“…Non-parametric correlation provides a single summary statistic (rho) that quantifies the strength of the association between a pair of variables. This is clearly useful when there are a number of factors under study (such as the seven potential barriers and four provider characteristics in Kulczycki et al's study) 1 and there is an interest in ascertaining/reporting the extent to which the provider characteristics are associated with barrier responses (28 associations in all). Furthermore, if there is a wish to draw inferences about the true strength of a correlation (rho), in the population from which the study sample was drawn, then the population estimate for rho can be reported with a confidence interval (CI), in accordance with good statistical practice.…”