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: hierarchical logistic regression command


From   Richard Williams <richardwilliams.ndu@gmail.com>
To   statalist@hsphsun2.harvard.edu, <statalist@hsphsun2.harvard.edu>
Subject   Re: st: hierarchical logistic regression command
Date   Fri, 17 Feb 2012 08:19:01 -0500

At 05:56 AM 2/16/2012, Maria Niarchou wrote:
Hi,
>
> I would like to perform a hierarchical logistic regression analysis in which > independent variables are entered in blocks. 'Hireg' doesn't seem to work with categorical outcomes.
> Could you please let me know if there is an alternative command to do this?
>
> Thanks,
> Maria

There are potential problems with comparing logit coefficients between nested models. Besides the pieces already cited, see

http://www.nd.edu/~rwilliam/stats/Oglm.pdf

http://www.nd.edu/~rwilliam/stats/Oglm-Handout.pdf

However, I wouldn't say that you should never use commands like -nestreg-. It is very common to do chi-square contrasts between nested models, e.g. having first entered demographic variables, maybe you want to then test whether social psychological measures add anything to the fit of the model. It is not unusual to present a series of chi-square contrasts and then present only the coefficients for your preferred model; see, for example, the Hauser and Andrew piece referenced in my handout.

In addition, comparison problems may not be that important in practice; there is a good chance you will actually be understating the importance of any changes you detect; and there are ways to deal with the problems. Y-standardization has been the most widely used practice in the past, but I think the new KHB method mentioned in the handout has a lot of potential.

Finally, if you do present the coefficients for several nested models, you'll have a lot of company. It took my grad student about a week to find dozens of examples where this is done. He has since been getting ahold of a lot of the data sets so we can see how much harm has been done in practice. It is easy to show the problems in theory, but showing that any horrendous mistakes have been made in practice is another matter.


-------------------------------------------
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