Comparing Regression Coefficients Across Subsamples

The comparison of regression coefficients across subsamples is relevant to many studies. Often, the same regression model is fitted to several subsamples and the question arises whether the effect of some of the explanatory variables, as expressed by the linear model, is the same for all subsamples. Because there are sometimes misunderstandings as to the statistical procedure which ought to be applied and consequently, wrong formulae are sometimes adopted by researchers. The purpose of this article is to describe in detail the relevant statistical procedures, together with the assumptions underlying the statistical tests. Some examples are quoted, in which wrong formulae were applied. It is demonstrated how the adoption of the wrong formulae might lead to mistaken conclusions.