# Help interpreting an interaction term from R output of linear regression

So I am familiar with the concept interaction. However in most instances that I have come across, a significant interaction term is in the same direction as the 2 individual terms. However I have come across a somewhat cancelling out effect and was wondering what someone with a little more data analytics experience makes of it.

Apologies I can't add more but I don't own the data. The top most term is the intercept. The second term is something that must be controlled for and so you can ignore this.

The third term is a fairly fundamental term that doesn't change. Perhaps you can think of this as gender? The fourth term is less fundamental, that may well change. Perhaps you can think of this as a treatment group? The last term is the interaction between the third and fourth terms, i.e. there are 2 binary variables at play.

Because the interaction term is in the opposite direction, should I interpret this as cancelling out the "treatment group" variable (-0.67)? Or does it cancel out both? The standard error is fairly large so I feel I could go either way.

Should I rely on my background knowledge of the system at play to answer this?

Any and all feedback would be welcome, even if it's, "you need to tell us what the data is representing".

• Apparently, the mathematical or statistical model must be specified in order to assess the interaction effect. I do not understand which sign can be assigned to variables individually ?
– user10619
Aug 1, 2018 at 11:10
• @SubhashC.Davar The model is just data ~ A + B + A:B or data ~ A*B. Aug 1, 2018 at 11:16
• Plot the group means. Then calculate them manually with the coefficients. That should help you understand this. Considering uncertainties is then an additional step. Aug 1, 2018 at 11:21
• @Roland The group means will be the same as the model, surely? Aug 1, 2018 at 11:25
• A model is a model, group means are group means. I don't understand your comment. Aug 1, 2018 at 11:28