The Constrained Categorical Regression (CCR) methodology was originally developed in a successful Phase I project. The long-term objective of this Phase II proposal is to develop-the CCR methodology as a user- friendly software package that naturally integrates prior knowledge as logical causal relationships directly into the analysis of categorical dependent-variable data. Such a statistical tool will greatly increase current knowledge of alcohol-related conditions by making more effective use of existing and future institutional, county, state, and national level data bases.
The specific aims of the Phase II project are: 1) utilize the CCR approach to analyze data from alcohol research, 2) develop a UNIX-based user-friendly CCR software package, and 3) further refine the Phase I CCR methodology through additional algorithm design and development. Thus, the expected Phase II results will include: 1) a better under- standing of a particular alcohol-related problem obtained through application of the CCR modeling approach, and 2) a marketable prototype CCR modeling software product. Because the incorporation of prior knowledge into the areas of epidemiologic, clinical, and social research, CCR modeling has tremendous potential for successful development as a commercial data analysis product.
