A look at the numbers in election polls

Sept. 30, 2018

By Arianne Robinson

A new Mainstreet Research poll this week suggests there could be a shift in tradition on Oct. 22 when it comes to top candidates for mayor. According to one poll published less than four weeks before the election, there are four noteworthy female candidates who are rising to the challenge of unseating incumbent mayor John Tory. If frontrunner Jennifer Keesmaat and one other are able to garner enough support over the next three weeks, then the 2018 municipal election will be the first in the history of the City of Toronto to have a majority of female candidates in the top three.

Since the 2000 election, the first after the city amalgamated its smaller municipalities, there have been only four female mayoral candidates: Enza Anderson, Barbara Hall, Jane Pitfield, and Olivia Chow. None of the four came very close to winning.

  • In 2000, the first amalgamated election, incumbent Mel Lastman was elected with 483,277 votes, Tooker Gomberg had 51,111, and Enza Anderson had 13,595.
  • In 2003, David Miller was elected with 299,385 votes, John Tory had 263,189, and Barbara Hall had 63,751.
  • In 2006, incumbent David Miller was elected with 332,969 votes, Jane Pitfield had 188,932, and Stephen LeDrew had 8,078.
  • In 2010, Rob Ford was elected with 383,501 votes, George Smitherman had 289,832 votes, and Joe Pantalone had 95,482 votes.
  • In 2014, John Tory was elected with 395,124 votes, Doug Ford had 331,006 votes, and Olivia Chow had 227,003 votes.

Much of the news about Toronto politics has focused on the provincial government cutting Toronto’s council from 44 city councillors to 25. A big legal challenge in August resulted in a court decision to uphold the initial plan for 47 wards, and was followed by a higher court decision in September to go back to the legislated 25 wards while the first decision is appealed.

Unlike his predecessor Rob Ford, the incumbent mayor John Tory is not controversial or outlandish. When considering the results of a public opinion poll such as the Mainstreet Research one this week, it’s worth considering how good a predictor the poll is, the effect the poll itself will have on potential voters, and whether the change in trend (more top female mayoral candidates) is something that voters care about.

In the automated phone survey conducted by Mainstreet Research on Monday and Tuesday, participants responded to the question, “Which mayoral candidate would you vote for, or might lean toward?” and listed the five names in a randomized order as well the option to choose ‘Another Candidate,’ and ‘Undecided.’ The five candidates who were listed were Jennifer Keesmaat, Sarah Climenhaga, Saron Gebresellassi, Faith Goldy and John Tory, as chosen by the poll company’s President and CEO, Quito Maggi. The third place contender will likely be the first of its kind: a young black female lawyer named Saron Gebresellassi, local mom and community advocate Sarah Climenhaga, or the right-wing extremist Faith Goldy. Each outcome is surprising and exciting in its own right.

The results published on Wednesday reported incumbent Tory in the lead for mayor. Out of 996 who completed the survey, 48.4 per cent supported Tory. Eliminate ‘undecided’ participants, and out of the 714 survey participants 63.7 per cent chose Tory.

Keesmaat, former chief planner of the City of Toronto, is in second place. Out of all 996 who completed the survey, 20.3 per cent chose her as the candidate they would vote for or might lean toward. Eliminate ‘undecided’ participants, and out of the 714 survey participants 30.7 per cent chose Keesmaat.

However, according to the survey results, second place could theoretically be anyone’s game, with 27.4 per cent of participants surveyed choosing the ‘undecided’ option.

The results of the Mainstreet Research poll, like any poll, media, or analysis about an election must be interpreted cautiously. In the case of the results of the Sept. 26 mayoral intention, the results show higher margins of error in certain sub-samples. For example, the 18- to 34-year-old subsample has a margin of error of about 5.2 per cent, where the margin of error for the entire poll is 3.15 percent.

On its face, the poll’s results show in the tables of data that no voters in the 18- to 34-year-old category would vote for, or lean towards voting for, Goldy or Gebresellassi, despite the fact that both candidates are themselves within that age bracket. However, factor in the 5.2 per cent margin of error for this sub-sample and it is more clear that there is a chance those surveyed in that age bracket could lean towards one of these candidates, but the poll didn’t capture their intention.

Reached by phone on Friday, Maggi explains how the margin of error works. “There’s a probability that five times out of a hundred it would be with plus or minus that number [5.2 per cent],” Maggi said. “It doesn’t mean it would happen, even if we did do it a hundred times, there’s a mathematical probability that it could come in within that range.”

It is also important to note that Goldy, Gebresellassi and Climenhaga are all neck and neck in Mainstreet poll results, within one per cent of each other. University of Toronto’s Randy Besco, Postdoctoral Fellow, Department of Political Science, explains that this along with the sample size are important when considering the outcome candidates may actually have on election.

“If you have two or three candidates who are pretty close to each other, they might be within the margin of error, in which case the positions could be flipped,” Besco said, reached via phone on Friday.

