Exit Polls: The Media's Dirty Little Secret
One of the most mystifying aspects of the coverage of the Wisconsin recall election has been the media's ongoing use of exit poll results in stories suggesting that -- despite Gov. Scott Walker's big win against the efforts of Democrats and Labor Unions to end his term early -- President Obama has a big lead over Mitt Romney in the crucial swing state.
The continued faith in the flawed Wisconsin survey is even more amazing when you consider the dreadful record exit polls have of matching up with the actual vote totals. In nearly every case of error, exit polls have oversampled Democrats, a fact almost never pointed out by the nation's news organizations.
As voting stations closed in Wisconsin on Tuesday, anchors and reporters at CNN, MSNBC and even Fox News cautioned, based on exit polls, that America was in for a “long night” of awaiting election results as exit polls showed the race was “too close to call.”
Then, suddenly, just 49 minutes after the polls closed, Fox, CNN and the rest of the networks all called the race for Walker over Democrat Tom Barrett.
The actual vote totals coming in showed the exit polls had been wrong. A race that the media's expensive exit polling apparatus had insisted was too close to call was not close at all.
And yet, reporters continued to report that, while Wisconsin voters had voted for Walker, they still preferred Obama over Mitt Romney in the presidential race -- based on the same exit poll results that were being shown to be worthless by the actual vote totals scrolling across the bottoms of their screens.
The disconnect was noticeable, as reporters acted as if the failure of the exit poll to accurately predict the Walker-Barrett numbers had no bearing on whether the same exit poll, of the same voters, would accurately reflect the Obama-Romney race in Wisconsin.
Even after the election, many media outlets are using the exit poll numbers on Obama-Romney as if they have credibility, without even mentioning the inaccuracy of the same poll in the recall race.
Wednesday's Baltimore Sun editorial is a good example: "The vote was widely billed as a preliminary skirmish in the November election between Mr. Obama and Republican Mitt Romney, but there is good reason to conclude that its predictive value is low," the Baltimore Sun said. "For starters, exit polling showed that even as voters handed Mr. Walker a convincing 53 percent to 46 percent victory over Milwaukee's Democratic mayor, Tom Barrett, a solid majority of them would have voted for Mr. Obama over Mr. Romney."
Do the Baltimore Sun's editors think about what they write before they publish it? They assert that exit polling on Tuesday showed “a solid majority” of Wisconsin voters would have voted for Obama over Romney, in an election that is five months away, while ignoring the fact that the very same exit polling couldn't even accurately predict the results of the election that was happening that day.
If the exit polls were off by 7 points in the Walker-Barrett race, it seems reasonable that they were off by about 7 points in the Obama-Romney race, too.
So, why did the media not acknowledge that the recall election results had shown their exit polls to be worthless, rather than forge ahead with stories about the exit polls showing Obama leading Romney?
The easy answer is, in the middle of a night of bad news for the Left, liberal journalists hung onto those exit poll results like a life-raft, desperate to find some small glimpse of hope for Obama amid the wreckage of the recall election. Many delusional Democrats did the exact same thing after a skewed survey by the liberal Public Policy Polling outfit released the day before the recall vote claimed that Walker had just a 3-point lead over Barrett, within the poll's margin of error.
During his radio show Wednesday, Rush Limbaugh made the case that CNN host Wolf Blitzer did the same thing once the voting ended, joyously hyping the incorrect exit polls and then dejectedly announcing the actual results shortly thereafter.
Liberal bias has some role in why exit polls continue not to be given the skepticism they deserve but there are other likely reasons that feed into it. Money is also a factor. The nation's largest media organizations have a lot invested in their exit polling machinery. It's so expensive, in fact, that they've all actually teamed up to split the costs through a service called the National Election Pool (NEP). ABC, Associated Press, CBS, CNN, Fox News, and NBC are the only members.
To toss away their own exit poll results simply because one part of them had been shown to be less-than-accurate would be to waste money – and leave the networks with little to talk about while waiting for the actual votes to be counted. If such data were truly reliable, it would certainly be very interesting to discuss.
But the fact is that exit polling has a long record of failure and the more experienced journalists out there are well aware of this.
