Don’t Freak Out (Yet) About Wisconsin — Poll Showing Trump Winning Is Heavily Weighted Toward GOP Voters’ Views
There was minor panic on social media this week, as a key swing state was polled showing President Donald Trump defeating all six of the top Democratic candidates vying to unseat him later this year.
Wisconsin was a state that, up until 2016, had voted “blue” all the way since the 1980s. Trump won the state in the last presidential election by a mere 22,748 votes — or just under 0.7 percent of the vote.
But a recent poll from Quinnipiac University shows that Trump is outperforming every viable Democratic candidate in the 2020 election.
According to the poll, Trump beats Joe Biden by 7 points, Bernie Sanders by 7 points, and Elizabeth Warren by 10 points.
He also trounces Amy Klobuchar by 11 points, Pete Buttigieg by 8 points, and Michael Bloomberg by the same margin.
Trump does indeed have a higher rate of popularity in the state than the nation rates him overall. But there’s one contextual thing missing from the polling data: it’s heavily weighted toward emphasizing Republican voters’ preferences.
The poll itself says as much. Of the voters it interviewed, 32 percent were Republican-leaning while 26 percent were Democratic-leaning.
For comparison, in the 2016 election, exit poll data found that Democrats and Republicans taking part in the presidential election were equal, with 32 percent of Wisconsinites saying they were aligned with one party, and 32 percent saying they supported the other.
— Quinnipiac University Poll (@QuinnipiacPoll) February 20, 2020
Quinnipiac writes that it “uses statistical weighting procedures to account for deviations in the survey sample from known population characteristics.” That means it tries to make up for the fact that its own sample might not be in line with what the actual state looks like.
But when it does that, the margin of error increases. “When including the design effect, the margin of sampling error for this study of registered voters is +/- 4.2 percentage points,” the survey says.
That means, if Quinnipiac does try to “account for deviations” between its sample of party identifiers versus what the actual state looks like, for many of the candidates listed above, it’s a statistical tie — not a blowout in favor of Trump.
For example, if you look at the race between Trump and Biden, Quinnipiac says the president leads the challenger by a spread of 49 percent to 42 percent, respectively. But if you account for the 4.2 percentage margin of error, the results could actually range anywhere between those numbers and Biden leading Trump, by a spread of 46 percent for the former VP to just under 45 percent for the incumbent.
Indeed, taking into account the widened margin of error, Biden, Sanders, Bloomberg, and Buttigieg could all theoretically be ahead of Trump.
Quinnipiac doesn’t say if partisan identifiers are weighted or not in its fine print. But at best, if those numbers from Wisconsin are weighted, the widened margin of error demonstrates that four of the six races are statistical ties; at worst, if political identifiers aren’t weighted, Quinnipiac made a questionable decision in looking at numbers that oversampled Republican voters.
Wisconsin will indeed be a close race — there can be no doubts about that. But other polling data, such as the Marquette Law School poll within the Badger State, has found the race to be much tighter than the Quinnipiac poll suggests.
Polls have to be looked at collectively. One poll showing a wide spread of numbers isn’t an indicator of realities on the ground. It COULD BE that Trump is going to win Wisconsin by such a wide spread — but with one poll saying as much, and all others saying it’s a closer race, it’s more likely that this recent poll is an outlier.