The Student Room Group
Uni students - Complete our survey >>
Uni students - Complete our survey >>
Uni students - Complete our survey >>

Mediating & Moderating Variables

Hi all!
I am currently writing up my third year dissertation, and all was well until I decided to google the differences between moderating and mediating variables. Prior to this, I was convinced that one of my variables had a mediating effect, but after reading up, I'm now really confused and was wondering if anybody could help me! :redface:

Basically, one of my predictor variables (P) has a positive and significant correlation with the dependent variable (B) in my Pearson's Correlation, so when taken alone and tested without the effects of the other variables in the model.
My hierarchical regression showed (P) to become non-significant after another variable (F) was added. The other two predictor variables (U) and (R) had little effect on (P), in that when they were added to the regression with (P), (P) remained significant. Only after addition of (F) did (P) become non-significant.

So, if anyone could tell me whether (F) has a MEDIATING or a MODERATING effect over the relationship between (P) and (B), I would be very grateful!
Also, sorry if this is a dummy question - the definitions for mod and med seem so similar, hence my confusion!!
Thank you :smile: x
Original post by rrreb
Hi all!
I am currently writing up my third year dissertation, and all was well until I decided to google the differences between moderating and mediating variables. Prior to this, I was convinced that one of my variables had a mediating effect, but after reading up, I'm now really confused and was wondering if anybody could help me! :redface:

Basically, one of my predictor variables (P) has a positive and significant correlation with the dependent variable (B) in my Pearson's Correlation, so when taken alone and tested without the effects of the other variables in the model.
My hierarchical regression showed (P) to become non-significant after another variable (F) was added. The other two predictor variables (U) and (R) had little effect on (P), in that when they were added to the regression with (P), (P) remained significant. Only after addition of (F) did (P) become non-significant.

So, if anyone could tell me whether (F) has a MEDIATING or a MODERATING effect over the relationship between (P) and (B), I would be very grateful!

Also, sorry if this is a dummy question - the definitions for mod and med seem so similar, hence my confusion!!
Thank you :smile: x



The definitions for mediator and moderator are not similar, its just hard to get your head around at first! Read http://www.ncbi.nlm.nih.gov/pubmed/11384888 for a description. Its a highly cited article (600+) so should be a good source. The picture on page 6 is very helpful.

I don't think you can test for mediation the way you described... see http://www.upa.pdx.edu/IOA/newsom/semclass/ho_indirect.pdf for one way of doing it

All you can say in this case is that P and F do not have unique effects on B. When one is controlled for, the other is non-significant. You haven't tested for mediation/moderation.

An example of a moderator may be sex: if P correlates with B in males but not females, then sex MODERATES the relationship since a moderator "According to Baron and Kenny (5), conceptually a moderator (A) specifies on whom or under what conditions another variable will operate to produce the outcome." (Kreamer article i cited).

Quick Reply

Related discussions

Latest

Trending

Trending

Articles for you