Hello,
I would like to know if interaction effect is equal to moderating effect. If not, what a difference there is. The problem I have is that a reviewer has advised me to use interaction instead of moderation.
Thank you
methodological doubt
-
- SmartPLS Developer
- Posts: 1284
- Joined: Tue Mar 28, 2006 11:09 am
- Real name and title: Dr. Jan-Michael Becker
Re: methodological doubt
Some people see a difference and some people don’t. The differences if there are any are only subtle.
Basically, the interaction effect is the statistical tool to test a moderating effect.
A moderating effect means that a variable M effects the strength of the relationship between a predictor P and a dependent variable Y.
The interaction effect is the effect of the product term (PxM) you include in the model between moderator (M) and predictor (P). Both variables interact. Just the by the interaction effect it is not distinguishable which is the moderator and the predictor. Both M and P you can also have opposite roles.
The moderating hypothesis distinguishes the moderator and the predictor and assigns them their role in the model.
Statistically you estimate a regression with Y = a + p1 P + p2 M + p3 PxM and p3 is the interaction effect.
This is equivalent to both
Y = a + (p1 + p3 M)P + p2 M
(where you can see that the effect of P on Y depends both on p1 and p3M hence the level of M, i.e. the moderator) as well as
Y = a + (p2 + p3 P)M + p1 P
where the effect of M on Y depends both on p2 and p3P hence the level of P, i.e. in this model P would be the moderator and M the predictor.
Therefore, you can see that p3 (the interaction effect) tests the moderating effect, but you need to assign which one is the moderator and predictor by ways of theory.
Please also see: http://jasemjournal.com/journal-of-appl ... t-al-2018/
Basically, the interaction effect is the statistical tool to test a moderating effect.
A moderating effect means that a variable M effects the strength of the relationship between a predictor P and a dependent variable Y.
The interaction effect is the effect of the product term (PxM) you include in the model between moderator (M) and predictor (P). Both variables interact. Just the by the interaction effect it is not distinguishable which is the moderator and the predictor. Both M and P you can also have opposite roles.
The moderating hypothesis distinguishes the moderator and the predictor and assigns them their role in the model.
Statistically you estimate a regression with Y = a + p1 P + p2 M + p3 PxM and p3 is the interaction effect.
This is equivalent to both
Y = a + (p1 + p3 M)P + p2 M
(where you can see that the effect of P on Y depends both on p1 and p3M hence the level of M, i.e. the moderator) as well as
Y = a + (p2 + p3 P)M + p1 P
where the effect of M on Y depends both on p2 and p3P hence the level of P, i.e. in this model P would be the moderator and M the predictor.
Therefore, you can see that p3 (the interaction effect) tests the moderating effect, but you need to assign which one is the moderator and predictor by ways of theory.
Please also see: http://jasemjournal.com/journal-of-appl ... t-al-2018/
Dr. Jan-Michael Becker, BI Norwegian Business School, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de