Measure of Association Dependent on Sensitivity and Specificity of Diagnostic Screening Tests in a Study Population

This paper proposes, develops and presents a measure of the strength of association between test results and state of nature or condition in a population, constructed using only the sensitivity and specificity of diagnostic screening tests or clinical trial that are independent of the population under study. The proposed measure of association which always lies between -1 and 1 inclusively enables the researcher to determine not only if an association exists between test results and state of nature or condition in a population and if such an association exists, whether it is positive and direct or negative and indirect thereby giving the measure an advantage over and above the traditional odds ratio method. Results obtained using the proposed measure are shown to be easier to interpret and understand than those obtained using the traditional odd ratio approach. Furthermore, using sample data, the proposed measure of association is shown to be at least as efficient and hence as powerful as the traditional odds ratio. The proposed method for comparison and completeness also develops modified sample estimate of the traditional odd ratio and its sample variance in terms of observable sample data from diagnostic screening tests or clinical trials. It also provides a test statistic in terms of observable sample data for testing the significance of the modified traditional odds ratio that would be properly used in the analysis of results obtained from diagnostic screening tests or clinical trials for which the traditional odds ratio may not always be properly used to analyze.