IM for time series with latent variables

[cg09]

The tabular results for a specified lag model with latent variables seems correct, but the implied covariance matrix gives 0.00 for all latent covariances. See attached. picture

Further, there should not be a "model statistics" tab for the IM...model statistics requires that the model be compared with data, and there is no data for the IM.

[jdramsey]

This suggests a more general facility in Compare where you give it a model and a dataset and it gives you back the stats. We discussed this for conditional Gaussian.