Smart data: Can Visualised Administrative Data Help Inform and Hold Public Stakeholders Accountable?

“Smart” data collection (using smartphones, biometrics, etc.) receives attention as a means to augment low-income states’ capacity. However, little attention has focused on how monitoring and accountability policies can harness extensive administrative data already being generated to improve programs and public accountability. We study how to turn big data into “smart” data in India’s large public works programme, MGNREGA, by examining how big data presentation affects administrators’ understanding of their performance and their appetites for, and usage of, programmatic data.

The randomised control trial, implemented with support from Bihar’s Rural Development Department, will provide district and block officials in half of Bihar’s 38 districts with visually appealing dashboards built from MGNREGA administrative data that highlight districts’ and blocks’ absolute and relative performance in MGNREGA. These dashboards will track year-to-date performance in major implementation areas such as execution against labor budget plans, timely wage payments, implementation of social audits, and female participation. Information will be presented in a user-friendly format (tables, graphs, chloropleth maps) so officials can quickly ascertain their performance over time and in reference to comparator groups. Block-level dashboards will provide similar information for blocks and Gram Panchayats, India’s lowest administrative division. Relevant officials in control districts will be sent a non-visualised version of this information in order to isolate the causal impact of the dashboards from changes in information provision or frequency of contact with study or government officials.

The study, begun in mid-2014, launched a baseline survey with district and block officials in the fall of 2014 to capture pre-intervention professional responsibilities and attitudes; perceptions of current performance in MGNREGA; personal attributes such as political aspirations, personality traits and public service motivation; and more. An endline survey will be conducted after one year of implementation.

Throughout the intervention, random subsets of officials will be queried on a regular basis to understand whether dashboards affect the accuracy of officials’ perceived performance in MGNREGA. Officials’ reported demand for administrative data and data use practices will also be tracked to understand if dashboards change appetites for, and use of, data. The study will also track local performance in MGNREGA to understand whether officials with visualised information improve programme outcomes compared to those in the control group. It is important to note that a priori, dashboards are not necessarily expected to impact implementation outcomes given they do not provide administrators with assistance in diagnosing the cause of implementation failures highlighted.

At its most basic level, this study will establish whether big data can be turned into “smart” data for administrators in capacity-constrained settings. By capturing information on officials’ demographic characteristics, personality traits, motivation to serve in the public sector, data usage patterns, and more, it will also generate rich insights on the circumstances under which such smart data initiatives are most effective.

Outputs