MTPmle: A SAS Macro and Stata Programs for Marginalized Inference in Semi-Continuous Data

Delia C. Voronca, Mulugeta Gebregziabher, Valerie Durkalski-Mauldin, Lei Liu, Leonard E. Egede

Main Article Content

Abstract

We develop a SAS macro and equivalent Stata programs that provide marginalized inference for semi-continuous data using a maximum likelihood approach. These software extensions are based on recently developed methods for marginalized two-part (MTP) models. Both the SAS and Stata extensions can fit simple MTP models for cross-sectional semi-continuous data. In addition, the SAS macro can fit random intercept models for longitudinal or clustered data, whereas the Stata programs can fit MTP models that account for subject level heteroscedasticity and for a complex survey design. Differences and similarities between the two software extensions are highlighted to provide a comparative picture of the available options for estimation, inclusion of random effects, convergence diagnosis, and graphical display. We provide detailed programming syntax, simulated and real data examples to facilitate the implementation of the MTP models for both SAS and Stata software users.

Article Details

Article Sidebar