92659 Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study MARCH 2014 LOGiCA Study Series No.2 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved First published March 2014 www.logica-wb.net Thiis Study Series disseminates the findings of work in progress to encourage discussion and exchange of ideas on gender and conflict related issues in Sub-Saharan Africa. Papers in this series are not formal publications of the World Bank. The papers carry the names of the authors and should be cited accordingly. The series is edited by the Learning on Gender and Conflict in Africa (LOGiCA) Program of the World Bank within the Fragile States, Conflict and Social Development Department. This paper has not undergone the review accorded to official World Bank publications. The findings, interpreta- tions and conclusions herein are those of the author and do not necessarily reflect the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, its Executive Directors, or the governments they represent. To request further information on the series, please contact ereessmith@worldbank.org LOGiCA. Papers are also avail- able on the LOGiCA website: www.logica-wb.net. Cover and layout design: Duina Reyes Photos provided by World Bank photo library and United Nations photo library Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study LOGiCA Study Series No.2 March 2014 Jhumka Gupta Kathryn L Falb Heidi Lehmann Denise Kpebo Ziming Xuan Mazeda Hossain Cathy Zimmerman Charlotte Watts Jeannie Annan Jhumka Gupta1,2,* Mazeda Hossain6 Email: jhumka.gupta@yale.edu Email: mazeda.hossain@lshtm.ac.uk Kathryn L Falb1,2 Cathy Zimmerman6 Email: kathryn.falb@yale.edu Email: cathy.zimmerman@lshtm.ac.uk Heidi Lehmann3 Charlotte Watts6 Email: heidi.lehmann@rescue.org Email: charlotte.watts@lshtm.ac.uk Denise Kpebo4 Jeannie Annan3 Email: dkpebo@poverty-action.org Email: jeannie.annan@rescue.org Ziming Xuan5 Email: xuanziming@gmail.com 1 Department of Chronic Disease Epidemiology, Division of Social and Behavioral Sciences, Yale School of Public Health, New Haven, CT 06520, USA 2 Center for Interdisciplinary Research on AIDS, Yale University, New Haven, CT, USA 3 International Rescue Committee, New York, NY, USA 4 Innovations for Poverty Action, Abidjan, Côte d’Ivoire 5 Community Health Sciences, Boston University School of Public Health, Boston, MA, USA 6 Department of Global Health & Development, London School of Hygiene & Tropical Medicine, London, UK * Corresponding author. Department of Chronic Disease Epidemiology, Division of Social and Behavioral Sciences, Yale School of Public Health, New Haven, CT 06520, USA Table of Contents Abstract....................................................................................................................... 3 Background.................................................................................................................... 3 Methods.......................................................................................................................... 3 Results............................................................................................................................. 3 Conclusions.................................................................................................................... 3 Trial registration............................................................................................................. 3 Keywords.................................................................................................................... 4 Background................................................................................................................ 4 Methods...................................................................................................................... 5 Design, setting, and participants.................................................................................... 5 Participants..................................................................................................................... 5 Data collection................................................................................................................ 6 Measures......................................................................................................................... 6 Analysis........................................................................................................................... 7 Results........................................................................................................................ 9 Discussion.................................................................................................................. 9 Conclusions..............................................................................................................14 Competing interests.................................................................................................14 Author’s contributions.............................................................................................14 Acknowledgments....................................................................................................14 References................................................................................................................16 Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 1 2 Abstract Background reporting past year physical and/or sexual IPV (OR: 0.92; 95% CI: 0.58, 1.47; not statistically sig- Gender-based violence against women, including in- nificant). Reductions in reporting of physical IPV timate partner violence (IPV), is a pervasive health and sexual IPV were also observed (not statistically and human rights concern. However, relatively little significant). Women in the treatment group were intervention research has been conducted on how significantly less likely to report economic abuse to reduce IPV in settings impacted by conflict. The than control group counterparts (OR = 0.39; 95% current study reports on the evaluation of the incre- CI: 0.25, 0.60, p < .0001). Acceptance of wife beat- mental impact of adding “gender dialogue groups” ing was significantly reduced among the treatment to an economic empowerment group savings pro- group (β = −0.97; 95%CI: -1.67, -0.28, p = 0.006), gram on levels of IPV. This study took place in north while attitudes towards refusal of sex did not sig- and northwestern rural Côte d’Ivoire. nificantly change Per protocol analysis suggests that compared to control women, treatment women at- Methods tending more than 75% of intervention sessions Between 2010 and 2012, we conducted a two- with their male partner were less likely to report armed, non-blinded randomized-controlled trial physical IPV (a OR: 0.45; 95% CI: 0.21, 0.94; p = (RCT) comparing t group savings only (control) .04) and report fewer justifications for wife beating to “gender dialogue groups” added to group savings (adjusted β = −1.14; 95%CI: -2.01, -0.28, p = 0.01) (treatment). The gender dialogue group consisted ; and both low and high adherent women reported of eight sessions that targeted women and their male significantly decreased economic abuse (a OR: 0.31; partner. Eligible Ivorian women (18+ years, no prior 95% CI: 0.18, 0.52, p < 0.0001; a OR: 0.47; 95% CI: experience with group savings) were invited to par- 0.27, 0.81, p = 01, respectively) . No significant re- ticipate. 