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]

st: Marginal effects after xtprobit

From   Elin Vimefall <>
To   "" <>
Subject   st: Marginal effects after xtprobit
Date   Wed, 2 Oct 2013 07:15:10 +0000


I'm interested in why some children start school and I analyze this with a probit model. 
Since I believe that there are unobserved characteristics at district level casing the error terms to be correlated I use cluster for the error term.
My model looks like this:

probit School x1 x2, cluster(district)

However now I instead use the panel setting like this

Xtset district

Xtprobit school x1 x2

The result is similar with some changes in significance.

However, when I calculate the marginal effects using

margins dydx(*)

the magnitude if the marginal effects in the panel setting becomes approximately 10 times larger than the marginal effects after the normal probit. 
Can someone help me explain why?

//Elin Vimefall

*   For searches and help try:

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