DEVELOPMENT OF ULTRA FRIENDLY EPIDEMIOLOGICAL STATISTICAL PA

THIS APPLICATION PROPOSES TO DEVELOP AND DISTRIBUTE AN ULTRA FRIENDLY, FULL-SPECTRUM MICROCOMPUTER STATISTICAL PACKAGE FOR EPIDEMIOLOGISTS WHICH CULMINATES IN A CORE OF STATE-OF-THE-ART REGRESSION TYPE MODELS FOR BINARY AND EVENT TIME DATA, AND WHICH INCORPORATES STATE-OF-THE-ART DIAGNOSTIC PROCEDURES AND COMPUTATIONAL ALGORITHMS. SPECIFICALLY INCLUDED ARE LOGISTIC REGRESSION, CONDITIONAL LOGISTIC REGRESSION, LOGISTIC REGRESSION WITH RANDOM EFFECTS, AND COX REGRESSION, AS WELL AS OTHER GENERALIZATIONS OF THE METHODS DESCRIBED IN ROTHMAN AND BOICE (1979). EMPHASIZED IS ULTRA FRIENDLINESS, SOPHISTICATED MODELING ABILITY, AND PORTABILITY. SPECIFICALLY PROPOSED IS THE DEVELOPMENT OF A STATISTICALLY ORIENTED, USER-FRIENDLY INTERFACE WHICH WILL BE MATED PRINCIPALLY WITH MAINFRAME AND MINICOMPUTER SOFTWARE THAT HAS BEEN MOVED TO A MICROCOMPUTER AND UPGRADED. INITIALLY THE PROJECT WILL DRAW UPON THE RESERVOIR OF LOCALLY AVAILABLE SOFTWARE. THE PACKAGE WILL BE EXTENDED BY INCORPORATING MORE LOCALLY DEVELOPED, PUBLIC DOMAIN, AND NCI SOFTWARE, AND BY INTERFACING WITH COMMERCIAL PACKAGES. THE PACKAGE WILL FILL A CRUCIAL NICHE FOR BIOSTATISTICIANS AND EPIDEMIOLOGISTS INVOLVED IN RESEARCH AND TEACHING IN THE UNITED STATES AND WORLDWIDE. IT WILL BE USEFUL FOR ANALYZING DATA FROM OBSERVATIONAL COHORT STUDIES, CLINICAL TRIALS, MATCHED AND UNMATCHED CASE-CONTROL STUDIES AND TOXICOLOGY EXPERIMENTS.