934 out of 981 (95.2%) partnered women ductions were observed for physical and/or sexual completed baseline and endline data collection. The IPV, or sexual IPV alone. primary trial outcome measure was an overall mea- Conclusions sure of past-year physical and/or sexual IPV. Past year physical IPV, sexual IPV, and economic abuse Results from this pilot RCT suggest the importance were also separately assessed, as were attitudes to- of addressing household gender inequities alongside wards justification of wife beating and a woman’s economic programming, because this type of com- ability to refuse sex with her husband. bined intervention has potential to reduce levels of IPV. Additional large-scale intervention research is Results needed to replicate these findings. Intent to treat analyses revealed that compared to Trial registration groups savings alone, the addition of gender dia- logue groups resulted in a slightly lower odds of Registration Number: NCT01629472. Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 3 Keywords en become more financially empowered and more willing to challenge household gender norms [11]. Gender-based violence, Randomized controlled Past evaluation studies of economic empowerment trial, West Africa, Economic Empowerment, Evalu- programs have yielded somewhat conflicting results, ation with programs pointing to either increased protec- tion from IPV or increased risk of violence among women participants in economic empowerment Background programs [12]. In response, efforts to combine both economic empowerment and gender equity have There is a paucity of evidence on effective strategies been recommended to reduce IPV [13,14]. The IM- to reduce intimate partner violence (IPV) against AGE study in rural South Africa was the first trial women in conflict-affected settings. In such contexts, to evaluate the impact on violence of a combined the prevention of gender-based violence (GBV) has micro-credit and participatory gender training in- not been a policy priority, and the few research and tervention. It demonstrated a 55% reduction in lev- programmatic efforts pertaining to GBV have pri- els of physical and sexual intimate partner violence, marily focused on sexual violence perpetrated by as well as reductions in levels of household poverty armed groups (i.e. rape as a weapon of war ) without and improved HIV communication [15,16]. These adequate attention to violence by intimate partners findings indicated that delivering interventions to [1]. However, recent research suggests that IPV may address a combination of structural factors (i.e. pov- be of greater prevalence than war-related violence erty and broader social norms) considered to enable victimization [1-3]. Moreover, programmatic data and sustain IPV may be essential components of from the International Rescue Committee (IRC), an programs seeking to empower communities, change international humanitarian organization dedicated behaviors and improve women’s safety [17]. to addressing GBV in conflict-affected regions, show that 63% of West African women assisted by IRC for As the knowledge-base on IPV programming in violence, sought help for violence committed by an low and middle income countries begins to grow, intimate partner [4]. Evidence-based approaches to intervention research is also needed on IPV within reduce IPV victimization among conflict-affected conflict-affected settings. However, to date, there is women are critical given women’s particular vulner- limited understanding of the effectiveness of these ability to IPV, [5] the potential long-term physical socioeconomic interventions within conflict-affect- and mental health sequelae, which have been well- ed communities where other structural factors, in- documented, [6,7] and the likely hindrances IPV cluding disruption of economic systems and liveli- poses to achieving the Millennium Development hoods, may play critical roles in women’s well-being Goals [8]. and impact programmatic outcomes [18,19]. To reduce IPV and mitigate its deleterious health, Côte d’Ivoire, once known as the ‘jewel of West economic, and social risks, economic empowerment Africa’ due to its relative economic stability in this strategies (e.g. group savings, livelihood efforts or precarious region, was affected by widespread con- microfinance) aiming to enable women to generate flict from 2002–2004, and again in 2010–2011. Like and save money have received substantial attention in other West African countries [20,21] IPV levels are development and health sectors. However, these ap- high. Regional estimates indicate that as many as proaches have been critiqued for their narrow focus 47.5% of women report past year IPV,[22] and a on altering economic structures without addressing community-based survey found that some 60% of the larger gender norms that perpetuate gender in- Ivorian women reported experiencing lifetime IPV equalities and IPV [9,10]. Concerns are commonly [23]. Both of these figures fall within the higher end voiced about microcredit programs that are imple- of global estimates of lifetime IPV, which range from mented in the absence of any broader attempts to 15-71% [24]. change the gendered views of male partners and the potential for increases in IPV, particularly as wom- The objective of the current study, a two-armed 4 randomized controlled trial (RCT) in rural Côte that would be randomized to receive the treatment. d’Ivoire, was to evaluate the incremental impact on Groups not randomly drawn during the lottery were levels of IPV of adding “Gender Dialogue Groups” told that they would receive the gender dialogue for women and their partners (aiming to change group upon completion of the study. An endline gender norms) to an economic empowerment pro- survey was conducted from July to August 2012. gram for women. Ethical approval was obtained for all study protocols through the Yale University Human Subjects Com- mittee (#1007007040) and Innovations for Poverty Methods Action (506.11September-003) Human Subjects Committee. Local, Côte d’Ivoire-based approval Design, setting, and participants was obtained by leadership committees of all par- Our study, Reduction of Gender-Based Violence ticipating villages. Against Women in Côte d’Ivoire, is a two-armed pi- Participants lot RCT implemented between October 2010 and August 2012 in north and northwestern rural Côte Eligible women were 18 and over and had no prior d’Ivoire. The study was led by Yale School of Public participation in group savings programs. Both part- Health (YSPH) in partnership with Innovations for nered (e.g. married or in a relationship with a male Poverty Action (IPA) and IRC. for at least 1 year) and non-partnered (e.g. single, divorced, widowed) women were eligible to partici- Thirty rural villages were selected for inclusion into pate in the