The link you cite starts by recommending NOT doing the Sobel test unless you have a large amount of data. Is that the case in your study? Boostrapping, instead, is recommended. The Preacher and Hayes article cited there goes into a lot of detail on these methods and is worth reading. It points out that Amos can do these analyses, which may be easier than working through the math using Statistics. Amos reads SPSS sav files, but it is a separate install. Your university may be able to provide an Amos license.
------------------------------
Jon Peck
------------------------------
Original Message:
Sent: Sun June 07, 2020 11:30 AM
From: Maaike van der Marel
Subject: Doing the Sobel Test when you hav used multiple dummies as IV's in your multiple regressions
Dear Experts,
My thesis instructor advised me to do a Sobel Test to test the degree of mediation in the model that I propose. I'm investigating the extent to which depression mediates between sexual orientation of women and BMI. Thus, as independent variable I have used three dummies: heterosexual women, lesbian women and bisexual women, meaning that I have two standardised and unstandardised coëfficients per model. How do I incorporate these in the Sobel Test; http://quantpsy.org/sobel/sobel.htm ?
I hope you can help me.
Kind Regards,
Maaike
------------------------------
Maaike
------------------------------
#SPSSStatistics