Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]


From   "Scott Merryman" <[email protected]>
To   <[email protected]>
Subject   st: Re: R-SQUARED AND XTGEE
Date   Tue, 28 Oct 2003 23:15:18 -0600

----- Original Message ----- 
From: "Clive Nicholas" <[email protected]>
To: <[email protected]>
Sent: Tuesday, October 28, 2003 7:46 PM

> (1) I cannot successfully fit an -xtgee- model (ie, xtgee [depvar]
> [varlist], family(binomial) link(probit) i(id) t(waves38)). All I get is
> this error message: "estimates diverging (absolute correlation > 1)
> r(430);". Uhh? And how can this be got round?

Is the model identified?  Does your id variable perfectly predict the dependent

> (2) An expert in pooled analysis taught us in the summer that if you're
> fitting fixed-effect models, you *must* choose logit. If a random-effects
> model is chosen, then *probit* must be used. Can anybody tell me why this
> is, especially since you can, eg, fit random-effect logit models with
> Stata?

The fixed effects probit model suffers from the incidental parameter problem
(see Lancaster, 2000 "The incidental parameter problem since 1948" Journal of
Econometrics)  As N goes to infinity, for a fixed T, the number of fixed effects
increase with N.  This means that they cannot be consistently estimated given a
fixed T.

A fixed effects (conditional) logit is able to get around this problem by
maximizing the conditional likelihood function that sweeps away the individual
effects.  In the logit model the conditional distribution does not depend on the
individual effects.

Hope this helps,

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index