Data for Progress?

Enter Sean McElwee, founder of Data for Progress (DFP), the combination polling firm and self-described “think tank for the future of progressivism.” Some background for the uninitiated: McElwee, a skillful operator with a massive Twitter following, has attracted much attention in recent years. In a New York magazine article that painted NYC Democratic Socialists of America as a kind of lifestyle brand, McElwee loomed large. His weekly “socialist happy hour” in the East Village was described in breathless detail (“Democratic politicians in nice clothes Uber in to kiss McElwee’s ring and gain the trust of New York’s young socialist power elite”) with the reporter noting that “McElwee boasted to me that Senator Kirsten Gillibrand had been by recently, drinking the ‘piss bar wine.’” McElwee’s latest venture, a new organization called “Secure Elections for America Now,” has an acronym that happens to spell out “SEAN.”

There is no doubt, however, that McElwee’s ventures have produced useful work, especially Data for Progress’s academic-style studies on voting and public opinion.

But twenty days before New York’s primary election, DFP announced that it had conducted a poll of the race for the 15th Congressional District. The byline on the poll’s accompanying memo had only two names: Sean McElwee and Ethan Winter, an analyst. According to them, it was “currently a two-way race.” Rubén Díaz Sr, the conservative, was winning with 22 percent, followed by Ritchie Torres in a close second place with 20 percent. According to Data for Progress, Samelys López was in a distant sixth place, with a mere 2 percent of the vote. McElwee’s personal Twitter account was the first to tweet the poll, and he framed the results with a call to action: “Progressives have three weeks to prevent an anti-choice homphobe from being elected in the most Democratic district in the country by consolidating behind @RitchieTorres” [sic].

According to the memo DFP released, the poll was a “text-to-web and web-panel survey of 323 likely voters” residing in the 15th District. Right away, one has to question how much confidence we can place in the results of any poll of a downballot primary election, especially one with a limited sample size. The problem is compounded by the fact that the 15th Congressional District is the poorest in the entire country, and according to a recent study, 29.4 percent of households in the district either only have dial-up internet or don’t have any internet access at all. In a statement to Jacobin, McElwee defended the poll’s methodology, arguing that the issue of internet access was addressed by the poll’s additional use of text messaging to contact respondents.

But, besides the poll’s methodology, there was another problem: McElwee and DFP seemed to have a dog in the race. Almost a year before the election, an interview with Torres was featured in the Data for Progress newsletter. It was promoted by both DFP and McElwee’s personal Twitter account. Of course, the reams of evidence that Torres was far from the progressive he claimed to be went completely unmentioned. (In his statement to Jacobin, McElwee said he had no long-standing relationship with Torres, pointed out that Torres has endorsement of the city’s Progressive Caucus and support of progressives like Brad Lander, and said that Data for Progress wasn’t alone in warning of the danger of a split left vote in the race.)

Despite the poll’s methodology and the fact that it came from an organization that was, at the very least, friendly to Torres, the DFP poll was taken at face value by the media. It was described as an “independent” poll, and it quickly started driving the entire narrative of the race.

The very same day the poll was released, an article was published in the Washington Post citing it and saying that “some on the left fear Lopez, who had struggled to raise money, will split progressive votes and create a path for Rubén Díaz Sr., a conservative Democrat who calls himself ‘the opposite of AOC.’” Naturally, McElwee himself was quoted, taking a swipe apparently aimed at López: “Too often people get excited about someone because of their social media clout without recognizing that they have very little chance of winning.”

The media dominoes fell rather easily from there. Even the Intercept fell for the narrative, with an entire article about how the poll had “scrambled” the race, arguing that López was a potential spoiler: “Votes cast for López could conceivably end up electing Díaz over the far more progressive Torres.” The New York Times endorsement of Torres parroted the same argument: “There are several impressive candidates in the race. But coalescing Democratic support around Mr. Torres is especially important because of the presence of Rubén Díaz Sr. on the ballot.” The Times endorsement bought into the fearmongering, claiming: “Polls show that Mr. Díaz may be poised to win . . .” Of course, “polls” shouldn’t have been in plural — the Data for Progress poll was the only public poll of the race.

