[Limdep Nlogit List] Question regarding generalized mixed logit (GMXLOGIT)

[Yau-Huo Shr]

Hi

I am running generalized mixed logit models with different specification
using the GMXLOGIT command in NLOGIT 5.  Because I am polling two samples
together and interested to know if the scale and its variance are different
between two samples, I have the option HFT and HFR included.

After getting the results, I then wanted to do a likelihood ratio test to
see if the theta and lambda are jointly zero and so remove the option HFT
and HFR.

Then I found out the restricted model have higher (less negative)
likelihood ratio compared and pseudo R-square to the model with the option
HFT and HFR, which I have no idea why it is the case.  Could someone
provide me some advice?

Here is the essential part of my code:

CREATE; IF(VERSION=1) V1=1 $
CREATE; IF(VERSION=2) V2=1 $

GMXLOGIT; lhs=CHOICE ; choices=1,2
                ; rhs= DSG2, DSG3, SPC2, SPC3, WTR2, WTR3, Cost,
                ; pds=12; maxit=1000; halton; draws=250; Wts=Weight;
                ; fcn= DSG2(n), DSG3(n), SPC2(n), SPC3(n), WTR2(n), WTR3(n)
                ; hft = V2; hfr=V2; cor =
1,
1,1,
0,0,1,
0,0,1,1,
0,0,0,0,1,
0,0,0,0,1,1
$

Thanks!
Jimmy Shr