Credible Granger-causality analysis appears to require post-sample inference, as it is well-known that in-sample fit can be a poor guide to actual forecasting effectiveness. However, post-sample model testing requires an often-consequential a priori partitioning of the data into an "in-sample"...

As a contribution toward the ongoing discussion about the use and mis-use of p-values, numerical examples are presented demonstrating that a p-value can, as a practical matter, give you a really different answer than the one that you want.