Development of Integrated Statistical Software for Clustered & Longitudinal Data

Scientific Software International, Inc., Lincolnwood IL

Investigators

Abstract

A common theme in mental health is multi-stage or multi-level sampling. For example, in logitudinal studies, subjects are repeatedly sampled and rated on measures of mental and physical level of functioning, well-being, and quality of life. These subjects are sample from a population, often stratified on the basis of the services they receive. Outcome measures typically exhibit individual differences that are of great interest to mental health researchers. In Phase I, we developed a prototype computer program SuperMix for the analysis of mixed-effects regression models with continuous outcome variables that has established the solid ground for a Phase II project. However, in practice, the focus of mental health services research is often mixed-effects regression models with categorical response variables or models for the analysis of time-to-event data and counts. The objective of this SBIR proposal is to develop and market SuperMix as a user-friendly integrated platform that includes level-2 and level-3 modeling of all the response variable types, graphical display of data and online help documentation. A SuperMix manual and primer, outlined in Phase II will be written; the primer to be translated into Spanish. These, and additional training and educational material will be based on health sciences data.