LPM/Probit/Logit

[kirill.serykh]

Hello. I've built 3 regressions (LPM, Logit, Probit) on households decisions on stock participation, and I would like to know what model is better (all covariates are the same). My professor said that it should be related to the loglikelihood value, but I don't know what test to use, because pseudo R^2 and likelihood ratiotest don't take place here. It might be reasonable to look on absolute values and to choose the smallest (in absolute values), but I think that it's not enough. Can you help me with this issue: what extra test to use and how it should be done in Eviews? (or maybe, I need to check for some other assumptions in order to say what model is better?). Thanks in advance.

Loglikelihood for LPM: -2500
Loglikelihood for Logit: -2380
Loglikelihood for Probit: -2377