Here we propose to implement recent methodological advances on generalized linear mixed effects models (glmm's) in response to research topics 7B2 and 8C6 in the program description for NIMH. Generalized linear mixed effects models (glmm's) can be used to analyze data commonly occurring in longitudinal studies. Removing the random effects, these models are also common in the analysis of cross-sectional data. With additional research, recent methodological advances should make it possible to develop algorithms that routinely compute glmm estimates for use on longitudinal data. We propose to develop comprehensive and easy- to-use commercial software implementing these algorithms. Because these algorithms will be difficult for novice users to understand and use, we further propose to deliver expert data-analytic advice on the use of the glmm algorithms. This advice, directly written in HTML, will include an online tutorial with laboratory exercises illustrating and explaining the models and algorithms, a decision network that will make it easier for users to select an appropriate analysis, worked examples that users can use as a template in their own analyses, links to related information on the web, and extensive online help. Multimedia materials will be included where appropriate.
Generalized linear mixed effects models are a psychometric method widely used in psychology, the social sciences, and other areas of behavioral and scientific research. The models we propose to implement have significant advantages over existing techniques. The tutorials, decision network, and guidance provided with the system will make these methods accessible to a wider audience of researchers. The proposed product should find a ready market.
