A Comparison of Two Drop-out Weighting Schemes in the Analysis of Clustered Data with Categorical and Continuous Responses

A model previously proposed for analyzing clustered data with a bivariate discrete and continuous response jointly is extended to a model that jointly analyzes categorical and continuous responses. As well, two previously proposed weighting schemes for generalized estimating equations that compensate for drop-out effects are investigated for unbiasedness and efficiency revealing conditions under which one scheme can be less efficient and more susceptible to influential observations. The data set used for demonstration is longitudinal, containing information on nests of young bass gathered over the life of the nest.