Practical use of correlation coefficients in the Social Sciences

Abstract

The Pearson correlation coefficient (r) is usually the first measure of association taught at elementary statistics courses. The usual presentation includes scatterplots, computation and interpretation of r, properties, examples, and warnings about inferring causality from high association between two variables. On this last aspect, few introductory textbooks go deeper into the criteria for establishing causation, and there is a lack of convincing examples in the area of the Social Sciences. Although some textbooks give adequate explanations, most of their examples belong to the field of Biostatistics. There is a need to incorporate convincing cases of the practical use of correlation as supporting evidence of causal relationships in the Social Sciences. We contribute with two examples that could be useful for teaching purposes.