THE NUMBERS GAME
A RECORD NUMBER OF OPINION POLLS IN THE CAMPAIGN HAS LED EXPERTS TO ASK IF THEY SWAY VOTERS
The poll results caused grave concerns in Conservative headquarters in Ottawa. On Nov. 7, two weeks before election day, the Gallup poll reported that 43 per cent of decided voters favored the Liberals, 31 per cent chose the Conservatives and 22 per cent supported the New Democratic Party. It indicated an astonishing, shockingly swift turnaround in the two leading parties’ standings, and, as the bleak media accounts spread across the nation, many Conservative riding workers slumped into depression. Then, Tory pollster Allan Gregg, president of Decima Research Ltd. in Toronto, personally reassured Prime Minister Brian Mulroney that his soundings were far more favorable. Conservative officials hastily distributed a memo to all regional headquarters, claiming that Gallup’s poll was “wrong.” The memo added that “our own polls show that this election is very much a horse race: we and the Liberals are neck and neck on a national basis. Tell your workers not to panic.” Looking back on the uproar, a senior Tory concluded last week, “Political parties are now poll junkies.” And indeed, there is evidence that ordinary Canadians as well have become poll junkies, avid consumers of their own opinions and perhaps unwitting victims of their own preferences.
During the seven-week campaign, news organizations produced 26 national polls to show voters’ party preferences—up from 12 during the 1984 contest. On top of that, the Tories, Liberals and New Democrats conducted their own heavy polling, producing frequent snapshots of Canadian public opinion on everything
from the leaders’ images to party standings. The result, for the average voter, was a blizzard of widely trumpeted and often contradictory public samplings and leaked private surveys. As Vancouver housewife Pamela MacKenzie, 65, told Maclean’s: “It is very confusing. There are too many, and we do not know where we stand.”
The effect of those polls on the political process was unclear. Most experts agreed that polls affect the performance of party workers: high standings delight campaign loyalists, driving them to greater efforts; low standings discourage them. But no one has measured the subtle effect that polls have on the campaign
strategists themselves, on media coverage of the politicians and on voter behavior at the ballot box. Lome Bozinoff, vice-president of Toronto-based Gallup Canada Inc., claimed that polls “do not have much effect in terms of influencing opinion or changing the election.” In contrast, Queen’s University political scientist George Perlin contended that polls do influence the campaign, allowing voters to make strategic choices, voting to defeat a front-runner, perhaps, instead of supporting their party of choice. Said Perlin: “In a sense, it is impossible to test. But I think that survey results can influence people’s behavior.”
That suspicion focused attention on the pro-
Iteration of public opinion surveys during the 1988 campaign. As a result, there have been calls for controls, or even bans, on the publication of polls. Experienced pollsters such as Martin Goldfarb, chairman of Goldfarb Consultants of Toronto, say that governments should create an agency to license pollsters. As well, many pollsters and political scientists have concluded that both journalists and voters must become far more knowledgeable about the differences among polls. As Michael Adams, president of Environics Research Group Ltd., told Maclean ’s: “The Inuit have 100 names for snow. Voters must end up having 100 names for dealing with polls, phrases like ‘random sampling’ and ‘margin of error.’ ” Added Adams: “We must see a more critical approach to polls. We must learn to take some more seriously than others.”
But for many Canadians, the problem in the 1988 campaign was simply keeping up with the blizzard of poll information. The Gallup organization did seven polls for eight newspaper clients, including The Toronto Star. Goldfarb Consultants, the pollster for the federal Liberals, also did three polls for Baton Broadcasting Inc. Environics conducted four polls for The Globe and Mail. Angus Reid Associates of Winnipeg did four polls for the 22 newspapers that subscribe to Southam News, including Montreal’s Gazette. Insight Canada Research Inc. of Toronto conducted six polls for the CTV Television Network Ltd., and Canadian Facts of Toronto did two polls for the CBC.
Among the parties, the most ambitious pollsters were the Conservatives. Decima sampled a total of 18,000 voters during the campaign. Questions ranged from simple party preference to such inquiries as “Who has the best team?” As well,
Decima employed sampling groups of representative voters to test the effect of Conservative advertising and its themes against the Liberals’ ads. For the Liberals, Goldfarb conducted daily polls of 300 voters across the nation and sampled 12 so-called swing ridings, which have a history of electing members from the party that is going to form the government. Pollsters asked not only about the voters’ preferences but about their depth of commitment.
