Discrepancy between Elections Canada, StatsCan reports likely due to social desirability bias

Of note. The limits of self-reporting and social bias:

Despite an 11-percentage-point discrepancy between self-reported and actual voter turnout, a recent Statistics Canada survey still provides valuable information on the electorate and voting trends, experts say.

The StatsCan survey, which relies on self-reporting, collected data by adding five election-related questions to the 2019 Labour Force Survey, which is distributed to approximately 56,000 households. The survey does not include Indigenous people living on reserve, full-time members of the Canadian armed forces, prisoners, and households in remote areas with very low population density.

Because the survey misses certain groups, it actually looks more like the electorate than the entire population, Richard Johnston, professor at the University of British Columbia and Canada Research Chair in public opinion, elections, and representation.

“The people who are missed by the survey tend to be the sort of people who are generally socially disconnected and are least likely to be subject to kinds of social pressure that get people to the polls,” Prof. Johnston said.

There was a similar gap between the reported turnout numbers after the 2015 election. Actual turnout in 2015, as reported by Elections Canada, was 68 per cent. The StatsCan post-2015 election report had self-reported turnout at 77 per cent, a difference of nine percentage points.

The data in the Elections Canada post-election survey is more accurate, said Lydia Miljan, University of Windsor political science professor, as the StatsCan survey relies on self-reporting. Prof. Miljan said the social desirability bias explains much of the discrepancy between the StatsCan survey and Elections Canada report.

“It’s not socially desirable to say, ‘I don’t vote’, so that’s why you always end up having a higher rate of self-reporting as opposed to what’s actually happening,” Prof. Miljan said.

Despite the discrepancy, Prof. Miljan said StatsCan’s report is valuable for the details it offers on demographic splits, which can “give a good trend analysis from one election to another.”

“If you’re trying to get inside the guts of social, psychological, or political differences in turnouts, these surveys are pretty good. It’s just that the baseline is too high,” Prof. Johnston said.

No interest in politics still top reason

A disinterest in politics was the top reason voters, in every age group except non-voters 75 years and older, cited for skipping out on the 2019 federal election, at 35 per cent, StatsCan’s report suggested. The same reason topped the list in the 2015 and 2011 federal elections. No data exists for prior elections, according to the agency, as the survey was inaugurated after the 2011 election. In 2019, the surveyed showed 23 per cent of Canadians did not vote.

Non-voters between 55 and 64 were the most likely to cite no interest in politics as the reason for not voting, at 38 per cent. Non-voters between 18 and 24, and 25 and 34, commonly thought of as the least-engaged age groups, were actually less likely than older voters to cite no interest.

Interest in politics appears to sharply increase between those who are 65 to 74 years old and those 75 years old and older. For voters between 65 and 74, 34 per cent said they lack sufficient interest, but that number drops to 21 per cent for voters 75 and up.

Women also appear to be generally more interested in politics than men, with 32 per cent of women and 37 per cent of men reporting a lack of interest as the prime reason for staying home.

Among the provinces, voters in Newfoundland and Labrador, Nova Scotia, and Quebec were the most likely to say they lack an interest in politics. Quebecers appear to be the most disengaged, with 41 per cent lacking an interest, compared to 40 per cent in Nova Scotia and 39 per cent in Newfoundland and Labrador.

Other voters reported they were too busy to vote, making it the second most-common reason at 22 per cent, which is also consistent across the three elections surveyed.

Younger voters were much more likely to cite being too busy than older voters. Voters between 25 and 34 years old were the most likely to be too busy, with 30 per cent reporting it as their reason. As voters get older, it drops precipitously. Just 16 per cent of voters between 55 and 64, seven per cent between 65 and 74, and four per cent older than 75 report being too busy to vote. Discrepancies in gender are virtually nonexistent, with 22 per cent of men and 21 per cent of women reporting being too busy.

The third most-common reason was suffering from an illness or disability. In 2019, 13 per cent of non-voters said an illness or disability prevented them from voting, up from 12 per cent in 2015 and nine per cent in 2011.

In a supplementary post-2015 report from Elections Canada that broke down turnout by demographics, youth voter turnout was actually 57 per cent. A similar supplementary report for the 2019 election has not yet been released.

Self-reported turnout amongst voters aged 55 and up has held steady around 80 per cent over the past three elections, but self-reported turnout amongst those 44 and younger jumped at least 10 points between 2011 and 2015, and remained high for the 2019 election.

“In 2015, there was a sort of social movement quality to the Trudeau victory, and the evidence suggests that the turnout surge in 2015 was a surge of younger people looking for a new kind of politics. And a lot of those younger people stuck around in 2019,” Prof. Johnston said.

Newfoundland and Labrador had the lowest self-reported turnout in the 2011, 2015, and 2019 general elections. In 2019, provincial turnout was 68 per cent, seven points lower than Manitoba at 75 per cent, the province with the second lowest turnout rate in 2019. Manitoba faced severe storms during advance polling time, causing evacuations, power outages, road closures, flooding, and some polling stations to close. Elections Canada set up an additional polling station at the University of Winnipeg for voters from four electoral districts, and teamed up with the Canadian Red Cross to transport voters. Elections Canada reported that 270 people used this option. Emergency workers helping with disaster response were also provided with additional polling stations, and 592 voted at the additional stations.

Prince Edward Island had the highest turnout in the 2011, 2015, and 2019 elections, topping 80 per cent each time. In 2019, turnout was 82 per cent, down from 86 per cent in 2015. Prof. Miljan and Prof. Johnston said P.E.I is usually the most turnout-heavy province in both federal and provincial elections.

Despite P.E.I.’s high turnout, the rate actually decreased the most between the 2015 and 2019 elections, from 86 to 22 per cent. Quebec, from 78 to 76 per cent, and British Columbia, from 79 to 76 per cent, also had turnout drops. Turnout largely remained the same in the remaining provinces.

Prof. Johnston provided an anecdotal explanation for the Atlantic provinces turnout numbers. He said the social pressure to vote in P.E.I is potentially higher given the population density, 25.1 people per square kilometre, which is the highest in the country. Newfoundland and Labrador is the province with the lowest population density, at 1.4 people per square kilometre.

“There’s a sense in which someone from P.E.I is going to feel social pressure to turn out because they see each other more regularly and they know each other. There are social networks that reinforce participation,” Prof. Johnston said.

Turnout increased the most between 2015 and 2019 in Alberta and Saskatchewan. Alberta turnout rose from 77 to 80 per cent, and Saskatchewan from 77 to 81 per cent. 

Prof. Miljan suggested one reason for increased turnout in Western Canada was due to frustration with the Trudeau government.

“When people don’t vote, it means they’re pretty happy with the regime and they don’t feel it [their vote] matters one way or another,” she said. This theory suggests that Western voters are “not happy with the regime and they really wanted to make sure their voices were heard.”

Source: Discrepancy between Elections Canada, StatsCan reports likely due to social desirability bias

About Andrew
Andrew blogs and tweets public policy issues, particularly the relationship between the political and bureaucratic levels, citizenship and multiculturalism. His latest book, Policy Arrogance or Innocent Bias, recounts his experience as a senior public servant in this area.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

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

%d bloggers like this: