Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: compare regression coefficients between 2 groups (SUEST) across time and across subgroups in a data set


From   Richard Williams <richardwilliams.ndu@gmail.com>
To   statalist@hsphsun2.harvard.edu, statalist@hsphsun2.harvard.edu
Subject   Re: st: compare regression coefficients between 2 groups (SUEST) across time and across subgroups in a data set
Date   Tue, 02 Aug 2011 12:46:24 -0500

At 10:55 AM 8/2/2011, Roland Teitzer wrote:
We would like to find out wheather the regression coefficients we get from
the "logit" command (binary logistic regression) differ significantly from
each other.

This is a far more difficult problem than people realize. Paul Allison described the problem well in this piece:

Allison, Paul. 1999. "Comparing Logit and Probit Coefficients Across Groups." Sociological Methods and Research 28(2): 186-208.

However, I (and others) had problems with Allison's proposed solution. I discussed these issues & offer solutions in

http://www.nd.edu/~rwilliam/oglm/RW_Hetero_Choice.pdf

http://www.nd.edu/~rwilliam/oglm/oglm_Stata.pdf

Quoting from the abstract to my first paper,

Allison (1999) notes that comparisons of logit and probit coefficients across groups can be invalid and misleading. He proposes a procedure by which these problems can be corrected, and argues that "routine use [of this method] seems advisable" and that "it is hard to see how [the method] can be improved." We argue that, as originally proposed, this method can have serious problems and should not be applied on a routine basis. However, we also show that the model used by Allison is part of a larger class of models variously known as heterogeneous choice or location-scale models. We illustrate that there are several advantages to turning to this broader and more flexible class of models. Dependent variables can be ordinal in addition to binary, sources of heterogeneity can be better modeled and controlled for, and insights can be gained into the effects of group characteristics on outcomes that would be missed by other methods.

However, not everyone is crazy about my proposed solutions. For Scott Long's take, see

http://www.indiana.edu/~jslsoc/research_groupdif.htm

Maarten Buis has also made a case for using odds ratios when interpreting interactions:

http://www.maartenbuis.nl/publications/interactions.html

There seems to be a lot of work on this lately. I think the description of the problem is more clear-cut than the solutions are. I'm coming to the conclusion that the best thing to do is to make sure you only have continuous dependent variables...

therefor we used the suest command but we are not quite sure if we used it
correctly, as in the stata help the command is only described for linear
regressions.

we have two different kinds of logit model we want to compare.

1.)in the first case we have two independent subgroups of persons in a
dataset (persons with german nationality and foreigners).

we tried out this command ("poor" is the dependent, the others are the
independant variables):

logit poor women young old if german==1, or

estimates store german

logit poor women young old if turkish==1, or

estimates store turkish

suest german turkish
test [german_poor]women= [turkish_poor]women
test [german_poor]young= [turkish_poor]young
test [german_poor]old= [turkish_poor]old

2.)in the second case we have date were there are mostly cross-sectional
data, but also some panel-data (EU SILC-data)

therefore we used this command:

logit poor women young old if 2004==1, or

estimates store data2004

logit poor women young old if 2009==1, or

estimates store data2009

suest data2004 data2009

test woman
test young
test old

--> we looked up some helps for this problem, but they are written so
complicated that we were even more confused afterwards.
we would need a clear syntax.

Could somebody tell us which syntax we schould use for which problem?

Thank you and sorry for asking such "simple" questions...




*
*   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/

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index