Generalized p-Value in Two-Way ANOVA Models

Generalized pivotal model is drawn by Fiducial inference,by which the generalized p-value is defined directly.Then the hypotheses testing problem is solved and this test is proved to be valid at the boundary.In simulation,the method of generalized p-value inference is compared with generalized F-test.The results showed,that the method of generalized p-value is better than the generalized F-test in heterogeneous variances cases.