IRC program to preserve community so- the trial based on their history of not having previ- cial cohesion. However, non-partnered women were ous experience with economic empowerment pro- not considered for the IPV analytic sample. Given gramming and their status as being a priority area for that the gender dialogue groups’ potential impact on intervention by IRC, the implementing agency. Six IPV had not been evaluated in prior work at the time villages were excluded due to challenges with mobi- the study was being planned, effect estimates were lizing village leaders and participants, thus yielding largely unavailable. However, power calculations a final set of 24 villages. The IRC Côte d’Ivoire field were conducted based on the expected number of staff met with village leaders and eligible women to women who would be eligible for the analytic sam- introduce the program and study. Women and village ple per IRC projections. Power calculations revealed leaders were told that all women would receive the that the minimum detectable difference would be economic empowerment program at the same time, 13-16% at a significance level of .05 and 80% power, while half of the groups would receive an additional with a total minimum sample of 1008 eligible wom- discussion group at an earlier time point than oth- en. In total 1,271 women completed the baseline ers (ie., a waitlist control). Women were then placed survey (96% response rate), of which 981 (77.2%) into 47 groups of 15–30 women. A baseline survey were partnered and thus eligible for the study. Of was subsequently conducted, in October 2010. All these 981 women, 934 (513 treatment; 421 con- groups began economic empowerment program- trol) were also included in the follow-up (95.2%); ming activities (control) in December 2010. How- thus yielding the final analytic sample (CONSORT, ever, due to post-election violence that occurred af- Figure 1). Women with no children were more likely ter the baseline survey, randomization to receive the to have both missing data and drop out of the inter- Gender Dialogue Group (treatment) in addition to vention; no other missingness or attrition varied by ongoing economic empowerment activities versus demographics or baseline IPV outcomes. Control continuing with economic activities only (control) group participants were significantly more likely to was delayed until September 2011. To preserve so- drop out of the program; IRC administrative records cial cohesion within villages, random assignment revealed that financial issues/lack of confidence in was done via public lottery. IRC staff held a public group savings activities was more frequently cited event in each participating village where each village by control women as reasons for leaving the pro- chief drew the names of groups within each village gram (no statistical testing conducted). Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 5 Table 1. Description of intervention components Economic Empowerment Program: Village Savings Gender Dialogue Groups (GDG) Aiming to Change and Loans Associations (VSLA) Gender Norms The VLSAs provide women with a local, safe, and Participants and their male partners (or male convenient place to save money, access small family member if the participant does not have a loans, and a critical safety net in the form of an male partner) were randomized to receive GDG “emergency fund or social fund”. The VSLA model is or wait-listed until after the study was completed. simple and practical. A group of 15-30 individuals GDGs create an opportunity for bringing together decide to save money together and contribute to VSLA members and their spouses to reflect on a shared fund weekly. Individual members borrow their financial decisions and goals, the value from this common fund and pay the loan back of women in the household, and alternatives at a modest interest rate, helping the fund grow to violence. While the overall focus of GDGs is over time. The group agrees on a pay-out date household financial well-being, each session is (generally 8-12 months after savings begins). At designed to raise underlying issues that condone this time, each member receives their accumulated IPV and challenge participants to equalize balance savings plus a percentage return on their savings. of power between themselves and their spouses. Managed appropriately, VSLAs provide affordable These discussions in turn provide an opportunity credit for borrowers and interest rates for savers that to promote women’s participation in household typically exceed those that any formal institution decision-making and encourage a shift towards could provide. VSLAs employ participant-driven more equitable spousal power relations. This management which fosters sustainability, and also approach was developed by IRC and was first make this form of savings more feasible in rural piloted in Burundi in 2009 [28] Groups met every regions (including conflict affected settings) lacking other week and took place between December other formal finance institutions or where the 2011 and April 2012. income level of women would not allow access to financial institutions [27]. Source: author; survey data. Note: *This figure includes ex-combatants only, as all the members of the community answered “a little” to this question. Data collection year physical and/or sexual intimate partner violence reported by women. For this we used the items from Trained local female research staff were matched the WHO Multi-Country study on Women’s Health to participants based on language and ethnicity. In and Domestic Violence [30]. In this instrument, private locations, they completed verbal informed and as is best practice in violence research, [30] consent with participants, verbally administered women are asked explicit questions about whether paper-based surveys and recorded respondents’ re- they have experienced different acts of violence. The sponses. Survey interviews were conducted in line past year prevalence of physical partner violence with WHO ethical and safety guidelines for research was assessed via whether a woman reported that on IPV [29]. Surveys were translated into French her partner had slapped her or thrown something and back-translated into English. Research staff ver- at her that could hurt her; pushed, shoved, kicked bally translated surveys and informed consent into or dragged her; choked her or burned her intention- eleven local languages for women, as necessary. A ally; threatened to use a gun, knife or other weapon list of local medical, legal, and psychosocial support against her; and used a gun, knife or other weapon services available for referral services was given to against her in the past year. Sexual intimate partner participants upon survey completion. violence was assessed through whether women, in Measures the past year, had been forced to have sex because of threats or intimidation or physically forced to have The main outcome measure of the study was past- sex by their intimate partner when she did not want 6 to. An affirmative response to any item was coded as [23] which was used in