I am estimating a fixed effect logit regression
clogit D X P, group(id)
where D and P are dummy variables. All id groups have size 2, and in each group D is once equal to 1, once equal to 0.
I need to recover the marginal effect of the dummy variable P on the probability that D=1.
Can anybody help with this?
Thanks, Marco
(I have already tried with "predict phat if e(sample)". For each group, in correspondence with D=1, this gives me the probability to observe D once equal to 1. But what I want to know is the effect of P.
I have tried creating: a) an additional individual, with two observations, all X equal to the e(sample) mean, and with P=1; b) a second individual with same characteristics, but P=0;
However, as P is time-invariant for these individuals, the probability to observe D=1 and D=0 is for both 0.5.
I have then created an individual with P=D=0 in one observation, and P=D=1 in the other. It seems to me that the probability obtained with predict, minus 0.5 (the probability for the individual with P=0 in both periods) is what I am looking for).
____________________________________________________________________________________
Now that's room service! Choose from over 150,000 hotels
in 45,000 destinations on Yahoo! Travel to find your fit.
http://farechase.yahoo.com/promo-generic-14795097
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/