They also asked strategic questions: how effective was the business community’s pro-free-trade intervention? how did the voters react when the Prime Minister dismissed Liberal Leader John Turner as a “liar?” For the New Democrats, veteran U.S. pollster Victor Fingerhut, of Fingerhut/Madison
Opinion Research in Washington, D.C., analysed the raw data from surveys conducted in key regions and national polls.
The polling industry dates from the 1930s. But its pivotal role in the federal political process began in the 1974 election when parties, particularly the Liberals, started to measure the depth of feeling about issues and about the values that were implicit in a voter’s selection. The process became more sophisticated in subsequent elections as the parties and their pollsters acquired faster and more efficient computers to analyse the complicated data. But strategists from all three parties emphasized that a poll still cannot prescribe a remedy for a political dilemma. Said Conservative operations director Harry Near: “They tell you what is, not what will be. You have to take the ‘is’ and extrapolate—and that is where judgment comes in.”
During the campaign, senior Conservatives knew from their own polling that the public had considerable confidence in Mulroney as a manager—and suspicions about Turner’s motives for attacking the free trade deal. As a result, when Liberal support soared and Conservative support declined after the French and English televised debates on Oct. 24 and 25, new Conservative ads cast doubts on the ability of the Liberals to put together a strong cabinet. At the same time, the Tory advertisements accused the Liberal leader of campaigning against free trade simply to protect his own job, not those of Canadians.
Liberal strategists said their polls indicated
that the voters were suspicious because Turner had not put a price tag on his 40-point election program. Turner planned to have announced the costing—$18 billion over five years—late last week, but because of the poll findings the party moved the announcement forward to the beginning of the week. Still, the Liberals did not always have the resources to act on their poll results, which also showed that the business community’s pro-free-trade campaign was extremely effective. The party wanted to counterattack but, said a senior strategist, “we did not have the firepower.”
But even good polls do not guarantee that parties will produce an effective strategy. NDP polls taken before the campaign indicated that the party would not gain many votes with an anti-free-trade platform. The reason: voters regarded free trade as an economic and managerial issue—and they did not believe strongly in the party’s ability to manage the economy. As a result, the New Democrats largely ignored free trade in the first four weeks of the campaign. But that, apparently, was the most important issue to the voters, and party support diminished.
As party polling became more intense, the media also sought a snapshot of the Canadian electorate. But many news organizations were uneasily aware that their polls could eclipse the substance of the campaign. The Toronto Star regularly ran its Gallup polls on the front page—and generally relegated other polls to the inside pages. “Otherwise,” said managing editor Ian Urquhart, “it becomes horse-race journalism.” The Globe and Mail placed only the Environics polls on the front page, always including a lengthy explanation of methodology that indicated the margin of error, the sample size and the polling technique used. As well, the pollsters themselves wrote the descriptions of their poll results. The Globe and Mail also stopped its polling before some other media outlets did. Said managing editor Geoffrey Stevens: “We do not poll in the last two weeks. I subscribe to the view that a published poll can affect voters.”
The CBC also refused to release results of its own polls during the last 10 days of the campaign. Said television news director William Morgan: “[We] fear that it will affect people’s voting behavior.” But Morgan defended the use of public polls: “Polls are a step forward for the democratic process, and I do not see why the public should not have them.”
Still, polls can certainly affect media coverage, creating the image of winners and losers. In Winnipeg, John Drabble, the senior news producer of CBWT TV’s 24 Hours news program, told Maclean ’s that the polls affected the treatment of a party—not the actual amount of television air time devoted to a party. “We have treated the Liberal party much more seriously than we would have if the polls had shown the party to be trailing,” he said. 24 Hours had the resources to do in-depth profiles on only five of Manitoba’s 14 ridings: poll standings were a factor in that selection. As well, the program did more than a dozen indepth stories on the free trade pact. Said
Charting the election
How the pollsters tracked the parties
Drabble: “I do not think that we would have done that many if the polls had not shown us that it was an issue.”
Yet no one knows the full effect of the surveys, and the subsequent media coverage, on the voters. Indeed, a Gallup poll on that subject, released on Nov. 4, indicated that only 11 per cent of Canadians said that polls influenced their voting decision, while 85 per cent insisted that the polls made no difference. Said Bozinoff:
“We did the poll on polling because we knew that we would get these kinds of criticisms. We have to take Canadians at their word.”
