Polls show substantial support for candidates like RFK Jr. How much of it is real?

Third-party and independent candidates like Robert F. Kennedy Jr. are making an unusually strong showing in early polls of this fall’s presidential race – a finding that not only highlights a strain of dissatisfaction with Joe Biden and Donald Trump but also adds further uncertainty to what is shaping up to be a close contest between them.

Across five national polls released in March and April – from Quinnipiac University, Fox News, Marquette Law School, NBC News and Marist College – Kennedy received an average of 13% support for his independent presidential bid when his name was explicitly included in the survey question, with independent candidate Cornel West and Green Party candidate Jill Stein taking an average of 3% each. Given the narrow margin between Trump and Biden, who are effectively deadlocked in many surveys, even a fraction of that support could prove crucial to the election’s outcome.

Historical precedent, however, suggests that third-party and independent candidates’ election performances rarely live up to their polling.

“When people tell us that they’re going to vote for a third-party candidate, they’re actually telling us one of two things,” said Dan Cassino, the executive director of the Fairleigh Dickinson University Poll, who has researched the overestimated support for such candidates in surveys. “Some proportion are saying, ‘I hate both of the major party candidates, and I’m going to illustrate that by saying that I’ll vote for literally whoever you put in the third spot (whether I’ve heard of them or not).’ Others are saying, ‘I really like that third party candidate, and I’m going to vote for them!’”

Given the relative unpopularity of this year’s two major presidential candidates, Cassino noted, the first group – those who treat being polled as an opportunity to vent their discontent with the system – are likely to far outnumber those who are fully committed to backing a specific alternative. In a national CNN poll last year, 39% of voters who said they would support Kennedy for president also said, in a separate question, that they didn’t know enough about the candidate to offer an opinion of him.

But finding the right questions to disentangle those sentiments is difficult, and past polling has frequently struggled to estimate the support for such challengers. In the 2014 midterms, most surveys overstated the eventual support for third-party candidates. But while pre-election polls that asked voters about third-party candidates by name tended to inflate their support, those that omitted them often underestimated it, according to data compiled by Joe Lenski, the executive vice president of Edison Research. Estimations of third-party support were less accurate further out from Election Day, the report also found.

More broadly, while the public have long expressed enthusiasm for new political alternatives as a concept, that fervor has rarely translated into traction for any specific movements. For more than a decade, Americans have told Gallup that the Republican and Democratic parties “do such a poor job that a third major party is needed,” while continuing to vote overwhelmingly for major-party candidates in elections.

Even in the 2008 president contest between two broadly liked candidates, Barack Obama and John McCain, nearly half of registered voters said in Gallup’s polling that they wished they had additional choices to pick from.

Polling on third-party and independent candidates could be especially tricky to navigate this year, for a number of reasons:

Dissatisfaction with the major-party candidates is unusually high

In a recent Gallup poll, 29% of US adults said that neither Trump nor Biden would make a good president – still a distinct minority of the populace, but a slight uptick from 25% in the summer of 2020 – and significantly higher than the 8%-to-19% shares who said the same of the candidates in 2004, 2008 and 2012. (Gallup did not ask the question in 2016, another year that featured relatively unpopular major-party nominees in Trump and Hillary Clinton. Libertarian Gary Johnson took about 3% nationally in that contest, the best showing for a third-party candidate since Ross Perot in 1996).

Kennedy’s name may be giving him a superficial boost in recognition

Kennedy, who carries a storied surname, enjoys the sort of name recognition that most presidential challengers would envy. In the March Quinnipiac poll, 31% of registered voters said they hadn’t heard enough about him to form an opinion, compared to 71% for West, and 69% for Stein. But other polling suggests a significant gap between awareness of his existence and deeper knowledge of his positions: In a December Monmouth poll, roughly half of registered voters said they hadn’t heard anything about Kennedy’s positions on public health issues such as Covid-19 and vaccines, with a similar share saying they weren’t aware of Kennedy’s support for unsupported “claims that autism is linked to vaccines.”

Independent candidates’ level of ballot access is still unsettled

Each state has its own process for candidates hoping to appear on the ballot. While Kennedy has said he hopes to make the ballot across all 50 states and Washington, DC, it’s possible that some voters now backing him or other independent candidates in national polls won’t actually have the option to do so in their state this fall.

The ways polls are conducted has changed

For decades, most public polling was conducted over the telephone by live interviewers. That gave respondents a chance to volunteer responses that weren’t explicitly included by the pollsters – for instance, answering a question about their preference between two main candidates by offering up that they’d actually prefer a third one, or that they wouldn’t vote at all.

Many pollsters, however, have moved at least in part to conducting online surveys, making it more difficult to incorporate similarly off-the-cuff responses. Instead, pollsters generally have to make a clear-cut decision as to whether an option should be present or not. While neither mode of polling is inherently better or worse, the shift is a further complication in trying to compare current polling with what we’ve seen in the past.

What’s the best way of demonstrating this uncertainty? Cassino suggested starting with an experimental approach: asking half of a poll’s respondents to choose between Biden, Trump and Kennedy, with the other half instead choosing between Biden, Trump and a different third-party candidate with little name recognition. The difference in support between the third options would then indicate the share of active support for Kennedy, rather than disgruntled respondents casting the equivalent of a protest vote.

Many public pollsters, meanwhile, are currently asking two different questions about the general election – one naming just Biden and Trump, and another including additional candidates – and separately reporting the results of both. Polls are often described as providing snapshots in time of public sentiment. But even snapshots taken at identical moments can reveal different facets of their subjects, depending on the framing. That’s a good reminder of the limits to how precisely polling can capture the current state of any race, let alone predict how it might evolve in the months to come.

For more CNN news and newsletters create an account at CNN.com