Empowering women through Family Visioning: a randomized experiment in Uganda Kate Ambler, Kelly Jones, and Michael O’Sullivan December 2018 1. Introduction Women’s empowerment programs have gained popularity in many low-income countries as a means to improve women’s economic status, health and livelihood. Such programs vary widely, ranging from purely economic to solely focused on changing hearts and minds. The successes and failures of economic approaches have been reviewed based on more than 100 randomized controlled trials (Buvinic and O’Donnell, 2016). Several intervention types have been documented as well – proven successes, including savings, child care, conditional cash transfers, land rights, and electrification. But others have been less successful, such as financial literacy training, business training, and micro-credit. Recently, there has been an increasing focus on integrating couples-based, psychosocial components with economic strategies for improving women’s empowerment. In Burkina Faso, a family coaching program regarding child protection, gender norms, and decision making was added to an economic intervention and resulted in increased women’s financial autonomy, improved marital quality and reduced emotional spousal violence (Ismayilova et.al., 2017). Similarly, in Côte d’Ivoire, adding couples’ trainings on gender inequality to a savings program resulted in a significant reduction in spousal violence (Gupta et al., 2013). In this study, we examine the impacts of a couples’ workshop that encouraged individual and joint planning for the future and discussions of gender balance in terms of responsibilities and access to resources. We examine the impacts of this workshop on gender norms, decision making, and women’s economic participation in rural Uganda. In some households, this intervention was combined with an economic intervention. However, due to our cross-randomized design we are able to identify the impacts of the workshop separately from those accruing to the economic intervention. The context of our study is an area where sugarcane farming represents the majority of household income. As such, economic empowerment is closely linked with the extent and nature of one’s participation in sugar production and marketing. This is common in agrarian economies, where increasing women’s participation in high-value agriculture is often seen as a promising pathway to economic empowerment. However, participation in production may not improve empowerment without increases in decision-making and control over income. For example, in Cote d’Ivoire, a couples’ training was found to increase the wife’s salience in the action plan for cultivation of rubber. However, her control over income only improved if she was assigned to high-level management tasks (Donald and Rouanet, 2018). In addition to the outcomes of interest noted above, we examine not only the extent but also the nature of women’s involvement in this industry. 2. Project Design 2.1. Intervention In this paper we study an intervention targeted at married couples that aims to increase gender equity and women’s empowerment within the household. The intervention was conducted in Jinja, Uganda. Jinja is in Eastern Uganda, near Lake Victoria and the border with Kenya. The agriculture in this area is dominated by production of sugarcane. One of the largest commercial buyers of sugarcane is Kakira Sugar Limited. While Kakira produces its own sugarcane, the majority of cane processed by the company is produced by approximately 6,000 outgrowers. This list of Kakira outgrowers served as the sampling frame for this study. The intervention consisted of Family Vision Workshops, implemented by Oxfam Novib, which were based closely on the Change Catalyst Workshops from the Gender Action Learning System (GALS). GALS was designed and has been previously implemented elsewhere by Oxfam Novib in Uganda, as well as in other countries. GALS seeks to teach couples participatory planning techniques and develop “new visions for relationships between women and men [and change] gender inequalities in resources and power” (Mayoux, 2014). The workshop was a three-day couples-based participatory training that focused on recognizing contributions of each member and arriving at a balanced approach to household (or farm) management and access to resources. The goal of the workshop was to alter the way in which spouses communicate and organize their households to become more inclusive and, in particular, involve women in sugarcane production. The intervention was implemented by experienced GALS trainers from another region in Uganda. For more details on the contents of the workshop, see Appendix A. Approximately one half of the sample was randomly assigned to receive the workshop treatment following the baseline survey. We stratified treatment assignment on the following variables: whether a household