R help - Jul 2012 - Mediation analysis

Hello all,

I apologize for the simplistic question, but I have been having some trouble
learning how to do mediation analysis in R. Ideally, I would like to use
Preacher's Bootstrapping test for mediation (Preacher & Hayes, 2004). I
have
attempted to use the mediate package to set this up, using code that looks
basically like this:

model.m <- lm(data$outcome ~ data$mediator + desirdata$predictor)
model.y <- lm(data$outcome ~ data$predictor)
age1 <- mediate(model.m, model.y, treat = "predictor", mediator
"mediator", conf.level=.95, boot=TRUE, sims=50)

 However, I get an error indicating that the number of observations doesn't
match across the two models. This may be true,  because it's a somewhat
large sample (13,000) with a pretty substantial amount of missing data, but
I'm not sure how to deal with that issue. Is there any way to get this to
run given that there may be different numbers of observations in the two
models?

Thanks,

Jessica



--
View this message in context:
http://r.789695.n4.nabble.com/Mediation-analysis-tp4638577.html
Sent from the R help mailing list archive at Nabble.com.

Subset your data first and/or use the data= and subset= arguments to 
lm() so the same data is used in all models.

On 7/31/2012 2:14 PM, jawortman wrote:
> Hello all,
>
> I apologize for the simplistic question, but I have been having some
trouble
> learning how to do mediation analysis in R. Ideally, I would like to use
> Preacher's Bootstrapping test for mediation (Preacher & Hayes,
2004). I have
> attempted to use the mediate package to set this up, using code that looks
> basically like this:
>
> model.m <- lm(data$outcome ~ data$mediator + desirdata$predictor)
> model.y <- lm(data$outcome ~ data$predictor)
> age1 <- mediate(model.m, model.y, treat = "predictor",
mediator > "mediator", conf.level=.95, boot=TRUE, sims=50)
>
>   However, I get an error indicating that the number of observations
doesn't
> match across the two models. This may be true,  because it's a somewhat
> large sample (13,000) with a pretty substantial amount of missing data, but
> I'm not sure how to deal with that issue. Is there any way to get this
to
> run given that there may be different numbers of observations in the two
> models?
>
> Thanks,
>
> Jessica
>
>
>
> --
> View this message in context:
http://r.789695.n4.nabble.com/Mediation-analysis-tp4638577.html
> Sent from the R help mailing list archive at Nabble.com.
>