When fitting a model with random effects using the expected mean squares method, JMP® tests the effects correctly. The "Test Denominator Synthesis" output may be different than what you think it should be. However, JMP follows the references of Hocking and Hartley and Searle (listed below) to determine the appropriate error terms in models with random effects. This issue is an item of controversy among statisticians and relates to the formulation and parameterization of the model.
- Hocking, R.R. (1973), "A Discussion of the Two-Way Mixed Model," The American Statistician, 27: 148-152.
- Hartley, H.O. and Searle, S.R. (1969), "A Discontinuity in Mixed Model Analysis," Biometrics, 25: 573-576.
[Previously JMP Note 37928]