I'm trying to choose between a conditional logit model vs. binary logit with random effects for discrete choice data. I've received a series of errors using first the Hausman test (asymptotic error), then after adding sigmamore (Estimators do not save e(sigma) or e(rmse)) and now with the suest command: it "requires that predict allow the score option."
My commands are:
All of my data are factor variables.
Many thanks for any help.
My commands are:
Code:
clogit choice agegrp_0 agegrp_2 popsz_1 popsz_2 svrty_1 svrty_2 ce_1 inc_0 inc_2 race_1 race_2 constant, group(obsid) estimate store fe xtlogit choice agegrp_0 agegrp_2 popsz_1 popsz_2 svrty_1 svrty_2 ce_1 inc_0 inc_2 race_1 race_2 constant, re estimate store re suest fe re
Many thanks for any help.
