(10/07/11) -

Provincial Elections: What We Saw and How We Acted

This week, we were the only Canadian pollster to issue a prediction in the two provinces that held elections: Manitoba and Ontario.

Angus Reid – This week marked a special milestone in the online polling program that my company initiated in 2007, when we covered our 19th and 20th elections in North America. I began my career as a pollster in the days of door-to-door interviewing more than four decades ago, and I lived through the transition from the face-to-face meeting to the telephone call centre. In the early years of the 21st Century, my company began to develop the tools that would allow for accurate electoral predictions to be issued through online research.

From the start, the new technology had its critics, who focused on issues such as Internet access, random selection and the uncanny assumption that we were merely replicating the thousands of “quickie polls” where any person can participate as many times as he or she pleases. Our approach was simple: ensure that the make-up of a constituency is carefully represented within the sample, and issue invitations for particular surveys to a randomly selected portion of the online panel.

Our record has proven that online panels—when combined with proper sampling techniques, thorough discipline in questionnaire writing and development, and an openness to work until the final moments of a campaign—can be used to accurately predict the views of the Canadian electorate. In fact, no other company in the country comes close to matching our record of both participation and accuracy.

Since 2007, we have covered 13 different federal and provincial electoral processes in Canada. In 10 of these elections (Quebec 2007, Manitoba 2007, Saskatchewan 2007, Canada 2008, Quebec 2008, British Columbia 2009, Nova Scotia 2009, Canada 2011, BC Referendum 2011 and Manitoba 2011), we have provided the most accurate prediction of all pollsters relying exclusively on our online methodology. In the remaining three elections (Ontario 2007, Alberta 2008 and Ontario 2011), only one of the 12 contending parties was called outside the margin of error.

This week, we were the only Canadian pollster to issue a prediction in the two provinces that held elections: Manitoba and Ontario. In Manitoba, our survey showed a high level of support for the governing New Democratic Party (NDP) in Winnipeg, which would translate into a first place finish. We also saw a slight improvement for the provincial Greens, and the Liberal Party losing one quarter of its voters from the 2007 contest. When all the votes were tallied, our prediction for three of the four contending parties was exact, with a difference of only one point for the remaining party.

Ontario provided a different challenge, due to the complexity of the campaign. In previous occasions, our company has gone into field as late as possible to review whether some shifts occur as voters ponder their choices in the final hours of a campaign. In 2008, we were the only pollster to conduct interviews on Thanksgiving Monday, publishing the results on our website a few hours before the blackout period for federal election polls came into effect. In the end, our prediction was the closest to the final outcome.

A similar situation ensued this week in Ontario. Our first survey on the final week of the election showed a very tight race. There were two countervailing forces at play: a highly motivated Tory base hitting up against a Liberal Party that had been gaining noticeable momentum in the 416 area. We therefore decided to go into field with a final poll to ascertain which, if any, of those forces might give either party an edge on election day. Our final survey—the last one published before the Elections Ontario blackout—featured online interviews with 747 absolutely certain Ontario voters, and disclosed that the Liberal momentum had continued to increase as the week wore on, giving the Grits a slight but significant edge. Support for the governing party across Ontario was pegged at 37 per cent and the Progressive Conservatives were pushed to second place. This exact scenario materialized on election night.

Over the past four years, we have witnessed many changes in the Canadian polling business. For starters, more companies are embracing online tools for data collection—a practice that many used to despise not too long ago—including five of the nine firms that covered the 2011 Canadian federal election. However, advances in technology are not enough to ensure accuracy. We are firm believers in taking the pulse of Canadians on a daily basis, not only on specific occasions or just in a few of the country’s provinces. We have made it a habit to look at Canada through unique syndicated studies and assessments of provincial leaders that go beyond the common practice of amalgamating the Prairies into a single entity. And we made the conscious decision to spend more time looking at designing new and innovative question types and focusing on the quality, representativeness and health of our online panel. Rather than partake in the unsubstantiated posturing and pointless bickering that has sadly enveloped much of the industry, ours is a more forward-thinking approach that stresses continuous improvement and thought leadership in establishing best practices for the online polling industry.