INVESTIGATION OF NONRESPONSE BIAS IN NHANES II1

Abstract

Forthofer, R. N. (U. of Texas School of Public Health, Houston, TX 77025). Investigation of nonresponse bias in NHANES II. Am J Epidemiol 1983; 117: 507–15. In the second National Health and Nutrition Examination Survey (NHANES II), there was a 27% nonresponse rate in the examination phase. This report investigates the potential bias in these data due to this large nonresponse rate. Data from a household and medical history interview are used In the investigation of factors related to examination status. In addition, data from the examined group are compared to data from the 1976 National Health Interview Survey (NHIS). Since there was only a 3.7% nonresponse rate for the 1976 NHIS, proportions calculated from these data represent reasonable estimates of the true population values. Several variables have a significant association with the interview and examination status. However, it appears that the nonresponse and poststratiftcation adjustments performed by the National Center for Health Statistics have removed most of these factors as sources of bias. There Is excellent agreement in the marginal distribution of variables between NHANES II for examined persons and the 1976 NHIS.