[Limdep Nlogit List] Need Multinomial outcome/Probit selection model
james benson
jbenson at ssc.wisc.edu
Sat Apr 7 01:38:18 EST 2007
Hi,
For my dissertation, I am preparing to evaluate the effects of several
state policies on the likelihood of success for community college
students. Success will be operationalized as a set of
mutually-exclusive outcomes: degree completion; transfer; persistence
with no degree; and dropping out (failure; the excluded category). I
need to compute a multinomial logistic regression model, while taking
into account selection into the sample. The selection model will have a
binary outcome (whether or not a student chose to attend a community
college), and thus can be either a logistic regression or probit model.
Does LIMDEP have a procedure by which I can estimate these two models,
either simultaneously or through a two-stage procedure? It would be
especially helpful if I can do at least some of the following: make the
computations using maximum-likelihood estimation; use design weights;
account for clustering at the state level; and output the inverse Mills
ratio from the selection model, for use in subsequent HLM analyses.
Any specific answers, general insights, or reading suggestions will be
appreciated.
James
James Benson
Research Assistant/ITP Fellow
UW-Madison
1025 West Johnson St #652
Madison, WI 53706
More information about the Limdep
mailing list