is polygamous (34%), whether the household farms more than the median number of sugarcane blocks (2), whether the tenure of marriage with the designated wife is greater than the median of 20 years, whether the designated wife can read and write (63%), and measures of cooperation and preference alignment measured through a baseline survey module of incentivized decision making. Selected households were invited in person by a mobilization team that made up to three visits. A second intervention focused on sugarcane contracts was assigned to half of the sample, stratifying on the same characteristics as above, as well as workshop treatment assignment. As such, a quarter of the household sample received both treatments, two quarters each received one treatment or the other, and a quarter of households are in the control group (see Figure 1). The contract intervention entailed household visits by specially-trained staff to remind/inform each farmer that he is allowed to transfer one or more of his contracts into the name of his wife at any time. Farmers can also register in the wife’s name any unregistered blocks. Making the wife the contract holder entitles her to inputs, cash advances, and the final payment. The project facilitated the paperwork to make these transfers and registrations easy and paid the small costs associated with them. Figure 1. Experimental Design Contract Intervention No Contract intervention Workshop Both interventions Workshop only 586 couples 594 couples No Workshop Contract only Pure Control Group 591 couples 589 couples Figure 2. Study Timeline 2.2. Study timing Details of study timing are presented in Figure 2. Sample recruitment and the baseline survey took place in July to September 2016. Workshops were conducted between October and December 2016. The contract intervention began in late-February 2017 and the field visits concluded in early May. Office approval of all paperwork and return visits to households to deliver Kakira outgrower ID cards took place between late March and mid-September 2017. In this paper we rely on midline data collected October to December of 2017, between 10 and 14 months after a couple attended a workshop (if applicable). Given the short time horizon between the completion of the contract intervention (September 2017) and the collection of this data, we leave investigation of the impacts of the contract intervention for future work. 2.3. Sample Households were recruited to participate from a list of active, male Kakira sugarcane farmers. An eligibility questionnaire was followed by the baseline survey with eligible households. Households had to meet all of the following criteria in order to be eligible to participate: they had to currently farm sugarcane, have at least one wife, have at least one block of sugarcane unregistered or registered to Kakira in the husband’s name, and have no outstanding loans against sugarcane blocks. The final baseline sample contains 2,370 households who were eligible for the project and who completed the baseline survey.1 In the case of a polygamous household, we asked the male household head to identify which of his wives he would most like to involve in sugarcane production, and who will participate in the rest of the survey and intervention. This wife is referred to as the “designated wife” for the purposes of the project. During the baseline survey, we interviewed both the male respondent and his designated wife, both together and privately. Table 1 provides some baseline demographic characteristics of the sample. Husbands are 47 years old on average, compared to 38 for designated wives. Approximately 34% of the sample is polygamous, and the average couple has been married for 20 years. Men have on average 1.79 years more education than their wives. Average household size is 8.7, including 5 children under the age of 18. Data on work that men and women do relating to sugarcane indicate that women are involved in sugarcane labor, but have little involvement in marketing activities at baseline. Additional details about the sample at baseline are presented in Ambler, Jones and O’Sullivan (2018). The midline survey interviewed wives in-person and collected information on intermediate outcomes, such as time use, personal expenditures, personal savings, gender norms, marital communication, and decision-making. We also queried knowledge of sugarcane production and sales, and details of production and harvests for sugarcane blocks in which the wife was involved. Additionally, we spoke with husbands in person or by phone to elicit cursory information about harvests of their own cane blocks since the time of the baseline. We were able to contact 2,256 wives from the baseline sample (95%) and completed interviews with 2,229 (1% of women refused). 2.4. Analysis of take-up Figure 3 shows the take-up by gender. 