[Limdep Nlogit List] multinomial logit with proportion data
Kandice L. Kleiber
kleiberk at onid.orst.edu
Sun Apr 12 10:49:22 EST 2009
I'm running a conditional logit model as seen in Greene, Econometric
Analysis (p846 new edition). The log-likelihood function contains an
indicator variable (dij) where dij = 1 if Yi = j and 0 otherwise. However,
my LHS variable in estimation is a proportion for each choice that all sum
to 1 for a given observation. Greene states that to account for the
proportion data, dij only needs to be defined as the proportion. I think I
have my program set up correctly, however I just want to make sure that it
is picking up my proportion data in estimating the log-likelihood function.
Can someone verify?
My code is:
lhs = lev2;
choices = corn, soy, wheat;
rhs = one, ed1, ed2, age1, age2, ... liv1, liv2;
wts = p30$
Thanks a lot!
More information about the Limdep