I promised some comments on the methodology behind the recent UNH Green Mountain poll showing Amanda Janoo leading @PhilScott4VT 51-43 (including leaners).
Because UNH doesn't publish its weighted likely-voter age distribution (or at least I couldn't find it) the calculation used its published age figures and assumed uniform age distributions where the polls' age brackets differed.
The discrepancy is potentially significant, because younger voters overwhelmingly support Janoo, so including more of them in the turnout model is going to increase her support.