a complementary evaluation experiencing past-year physical and/or sexual IPV study in Côte d’Ivoire. in the final outcome variable; while “no” to all items was coded as no IPV. Analysis Secondary outcomes included: 1) any past-year Data were double-entered into a Microsoft Access physical IPV violence; 2) any past year sexual IPV database [32]. The distribution of baseline socio- violence; 3) any past year economic abuse from demographic variables were compared using Pear- their partner; and 4) gender norms. Past-year physi- son chi-squared or two sample t-tests by treatment cal IPV and sexual IPV were derived from the sum- status to ascertain the results of randomization; no mary measure of any physical and/or sexual IPV as significant differences emerged (Table 2). described above. The secondary outcomes examine To address clustering inherent in the data (baseline past-year sexual IPV and physical IPV indepen- and endline outcomes were repeated measures nest- dently, rather than in a summary measure of past- ed within individuals that were nested within groups year physical and/or sexual IPV. Economic abuse which were nested within villages), multilevel analy- was measured through the following three items: if sis was used to model changes in IPV by treatment in the past-year the partner: (1) had taken money status. Specifically, 4-level random intercepts mod- against her will; (2) refused money for household els were used to evaluate the significance of the in- necessities; or (3) obliged the woman to give him all teraction term treatment status X time (e.g. baseline or part of the money she earned [31]. These second- vs. endline) while using random effects to adjust for ary outcomes were operationalized as a ‘yes to any’ / both autocorrelation between the two time points ‘no to all’ summary measure. Gender norms were as- within individuals as well as clustering among in- sessed continuously via fourteen adapted items that dividuals nested in groups nested in villages. The asked if a husband was justified in beating his wife generalized mixed model in GLIMMIX procedure in different scenarios (Cronbach’s α = 0.91) [30]. in SAS v9.2 [33] was used to fit the multilevel mod- Scenarios included if she [the wife] disobeys him el. A significant interaction term (time X treatment [the husband], he suspects she was unfaithful, finds status) was indicative of statistically significant dif- out she was unfaithful, she gossips with the neigh- ferential effects of the treatment status on changes bors instead of taking care of children, she does not in outcomes from baseline to end-line. Odds ratios prepare the meals on time, she refuses to have sex (OR), 95% confidence intervals (CI), and p-values with him, she does not complete her housework to (at the p < .05 level) were calculated to assess signifi- his satisfaction, she neglects the children, she argues cance for models with binary outcomes. Betas (β), with him, she burns the food, she argues with her 95% CI’s, and p-values (at the p < .05 level) were in-laws, she disobeys her in-laws, she does not com- computed for linear GLIMMIX models assessing plete her household work to her in-laws’ satisfaction, continuous outcomes. and she cannot have children. A second adapted scale, measured continuously, assessed if a woman To examine the intervention effects, two types of had the right to refuse sex through eight items (if she analyses were conducted: intent to treat (ITT), and did not want to, he is drunk, she is sick, he mistreats secondarily, per-protocol (PP). No covariates were her, she suspects he has been unfaithful, she knows included in the models for ITT analysis as random- he has been unfaithful, he refuses to use condoms, ization was successful. PP analysis utilized a 3-level she has pelvic or menstrual pain) (Cronbach’s α = intervention variable: VSLA only (comparison/ 0.68) [31]. For both of the above scales, women referent); VSLA + GDG low adherent (where both were asked if they agreed or disagreed, and were as- women and their partners attended less than 75% of signed one or zero points based on their response. sessions; VSLA + GDG high adherent (where both Women’s scores were then summed for the continu- women and their partners attended at least 75% of ous measures. The study instrument was adapted sessions). As the intervention was intended to target from a questionnaire developed by researchers at the couples, adherence was determined based on the at- London School of Health and Tropical Medicine, tendance of both women and their male partner. In Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 7 Table 2. Baseline characteristics of study sample, by treatment arm (N = 934) VSLA Only VSLA Plus GDG Overall N = 934 a p-value b (Control n = 421) a (Treatment n = 513) a Age in years 37.7 (s.d. = 11.5) 37.7 (s.d. =12.1) 37.7 (s.d. = 10.9) 0.96 Marital Status Married 767 (82.1) 342 (81.2) 425 (82.9) Not married 167 (17.9) 79 (18.8) 88 (17.2) Lives with partner 124 (13.3) 61 (14.5) 63 (12.3) 0.58 Does not live with partner 43 (4.6) 18 (4.3) 25 (4.9) Women’s Occupation Farmer Only 145 (15.5) 61 (14.5) 84 (16.4) Small business owner only 425 (45.5) 194 (46.1) 231 (45.0) 0.08 Farmer and small business owner 308 (33.0) 132 (31.4) 176 (34.3) Other 56 (6.0) 34 (8.1) 22 (4.3) Ethnicity Yacouba 585 (62.6) 275 (65.3) 310 (60.4) Senoufo, Dioula, or Guere 140 (15.0) 63 (15.0) 77 (15.0) 0.19 Other 209 (22.4) 83 (19.7) 126 (24.6) Education c None 657 (70.6) 288 (68.7) 369 (72.1) Primary 212 (22.8) 97 (23.2) 115 (21.5) 0.24 Secondary and above 62 (6.7) 34 (8.1) 28 (5.5) Religion c Christian 409 (44.3) 176 (42.3) 233 (46.0) Muslim 139 (15.1) 73 (17.6) 66 (13.0) 0.19 Traditional 161 (17.4) 67 (16.1) 94 (18.5) Other / None 214 (23.2) 100 (24.0) 114 (22.5) Number of pregnancies 0 29 (3.1) 9 (2.1) 20 (3.9) 1-3 220 (23.6) 111 (26.4) 109 (21.3) 0.07 ≥4 685 (73.3) 301 (71.5) 384 (74.9) Partner’s Occupation c Farming 729 (79.8) 318 (77.2) 411 (81.9) 0.08 Non-Farming 185 (20.2) 94 (22.8) 91 (18.1) a Column percentages. b P-value presented from chi-square or t-tests, where appropriate. c Numbers do not add to 934 due to missing data. 8 the PP analysis, number of pregnancies and religion p = 01; a OR: 0.31; 95% CI: 0.18, 0.52, p < 0.0001, was adjusted for, as they were the only two variables respectively). Women in high adherent couples were statistically associated with adherence to protocol. also significantly more likely to report reduced justi- fication of wife beating (aOR: -1.14; 95%CI: -2.01, -0.28), while the reduction in wife beating justifica- Results tions for women in low adherent couples was not Demographics of participants are presented in Ta- significant (aOR: -0.19; 95%CI: -1.13, 0.74). Atti- ble 2 for the overall sample and by treatment group. tudes toward women’s ability to refuse sex did not No statistically significant demographic differences statistically change in either adherent group. were found between treatment arms. Table 3 indicates frequencies of different forms of Discussion IPV (physical and/or sexual; physical; sexual; and In this randomized intervention study