A few weeks ago, Eric Duhaime wrote an op-ed for QMI arguing for a voting age of 16 in Canada. I’m a leading ridiculer of this idea, partly because the same people who propose it are the same nerds who worry endlessly about low voter-turnout figures. It so happens our voter-turnout figures are low by historical standards partly because we made a decision, in 1970, to lower the voting age from 21 to 18. Lowering the cutoff still further as a cure for perceived turnout malaise is like doubling a dose of poison. I shouldn’t single out Duhaime here, but… well, let’s single him out, because he wrote this:
Those who oppose [a lower voting age] point to the already very low turnout among 18- to 24-year-olds, and the fact that, at 16, potential voters are still heedless and irresponsible. Well, the sooner you start voting, the higher the probability you will keep doing your duty as a citizen for the rest of your life.
This kind of wishful thinking disguised as an assertion of fact is characteristic of the turnout nerd. (It’s not uncharacteristic of op-ed pages in general.) What struck me this time is: isn’t Duhaime’s Law checkable? Our previous lowering of the voting age provides us with the material for a natural experiment. We can pick out a series of birth cohorts whose members had to wait until they were 21 to vote in any election: those who were born before 1948 didn’t see the voting age lowered until they were at least 22, and presumably never developed those good civic habits Duhaime is talking about. Those born in 1954 or later, on the other hand, were turned loose on the electoral system immediately upon turning 18. Whee! Presumably they kept doing their duty, and their turnout numbers are higher.
Well, if you have any acquaintance with voter-turnout literature, you can already see the train wreck coming. There is a very strong tendency, across the age spectrum, for older voters to vote more, not less. But I felt I should give Duhaime the benefit of the doubt, and test his assertion against some evidence.
This involved giving a LOT of benefit. Which I’ll summarize here as quickly as possible.
In Cycle 22 of the longitudinal General Social Survey, which was executed in 2008, Statistics Canada asked about 20,000 Canadians various questions about civic engagement and social networks. These included questions about whether the respondents were eligible to vote, and whether they had voted at the last opportunity—not only in the most recent federal election, but in the most recent provincial and local elections relevant to them. You can thus look at Cycle 22 and get very strong, tightly bounded estimates of self-reported voter turnout for various age groups (in 2008). Statcan breaks ‘em down into five-year blocks.
So, with data from 2008, we’re interested in comparing the over-60s, who were born in 1948 or earlier, with the under-55s, born in 1954 or later. These correspond conveniently to the “60 to 64” age bracket in the survey and the “50 to 54” bracket. We’ll call them the Olds and the Youngs, though by now they’re all pretty damn long in the tooth. Duhaime insisted as a certain matter of fact that the Olds must have a (counterintuitively) lower propensity to vote than the Youngs, since they didn’t get the chance to form the habit as teenagers. These are the turnout numbers, with turnout defined as “estimated number of Canadians in the age bracket who claimed to have voted at the most recent opportunity” divided by “estimated number of Canadians in the age bracket who claimed to have been eligible to vote at that time”.
Federal Provincial Municipal/Local Olds 88.1% 89.4% 78.4% Youngs 84.0% 86.1% 71.8%
(Important disclaimer here.) These figures, since they are self-reported, are NOT consistent with the official, particular estimates of voter turnout by age group that Elections Canada produces by cross-referencing survey answers with the electoral register. Elections Canada knows this, as Statcan certainly does. (“A well-known problem encountered by researchers who use survey data to study voter participation is that self-reported turnout is consistently and significantly higher than the official turnout.”) But we’re comparing self-reports (Olds) to self-reports (Youngs) here, and still finding the same damn thing Elections Canada almost always finds with its superior methodology: older people turn out more often than younger people, even when the younger people had the supposed advantage of earlier civic engagement. If nothing else, the self-reporting indicates that more of the older people care about voting—if they’re not actually turning out in higher numbers, the best alternative explanation for the data is that they were more reluctant to admit not having done it.
What evidence we can lay hands upon runs contrary to, and not in favour of, the proposal that “the sooner you start voting, the higher the probability you will keep doing your duty as a citizen for the rest of your life.” And, moreover, if you are familiar at all with age trends in voting, your first reaction to that pronunciamento must inevitably be “That can’t possibly be right.” Duhaime wrote it because he wanted it to be true.