Hi Patrick,

there is nothing special to a CLPM. You simply add the t1-versions of each variable as autoregressive predictor and add the cross-lagged effects or--alternatively--the two synchronous effects (X_t2 --> Y_t2 and vice versa). In this case, you should test the error covariance between both).

If you have latent variables, correlate the error of each indicator with its successor.

Hope that helps. I hope the old Phil 2 in Gießen is still alive. Miss that a lot.

All the best,

--Holger