|
Discriminating among Linear Models with Interdependent Disturbances
Kenneth M. Gaver
Martin S. Geisel
Abstract
A Bayesian procedure for comparing linear models with non-scalar covariance matrices is developed. For the case of first order auto-regressive disturbances, an approximate expression for the error in the posterior odds due to ignoring the serial correlation is given, and it's accuracy is investigated via sampling experiments.
|