But other pollsters and political scientists suspect that there is a wider—although subtle—influence. Winnipeg pollster Angus Reid said that polls during the 1988 campaign may have hurt the third-place NDP. He added,
“Given the impact of the free trade issue, some supporters might have moved in the direction of the Liberals because they perceived them as the party that could unseat Brian Mulroney and kill the free trade deal.” Political scientist Perlin added that polls can create a bandwagon effect. He contended that some voters hardened their stand against free trade when polls during the third week of the campaign picked up growing opposition to the deal. Declared Perlin: “For people who were wavering, the polls might have made a difference.”
If the polls do, in fact, make a difference, Canadians will have to find ways of using them constructively. Political scientist Agar Adamson at Nova Scotia’s Acadia University said that Canada should adopt France’s practice of banning the publication of polls during the last 10 days of the campaign.
“Polls become a self-fulfilling prophecy,” he said. For his part, Goldfarb recommended that universities develop a specialized degree for pollsters. Governments, in turn, should create a commission that would license only degree-holding pollsters, he said.
Most experts say that, asa first step, the public and the media should learn how to read polls in order to keep them in perspective. They should also examine the methodology of each pollster.
In its last campaign poll, released on Nov. 19, Gallup conducted 2,097 personal in-
terviews and 1,970 telephone interviews— believed to be the largest public opinion survey ever published in Canada. But throughout the rest of the campaign, Gallup reached 1,000 voters per poll, occasionally alternating polls conducted through at-home interviews with polls conducted by telephone. According to generally accepted statistical theory, that should produce a margin of error of four per-
centage points above or below each resulting figure in 19 out of 20 polls. As Carleton University journalism professors Anthony Westell and Alan Frizzell explained in The Canadian General Election of 1984, “This is merely an estimate that if the poll were repeated over and over again, the results would be the same within a range of four percentage points almost all of the time.”
Even given Gallup’s margin of error, its Nov. 7 poll, which disturbed the Conservatives with its indication of a Liberal surge, surprised political strategists from all three parties. But Bozinoff insisted that it was not wrong. He said that Gallup had surveyed voters throughout the first week of November— even though the poll came out at the start of the second week. So competing polls, taken in a subsequent week and released after Nov. 7, which showed that the two parties were neck and neck,
simply reflected the volatile mood of the electorate. Added Bozinoff: “Our poll, which reflected the previous week’s happenings, caught the crest of an emotional Liberal wave after the debate.”
Still, the margin of error is usually lower than Gallup’s four per cent in other national polls. One Canadian Facts poll for the CBC reached 2,199 voters by telephone, for a margin of error of plus or minus 2.24 per cent. A second Canadian Facts poll reached 2,467 voters, for a margin of error of plus or minus 2.09 per cent. The margins of error in the other national polls ranged between Gallup’s and Canadian Facts’.
At the same time, most experts said that it is essential to separate the numbers from the pollsters’ interpretation of the numbers. In July, 1987, Angus Reid predicted that the NDP’s high standing in the polls—then 40 per cent of decided voters—indicated that it would win at least 100 seats in the next election. As ^ Reid told The Toronto Star at g the time: “We are not dealing I with a one-day wonder. We z are dealing with a party that g will gain up to 100 seats in I the next election, and maybe “ more.” Citing that remark, I journalism professor Frizzell g said, “Problems occur when a o pollster goes beyond what his “■ numbers tell him.”
Analysed properly, the polls might be more useful generally. Insight Canada president Michael Marzolini contended that polls enabled the average voter to understand why parties are changing their campaign strategies. “The parties are obviously trying to influence the electorate,” Marzolini reasoned. “Why should the voters not be aware of it?”
In the recently published book Marching to a Different Drummer, Goldfarb and Liberal strategist Thomas Axworthy argued that Canada is moving from a “representative democracy,” where MPs looked after voters’ interests, to an “information democracy,” where Canadians will be better able to understand and judge political decisions. “Voters today, by means of poll-generated information, are far better critics of the political decision-making process than those who came before,” they wrote. “Today’s politicians have access to very little information that is not also available to the public.” It is an assessment that poll-weary voters can weigh in the wake of the campaign of 1988.
MARY JANIGAN with MARC CLARK, PAUL KAIHLA and HILARY MACKENZIE with the leaders’ campaigns, GLEN ALLEN in Halifax, ROSS LAVER and THERESA TEDESCO in Ottawa and HAL QUINN in Vancouver