Rarely has a single poll of a single congressional district prompted so much media coverage. And the consequences were real: the media narrative kneecapped López in the crucial final weeks. Though she was endorsed by Bernie Sanders just days after the poll, the damage had been done — the narrative was perfectly timed to dull any momentum she might have gained. In the eyes of potential volunteers and donors, López seemed like a lost cause — at best not worth it, and at worst, a potential spoiler.

As a López campaign staffer myself, I noticed the effects firsthand: volunteers vocally expressed concern about the poll and became less willing to sign up for phone-banking shifts, and it grew increasingly difficult to involve new people who hadn’t volunteered yet. Why would anyone want to devote their time or money to a candidate who was only going to get 2 percent of the vote? The poll had a chilling effect on a campaign that, until that point, had been catching fire.

Instead, volunteer energy and money flowed to another excellent candidate: Jamaal Bowman, who was running in the Bronx’s 16th District. A DFP poll in his race had helped convince lefties from all over the country that he was the one to help. The poll’s influence didn’t go unnoticed: Shane Goldmacher of the New York Times remarked: “Does anyone put out polls to push media narratives more effectively than @SeanMcElwee?”

This massive inflow of energy and attention in the final stretch was likely crucial to Bowman’s win. To many observers, Data for Progress seemed to have proven its thesis — that polling can be used to help advance a left-wing political agenda and help secure primary wins for the “progressive” wing of the Democratic Party. McElwee has been taking an extended victory lap in the media and on Twitter.

In a recent Atlantic magazine profile, he was quoted giving himself glowing reviews: “Yeah, I’m good at politics,” he modestly told reporter Elaine Godfrey. “Listen to Sean. This is the lesson.” In the piece, McElwee is effusive about the NY-15 race in particular, saying that Torres’s win “is a point for the normie-progressive theory of change.”

The Snapshot Heard Round the World

But there’s a major problem with this neat and tidy story. When the voting results finally rolled in, it emerged that DFP’s poll in the López/Torres race had been wrong in almost every possible way.

The poll that had pretty much singlehandedly driven media coverage in the final weeks of the campaign — anointing Torres and casting López aside — turned out to be completely off the mark, an almost ludicrous misrepresentation of the state of the race.

In reality, Rubén Díaz Sr (the right-wing bogeyman who had been in first place in the poll and whom progressives supposedly had to unite to defeat) hadn’t been a threat at all. He ended up in third place, beaten handily not only by Ritchie Torres, but also by another candidate, Michael Blake. Blake, who in the DFP poll had been credited with a meager 6 percent — a whopping 16 points below Díaz — actually ended up with 18 percent of the vote, almost 4 points above Díaz. As for Samelys López, her share of the vote turned out to be seven times higher than the DFP poll indicated — 14 percent rather than 2 percent, a virtual tie with Díaz’s 14.4 percent. Contrary to McElwee’s fearmongering, progressives didn’t need to “unite” to defeat Díaz. Samelys López had opened up a democratic-socialist lane all her own.

To put this in perspective: despite the massive last-minute media hype about the danger of the “unelectable” López splitting the vote and electing Díaz, the socialist López essentially tied Díaz herself. Running a grassroots campaign with a tiny fraction of Torres’s money, having no prior name recognition, campaigning in a pandemic, and fighting against a contradictory media narrative that painted her as somehow both inconsequential and a potential spoiler, López defeated all but three of the twelve contenders in the race, receiving more than three times as many votes as the well-known former speaker of the City Council.

The media narrative of the race had been a sham — and it had largely been based on a single poll from Data for Progress that turned out to be completely wrong.

Wrong or not, however, the survey served its intended purpose: Torres won. McElwee even confessed on Twitter that the poll had an agenda, acknowledging it was intended as a wake-up call to progressives to consolidate around Torres. Responding to a Jacobin request, McElwee said that “polls are snapshots, and the poll was taken a month out. Our polling close to elections has proven to be incredibly accurate.”

Of course, this particular snapshot came at the expense of López, the candidate that DSA had democratically decided to endorse and behind whose campaign it put massive amounts of volunteer energy. Had the narrative been different, and had the spending and media hype flowed her way, as they did to Bowman and Torres, the outcome probably would have been very different, too.