with rural economic) at baseline and endline in both arms. In Ivorian women, the addition of a dialogue compo- ITT analysis, while not reaching statistical signifi- nent explicitly addressing gendered social inequali- cance, the odds of reporting physical and/or sexual ties and norms alongside economic empowerment IPV in the past year was lower in the VSLA + GDG programming significantly reduced past year physi- in comparison to the referent (OR: 0.92; 95% CI: cal IPV among women who participated in more 0.58, 1.47, p = .72). Reductions in the likelihood than 75% of the program with their male partner. of reporting of physical IPV and sexual IPV were The combined intervention also significantly re- also observed in the treatment vs. control; although the decreases did not reach statistical significance. duced economic abuse and altered attitudes regard- VSLA + GDG women were significantly less likely ing the justification and acceptance of IPV among to report economic abuse than VSLA-only counter- all women in the study. Encouragingly, reductions parts (OR = 0.39; 95% CI: 0.25, 0.60, p < 0.0001). in IPV were also observed in ITT analyses, though Acceptance of justification towards wife beating was decreases were not significant and the effect size for significantly reduced in the VSLA + GDG group (β the overall physical and/or sexual IPV was small. = −0.97; 95%CI: -1.67, -0.28, p = 0.006), while at- Importantly, this research demonstrates that IPV titudes towards the ability of a woman to refuse sex reduction programs can be rigorously evaluated in did not significantly change. a conflict-affected setting—even in the midst of a period of heightened post-election violence, and, In total, 234, or 46% of both women and men at- moreover, that it is possible to observe reductions tended 75% of sessions. In PP analysis (Table 4), in in this under-addressed yet very prevalent form of comparison to VSLA only, women part of a high ad- violence in such challenging and unstable contexts. herent VSLA + GDG couple were significantly less likely to report physical IPV (aOR: 0.45; 95% CI: The current RCT findings are broadly consistent 0.21, 0.94; p = .04). In comparison to VSLA only with the one prior South Africa-based study (the women, women who were part of a high adherent IMAGE intervention) that examined the impact VSLA + GDG couple were also less likely to report of combining gender equity components with eco- the summary measure of physical and/or sexual IPV nomic empowerment programming on IPV, [11,16] and sexual IPV, although these reductions were not offering further strength to the evidence that the in- statistically significant. Women who were part of clusion of gender training to socio-economic pro- low adherent couples had reduced odds of report- gramming can offer social and health benefits, and ing physical IPV or sexual IPV and slightly increased also extending such findings to a conflict-impacted odds of physical and/or sexual IPV, although no out- region. However, it should be noted that the cur- come was statistically significant. Both high and low rent intervention possessed unique features that in- adherent women were significantly less likely to re- cluded activities that specifically involved male part- port economic abuse in comparison to their VSLA- ners and the use of a financial lens with couples to only counterparts (a OR: 0.47; 95% CI: 0.27, 0.81, talk about gendered power dynamics in the home. Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 9 Table 3. Distribution of study outcomes at baseline and endline, by treatment group and unadjusted effect estimates of past-year intimate partner violence (Intent to Treat Analysis) (N = 934) Baseline N Endline N OR† Treatment type p-value (%) (%) (95% CI) VSLA Only Physical and/or Sexual IPV 93 (22.1) 78 (21.0) -- (Comparison) a VSLA + GDG b 119 (23.2) 100 (20.7) 0.92 (0.58, 1.47) i 0.72 VSLA Only Physical IPV 65 (15.4) 55 (14.8) -- (Comparison) a VSLA + GDG b 80 (15.6) 53 (11.0) 0.69 (0.39, 1.21) i 0.19 VSLA Only Sexual IPV 44 (10.5) 53 (14.3) -- (Comparison) a VSLA + GDG b 71 (13.8) 68 (14.1) 0.71 (0.40, 1.25) i 0.24 VSLA Only Economic Abuse 113 (27.4) 128 (34.6) -- (Comparison) c VSLA + GDG d 163 (32.5) 104 (21.5) 0.39 (0.25, 0.60) j <0.0001 Mean (SD) Mean (SD) β (95% CI) p-value VSLA Only Justification for wife beating 4.5 (4.3) 4.0 (4.0) −0.97 (−1.66, -0.28)k 0.006 (Comparison) e VLSA + GDG f 4.9 (4.4) 3.4 (4.0) VSLA Only Ability to refuse sex 5.7 (1.7) 6.2 (1.5) (Comparison) g VSLA + GDG h 5.7 (1.8) 6.3 (1.5) 0.10 (−0.19, 0.39)l 0.49 † Adjusted for clustering. a Total n for comparison group at baseline is 421; Total n for comparison group at endline is 371. b Total n for intervention group at baseline is 513; Total n for intervention group at endline is 483. c Total n for comparison group at baseline is 412; Total n for comparison group at endline is 370. d Total n for intervention group at baseline is 501; Total n for intervention group at endline is 483. e Total n for comparison group at baseline is 419; Total n for comparison group at endline is 401. f Total n for intervention group at baseline is 511; Total n for intervention group at endline is 502. g Total n for comparison group at baseline is 421; Total n for comparison group at endline is 403. h Total n for intervention group at baseline is 512; Total n for intervention group at endline is 503. i Total observations used in model is 1788 / 1868. j Total observations used in model is 1766/1868. k Total observations used in model is 1833/1868. l Total observations used in model is 1839/1868. 10 Table 4. Distribution of study outcomes at baseline and endline, by treatment group and unadjusted effect estimates of past-year intimate partner violence (Per protocol Analysis) (N = 934) Baseline N Endline N Treatment type (%) (%) Adjusted OR,q 95% CI p-value Physical and/or Sexual IPV VSLA Only (Comparison) a 93 (22.1) 78 (21.0) -- -- VSLA + GDG b (Low adherence) 64 (22.9) 63 (24.6) 1.19 (0.69, 2.05) 0.64 VSLA + GDG c (High adherence) 55 (23.5) 37 (16.3) 0.64 (0.35, 1.16)l 0.14 Physical IPV VSLA Only (Comparison) a 65 (15.4) 55 (14.8) -- -- VSLA + GDG b (Low adherence) 44 (15.8) 36 (14.1) 0.93 (0.49, 1.77) 0.82 VSLA + GDG c (High adherence) 36 (15.4) 17 (7.5) 0.45 (0.21, 0.94)l 0.04 Sexual IPV VSLA Only (Comparison) a 44 (10.5) 53 (14.3) -- -- VSLA + GDG b (Low adherence) 38 (13.6) 41 (16.0) 0.85 (0.44, 1.64) 0.63 VSLA + GDG c (High adherence) 33 (14.1) 27 (11.9) 0.54 (0.27, 1,10)l 0.11 Economic Abuse VSLA Only (Comparison) d 113 (27.4) 128 (34.6) -- -- VSLA + GDG e (Low adherence) 99 (36.3) 56 (21.9) 0.31 (0.18, 0.52) <0.0001 VSLA + GDG f (High adherence) 64 (28.1) 48 (21.2) 0.47 (0.27, 0.81) m 0.01 Mean (SD) Mean (SD) Adjusted β (95% CI) p-value Justification for VSLA Only (Comparison) g 4.5 (4.3) 4.0 (4.0) wife beating VSLA + GDG i (High adherence) 4.5 (4.2) 2.9 (3.6) −1.14 (−2.01, -0.28) 0.01 Ability to VSLA Only (Comparison) j 5.7 (1.7) 6.2 (1.5) refuse sex VLSA + GDG (Low adherence) h 5.7 (1.8) 6.3 (1.6) 0.07 (−0.32, 0.46) o 0.72 VSLA + GDG (High adherence)k 5.7 (1.7) 6.4 (1.4) 0.12 (−0.24, 0.48) 0.50 