Career, education, ethnic roots: Demographics tell the tale of the 2019 federal election

By the time polls closed on Oct. 21, 2019, more than 18 million Canadians had cast their ballots to fill each of the 338 seats in the federal House of Commons. It had been a fairly tight race in which a minority government seemed likely, but uncertainty remained over whether that government would be headed by incumbent Liberal prime minister Justin Trudeau or Conservative opposition leader Andrew Scheer.

By the end of the night, it was clear that the Liberals would remain in office for at least the time being, their robust 2015 majority being reduced to 13 seats short of the majority-government threshold. Although they outnumbered the Conservatives in the House by 157 seats to 121, the Liberals won fewer votes overall.

Some of the answers as to why the 2019 election turned out as it did will be revealed when a team of academics completes the Canadian Election Study, a survey of Canadians conducted both before and after each federal election to analyze the patterns of public opinion underlying the result.

Yet there is another way to analyze the election, and that is by merging together two vital datasets: Elections Canada’s tally of votes received by electoral district, and Statistics Canada’s 2016 census results by electoral district. Merging this data can be fairly easily done by anyone with an understanding of how Excel’s Pivot Tables feature works. With the same constituencies being in play in both 2015 and 2019, the data files could be matched up quite readily.

Using Pearson correlations, it then becomes possible to look for patterns in the vote. For example, if 35 to 39 year olds are disproportionately likely to produce votes for Party A, this will show up as a solidly positive relationship across the ridings in which that party fields candidates with a correlation of perhaps 0.400 or better on a scale of -1.000 to +1.000. But if they are no more likely than the population as a whole to vote for Party A, the correlation would be closer to zero.

The Liberals’ core constituency is the country’s university graduates. Their presence was the single strongest predictor of how well the Liberals did in a constituency, as about one-third of the difference in the Liberal vote take from one riding to the next could be explained by the number of residents with a university certificate or diploma above bachelor level. Liberal support was particularly strongly associated with the presence of people who had degrees in history, law, the social sciences, computer and information sciences, mathematics, and statistics. Among these groups, they largely compete with a significantly weaker NDP and a weaker-still Green Party, while these groups show signs of mild aversion to the Conservatives and the Bloc.

Secondary sources of Liberal support appear to have been 20 to 24 year olds and 45 to 54 year olds, singles, immigrant communities, some very low income earners, those spending more than 30 percent of their income on shelter, and those employed in office jobs, particularly finance and insurance. Francophones also showed signs of being more likely to be Liberal voters.

Canadians of Indigenous descent, particularly Métis, appeared to be averse to voting Liberal in 2019, as were tradespeople such as those employed in construction or the mechanic trades. As has long been the case, those in agriculture and natural resources also appeared to be Liberal-unfriendly.  Their presence in a riding correlated favourably with Conservative fortunes.

Several subgroups formed the Conservatives’ core constituencies, in some cases these being the diametrical opposite of the Liberal support base. The Conservative vote in ridings throughout the country tended to be favourably influenced by the number of married people, native English speakers, households earning $80,000 or more annually, residents of northwestern European ancestries, and homes with four or more bedrooms. Those who had studied construction or theology also showed signs of being more strongly attracted to the Conservative Party in 2019.

Half of the difference in how the Conservatives polled across the 338 ridings could be explained by the number of residents claiming to be of non-French Western European origins, particularly German.

Electoral districts with higher numbers of people working in management occupations, trades, transport and equipment operation, natural resources, agriculture, forestry, fishing, hunting, mining, quarrying or oil and gas extraction also tended to be more likely to provide the Conservatives with favourable election night numbers. As noted before, there were indications that the larger a constituency’s Indigenous and particularly Métis population, the better the Conservative candidate tended to poll.

Forces that tended to hold down the Conservative vote included French-speaking populations and  low incomes. There are also indications that university-educated voters swung away from the Conservatives between 2015 and 2019. The ability of the next Conservative leader to make inroads in Quebec and among not just recent but past university graduates will play a role in how well the party does in the next election.

The Bloc Québécois is essentially Quebec’s blue-collar party, similar to what the NDP used to be in the rest of Canada. Its most critical support base is the francophone Quebecker — quelle surprise! — and those in the trades, particularly construction. Solid secondary support groups include those aged 50 to 64 years, those living in lower population density areas, and middle-income households in the fifth, sixth and seventh deciles.

There is a distinct aversion to voting for the Bloc among Quebec’s immigrant communities, specifically those of non-francophone origin, and of course among anglophones. University graduates also appeared to be negatively inclined toward the Bloc. These patterns were not just exhibited in the 2019 vote, but in how that vote differed from 2015, suggesting a further polarization of the Quebec electorate.

There are indications that there might be a chilly relationship between the Bloc and the younger Quebec voter, as about one-fifth of the difference between ridings in terms of how well the Bloc polled could be explained by the number of 25 to 29 year olds — the more of them there were, the worse the Bloc tended to do.

The 2019 campaign was not a great one for the NDP, securing the party just 24 seats in the House of Commons. A big part of the problem here is that, unlike the Liberals, Conservatives and the Bloc, the NDP didn’t seem to have a distinct constituency of its own to monopolize. Across the ridings, no demographic group showed signs of being strongly NDP-oriented, though at the same time, none seemed to be outwardly hostile, either.

