Sanders’s surprise win in Michigan keeps pollsters guessing

Sanders’s surprise win in Michigan keeps pollsters guessing
© Getty Images

The shocking results of Tuesday’s Michigan Democratic primary have renewed the debate over the reliability of presidential polling and called into question one of the primary tools news outlets use to cover political horse races.

Hillary ClintonHillary Diane Rodham ClintonCommunion vote puts spotlight on Hispanic Catholics Trump's biggest political obstacle is Trump The Memo: Some Democrats worry rising crime will cost them MORE entered election day in Michigan with a 21-point lead over Bernie SandersBernie SandersSenators say White House aides agreed to infrastructure 'framework' Briahna Joy Gray: Biden is keeping the filibuster to have 'a Joe Manchin presidency' On The Money: Biden to fire FHFA director after Supreme Court removes restriction | Yellen pleads with Congress to raise debt ceiling MORE in the -RealClearPolitics average of polls.


Not one public opinion survey conducted this cycle found Sanders in the lead there. The closest he ever got was within 7 points in a Fox 2 Detroit-Mitchell survey from last September.

That led many to believe that Tuesday would be the day Clinton secured her grasp on the nomination and could pivot fully to the general election. Many media outlets turned their attention to the more chaotic scene on the GOP side, believing that to be the hotter race.

But in what he called a “repudiation of the pundits” and a “repudiation of the polls,” Sanders defeated Clinton in Michigan by 1.5 points in a race that couldn’t be called until nearly all votes were counted.

A chastened Clinton campaign held an impromptu conference call with reporters on Wednesday to discuss the results. Campaign manager Robby Mook issued a stark warning to reporters about relying on public opinion surveys to shape their coverage.

 “Whatever the public polling is saying right now, I want to reiterate, it was very inaccurate last night,” Mook said. “We would all be well advised to treat the public polling coming out today and this week with skepticism.”

Michigan is not the only example of a polling failure this cycle.

Surprising disparities between the polls and the election day outcomes have repeatedly roiled the races in both parties, provoking new storylines from the media about candidates gaining or losing steam.

Republican front-runner Donald TrumpDonald TrumpIran claims U.S. to lift all oil sanctions but State Department says 'nothing is agreed' Ivanka Trump, Kushner distance themselves from Trump claims on election: CNN Overnight Defense: Joint Chiefs chairman clashes with GOP on critical race theory | House bill introduced to overhaul military justice system as sexual assault reform builds momentum MORE was favored to win the Iowa caucuses, but instead Ted CruzRafael (Ted) Edward CruzNew Jersey governor tweaks Cruz on Cancun over moving truck quip Hirono tells Ted Cruz to stop 'mansplaining' Senate Republicans: Newly proposed ATF rules could pave way for national gun registry MORE delivered a record-setting victory.

Trump was heavily favored in Louisiana and Kentucky last Saturday but again had to withstand a late charge by Cruz in both states, only narrowly winning in both.

In Louisiana, it appeared that some networks might have to retract projections of a victory for Trump, who had built up a big lead among early voters that nearly disappeared as later voters broke sharply in favor of Cruz.

Trump was also favored to win the Kansas caucuses, but Cruz triumphed in a rout.

This led some pundits to believe that Trump’s polling was inflated heading into the March 8 contests. One survey found John Kasich ahead of Trump in Michigan, but the front-runner reasserted his dominance and sailed past Kasich in the state.

Some of these disparities can be chalked up to a scarcity of polling — there was only one public opinion survey in Kansas before voters caucused there.

But industry experts readily acknowledge that the 2016 race presents a unique set of challenges for pollsters that are specific to this cycle.

There has been record-breaking turnout in many contests, with scores of new voters — many of them young — participating for the first time.

Furthermore, the same disruptions that outsider or insurgent candidates like Trump, Cruz and Sanders bring to the race the as a whole have had a similarly destabilizing affect on polling.

Democrats and Republicans are crossing lines to participate in open primaries, either swayed by the energy of the insurgent or intent on stopping a candidate in another party by casting a protest vote.

That, in addition to the furious pace of sharing on social media, has fostered an unstable political environment where races are more likely to break sharply at the last minute than they have in the past.

“Pollsters are scrambling to measure the intensity of emotions voters have with these outsider candidates,” said pollster John Zogby. “Plus, you have the complete breakdown of the establishment. People are no longer predictably Republican or Democrat.”

And pollsters continue to struggle with the issues they’ve long dealt with in the presidential primary process.

State caucuses are notoriously difficult to survey, and pollsters are still seeking the right sampling balance between landlines, cellphones, Internet surveys and new mobile technologies.

That confluence of complexities conspired to keep Gallup, one of the most admired polling outlets in the nation, out of the 2016 game after a 2012 effort in which they were off the mark and widely criticized.

“The polling will be more erratic with these sociopolitical changes and new technologies,” said Zogby.

Of course, sometimes the reasons for a poll being wrong are simpler.

“Events still happen that can change things,” said GOP pollster David Winston, a veteran of Newt Gingrich’s 2012 Republican presidential campaign.

That could be what happened in Michigan between Sanders and Clinton.

While it’s true that Sanders outspent Clinton in Michigan and campaigned there more often, the debate in Flint just days before the election might have had an impact.

There, Sanders hammered Clinton for supporting trade policies he said have eliminated huge numbers of jobs in the industrial Midwestern state.

Clinton, meanwhile, launched a questionable attack against Sanders for purportedly opposing the auto industry bailout. That line of attack was deemed suspect by many and might have backfired.

Clinton also did not do as well among African-American voters in Michigan as she has done in the South. That has pollsters scratching their heads as to whether Michigan was a fluke or a cause to recalibrate their models.

The sum total should offer hope to the underdogs, like Marco RubioMarco Antonio RubioOvernight Defense: Joint Chiefs warn against sweeping reform to military justice system | Senate panel plans July briefing on war authorization repeal | National Guard may have 'training issues' if not reimbursed Senate panel plans July briefing on war authorization repeal Rising violent crime poses new challenge for White House MORE, who must win in his home state of Florida on March 15 but trails Trump by double digits there in polls.

“There seems to be equilibrium in Florida right now, and if that is sustained, then Trump wins,” Winston said. “Marco Rubio’s whole strategic plan is to disrupt that equilibrium in a way that’s possible. It’s hard, but in this cycle it’s doable.”