How Polls Influenced the Outcome of the 2011 Canadian Federal Election

Canadians gave Stephen Harper’s Conservatives a majority government in possibly the strangest election in Canadian electoral history.  Conservatives ended up with 167 seats, 24 more than before, while both Liberals and Bloc Québecois were demolished with their respective leaders, Michael Ignatieff and Gilles Duceppe, resigning in the aftermath.  The Liberals lost over half their seats ending up with 34, while the Bloc Québecois lost almost all their seats.

The big winner was the NDP.  It gained 102 seats nationally, almost 3 times their seat count in the 2008 election and the highest total ever achieved by the NDP party.  This stunning performance also made them the official opposition to the Conservatives.  Even more remarkable was their achievement in Québec.  In winning 58 seats in Québec (up from only one in 2008), the NDP crushed the Bloc Québecois leaving it with an embarrassing four seats from their previous majority of 49.

The additional 24 seats that lifted the Conservatives from a minority to a majority status represent an increased seat count of 17%.  Remarkably, Elections Canada data show the popular vote for Conservatives increased by only 2% from 2008 (from 37.6% to 39.6% in 2011).

These extraordinary results were not predicted by the polls.

Poll predictions

While showing Conservatives as the most popular party, the polls predicted a minority, not a majority government.  Four polls completed just prior to Election Day showed Conservative support ranging from 37.1% to 33.9%, with a mean value of 35.8%.  The latter figure was below the Conservatives popular vote in 2008, leading the experts to the prediction of a minority government.

Secondly, while showing reduced popular support for the Liberals in Ontario, none of the polls predicted the utter collapse of the party in Ontario, and in particular Toronto, their historic stronghold.  It is this collapse that gave Harper his majority.

Thirdly, while the polls accurately captured the rise of the NDP wave and there was speculation this would translate into “a lot of seats”, nobody was predicting that the party, an anglophone federalist party, would effectively wipe out the sovereigntist Bloc in Québec.  Historically, NDP support during a campaign would dissipate by Election Day.

The unexpected massive and sudden rise of NDP popularity left both the pundits and the public scratching their heads.  What was the tectonic event that triggered the NDP wave?

Was there an issue during the campaign that stirred the public and attracted support for the NDP?  Not that anyone can recall.  The campaign was notable for the absence of any issue debate that might distinguish the parties.

In the absence of issues, the sole discriminating factor was the public’s perception of the party leaders… and a mechanism that catapulted one of them into public favor.

The public’s perception of party leaders

According to polls, the leader Canadians trusted most was NDP’s Jack Layton.  A Nanos poll found that Jack scored with 33% of Canadians compared to 24% for Stephen Harper and only 11% for Michael Ignatieff. An Angus Reid poll echoed these results, showing 49% of Canadians were satisfied with how Jack was doing his job, compared to 31% for Stephen and only 22% for Michael.

Walking with a cane due to a recent hip surgery and recovering from a bout of prostate cancer, pundits at the start of the campaign questioned whether Jack had the physical stamina to run a campaign and whether the need for a cane would portray him as weak.  The public saw him quite differently.  Here was a man who would fight for a health system that gave him a future.  When he spoke of his support for universal health care in Canada, no one doubted his sincerity.  It wasn’t political.  It was personal.  When he raised his cane in front of a cheering public, the cane became a shibboleth.  It wasn’t an instrument of weakness; it was a weapon of battle.  You can’t buy better advertising than that.

The mechanism for the NDP wave

So how did this trusted leader become the rage in Québec and beyond?

The love affair between Jack and Canadian voters started on April 12, the night of the leaders English language debate.  A before and after poll by Angus Reid indicated that while 42% thought that Stephen Harper had won (versus 25% for Jack Layton and 23% for Michael Ignatieff), 55% of respondents said their impression of Jack Layton had improved as a consequence of the debate.  Perhaps more importantly, on the following night when the leaders debated in French, Léger Marketing reported that while 41% felt that Gilles Duceppe had won, Jack Layton came in second with 20% followed by Michael Ignatieff with 15% and Stephen Harper with 11%.  These numbers came at a time when Jack’s party had a national support of about 15% of Canadian voters compared to about 40% for Stephen Harper’s Conservatives.

However it wasn’t until about a week later on about April 19, after the public and the press had a chance to digest this event, that polls began to show a substantial uptick for the NDP in Québec.  At the same time the polls were reporting that Bloc Québecois popularity was continuing to deteriorate.  This upward trend caught the attention of the press, particularly in Québec, but nobody could say at the time if this was a transient effect or something significant.

For many voters, seeing and hearing about this uptick was an important signal.  It provided confirmation that there were other voters who had decided that the NDP was an acceptable alternative to the Bloc Québecois.  In the minds of many, it validated their up to this point private decision to support the party.

As more voters were drawn into the NDP camp, their increasing numbers were measured by numerous polls that were tracking party support.  This created a feedback loop in which press stories about the rising popularity of the NDP emboldened increasing numbers of Québec voters to cast their support in favor of the NDP.  It is this feedback loop involving the interaction of the polls, the press and the public that amplified the initial trickle of NDP support to a veritable tsunami that stunned the province.