a Total n for VLSA only at baseline is 416; Total n for VLSA only at endline is 368. b Total n for VSLA + GDG low adherence at baseline is 275; Total n for VLSA + GDG low adherence at endline is 252. c Total n for VSLA + GDG high adherence at baseline is 232; Total n for VSLA + GDG high adherence at endline is 226. d Total n for VLSA only at baseline is 408; Total n for VLSA only at endline is 367. e Total n for VSLA + GDG low adherence at baseline is 270; Total n for VSLA + GDG low adherence at endline is 252. f Total n for VSLA + GDG high adherence at baseline is 227; Total n for VSLA + GDG high adherence at endline is 226. g Total n for VLSA only at baseline is 419; Total n for VLSA only at endline is 401. h Total n for VSLA + GDG low adherence at baseline is 278; Total n for VSLA + GDG low adherence at endline is 273. i Total n for VSLA + GDG high adherence at baseline is 233; Total n for VSLA + GDG high adhesrence at endline is 229. j Total n for VLSA only at baseline is 421; Total n for VLSA only at endline is 403. k Total n for VSLA + GDG high adherence at baseline is 234; Total n for VSLA + GDG high adherence at endline is 230. l Total observations in model is 1769/1868. m Total observations in model is 1750/1868. n Total observations in model is 1812/1868. o Total observations in model is 1818/1868. p Adjusted for categorical number of pregnancies and religion as they were statistically associated with adherence. Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 11 Moreover, the current Cote d’Ivoire intervention served between treatment arms per administrative was a shorter intervention than the IMAGE trial in- records. Also, in eight villages with a smaller popula- tervention, which also included a larger community tion, groups were only randomized to one arm due mobilization component. to contamination concerns. Regarding external va- lidity, while a community-based sample was recruit- There are also differences regarding study design. ed, participants may differ from women who chose In this study, we assessed the incremental impact not to participate in the investigation. of adding gender equity components onto a VSLA program, while the IMAGE intervention assessed The study was likely to be underpowered as certain the impact of a combined micro-credit and gender villages had more widowed participants than an- training program compared with no intervention at ticipated and because of our inability to mobilize as all. In the current study, we saw significant changes many villages as expected, thus reducing the analytic among those women who were highly adherent to sample. Moreover, given the relatively scarce num- the intervention (or who achieved high interven- ber of interventions conducted to date that focus tion exposure), while in IMAGE, the impact was on the program components and the populations in significant among all participants. Secondary analy- the current study, findings reported herein are pre- ses of IMAGE data suggest that the gender training liminary in nature. Additionally, since some of the component of the intervention was critical to its villages were inaccessible during a period of post- success [11]. Future research is needed to identify election violence in 2010, regular fidelity monitor- exact pathways of change for the current interven- ing of VSLA activities was not possible. However, in tion. Taken together, findings illustrate the potential anticipation of possible post-election violence, the benefits of adding gender sensitization components IRC team conducted training prior to the start of onto livelihood programs for women in both con- activities. Attendance records indicate regular meet- flict affected and non-conflict affected settings. ings, including meeting at undisclosed locations during threats of intense violence. While widespread Study findings must be interpreted within the con- violence may have influenced the physical mobility text of limitations. Firstly, as with most stigmatized of our study population, which may in turn have af- health issues, IPV self-reporting may be subject to fected whether couples stayed together physically social desirability bias. Also, prior research has sug- during times of insecurity, post-hoc analyses indi- gested that participants in IPV reduction interven- cated that cohabitation of partners in the year pre- tions may over-report IPV due to increased aware- ceding the endline survey did not significantly differ ness [16]. However, the directionality of such bias by treatment group, adherent group, or reporting of is difficult to determine as it is unclear whether one IPV. Although financial issues/lack of confidence arm would be more or less likely to under or over in VSLA activities were frequently cited as reasons report; biases away from the null would be present for leaving the program among those who dropped if over-reporting was only present in the treatment out, it is unclear if this fully explains why drop-outs arm. Future research in post-conflict settings that in- differed by treatment status given that all women cludes similar quantitative assessments of IPV per- received VSLA simultaneously. A related potential petration and gender attitudes among men would limitation is that we were unable to assess how the strengthen understanding of the impacts of the type success of VSLA activities influenced overall group of programming that was evaluated herein. Second, dynamics or effectiveness of the GDGs, as the incre- participants, facilitators, and researchers were not mental effects of GDGs may be correlated with the blinded to treatment status. Given that both treat- success of the economic component of the interven- ment and control groups took place in the same vil- tion. Notably, attrition was not significantly related lages, there is a chance of contamination, and thus a to IPV at baseline or endline, nor were groups with bias towards null findings. However, pre-study con- high levels of drop-outs collapsed, which minimized sultations with community leaders suggested that contamination concerns. Moreover, due to post- the use of a waitlist control design in large villages election violence, the start of the GDGs were de- would minimize such threats. No crossover was ob- layed and could not be completed as of August 2011 12 as planned initially. Thus, the past year assessment may precede changes in IPV [19]; longer-term re- of IPV is inclusive of 8 months in which the GDG search is needed to determine this theory of change. was being delivered and it is unclear if the GDGs, Further, trends in IPV reduction are encouraging, in the context of economic empowerment program- given the overlap of follow-up period and delayed ming, would have been able to reduce IPV in this intervention delivery. Related to the short follow-up shortened period. In addition, while per protocol time, the effect size for the overall outcome measure analyses assessed adherence to the GDG sessions, (physical and/or sexual IPV) may have been small we do not know which sessions were skipped. It can due to overlap between women who reported both be