A Test for Model Specification in the Absence of Alternative Hypothesis
Herman J. Bierens · University of Minnesota Digital Conservancy (University of Minnesota) · 1981
In recent papers ([6], [4]) several tests have been proposed for testing the truth of a nonlinear specification of a regression model against one or more well-specified alternative hypothesis. In this paper we first show by an example that these tests may be defective if all the hypotheses are false, and then we propose a new test that does not need alternative hypotheses in the sense that we only test whether or not a given specification is true.