Polls Show Opposing Results
Calgary Mayor Naheed Nenshi might well be facing defeat in his attempt to win a third term on Monday. There have been conflicting polls on the mayoral race. The strangest thing about the conflicting polls is they all show double-digit leads. Some of the polls show Nenshi with a big lead while others show challenger Bill Smith leading. The recent history of polling misfires in elections lead to the need for further scrutiny of these polls.
The most credible of three different sets of polls show that Nenshi will be defeated by challenger Bill Smith. Mainstreet Research’s set of polls consisting of three polls that all put Nenshi up by more than ten points. The Mainstreet polls were commissioned by the Calgary Herald and Calgary Sun.
The polls use census data and are the only polls sponsored by media organizations.
Mainstreet was extremely close to the actual results in the final poll before the federal election in 2015 as they missed the Liberals final score by only half a percent and the Conservative’s final share of the vote by one-tenth of a point.
However, Mainstreet didn’t do as well in the Alberta provincial election of 2015. While they did predict an NDP majority, they missed the results of the three major parties by between two and seven points.
Nenshi Leads but Methodology Important
There are two recent polls that show Nenshi with a lead. The methodology and motivations of these polls are more questionable. A poll commissioned by a group favouring transit asked a number of questions about transit before asking about mayoral vote choice. The company who conducted the poll, Asking Canadians, offers ‘incentives’ in the form of potential prizes to poll respondents. Asking Canadians even state this particular poll is “representative but a non-random sample, therefore the margin of error is not applicable.”
Additional concerns with the Asking Canadians poll are two easily seen forms of potential bias in the poll design. The poll asked four questions about the Green line LRT. The fifth question asked who the respondent would vote for. The way in which the poll was conducted leads to what is called question order bias. The first four questions make it more likely that the respondent is going to favour the more pro-LRT candidate, in this case, Naheed Nenshi. The second potential bias is self-selection bias as the respondents chose to fill out the poll. It is reasonable to conclude people who voluntarily choose to answer a poll on the LRT would be more likely to support Nenshi.
A poll conducted as part of an academic study of mayoral races in multiple cities was released showing Nenshi with a double-digit lead. The poll was conducted by Forum research. Forum was approximately as accurate as Mainstreet in the last federal election. Forum was closer than Mainstreet on the NDP vote but did a little worse on the Conservative vote. The two firms split the difference on the Liberal vote. In the last Alber, a provincial election Forum did a little better than Mainstreet on both the Progressive Conservative and Wildrose vote but did a little worse on the NDP vote.
Since Mainstreet and Forum performed relatively equally in the two major elections in 2015 the sponsors of the poll needs to be examined. As highlighted above, the Mainstreet polls were sponsored by the two major newspapers in Calgary. The Forum poll is part of a study of eight cities and their municipal governance, The Canadian Municipal Election Study. This means that the Forum poll wasn’t commissioned to be released as part of the coverage of the mayoral race. The Forum poll has a larger margin of error, +/- 3.4% than the Mainstreet poll at +/- 2.5%.
The media always likes a horse race. Conflicting polls certainly add to that narrative. The Asking Canadians poll appears to have a number of problems and it is the least credible of the three organizations who polled the Calgary mayoral race. The Forum polls was not meant to be publicly released, and the fact that it was raises eyebrows. The Mainstreet polling feels most credible.
Recent history has proven that polling does not always produce accurate forecasts of election results. Given the way in which the conflicting polls on this race were conducted it appears that Bill Smith is poised to win the Calgary mayoral election on Monday.