Using Administrative Data to Improve Decision Making.

Position:News & Numbers

States traditionally use administrative data to prepare annual reports showing how funds were spent and the impact of a particular program, to demonstrate transparency in describing what a state agency does, and to comply with performance measures set by the federal government, state legislature, governor, or an agency.

More recently, states have begun harnessing existing information through data analytics--procedures that review data to identify meaningful information and correlations. Such efforts provide additional opportunities for governments to make effective decisions. Analysts can uncover important insights by employing techniques such as integrating and cross-referencing data sets, undertaking calculations to show trends, finding correlations, running statistical experiments, mapping geographical data to show areas of high activity, and visualizing data in charts and graphs. Additionally, data analytics can reveal the root cause of a persistent issue, diagnose breakdowns in a system, highlight obstacles, and predict future phenomena, helping leaders make more strategic decisions.

In a recent study, How States Use Data to Inform Decisions, Pew Charitable Trusts used data collected from interviews with more than 350 state officials to highlight ways in which some government leaders have employed sophisticated data analytics to craft policy responses to complex problems, improve service delivery, manage existing resources, and examine policy and program effectiveness. In one example, the District of Columbia performed a randomized controlled trial using administrative data to assess the best way to boost participation in its Summer Youth Employment Program. The trial revealed the effect of different strategies on program attendance, helping administrators choose the most effective course of action.

The study identified five actions state leaders can take to work through the challenges they face in maximizing the value of administrative data...

To continue reading