Abstract:

Information from various public and private data sources of extremely large sample sizes is now increasingly available for research purposes. Statistical methods are needed for utilizing information from such big data sources while analyzing data from individual studies that collect more detailed information to address specific problems. In this talk, we consider regression analysis with individual-level data from an “internal” study while utilizing summary-level or crude information, such as information on reduced models, from an “external” big data source. The constraints that link internal and external models are identified and used to develop a semiparametric maximum likelihood inference framework, which allows for both the settings where the covariate distribution in the internal sample is the same as or different from that in the external data sources. Extensions for handling complex stratified sampling designs, such as case-control sampling, for the internal study are also considered. Asymptotic distribution theory is developed. We use simulation studies and a real data application to assess the performance of the proposed methods.