed to are detailed below. better understand the effect of the Power to Girls methodology is a self-administered questionnaire with boys and girls who Quantitative: are enrolled in schools where the Power to Girls curriculum is Component 1 & 4 being employed. This will also occur at baseline and endline • Community-based Survey: A cross-sectional community survey (repeated cross-sectional surveys). will be employed at baseline, midterm and endline of the • Qualitative: Participatory FGDs will be undertaken with school program. This questionnaire has been developed based on age girls and boys as well as (separately) teachers and school the tools used in the original SASA! RCT study undertaken by stakeholders to understand the effect of this intervention on the London School of Tropical Hygiene and Medicine. The their lives. Key informant interviews will be undertaken with survey will measure community members’ (both men and school administrators and other school-based stakeholders. women - aged 15-64) knowledge, attitudes, and behaviors on key indicators related to VAWG. Survey questions will be Component 4 – Understanding the effect of RP on Women drawn from indicators related to both SASA! and Power to and Girls with Disabilities: Girls. Overall knowledge, attitudes, skills, and behaviors will be • Quantitative: This will be the same cross-sectional population- measured for the community-at-large, but with a great enough based household survey with both women and men aged sample size to understand the experiences of violence of 15-64 of Component 1 for the midterm and endline – in both women and girls. Some questions about the mental and both intervention and comparison communities. Questions physical health of the participants and about attitudes around about the health of participants to measure the situation disabilities will also be included as explained in the previous of any physical or mental disabilities as well as questions section. about attitudes of stigma and knowledge of rights to people Component 2 with disabilities will be included to the survey tool that was • Girl’s Club Survey: As part of Component 2, a separate smaller used during the baseline. The questions about the health of survey will be employed with girls (aged 10-19) who participate participants will be based on the Washington Group Scale and in RP girl’s clubs. This survey will specifically focus on changes the Kessler Scale, which have been largely validated in different in knowledge, attitudes, and behaviors related to the Power to studies, including the WHO Multi-Country Study on VAWG. The Girls methodology (e.g., empowerment, self-esteem, gender questions about attitudes of stigma and knowledge of rights norms, etc.). Participants will not be asked about their own will be based on several validated scales, covering aspects experiences of violence in the home or community. of knowledge on rights, discomfort around people with Component 3 disabilities, interaction with people with disabilities, sensitivity, • School-based Survey: As part of Component 3, a self- and knowledge of cause. administered survey will be employed with girls and boys • Qualitative: This will be the same participatory focus group (aged 10-19) enrolled in schools using the Power to Girls discussions (FGDs) with women and men in the community that curriculum. This will be similar to the survey used with girl’s cover issues around VAWG and are mentioned in Component clubs and specifically used to measure knowledge, attitudes, 1. To the FGD tool, some questions that will cover the area of and behaviors of this sub-group, specifically related to the norms around people with disabilities and violence against newly created Power to Girls methodology. Participants will not them will be added to the already existing tools of Component be asked about their own experiences of violence in the home 1. In addition, some FGDs with people with disabilities and with or community. people representing the rights of these groups will be added. A new tool will be developed for this purpose. Rethinking Power Program Evaluation in Southeast Haiti | 58 Qualitative: Sample Size Components 1, 2 3, & 4 Quantitative: • Focus Groups Discussions: Focus group discussions will Component 1 – Community-based Survey: The goal of the study concentrate on three specific groups. First, general community is to assess the effectiveness of the RP program in preventing focus groups (targeting men and women separately) will VAWG. In order to achieve this, the primary outcome of interest be held to validate and contextualize the findings of the in which the sampling strategy is based is a reduction in IPV in community based survey. Second, school and girl focus groups a 12 month period. Based on available national statistics and (engaging girls as well as key stakeholders) will gather further knowledge of the area, the research team expects that the true information on the experience of girls, particularly in relation prevalence of physical IPV in the program area is 20% in the past to their experience with the Power to Girls. Third, focus groups 12 months. Over the course of the program, it is anticipated that with parents will deepen and contextualize our understanding this rate will be reduced to at least 14%. Therefore, the sample of any change in their attitudes or behaviors related to size has been calculated to ensure each arm of the study has girls. These groups will utilize interactive and participatory sufficient power to detect this change. Based on this, an overall methods, such as free-listing, open-ended stories, role playing, total sample of at least 615 completed surveys per arm has been community mapping, etc. These discussions should take no determined. After adjusting due to the expected intra-cluster longer than 2 hours and the draft tools can be found in Annex correlation and expected non-response, the final sample will be 2. An additional FGD will include people with disabilities and/ 1,500 (1,000 women and 500 men) households per arm – for a or groups that work on the rights of people with disabilities. total of 3,000 households overall in the survey. • Key Informant Interviews: Key informant interviews will be held with women, girls, community activists, school administrators, Component 2 – Girl’s Club Survey: In order to provide teachers, community leaders, and service providers. This data complementary information to the main community-based will help contextualize and triangulate the findings from the survey, a smaller survey of girl’s clubs participants will be community-based and stakeholder surveys. Semi-structured undertaken. The program expects to target approximately 320 interview guides have been developed to give a general girls (16 clubs with approximately 20 participants per group – framework for the interviews and include opening questions aged 10-19). Given the limited number of girls enrolled in girl’s that will help guide the conversation towards answering the clubs, the research team will attempt to survey all participants for research questions, but still allow flexibility in the conversation. an estimated sample of 320. Component 3 - School-based Survey: In addition, a self- administered