84% of invited husbands and 82% of invited wives attended any part of the workshop. Nearly all of these attended all day of all three days of the workshop, with a small share attending at least half but not all of the total meetings and very few attending less than half. In the midline interview we included several questions to assess the degree to which the workshops were memorable or salient for the participants. 86% of women who attended any part of the workshop (90% of those who attended the full workshop) reported they had ever attended any type of training or workshop that encouraged spousal communication about a vision or plan for the future of the family. In contrast only 43% of non-attendees (including control-group women) reported ever attending one. 1 Additional details about creation of the sample and baseline data collection are available in Ambler, Jones and O’Sullivan (2018). Figure 3. Workshop Attendance 3. Impact of Intervention 3.1. Main Results In this section we examine the impacts of the workshop 1 year later. We estimate these impacts at the household level with the equation = + + + (1) Where is an outcome of interest for couple from stratum , indicates that the couple was assigned to the workshop intervention, is a stratum-fixed effect, and is a normally distributed error term. We present estimates using ordinary least squares estimation. However, we also estimate the impact of actually attending the workshop, as instrumented by treatment assignment, in a two- stage least squares framework. As expected given the high level of compliance, the results are not measurably different. For binary outcomes we also estimate the equation using a probit specification and find that the results are unchanged in terms of direction, significance, and magnitude of marginal effects at the mean. 3.1.1 Gender Norms & Household Balance The first set of results examines the impacts of the intervention on personal views about gender norms, marital communication, and marital decision-making. We also examine women’s balance of housework and leisure. If the intervention induced men to increase their participation in housework this would decrease women’s contribution to housework. However, it is also of interest whether such a shift would increase women’s time spent in income-generating activities or increase women’s leisure time. Further, if women increase time spent on sugarcane as a result of the intervention, that could also affect leisure time. The measures of gender norms are detailed in Table 2with the share of women reporting strong agreement (or disagreement) with each statement. While 95% of respondents strongly disagree that it is better to send a son to school than a daughter, other measures show significant variation. These measure topics such as decision making, domestic violence, household chores, and the right to express disagreement. Marital communication is measured by two questions, each on a scale of 1 to 10 regarding agreement with statements that the respondent likes to discuss personal concerns with her husband, and that they usually discuss major household decisions. The mean response is 7 or 8, with a standard deviation of 2. Household decision making is measured with eleven questions, each measured on a scale from 1 to 5 where 5 indicates the wife makes the decision independently, and 1 indicates that the husband makes the decision independently.2 These include questions about decisions regarding how the household will earn and use money, major and everyday purchases, allocation of responsibilities, land use, healthcare respondent and her children, family planning, and family visits. The decision-making index is the average score across the eleven items. Table 3 presents the results from Equation 1. We find that invitation to the workshop induced a small but statistically significant increase in women’s belief in gender equality. We also find that the workshop improved some aspects of marital communication, specifically, the probability that the couple discusses a vision for their family at least once per month. The average of other aspects of communication, such as discussing personal concerns or major household decisions, are not increased. The negative coefficient on these may indicate that women’s expectations of communication were increased by the workshop, so that a lack of improvement in actual communication results in a lower score on the 10- point scale. The index of eleven decision making measures does not indicate that the workshop increased women’s decision-making power within their households. Similarly, we see no statistically significant impact on hours per week spent on housework or leisure by women. In Table 4 we further explore what change in norms is driving the impact shown in column 1 of Table 3. We estimate Equation 1 separately for each of the four norms included in the average. We find that the impact is driven by changes in women’s opinions about responsibility for household chores. Women invited to the workshop agree that husbands should help around the house when wives work outside the home at a rate that is 4% higher than the mean in the control group. 