Comment by arrrg

Election polls were only ever off a couple percentage points. US elections are hard to predict because they are so close. And because of the electoral system. So missing by a bit can mean making a wildly wrong prediction.

This does not apply to opinion questions that show huge differences (not single digit percentage point differences), though there validity, not reliability is a bigger concern, especially since there exists no voting benchmark you can measure against.

Still: I think it’s awfully convenient to just wholly discount actual empirical evidence whenever you feel like it because it might not be perfect. Why exactly do you think your gut feeling is better?