ABSTRACT
In this chapter, the author explores research interests lie in the area of nonparametric statistics. Some researchers might prefer a measure of partial correlation in which pairs are not simply classified as "matched" or "not matched" but are weighted according to their closeness with respect to the control variable. It is perhaps a disadvantage that the calculation of a matched partial correlation must always begin from scratch, since there is no formula by which it can be determined from total correlations. Yet the partial correlation formula is sometimes deceptively easy, since its numerical instability in the presence of highly correlated variables is not always obvious. The first and best-known index is undoubtedly the classical product-moment correlation of Karl Pearson, which may be defined by the formula. Many of the indices originally based on other concepts of correlation can also be given proportional reduction in error interpretations.
