Correcting for repeated testing with 4 mediation analyses?

[Yulir1993]

Hi guys,

First time poster here, sorry if I post this at the wrong spot or something like that.

For my thesis if have the following 4 hypotheses:

1. Emotion regulation (partially) mediates the relationship between cognitive behavioural therapy (CBT) and social anxiety severity.
2. Emotion regulation (partially) mediates the relationship between schema therapy and social anxiety severity
3. Emotion regulation (partially) mediates the relationship between CBT and avoidant personality disorder severity
4. Emotion regulation (partially) mediates the relationship between schema therapy and avoidant personality disorder severity

To test these hypotheses, I want to perform 4 mediation analyses. For that purpose, I need to perform 3 regression analyses per hypothesis to test if all the paths (a, b, c) are significant before I can do the mediation analysis.

This means I'll have to do at least 12 regression analyses, and 3 mediation analyses.

Do I have to correct for multiple testing?
If yes, how can I do that?
I can't really find anything on correcting for multiple testing with mediation analyses online...

I use IBM SPSS Statistics version 23.

Thanks in advance!

 

[psychcy]

Use Structural Equation Modeling if you can. There's a SPSS module called AMOS (you'd have to pay for it); or you could use R to analyze it. It'll allow you to test a global fit. Just 1 test essentially.