Intervention (Hidden)
The Smart Data project examines how providing administrators with visualized administrative data affects their knowledge of their locality’s implementation of MGNREGA, India’s large-scale public works program, and whether receiving visualized versions of administrative data affects administrators’ data appetite and usage. The study population for this project is Bihar’s District Magistrates, District Development Commissioners, and MGNREGA Block Program Officers, of which there are approximately 575 officers throughout the state. Our study partner is Bihar’s Rural Development Department.
The study is a randomized control trial that randomizes receipt of MGNREGA ranking dashboards at the district level. The state’s 38 districts will be stratified based on agroclimatic zones (Northern West, Northern East, Southern East, and Southern West), followed by pair matching within stratum based on districts’ average previous ranking for available months (June 2013 – February 2014). The ranking system, which tracks year-to-date performance in major implementation areas such as works completion , timely wage payments, and implementation of social audits, was developed by Bihar’s Rural Development Department. Ranking dashboards provide administrators with easy access to user-friendly information on their performance over time with respect to the different indicators and compared to comparator groups using tables, graphs, and maps. Block-level dashboards provide this information for blocks and Gram Panchayats, India’s lowest administrative division, in a similar fashion. The control districts in the study will be sent a non-visualized version of this information in order to isolate the causal impact of the dashboards from changes in information provision or contact regarding MGNREGA rankings. District and block rankings will be constructed monthly by our study team using MGNREGA administrative data. PDF versions of the rankings dashboards will be sent by email each month to officials in treatment districts, and an Excel spreadsheet with ranking information will be sent to officials in control districts by email at the same time each month.
We will investigate several areas in which officials may be affected by the dashboards. First, do dashboards change the accuracy of district and block officials’ perceptions on their performance in MGNREGA? To answer this, we will track accuracy of officials’ reports of their absolute and relative performance in MGNREGA throughout the study period. Belief accuracy will be investigated using deviations and absolute deviations of officials’ reports of how well they perform in MGNREGA on specific indicators (ranking and actual values of indicators) relative to administrative data and rankings.
We also will examine whether dashboards change officials’ demand for administrative data and data use practices. Behaviors will be examined comparing officials’ self-reported methods of obtaining information, frequency of consulting administrative data, and related measures. Web usage statistics of the MGNREGA MIS at the district level will also be used to understand if receiving visualized reports increases demand for administrative data more generally.
Finally, we will track performance in MGNREGA, both generally and with respect to specific rankings, over time to understand whether officials with visualized information more effectively improve program outcomes compared to those in the control group. It is important to note at the outset that we do not a priori expect dashboards will impact these outcomes since although dashboards may help officials identify program successes and failures, they do not highlight how to improve implementation.