Let’s talk about polls. More specifically, let’s talk about the polls that have recently generated headlines claiming Terry Goddard has erased what some said was a 20-point lead for Jan Brewer, and that Felecia Rotellini and Tom Horne are neck and neck in the contest for attorney general. There are two things driving those stories: inattention to detail by reporters and sleight of hand by the pollster.
The poll on the governor’s race was released Oct. 11 by Behavior Research Center as part of its periodic Rocky Mountain Poll series, which largely focuses on politics and campaigns. Publications across the country latched on to the story, as it makes for good copy: Jan Brewer appeared invincible after signing SB1070, but now she’s about to be toppled by her Democratic challenger. That story line was promoted by BRC in its release announcing the poll, which touted Brewer’s slim three-point lead.
If that were true, it certainly would be news. But it’s not. The poll was of 555 registered voters, but a sizeable number of them said they didn’t plan to vote this year. In fact, only 405 of the respondents said they were certain they would vote. Among those people – really, the only ones that matter – Brewer was up by double digits, leading Goddard 46-35, a far cry from the 38-35 lead the poll’s narrative claimed she had.
Likewise, today’s Rocky Mountain Poll on the attorney general race touts the results from the larger pool of voters in its headline: “Attorney General a Dead Heat.” Among all of the surveyed voters, the race is tied at 34 points, while Horne has a 40-36 lead among the likely voters. (The margin of error on the likely voter figures are 4.96 percent, which means the race is very close – just not as close as the pollster indicates.)
Many of my colleagues in the Fourth Estate treated the polls as gospel and unflinchingly reported as fact in both print and on television the “news” that Brewer’s large lead had magically evaporated. That shouldn’t have passed the smell test for reporters, especially when Rasmussen Reports said less than two weeks ago that Brewer had a 55-39 lead. Yes, Rasmussen’s methodology – the firm uses autodial polling, not interviews from live operators – is routinely questioned, but its figures on the governor’s race largely match figures from private polling other reporters and I have heard about.
Some got it right: Both Howie Fischer, dean of the Capitol press corps, and Arizona Capitol Times reporter Jeremy Duda wrote stories that either ignored the poll results from the “registered voters” or correctly noted they were announced, but weren’t the most important figures in the poll.
We need to be more diligent, especially when it comes to reporting on polling. Polls are important, but they are not infallible. They are not gospel, but merely a snapshot of a specific moment in a race. Reporters have a duty to carefully examine polls, not just take their results at face value and parrot those numbers to the public.