questionnaire will be given to a sample of participants in 3 schools in the intervention area where the RP program is being employed, as well as 3 schools from the comparsion community (matched on characteristics, such as size, student composition, state/private, etc). The average number of pupils in schools in the area is 200 students. The research team will randomly select 3 classes within each school to adminster the questionnaire – for a total sample size of 120 per school. The questionnaire will be administered in 3 schools per arm for a total of 720 questionnaires across the entire study (360 per arm). Qualitative: Components 1, 2 3, & 4: For all qualitative data collection, purposeful sampling will be used to ensure a wide breath of knowledge and experience. Additional respondents may be found through snowball sampling if required. It is expected that the focus groups will include specific groups of women, men, girls and boys. Key informant interviews will focus on stakeholders, including community activists, teachers, girl’s clubs leaders, community leaders, and service providers. Rethinking Power Program Evaluation in Southeast Haiti | 59 Data Analysis The following principles will also be adhered to throughout the evaluation process: 1) engage women and girls who reflect the Quantitative diversity of the primary beneficiaries; 2) engage community Components 1, 2 3, & 4: The quantitative data from the study will members and researchers in a joint process in which each will be analyzed using descriptive statistics as well as bivariate and contribute equally; 3) facilitate a co-learning process; 4) involve multivariate statistical methods. Descriptive statistics will be used systems development and local capacity building; 5) facilitate to present the data on social norms, VAWG, and exposure to the an empowering process that validates participants’ experiences, program. Bivariate (t-tests, chi-square) and multivariate regression ideas, and opinions and through which they can increase control will be used to compare the situation in the intervention and over their lives; and 6) achieve a balance between research and comparison communities at baseline, midterm, and endline. For action.14 This approach will ensure that the data and findings are each endpoint our main analytic model will use a difference-in- relevant and useful not only for the program evaluation, but also difference specification: for local stakeholders and program managers.15 Yij = 0 + 1Time1ij + 2Time2ij + 3Treatj + 4Treatj*Time1ij + The evaluation will also apply a gender approach to achieve 5Treatj*Time2ij + Xij + uj + eij the ultimate goal of the program: to transform the underlying roots of gender inequality. Following Patton16, this evaluation Where i indexes the individual and j indexes the community; will have five components that characterize a broadly defined Time1 and Time 2 are dummy indicators of midline and endline gender approach. These include: 1) a central focus on gender (baseline serves as the reference); Treat is a dummy indicator of inequalities; 2) the conceptualization of inequality based on being an intervention as opposed to control community; X is a gender as systemic and structural; 3) the recognition that vector of individual-level control variables, including age, marital information and knowledge are powerful resources; 4) the status, parity, and household wealth; and uj is a cummunity fixed acknowledgement that the evaluator is not “neutral” but brings effect. The coefficients 4 and 5 measure the extent to which specific experiences, sensitivities, awareness, and perspectives; the expected value of Y increased more in intervention than in and 5) the acknowledgement that evaluation is not merely a control communities between baseline and midline and between technical activity but is political. baseline and follow-up. Qualitative Research Uptake Components 1, 2, 3 & 4: Interviews and FGDs from the The aim of the research uptake strategy for this study is to ensure qualitative study will be captured by note takers, translated and the translation of research to action. To ensure that the results transcribed. The data analysis will be carried out by GWI using of this evaluation inform not only the RP model, but also future the software Atlas.ti. GWI researchers will use a combination of a adaptation and implementation of community mobilization priori and grounded theory to develop and assign codes to the interventions, GWI will consult with BB, the TAG, and other data. The information gathered through the interviews and FGDs stakeholders and end users of our research to develop effective will allow the researchers to further document the situation of communications and dissemination activities. Such activities can women and girls in Southeast Haiti as well as the impact of the RP include online communications and media, conferences, working program. groups, and mass media campaigns. Participatory and Gender Approach Adequate attention must be given to disseminating results at the local, national, and international levels. This process will include The evaluation will apply a participatory approach that will initial reports of baseline results, regular six-month reports of the involve beneficiaries – women and girls in the intervention process, and a final report, which includes strategies for scaling communities, in particular – in the design and implementation of up and adaptation to other settings. Products could also include the research, as well as in the interpretation and dissemination articles in peer-reviewed journals, policy notes, working papers, of findings. GWI, BB, and local partners have been working and other informative documents disseminated both in print closely on the development of the research protocol and data form and electronically. Presentations (workshops, conferences, collection tools, both to ensure that the design is relevant in the seminars, etc.) at the national and international levels should local context and to strengthen local capacity. Local stakeholders, also be considered. These may include presentations at the community-based organizations, and local authorities will be Commission on the Status of Women, appropriate global involved in the planning and implementation stages through the seminars and international events on prevention, community creation of a TAG, which will be invited to provide input during safety, women’s rights, children’s rights, and urban planning, each stage of the process. governance and safety, as well as through regional events organized by local networks working in VAWG and public safety. 14 Ellsberg, M; Heise, L. (2005). Researching Violence against Women: A Practical Guide for Researchers and Activists. Washington, D.C.: PATH, WHO. 15 An example of how we have put these principles into action is found in the following article: Ellsberg, M. et al. (2009). Using Participatory Methods for Researching Violence Against Women: An experience from Melanesia and East Timor. The article is attached to this proposal. Rethinking 16 Patton, M. (2002). Qualitative Research and Evaluation Methods. Thousand Oaks, California: Power Program Evaluation in Southeast Haiti | 60 SAGE Publications.