Building Better Longitudinal Surveys (on the cheap) Through Links to Administrative Data

I make four recommendations regarding the longitudinal surveys of the National Center for Education Statistics. I recommend that NCES: (1) Supplement its surveys with administrative data, focusing survey efforts on collecting data that is not contained in administrative sources. (2) Use administrative data to convert what are now cross-sectional surveys into longitudinal surveys. (3) Support researcher-initiated requests to link existing NCES surveys to administrative data sources. (4) Explore methods to make microdata more widely available to researchers, particularly by replicating successful practices at Census. I use the example of student borrowing to explore the rationale for these recommendations, discussing the data needs of researchers and policymakers in this arena. I describe prospects for fulfilling those needs with a combination of NCES surveys and administrative data