A headline in the May 2, 2014 Halifax Chronicle-Herald caught my eye. It appeared over a story about the highly controversial, $164 million, second convention centre that may be built at taxpayers’ expense in the downtown core of Nova Scotia’s capital city:
Poll: Most back convention centre
The story under the headline reported that 61% of those who live in the Halifax Regional Municipality support the new convention centre while 33% oppose it.
Wait a minute. Hadn’t I just read about a poll in which a majority opposed the new convention centre? Sure enough, when I checked the Herald archives, I found this headline on April 25, 2014:
Poll: People oppose Halifax convention centre
The story under that headline reported that 86% of Nova Scotians agreed the provincial government should seek a second opinion before spending public money on a new convention centre, while 89% felt that the federal money set aside for the centre would be better spent on roads, bridges and public transportation across the province.
“Holy liftin’ Jehosaphat!” I exclaimed. “What in hell happened to turn public opinion around within seven short days?”
Battle of the polls
Well, when you look more closely, it turns out that the two polls were released by supporters and opponents of the convention centre and the Herald rather uncritically repeated their claims.
The poll that the Herald wrote about on May 2nd was from a polling company, whose Chairman and CEO, is a vocal supporter of the convention centre. No surprise then that the company’s poll showed strong support. The April 25th story was based on a poll commissioned by a coalition of non-profit groups that are vigorous opponents of the project. No surprise there either. Their poll showed strong opposition.
But if polling is based on well-established statistical science, how could a majority be both for and against the convention centre in less than a week?
Part of the answer is that “public opinion” is an abstraction. It’s created when it’s measured and also according to how it’s measured. People who may have given little or no thought to an issue often answer a pollster’s questions by responding to the information the pollster supplies or omits.
Let’s say, for example, that a pollster phones a random sample of people and asks: “As you may or may not be aware, a distinguished scientist says the moon was once part of the Earth, do you completely agree, mostly agree, mostly disagree or completely disagree?”
We might learn from this poll that 66% of Canadians side with the scientist while 30% say no, they don’t believe the moon was once part of the Earth with the other 4% in the “don’t know” or “refuse to answer” category. This result could yield an interesting headline: “Two-thirds believe moon once part of Earth.” But that two thirds would exist only because of the poll since most people haven’t spent a lot of time considering the question and most know little or nothing about the various theories about how the moon originated. In other words, there is no “public opinion” out there about the origin of the moon. But that opinion could be created and measured by polling.
This poll would be a harmless oddity unless organized groups had a vested interest in its results. Perhaps a multi-national mining company wants the US government to pay for an expedition to the moon to extract the precious metals that were torn from the Earth when the moon broke away.
What is the question?
The company whose CEO, Don Mills, strongly backs the convention centre asked the following question to a surprisingly small sample of 377 adults who live in the Halifax Regional Municipality, the area closest to the project:
As you may or may not be aware, the provincial government began construction on a new convention centre in downtown Halifax. All things considered, do you completely support, mostly support, mostly oppose or completely oppose the development of the new convention centre in downtown Halifax?
The information in the question is somewhat misleading. Yes, the Nova Scotia government gave permission to the Halifax Regional Municipality to allow a private developer to begin preliminary underground construction. But the municipal council has yet to approve the final design for the half-billion dollar, two-block complex that would include not only the convention centre, but a swanky hotel, office towers, restaurants, stores, apartments and underground parking.
And there’s no hint in the pollster’s question that municipal taxpayers would be contributing $56.4 million in construction costs for the new convention centre plus half of any operating losses.
The coalition opposing the convention centre cast their net wider asking another relatively small sample of 595 people across Nova Scotia questions that raise doubts about the project:
Justification for the second convention centre is based on a report written by the management of the present convention centre.
However, the provincial Auditor General has questioned the conclusions in the current report, and recommend (sic) that the provincial government commission an independent report before going ahead with the project to ensure that the project is a good use of taxpayers (sic) money.
Which of the following is CLOSEST to your view?
Statement 1: Halifax needs a second convention centre and there should be no delays in proceeding with its development.
Statement 2: The provincial government should get an independent second opinion before spending public money on this project.
No wonder 86% supported getting “an independent second opinion.” The question is not just a question, it’s a forceful, one-sided argument against the new convention centre.
Journalists and polls
Over the last four decades, journalists have gradually become more and more dependent on opinion polls especially in political reporting.
Instead of talking to a broad, cross-section of voters during election campaigns, for example, they now rely almost exclusively on polls to tell them how the campaign is unfolding. The pollsters themselves are frequently asked to comment on political questions and news outlets collaborate with polling companies to produce their own surveys. But, in spite of their reliance on polling, many reporters and editors have not developed the critical skills they need to judge the work of pollsters.
In the case of the polls showing majority support for and against the new Halifax convention centre, reporters and editors at the Herald accepted the competing polls and published the results without digging into the polls’ weaknesses. To be fair, the Herald did ask pollster Don Mills to critique the rival poll commissioned by the coalition opposing the convention centre. But Mills is hardly an impartial observer and no one was quoted in the Herald about the limitations of his poll.
During the federal campaign of 2006, a professor of communications studies at Toronto’s York University criticized the use of polls to predict the election outcome. Bob Hanke, who wrote a study on polling in the 2004 campaign, is quoted by the Montreal Gazette as deploring what he terms “media poll-itics.” Hanke makes it clear he’s referring to the role of pollsters in stage-managing electoral politics. He added they’re the kind of people Marshall McLuhan called “pollstergeists — the culture mind readers.”
I wish journalists would hop out of bed with the mind readers — or at the very least, start reporting critically on the pollstergeists’ stagecraft.