3.1.2. Integration in Cane Activities This set of results examines the extent to which the interventions increased women’s participation in and ownership of sugarcane farming activities. We focus on management activities, in particular those related to marketing and sales, which are determining factors in the receipt and distribution of cane revenues. We estimate several different outcomes using Equation 1; the results are presented in Table 5. We find that invitation to the workshop has a small, negligible effect on women’s participation in management of production activities. Sixty-eight percent of women in the control group report managing production, and the workshop increases this by only 2 percentage points (not statistically significant). In contrast, we 2 2=Decide together but husband’s preferences matter more; 3=Decide together and preferences matter equally; 4=Decide together but wife’s preferences matter more. estimate a much larger relative increase in management of sales activities: an increase of 2 percentage points on a basis of only 6.5% among the control group (significant at the 10% level). Similarly, we do not find significant increases in women reporting that they perform any management activity for any block, but we do find a large and significant increase in the share of women reporting to be the primary manager of any block. This is true for only 33% of women in the control group and is higher by 12% (3.9 ppts) among those invited to workshops (significant at 5% level). Relatedly, we see a similarly-sized increase in women reporting being the registrant or contract-holder for any block. Note that women invited to the workshop were also invited to the contract intervention at equal rates with the control group (50% of each), so these effects cannot be ascribed to the contract intervention alone. Rather, these indicate an impact of the workshop intervention that is over and above any impact of the contract intervention.3 Given that the couples enrolled in this study use sugarcane as their primary means of income, it is not unexpected that changes in women’s economic empowerment would result in increased participation in cane management and sales, regardless of whether the workshop specifically encouraged that dynamic. In reality, the workshops did not avoid discussions of women’s roles in cane production, given its dominance in the context, but encouraging couples to increase women’s participation was not the main focus of the workshops. 3.2. Heterogeneous Impacts At baseline we collected incentivized decision measures of efficiency and bargaining power within each couple. Individuals were asked which spouse they would prefer to receive 10,000 USh on behalf of the household. They were then asked whether they would change their decision if the household would receive 15,000 USh to the less-preferred recipient versus 10,000 to the preferred recipient. In this way, we measured individuals’ willingness to pay to control household resources. We interpret this as both a measure of household inefficiency, but also baseline access to resources. Individuals with insufficient access to household resources at baseline should be more willing to pay to control new incoming resources. This measure is more reliable than survey-based questions about resource access because the answers individuals give actually determine how much money their household will receive and who will receive it. Additionally, we asked each individual to divide a sum of 28,000 USh between oneself and one’s spouse. Later, couples made a joint decision about how such a sum should be divided. In this way, we measured individuals’ preferences for resource allocation and the weight of their preferences in a joint decision, a measure of one’s bargaining power. The summary statistics from these measures are shown in Table 5. Impacts on participation in cane production, management, and sales 3 Since these interventions were cross-randomized, assignments are fully orthogonal. Therefore, as expected, these results are unchanged regardless of whether we control for invitation to the contract intervention. Any management role For any block Production Sales Does labor Does Mgmt Primary Mgr. Registrant (1) (2) (3) (4) (5) (6) Workshop 0.023 0.018* 0.029* 0.011 0.039** 0.043** (0.019) (0.011) (0.015) (0.019) (0.020) (0.020) Control Mean of Dep.Var. 0.684 0.065 0.825 0.704 0.331 0.342 N 2370 2370 2370 2370 2370 2370 R 0.033 0.038 0.032 0.038 0.035 0.026 Note: All regressions control for stratum fixed effects. Standard errors are shown in parentheses. Coefficients are statistically significantly different from zero at the 10% (*), 5% (**), or 