As liberal blogger Nate Silver has noted, exit polls by the Voter News Service, predecessor to the NEP, incorrectly predicted that Bill Clinton would win Indiana and Texas in the 1992 presidential election. Later in 2000, VNS started the Bush-Gore recount controversy by calling the state of Florida for Al Gore before the polls in the Republican western portion of the state had even closed. According to a report commissioned by CNN following the 2000 debacle, VNS exit polls incorrectly predicted the winner in eight separate states. Two separate studies indicated that Republican turnout was lower by between 3 and 4 percent thanks to GOP voters becoming discouraged that their candidate was destined to lose.
In 2002, VNS dropped the ball again by failing to provide any exit polling data on Election Night at all. That led to its eventual disbandment and the creation of the National Election Pool.
NEP has had its own record of failure as well. In 2004, bad NEP data led the media and many Democrats to erroneously believe that John Kerry was going to defeat George W. Bush in his attempt to be reelected. Bad data collection led NEP to incorrectly overestimate the John Kerry vote in 26 states. By comparison, the Bush vote was overstated in just 4 states.
Believing the bad NEP exit polls, Kerry campaign manager and perennial loser Bob Shrum even went so far as to congratulate the man he thought was a shoe-in as the next president. Shrum was far from the only Democrat led astray by NEP's false data, due to widespread uncritical reporting of the imaginary Kerry lead, many grassroots liberals on sites like Daily Kos or Democratic Underground began building up elaborate conspiracy theories about how "Bushco" had "stolen" yet another election from them.
And now we have Wisconsin.
The common denominator in each of these failures was that exit polling data is usually skewed leftward. That was the same conclusion reached by NEP itself in a secret memo to members about the 2004 failure:
“Our investigation of the differences between the exit poll estimates and the actual vote count point to one primary reason: in a number of precincts a higher than average Within Precinct Error most likely due to Kerry voters participating in the exit polls at a higher rate than Bush voters. There has been partisan overstatements in previous elections, more often overstating the Democrat, but occasionally overstating the Republican.”
More traditional polling about exit poll participation backs this conclusion up. As pollster Scott Rasmussen stated in 2008, Democrats are simply more eager to respond to exit pollsters than Republicans or independents:
The bottom line is that in every state we polled—Colorado, Florida, Missouri, North Carolina, Ohio and Virginia— Democrats are a lot more eager to take exit polls than Republicans.
In five of the six states, a majority of Democrats say they would be Very Likely to participate in the exit polling process. At the same time, in five of the six states, fewer than 40% of Republicans would be willing to do the same.
In every state, Republicans are at least twice as likely as Democrats to say that they are not at all willing to take an exit poll.
Unaffiliated voters tend to align more closely to Republicans in all six states in both willingness and unwillingness to participate in exit polls.
Exit poll acuracy has also been negatively impacted by the increase in mail-in voting and early voting. In the case of postal voting, surveying voters in person after they’ve cast their ballots is impossible. In the case of early voting, it is impossibly expensive to have staffers present at polling locations for days or weeks on end.
Since exit polls have to be done on Election Day, the increase in early voting also presents a disadvantage in exit polling to Republican candidates because historically (with the exception of 2008), Republicans have been more likely to take advantage of it. With fewer Republicans showing up on Election Day, the natural Democratic skew of an exit poll is likely to be even more pronounced.
Given this record of unreliability, exit polls should always be regarded with caution. TV networks that use them should frequently remind viewers that such data may not be reliable and occasionally mention their Democratic tilt. Unfortunately, such skepticism is almost entirely absent from television. In reality, it appears it will be up to conservatives to spread the word about them so that their use will be discontinued.
My hope is that as awareness of exit polls’ inaccuracy grows, we’ll see more comments like those made by CNN’s Roland Martin who tweeted at 9:21 p.m. on Tuesday a caution against paying too much attention to the exit poll results on the Obama-Romney race:
"Why are Obama supporters touting exit polls saying Obama preferred over Romney 53-42? Same exits had this race 50-50. IGNORE THOSE POLLS!" he tweeted.
Indeed. You can bet that if exit polls oversampled Republicans, this situation would not have been allowed to stand. But that's not the reality we live in.
Fortunately, it is possible to change public perception in the same way that conservatives have done with the broader topic of liberal media bias. Before groups like the Media Research Center and Accuracy in Media got started, most Americans were unaware of the pervasive liberal bias at places like CBS or the New York Times. In 2008, that had changed with 70 percent of Americans correctly believing that most journalists wanted Barack Obama to defeat John McCain.
It's time we raised similar awareness about the pervasive liberal bias and inaccuracy of exit polls.