Re: [Gretl-users] Logit-model

Alexander, By "an ols method" do you mean a linear probability model? If so, just do ols Y 0 MDR Age Selfempl Income or use the gui as usual. PS -----Original Message----- From: gretl-users-bounces(a)lists.wfu.edu [mailto:gretl-users-bounces@lists.wfu.edu] On Behalf Of Alexander Rogalsky Sent: Wednesday, December 29, 2010 10:27 AM To: gretl-users(a)lists.wfu.edu Subject: [Gretl-users] Logit-model Dear Gretl-community, I try to estimate a logit model using gretl according to the user's guide. I used an embedded example for the calculation (credscore.gdt). The dependent variable (Y) = 0 or 1. Four independent variables: MDR, Age, Selfempl, Income. The package performs calculations using maximum likelihood method. But I would like to use an ols method. How can I do it? Variable sustitution does not work: y* = ln (P / 1-P) = a + bx but y* is unobserved variable (latent). P^ = 1 / (1 + exp (-y*)) ln (1/P^ -1) = a + bx May I use Y (0 or 1) instead of P^? Any help will be appreciated. With best wishes for Christmas and the New Year, Alexander Rogalsky _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users