Political campaign polling always has had its problems. Back in 1948, polls conducted by the Gallup organization showed Thomas Dewey ahead of Harry Truman in the presidential race, so when early returns on Election night had Dewey in the lead, the Chicago Daily Tribune was confident about the final result. Of course, Truman had the last laugh when he held aloft the Daily Tribune edition headlined “Dewey Defeats Truman” — at his victory party.
That was 56 years ago. You would think that polling accuracy would have improved since then.
However, even though great strides have been made in public opinion research, most political polls do not utilize the best methods, either because of bias, lack of time and resources, or all of the above.
Last week, the Republican Party of Arkansas touted a Survey USA poll commissioned by KTHV that showed President Bush leading John Kerry 53-44 percent among likely Arkansas voters. The result was a significant departure from other polls that indicated the race was a dead heat.
The significance of these opinion surveys goes beyond bragging rights for one campaign or another. They influence a candidate’s decision about whether to invest in a particular area, and they affect the tenor of news coverage, by casting the race in terms of “front-runners” and those who “trail behind.” (The Democrat-Gazette ran a five-column headline above their article about the KTHV/Survey USA poll.) All of this combines to form impressions in the minds of voters, who can either be energized or demoralized by how their preferred candidates are polling.
Which is why news organizations especially should insist on the best methods for the polls they commission. However, the reality is that newspapers and television stations across the country are only interested in sharing credit for the surveys and making news by announcing the results. They do not get involved with how the polling organization obtains its results.
KTHV’s news director, Mark Raines, said he is not familiar with Survey USA’s methodology, and doesn’t know much about polling. However, he said that KTHV works with Survey USA because the firm has a “good track record.”
According to the Survey USA website, they conduct their polls by using automated calls to random households. A recorded voice asks the questions (a KTHV anchor did the work for Arkansas), and respondents press buttons to indicate their answers (“Press one for Bush, two for Kerry”).
Michael Traugott, a professor of political science and communications studies at the University of Michigan, has his own term for this kind of polling: CRAP, for Computerized Response Automated Polling.
“It is not a good way to be conducting election polling,” Traugott, a survey research expert, said from his office in Ann Arbor. He pointed out that response rates for CRAP polls are lower than average, probably because there is no live person on the other end of the line. For the same reason, respondents cannot ask for clarification or repetition of the question, and there is no way to determine who is answering in behalf of the household. It may be a true likely voter, or it may be a 10-year-old kid.
Traugott says that this automated method allows Survey USA to offer a cheaper product to clients like KTHV. “You get what you pay for,” he added.
Unfortunately, what you get is a bad poll, and one that may be illegal as well. Section 5-63-204(a)(1) of the Arkansas criminal code says that it is unlawful for any person to use a telephone for the purpose of gathering data “when such use involves an automated system for the selection and dialing of telephone numbers and the playing of recorded messages when a message is completed to the called number.”
Survey USA is not the only polling firm with suspect methodology. Zogby International has used an “interactive” survey in Arkansas, which limits participation to those who have access to the Internet. Gallup is under fire for oversampling Republican voters in its nationwide polls. And even the best-designed public opinion research conducted by telephone overlooks the growing population of Americans who exclusively use mobile phones.
Therefore, whether the polls show Bush or Kerry ahead, at this point they are probably inaccurate, and they are definitely unreliable predictors of the actual outcome. The race is very fluid, and important events like the debates will radically and quickly shift opinions.
With that in mind, it is more important than ever to question everything you hear, gather as much information as possible, and decide for yourself.