Voter turnout has declined in Canadian elections, dropping from nearly 80 per cent in the 1950s to around 60 per cent in the most recent elections. This decline makes it more important to know who is going to turn out to vote, since who is and isn't going to show up has major implications for which parties get into power and how pollsters survey the population.
Who is the likely voter?
Research into factors associated with turning out to vote uniformly point to three key factors that affect turnout. The first is age. According to a Statistics Canada study of the 2011 election, 81.5 per cent of Canadians aged 65 to 74 turned out to vote, while just 49.7 per cent of those 18 to 24 did the same.
The second, education, was another important factor in 2011, according to the same study. For almost every age group, the more well-educated the respondent, the more likely they were to vote. This was especially true for younger Canadians: nearly two-thirds of those 18 to 24 with a university degree turned out to vote in 2011, while just over one-third of those in the same age group but with less than a high-school diploma did so. The effect of education is so large that a university-educated young person and a high-school-educated Canadian aged 45 to 54 have the same likelihood of going to the polls.
Being economically well-off and having more free time also explained why Canadians went to the polls. Employed Canadians were nine percentage points more likely to vote than their unemployed neighbours, while single parents with children under the age of five were less than half as likely to turn out to vote as a married Canadian with no small children (36 per cent turnout against 74 per cent).
How does this affect elections?
The fact that older Canadians turn out to vote at a higher rate than younger ones is an advantage for the Conservatives. In the latest Ipsos poll, which had the Conservatives just one point ahead of the Liberals nationwide, it was reported that for Canadians over the age of 55, the governing party had a lead of 8 points, but among those 18 to 34, they trailed in third place, 17 points behind the Liberals and 6 behind the New Democrats. This is broadly consistent with most other studies.
Voter turnout increasing with education has the opposite effect on age. A recent EKOS poll suggested that respondents with higher levels of education are less likely to vote Conservative and more likely to vote for either the Liberals or NDP.
Differences in economic status have a smaller effect on the election outcomes than age. While according to the same Ipsos poll the Conservatives are doing worse than average among those making less than $40,000 a year and better among those making more than $100,000, the gap is much smaller than it is for age.
What about polls?
When you read polls that look at both likely and eligible voters, there is sometimes a strong difference between the two. Making statements about eligible voters requires polling firms to sample among all those eligible to vote. On the other hand, likely voter models vary from pollster to pollster, but generally involve asking a series of questions about how engaged the respondent is with the current election, as well as about their past history of voting, since past democratic participation is a great predictor of future participation.
Given the knowledge that older voters are both more likely to vote and more likely to vote Conservative, it would seem that developing a good eligible voter model for Canada should be a high priority, since a tie in a poll of eligible voters would suggest a Conservative lead among likely voters.
Yet, it is important not to place too much faith in likely voter models, since they have often led pollsters (and the public) astray. For example, in 2012, Gallup's final U.S. presidential poll called a result of Mitt Romney 49 per cent, Barack Obama 48 per cent (compared to the actual result which saw Mr. Obama win 51 to 47). While this might have been the result of just sampling error alone, Gallup's own investigation of their data found that their likely voter model rated the chances of white voters higher than they were.
Many pollsters have reported that, with the benefit of hindsight, sampling eligible voters is often better at calling the result than weighting the sample for eligible voters. This failure of eligible voter models is also often the cited reason behind the failure of polls to call the British Columbia provincial election of 2013 correctly.
Eligible voter models are also generally backward-looking rather than forward-looking. They tend to get their insights from how different demographic groups voted previously, and then apply those findings to the current election. However, just as voter intentions can be swayed by an election campaign, so too can voter turnout among demographic groups. Ultimately, while paying attention to factors like voter intention by age can be helpful in figuring out what is going on, it must also be done with caution, given recent problems with likely voter models.
Paul Fairie is a political scientist in Calgary who designed The Globe's Election Forecast.