Roundtable:
Big Puzzles, Big Promise: Using Linked State Administrative Data for Policy Research
(Methods and Tools of Analysis)
*Names in bold indicate Presenter
Policy researchers have long drawn on administrative data to examine outcomes such as employment and earnings using Unemployment Insurance records or educational outcomes using students’ transcripts and enrollment records. Using records created in the process of administering routine public functions is cost-effective and often allows describing conditions or examining policy changes for a full population rather than a select group of survey respondents. Despite these advantages, administrative data comes with big puzzles in the form of limited or lacking data. Administrative records often contain partial information about possible outcomes (earnings rather than all income, for example) or lack demographic or contextual information that important to understanding complex social processes (for instance, school records largely lack data on children’s home environments).
Recent methodological and infrastructure advances allowing for new linkages between data sources hold promise for addressing some of these limitations. This roundtable will bring together leaders from four groups using innovating linked state administrative data sources. Participants include academic researchers and a state practitioner, and the chair has a distinguished record spanning policy and academic settings. The group will discuss questions about the logistical, relational, and technical challenges associated with using state data; share some ways in which they have met those challenges; and consider how state data resources and research uses will evolve over the next decade.
Rodney Andrews (UT-Dallas) will discuss the Texas School Project, which supports and conducts research using individual level administrative data following Texas students from pre-K through college and into the Texas workforce. For example, Andrews and colleagues have documented the extent of heterogeneity in the effects of college quality on earnings as well as variation in the pathways by which students have navigated their post-secondary enrollment.
Christina Gibson-Davis (Duke) will discuss matching individuals both within and across data sets, using common markers such as names, addresses, etc. She will also discuss how to perform limited matches when confidentiality concerns prohibit matching on these common markers. As a case study, she will discuss her work with the North Carolina Detailed Birth Record Database, which contains the records of all births that occur in the state.
Sarah Leopold (Colorado Department of Human Services) will discuss her group’s efforts to merge survey records with TANF and child support case records to better understand conditions of TANF leavers and evaluate a child support pass through experiment.
Mark Long (University of Washington) will discuss collaboration with the state’s Department of Social and Health Services to build a new data resource, the Washington Merged Longitudinal Administrative Data, which was developed to better understand the impact of minimum wage ordinances in Seattle and elsewhere on family well-being. A key challenge in developing the data for this question is the need to construct families from records that do not identify all family member’s locations contemporaneously.
Chair Amy O’Hara, the Former Chief of the Center for Administrative Records Research and Applications at the U.S. Census Bureau, is an expert in administrative data use in general and record linkage and validation in particular.
