Evaluating the Effect of Program Intervention on Network Building
Manoj ShresthaHow networks can be manipulated to effect change has been a question of great interest to both scholars and practitioners in recent years. Two related questions regarding network effectiveness are (a) evaluating antecedents of networks, and (b) evaluating consequences of networks. This paper addresses the first question by applying network-based comparison group evaluation to determine whether rural communities (villages) with program exposure are different compared to their counterpart without the program in their preference of or proclivity for network structures.
To investigate the proposition, this research uses the case of rural communities in Nepal that applied for funds for their drinking water project from the World Bank supported Rural Water Supply and Sanitation Program (the Program). While the immediate goal was to improve access to clean drinking water, the program also hoped that the communities will learn from this assistance and will pursue self-help development activities on their own after the completion of the funded project. The program assistance included engineering support and training in project management and communication skills. The program also encouraged communities to build contacts with organizations to mobilize various resources to help implement their projects. Those supports involved meeting regulatory compliance, securing matching fund, resolving conflicts, and mobilizing political support.
The sample consists of 125 communities from five districts in the Central Development Region of Nepal that applied for funds from the program. Out of 125 communities that applied, 62 communities received funds to implement the water projects. Those 62 funded communities served as communities with program exposure, and the unfunded ones served as communities without the program exposure. The data on the communities’ networks with organizations was collected from the field survey of both groups of communities after two years from the time of funding and again after eight years from the time of funding. Each community, represented by the elected Water User Committee, was interviewed to collect data on who they contacted and how often for different resources needed to pursue self-help community development activities such as village roads, irrigation channels, school building, and community centers.
Bipartite exponential random graph models will be estimated using XPNet program available at http://www.melnet.org.au/pnet/. These models will include network structures and a binary attribute, coded 1 for funded communities and 0 for unfunded communities. The binary attribute effect’s interaction with the network structures will provide a test of the proposition whether the network structures and the proclivity for those networks of the funded communities were different from the unfunded ones. Community attributes such as their size and geographic distance to the district headquarters will be included to account for the differences across the communities. The preliminary data analyses indicate that the funded communities tend to prefer greater expansive and bridging networks compared to the unfunded communities.
This research contributes to the network effectiveness literature. It shows how programs can be used to manipulate networks to affect or manage change. It also advances the development of network-based evaluation methods.