ROUNDHOUSE ROUND-UP: POLLS, INFORMED AND UNINFORMED

A version of this was published in The Santa Fe New Mexican
April 7, 2006

Poll results released by Geno Zamora’s campaign for New Mexico attorney general show that 44 percent of 400 likely Democratic primary-election voters contacted during the last week of February preferred candidate Gary King. Zamora was the choice of 9 percent, while 8 percent liked Lem Martinez, and a whopping 39 percent were undecided.

But wait. That’s only the first round.

Zamora’s next set of poll numbers — obtained “after comprehensive information on all three candidates is presented” to the same 400 Democrats — has Zamora pulling ahead significantly, leading King 33 percent to 24 percent, with Martinez lagging behind with 19 percent.

As Gomer Pyle used to say, “Surprise, surprise, surprise!”

So what is this “comprehensive information” that so boosts Zamora (and Martinez) and so deflates King? What could the pollsters have said to cause poll participants to give Zamora an extra 24 points and cause nearly half of King supporters to flee?

Did they claim that Gary King shot a man in Reno just to watch him die?

Nothing like that, the Zamora camp says.

But they won’t say exactly what mysterious “information” was used in the poll conducted by the Washington, D.C.-based Bennett, Petts & Blumenthal firm, which is described as having a statistical margin of error of 4.9 percent.

Although Zamora himself told me early Wednesday afternoon that he had no problem with releasing the information used in conducting the poll, later in the day campaign spokesman Allan Oliver refused to hand it over.

“I hope you understand that the poll is a part of our strategy,” Oliver said.
I understand.

In general, Oliver said, the “information” consisted of the pollsters “progressively going through each of the candidates’ experience.”

There also was “information on each of the issues.”

Oliver said the information was “even handed.”

But when asked if he could categorically say that there was no negative information given about King or Martinez, Oliver wouldn’t answer with a simple yes or no.

“We talked about all their experience,” he said. “I wouldn’t characterize it as positive or negative. I don’t want to characterize it either way.”

I understand.

Pushing too hard? Oliver insisted that this was not a “push poll.”

And maybe that’s right.

While that term is bandied about a lot, according to the Wikipedia, “True push polls tend to be very short, with only a handful of questions, so as to make as many calls as possible. The data obtained is discarded rather than analyzed.”

The example they cite is the 2000 South Carolina Republican primary where a telephone pollster asked GOP voters, “Would you be more likely or less likely to vote for John McCain for president if you knew he had fathered an illegitimate black child?” McCain lost that primary to George W. Bush.

However, the Wikipedia article says, “The term is also sometimes used incorrectly to refer to legitimate polls which test political messages, some of which may be negative.”

Sounds like “strategy.”

King responds: Contacted Wednesday, King shrugged off Zamora’s poll results.

“I’ve had a pollster tell me that they can get me any answer they want if they just ask the right questions,” he said.

The “uninformed” poll results obtained by Bennett, Petts & Blumenthal are consistent with his own poll numbers, King said.

But King said the “informed” poll results “probably indicate that if I sat on my hands and didn’t do anything, it would be possible for an opponent to close in. I’m not going to let that happen.”

See the Zamora poll results on Joe Monahan’s blog.

UPDATE: Due to some kind of human snafu at The New Mexican, this column did not appear in Thursday's paper as it usually does. It should be in Friday's paper. (Keep your fingers crossed.)

I changed the date of publication at the top of the post.

On Thursday, I spoke with Albuquerque pollster Brian Sanderoff, who told me that Zamora’s poll doesn’t sound like push polling.

“It sounds like aggressive message testing,” he said.

Of the “information” the poll-takers gave, Sanderoff said, “If it was equally balanced between negative and positive information about the candidates, there’s no reason for (the numbers) to have changed that much.”