Differences can occur when comparing the SAS PROC GLM and JMP systems when the number of data points per factor combination is unequal or when random effects are present. These differences arise because of differences in the parameterization (coding) methods and algorithms for calculating statistics. The JMP algorithm is more efficient in its use of computer resources, while the GLM algorithm is more general. This does not mean that one is wrong, just that the methods are different.
For more information on the difference between the two, see "Singularity and Missing Cells in Nominal Effects" in the "Statistical Details" chapter of the JMP online help for "Finding Regression Models."
FAQ # 462