argued that the sessions regarding financial stress sexual IPV and physical IPV. This overlap, combined and household economy may have more directly with the possibility that it may be more feasible to addressed IPV and inequitable gender norms than influence physical IPV levels in shorter time frames other sessions. Therefore, non-attendance to these compared to sexual IPV, may in part explain why sessions may have reduced the impact of the GDG effect sizes for the overall outcome (physical and/ component. Notably, while not all sessions were or sexual IPV) were small and did not reach signifi- developed to explicitly discuss IPV, reports from fa- cance. Longer follow-up time is needed for future cilitators indicate that the topic of IPV was sponta- work. Findings that attitudes related to the justifi- neously discussed by groups at each session. As we cation of physical IPV were significantly improved, were able to assess adherence to the GDG interven- but not sexual IPV further suggest the need for fu- tion and impacts on summary measures of different ture research to investigate the relative difficulty of forms of IPV, we could not examine the severity or reducing sexual IPV in comparison to physical IPV. frequency of IPV events in our data in accordance In addition, while significant reductions in sexual with a dose–response relationship. IPV were not found, it is possible that the GDGs may have prevented increases in sexual IPV. These limitations notwithstanding, the current RCT has important strengths. It was done in partnership with a non-governmental organization with a long Conclusions working history in Côte d’Ivoire, and incorporated specific components to maintain positive relation- The current study has important programmatic im- ships with the community (e.g. inclusion of non- plications for addressing IPV in conflict- affected partnered women, non-use of a non-interventional settings, including areas devoid of formal financ- control group due to voiced ethical concerns). ing institutions. While more research is needed to Moreover, despite the use of a comparison arm investigate the sustainability of the intervention that received only the economic intervention (ver- effects in the longer term, underlying mechanisms sus a pure control), reductions in IPV and changes that give rise to the observed effects, and the costs in attitudes towards justification of IPV were ob- of scaling up, findings on the reduction of IPV and served. All responses were prospectively assessed, improvements in attitudes toward justification of and loss to follow-up was minimal despite ongoing IPV in the GDG + VSLA group compared to VSLA post-election violence. Finally, although not all re- alone are promising. This research has also shown sults reached statistical significance, the findings that a subtle and feasible approach to addressing from both ITT and PP analyses are consistent with IPV, such as the one employed by GDGs can also one another. As argued by other violence research- reduce women’s exposure to partner violence, even ers and community interventionists, for complex in war-affected settings where violence may be more interventions such as the one evaluated in Côte prominent and individuals’ reluctance to discuss d’Ivoire, consistency and directionality of findings, abuse may be exacerbated for fear of seeming divi- in additional to statistical significance, are important sive during community-wide attempts to maintain components to consider [15,31,34]. The significant peace and reconciliation [35]. In addition, in con- finding regarding improvement in attitudes towards flict-affected situations, where social and financial justification of IPV is particularly important, as at- structures might become altered and women may, titudinal changes may be proxies for norms, which out of necessity, take on what are traditionally male Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 13 roles, women’s post-conflict safety may depend on principal investigator, had full access to all the data fostering equitable gender norms to reduce the in the study and takes responsibility for the integrity threat of backlash [36]. Effectively reducing IPV in of the data and the accuracy of the data analysis. combination with creating greater financial oppor- tunities—whether organically produced or formal- ized through economic empowerment efforts— Acknowledgments may lead to sustained improvements in women’s This study was funded by the World Bank’s State status in a post-conflict period. Moreover, while and Peace-building Fund, Contract #1007007040 economic empowerment programs targeting wom- (PI J Gupta). This work was supported, in part, by en are increasing as a means to advance women’s sta- Yale University’s Center for Interdisciplinary Re- tus, health, and livelihood, these findings showcase search on AIDS (CIRA), through grants from the that the addition of an intervention for women and National Institute of Mental Health, Paul Cleary, their male partners that promotes gender equitable Ph.D., Principal Investigator (P30MH062294). The norms yields more reductions in IPV than econom- views presented are those of the authors and do not ic programming alone. Results from this trial should serve as a part of an emerging evidence base to in- necessarily represent the views of the World Bank, form policy and programs on promising strategies NIMH, NIH, or IRC. Study sponsors had no role to reduce IPV in conflict-affected settings. This type in the planning of the study or in writing/approving of innovative multi-sector programming is likely to the manuscript. translate into better health and safety [6] and greater The authors would like to thank the following social and economic advancements for women and people: Ms Mathilde Dubois (Economic Recovery their families [8]. Coordinator) and Ms Monika Bakayoko-Topolska (Gender-Based Violence Coordinator) of Interna- Competing interests tional Rescue Committee (International Rescue Committee, Cote d’Ivoire) for their invaluable assis- The authors have no competing interests to declare. tance in planning field activities; Ms Kelly Bidwell (IPA, New Haven) for her excellent oversight in Author’s contributions coordination; Ms Lauren Young (Columbia Univer- sity) for her invaluable assistance during the start-up JG and JA led the conceptualization of the study. phase of the project; and Mr Elder Marc Gonkanou MZ, CZ, and CW contributed to the initial study Lie for his supervision and project coordination. The conceptualization. JG, DK, JA, and KF carried out authors would also like to thank Elizabeth Bradley, data collection and the evaluation. ZX completed PhD; Trace Kershaw, PhD; Rafael Perez-Escamilia, statistical analyses. JG and KF led the manuscript PhD; Linda Niccolai, PhD, Sarah K Calabrese, PhD, writing. All authors have been involved in critically Ann Levie, MSW, MPH, and Russel Barbour, PhD revising the manuscript. All authors have read and for their helpful comments for improvement of the approved the final version of the manuscript. JG, manuscript. 