Hey, I am trying to figure out how to see if a particular variable is a mediator in SPSS. Does anyone know how to do this?
Thanks
Thanks
Mediators - finding
- Thread starter edieb
- Start date
mmonte4 said:i believe Dr. Davis would know
Unfortunately, there's no magic button for mediation.
Ideally, you would examine mediators through Structural Equation Modeling, which you cannot do on SPSS.
However, a quick and dirty way to examine mediators is to run a series of regressions.
So let's say:
IV = A
Mediator = B
DV = C
you will first need to demonstrate that A significantly predicts B (simple correlation or simple regression), A significantly predicts C, and that B significantly predicts C. If those basic assumptions are met, you can then proceed with the multiple regression model regressing C onto both A and B. If B continues to significantly predict C, whereas A no longer predicts C, then you can conclude mediation.
However, there is a possibility that A will still continue to significantly predict C, but that by adding B, there was a statistically significant decrease in the amount of variance in C that A predicts. This is partial mediation, and can be tested using other kinds of statistics (e.g., Sobel Test).
Essentially, it is wise to proceed with these analyses only if you a) really understand, conceptually, what it is you're looking at, and b) if you really understand the statistics behind what you are doing. Cookbook statistics is never a good idea.
edieb said:
You're welcome.
On this website, there is a great list of references. I would recommend reading through several, if you have the time. I particularly like the Kenny, Kashy, & Bolger 1998 chapter.
http://davidakenny.net/cm/mediate.htm
