[Limdep Nlogit List] RPL - heterogeneity in mean for a subset of random parameters

[Hunt, Len (MNRF)]

Hi folks,

I am trying to estimate an RPL whereby only some random parameters have mean shifters (and consequently, I cannot use RPL= ). The real problem for me arises from the fact that I specified several of these random parameters as lognormally distributed to ensure that the parameters are strictly positive. So ... I cannot simply create new variables that interact the mean shifter and random parameter variables; and estimate parameters for these interactions directly in the RPL.

Is there a way to fix (to 0) some of the mean shifter parameter estimates in an RPL using the "RPL=" option? If yes, please explain how. If no, does anyone have any other potential solution?

Thanx in advance,

Len