based on this wonderful post, I am trying to run my own moderated mediation, but my model has a dichotomous moderator only on b path.
The code goes like this:
library(mediation)
Model.m<-lm(mpg ~ drat, data = mtcars)
Model.y<-lm(disp ~ mpg*vs + drat, data = mtcars)
Mod.Med.vs1 <- mediate(model.m = Model.m, model.y = Model.y,
covariates = list(vs = 1),
boot = TRUE,
boot.ci.type = "bca", sims = 2000, mediator="mpg", treat="drat")
# Here I get the error
Mod.Med.vs0 <- mediate(model.m = Model.m, model.y = Model.y,
covariates = list(vs = 0),
boot = TRUE,
boot.ci.type = "bca", sims = 2000, mediator="mpg", treat="drat")
Mod.Med <- mediate(model.m = Model.m, model.y = Model.y,
boot = TRUE,
boot.ci.type = "bca", sims = 2000, mediator="mpg", treat="drat")
I get an error right after running Mod.Med.vs1:
Error in `[.data.frame`(pred.data.t, , vl) : undefined columns selected
Importantly, if I multiply drat with vs also in Model.m it works fine, but again, I am not interested in investigating the influence of the moderator prior to the mediator.
Any suggestions?