> When I run the causality analysis, do I simply include the whole time series,
> but covary out the regressors I am NOT interested in (e.g., NI and PI)...OR, do
> I need to enter in the data from the regressors for only those conditions I am
> interested in (which would be difficult due to overlap between regressors of
> interest and those of no interest)?
Both approaches are fine, but covarying out the effects of no interest as part of the vector auto-regressive model is usually preferable.
Gang