GENERATION OF BIOSTATISTICAL DISTRIBUTIONS

IN PHASE I, IT IS PLANNED TO SELECT AND PROGRAM IN THE C LANGUAGE STATE-OF-THE-ART ALGORITHMS FOR THE COMPUTATION OF THE CUMULATIVE DISTRIBUTION FUNCTIONS OF THE MOST IMPORTANT PARAMETRIC AND NONPARAMETRIC STATISTICAL DISTRIBUTIONS, INCLUDING THE NORMAL, CHI-SQUARE, STUDENT'S T, 1- AND 2-SAMPLE WILCOXON, AND SPEARMAN, AS WELL AS FLEXIBLE RANDOM-NUMBER GENERATION FACILITIES. THESE DISTRIBUTIONS ARE ESSENTIAL FOR THE STATISTICAL EVALUATION OF EXPERIMENTALDATA FROM BIOMEDICAL RESEARCH. ACCESS TO THESE PROGRAMS WILL BE PROVIDED THROUGH A MENU SYSTEM, AS WELL AS THE MLAB ENVIRONMENT, WHICH OFFERS A SIMPLE BUT POWERFUL INTERACTIVE USER INTERFACE, INCLUDING FACILITIES FOR TABLE GENERATION, HIGH-QUALITY 2- AND 3-D GRAPHICS, AND DATA MANIPULATION. PHASE II RESEARCH WILL: (1) EXTEND MLAB BY INCORPORATING USER-FRIENDLY FACILITIES FOR PERFORMANCE OF THE USUAL AS WELL AS ROBUST PROCEDURES FOR STATISTICAL ANALYSIS OF EXPERIMENTAL DATA, USING THE DISTRIBUTION FUNCTIONS DEVELOPED IN PHASE I; (2) PROVIDE MODERN COMPUTATIONALLY INTENSIVE TOOLS SUCH AS JACKKNIFE AND BOOTSTRAP STATISTICAL ANALYSIS; (3) DEVELOP GRAPHICAL DISPLAY OF CONFIDENCE REGIONS FOR NONLINEAR PARAMETERS; AND (4) EXTEND THE PROBABILITY CALCULATIONS OF PHASE I TO INCLUDE THE CASE WHERE THE NULL HYPOTHESIS IS FALSE, FOR POWER AND SAMPLE SIZE CALCULATION. THE RESULTING EXTENDED MLAB PROGRAM WILL BE SUITABLE FOR GENERAL STATISTICAL APPLICATION, AS WELL AS FOR TEACHING STATISTICAL ANALYSIS TO STUDENTS AND RESEARCH WORKERS.