It’s Time to Unskew the 2014 Election Polls

Early polling on Senate and House races may be underestimating Republican support.

Voters wait in line to pick up their ballots inside the Hamilton County Board of Elections after it opened for early voting, Tuesday, Oct. 2, 2012, in Cincinnati. Ohioans can cast an early ballot by mail or in person beginning Tuesday for the Nov. 6 election. 
National Journal
Steven Shepard
Jan. 29, 2014, 3:04 p.m.

Dur­ing the last pres­id­en­tial elec­tion, Demo­crats mocked Re­pub­lic­an com­ment­at­ors for sug­gest­ing that pub­lic polling show­ing Pres­id­ent Obama lead­ing Mitt Rom­ney was flawed be­cause more Demo­crats than Re­pub­lic­ans were in­ter­viewed. Con­ser­vat­ives ques­tioned the motives of poll­sters and the me­dia out­lets that com­mis­sioned the sur­veys for pub­lish­ing (they said) such ob­vi­ously biased res­ults, and web­sites sprang up that “un­skewed” the num­bers to re­flect what Re­pub­lic­ans thought was closer to real­ity.

Ul­ti­mately, their cri­ti­cism was un­foun­ded. The polls cor­rectly pre­dicted Obama’s vic­tory.

But it turns out that GOP cri­tique may have come two years too early. Demo­crat­ic and Re­pub­lic­an poll­sters alike agree that most of the pub­lic sur­veys on the big 2014 con­gres­sion­al races are un­der­es­tim­at­ing the level of Re­pub­lic­an sup­port in a midterm elec­tion year, which tends to be more con­ser­vat­ive than the rest of the vot­ing-age pop­u­la­tion.

That’s be­cause most pub­lic polls con­duc­ted for me­dia out­lets or by aca­dem­ics are sur­vey­ing the en­tire uni­verse of re­gistered voters, with little re­gard for wheth­er those voters will ac­tu­ally cast a bal­lot on Elec­tion Day. And in the past few midterm elec­tions, Demo­crat­ic-lean­ing voters haven’t turned out at close to the same rates as those who typ­ic­ally back GOP can­did­ates.

Cam­paign poll­sters, on the oth­er hand, con­tact only those who are likely to vote.

“We know a lot about the dif­fer­ence between” the likely midterm elect­or­ate and the over­all pool of voters, said Demo­crat­ic poll­ster John An­za­lone, whose firm works for scores of Demo­crat­ic can­did­ates run­ning this year. “It’s gonna be older, it’s gonna be whiter, it’s gonna be more Re­pub­lic­an.”

Re­pub­lic­an poll­ster Glen Bol­ger offered vir­tu­ally the same as­sess­ment. “Likely voters tend to be a little older, a little bit more Re­pub­lic­an, a little more white,” he said. “And that’s the nature of the elect­or­ate, par­tic­u­larly in non­pres­id­en­tial elec­tions. Re­gistered voters are a little more likely to match a pres­id­en­tial-year look.”

Exit polls aren’t in­fal­lible meas­ures of the com­pos­i­tion of the elect­or­ate, but they are in­struct­ive in show­ing the dif­fer­ence between a pres­id­en­tial- and midterm-level turnout. In 2006, 79 per­cent of voters were white. That dropped to 74 per­cent in 2008, but jumped back up to 77 per­cent in the 2010 midterms. In 2012, just 72 per­cent of voters were white, a re­cord low.

Young­er voters, in par­tic­u­lar, drop off in midterm years. Voters un­der 30 made up 12 per­cent of the 2006 and 2010 elect­or­ates, com­pared with 18 and 19 per­cent of the elect­or­ate in the last two pres­id­en­tial elec­tions, re­spect­ively.

But pub­lic poll­sters ar­gue that their sur­veys at this stage of the cam­paign aren’t meant to be pre­dict­ive. They are a snap­shot of where the elect­or­ate stands now, and just be­cause the midterm elect­or­ate has his­tor­ic­ally been older and whiter than in pres­id­en­tial years doesn’t mean it’s cor­rect to as­sume it will hap­pen again this year.

“His­tor­ic­ally, the elect­or­ate is more Re­pub­lic­an in off-year elec­tions, it’s whiter, it’s older,” said Doug Schwartz, dir­ect­or of the Quin­nipi­ac Uni­versity Polling In­sti­tute in Con­necti­c­ut. “First of all, you don’t know for sure that those his­tor­ic­al pat­terns are go­ing to hold up, and even if they do hold up, you don’t know how much more Re­pub­lic­an, how older, how much more white” the elect­or­ate will be.

Schwartz says Quin­nipi­ac, which will be sur­vey­ing the com­pet­it­ive gubernat­ori­al races in Con­necti­c­ut, Flor­ida, and Pennsylvania, among oth­ers, won’t start screen­ing for likely voters un­til after the sum­mer.

“The main reas­on is that voters are not really tun­ing in closely to the cam­paigns un­til after Labor Day,” he said. “In a lot of races, you don’t have the two can­did­ates set yet…. You don’t know who’s go­ing to be a likely voter un­til that time. It’s too early to as­sess who’s go­ing to be likely to vote.”

