[Limdep Nlogit List] Scale parameter in RPL

Emmanouil Mentzakis ementzakis at gmail.com
Tue Nov 9 10:08:21 EST 2010


I have data on the same DCE survey from two different samples and I
want to estimate a random parameter logit where I can allow for
differences in the scale. I know I can estimate an RPL with
heterogeneity in mean and heteroscedasticity in random parameters
using the -RPL- and -Hfr- options:

;Lhs = choice;Choices = alt1,alt2,none
;RPL = sample2
;Fcn=X1(N), X2(N)
U(alt1) = b1*X1 + b2*X2 /
U(alt2) = b1*X1 + b2*X2 /
U(none) = asc3
;Hfr= sample2$

-sample2- is a dummy indicating the second sample.

However, I have seen in the literature RPL models (e.g. Hanley et al.
2005. Price vector effects in choice experiments: an empirical test.
Resource and Energy Economics) where they allow for differences in
scaling using just one parameter (i.e. scale); very similar to an
HLOGIT with the -Hfn=sample2- and -Sdv=1,1,1- options.

Could anyone please advice on how to estimate such RPL? Thank you in advance.

Best regards

More information about the Limdep mailing list