Globe and Mail Poll Called 'Unethical' by Expert

On the Friday before the tightest provincial election facing British Columbia this decade, Canada’s most influential newspaper published dramatic poll numbers that appear to be based on biased methodology.

“This is unethical,” Vancouver pollster Angus McAllister told The Tyee. “The Globe and Mail ought to be responsible enough not to publish a poll where the ballot question is preceded by 14 questions that influence people’s answers.”

The poll, which was released on Friday the 13th, predicted that the BC Liberal Party would beat the New Democrats by a 13 per cent margin on May 17. Co-sponsored by The Globe and Mail as well as CTV, the poll was conducted by a Toronto firm called The Strategic Counsel.

The wide lead found by the Strategic Counsel poll not only exceeded both the 8-point lead predicted by an Ipsos-Reid poll and the 5-point lead predicted by a Mustel Group poll (both of which were conducted the same week), it also differed from those polls in that its Liberal lead exceeded its margin of error. At a critical moment in the campaign, the Globe’s 13-point prediction provided an appearance of invincibility that buoyed BC Liberals and demoralized New Democrats.

“With four days to go until the election,” The Globe and Mail reported on the front page of its BC section, “the governing BC Liberals have surged into a commanding lead that will produce another solid majority for the party.”

Ratcheting up bias?

Professional pollsters take considerable effort not to influence the thinking process of people they are questioning. For this reason, the professional practice is to ask how the respondent intends to vote before saying anything that might influence that decision. Pollsters generally ask about voting intent within the first three questions.

The Globe’s poll asked 14 questions before inquiring how respondents planned to vote.

“If you’re really trying to find out where people stand, you don’t want to ask them questions that will lead them in any one direction,” observed Steven Rosell, who is a partner to highly respected American pollster Dan Yankelovich.

Equally unusual were the questions themselves, and the sequence in which they were asked:

Question 3: “Some people say the election is about trust and keeping election promises. Other people say that the election is about providing effective government and ensuring that the economy continues to grow. Which of these two points of view best reflects your own?”

Question 4: “Some people have been saying it’s time for a change and a new government should be voted in. Other people have said that now would be the wrong time to make a change and we should return the Liberals to power. Which of these two views best represents your own?

Question 5: “Most people have been telling us that the BC economy is in pretty good shape right now. Some say the provincial government had very little to do with this improvement, so there’s no risk in changing to an NDP government. Others say that without the right approach, the economy will slow down and it’s too risky to switch to the NDP. Which one of these two views best represents your own?”

“This is outrageous,” McAllister responded, after The Tyee directed him to a document listing the questions. “They are using a sequence of questions to build an argument for the BC Liberal government.”

‘Definitely not accepted technique’

The Strategic Counsel poll asked nine more questions before inquiring which candidates respondents would support. Those questions included: “Which party had the best advertising?” and “Which party has fielded the best team of candidates?”

“When you ask a whole bunch of questions like these in a row, you are stimulating a thinking process that was not already there,” explained McAllister, a former vice president with Ipsos-Reid and Environics who now heads McAllister Research.

“This sort of questioning is a message-testing technique. It’s a way of testing messages and determining the impact of those messages on voter decisions. It is most definitely not an accepted technique for predicting election results.”

All but one of the 19 questions asked in the poll conducted from May 9-11 can be found in a document posted on Strategic Counsel co-founder Allan Gregg’s personal web site.

One question is curiously absent from the document posted on Gregg’s site, however. The document includes details about every question and answer except for number 14. The document offers no explanation as to why the question was redacted.

The missing question immediately precedes question 15, in which the Globe poll finally inquires as to which party candidates the respondent would support “if the election were being held tomorrow.” Neither Allan Gregg nor The Strategic Counsel was available for comment on this Sunday afternoon.

‘Definitely unethical’

Also unexplained is why The Globe and Mail sponsored the unusual poll.

The National Council on Public Polls is an association of polling organizations established in 1969. The US-based organization sets professional standards for public opinion pollsters, and advises media on how to interpret poll results. The council’s website includes a guide entitled, 20 Questions A Journalist Should Ask About Poll Results.

These include “What questions were asked?” and “In what order were they asked?”

“Sometimes the very order of the questions can have an impact on the results,” the polling association’s guide warns. “Often that impact is intentional; sometimes it is not. The impact of order can often be subtle.”

Rod Mickleburgh is bureau chief for The Globe and Mail's British Columbia office. He did not know who chose the poll’s questions. “The polling firm was selected in Toronto,” Mickleburgh said. “It’s only recently we’ve been using Strategic Counsel.”

“It’s possible we published a rogue poll,” he added. “Or, our poll could be right.”

“It’s the Globe’s responsibility,” McAllister concluded. “The Globe and Mail appears to have misrepresented a message-testing poll as a public opinion poll. And they’ve released those results just four days before an election. This is definitely unethical.”

Veteran political journalist Monte Paulsen is managing editor of The Tyee’s Election Central.