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

RE: st: Test of coefficient equality


From   "Dimova, Ralitza" <Ralitza.Dimova@econ.kuleuven.ac.be>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: Test of coefficient equality
Date   Fri, 5 Mar 2004 14:55:22 +0100

Thanks so much!!! I'll try and see how it goes, possibly contact you later about developmens/ more questions. It would be great if I can compare coefficients both across the MLOGIT categories (Wald test does this for the whole coefficient vector, but I am really interested on the effect of "education" on each category) AND across the years. Good luck with the handout; hope to be hearing more from you over the next stages....

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu]On Behalf Of Richard Williams
Sent: Friday, March 05, 2004 2:37 PM
To: statalist@hsphsun2.harvard.edu
Subject: RE: st: Test of coefficient equality


At 01:58 PM 3/5/2004 +0100, Dimova, Ralitza wrote:
>Thanks for the quick response. Yes, I have THREE INDEPENDENT SAMPLES for 
>1995, 1997 and 2001. But for TWO of them (1995 and 1997) SOME of the 
>people are common, so I can also create a quasi panel just for those two 
>years. Right now I am trying to use all three independent samples, so it 
>would be useful for me to know how to do the testing with the time trend 
>that you had proposed. But it won't hurt to know also whether something 
>similar can be done over the quasi-panel. THANKS A LOT. I'll be looking 
>forward to hearing from you.

Multinomial logit models always make my head hurt because of all the 
equations and coefficients floating around!  To see how to approach such a 
problem when regular OLS regression is appropriate, you can see my handout at

http://www.nd.edu/~rwilliam/xsoc593/lectures/l11b.pdf

UCLA has a similar (possibly better) handout at

http://www.ats.ucla.edu/stat/stata/faq/compreg3.htm

The sequence of models I would estimate would be something like

1.  Totally constrained, i.e. no differences across years.  Y is IV, X is DV
2.  Model 1 + dummy variables for year.  Allows intercept to differ across 
years but not slopes
3.  Model 2 + interaction terms.  Allows both intercepts and slopes to 
differ across years.

I'll think about how exactly to adapt this to the mlogit case (I need to do 
this for an upcoming handout anyway!) and write back, if somebody doesn't 
beat me to it.  (You could probably do it the exact same way as in these 
handouts, substituting -mlogit- for -regress-; but the -test- command is 
not the optimal way to do things for stuff like logistic regression.)


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
FAX:    (574)288-4373
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW (personal):    http://www.nd.edu/~rwilliam
WWW (department):    http://www.nd.edu/~soc

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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   |   What's new   |   Site index