American Journal of Political Science,July 2008,v. 52,iss 3,pp. 585-602. Available From:Link to Source
This paper evaluates the Women’s Group Project, an intervention implemented by the Dutch nongovernmental organization International Child Support in the Busia and Teso districts of rural western Kenya. Weak civic organizations and low participation rates in community associations impede the political representation and economic advancement of poor and disadvantaged groups, especially women, in many developing countries. To address this problem and strengthen local women’s agricultural organizations, the International Child Support project provided two days of leadership and management training for group leaders, agricultural tools and seeds for member use and training in agricultural techniques.
This study examines the programme’s impact on agricultural production; group strength as measured by member evaluations, attendance rates, frequency of meetings and assistance to members; groups’ interactions with local political and social institutions as measured by community fundraising contributions, visits to group members by extension workers and local government officials and grants received by the group; and group membership patterns, including applications for membership, entry and exit.
The basic methodology is a randomised controlled trial (RCT) that exploits variation in the timing of programme implementation to evaluate impact in the project’s first phase. During this phase, 40 of 80 operational women’s associations involved in agricultural activities were randomly assigned to receive treatment. The remaining associations received treatment 2 years later but, for the purposes of this evaluation, comprised the comparison group. This study employs both group- and individual-level data collected from a baseline survey administered prior to randomisation and implementation and post-intervention surveys conducted 14 and 20 months after the initial survey. Impact is estimated by regressing outcomes on a dummy treatment indicator using either ordinary least squares (OLS) or probit, as appropriate for the outcome measure.
This study reports insignificant post-project differences between the treatment and comparison groups on nearly all measures of agricultural output, group strength and community interaction, with the exception of visits from agricultural and health extension agents and local government officials. On these community interaction indicators, treated associations were 75 percent and 50 percent more likely to have received visits, respectively, than the control group.
The programme was associated with significant effects on group membership patterns. The number of applicants to programme groups was 40 percent higher than to control groups, and treated groups reported twice as many new members during the project period. Further, new members of programme groups were generally of higher socioeconomic status as measured by formal-sector income and educational attainment. There was, however, no significant difference in overall exit rates between treated and control groups, but exits were more likely to be due to conflict in the treated groups, and women over age 50 were significantly more likely to leave treated associations. Finally, leadership turnover was 20 percentage points greater in treatment associations, and both men and well-educated women were significantly more likely to take on leadership roles in treated groups.
These findings suggest that the project failed to achieve its goal of strengthening women’s organizations and had negative effects on participation by socioeconomically disadvantaged persons, especially older women. These findings are generalised in a formal model specifying the impact of external funding on group membership dynamics. This model suggests that external funding makes membership more attractive to higher-status persons, thereby crowding out the poorest and most disadvantaged members. From a policy perspective, this suggests that external funding for associations serving disadvantaged populations should be designed in ways not to appeal to elites or to disproportionately benefit the most disadvantaged members so as to discourage elite capture.