After the polls failed in 2016 to predict the outcome of the Presidential election, there's been some questions about how accurate they will be in the 2020 election cycle.
After the polls failed in 2016 to predict the outcome of the Presidential election, there’s been some questions about how accurate they will be in the 2020 election cycle.
So far, the answer is: not very.
To be fair, Super Tuesday was a tough one to predict with three candidates dropping out after early voting started. It’s hard to predict how voters will behave when the question you’re asking dramatically changes after you ask it.
However, the polls around the Bay State in the three days before the election did not reflect the surge that pushed former Vice President Joe Biden to the top.
Here are the actual results at 78 percent reporting:
Joe Biden, 33.6% Bernie Sanders, 27.3% Elizabeth Warren, 20.7% Michael Bloomberg, 11.7% Pete Buttigieg (dropped out before Super Tuesday) 2.6% Amy Klobuchar (dropped out before Super Tuesday), 1.2% No Preference, .8% Tulsi Gabbard, .8% Deval Patrick (dropped out Feb. 12), .5% Tom Steyer (dropped out before Super Tuesday), .5%And here are some of the predictions:
Swayable polled 917 people in Massachusetts from March 1 and 2, predicting:
Sanders, 27% Bloomberg, 18% Biden, 17% Warren, 15% Buttigieg, 11% Klobuchar, 5% Steyer 4% Gabbard, 1%Data Progress polled 301 people from Feb 28 to March 2, predicting:
Warren, 28% Sanders, 26% Biden, 26% Bloomberg, 15% Buttigieg, 2% Klobuchar, 1% Gabbard, 1%FiveThirtyEight averaged all of the polls that have been run together, finding for active candidates:
Sanders, 24% Warren, 20% Biden, 18% Bloomberg, 14% Gabbard, .4%None of the polls got the order right. They didn’t predict Biden’s rise. They all over-estimated Bloomberg compared to his performance. The Swayable poll over-estimated Warren’s performance, and the Data Progress poll under-estimated it.
Where they were most accurate was in predicting Sanders’s performance, as while he didn’t win as the Swayable polls indicated, he did take 27% of the vote, which is what Swayable predicted and very close to Data Progress’s 26% prediction. This seems to indicate Sanders supporters made up their minds earlier and were more consistent than backers of other candidates.
The other one they got right was Gabbard, who took just under 1% of the vote even with three candidates dropping out leading up to the election.