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 at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: Fwd: Comparing marginal effects of two subsamples


From   Jianhong Chen <jianhongchen1985@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Fwd: Comparing marginal effects of two subsamples
Date   Mon, 24 Oct 2011 11:01:33 -0400

Hi, Maarten

Thank you very much for your response. They are very helpful. I really
appreciate it. I have a following question about reporting.

First, I used nbreg, dv iv moderator iv*moderator

The results are:

DV                    Coef.          robuts Std. Err.          z
   P>z               [95% Conf.Interval]

iv                  0.02020         .0565                      0.3600
  0.72            -0.0905   0.1309
moderator         0.0267        0.0768                  0.3500
0.728           -0.1237  0.1772
interaction          0.18260      .0806                    2.2700
0.0230         0.0247   0.3406


Then, I used nbreg, dv iv moderator iv* moderator, irr

The results are:

Robust
DV                   IRR         Std. Err.             z           P>z
            [95% Conf.Interval]

iv                 1.020421    0.057651     0.36           0.72
     0.913461.139907
moderator   1.027098    0.078834     0.35           0.728
0.8836461.193838
interaction   1.200385     0.096749    2.27           0.023
1.024981.405809


So when I report the results in the table, I will use IRR, not coef,
(1.20, not 0.81), right? Do you have a published paper who also used
the same approach for my reference?

Thanks again

Best,

On Mon, Oct 24, 2011 at 4:12 AM, Maarten Buis <maartenlbuis@gmail.com> wrote:
> On Fri, Oct 21, 2011 at 9:23 PM, Jianhong Chen wrote:
>> Thank you very much for your response. I read your article on Stata
>> Journal and I think you are completely right.  Just to make sure that
>> I understand right, are the exponentiated coefficients you mentioned
>> the same coefficients which I can get after runing negative binomial
>> model?
>
> The magic option is -irr-, just as I did in the example I posted earlier.
>
>> Also, since the reviewers require that we should compare marginal
>> effect, we have no chance to avoid that and we plan to use the
>> marginal effects as supplemental analysis, not main analysis. Do you
>> have any idea to do t-test of marginal effects of two subsample?
>
> Typically, my response to the reviewer would be something like:
> "We agree with the reviewer that a substantive interpretation of the
> parameters would improve the article. We have chosen to do so in the
> form of rate ratios rather than marginal effects as that way we avoid
> the problem that correct marginal effects of interaction terms tend to
> be so variable across individuals that a one number summary, like a
> mean marginal effect, is hard to justify (Ai and Norton 2003)."
>
> Hope this helps,
> Maarten
>
> Ai, C., and E. Norton. 2003. Interaction terms in logit and probit
> models. Economics Letters 80: 123–129
>
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
>
> http://www.maartenbuis.nl
> --------------------------
>
> *
> *   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/


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