1% (***) level as indicated. Table 6. We use the baseline measures of efficiency and empowerment to test for heterogeneous impact of the workshop on the outcomes shown in Table 3. We disaggregate the results from Table 3 by a key variable from Table 5. Impacts on participation in cane production, management, and sales Any management role For any block Production Sales Does labor Does Mgmt Primary Mgr. Registrant (1) (2) (3) (4) (5) (6) Workshop 0.023 0.018* 0.029* 0.011 0.039** 0.043** (0.019) (0.011) (0.015) (0.019) (0.020) (0.020) Control Mean of Dep.Var. 0.684 0.065 0.825 0.704 0.331 0.342 N 2370 2370 2370 2370 2370 2370 R 0.033 0.038 0.032 0.038 0.035 0.026 Note: All regressions control for stratum fixed effects. Standard errors are shown in parentheses. Coefficients are statistically significantly different from zero at the 10% (*), 5% (**), or 1% (***) level as indicated. Table 6: whether the wife is willing to pay to control resources, an indicator that she lacks sufficient access to resources within her household. These results are presented in Table 7. We observe that the impacts of workshop invitation are strongest among couples where the wives are disempowered at baseline. The coefficients in the top row indicate the treatment effects for women who are willing to pay to control resources. These are generally larger and more precisely estimated than those in Table 3. In addition, the interaction effects are consistently negative, indicating that treatment effects are weaker in household with more empowered wives. However, only for the outcomes of primary block manager and block registrant can we reject that the treatment effects are the same for more and less empowered women. In results not shown, we explore this interaction model using other indicators presented in Table 6. We find similar results for other measures of empowerment, for example, effects are more concentrated among households where the joint decision exactly matches the husband’s preference, or where the wife has zero bargaining power.4 We also find that other measures of household inefficiency, for example an indicator for whether either spouse is willing to pay to control resources, do not exhibit the same pattern. From this we conclude that the results shown in Table 7 provide evidence more related to wife’s empowerment than to household efficiency generally. 4. Discussion & Conclusions In this study we have examined the impacts of a Family Vision Workshop for couples on several outcomes of interest. We find that the workshops significantly increased women’s belief in gender equality, as measured by an index of four gender norms. In particular, we see the strongest impacts on women’s belief that husbands should help with household chores when the wife works outside the home. This effect is entirely driven by women who are less empowered at baseline. Women who were willing to pay a 50% penalty at baseline for the right to control incoming resources account for the full impact of the workshop on changes in views about gender norms. At first glance one might assume that this is simply reversion to the mean; perhaps more empowered women had higher belief in gender equality at baseline and so there is more room for improvement among less empowered women. However, we do not observe this to be the case. At baseline there is no correlation between views on gender norms and any experimental measures of empowerment. This suggests that the impact of the workshop on gender norms is truly stronger among less empowered women, indicating that it may be a useful tool for increasing women’s empowerment. Similarly, we find that the workshop increased the probability that a woman reports discussing a shared vision for the future with her husband on a regular basis (at least once per month). One might be concerned that this is simply due to differential reporting based on desirability bias following attending the workshop. However, we exploit the fact that 11% of attendees report never having attended a training of this type (reflecting memory lapses after one year). While this sample is too small for hypothesis testing we do find a positive coefficient that is similar in magnitude to the original estimate, indicating that the effect is not entirely due to differential reporting. This suggests that the workshops did exert a measurable effect on spousal communication. As with the impact on norms, this effect is larger for (though not entirely due to) women who were less empowered at baseline. In addition, we find that the workshops did increase women’s participation in cash crop production and sales. Though we find no statistically significant increase in the share of women with a management role in production and only a small relative (3.5%) increase in the share of women participating in physical labor on sugarcane, we find much larger increases in women’s participation in management