Federal agencies want to release data with fine resolution and many attributes; yet, it is increasingly expensive to mount new data collections with such detail. The Census Bureau and many federal agencies recognize that integrating information from multiple data sources, such as administrative records or data from other agencies, is a potentially cost-effective way to increase accuracy, usefulness, and timeliness of data products. However, integrating information across data sources raises a variety of practical and technical issues, the latter of which will be a focus of TCRN research. Specifically, the TCRN will
- develop methods that agencies and secondary analysts can use to properly account for uncertainty in inferences in imperfect record linkage settings, as well as to pass on that uncertainty in public use data products via multiply-imputed datasets; and,
- develop statistical approaches to combining information from multiple data sources that do not depend on record linkage.
These methodologies will be applied to create two data products. First, we will use administrative records to improve imputation of missing and erroneous data in the SIPP. This project will serve as a case study for federal agencies on how to utilize administrative records to improve survey data. Second, we will link Census, Medicare, and death certificate data to improve estimates of mortality and its correlates for population subgroups, in the process developing improved methodology for demographic analyses.