Interaction term two dummy variables: mean or standardized?

Research topics can be discussed in this area.
Post Reply
tyskevdb
PLS Junior User
Posts: 1
Joined: Mon Nov 17, 2014 3:06 pm
Real name and title: Tijs van den Broek

Interaction term two dummy variables: mean or standardized?

Post by tyskevdb » Mon Nov 17, 2014 3:21 pm

Dear all,

As a new member of the SmartPLS community, I have a question about the type of product term generation I should use when adding an interaction effect to my model.

I developed a model about an online protest campaign with three dummy IDVs (Interactivity, Vividness and Surveillance of communication), each coded with 0 (low) or 1 (high). I expect that Vividness and Surveillance have an interaction effect on the DV Effort of participation, which is reflected by two Likert-scale indicators. When adding the interaction effect to the model, should I use standardized or mean-centered product term generation? And more importantly: why?

I have read Henseler and Fassott's (2010) book chapter on moderating effects in PLS path models, but still I am not sure about my situation. They mention that standardization can lead to significant bias, but I guess that is not the case with two dummy variables.

Thanks in advance for your help!

Best,
Tijs

References
Henseler, J., & Fassott, G. (2010). Testing moderating effects in PLS path models: An illustration of available procedures. In Handbook of partial least squares (pp. 713-735). Springer Berlin Heidelberg.

User avatar
Hengkov
PLS Super-Expert
Posts: 1619
Joined: Sun Apr 24, 2011 10:13 am
Real name and title: Hengky Latan
Location: AMQ, Indonesia
Contact:

Re: Interaction term two dummy variables: mean or standardiz

Post by Hengkov » Mon Nov 17, 2014 11:28 pm

Hi,

If dummy variables is moderating, you need to run multigroup analysis.
Note: non-metric variables cannot calculate to interaction term (0*1).

Best regards

Post Reply