I'd like to estimate a multi-level AR model, in which Y at t is predicted by Y at t-1 and X at t. This model is equivalent to Figure 5 in McNeish & Hamaker (2019).
If I include the INTERVAL option, I receive the following error:
“ When expanding time series data with the TINTERVAL option, all variables on the within level should be dependent variables. Problem with: X”
I don't receive an error if: - Estimate a multi-level AR model with a predictor, as described above, without using the TINTERVAL function. - I estimate a model with an autoregressive effect for Y and using TINTERVAL (without any other predictors) - Estimate a multi-level VAR model with 2 outcomes (Y and X), and using the TINTERVAL function.
Is it correct that the TINTERVAL function could only be used to estimate multi-level VAR models, but cannot be used to estimate multi-level AR models with any other predictors in the model?