I am trying to do a moderation analysis for an assignment and am struggling to find an answer from googling. I have tried tooling around with lm() but am not sure if I am doing so correctly.
Essentially, I have 4 variables that I am working with. The dependent variable (selfesteem) is numerical, and I have two numerical grouping variables (bodysatis AND importance) and grouping variable that is a factor with two levels (agegp: "old" and "young").
I am trying to work out if the strength of the relationship of the two numerical grouping variables to the dependent variable differ between the two age groups.
Is this how I would specify it?
lm0 <- lm(selfesteem ~ bodysatis * importance + agegp, data)
Or should I break them up and do
lm1 <- lm(selfesteem ~ bodysatis * agegp, data)
lm2 <- lm(selfesteem ~ importance * agegp, data)
and then compare the R2 strengths?
I could and should I go further still and break up each one individually and compare the results, as in
lm1 <- lm(selfesteem[agegp == "old"] ~ bodysatis[agegp == "old"] * agegp[agegp == "old"], data)
lm2 <- lm(selfesteem[agegp == "old"] ~ importance[agegp == "old"] * agegp[agegp == "old"], data)
And then the same for the younger group?
Or am I going about this the wrong way?