Dummy Variable Regression Models and Analysis of Responses in Diagnostic Screening Tests with Replications

This paper develops and presents use of dummy variable regression models in the analysis of results from diagnostic screening tests or clinical trials with several case-control subject response options or categories, with replications in each case-control response category or combination. The proposed method can also be used in comparing differences between proportions of case and control subjects responding positive under various case-control response categories or options. Test statistics are developed for testing the null hypotheses that the regression model is an adequate representation of the differences in scores or responses between pairs of case and control subjects in diagnostic screening tests and for comparing the responses of cases and controls in the paired case-control subject response categories. Test statistics are also developed for use in pairwise comparisons of the effect of scores by case-control subject at various response categories. Corresponding probabilities, odds and odds-ratios of positive responses are also estimated using the resulting regression coefficients. The proposed method is illustrated with some data.