Interactions Interpretations

[Alessia Green]

Hi everyone,
I really need an explanation about this topic. I'm doing a regression with an interaction between two dummies, and i want to know if it is better to put ## or just #. What's the difference between the results?

If the result for A#B is:

A#B. p-value
0 1 not significant
1 0 0,01
1 1 0,01

Can I say that when A is equal to 1 and B is equal to 0 the correlation is significant, and the same for A=1 and B=1?
I'm not sure if it is the right interpretation.

I will be so glad if you will find time to answer me!
Thank you!

Alessia

[Alessia Green]

please can someone help me?

[Sem Vandekerckhove]

The result should be the same whether you use a single or double hashtag. After all, the interaction includes all possible combinations. You can easily verify this.

[Phil Bromiley]

You might look at Robert Friedrich's American Journal of Political Science article in 1982 as a nice commentary on interpreting interactions.

We would be better able to help you if you followed the FAQ on asking questions – right Stata code readable Stata output, and sample data using dataex.

With interactions, the tests in the standard output are testing essentially if something differs from the main effect. This is different than saying the marginal effect in a particular condition is significant. This is why we use factor variable notation and the margins command because it makes doing those tests easier.