Hi all! We are struggling to do a moderation analysis as we chose to do a 3 group experiment instead of a 2x2 factorial design.

We are trying to find out whether social observability (i.e a public or private act) would moderate the effect of an online act of support (e.g. signing a petition) on subsequent prosocial behaviour (e.g. donation). We have 3 experimental groups: Public act of support, Private act of support, and no act of support.

We would like to find out how to compare the difference between public and no act vs. private and no act of support. We were asked by our professor to find out how to test the difference between the 2 beta coefficients, but we're not sure what that means.

In a 2x2 factorial design, we would be able to code Private and Public as 0 and 1, and no act vs an act of support as 0 and 1, then we run process macro. However, with just 3 groups, we can't run the same analysis, as the private and public group would not have an overlap with no act of support group.

Is it impossible to run a moderation analysis if it's not a 2x2? Or is there a way where we can find the difference in the pairwise interaction of these 2 groups (i.e public and no act vs. private and no act of support).

We would deeply appreciate any suggestions!! If it's more convenient, you can email me at [email protected]

Here are some additional information:

We've ran a t-test between public&private act vs. no act of support, and it was statistically significant t(1,191), p=0.035.

Then we ran t-tests between private vs. no act of support t(1,126), p=0.008, confidence interval (-.657,-.102), and between public vs. no act of support t(1,127), p=0.30, confidence interval (-.452, .139).

We also ran a one way ANOVA on the 3 groups, F(2,190), p=0.03.

Bonferroni Post-hoc:

Private v Public p=.37

Private v Control p=.028

Public v Control p= .84