How polls can be distorted


People with a keen interest in federal elections pay close attention to polls, but pollsters and pollsters can mislead voters with poll questions, available answers, and distorted headings. It states that there is.

Ongoing polls during the campaign help candidates, political parties, and political strategists evaluate their support and adjust their approach accordingly. However, according to Tom Flanagan, a political scientist who led campaigns for various candidates and political parties, the results of public polls can also affect the actions of voters themselves.

“Voting results can be effective in encouraging strategic voting in multi-party races,” Flanagan said in an interview.

“Current polls showing that the Conservatives are in the lead may encourage some [NDP supporters] Hold their noses and vote for liberals. I think there is some political science literature on this, but it is difficult to study strategic voting because people do not like to give a direct answer to voting for the second option. “

An emeritus professor at the University of Calgary said it would be in the financial interest of pollsters to get accurate voting results because credibility leads to clients. Still, pollsters may fail.

“Polsters are people with their own political views. No matter how objective they may be, their personal commitment may color their question choices and wording. , I don’t think they are consciously aiming to distort their findings. There is no reward. “

Jeffrey Hale, a professor of political science at the University of Lethbridge, agrees that most pollsters are trying to get it right, but for those who fail, it’s hard to conclude why.

“Polsters can show’house prejudice’, but they are methodological and how much’torque’is open to interpretation by those who study such things.” Hale said in an interview.

“Some companies haven’t paid much attention to this fall this fall. Poll Aggregation … There may be a methodology for predicting certain assumptions about the development of the race. This may actually work. It may not be, but [influence voter perception and behaviour] Combined with the editorial stance of a particular newspaper. “

Given how far away pollsters were about the Progressive Conservative Party’s upset victory in Nova Scotia last month, “reputation pollsters need to stay as close as possible to the election day to catch the last-minute shift. There is, “said Mr. Hale. ..

Challenges for pollsters

Research Co. Mario Canseco, the president of the company, told The Epoch Times that the smaller the sample size, the more likely it is that regional variability will increase. He said a poll of 1,000 voters achieved a reasonable +/- 3.1% confidence interval at the national level, but not very well locally.

“Let’s say we have 150 people in BC and about 110 people in Alberta, which gives us a margin of error close to 7-8 percent, depending on the population,” says Canseco.

Sean Simpson, vice president of public relations at Ipsos, said asking the right people with the right number of people is the basis for accurate results.

“The first key is to reduce the causes of sampling errors, which is to make sure we’re talking to a sample that reflects the population we’re studying,” he said in an interview.

Simpson said Ipsos’ national polls are careful to maintain a representative number of regions, genders and education levels. A failed pollster “takes a very biased sample and requires a very large weight to produce an essentially representative sample,” he says.

According to Simpson, even demographically accurate and large samples could be challenged by changing turnout, and this election “increased the complexity of COVID-19.” is.

“Global news polls show that one in four Canadians says it’s unsafe to vote directly.”

Canseco approaches pollsters with very specific ideas for some clients asking questions and the answers that will be available, but sincere pollsters need to insist on the right approach. Said.

“It must be the right question …. You shouldn’t actually load the questionnaire in a way that people support what you’re saying.”

He recalls, along with his former employer, when the then conservative federal finance minister Jim Flaherty’s political opponent proposed a voting question on the 2006 income tax decision.

“I remember saying,’There is no way to ask this. It’s a very long preamble, the question is wrong. You’re essentially putting a lot of words in people’s mouths. “He said.

Two weeks later, the client found another company to ask a question.

“Sometimes you turn down people, and … the data collection agency will say,’It’s okay. We have quotas, so we’ll ask you whatever you want.” And that’s bad for the reputation of the industry as a whole. .. “

Misrepresentation of media

Some pollsters have stated that the media could also abuse poll results in a way that is generally interpreted. Simpson believes political experts are particularly vulnerable to this problem.

“If a party is polling more people [else] If is down 2 points, they are making a firm conclusion trying to explain why we see it, [yet] It doesn’t really matter [political] Did anyone do this or say it, “he explains.

“In fact, it’s all variable …. It can be a random sampling error because you haven’t done a full poll.”

According to Canseco, party voters may protest the headlines used by pollsters in press releases. He is even more annoyed when the media combines poor poll questions with misleading headlines.

“We allow well-known people to ask undecided questions or just say A or B. And the headline says,” 70% of Canadians feel this way. Is written.Well, what happened to the other 30 people [percent]?? Are they undecided? Did they choose something different? ” He said.

“If the data is not collected properly, it will be reported as if it were genuine and there is definitely a problem.”

Lee Harding

Lee Harding is a Saskatchewan-based journalist and think tank researcher and contributor to The Epoch Times.