4 ms·

So you exclude any pollster that's been wrong once? Doesn't that strike you as very un-statisticianlike?