Predictive margins with survey data

In the analysis of covariance, the display of adjusted treatment means allo ws one to compare mean (treatment) group outcomes controlling for different covariate distributions in the groups. Predictive margins are a generaliza tion of adjusted treatment means to nonlinear models. The predictive margin for group represents the average predicted response if everyone in the sam ple had been in group r. This paper discusses the use of predictive margins with complex survey data, where an important consideration is the choice o f covariate distribution used to standardize the predictive margin. It is s uggested that the textbook formula for the standard error of an adjusted tr eatment mean from the analysis of covariance may be inappropriate for appli cations involving survey data, Applications are given using data from the 1 992 National Health Interview Survey (NHIS) and the Epidemiologic Followup Study to the first National Health and Nutrition Examination Survey (NHANES I).