of cane 4 Bargaining power is defined (for the wife) as the distance between the husband’s preference and the joint decision, minus the distance between the wife’s preference and the joint decision. That is how much the husband had to compromise, relative to how much the wife had to compromise. sales (28%). We also find 12% increases in the probability that a woman is the primary manager or registrant for a cane block, which are both significant at the 5% level. Consistent with our other findings, these effects are also larger for women who are less empowered at baseline, representing effect sizes of 43% for cane sales management and 18% for being the primary manager or registrant for a block, effects significant at the 5% and 1% levels respectively. However, beyond the impacts on communication, norms, and cash cropping we did not find any other statistically significant impacts of the workshop. In particular, we did not observe changes in other types of marital communication, beyond the family visioning taught at the workshop, nor changes in women’s role in household decision making. Not only were these estimated effects not statistically different from zero, the coefficients were also quite small in magnitude, representing effect sizes of less than 1%. Further, we did not observe the hypothesized changes in women’s time allocations to housework and leisure. In fact, the estimated coefficients indicate possible increases in housework and decreases in leisure, though we cannot reject that the effects are zero. We find that the workshops did have impacts on meaningful indicators of women’s empowerment within their households. Women have stronger beliefs in gender equality, which is a foundational element of achieving equality, and women have stronger roles in the key income-generating activity of the household, which increases their access to resources. However, the downstream impacts of these initial changes had not yet manifested by the time of the midline data collection. Perhaps more time is required for increased communication about family vision to spill over to other types of marital communication, or for increased access to income-generating activities to affect bargaining weight in household decision making. Given more time, women’s changed views about gender norms may shift the balance of housework within their households. However, it may be true that men’s views of gender norms are far more important in these determinations than women’s. Given the limits of the midline data collection to interviewing only women, we are unable to make further conclusions about this in this study. In future work we will use the endline data to further these analyses and draw more nuanced conclusions. References Donald, A. & Rouanet, L. (2018). Status Update? The Impacts of Making Women’s Work Visible on Their Intra-Household Status. Gender Innovation Lab, Office of the Chief Economist, Africa Region, World Bank. Draft paper. Gupta, J; Falb, KL; Lehmann, H; Kpebo, D; Xuan, Z; Hossain, M; Zimmerman, C; Watts, C; Annan, J (2013) Gender norms and eco- nomic empowerment intervention to reduce intimate partner violence against women in rural Cote d'Ivoire: a randomized controlled pilot study. https://bmcinthealthhumrights.biomedcentral.com/track/pdf/10.1186/1472-698X-13-46 Ismayilova, L., Karimli, L., Gaveras, E., Tô-Camier, A., Sanson, J., Chaffin, J., & Nanema, R. (2017). An Integrated Approach to Increasing Women’s Empowerment Status and Reducing Domestic Violence: Results of a Cluster-Randomized Controlled Trial in a West African Country. Lecoutere, E. & Van Campenhout, B. (2018). Joint forces - The impact of intrahousehold Cooperation on welfare in East African agricultural households. IOB working Paper 2018-11agricul Lowe, M & McKelway, M. (2017).tBargaining Breakdown: Intra-Household Decision-Making and Women’s Employment. http://barrett.dyson.cornell.edu/NEUDC/paper_376.pdf Mayoux, Linda (2014). GALS at Scale. www.galsatscale.net. Oxfam Novib (2014) Rocky Road to Diamond Dreams. GALS Phase 1 Visioning and Catalysing a Gender Justice Movement Implementation Manual, V1.0. https://www.oxfamnovib.nl/redactie/Downloads/English/SPEF/140701_RRDD_manual_July_small(1).pdf Riisgaard, L., Fibla, A.M., Ponte, S. (2010). Gender and Value Chain Development. Ministry of foreign Affairs of Denmark, International Development Cooperation. Tables Table 1. Baseline Demographics Mean SD Min Med Max N Husband age 46.61 12.67 21 45 96 2,370 Designated wife age 38.32 11.08 18 38 89 2,370 Household is polygamous 0.34 0.47 0 0 1 2,370 Length of marriage to designated wife 19.52 11.74 0 19 64 2,370 Husband is literate 0.85 0.36 0 1 1 2,370 Husband's years of formal schooling 7.38 3.68 0 7 17 2,369 Designated wife is literate 0.63 0.48 0 1 1 2,370 Designated wife's years of formal schooling 5.59 3.65 0 6 17 2,368 Absolute age difference between husband & wife 8.46 