Understanding the difference
PRINCETON, NJ -- We have received a number of questions about the difference between registered voters and likely voters when Gallup reports presidential election trial-heat results. Here's an explanation.
Not everyone votes in presidential elections. There are several ways to calculate voter turnout figures for an election, but in 2004, for example, experts estimate that only about 55% to 60% of the eligible voting population actually voted. (That's actually higher than was true in the 2000 election.)
When we interview a typical Gallup Poll random sample, we are estimating the responses of all adults 18 years of age and older. We call these national adults. In our recent poll conducted the weekend of Sept. 5-7, for example, John McCain led Barack Obama among these national adults by a 48% to 46% margin.
But we almost never report this figure. It's unrealistic to do so because we know that a percentage of these national adults not only won't vote, but can't vote -- because they are not U.S. citizens or are not registered to vote in their local areas.
So we narrow down the national adult sample to registered voters. This is the group who in response to a standard poll question say they are "registered to vote in their precinct or election district." This is the group whose data we report most often, because it represents an estimate of Americans who in theory are eligible to vote and could vote if they want to. In the Sept. 5-7 poll, this group divided 50% for McCain and 46% for Obama.
Of course, we know that in the final analysis, not all of these registered voters will actually end up voting. So Gallup has over the years created systems to isolate likely voters, that group of individuals who we can estimate are most likely to actually turn out and vote.
There are a variety of ways to estimate likely voters. Different pollsters and polling groups have different approaches. Gallup has spent decades developing our system, which we have found in election after election helps improve our accuracy in terms of how our final poll before an election compares to the actual vote percentages on Election Day itself.
Gallup's system consists of asking respondents a battery of questions about past voting, current interest in the election, and self-reported interest in voting. These include such questions as "How much thought have you given to the upcoming election for president?", "Do you happen to know where people who live in your neighborhood go to vote?", "Have you ever voted in your precinct or election district?", "Do you yourself plan to vote in the presidential election this November?", and "Rate your chances of voting in November's election for president on a scale of 1 to 10."
Putting all of this information together, we can assign each voter a score based on our estimate of his or her probability of actually voting. Based on assumptions about actual turnout, we use the scores to select the pool of voters that we think best represents a realistic pool of likely voters come Election Day. In the Sept. 5-7 poll, this group of likely voters went 54% for McCain, 44% for Obama.
Comparing across national adults, registered voters, and likely voters, one can see that at this point, shortly after the Republican National Convention, the more we winnow the sample down to voters with the highest likelihood of voting, the better McCain does. This is not unusual. The Republican candidate often benefits from a turnout advantage.
Here's an example. Gallup's final poll before the 2004 election showed the following:
George W. Bush 46%
John Kerry 48%
George W. Bush 49%
John Kerry 47%
Kerry was ahead among registered voters by 2 points, while Bush was ahead among likely voters by 2 points.
The final election result? Bush won the popular vote over Kerry by about 2.5 percentage points, almost exactly what our likely voter estimate predicted. Had we reported only registered voters, we would have estimated a Kerry victory. In other words, had all registered voters turned out in 2004, Kerry would have been elected president. But all registered voters didn't turn out. There was a Republican advantage among those who did turn out. And Bush won.
Two final notes.
Some observers have argued that the Obama campaign will successfully increase turnout among groups that typically are less likely to vote, including in particular young voters. We are keenly aware of these hypotheses and are continually and carefully analyzing our data to make sure we pick up any unique or unusual surge of turnout potential among certain subgroups of the population.
Second, we are at this point reporting likely voter estimates on only an occasional basis. We feel that the trends among registered voters give us the best way to track election preferences in our daily poll, in part because many voters are not yet in a position to accurately estimate their chances of voting on Election Day. But from time to time, we do estimate (and report) likely voter results to give us a feel for the potential difference turnout could make in November. So far this summer, there have been occasions when -- as was the case this past weekend after the GOP convention -- likely voters were decidedly more Republican. But there have also been occasions when there was little difference between the vote patterns of likely voters and those of registered voters. We will continue to monitor these patterns as Election Day draws closer.
To provide feedback or suggestions about how to improve Gallup.com, please e-mail firstname.lastname@example.org.