While that’s true, it also means that plenty of voters who won’t cast bal­lots are be­ing in­cluded in these sur­veys — and that Re­pub­lic­an can­did­ates might be in stronger po­s­i­tions than pub­lic polls in­dic­ate. In Oc­to­ber 2010, the re­spec­ted NBC/Wall Street Journ­al poll showed more re­gistered voters favored a Con­gress con­trolled by Demo­crats, by a 2-point mar­gin. In Novem­ber, Re­pub­lic­ans romped to a his­tor­ic midterm land­slide that the sur­vey didn’t an­ti­cip­ate. Like­wise, the same sur­vey this month showed Demo­crats with the same 2-point edge.

Demo­crats aren’t auto­mat­ic­ally doomed be­cause polls of re­gistered voters overrep­res­ent voters who won’t turn out on Elec­tion Day. The demo­graph­ics of the 2010 elect­or­ate looked just like 2006, but the res­ults were vastly dif­fer­ent. Some Demo­crats also point to the Obama cam­paign’s soph­ist­ic­ated turnout op­er­a­tion, and the chances that ef­fort could com­pensate — at least in part — for the drop-off that usu­ally oc­curs in midterms.

So how can in­ter­ested ob­serv­ers ac­count for this dis­crep­ancy? For one, polit­ic­al pro­fes­sion­als pay less at­ten­tion to the pub­lic sur­veys than most would be­lieve, es­pe­cially judging by the volume of press re­leases gen­er­ated by the can­did­ate lead­ing in such a sur­vey.

“I nev­er get a false sense of hope from some pub­lic poll,” said An­za­lone, the Demo­crat­ic poll­ster. “As a pro­fes­sion­al, the only thing I look at is Pew,” re­fer­ring to the Pew Re­search Cen­ter’s na­tion­al polling on the gen­er­ic House bal­lot and oth­er is­sues.

And us­ing a crude in­stru­ment to add points to the GOP can­did­ate’s vote share to re­flect an ar­bit­rary turnout tar­get — see 2012’s “Un­skewed Polls” move­ment — isn’t the an­swer, either.

Still, if re­cent his­tory holds, when pub­lic poll­sters start screen­ing for likely voters, those track­ing these races should ex­pect a shift to­ward Re­pub­lic­ans. Un­til then, pub­lic sur­veys are best viewed as snap­shots of the over­all elect­or­ate, with­in which par­tis­ans on each side can de­term­ine at what turnout levels their can­did­ates might pre­vail.

Un­skew away.

What We're Following See More »
TAKING A LONG VIEW TO SOUTHERN STATES
In Dropout Speech, Santorum Endorses Rubio
2 days ago
THE DETAILS

As expected after earlier reports on Wednesday, Rick Santorum ended his presidential bid. But less expected: he threw his support to Marco Rubio. After noting he spoke with Rubio the day before for an hour, he said, “Someone who has a real understanding of the threat of ISIS, real understanding of the threat of fundamentalist Islam, and has experience, one of the things I wanted was someone who has experience in this area, and that’s why we decided to support Marco Rubio.” It doesn’t figure to help Rubio much in New Hampshire, but the Santorum nod could pay dividends down the road in southern states.

Source:
‘PITTING PEOPLE AGAINST EACH OTHER’
Rubio, Trump Question Obama’s Mosque Visit
2 days ago
WHY WE CARE

President Obama’s decision to visit a mosque in Baltimore today was never going to be completely uncontroversial. And Donald Trump and Marco Rubio proved it. “Maybe he feels comfortable there,” Trump told interviewer Greta van Susteren on Fox News. “There are a lot of places he can go, and he chose a mosque.” And in New Hampshire, Rubio said of Obama, “Always pitting people against each other. Always. Look at today – he gave a speech at a mosque. Oh, you know, basically implying that America is discriminating against Muslims.”

Source:
THE TIME IS NOW, TED
Cruz Must Max Out on Evangelical Support through Early March
2 days ago
WHY WE CARE

For Ted Cruz, a strong showing in New Hampshire would be nice, but not necessary. That’s because evangelical voters only make up 21% of the Granite State’s population. “But from the February 20 South Carolina primary through March 15, there are nine states (South Carolina, Alabama, Arkansas, Georgia, Oklahoma, Tennessee, Kentucky, Mississippi, and North Carolina) with an estimated white-Evangelical percentage of the GOP electorate over 60 percent, and another four (Texas, Kansas, Louisiana, and Missouri) that come in over 50 percent.” But after that, he better be in the catbird’s seat, because only four smaller states remain with evangelical voter majorities.

Source:
CHRISTIE, BUSH TRYING TO TAKE HIM DOWN
Rubio Now Winning the ‘Endorsement Primary’
2 days ago
WHY WE CARE

Since his strong third-place finish in Iowa, Marco Rubio has won endorsement by two sitting senators and two congressmen, putting him in the lead for the first time of FiveThirtyEight‘s Endorsement Tracker. “Some politicians had put early support behind Jeb Bush — he had led [their] list since August — but since January the only new endorsement he has received was from former presidential candidate Sen. Lindsey Graham.” Meanwhile, the New York Times reports that fueled by resentment, “members of the Bush and Christie campaigns have communicated about their mutual desire to halt … Rubio’s rise in the polls.”

Source:
ARE YOU THE GATEKEEPER?
Sanders: Obama Is a Progressive
2 days ago
THE LATEST

“Do I think President Obama is a progressive? Yeah, I do,” said Bernie Sanders, in response to a direct question in tonight’s debate. “I think they’ve done a great job.” But Hillary Clinton wasn’t content to sit out the latest chapter in the great debate over the definition of progressivism. “In your definition, with you being the gatekeeper of progressivism, I don’t think anyone else fits that definition,” she told Sanders.

×