Exploding coefficients in logit PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Jintai Li posted on Tuesday, November 28, 2017 - 12:43 pm
I am trying to run a discrete choice model (logistic regression) with two latent classes, with no constraints on the coefficients across classes (also fair to say that I am running two structurally similar logit models within each of the two latent classes). I am getting very reasonable estimates for coefficients of the larger class, but the coefficients of the smaller class are usually several hundred times the magnitude. Is this to be expected? How do I fix this problem?
 Jintai Li posted on Tuesday, November 28, 2017 - 12:51 pm
My code in the MODEL part is as follows:

MODEL:
%OVERALL%

c ON HHCars-FullTime;
PMD ON TT_WALK;

PMD#1 ON PTPass(1)
IVTT_BUS OVTT_BUS
TT_WALK(2);
PMD#2 ON PTPass(1)
IVTT_TRAIN OVTT_TRAIN
TT_WALK(2);
PMD#3 ON PTPass(1)
IVTT_MIX OVTT_MIX
TT_WALK(2);
PMD#4 ON HHCars-FullTime PTPass@0
TT_CAR
TT_WALK(2)
CUPB;

%C#1%

PMD#1 ON PTPass(3)
IVTT_BUS OVTT_BUS
TT_WALK(4);
PMD#2 ON PTPass(3)
IVTT_TRAIN OVTT_TRAIN
TT_WALK(4);
PMD#3 ON PTPass(3)
IVTT_MIX OVTT_MIX
TT_WALK(4);
PMD#4 ON HHCars-FullTime PTPass@0
TT_CAR
TT_WALK(4)
CUPB;
 Bengt O. Muthen posted on Tuesday, November 28, 2017 - 2:48 pm
We ask that postings be limited to one window - for longer messages, send to Support.

We need to see the full output; send to Support along with your license number.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: