[R] R squared change value for a moderation effect

[Marcel Meyer]

Hello,

I want to test a regression model with neuroticism as focal predictor, agreeableness as moderator and RT variability as dependent measure (covariates: attentional control and mean RT). Previously, I have used the modprobe macro in SPSS by Andrew Hayes for this (for full reference see end of message). I am in the process of transitioning to R, however, and would like to learn how to run a similar routine there. I have set up my regression model as follows:

    m3<-lm(data=stp2_sub2, all_SD~Neuroticism*Agreeableness+Attentional.Control+all_RT, na.action=na.omit) # full interaction model
    m33<-lm(data=stp2_sub2, all_SD~Neuroticism+Agreeableness+Attentional.Control+all_RT, na.action=na.omit) # reduced model

I know that I can obtain F-change and p-change, using:

    anova(m3, m33) # provides F-change and p-change

What I still don’t know yet is how to obtain the R squared change value, which gives me the effect size of the interaction effect. Any advice on this would be much appreciated.

Best,
Marcel

Reference:
Hayes, A. F., & Matthes, J. (2009). Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementations. Behavior Research Methods, 41(3), 924–36. doi:10.3758/BRM.41.3.924
	[[alternative HTML version deleted]]