Dependence and Independence: Structures, Testing, and Measuring
In statistical analyses of medical data, exploring dependence between variables plays a fundamental role. For example, it is important to detect and quantify the dependence between disease status and a variety of potential predictors to determine significant risk factors. This chapter presents the results of Vexler, with which practitioners can choose an efficient test of independence. It presents various tests of independence. Dependence measures that can be applied in general cases. The chapter considers various sorts of dependence structures. It conducts simulation studies to compare powers of the tests of independence. The chapter discusses application of tests and measures of dependence to real data examples. It also presents a broader discussion on choosing appropriate tests of independence. Classical dependence measures, such as the Pearson, Spearman, and Kendall rank correlation coefficients, and their accompanying tests of independence, technically target only specific subsets of dependence structures.