Hi Clive, the reason why I don't pool both groups together is that the estimated cut point point paramters are much different for immigrants and the native-born. And you cannot interact them with an immigrant dummy in the ologit command. Thus it constrains them to be the same across groups. I find it makes a large difference, in terms of the results, between pooling them and running them separate. Is my thinking on this wrong?
________________________________________
From: [email protected] [[email protected]] On Behalf Of Clive Nicholas [[email protected]]
Sent: Wednesday, May 06, 2009 11:19 PM
To: [email protected]
Subject: Re: st: St: Ordered Logit Question
Jason Dean wrote:
I am running two ordered logits equations. One for immigrants and one for the native-born. > Each has the exact same independent and dependent variables. There are 3 categories for > the depedent variable. I find that the threshhold parameters are quite different for these two > groups. Specifically, both cutpoints are much lower for immigrants. Can anyone enlighten > me as to how I should interpret this? To me this means, all else equal, immigrants are
much more likely to be in the highest category and much less likely to be in the lowest
category. Can I just interpret this in a similar manor as if these two groups had different
intercepts in a linear regression? Also, is it appropriate to compare marginal effects
between immigrants and the native-born.
My first reaction to this would be to run the one model only for all
of your cases, if all of your variables are the same in both models,
including a dummy variable for ethnic origin (say: 0=non-native;
1=native). Then you only have to interpret one set of thresholds.
Running -predict- after -ologit- will give you the estimated scores on
Y* (the latent construct of your dependent variable whose values are
measured continuously) against which you can compare the thresholds.
Hamilton (2004: 278-80) has some concise stuff on interpreting the
thresholds (although my copy is old), whilst Jaccard (2001: 17)
explains why it really isn't a good idea to run seperate logistic
regressions for discrete groups.
--
Clive Nicholas
[Please DO NOT mail me personally here, but at
<[email protected]>. Please respond to contributions I make in
a list thread here. Thanks!]
Hamilton LC (2004) "Statistics With Stata 8", Belmont, CA: Thomson.
Jaccard (2001) "Interaction Effect In Logistic Regression", QASS
Series Paper 135,
Thousand Oaks, CA: Sage.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