6.78 0 7 54 2,370 Absolute difference in spouses’ years of schooling 3.46 2.79 0 3 17 2,367 Number of household members 8.7 3.39 2 8 28 2,370 Number of children (under 18) 4.99 2.53 0 5 19 2,370 Number of children 5 years old or under 1.4 1.15 0 1 7 2,370 Table 2. Means of outcomes for full sample Std. Mean Dev. Min Max Obs Strongly disagree that The important decisions in the family should 0.45 0 1 2,225 be made only by the men in the family A wife should tolerate being beaten by her 0.50 0 1 2,226 husband in order to keep the family together It is better to send a son to school than it is to 0.95 0 1 2,225 send a daughter Strongly agree that If the wife is working outside the home, then 0.58 0 1 2,227 the husband should help with household chores A wife has the right to express her opinion, 0.74 0 1 2,228 even if she disagrees with her husband Average of belief in gender equality 3.18 0.62 1 4 2,228 (4=strongly support) Ever talked about family vision 0.96 0 1 2,228 Talk about family vision at least a few 0.79 0 1 2,228 times/year Talk about family vision at least once/month 0.43 0 1 2,228 Usually discuss major household decisions 8.19 2.20 0 10 2,223 (10=strongest agreement) Like to discuss personal concerns with 7.43 2.34 0 10 2,223 husband (10=strongest agreement) Decision making power index 2.98 0.61 1 5 2,227 Hours per week on housework 25.45 13.73 0 97 2,227 Hours per week on leisure 8.94 8.26 0 84 2,227 Table 3. Main Results Norms Discuss Other Decision House- Average Vision Communication Making work Leisure (1) (2) (3) (4) (5) (6) Workshop 0.043* 0.047** -0.040 0.017 0.940 -0.236 (0.026) (0.020) (0.084) (0.026) (0.582) (0.350) Control Mean of Dep. Var. 3.158 0.376 7.828 2.967 24.954 9.056 N 2228 2370 2223 2227 2227 2227 R2 0.033 0.039 0.041 0.039 0.031 0.034 Note: Dependent variables are given in column headers. All regressions control for stratum fixed effects. Standard errors are shown in parentheses. Coefficients are statistically significantly different from zero at the 10% (*), 5% (**), or 1% (***) level as indicated. Table 4. Impact on beliefs about gender norms Decisions Chores Opinions Abuse (1) (2) (3) (4) Workshop 0.055 0.133*** -0.023 0.006 (0.051) (0.045) (0.041) (0.044) Control Mean of Dep.Var. 2.810 3.092 3.322 3.407 N 2225 2226 2227 2228 R 0.028 0.032 0.027 0.041 Note: Dependent variables are shown in column headers and refer to the questions about gender norms shown in Table 2. The question regarding child schooling is not included due to lack of variation. Variables are ratings on a 4-point scale, where 4 indicates agreement or disagreement as shown in Table 2. All regressions control for stratum fixed effects. Standard errors are shown in parentheses. Coefficients are statistically significantly different from zero at the 10% (*), 5% (**), or 1% (***) level as indicated. Table 5. Impacts on participation in cane production, management, and sales Any management role For any block Production Sales Does labor Does Mgmt Primary Mgr. Registrant (1) (2) (3) (4) (5) (6) Workshop 0.023 0.018* 0.029* 0.011 0.039** 0.043** (0.019) (0.011) (0.015) (0.019) (0.020) (0.020) Control Mean of Dep.Var. 0.684 0.065 0.825 0.704 0.331 0.342 N 2370 2370 2370 2370 2370 2370 R 0.033 0.038 0.032 0.038 0.035 0.026 Note: All regressions control for stratum fixed effects. Standard errors are shown in parentheses. Coefficients are statistically significantly different from zero at the 10% (*), 5% (**), or 1% (***) level as indicated. Table 6. Baseline incentivized decision measures of household efficiency and bargaining power Std. Mean Dev. Min Max Obs Wife prefers to control resources 0.44 0 1 2,370 Wife is willing to pay to control resources 0.24 0 1 2,370 Husband prefers to control resources 0.40 0 1 2,370 Husband is willing to pay to control resources 0.20 0 1 2,370 Share husband prefers to give wife 0.44 0.24 0 1 2,365 Share wife prefers to give wife 0.52 0.25 0 1 2,365 Share joint decision gives to wife 0.46 0.20 0 1 2,365 Distance from husband's preference to joint decision 0.15 0.20 0 1 2,365 Distance from wife's preference to joint decision 0.22 0.21 0 1 2,365 Table 7. Heterogeneous effects by wife’s access to resources at baseline Norms Discuss Any management role For any block Average Vision Production Sales Does Does Primary Registrant labor Mgmt Mgr. (1) (2) (1) (2) (3) (4) (5) (6) Workshop 0.056* 0.054** 0.035* 0.026** 0.029* 0.025 0.059*** 0.060*** (0.029) (0.023) (0.021) (0.012) (0.017) (0.021) (0.022) (0.022) Wife not willing to 0.074 0.040 0.017 0.022 0.001 -0.014 0.036 0.030 pay (0.046) (0.037) (0.033) (0.019) (0.027) (0.033) (0.034) (0.035) Workshop x Not -0.056 -0.028 -0.065 -0.037 -0.001 -0.075 -0.095* -0.085* willing to pay (0.065) (0.052) (0.047) (0.027) (0.038) (0.046) (0.049) (0.050) Mean 3.147 0.392 0.679 0.061 0.823 0.706 0.324 0.335 N 2,228 2,228 2,370 2,370 2,370 2,370 2,370 2,370 R2 0.005 0.010 0.007 0.012 