Presented by:

T.E. Raghunathan & Roderick J. Little

University of Michigan

Thursday, April 20, 2000

Abstract

Many probability surveys conducted by, or on behalf of, the National Center for Health Statistics (NCHS) repeatedly sample subjects from the same geographical areas or Primary Sampling Units (PSUs).  For example, the National Health and Nutritional Examination Survey (99-03) will be sampling subjects from the same PSUs as in the National Health Interview Survey (NHIS).  A tighter link occurred when the National Survey of Family Growth (Cycle V) sampled women in the age range 15-44 from the pool of the NHIS respondents.  Such linked studies may lead to useful auxiliary data for nonresponse adjustments.  The linkages provide a rich set of contextual or neighborhood level variables that may inform individual level relationships and improve the efficiency of estimates by borrowing strength across surveys.  This talk, third in a series sponsored by the Michigan Center for Excellence in Health Statistics (funded by NCHS), will focus on methodological issues in the analysis of linked survey data sets.  Methodological approaches will be illustrated primarily using NSFG-NHIS linked survey data, but general methodological issues will be discussed using other national surveys.