Most Super Tuesday polls did not predict the Joe Biden sweep that was about to occur.
But that wasn’t necessarily the fault of the polls. It was a combination of a rapidly changing race that saw three candidates — businessman Tom Steyer, former South Bend, Indiana, Mayor Pete Buttigieg, and Sen. Amy Klobuchar — drop out from Saturday to Monday. There were also large numbers of voters who decided late, which exit polls showed benefited Biden hugely.
The morning after Super Tuesday, some were wondering aloud why most of the polls seemed to miss Biden’s strength not just in Southern states where it was predicted he would do well, but in Northeast states like Maine and Massachusetts, both of which Biden won.
If you combine tonight’s results with the 2016 general election results, the question no one seems to be asking is:— michaelharriot (@michaelharriot) March 4, 2020
Why are the polls getting it so wrong?
First of all, we should note that polls shouldn’t be judged for whether or not they predict a winner outright; rather, they should be judged on how close they are to the margin of victory, as election watcher Nate Silver has written about. That metric becomes less clear, however, with candidates dropping out and other big events.
With a crush of news happening in the days leading up Super Tuesday, some pollsters openly admitted they had absolutely no clue what might happen. Buttigieg and Klobuchar were taking votes away primarily from Biden (especially in whiter states), so their dropouts and subsequent Biden endorsements meant a serious consolidation around the former vice president.
A few polls did contain some tea leaves of what was to come. Progressive firm Data For Progress predicted Sanders’s wins in Utah and Colorado, as well as his likely California win. They also predicted Biden’s romp in the South. And while that poll didn’t capture his wins in Maine, Minnesota, or Massachusetts, it did show him suddenly surging into second place in those first two. And Morning Consult’s national poll released Tuesday morning (after Klobuchar and Buttigieg dropped out) picked up on a sudden Biden surge after waves of establishment endorsements.
It’s also worth noting that, despite the narrative driven by Biden winning states, delegates are the actual metric that matters in this phase of the primary. By that metric, Biden and Sanders are closer; as of Decision Desk’s latest count, Biden has won 461 delegates overall, compared to 404 for Sanders. But Sanders’s number is likely to go up because he’s ahead in California — the only state left to count, and the one with the biggest cache of delegates.
So, while Biden definitely overperformed, polls generally showed a two-man race for delegates between Biden and Sanders going into Super Tuesday. And that’s essentially what it was.
In specific states, the polls were off simply because they were out of date after less than a week. The news was moving too fast for them to keep up. Perhaps the starkest example of this occurred in Massachusetts, where Biden won despite most polls showing him somewhere in the middle of the pack, behind Bernie Sanders and Sen. Elizabeth Warren (Massachusetts is Warren’s home state).
When I talked on Monday to MassINC Polling’s president Steve Koczela, who polls with Boston NPR affiliate WBUR and conducted his most recent one from February 23 to 26, he freely admitted he had no idea what was going to happen on Super Tuesday — and literally didn’t have enough time to poll to find out.
“We did the poll just last week but since then it feels like two years ago,” Koczela said. “The news is moving so fast.”
In fact, Koczela said, using a poll after three candidates dropped out would be “a pretty dicey proposition.”
“It’s fun to watch, honestly,” he said. “There’s no need for us to know tonight what might happen tomorrow. We’ll know tomorrow what’s happened.”
What ultimately happened was Biden went from fifth to first place in the poll Koczela conducted, with Sanders in second and Warren in third.
Late deciders went for Biden
Part of the reason the election was so volatile leading up to Super Tuesday was the number of people who decided late. Exit polls show these voters were a big boost to Biden.
Early NBC exit polls showed 30 percent of Super Tuesday voters chose a candidate in the last few days before the primaries; of that group, 47 percent chose Biden compared to 18 percent who chose Sanders. (A necessary disclaimer about exit polls: They don’t always capture the results accurately and it can take months to get an accurate picture from all the data collected.)
A similar story played out in the Washington Post’s exit polling, which showed 47 percent of voters who said they decided in the last few days went to Biden compared to 17 percent who went for Sanders. Sanders had a slight lead with voters who had decided earlier than that at 33 percent compared to Biden’s 29 percent.
This also seems to be backed up by the disparities seen in early voting versus day-of voting in Texas, another state Biden won. When early vote returns were coming in, Sanders had a several-point lead. But as day-of voting returns streamed in, Biden eventually took the lead and ended up winning the state by close to four points.
Election watchers were startled at how disparate the late deciders were from early voters, as surmised by this tweet from Cook Political Report editor Dave Wasserman.
In my lifetime, I've never seen an election where late deciders have broken so sharply from early voters than this one.— Dave Wasserman (@Redistrict) March 4, 2020
And as CNN’s Harry Enten tweeted, late deciders even provided Biden a boost in Bernie Sanders’s home state of Vermont.
Here's a stat for you... Joe Biden won the late deciders (last few days)... in Vermont.— (((Harry Enten))) (@ForecasterEnten) March 4, 2020
All of this paints a picture of an extremely unpredictable and fast-moving news cycle. Polls are a snapshot in time, but Super Tuesday polls couldn’t move fast enough to give us the bigger picture.