0.004 0.013 0.009 0.001 Note: Willingness to pay to control resources indicates inability to access sufficient resources in the household. Not willing to pay therefore indicates a resource-empowered wife. All regressions control for stratum fixed effects. Standard errors are shown in parentheses. Coefficients are statistically significantly different from zero at the 10% (*), 5% (**), or 1% (***) level as indicated Appendix A This section describes the workshop in greater detail. It heavily references the document provided by Oxfam Novib (2014) that is available here. Family Vision Workshops Developed jointly by IFPRI and Oxfam August, 2015 Family Vision Workshop vs. GALS Change Catalyst Workshop Similarities & Differences We summarize here some key differences between the FVW and traditional GALS: both the GALS methodology generally and the GALS CCW specifically. 1. Objectives The aims of FVW are the same as the aims for GALS CCW, with the exception of the third bullet point, which is NOT an objective of FVW (see box). The hope is that by accomplishing these aims, the workshop will indirectly open the minds of men to the possibility that it may benefit them and their household for their wives to have greater participation in the sales of sugarcane and the decision-making regarding use of the profits. Another possible indirect impact may be to empower women to feel capable of contracting directly with the sugar company and to exert their preferences alongside their husbands in terms of how the profits should be used. 2. Participation Participation in the FVW will be based on assignment by the study team according to specific protocols that will allow us to analyze impact. It is very important that the households mobilized by our staff are the beneficiaries of the FVW. In each selected household we will aim to have both the husband and (at least one) wife participate. 3. Dissemination The FVW is intended as a one-time workshop to spark change within the participating households. At this stage, we are not planning to conduct the other GALS activities, such as Phase 1 parts 2 & 3 (community action learning & participatory gender reviews) or later phase activities, as discussed above. Therefore, significantly less workshop time will be dedicated to preparing participants to disseminate what they have learned. While we recognize that community dissemination may be beneficial, it is not the focus of this study and we must use the limited workshop time carefully to achieve the FVW aims. Preparing for the Workshops In order for the facilitators to become familiar with the sugarcane value chain and the current gender dynamics in target households, we are planning for one day in each target village for key informant meetings. This may include activities normally conducted during the “mapping the value chain” activities of GALS. Facilitators will visit the villages in which workshops will be held, and will be accompanied by the Project Manager, Ivan Ssenkubuge, and the Field Manager, Dennis Baliddawa. The team should begin by introducing themselves to the relevant local authorities in order to identify key informants. Both in this initial visit and during all field activities, staff should be aware that revealing too much detail regarding the nature of the research can affect the extent to which we are able to accurately answer the research questions. For example, if all villagers are aware of the exact research questions, this may impact willingness to attend workshops or could bias the answers given during data collection interviews. FVWs should be presented as a service provided by Oxfam, funded by IFPRI and Kakira Sugar. The aims of the FVWs should be described as in the box above. During the initial village visit, the team should, in collaboration with local authorities, identify a suitable location for conducting the workshop. This may be a church or school room. An ideal location will be central to the village, perhaps near a major marketplace so that it is not burdensome for the participants to come and go each day. This would also be an ideal time to identify a local community member willing to provide a hot lunch and normal refreshments during workshop days at a reasonable rate. In short, as much as possible should be prepared in advance during this visit. Capacity Building Workshop Tools As noted above, the FVW will make use of existing GALS CCW tools, with the exception of the “taking GALS home” tools that focus on dissemination. Below is a proposed agenda. Day 1 CCW Session 1 Pairwise Introductions & Soulmate Visioning CCW Session 2 Vision Journey Day 2 CCW Session 3 Gender Balance Tree *potential additional activity this day related to men’s and women’s roles in sugarcane production, marketing, sales, and use of profits. Day 3 CCW Session 4 Empowerment Leadership Map CCW Session 6 Multilane Highway Action plan (6.1 and 6.3 only)