PROCESS macro help!

[psychoptimist]

I doing simple mediation using PROCESS. My IV and mediator are continuous but my DV is dichotomous.
The output shows that path a is significant, that path b is significant and path c' is not significant. However, I do not know where in the PROCESS output it shows the direct effect (path c)?

Any help is appreciated! Thanks!

---

Members don't see this ad.

 

[MCParent]

I doing simple mediation using PROCESS. My IV and mediator are continuous but my DV is dichotomous.
The output shows that path a is significant, that path b is significant and path c' is not significant. However, I do not know where in the PROCESS output it shows the direct effect (path c)?

Any help is appreciated! Thanks!

The direct effect c is just a regression with X as the IV and Y as the DV.
But you don't need it. You're about to check whether the c path drops to non-significance after including the mediator, which is an old approach. The significance of the indirect effect is all you need to establish medication. The sig value of c delta is dependent on sample size; you could have a drop from .85** to .25* and that's "partial" mediation and a drop from .25* to .24 (ns) and that's "full" mediation, which makes no sense

 

[psychoptimist]

The direct effect c is just a regression with X as the IV and Y as the DV.
But you don't need it. You're about to check whether the c path drops to non-significance after including the mediator, which is an old approach. The significance of the indirect effect is all you need to establish medication. The sig value of c delta is dependent on sample size; you could have a drop from .85** to .25* and that's "partial" mediation and a drop from .25* to .24 (ns) and that's "full" mediation, which makes no sense

You are a lifesaver! Thank you!

 

[MCParent]

You are a lifesaver! Thank you!

FYI, Kenny writes about this in more detail on his web site:
http://davidakenny.net/cm/mediate.htm#IE

You might still encounter thesis committee members, reviewers, etc., who expect to see c and c' so maybe you have a reason for wanting to do a Sober test anyway, or you can read that and more recent papers on mediation to be able to tell committee members etc why you're not looking at it.

HTH!

 

[cara susanna]

Yup, use the confidence intervals of the indirect effect. If the confidence interval doesn't include zero, it's significant and you can say there was mediation.

 

[EmotRegulation]

Yes, though also see this: http://onlinelibrary.wiley.com/doi/10.1002/jclp.22298/full

 

[Ollie123]

Yes, though also see this: http://onlinelibrary.wiley.com/doi/10.1002/jclp.22298/full

A citation worth reiterating. Statistical mediation is not synonymous with causation.

The amount that these concepts are conflated in the literature is scary.