THE MUSCATINE CHILDREN OBESITY DATA REANALYZED USING PATTERN MIXTURE-MODELS

A set of longitudinal binary, partially incomplete, data on obesity am ong children in the USA is reanalysed. The multivariate Bernoulli dist ribution is parameterized by the univariate marginal probabilities and dependence ratios of all orders, which together support maximum likel ihood inference. The temporal association of obesity is strong and com plex but stationary. We fit a saturated model for the distribution of response patterns and find that non-response is completely at random f or boys but that the probability of obesity is consistently higher amo ng girls who provided incomplete records than among girls who provided complete records. We discuss the statistical and substantive features of respectively pattern mixture and selection models for this data se t.