This article is an excellent example of how the so-called "social sciences" struggle with the problem of determining dependent versus independent variables and the author is to be commended for trying to accommodate the inevitable ambiguities that arise therefrom. One thing not mentioned, however, is that datasets are often unreliable either because of sample bias or because the number of points included are too few to be statistically significant. A great many false inferences have been drawn by "social scientists" on the basis of poorly-assembled data.