[Limdep Nlogit List] Is possible to weight observations in logit models

scolombo at ugr.es scolombo at ugr.es
Tue Apr 20 17:14:29 EST 2010


Dear all,

I wish to weigh the observations in a conditional or random parameters
logit model according to a weight variable which is observation specific
(i.e invariant within each observation (one choice) but varies among the
observations (choices) made by a respondents).

I wonder if is possible to used the ; WTS command to consider the weight
of an observation like in a "choice based Sampling". It would be something
like to give more importance in the likelihood function to some
observations than others. I observed that in the resulting model the
log-obs of each observation is "scaled" with the weight variable and this
seems correct to attain more information to some observation.
Has somebody had any similar experience/question?

Thanks for any advises.
Sergio.



More information about the Limdep mailing list