This study will develop statistical methods for the analysis of data from complex sample surveys and test them empirically using the National Health and Nutrition Examination Survey I and II (NHANES). The research concentrates on ways to test hypotheses about the coefficients that derive from multiple linear regression of survey data. Existing methods which use Wald test statistics are compared to newly developed approaches based upon Bonferroni t-statistics and jackknifed Wald statistics. These regression methods have been applied to the NHANES data sets to determine if they can be used to provide ways to analyze the complex relationships of growth and nutrition. Survey statistical methods proper for the National Health Interview Survey (NHIS) have begun. Data sets from the Child Health Supplement of NHIS 1988 will also be examined. The project is proposed to be completed by the end of 1993.
