Log likelihood ratio test for model selection by using mixed logit model

[Yaw Sarfo]

Dear Colleagues,

I would like to use the log likelihood ratio test to select between two models: one with only choice attributes, and the other includes choice attributes and interaction terms. I followed the procedure and applied the code lrtest m1 m2 for the log likelihood ratio test. However, this message is displayed: log likelihood of null models differ: -1062.44 vs. -1106.533. Please, has anybody encountered such issue before and how did S/he address it? Thanks for your support.

Best regards,
Yaw

[Carlo Lazzaro]

Yaw:
welcome to ths forum-
As per FAQ, please repost your query (following the FAQ requirements) on the General forum, as this one is for practising purposes only. Thanks.