To put it bluntly, without the timely, positive NDP polls, there would not have been this massive NDP wave.

In turning to the NDP, Québec voters not only revealed their willingness to trust Jack Layton and, by association only, his young band of politically inexperienced and unknown candidates, they were also telling the traditional recipients of their vote — the Bloc Québecois, the Conservatives and the Liberals — that they wanted a new vision for Québec, a vision they did not trust these parties to deliver.  In this context it’s worth noting that while the Bloc Québecois lost all but four of their 49 seats, both the Liberals and Conservatives lost about half of their Québec seats.  It was a truly bold, risky move by Québec voters who once again showed English Canada just how politically astute and sophisticated they are.

The effects of this tsunami were not restricted to Québec where the NDP’s popular vote increased to a whopping 43% in 2011 election, up from a mere 12% in the 2008 election.

Substantial increases in the NDP popular vote were found in most other provinces as well. In Ontario, the NDP vote increased from 18% to 26%; in British Columbia, from 26% to 33%; in New Brunswick, from 22% to 30%; and in Saskatchewan, from 26% to 32%.  The orange wave even affected the bastion of conservatism Alberta, where NDP support increased from 13% to 17%.  This tsunami created waves everywhere.

Not only did the magnitude of the NDP wave reveal the power of this feedback process to move public opinion, it also refuted the claim of some experts that the conduct and dissemination of polling information does not influence election results.

The Spiral of Silence

The election results are in fact consistent with the theory of political scientist Elisabeth Noelle-Neumann of how mass media impacts public opinion.  In her book “The Spiral of Silence: Public Opinion – Our Social Skin” she proposes the idea that there is a gap between the opinion held by an individual and the social consensus on that opinion.  She theorizes that most people have a general sense of how far removed their opinion is from consensus.  If they feel the gap is large, they tend to keep the opinion to themselves to avoid potential social isolation.  If the gap is small, social isolation is not an issue and the opinion is communicated publicly.

The latter was the case in Québec where increasing public support for the NDP was reflected in broad-based press coverage and in increasing poll numbers.  By the end of the campaign, the poll numbers were close to the popular vote of the NDP in Québec.

In Ontario however, this gap was large as the press was predicting some Liberal weakness but definitely not a total collapse.  Also there was no prediction of a Conservative sweep.  On Election Day, the Conservative popular vote was 44.4%, substantially higher than the 39-41% figures reported by final polls.  Also, most of these polls overestimated Liberal and NDP support in the province compared to the popular vote.

Again this is consistent with the spiral of silence theory whereby some voters decided to switch to the Conservative side but were not prepared to admit this in polls since it was at odds with the prevailing media opinion.

Polls can and do influence election results, mainly because they can create media driven narratives that influence voter decisions, as was the case in this election.

While the idea of polls influencing election results may seem farfetched for many Canadians, in the US there is no such naïveté.  Partisan polling is a reality in many US state and federal elections.  Polling companies run by Republican pollsters have been found to produce results with a Republican bias, and ditto for Democrats.  For example, Democratic Senate Majority Leader Harry Reid’s re-election was almost derailed by polls that during the campaign showed him consistently trailing his Republican, Tea Party supported adversary Sharron Angle.  While on Election Day Harry Reid emerged victorious, the biased polls managed to create a false narrative that his competitor had much stronger voter support than she actually did.  In the minds of Nevada voters, this created more legitimacy to Sharron Angle as a viable alternative to Harry Reid. In this case the polls were wrong.

Clearly, polls can influence election results, either for the good or the bad.  For pollsters this places more responsibility on their shoulders to ensure that their polls are accurate and are not responsible for the creation of false narratives that can negatively influence results on Election Day.  Given their sketchy record on accuracy or even if, according to the spiral of silence theory, accuracy is possible, one wonders whether the pollsters can rise to the challenge.

About Oleh Iwanyshyn

Oleh Iwanyshyn has been involved professionally with surveys from the mid-70s when he started as a methodologist at the Institute for Behavioral Research at York University. Later, while at the Canadian Broadcasting Corporation his focus shifted to media and election surveys. He now runs his own survey research company, ViewStats Research, established in 1997 and specializing in online surveys. Public opinion surveys is an essential communication tool between a democratic society and its leaders. Unfortunately, surveys can be very easily manipulated. Revealing such manipulations and their consequences is the raison d'être of the poll stuff blog.
This entry was posted in Polling and tagged , , , , , , , , , , , , , , , , , , . Bookmark the permalink.

3 Responses to How Polls Influenced the Outcome of the 2011 Canadian Federal Election

  1. Pingback: Do polls work? | iPolitics

  2. Pingback: Do Polls Work? | poll stuff

  3. Pingback: Searching for Accuracy in Election Predictions: More Regulation of Polls or More Competition? | poll stuff

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.