Re: st: Re: Why main effects are significant but interction term is not signficant

[Nick Cox <njcoxstata@gmail.com>]

You need to show us detailed results, because unless you feel bound to
jump one way or the other just because P is above or below some
critical level, you do have choices, and your choices will depend on
more than you are telling us.

1. If you think there should be or could be an interaction, you can
still estimate its magnitude.
If P were just greater than 0.05, many researchers would keep it in
and flag caution.

2. It may be that there is an interaction, but you need something more
subtle to catch it. You should be plotting data and results to check
for structure you have missed. Should income be modelled on some other
scale? What is educational level: years of schooling? Should it be
left as is or are there threshold or nonlinear effects for college
degrees?

On a small point, I'd advise against saying OLS when you mean
regression. The model is more important than the estimation method.
(You would not say, I imagine, that you used OLS when you present a
mean or anova.)

On Tue, Mar 8, 2011 at 12:22 PM, Mike <quangdata@gmail.com> wrote:
>  y= beta+beta1*x1+beta2*x2+beta3*x1*x2+epsilon
>
> You can think of y as income, x1 is gender (1 for male) and x2 is the
> educational level.
>
>  The OLS gives a significant results for beta1 and beta2 but not
> beta3. In the context of the example, male and higher education help
> having higher income. But the interaction of male and higher education
> doesn't have any significant effect on income. Can you provide some
> insights?

*
*   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/