Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: Marginal Effect

From   "[email protected]" <[email protected]>
To   "[email protected]" <[email protected]>
Subject   RE: st: Marginal Effect
Date   Wed, 5 Jan 2011 20:36:52 +0000

Hi Marten,
I thought if the unobserved individual effect i. e u_i = 0 implies that rho which is the relative importance of the unobserved effect  is 0. I think it is same as the neglected heterogeneity in the estimation of partial effect  Wooldridge was refering to in his book on chapter 15. 

From: [email protected] [[email protected]] on behalf of Maarten buis [[email protected]]
Sent: Wednesday, January 05, 2011 1:46 PM
To: [email protected]
Subject: Re: st: Marginal Effect

--- On Wed, 5/1/11, sazas  wrote:
> I estimated a random effect probit model using xtprobit in
> stata 11.
> Marginal effects using margins, dydx(*) predict(pu0)
> assumes u_i =0 which l think implies rho =  0 since rho = 0.
> Please is there any command that l can use to change this
> assumption?

It is not true that rho (proportion of variance due to group
level variance) is assumed to be 0. The way to think about it
is that there is an (unobserved) group level variable added
to your model and that you compute your marginal effects for
individuals that have an average value on this variable. I do
not think that is a too problematic assumption, but there is
always a bit of friction when using marginal effects for this
type of models. In essence you are trying to fit a linear
line to a non-linear one, which will often (but not always)
produce an ok summary of the non-linear line, but it will
never be exactly right. If you are a purist, then you should
probably use -xtlogit- and interpret the odds ratios rather
than the marginal effects. In practice, I would use which ever
model and effect size I prefer, and look at tables of
predicted probabilities and odds, look at graphs of predicted
probablities and odds and look if I can (approximately) match
those with the effect sizes I found. If you can do that, then
there isn't much of a problem.

Hope this helps,

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen

*   For searches and help try:
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index