![Statalist](/includes/images/head_statalist.jpg) |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: probit with interaction dummies (significance and marginal effects)
From |
Andrea Bennett <[email protected]> |
To |
[email protected] |
Subject |
st: probit with interaction dummies (significance and marginal effects) |
Date |
Thu, 17 Jul 2008 16:12:06 +0200 |
Dear all,
I am a little confused after reading multiple posts from the Statalist
how I can make sure I do interpret interaction dummies correctly when
using a probit estimation. For example I would like to estimate -
male*low_education-, -male*mid_education- and -male*high_education- in
a probit model. But, in this article posted on Statalist (http://www.stata.com/statalist/archive/2005-07/msg00559.html
) it says that "surprisingly, the sign may be different for difierent
observations. The statistical significance cannot be determined from
the z-statistic reported in the regression output." Does this really
mean that even if the interaction dummy shows a P-value of "0.000" I
cannot take it for being significant?
The authors then state that one should instead use -inteff- for these
kind of tests. But another article on the Statalist (http://www.stata.com/statalist/archive/2008-03/msg00549.html
) says that -inteff- cannot do interactions like the one above and
refers back to -mfx- or -margeff-.
So, if anyone knows an easy to understand procedure how I can tell
that an interaction dummy in a probit estimation is significant and
additionally how I can compute marginal effects for it, that would be
great. I'm rather new to probit estimations, so solid but not all too
fancy solutions would be preferred (if there are any).
Many thanks for your considerations,
Andrea
*
* 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/