Trying to figure out the state of the Democratic presidential race? Get in line.
Public opinion surveys have varied by the day, particularly in the early voting states of Iowa and New Hampshire, where some candidates can point to individual polls to argue they are building momentum.
Former Vice President Joe Biden has held a relatively steady lead in national polls, though an Economist-YouGov poll this week found him holding just a 1 point lead over Sen. Elizabeth Warren (D-Mass.).
{mosads}A poll from The Hill-HarrisX found Biden with a healthier 15 point lead over Sen. Bernie Sanders (I-Vt.), who stood in third place in the Economist poll. Warren was further behind, winning 10 percent support compared to 31 percent for Biden and 16 for Sanders.
Most polls have found Biden with a lead in New Hampshire, though a Gravis poll this week put Sanders in first place.
A Monmouth poll found Sanders trailing badly in Iowa, behind not only Biden and Warren but also Sen. Kamala Harris (D-Calif.).
The different signals in the polls has even experts questioning who is up, who is down and whether the horse race surveys can be trusted.
Anxiety is also growing among the campaigns, as candidates and aides rush to highlight their incremental gains or question surveys that show them losing support.
Some say the political amateurs and professionals should just give it a rest when it comes to polls.
“It’s August of 2019, the horse race polls are absolutely worthless,” said Chris Kofinis, a Democratic pollster.
Pollsters interviewed by The Hill defended their findings and methods, even as they warned against reading too much into their results.
An Iowa Starting Line-Change Research survey of Iowa drove red-siren headlines this week, finding Warren jumping out to a double-digit lead with 28 percent support, followed by Biden at 17 and Sanders at 17. The same poll from May found South Bend, Ind., Mayor Pete Buttigieg with a small lead in Iowa, the only survey of the cycle to find him leading anywhere.
The Change Research online survey recruited respondents through ads on social media, which some experts say is prone to manipulation, particularly around the Iowa State Fair, when enthusiastic supporters are eager to pump up their preferred candidates.
A Pew Research study from 2018 warned that “even the most effective adjustment procedures were unable to remove most of the bias” from internet polls and that “very large sample sizes do not fix the shortcomings of online opt-in samples.”
Pat Reilly, the co-founder of Change Research, told The Hill they have a “unique patent-pending” bias correction mechanism that “enables us to quickly and dynamically build a sample that reflects the population.”
The weekly Economist-YouGov poll found Warren in a statistical tie with Biden for first place nationally. That survey put Biden at 21 percent and Warren at 20 percent.
{mossecondads}It was a striking finding since only a handful of national surveys conducted this cycle have registered Biden with less than 30 percent support nationally, and none have found a second-place candidate that close to the front-runner.
Representatives for the Economist-YouGov poll were not available to comment. That survey, which has consistently found Biden in the low 20s, is not a traditional likely voter model, but rather is pulled from a recruited panel of more than 10,000 respondents who are surveyed regularly.
The Hill-Harrix X survey found Warren with 10 percent support nationally and in third place overall, a much more dismal result than several other recent surveys.
Pollsters attribute the different findings in various polls to the generally soft support for candidates at this early stage in the race.
They say it is causing a “sloshing” effect that results in support shifting back and forth between the candidates.
Some also point the finger at the media, saying voters are quick to change their minds based on the latest news cycle.
“This particular election has been nationalized in the media unlike any before, from the logistics of getting into debates to the national media environment in the age of Trump,” said Monmouth University pollster Patrick Murray.
“Talking to voters in Iowa, they will rattle off six or seven names sometimes, very few people are fully committed,” he continued. “They say it’s difficult to focus on just one candidate, so I don’t see the polls right now as predictive at all of what will happen here in February. Things will shift right up until the last minute. Many of these voters will change their minds from their first choice and have no idea who their second choice will be.”
The Sanders campaign has flat-out rejected polls that find him losing ground, saying the surveys don’t reach his young supporters or capture the activist energy around his campaign.
They were irritated with the Monmouth poll that found him with just 9 percent support in Iowa.
Murray said he stands by the poll’s methodology, noting that his sample reflects the voters that traditionally turn out to caucus. If Sanders is drawing new voters, Murray is confident that will show up in the polling data as the caucuses near.
Of course, Sanders has also benefited from a survey that is an apparent outlier.
The Gravis Marketing poll released this week found him leading Biden 21 percent to 15 percent in New Hampshire, the first Granite State poll to find him in the lead.
Sanders won big in New Hampshire in 2016 and his 21 percent in that poll is not far off from where other polls have found his support. But the survey raised questions among experts about the small sample size of only 250 Democrats.
“Biden and Sanders are within the margin of error in our poll, so this just tells you that it is a very, very close race,” said Doug Kaplan, the president of Gravis Marketing.
The polling industry is still grappling with big picture challenges, such as the migration away from landlines to cellphones and the proliferation of online polls, which some view as unreliable.
It is harder than ever to get likely voters to answer their cellphones and respond to survey questions, which oftentimes results in smaller sample sizes or polls of adults who are unlikely to vote in the primary.
The massive field of contenders is also a challenge for pollsters, particularly when the 2 percent threshold to qualify for the fall debates might require a pollster to randomly stumble upon six individual supporters for a candidate.
The Gravis Marketing survey of New Hampshire found eight low-polling candidates combining for 22 percent, a large chunk of support that most believe will ultimately go elsewhere and fundamentally reshape the race.
“Voters are overwhelmed,” said Kofinis. “The size of the field is having a freezing effect on a lot of people, so contrary to the way these polls are sometimes used and abused by the media, these current surveys are not meant to predict who is going to win where in 2020. These are snapshots of a race at a very early stage.”
Chris Mills Rodrigo contributed.