Testing Alternative Hypotheses in Economic History

This paper employs a statistical technique comparing non-nested quantitative models in order to address an important problem in economic history, namely, the appropriate role of microeconomic models in historical analysis. Discussion proceeds by example. As described in the paper, three research groups have offered explantations of the shift from corn to cotton production in the post-bellum US South but because their research methods are different, it is difficult to evaluate them effectively. This paper suggests a method for comparing them. The method is based on the Neyman–Pearson likelihood ratio and proceeds by focusing on the three groups’ models of crop choice, in turn hypothesizing each model as ‘truth’, and testing all other against it. Though restrictive in its own right, this excercise suggests one data-oriented approach to all-too-common problem of model proliferation in economic history.