When Donald Trump insists that he’s winning the presidential election, sometimes he’s correct. He’s simply cherry-picking the three daily tracking polls that have him jockeying with Hillary Clinton for the lead.
While the bulk of other polling shows Clinton building a significant advantage, the three polls reflect a vastly different race — one where Trump is frequently ahead. A joint project between the University of Southern California and the Los Angeles Times reports Trump is in the lead by a single percentage point. Rasmussen Reports, which had Trump leading by 2 points as recently as two weeks ago, now has Clinton 1 point ahead, 44 percent to 43 percent. The polling partnership between Investor’s Business Daily and the polling company Technometrica also finds a neck-and-neck race — the latest results released Wednesday put Clinton in the lead by 0.6 points.
The three polls are more obscure and, at the same time, more prolific than the major national surveys that dominate news coverage of the campaign and are often conducted by the nation’s most prominent news organizations. And their divergent results have put their unique methods under serious scrutiny.
The USC Dornsife/Los Angeles Times poll consists of a weekly average of interviews conducted among a static panel of respondents surveyed throughout the campaign. That panel, which may not be reflective of the overall electorate this year, is weighted to the self-reported 2012 vote choice — a controversial decision that seeks to achieve the right balance between voters who usually cast ballots for Democrats and Republicans.
As of Wednesday morning, Trump was ahead by 1 point in the new model. It has been, by far, the friendliest poll for Trump, even showing him ahead of Clinton in the two weeks following the first debate last month, when Clinton began to pull away in other polls.
Rasmussen, which relies mostly on automated phone calls to landlines, is a more familiar presence: it has had a persistent and often inaccurate Republican lean for most of the decade. The Investor’s Business Daily/TIPP poll is more opaque, using more traditional methods but also peculiar and not-totally transparent weighting procedures, as outlined Tuesday by Washington Post pollster Scott Clement.
Investor’s Business Daily — a weekly, California-based business newspaper — and TIPP conduct live telephone interviews with likely voters on a daily basis, adding up to a six-day, rolling average.
Both Investor’s Business Daily and TIPP claim their survey has been the “most accurate presidential poll in America” — a tenuous boast based on their assessment of combined performance from the past three presidential elections. Like most polls in 2012, they underestimated the incumbent’s margin of victory. Their final poll in 2012 showed Obama leading by 1 point; he won by nearly 4 points. Their results were closer to the mark in 2008 and 2004.
TIPP pollster Raghavan Mayur addressed the Post’s critiques in a phone interview with POLITICO on Wednesday. First, Mayur said, TIPP does not weight their poll to ensure they have the right mix of voters by education level — despite the fact education has been a defining fissure in this race, especially among white voters.
“I have done extensive testing about that quite a while back,” he said. “It never gave me good results. … I also think that people tend to overstate their education. That’s why I feel it’s not a very good measure.”
The Post also raised questions about TIPP’s weighting by party identification — a flexible attitude rather than more fixed demographic classifications. Mayur did not divulge to the Post how he determines the proper party-ID balance, but he said Wednesday it is divined from a number of sources.
“I look at the last six months of our polling and see what we are getting,” Mayur said. “I look at published numbers from people like Gallup and Pew [Research Center]. I kind of look at them, and I have a model that makes a consensus based on all of this.”
The Trump campaign complains that the media coverage is consistently and inaccurately driven by the polls showing Clinton well ahead rather than those portraying a closer race.
“He likes to cite the three polls,” said Kellyanne Conway, Trump’s campaign manager and a GOP campaign pollster Monday on MSNBC. “The most accurate poll in the last elections and the L.A. Times poll, and everybody said we can’t cite those.”
In recent days, Trump — who frequently bragged on the stump and on Twitter about his standing in the polls during the primary season — has cited the three surveys to claim he is poised for an upset victory on Election Day.
On Monday, Trump tweeted a graphic citing the previous Rasmussen Reports results giving him a 2-point lead over Hillary Clinton. A day earlier, Trump linked to a poll from Investor’s Business Daily also showing him ahead by 2 points at that time. And throughout the campaign, Trump has clung to the University of Southern California/L.A. Times study.
“When the polls are even, when they leave them alone and do them properly, I’m leading,” Trump claimed Monday in a campaign event in northern Florida. “But you see these polls, where they’re polling Democrats: ‘How’s Trump doing? Oh, he’s down.’ They’re polling Democrats.”
The final weeks of a presidential campaign used to mean a glut of daily tracking polls from a number of news outlets and other public sources. In 2008, the organizations conducting daily tracking polls at the end of the race included ABC News/Washington Post, NBC News/Wall Street Journal, CBS News/New York Times, Gallup and Diageo/Hotline.
But other than ABC News — which has Clinton ahead by 9 points in its most recent tracking poll — none of those news organizations has launched a regular, daily tracking poll. Some have polled the race once in the past few weeks, and others, like Gallup, are no longer conducting horse-race polls at all. NBC News is working with an online pollster, SurveyMonkey, to conduct a weekly tracking poll, with Clinton leading by 5 points in the most recent results. CNN conducted a national poll last weekend and also found Clinton ahead by 5 points, and a new USA Today/Suffolk University poll out on Wednesday has Clinton ahead by 9 points.
The volume of major national telephone polls, however, has declined precipitously.
“It really comes down to budget,” said Patrick Murray, director of the Monmouth University Polling Institute in New Jersey. “Polling costs have increased dramatically over the past five to 10 years, and they just continue to spiral upwards. There’s been a big cost increase in the past two years alone.”
If the wide variance between the two kinds of poll results holds, that sets up an unpredictable final two weeks of the campaign. Clinton could continue to hold a commanding lead in most conventional polls, but a wide spread between those surveys and the high-volume polls cited by Trump could create a second reality, creating uncertainty about the true state of the race.
“The average is kind of misleading,” said Murray, the Monmouth pollster, pointing to tracking polls this week ranging from Clinton’s 12-point lead in the first ABC News tracking poll over the weekend to Trump’s 2-point advantage in previous iterations of the Rasmussen and IBD/TIPP polls. “That’s a 14-point variance. If I look back at four years ago, it was only about 4 points then.”
Nolan D. McCaskill and Madeline Conway contributed to this report.