Sometimes, Bloomberg Pollsters Ask Negative Questions Because They Want to Know the Answers

This morning’s Times story about a series of polling calls that included negative information about Anthony Weiner gives the impression that the calls were a form of Bloomberg-campaign “push-polling.” The article never asserts this directly.

The piece leaves that to a Weiner aide, who says that the calls sound “suspiciously like” a push-poll, and to Weiner pollster and adviser Joel Benenson, who says, as an apparently general observation, that push-polling is a “dishonorable” practice. Also, tellingly, the the headline—”‘Survey’ Calls Attack Bloomberg Rival”—puts quotes around “Survey.”

Taken overall as a piece of journalism, it’s a useful article (by a good reporter), revealing what seems to be the Bloomberg campaign’s thinking about Weiner’s potential vulnerabilities—and indicating the fact that the Bloomberg campaign also continued to take Weiner somewhat seriously until at least mid-March, despite his quasi-withdrawal from the race right around that time. In addition to the negatives reportedly listed by the pollster (“pollster”?) on the calls, Bloomberg spokesman Howard Wolfson also obliges—in what I figure constitutes the most spectacular public attack to date by one Schumer alumnus against another—by repeating the charges, and then some:

“Unfortunately for Congressman Weiner, the fact that he takes money from lobbyists and special interests, misses votes and has not passed any significant legislation isn’t a push poll — it’s his record,” Mr. Wolfson said.

But whatever the calls are, they’re almost certainly not push-polling.

As former pollster Peter Feld explains on his blog, the calls didn’t go to enough voters to sway a significant portion of the electorate. And, as Feld points out, it’s not uncommon for campaigns to test negative messages.

Here’s Feld:

I conducted hundreds of similar surveys on behalf of various liberal Democrats over the years.


The results are used to rank messages and select the most effective, as well as to gauge how moveable the electorate is (i.e., do voters change their minds after hearing info?) and to identify those voter groups who are most persuadable.


To be effective, push-polling has to contact thousands — tens or hundreds of thousands — of voters, not hundreds. Further, those “push-polling” calls are brief — usually under one minute, just enough time to pretend to be a real survey and then to spread a single piece of pre-tested dirt — and they are invariably done close to the election. Rather than a random sample, those polls target just those swing voters who are thought (demographically) to be most susceptible to influence.


Unless he has evidence that many thousands of voters received these calls, and that the Bloomberg survey didn’t follow the template of standard message-testing research, he is simply fooling a gullible reporter.

There could hardly be a less effective or efficient way of influencing an election than to call a random sample of several hundred voters, and keep them on the phone for 10, 15 or 20 minutes, eight months before voters go to the polls.

Sometimes, Bloomberg Pollsters Ask Negative Questions Because They Want to Know the Answers