> It outputs 19 GLTs, 18 for each row of the glt matrix plus an overall one, and I
> assume this last one (three_levels_GLT_Fstat) is the one I should be looking at?
The 1st 18 tests are 9 lags by two comparisons per lag, and, yes, the last one (F-statistic) is an overall test.
> Incidentally, this Fmap is very similar to the overall R^2 (Full_R^2) map. I guess this is expected?
That is right because this F-statistic also tests a model with all the regressors of interest + baseline versus a model of pure baseline.
Gang