14 References 1. Stark L, Ager A: A systematic review of prevalence studies of gender-based violence in complex emer- gencies. Trauma Violence Abuse 2011, 12(3):127–134. 2. Parmar P, Agrawal P, Greenough PG, Goyal R, Kayden S: Sexual violence among host and refugee pop- ulation in Djohong District, Eastern Cameroon. Glob Public Health 2012, 7(9):974–994. 3. Peterman A, Palermo T, Bredenkamp C: Estimates and determinants of sexual violence against women in the Democratic Republic of Congo. AJPH 2011, 101(6):1060–1067. 4. International Rescue Committee: Let me not die before my time: domestic violence in West Africa. New York, NY: IRC; 2012. 5. Annan J, Brier M: The risk of return: intimate partner violence in Northern Uganda’s armed conflict. Soc Sci Med 2010, 70:152–159. 6. Campbell JC: The health consequences of intimate partner violence. Lancet 2002, 359:1331–1336. 7. WHO Multi-County Study on Women’s Health and Domestic Violence against Women Study Team, Ells- berg M, Jansen HAFM, Heise L, Watts CH, Garcia-Moreno C: Intimate partner violence and women’s physical and mental health in the WHO multi-country study on women’s health and domestic vio- lence: an observational study. Lancet 2008, 371(9619):1165–1172. 8. World Health Organization: Addressing violence against women and achieving the Millennium Development Goals. Geneva: WHO; 2005. 9. Kabeer N: Is microfinance a ‘magic bullet’ for women’s empowerment? analysis of findings from South Asia. Econ Pol Wkly 2005, 40(44/45):4709–4718. 10. Krishnan S, Rocca CH, Hubbard AE, Subbiah K, Edmeades J, Padian N: Do changes in spousal em- ployment lead to domestic violence? Insights from a prospective study in Bangalore, India. Soc Sci Med 2010, 70(1):136–143. 11. Kim J, Ferrari G, Abramsky T, Watts C, Hargreaves J, Morison L, Phetla G, Porter J, Pronyk P: Assessing the incremental effects of combining economic and health interventions: the IMAGE study in South Africa. Bull World Health Organ 2009, 87(11):824–832. 12. Vyas S, Watts C: How does economic empowerment affect women’s risk of intimate partner vio- lence in low and middle income countries? A systematic review of published evidence. J Int Dev 2009, 21:577–602. 13. Goetz AM, Gupta RS: Who takes the credit? gender, power, and control over loan use in rural credit programs in Bangladesh. World Dev 1996, 24(1):45–63. 14. Kim JC, Watts CH: Gaining a foothold: tackling poverty, gender inequality, and HIV in Africa. BMJ 2005, 331(7519):769–772. Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 15 15. Kim JC, Watts CH, Hargreaves JR, Ndhlovu LX, Phetla G, Morison LA, Busza J, Porter JD, Pronyk P: Understanding the impact of a microfinance-based intervention on women’s empowerment and the reduction of intimate partner violence in South Africa. Am J Public Health 2007, 97(10):1794–1802. 16. Pronyk PM, Hargreaves JR, Kim JC, Morison LA, Phetla G, Watts C, Busza J, Porter JD: Effect of a struc- tural intervention for the prevention of intimate-partner violence and HIV in rural South Africa: a cluster randomised trial. Lancet 2006, 368(9551):1973–1983. 17. Gibbs A, Willan S, Misselhorn A, Mangoma J: Combined structural interventions for gender equal- ity and livelihood security: a critical review of the evidence from southern and eastern Africa and the implications for young people. J Int AIDS Soc 2012, 15(Suppl 1):1–10. 18. Glass N, Ramazani P, Tosha M, Mpanano M, Cinyabuguma M: A Congolese-US participatory action research partnership to rebuild the lives of rape survivors and their families in eastern Democratic Republic of Congo. Glob Public Health 2012, 7(2):184–195. 19. Heise L: What works to prevent partner violence? An evidence overview. London: London School of Hygiene and Tropical Medicine; 2011. 20. Coker AL, Richter DL: Violence against women in Sierra Leone: frequency and correlates of inti- mate partner violence and forced sexual intercourse. Afr J Reprod Health 1998, 2(1):61–72. 21. Salihu H, August E, Salemi J, Weldeselasse H, Sarro Y, Alio A: The association between female genital mutilation and intimate partner violence. BJOG 2012. epub ahead of print. 22. Institut National de la Statistique (INS) et Ministere de la Lutte contre le Sida [Cote d’Ivoire] et ORC Macro: Enquete sur les indicateurs du sida, Cote d’Ivoire 2005. Calverton, Maryland, USA: INS and ORC Mac- ro; 2006. 23. Hossain M, Zimmerman C, Kiss L, Watts C: Violence against women and men in Cote d’Ivoire: a cluster ran- domized controlled trial to assess the impact of the ‘Men and Women in Partnership’intervention on the reduction of violence against women and girls in rural Cote d’Ivoire - results from a community survey. London, UK: London School of Hygiene and Tropical Medicine (LSHTM); 2010. 24. WHO Multi-country Study on Women’s Health and Domestic Violence against Women Study Team, Garcia-Moreno C, Jansen HA, Ellsberg M, Heise L, Watts CH: Prevalence of intimate partner violence: findings from the WHO multi-country study on women’s health and domestic violence. Lancet 2006, 368(9543):1260–1269. 25. Prochaska JO, Velicer WF: The transtheoretical model of health behavior change. Am J Health Promot 1997, 12(1):38–48. 26. Beyene B: EA$E discussion series: Facilitator guide. Abidjan: International Rescue Committee; 2010. 27. VSL Associates. What is VSLA? Accessed October 31, 2013. http://www.vsla.net/home. 28. Iyengar R, Ferrari G: Discussion sessions coupled with microfinancing may enhance the role of women in house- hold decision-making in Burundi. Working Paper 16902. Cambridge, MA: National Bureau of Economic Re- search; 2011. 29. World Health Organization: Putting women first: ethical and safety recommendations for research on domestic violence against women. Geneva: WHO; 2001. 16 30. Garcia-Moreno C, Jansen AFM, Ellsberg M, Heise L, Watts C: WHO multi-country study on women’s health and domestic violence against women: summary report of initial results on prevalence, health outcomes, and women’s responses. Geneva: WHO; 2005. 31. Abramsky T, Devries K, Kiss L, Francisco L, Nakuti J, Musuya T, Kyegombe N, Starmann E, Kaye D, Mi- chau L, et al: A community mobilisation intervention to prevent violence against women and reduce HIV/AIDS risk in Kampala, Uganda (the SASA! study): study protocol for a cluster randomised con- trolled trial. Trials 2012, 13(96):1–22. 32. Microsoft Corporation: Access. Seattle: Microsoft; 2010. 33. SAS Institute Inc: SAS 9.2. Cary, NC: SAS Institute Inc; 2010. 34. Sorensen G, Emmons K, Hunt MK, Johnston D: Implications of the results of community interven- tion trials. Annu Rev Public Health 1998, 19:379–416. 35. Silber IC: Mothers / fighters / citizens: Violence and disillusionment in post-war El Salvador. Gend History 2004, 16(3):561–587. 36. Bop C: Women in conflicts, their gains and their losses. In The aftermath: women in post-conflict trans- formation. Edited by Meintjes S, Pillay A, Turshen M. London: Zed Books; 2001. Gender Norms and Economic Empowerment Intervention to Reduce Intimate Partner Violence Against Women in Rural Côte D’ivoire: A Randomized Controlled Pilot Study 17 THE WORLD BANK 1818 H, Street N.W. Washington, D.C. 20433 www.logica-wb.org