Resumen de Model checks for parametric regression models

Eckhard Liebscher

• In the paper we deal with the problem of model selection among fixed-design regression models. We establish a new test that indicates whether or not the model fits the data. The test statistic is based on the difference between a parametric estimator for the model variance and a nonparametric difference-based estimator, see Hall et al. (Biometrika 77:521–528, 1990). The weights in the nonparametric estimator depend on n, and they are chosen by solving an optimisation problem in order to obtain a test with high power.