Re: st: Compute and grapth the interaction after logit

[Maarten Buis <maartenlbuis@gmail.com>]

On Wed, Jun 5, 2013 at 2:41 PM, Alexandre James  wrote:
> I just received back from a journal the reviwers comments about a paper I have subbmited. I am using a nested logit model and one of the reviewers mentioned that I need to be more parsimonious to interpret the interaction effects because it is a non-linear model. So, he asked to Compute and grapth the interaction effects. I checked the inteff commend but it does not run after nlogit. Would someone have a suggestion how I could graph it?

A graph could certainly be useful, but I would not describe that as
parsimonious. Especially, the graphs returned by -inteff- are really
far from parsimonious as those basically show different marginal
effects for each observation. So either "parsimonious" was not the
right word to describe what you or the reviewer wanted to achieve or
graphs are not the way to achieve that.

A really parsimonious way to describe interaction effects is to stick
to the odds metric, as is explained here:
M.L. Buis (2010) "Stata tip 87: Interpretation of interactions in
non-linear models", The Stata Journal, 10(2), pp. 305-308.

Hope this helps,
Maarten

---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/