On measuring and comparing usefulness of statistical models

David Azriel
Yosef Rinott

Statistical models in econometrics, biology, and most other areas, are not expected to be correct, and often are not very accurate. The choice of a model for the analysis of data depends on the purpose of the analysis, the relation between the data and the model, and also on the sample or data size. Combining ideas from Erev, Roth, Slonim, and Barron (2007) and the well-known AIC criterion and cross-validation, we propose a variant of model selection approach as a function of the models and the data size, with quantification of the chosen model's relative value. Our research is motivated by data from experimental economics, and we also give a simple biological example.

October, 2014