
A Closer Look at Those Wisconsin Exit Polls
Published June 7, 2012 • Read time: 6m
Join the Conversation
0The Democrats emerged from Tuesday's recall election with two consolation prizes. The first is that they flipped a seat in the Wisconsin state Senate, taking control of that body (which won’t convene again until after the 2012 elections).
The second, and most frequently cited, is that exit polls found President Obama ahead of Mitt Romney by seven points in Wisconsin. As Markos Moulitsas of DailyKos summarized, “[a] seven-point lead, Obama over 50 percent, despite lacking participation of one of Obama's biggest constituencies (young voters), does not suggest a particularly close race this November. If Republicans want to dump their millions (and they'll have plenty of those) here, they're free to do so.”
But was this finding of the exit poll correct? We can’t know for certain, but there are four reasons to at least raise an eyebrow. The first is the wording of the question, which asked whether respondents planned to vote for Obama, Romney, or not vote. Missing from the choices? “Don’t know.”
In other words, what we may be seeing -- and what may explain some of the discrepancies we see below -- is that voters who haven’t given a lot of thought to the 2012 election are presently leaning Obama’s direction if forced to make a choice, but would respond to a typical horse-race poll that they are “undecided.”
Second, the independent polling shows a closer race. If we look at the three pollsters who surveyed both the general election and the recall election in the closing days of the latter (Marquette, WPR/St. Norbert, and DailyKos/PPP), we find a 5.6-point Scott Walker lead -- about two points worse than the actual result.
The FBI Gained Access; We'd Already Lost Our Privacy
Is Trump Right About NATO?
Arizona's Voting Rights Fire Bell
U.S. Officials Say 2 Americans Died in Brussels Attack
For the general election, however, these three pollsters found Obama with a five-point lead, 49 percent to 44 percent. When you consider that the recall polls underestimated Walker’s advantage by a couple of points, you would expect a three-point Obama lead to have emerged in the recall electorate.
It’s tough to reconcile this: What type of voters could these pollsters have missed to understate both Walker’s and the president’s margins? I suppose there is some distinct demographic group, but it seems at least as likely that something is screwy with the exit poll.
Third, consider what the exit polls are suggesting: that Obama leads by seven points in an R+1 electorate. To achieve this, he would have to either be doing disproportionately well among Republicans compared to how Romney performs among Democrats, or very, very well among independents.
This is inconsistent with what we’ve seen in just about every poll of the race between Romney and the president, who currently sports a 41 percent approval rating among independents. This isn’t to say it’s impossible that he’d be leading in an electorate with more Republicans than Democrats, but it is something of a red flag.
None of these would really be much cause for concern, however, without the larger context of the exit poll itself. Remember, the initial data that came in showed Walker and Tom Barrett tied, which was off by a full seven points.
In other words, the exit pollsters in the field missed a lot of Walker voters. Now, exit pollsters have ways to fix this. For one thing, they weight different regions of the state to the actual vote returns. For example, if northeast Wisconsin exit polls are showing a 50-50 race, and the actual results are 60-40 for Walker, they will simply assign greater weight to a Walker respondent in the region, bringing the reported result in line with the actual result.
In addition, poll takers on the ground keep track of various features of people who decline to respond, such as their gender, race, and approximate age. This allows these demographics to be weighted properly as well, and helps to ensure that the weights are representative of the demographic traits of non-respondents. In other words, if the non-respondents are disproportionately male, white, and older, the exit pollsters will make sure that the final weights account for those discrepencies.
But what about the other questions where exit pollsters cannot simply make visual assessments of non-respodnents? For example, “For whom would you vote in the presidential election if it were held today?” We have no idea if the non-respondents are somehow different than actual respondents of a similar gender, race, or age.
To understand why this might be a problem, consider the following (very exaggerated) hypothetical. Assume that the initial exit polls find that the race is tied, and that Obama trails Romney by 30 points among Walker voters.
As the results come in, it is apparent that Walker leads by 10 points. So these Walker voters are weighted more heavily, and the president’s lead over Romney shrinks.
But what if the non-respondents were also disproportionately the most conservative voters, who don’t trust media folk? In other words, what if the president trails by 50 points within this category? Weighting these non-respondents the same as actual respondents on the presidential horse-race question will actually overstate the president’s standing -- it won't shrink enough.
This is the problem with the “down-ticket” questions, especially in elections like this one where the top lines had to be heavily weighted. This is part of why exit pollsters only asked basic demographic and top-line questions until fairly recently. And, in fact, Warren Mitofsky found that conservative voters were, in fact, less likely to respond to exit pollsters than liberal or moderate voters, suggesting that our hypothetical isn’t purely hypothetical.
All of this is further compounded when you understand that absentee voters weren’t polled, and this group skewed toward Walker. We don’t know much about this group’s preference, and if its down-ticket preferences were different from actual “live” voters, they will be weighted improperly.
This is ultimately similar to the problem that pollsters faced with "cell-phone only" households. For a while, there wasn't thought to be a problem, as these were a small portion of the electorate. Moreover, they were young, and so the problem could be corrected by simply more heavily weighting young respondents who answered landline polling questions.
But today cellphone-only households are a larger section of the electorate, and we know that they are more heavily Democratic than young voters who have land lines. In other words, weighting landline respondents won't solve the problem. You have to find a way to reach out and account for cellphone-only voters, or your poll risks becoming skewed.
Of course, none of this is definitive proof, and I don’t want to overstate my case here. The exit poll might well be correct. If the president is doing this well against Romney in a Republican-leaning electorate, the Republicans really are pretty well doomed there in the fall. But there are real reasons to question whether he really is doing that well in Wisconsin right now.

Democrats Won the Midterms. Common Sense Won, Too
Jordanne Kemper • RealClearPolitics
The 120th Congress Shapes Up
Carolina Lumetta • RealClearPolitics
Flooding GOP Primaries With Democrats
Robert Natelson • RealClearPolitics
Washington Keeps Rewarding Oracle—Veterans Keep Paying the Price
Darin Gaub • RealClearPolitics
The Seeds of Unrestricted Warfare
John Teichert • RealClearPolitics
The Primary System Is Losing the Next Generation
Sean Trende • RealClearPolitics
RCP Today
Top stories on RealClearPolitics
The Takeaway
Polling intelligence and analysis delivered to your inbox
World News
Featuring the latest world news
Policy
Featuring the latest policy news
Support Our Mission
As part of RealClearPolitics, authors aim to provide balanced, non-partisan analysis by presenting top reporting and commentary from all sides of the political spectrum.
Trusted by policymakers and citizens alike since 2000
