out of comptrol

Could the Polls Be Underestimating Eliot Spitzer’s Support?

Comptroller candidate Eliot Spitzer campaigns at the West Indian Day Parade on September 2, 2013 in the Brooklyn borough of New York City. Over a million people are expected to attend the 46th annual parade.
Unabashed Spitzer supporter. Photo: Michael Loccisano/Getty Images

There’s a famous theory in polling known as the Bradley Effect. Named after Thomas Bradley, the former mayor of Los Angeles, it suggests that in a race between a white candidate and a black candidate, some voters will tell pollsters that they intend to vote for the black candidate, when in reality, they’ll actually vote for the white candidate. The idea is that, even in an anonymous poll, people want to give the answer that they think they’re supposed to give — in this example, the racially progressive answer. 

As New Yorkers head to their old-timey voting booths today,the comptroller race between Scott Stringer or Eliot Spitzer is essentially a dead heat. Quinnipiac has Stringer up seven points, Marist shows Spitzer leading by two, and PPP has Spitzer ahead by four. But it’s possible that these numbers fail to give an accurate snapshot of the contest. What if the same social-desirability bias that leads voters to lie about voting for the black guy also motivates them to lie about voting for the creep who resigned from office in disgrace after having sex with hookers?

I think there’s some possibility of that,” says Tom Jensen of Public Policy Polling. “All of the polling, both public and private, when Mark Sanford was running for Congress earlier this year underestimated how well he ended up doing. I think at least part of that was hesitation of people to tell pollsters they supported the Appalachian Trail guy, and you could see something like that with Spitzer too.”

Don Levy, the director of the Siena College Research Institute, says he’s “intrigued by the idea,” and that such an outcome “wouldn’t surprise” him.

I was talking about that earlier today, and sort of mulling that potentiality over in my thinking,” Levy tells Daily Intelligencer. “I really don’t have anything in the data that supports that — it’s a hunch, it’s a theory.” 

The Spitzer Effect, let’s call it, would presumably be most prominent in phone interviews with real live humans, as opposed to automated phone polls. And, indeed, of the three comptroller race polls conducted in the past week, Spitzer’s best numbers come from PPP’s robopoll.  

There’s a lot of data in polling, the gender effect of the interviewer,” Levy says. “We have approximately equal numbers of men and women working for us as interviewers [at Siena]. Is it a possibility that a respondent would be disinclined to tell a woman who’s interviewing him that he’s voting for Spitzer? I think you have to consider that as a possibility.”

Anecdotally, Daily Intelligencer has heard from various men across the city who intend to vote for Spitzer but would rather not tell their significant others about it. 

She would never support the idea of me liking or voting for someone who basically had no regard or care about their wife,” one secret Spitzer voter, the CEO of a philanthropic organization, told us, referring to his girlfriend of a few months.

I’m hoping that it doesn’t come up,” echoed a 29-year-old Williamsburg resident who is currently seeing someone. “She probably wouldn’t make a big deal, but maybe she would.”

Even some men entrenched in long-term and committed relationships are wary of revealing their voting preferences. “I’m voting for Spitzer but am waffling about telling my wife,” one anonymous New Yorker admitted. 

Not all pollsters are believers in the Spitzer Effect. 

sort of doubt it,” Mickey Carroll of Quinnipiac says. “I can’t imagine that there’s a lot of sympathy around for Spitzer by people who just don’t want to admit that they’re sympathetic to him, but there might be.”

Harry Enten, who blogs about electoral statistics for the Guardian, is also skeptical. He says Mark Sanford overperformed in his congressional race last year simply because “Republicans were pegged as less likely to vote than they actually turned out being.”  

As for Spitzer, ”I don’t think there’s any proof that people are holding back,” Enten says. “It doesn’t mean they aren’t. Yet, I would be more likely to think that if Spitzer did better, it would be because polls in NYC in general are not very good.”

We may find out tonight. Unlike Spitzer, Anthony Weiner is far out of contention in his race, but the same polling phenomenon would apply to him as much or even more so than it would to Spitzer. If both Weiner and Spitzer outperform the polls, that might be a good indication that the Spitzer Effect is real. 

Could Polls Be Undercounting Spitzer Support?