I have two tests applied concurrently to students in a longitudinal manner (3 occasions): one is theoretical and one is practical.
The theoretical and practical tests use the same items in all occasions. The theoretical test has 10 dichotomous items and the practical test has only 1 dichotomous item. These given conditions unfortunately cannot be changed.
I would like to assess how growth in theoretical knowledge impacts practical performance.
I have tried several models that terminated normally and quickly but I am not sure whether any of those attempts was an adequate strategy to achieve my objective.
Why don't you at each time point specify a factor measured by the 10 theoretical tests and at each time point let that predict the practical test item? So with 3 occasions you have 33 columns in your data, 11 per time point.