Ron Paul of Texas says he’s battling a false impression that he can’t triumph in a general election.
John King of CNN asked the candidates at the Feb. 22, 2012, Republican presidential debate to help "the voters who still have questions about you" and explain the "biggest misconception about you" in the public debate.
"I would say the perpetuation of the myth by the media that I can't win" while some statistics show the contrary, said Paul, a Lake Jackson congressman. "Just recently, there was a poll in Iowa, and it matched all the four of us up against (President Barack) Obama. And guess what? I did the very best."
Nationally, at the end of February 2012, Obama was outpacing each of the Republican hopefuls in one-on-one matchup polls, according to aggregations of results by RealClearPolitics.com. The "spread" was smallest for former Massachusetts Gov. Mitt Romney (4-point lead for Obama), followed by former U.S. Sen. Rick Santorum (5-point lead for Obama) and Paul (6.6-point lead for Obama). Former House Speaker Newt Gingrich was the furthest behind Obama, down 12.4 points.
But did Paul accurately recap an Iowa poll?
Selzer & Co., a West Des Moines firm, conducted the telephone survey of 800 Iowans Feb. 12-15, 2012. As part of the poll, 611 likely voters were asked to indicate who they would vote for in matchups between Obama and each of the four major Republican candidates. The questions were phrased this way: "If the election for president were held today, and the candidates were Barack Obama for the Democrats and (name of GOP candidate) for the Republicans, for whom would you vote?"
The results show all the candidates except Gingrich beating Obama in Iowa, with Paul winning by the greatest margin, as noted in the Register’s articles.
** Paul 49 percent, Obama 42 percent (7-point difference)
** Santorum 48, Obama 44 (4-point difference)
** Romney 46, Obama 44 (2-point difference)
** Gingrich 37, Obama 51 (14-point difference)
Not entirely. Interviews of pollsters led us to realize that there’s at least debate over the significance of the poll’s margin of error, which for the matchup questions in the Iowa Poll was plus or minus four percentage points.
Margins of error hint at every poll’s built-in weakness: It’s impossible to survey every voter.
In a write-up about polling, Selzer & Co. President J. Ann Selzer defines the margin of error as the estimated difference between a percentage obtained from a survey of a sample of voters — for example, the 49 percent in the Iowa Poll who said they would vote for Paul in a contest with Obama — and the percentage the firm theoretically would find if it interviewed every likely voter in the state.
She explains further: Consider, she says, a report from her firm that a candidate has captured 60 percent of the vote in a survey with a margin of error of plus or minus four points. That means, she says, that if the survey were repeated in the same way at the same time with different samples of voters, 95 percent of the time, the percentage of support for the candidate would fall between 56 percent and 64 percent (a range of four more and four less than the reported percentage).
Applying the margin of error in the Iowa Poll’s matchup questions demonstrates how much the actual results could swing.
For Paul vs. Obama, the possibilities range from a 15-point Paul lead — Paul 53 (reported result plus four), Obama 38 (result minus four) — to a 1-point Obama lead — Paul 45 (result minus four), Obama 46 (result plus four).
For Santorum vs. Obama, the matchup range goes from a 12-point Santorum lead (Santorum 52, Obama 40) to a 4-point Obama lead (Santorum 44, Obama 48).
For Romney vs. Obama, the range is from a 10-point Romney lead to a 6-point Obama lead.
And for Gingrich, the possibilities range from a 6-point Obama lead to a 22-point Obama lead.
We asked five polling experts about Paul’s characterization of the Iowa results. There was no argument from anyone that Paul had accurately recapped the poll’s reported results.
Still, Charlie Leonard, a visiting professor at Southern Illinois University who conducts polls for the school’s Paul Simon Public Policy Institute, told us that in his view, there’s no statistical difference between Paul and Santorum’s performance against Obama.
However, Selzer, head of the firm that conducted the poll, aligned with Paul’s interpretation of the results. She said the best estimate of Paul’s support — and that of the other Republicans and Obama — is the percentage found during the survey and reported in the poll. She acknowledged that Paul’s percentage of support, for example, could be one, two, three or four points lower — or higher — than 49 in a matchup with Obama. But, she said, technically speaking, that’s less likely to be the case, because each point you move away from the reported number has a lower probability of being true.
Jim Henson, director of the Texas Politics Project at the University of Texas and co-director of joint polling projects for the university and the Texas Tribune, described the difference between the Paul-Obama and Santorum-Obama matchup as marginal and said Paul’s statement "ignores the underlying statistics of those numbers" while being literally correct. Henson’s take on the margin of error: "What it tells you is that the most probable outcome is the two numbers (reported in the poll), and the farther you go off those numbers, the less likely they are to be true — but they might be."
Paul had the largest lead on Obama of any of the major Republicans in matchup polling in Iowa in February 2012. Still, experts raised questions about how statistically certain that conclusion is, considering how close the poll’s results were. To us, that is additional information needed to evaluate Paul’s claim, and so, his statement rates Mostly True.