Re: st: interaction effects vs. interaction coefficients using contrast?

[Maarten Buis <maartenlbuis@gmail.com>]

On Sat, Feb 4, 2012 at 2:51 PM, Colleen Nugent wrote:
> I'm running an event history model using discrete-time multinomial logit in Stata 12, and I have a three-category by three-category interaction.  I want to know if certain groups formed by combinations of those two variables are significantly different frome one another.  I've read through the Stata 12 manual and examples related to using "contrast" and "lincom" for testing differences between groups after fitting a model, but I haven't been able to figure out if either or both of these methods takes into account the multiplicative (or additive if a person is using probabilities instead of relative risk ratios) nature of overall interaction effects, which can involve two or three individual coefficient terms together rather than just a single coefficient.  Do contrast and lincom take this into account when they test the statistical significance of differences between particular groups?

-lincom- is very general, this means that it can do a lot but you need
to type the correct commands, and it is easy to get it wrong.
-margins- seems more to the point, this requires that you used the
factor variable notation to create the interactions, and than it will
take those into account. If you add the option -rrr- to your -mlogit-
command than you already got the interpretation in terms of ratios of
relative risk ratios without any need for further commands. A basic
discussion on that is:

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
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/