CLPM with multiple predictors

[AM]

Hi all,

This is more a theoretical question than a strictly lavaan question, but I don't know where else to ask. I have been searching the literature and I am failing to find an example which I am looking for.

If we are fitting a cross-lagged panel model for testing mediation across three waves, how does the setup look like if we have more than one predictor? The usual examples which I find in the literature have one predictor (X), one mediator (M) and one outcome (Y). I managed to find an example using 2 mediators (M1 and M2), but I can't find anything for the situation when you have several predictors (X1, X2, X3 etc.). Note that this is not the same predictor across three timepoints, but three distinct predictors.

Specifically, I am wondering about the following: In case of three time points, with one X, M and Y, the M in the second wave is regressed on the X and Y in the first wave. However, if we also have X2, would the X2 from the second wave have to be regressed on the X1 from the first wave? Or would the X2 in the second wave be regressed solely on the X2 from the first wave and nothing else?

[Ingo Man]

Hey AM, 

have you checked out this one? https://doi.org/10.4324/9781003263036

Newsom has written quite a bit there about CLMP. And also about covariates (synchronous, time-variying, time-invariant). Determine which kind of predictor you have determines how to include it into the clpm.

Hope this helps a little.

Kindly,

Marcus