1.5. Fixed Effects Methods in SAS/STAT

Different kinds of dependent variables require different kinds of fixed effects methods. SAS has a variety of procedures that can be used to implement fixed effects methods, although the original programmers may not have had such applications in mind. For linear models, PROC GLM is probably the most convenient procedure for performing a fixed effects analysis, although PROC REG and PROC MIXED can also be used with a little effort. For fixed effects logistic regression, PROC LOGISTIC is the most convenient procedure, but PROC GENMOD can be used in applications where there are only two observations per individual. For fixed effects Poisson regression, PROC GENMOD appears to be the only choice. For fixed effects survival analysis, PROC PHREG is the procedure of choice, although PROC LOGISTIC can be used in some circumstances. Finally, PROC CALIS can be useful for fitting fixed effects linear models with lagged endogenous variables.