“The political polling profession is done. It is devastating for my industry.”
— GOP pollster Frank Luntz, quoted by Axios.
I am posting this at the risk of it being made by obsolete by ongoing events, but Lutz may be wrong about what yesterday’s election means for the polling industry. In fact, at the moment it is looking like the final results in the presidential race will be consistent with the polling and at least a normal polling error.
The final polling average by The Upshot was Biden +8. A normal polling error – like the one we saw in 2016 – would be about three points. That means Biden +5 would be consistent with The Upshot’s final polling average. As it stands now, Biden has a two point lead in the popular vote with another 2.9 million or more votes to be counted. Most of that is expected to come his way. The end result will likely be Biden +4, but it could be even bigger (and it could be smaller). If it is Biden +4 then it would be nitpicking to say that the polls missed big.
The final electoral tally for Biden looks like it will be anywhere from 270 to 322, with a safe bet being 291. This is fewer votes than The Upshot or others projected, but not by a lot. And as long as Biden wins the electoral vote, then the forecasts from FiveThirtyEight and The Economist were correct. While they projected certain electoral vote probabilities, a win is a win.
We won’t know for a while how close or far the polling was from the actual topline results, let alone the demographic categories. We don’t know yet whether the polling misses this year were within normal ranges or worse. But, at the moment, Frank Lutz is overreacting - at least as concerns the presidential polling. He might be right about some of the Senate polling, but we have to remember that the polling in those races was not as frequent or diverse as we would need to feel really comfortable with polling averages for them.