Pretty soon, the country's top pollsters will make a subtle change that even some political junkies won't process: They will shift from reporting results of registered voters to only those most likely to vote in the 2014 election -- a.k.a. "likely voters."
For those who follow polling closely the distinction between the two is key to understanding the true state of play in a race. It's also likely to cause an apparent shift -- almost certainly in the GOP's favor -- that some will misinterpret as newfound momentum.
So what's the deal with the switch? And when does it happen? Below, we explain it all.
Why is the difference between registered and likely voters important?
Quite simply, not all registered voters are likely voters. The behavior and attitudes of the two groups can be slightly different -- a factor that can be especially important in midterm elections in which fewer than 50 percent of eligible voters cast ballots. In order to provide accurate estimates, polls must report opinions of the people who are most likely to participate in the election. That requires expertise and judgment in sorting out poll respondents who are more or less likely to vote.
The 2010 midterm election is a good example of why the distinction makes a difference. The final Washington Post-ABC News poll before the 2010 election found Democrats leading on the generic ballot by five points among registered voters but Republicans leading by four points among likely voters. The latter estimate proved far more accurate, with Republicans receiving 51 percent of the vote across all House contests to Democrats' 44 percent.
Why do results among likely voters look better for Republicans?
This isn’t always true. The likely voter population is different in midterm versus presidential elections. Midterm elections, in which there is no national candidate and less of a focus on a broad, national campaign attracts different types of voters. Specifically, non-white and younger voters have been less apt to turn out for midterm elections. These are two key Democratic constituencies that turned out reliably in 2008 and 2012 and gave a big lift to President Obama. These two groups were less reliable for Democratic candidates in the 2010 election.
In addition, early polls of the 2014 election show there is significantly more enthusiasm on the Republican side this year, which suggests likely voter models will be more favorable to the GOP.
Why are likely voters special?
The population of likely voters isn't known before Election Day, while the population of registered voters and voting-age adults are more or less fixed groups (you’re either registered to vote or you’re not). Because polls use different techniques to classify likely voters, comparing results among likely voters from two polls entails an extra layer of complication and unpredictability.
How do pollsters identify likely voters?
Pollsters use various strategies to identify likely voters, but three techniques dominate:
- Screening questions: Pollsters ask respondents screening questions about past voting and/or vote intention and set a threshold of who is considered a likely voter (this is the Post’s standard method). The Post begins by excluding those who are not registered to vote and uses questions to predict voting such as past voting behavior, self-reported vote intention, interest in the election and knowledge of one’s voting place to further thin the herd. A related method uses similar questions to assign each respondent a likely voter score based on how many questions they answer “correctly” (e.g. Did you vote in the most recent presidential election? Yes = 1 point). Gallup and the Pew Research Center employ methods like this.
- Voter list sampling: Campaign pollsters typically use voter lists to recruit respondents, targeting those who have voted in a particular set of recent elections according to voter registration files. This method is often combined with the screening questions above to establish voter likelihood.
- Probability-weighted models: Less commonly used is a “probability voting” model for identifying likely voters. Probability voting models assign probability of voting on a 0 to 1 scale for each respondent and weight results by this. In other words, if a respondent is considered 66 percent likely to vote and another 33 percent, the first respondents’ vote choice will receive twice the weight in data. CBS News, the New York Times and Marist College polls have employed such models in the past.
For more details on varying likely voter models, check out Mark Blumenthal's excellent explainers from 2004.
If they’re so special, why don’t pollsters always report poll results among likely voters?
The biggest reason pollsters report registered voters instead of likely voters is uncertainty. The further you are from an election, the more likely it is that the political dynamics have not gelled and voters themselves have not figured out whether they will cast ballots. Correctly identifying those who are likely to vote is more error-prone many months from the election, so early results among likely voters should be taken with a grain of salt.
Identifying the correct blend of likely voters out of a sample of poll respondents can improve a vote estimate. But a good vote estimate is just one (relatively small) reason for polling the entire public.
We conduct polls to report broadly on the public’s opinions on politics, policy and values that motivate the country, and non-voters have meaningful views on these issues too. Many people don't vote because they are disillusioned with the political process or too busy in their daily lives, but polls of the general population allow for better understanding of their views and how they get represented (or not represented) by officeholders.
So, when should pollsters switch from registered to likely voters?
The Post doesn't have a fixed schedule or specific date for when we move from reporting registered voters to likely voters for the congressional or presidential preference questions. We might report vote estimates among both registered and likely voters as early as the mid-summer before a fall election. But those early likely voters models rely on a simple self-report of certainty of voting. Our April poll, for instance, broke out results among respondents who said they were “absolutely certain” to vote, which is typically one component in our likely voter model.
By September, we begin to more consistently provide a likely voter estimate with a more detailed battery of measures to gauge vote likelihood. By mid- to late-October, we generally move to reporting the vote estimate only among likely voters. We produce and examine a wide range of likely voter models to look for consistent and coherent demographic profiles of potential voters.
Why not report results from both registered and likely voters all the time?
Building a strong likely-voter model requires asking a series of questions, which can be wasteful far away from an election when voters have yet to engage with the campaign. Earlier polls focus on major issues being debated on the campaign trail, approval of the president, views of the parties and interest in the election to tell a fuller story of the campaign than a horse race question. By contrast, polls often interview only likely voters close to an election because it allows for increased sample size (and greater precision) among this target group.
Some pollsters like to try and pin down likely voters far earlier in the process than others (particularly Republican pollsters), but it's a tricky thing to do and has a much higher potential for getting things wrong.
Scott Clement and Peyton Craighill run the Post's polling unit.