Some groups, however, showed signs of being more likely than average to be NDP voters. This included those who were separated from their spouses, those with extremely low incomes of $5,000 or less, those of American or European (including British Isles) origin, and those who had studied public administration, social services or cultural or gender studies in university. People who worked in arts, entertainment, recreation, accommodation or food services also showed signs of being more likely to be NDP voters.

And, yes, the NDP ostensibly did quite well with the cyclist vote: about a fifth of the difference in their vote across the 338 ridings could be explained by how many people commuted to work by bicycle.

A large proportion of their setback in the 2019 election could be explained by having lost votes to the Bloc Québécois. On an election-over-election basis, they appeared to offset this with gains in English-speaking constituencies nationwide among married and separated voters and among those of British Isles and northwestern European origin — groups among which they might have taken some support away from the Conservatives.

They also appeared to split the support of those who had studied theology or religion with the Conservatives, reflecting religion’s natural liberal-conservative divide.

More worrisome for the NDP might be the loss of the party’s blue-collar roots. Households earning $30,000 to $50,000 nationwide appeared to drift away from the NDP between 2015 and 2019, with this being much more pronounced among those with an apprenticeship or trades certificate or diploma.

Like the NDP, the Green Party had trouble finding a demographic group to monopolize in the 2019 election. On a secondary basis, however, there are indications of a northwestern European element to the Green vote, as ridings with larger numbers of voters of British Isles and Scandinavian ancestry tended to produce slightly better results for Green Party candidates.

The incidence of some west coast Indigenous languages such as Straits Salish and Halkomelem were also associated with better Green Party outcomes.

Now we come to the Peoples’ Party of Canada. When former Conservative leadership contender (and near-victor) Maxime Bernier announced that he would start his own party running on an immigrant-unfriendly populist platform, some wondered if it would draw votes away from the Conservatives.

When it came time to vote, however, the Peoples’ Party singularly failed to monopolize or even do particularly well with any demographic. The best it could do was to cultivate a few groups that tended to skew slightly in its favour, such as third-generation or greater voters (i.e., those without recent immigrant roots) and a bit of the blue-collar and agricultural vote. These groups largely stuck by the Conservatives, though.

There were also a few obscure parties that ran in the election. Even though a lack of money and organization prevented them from running candidates in more than a fraction of constituencies, it was still possible to glean insights into their support base by looking at the relationship between constituency demographics and their total vote.

The Christian Heritage Party (51 candidates) has been a consistent minor contender for power for decades, never coming close to electing a Member of Parliament. They have strong cultural roots in the Low Countries, as shown by the fact that about a third of the difference in how they polled from riding to riding could be explained by the number of Dutch and Frisian speakers, the latter being a language indigenous to the Netherlands’ northern coast.

There were also indications that the number of German or Flemish speakers, the number of Indigenous people in a riding, and the presence of a significant agricultural, forestry, fishing or hunting sector had a favourable effect on how the CHP polled, however modestly.

Canada’s hard-left is split between two minor parties, the Communist Party of Canada and the Marxist-Leninists. This split goes back many decades to debates over whether the true path to socialism was to be found in Moscow or Beijing, and later in the more liberal Yugoslav variant, the isolationistic Albanian variant, or the tropical Cuban variant.

This difference shows in the correlations. As small as their voter bases were, a correlation analysis was enough to detect the distinctions — one of the reasons why I highly recommend looking at correlations as a much more insightful form of research than looking at percentages.

The Communist Party vote in Canada (30 candidates) was strongly correlated with the speaking of various East Asian languages in the constituencies where it ran candidates, especially Chinese languages such as Cantonese and Hakka. The Communist vote tended to be higher in ridings with larger numbers of people originally from China, Hong Kong or Viet Nam, suggesting a familiarity aspect.

The Marxist-Leninist vote (50 candidates), conversely, tended to be associated with populations of Eastern European origin, such as Albanian, Croatian, Macedonian, Polish or Slovene.

The Libertarians (24 candidates), which attracted about eight thousand votes nationwide — about the same as the Communists and Marxist-Leninists combined — also had a vote with some curious, strong linguistic correlations. More than 60 percent of the difference in the Libertarian vote across the ridings could be explained by — strangely enough — the number of Lao, Mohawk, Assyrian Neo-Aramaic or Chaldean Neo-Aramaic speakers, the latter two being Middle Eastern languages. The Liberals might want to ask these communities if there’s a bit of name confusion happening here — or perhaps there’s something about the Libertarians that resonates with them.

Finally, let’s look at the Rhinoceros Party. Dating back many years as a satirical presence on the Canadian electoral landscape, they fielded 39 candidates and won nearly 10,000 votes. Their support tended to come from (perhaps even exclusively from) Quebec, with the party resonating a bit more with blue-collar voters in the $20,000 to $50,000 income range.

If you’re an Excel lover like I am and want to look at numbers, have fun. (Large .xlsx file — don’t run up your data bill!)

About theviewfromseven
A lone wolf and a bit of a contrarian who sometimes has something to share.

Leave a Reply

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

You are commenting using your 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

%d bloggers like this: