Working Papers, University of California, Davis, Department of Economics 01-8
When a significance test fails to disconfirm a hypothesis economist often interpret this as evidence that this hypothesis is valid. Six such examples are cited from recent journals. But this is a misinterpretation of what significance tests show. Presumably this misinterpretation is founded on the valid principle that every failure to disconfirm a hypothesis adds to its credibility. But that principle defines a failure to disconfirm in a way that differs sharply from the way that this phrase is used in the context of significance tests. Some ways of ameliorating this problem exist.
significance tests t values t coefficients confirmation