President-elect Donald Trump offers his accept speech throughout his choice night rally, Wednesday, Nov. 9, 2016, in brand-new York. John Locher/AP Photo

State-level polls in 2016 were a different story. End the past four years, professionals reached an scholastic consensus about those polls" greatest flaw: a fail by numerous to "weight for education."

"Weighting for education" is an attempt to compensate because that the reality that college-educated world are both much more likely come respond to polls and an ext likely to be Democrats. A vote that"s weight for education takes responses from civilization without college degrees and gives them much more "weight," therefore the poll more accurately to represent the demographics that the location it"s trying come survey.

You are watching: Clinton and trump in the polls

For example, here"s just how a Wisconsin poll that"s weighted for education and learning would play out:

Pollsters contact thousands of people, but many don"t answer their phones. That those who execute answer and take the survey, 60 percent have actually a college degree. Yet here"s the problem: only about 30 percent that Wisconsin"s populace has a university degree, according to the U.S. Census.

If the poll published those results without any type of adjustments, it would disproportionately represent the opinions that college-educated voters, who tend to be much more liberal than their neighbors without university educations.

So, the vote takes the responses from the world without college degrees and also gives them an ext "weight," till the demography ratios in the vote align through the demographics that the state.

According come Ashley Kirzinger, combine director of publicly opinion and also survey study at the Kaiser Family foundation and a member of the American Association because that Public Opinion Research"s poll transparency project, a number of state-level polls didn"t weight for education and learning in 2016.

But she said plenty of have now course-corrected.

"If polls obtain it dorn this year, it"s no going come be since they weren"t weighting because that education," Kirzinger said.

According to an analysis of 2020 state-level polls through the brand-new York TImes earlier this year, 46 percent the the more than 30 state-level polls released between March and also May weighted because that education, compared to 20 percent in 2016.

The Monmouth Poll, a survey based the end of Monmouth college in new Jersey the does state-level polls in very closely watched states choose Pennsylvania, Florida and Arizona, along with a number of others, was one poll the made the switch.

Patrick Murray, manager of the poll, stated Monmouth hadn"t load for education and learning before because it didn"t believe there was exact data available about education levels in specific states. Murray also said in an email interview that, "prior come 2016 there was no far-reaching difference in between how human being with and without a college degree voted."

"The latter condition adjusted dramatically in 2016," the said.

Now, Monmouth weights for education using self-reported data native respondents, in addition to U.S. Census data.

According come Franklin, the Marquette University law School poll has always weighted for education and learning (among other things, prefer age). He claims if his poll hadn"t load for education and learning in 2016, its last October 2016 prediction of a 6-point Clinton victory would have been even further off.

Charles Franklin is the director of the Marquette law School Poll. Angela Major/elafilador.net

"It aided us be less wrong than we would have actually been," Franklin said.

He said, together it turns out, the Marquette poll — and many others — had other major factors to complete with in 2016.

Complicating determinants In 2016: Undecided Voters and Turnout Changes

In the last Marquette poll of 2016, which to be done around a week before Election Day, Clinton led by 6 points among likely voters. However 7 percent of likely voters claimed they to be undecided — that includes people who stated they wouldn"t poll for any type of of the candidates provided by the pollster and those who refused come answer.

Ultimately, those undecided voters damaged heavily because that Trump, according to exit polls on election Day and a variety of post-election studies, including a substantial review the 2016 polling through the American Association because that Public Opinion Research.

In a push conference about the AAPOR study, Courtney Kennedy, manager of survey research study at Pew study Center, claimed undecided voter typically split down the middle in between the two significant party candidates. In 2016, that wasn"t the case.

"In 2016, they damaged decidedly for Donald Trump," she said.

Some world have attributed the so late surge the undecided voters to trump card in 2016 come the visibility of so-called "shy" Trump voters — world who plan to vote for Trump yet didn"t desire to phone call a pollster since of his controversial windy image.

However, a number of studies have refuted the "shy trump voter" theory. Many experts argue if awkward Trump voters existed, polls performed online, which feel much more anonymous, would have shown different results 보다 polls excellent over the phone.

"If there was a solitary methodological defect (with polls), choose shy trump voters, us should have actually seen the digital polls to be right and others to be wrong," Franklin said.

Aside from the Trump-heavy swing that undecided voters, unpredictable voter turnout is thought about the second notable complicating element for part state-level polls in 2016.

See more: Watch Corey Feldman Take A Stand Today Show Performance Sparks Backlash

"(Turnout is) no the biggest of the reasons, yet it was absolutely a factor," said Andrew Mercer, a an elderly research methodologist at the Pew study Center.