“The one who looks like they’re in second could actually be in first. The margin of error is related to the sample size. So if you have a big sample then you have a small margin of error. Or conversely if you have a small sample, which a lot of these polls of specific districts are often small samples, then the margin of error is bigger. Or if you’re looking at subgroups, if you look at just women, then your sample size gets smaller.”

The sample size for the Mainstreet Research mayoral candidate poll for this week, a thousand adults, is a tiny fraction of Toronto’s nearly three million population. The poll results do break out how participants responded according to different parts of the city: 441 surveyed downtown, 174 in Etobicoke, 207 in North York, and 174 in Scarborough. The total number of calls made and the rate of non-response (and how it compares to the actual sample of participants that did choose to respond) is not reported in the poll.

Understanding the nuances of how polls work is important, given that the effects the polls themselves can have on potential voters.

“Polls can affect how people vote. They don’t necessarily or not always, but they can… [polls tend to] affect them differently depending on the size – what the poll is,” Besco said.

“If there’s a big difference between the candidates, if it’s going to be a landslide for one candidate, then that tends to lead to lower turnout because it’s clear who’s going to win and why would you bother going and voting? On the other hand if it’s a very close race, then this is motivating for people. They think ‘every vote counts and so I need to get out there and vote’.”

There can also be “bandwagon effects” that result from polls and candidate debates, Besco says. “If a candidate is ahead, people like to vote for a winner.”

Organizations across the city are holding debates. Last week, Artsvote and Global News hosted the debates with Tory, Keesmaat, Climenhaga, and Gebresellassi. Operation Black Vote Canada (OBVC) is holding a debate this coming Monday with Tory, Keesmaat, Gebresellassi, and Knia Singh.

Reached on Friday, Velma Morgan, Chair of OBVC said the choice to have two black candidates was made in August when they organized the debate, when a number of black mayoral candidates were invited to participate.

“Our mandate is not to tell people who to vote for. Our mandate is to have a lot of black candidates run. Our mandate is to get our community to go out and vote,” Morgan said. “We don’t tell them who to vote for. We provide them with the opportunity to get the information to make an informed decision.”

The opportunity to be included in the debate is of particular value to Singh; the debate will be the first time in this election that he’s had the chance to face off against the other mayoral contenders, having not been included in the two last weeks. Singh is no stranger to Toronto municipal races; he was defeated by Neethan Shan in a Scarborough by-election for city councillor in 2017.

“When candidates are included in debates and in polls and on forums or in newspapers…  being included in that group of candidates is a clear signal that this is a candidate who is a serious candidate who you might consider voting for,” Besco said.

In the case of Gebresellasi, Morgan says her profile and place in the debate is rare and important, especially for young people when they see black candidates running for mayor.

“I think it is extremely valuable to young black people because they have something to aspire to, they can say ‘well you know what, I can do that.’ Given Saron’s history, where she grew up [in social housing] there’s not many role models there, and I like when she tells her story because little girls like her who were from that community can say… [if] she’s running for mayor, then I can do that,” Morgan said by phone. “It works in our favour to have someone like her be highlighted in the mainstream that she’s running for mayor.”

Exposure to the candidates, especially those who share the same background, may also help voters decide to support them.

“There’s evidence that people are more likely to support candidates of their own ethnic or cultural group,” Besco says. “In Canada, there’s evidence that ethnic and cultural groups matter. So, for example, Chinese voters are more likely to support Chinese-Canadian candidates than white candidates.

“On the other hand, there’s no evidence that suggests this works for gender. People have done research on this and it doesn’t look like women are more likely to vote for women candidates, at least at the federal level… you see these effects for ethnicity but not for gender.”

This point is salient when looking at the results of the Mainstreet Research poll that did ask participants to identify if they identify as a man or woman, but not what ethnic or cultural group they belong.

As for how much a poll actually matters to the outcome of an election, Besco says “it matters some… there are other things that matter more.” He gives examples: incumbency, name recognition, the individual’s interest in politics, how strong and clear their opinions and beliefs are. “There are effects from polls, but they’re medium-sized, they’re not gigantic,” Besco said.

“If you’re really interested in politics then you tend to have strong and clear opinions about who you like and what you believe already, and so therefore you’re also not very likely to be affected by the polls.”

Beso also notes the difficulty voters have just knowing candidates’ platforms in an election where they are not tied to a political party the voter can easily identify. For example, the mayor of the city, who was previously the leader of the Ontario Progressive Conservative Party between 2004 and 2009, has built his re-election platform with endorsements from former and current members of the provincial and federal Liberal and Conservative parties.

“There are lots of people who know what John Tory [and other candidates] stands for, but in general, most people don’t pay a lot of attention to politics, and so the fact that he doesn’t have a party label means it sort of takes more effort to figure out what they stand for and which side of the spectrum they are.

“It’s not that most people won’t know if he’s on the left of right, but a good chunk of people won’t, and certainly more [won’t know] than at the provincial or federal level.”