Ding dong, the polling industry is dead – 2020 Scorecard
Bottom Line: By now, not even most in the polling industry are attempting to pretend they aren’t awful. That’s helpful. The first step to recovery is acknowledgement of a problem. Although I don’t think the current polling models are sustainable as constructed regardless. Just how bad do you think the pollsters were this election cycle? Let me put it another way, out of 13 accredited national pollsters how many do you think were accurate to within 1% of the outcome? If you answered zero, you’re correct. That’s right, no national pollster was even within 1% of the national popular vote outcome. In fact, even applying the maximum margin of error implied in their polls doesn’t help much.
The average national pollster predicted Joe Biden would beat Donald Trump by 7.2% according to the RealClearPolitics final average. The actual outcome. 2.9%. This next factoid places the exclamation point on just how useless the industry truly happens to be. The maximum margin of error provided by any pollster was 3.7%. The average of all polls, which is supposed to account still further for margin of error, still wasn’t accurate if you provided maximum margin of error with any pollsters to it. This is to say the industry is demonstrably dead. In fact, these are the only three pollsters who were accurate when applying their maximum margin of error to the results:
- Harris Interactive X, TIPP, Rasmussen
Here are all of the pollsters who were wrong even when applying their maximum margin of error to the results:
- YouGov, Reuters/Ipsos, Change Research, Quinnipiac, RMG Research, Hart Research (NBC/Wall Street Journal), Survey USA, Fox News, Harvard-Harris, Emerson
What does it say when 77 percent of the polling industry misses by more than their largest possible margin of error? Anyone who paid these people should be provided refunds. And who proved to be the worst of the worst? Four pollsters missed the final results by at least 7%, or more than double their maximum margin of error. These utterly incompetent pollsters are:
- YouGov, Change Research, Quinnipiac and Hart Research (NBC/Wall St. Journal)
Going forward you are right to be skeptical of all pollsters. To the extent you entertain the thought at all considering the source has never been more important. If the polls aren’t coming from HarrisX, TIPP or Rasmussen, they’re not worth the time or consideration until proven otherwise. Also, you may recall on Election Day I stated that for as inaccurate as most pollsters were in 2016, they managed to do worse this cycle. It might have seemed odd for me to make that statement prior to knowing the outcome. Now you know why. Most pollsters didn’t provide transparency in their sampling and for those that did, it was generally significantly flawed. Garbage in, garbage out.
Photo Credit: Getty Images