Gang, thanks again for your assistance.

Actually, the 4th column is an alternative explanatory variable in lieu of my original one. I had thought that if I include it in my script but not reference it in the model, then there would be no harm. I hope that's the case!

The 5th column t-stat does seem generally in line with the whole model F-stat, although I haven't tested 5: 0 1 2 3 4.

I think I've sorted out the issue with that problematic covariate explaining a lot of my effects. As the coeff #2 is categorical, doing a quick and dirty ttest stratifying across that variable does yield similarily significant results in my cluster of interest.... oh well...