How to do Mediation across pre-post timepoints?

[roquera]

Hi all, I am looking into analyzing potential mediation and not sure how to approach it with two timepoints. For some context, participants are measured at two timepoints: pretreatment (T1) and posttreatment (T2). All participants receive the same treatment. At pre and post, they complete measures for anxiety, attention, and depression. I would like to address the following questions (all variables continuous):

Is change in anxiety from T1 to T2 mediated by attention at T1?

Is change in anxiety from T1 to T2 mediated by change in attention from T1 to T2?

I am thinking difference scores are going to be involved somehow? Any advice on how to proceed would be greatly appreciated!

 

[dhvalden]

Cross lagged SEM or Path analysis. Mplus or Lavaan (R). Also, there is a less powerfull ad-on for spss called "Process" that can do part of the analysis (can only do path). Difference scores is gonna need to be a new variable mediating the IV and the DV.