0
$\begingroup$

I have ran a (mixed) linear model, finding a strong two-way interaction effect between two predictors (p < 0.001).To better understand the interaction in point, I have produced an interaction plot, finding that the two lines in the plot do not cross each other.

To the best of my knowledge, the two lines must cross each other for an interaction effect to be present - this kind of "switch" gives in fact an immediate intuition of what the interaction is about.

Am I correct? Or is it possible to have an interaction effect in which the two lines do not cross?

Any elucidation on this would be much appreciated!

$\endgroup$

1 Answer 1

1
$\begingroup$

No, that's not correct. You can even have lines cross that does imply interaction. You can almost think of it as being whether or not the slope of the lines are statistically significantly different. It's just that if the slopes are very different then the lines are likely to cross.

$\endgroup$
1
  • $\begingroup$ So, if I understand it correctly, the condition for an interaction to be present is that the slopes of the lines involved must be statistically different. It is crystal clear, thank you so much for the insight! $\endgroup$ Oct 31, 2020 at 11:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.