currently I am dealing with the following problem:
I have a mediation model which I want to test in my multilevel data
X (=Stress)
Y (=Flow)
Mediator (=Stress-related behavior)
Moderator 1 (=Personality Factor 1)
Moderator 2 (=Personality Factor 2)
So my Hypothesis is that Flow (Y) occurs at a moderate stress level (X) and that this relationship is mediated via stress-related behavior (M). Further, I assume that the relationship between stress-related behavior and flow depends on 2 personality traits (=Moderators).
I am using mediation package in R to analyse data.
I am not quite sure, if I have specified the correct terms and maybe you can help me:
med.fit<-lmer(Mediator~X+I(X^2)+(1|VP01_01), data=MT3)
out.fit<-lmer(Flow~X+I(X^2)+Mediator*Trait1+Mediator*Trait2+(1|VP01_01),
data=MT3)
med.out<-mediate(med.fit, out.fit, treat="X", mediator= "Mediator", sims=100)
--> Right now, my code only specifies "X" and not "X" and "I(X^2)" as treat. How has the code has to look like so that quadratic effects are tested? --> Is the b-path moderation specified correctly?
Right now, I get this output, where it looks like that the mediation is not significant (ACME) but I'm not sure if I did it correclty:
Estimate 95% CI Lower 95% CI Upper p-value
ACME -0.00676 -0.01988 0.00777 0.34
ADE -0.10578 -0.18514 -0.02457 0.02
Total Effect -0.11253 -0.18747 -0.03499 0.00
Prop. Mediated 0.06570 -0.05545 0.26450 0.34
Thanks in advance for your help :)
