Skip to content

Editorial — Polls have problems

Polls are less reliable than they used to be, but more prominent than ever, particularly as an election campaign draws to a close.

There are literally hundreds of polls being conducted in individual ridings and nationally, even as some pollsters freely admit that polls are no longer as accurate as they once were.

Allan Gregg of Harris-Decima Research, who has been involved with polling since the 1970s, is among the most provocative of the internal critics.

“The dirty little secret of the polling business is that our ability to yield results accurately from samples that reflect the total population has probably never been worse in the 30 to 35 years that the discipline has been active in Canada,” he told an interviewer.

There are many reasons, but one major one is that traditional telephone polls have great difficulty in reaching the vast majority of people. Only 15 per cent of those contacted agree to take part in a poll, as compared to 70 per cent 30 years ago.

Many people do not have land lines any more, particularly young people, and they are thus less represented in poll sampling.

Perhaps one of the most troubling aspects of perpetual reporting on polls is that regional results are much less reliable than national results, yet are pulled out as showing a trend. This is now being demonstrated with the “surge” for the NDP in Quebec.  There is no doubt that the party is being treated more seriously by Quebec voters, but the initial reported “surge” in the polls came with a significant margin of error.

A survey conducted April 17 and 18, one of the first showing the NDP moving into second place in Quebec, had just 164 Quebec respondents and a margin of error of 7.8 per cent. Thus the reported 25 per cent support for the NDP in Quebec could have been as low as 17 per cent or as high as 33 per cent.

Yet the support level of 25 per cent is reported as fact and everyone jumps on the bandwagon.

Poll results are important information and in a democracy, information needs to be shared freely and disseminated widely.

However, the national media, the major players in publishing poll results, need to do a much better job in prominently reporting the number of people polled, and the margin of error both nationally and regionally. They must also explain why there is such a wide variation in many poll results.