Dummy variable regression models and analysis of responses in diagnostic screening tests with replications

Abstract

Often in controlled comparative either prospective, retrospective or cross–sectional study involving matched samples of subjects or patients, the response of a subject or patient to a predisposing factor in a retrospective study or to a condition or treatment in a prospective study may be dichotomous. But there may also be several other possible response options other than dichotomous but numbering some c possible mutually exclusive response options or categories. Specifically , suppose we have a total random sample of n- matched pairs of patients or subjects matched on certain characteristics to be exposed to two treatments, drugs, tests; suppose nil pairs of these subject pairs are in the response category by case and response category by the control subject, for responses. There may be no replications; that is there is only one pair of case-control subjects.