I have four variables including a pre-post variable (two columns), a group variable (X), and two mediation variables (Z1 and Z2). How can I perform a mediation analysis in ANCOVA?
What statistical method and what software do you suggest?
Same way as always. Just add the covariate as an IV to your regression model. Google mediation moderation research virginia R covariate for examples. Best, David Booth
You need to decide whether to have post-pre as the DV, or have post as the DV and pre as a covariate. There is lots on this. Google Lord's paradox. For example: Article Allocation to groups: Examples of Lord's paradox
. This paper notes that one of the things that will inform your choice is how variables influence your two Z variables.