by Mike McCabe, Executive Director
September 27, 2004
Almost every day we’re treated to the results of a new poll telling us how one of the races in Campaign 2004 is shaping up. What we’re not told is that the latest poll isn’t worth the paper it’s written on.
One of the great untold stories of this election season is the increasing unreliability of polls. The media continue to dutifully report the pollsters’ findings, even though doing so is a grave disservice to voters.
News stories about polls are a prime example of how news organizations unduly focus on who is winning the horse race rather than providing information voters can use to make up their own minds. Worse yet, the polls are illegitimate. They are deeply flawed barometers of public opinion.
Polling has been crippled by the rise of cell phone use. Telephone surveys are the staple of public opinion polling, and pollsters rely on something called “random digit dialing” to conduct their questionnaires. That means they use computer technology to randomly dial telephone numbers from published telephone directories.
The problem is that cell phone numbers are not published in those directories. So the large – and rapidly growing – ranks of cell phone users are excluded from these “representative” samplings of the public. The opinions of many young people in particular are not captured by pollsters because of this problem.
Accurate polling is further disabled by the growing revolt against telemarketing. The public’s hatred of nuisance phone calls has inspired millions to put their names on no-call lists. This phenomenon wasn’t caused primarily by public opinion polling firms but it affects them profoundly. It used to take maybe four or five calls to find someone willing to participate in a poll. Now pollsters will privately acknowledge that it can take 20 calls or more to find a willing participant.
That makes the people answering the pollsters’ questions oddballs by definition – they are doing something that 19 out of 20 people refuse to do. It also makes your average poll anything but random and hardly representative.
The final, insurmountable challenge for public opinion pollsters is trying to identify people who will actually vote. Ask 10 people if they plan to vote in the next election, and probably at least seven or eight will insist that they will. Then on election day you find out three or four of them were fibbing.
Recently a national polling firm conducted a three-day survey of “likely” voters and found President Bush leading John Kerry by 15 percentage points. A day later, the same polling firm started another four-day survey and this time found the race to be a dead heat. The pollster said the results show “voter opinion is unsettled.”
No way are voters that unsettled. What these results really show is that polls provide no meaningful insight into what voters are thinking.
Despite vexing social and technological changes that seriously undermine the legitimacy of the polling industry, gauging public opinion and predicting how voters will behave still is being passed off as science. In truth, it’s closer to palm reading or the daily horoscope.
The media can do something about the fraud that public opinion polling has become: They can stop reporting the pollsters’ findings.
If the media won’t do that, voters should take the polls with more than a few grains of salt. Or better yet, ignore them altogether. They are worthless.