Who votes? Age, interest in politics and turnout in Quebec • qesR

Version française

This page downloads 11 studies (about 16 MB) the first time it runs. options(qesR.cache = "disk") keeps them on disk for later sessions.

Turnout in Quebec provincial elections has declined. According to Élections Québec, it fell from 78% in 1998 to 66% in 2022, with a low of 57% in 2008. The conventional wisdom attributes this decline to the young, and not without reason: that young people vote less than their elders is one of the most robust findings in the study of elections, and in Canada it comes with a well-established explanation. Blais and colleagues (2004) show that the decline of turnout in federal elections comes mostly from the generations born after the baby boom, which vote less than their elders did at the same age, in good part because they pay less attention to politics and are less likely to see voting as a duty. However, most of what we know about the age gap comes from federal elections. Here, we use the Quebec Election Studies to ask two questions. How large is the age gap in reported turnout at each provincial election since 1998? And how much of it does interest in politics account for?

Overall, we find mixed support for the conventional wisdom. Respondents aged 18 to 34 report voting less than those aged 55 and over at every election, by 11 percentage points in 2007 and by 21 in 2018, and they are less interested in politics. Yet when we compare younger and older respondents who report the same interest, most of the gap remains: in 2018, it goes from 21 to 18 points. Interest is part of the story, but a small part at most elections.

Surveys overstate turnout

Before turning to age, we look at the reports themselves. Surveys overstate turnout, and the Quebec Election Studies are no exception. In 2022, 90% of respondents said they had voted, against an official turnout of 66%. Two mechanisms account for this (Selb and Munzert 2013). First, people who vote are also more likely to answer a political survey. Second, some who did not vote say they did. Weighting does not correct either: the weights bring each sample in line with census figures on age, sex and language (and, depending on the study, region or education), not with turnout.

Dumbbell chart, one row per study from 1998 to 2022: the official turnout of the election as a tick and the turnout reported by the study's respondents as a dot with its confidence interval. Every study is above the official turnout, by +9 pts to +30 pts. Values in the table view. — Surveys overstate turnout by 9 to 30 pointsReported turnout in each study (dot, 95% CI) against the official turnout of the election (tick)

Election	Study	Reported turnout, % [95% CI]	n	Official turnout, %	Gap	Weighting
1998	1998 polls	87.4 [85.6, 89.0]	1483	78.3	+9 pts	unweighted: no validated weight
2007	QES 2007	90.7 [89.0, 92.2]	2162	71.2	+20 pts	weighted
2007	2007 panel	85.3 [83.7, 86.8]	2054	71.2	+14 pts	unweighted: no validated weight
2008	QES 2008	87.2 [85.1, 89.0]	1131	57.4	+30 pts	unweighted: no validated weight
2012	QES 2012	93.2 [91.8, 94.4]	1486	74.6	+19 pts	weighted
2012	2012 panel	92.2 [90.2, 93.8]	844	74.6	+18 pts	unweighted: no validated weight
2014	QES 2014	88.9 [86.8, 90.8]	1499	71.4	+17 pts	weighted
2018	QES 2018	83.2 [81.5, 84.7]	2635	66.4	+17 pts	weighted
2018	2018 panel	83.7 [80.1, 86.7]	842	66.4	+17 pts	weighted
2022	QES 2022	90.0 [87.3, 92.1]	1215	66.1	+24 pts	weighted

Every study sits above the official figure, by 9 to 30 points, and the gap is largest in 2008, the election with the lowest turnout. More generally, reported turnout moves much less than official turnout. Across the Quebec Election Studies it ranges from 83% to 93%, while official turnout ranges from 57% to 78%. In other words, these surveys are of little use for measuring the level of turnout, and more useful for comparing groups, which is how we use them below. Of course, a comparison between groups is only as good as the reports behind it. If young non-voters say they voted more often than older non-voters do, the gap between the reports of younger and older respondents understates the real gap; if they say so less often, it overstates it. The surveys cannot tell the two apart. Selection calls for the same caution: the young who agree to answer a political survey are likely more engaged than their peers, which, if anything, would narrow the gap between reports.

The age gap, election by election

Line chart with a confidence band of the gap in reported turnout between respondents aged 55 and over and those aged 18 to 34, in points, at each election from 1998 to 2022. The gap is positive at every election: +11 pts in 2007, +21 pts in 2018 (the largest) and +8 pts in 2022. Values in the table view. — Older respondents report voting more than the young at every election; the gap peaked at 21 points in 2018Reported turnout of respondents aged 55 and over minus that of those aged 18 to 34, in points, with its 95% confidence interval

Election	Study	18 to 34, %	35 to 54, %	55 and over, %	Gap, 55+ minus 18-34 [95% CI]	Weighting
1998	1998 polls	81.6	88.8	90.3	+8.7 pts [3.9, 13.6]	unweighted: no validated weight
2007	QES 2007	84.8	91.4	95.7	+11.0 pts [6.9, 15.1]	weighted
2007	2007 panel	73.5	86.4	91.0	+17.5 pts [12.9, 22.2]	unweighted: no validated weight
2008	QES 2008	78.9	86.5	93.5	+14.5 pts [9.0, 20.0]	unweighted: no validated weight
2012	QES 2012	89.5	92.7	96.4	+7.0 pts [3.5, 10.4]	weighted
2012	2012 panel	89.0	91.7	93.7	+4.7 pts [-1.3, 10.7]	unweighted: no validated weight
2014	QES 2014	80.9	89.8	93.9	+13.1 pts [7.6, 18.5]	weighted
2018	QES 2018	70.3	81.8	91.4	+21.2 pts [17.1, 25.2]	weighted
2018	2018 panel	73.8	81.9	91.4	+17.5 pts [8.3, 26.8]	weighted
2022	QES 2022	86.5	85.8	94.9	+8.4 pts [2.6, 14.2]	weighted

The age gap is present at every election. Respondents aged 55 and over report voting more than those aged 18 to 34 by 11 points in 2007, 21 in 2018 and 8 in 2022. Its size, however, is far from stable. The gap is smallest in 2012 (7 points), when official turnout was 75%, and largest in 2018, when it fell to 66%. In 2018, the drop in turnout shows among the young and barely among their elders. Compared with 2014, reported turnout fell by 11 points among 18- to 34-year-olds (to 70%) and by 2 among those aged 55 and over (to 91%). One might conclude that when turnout falls, it falls among the young. The 2022 election does not fit this reading. Official turnout was 66%, as in 2018, but the gap was less than half as large, and those aged 18 to 34 reported voting about as often as those aged 35 to 54 (both at 86%, although the figure for the young rests on only 269 respondents).

Interest in politics matters most before 55

Line chart, one line per age group (18 to 34, 35 to 54, 55 and over), of the gap in reported turnout between respondents with high and with low interest in politics, in points, at the 2007, 2012, 2014, 2018 and 2022 Quebec Election Studies, with confidence intervals. Under 55 the gap reaches +30 pts; among those aged 55 and over it is +11 pts at most. Values in the table view. — Under 55, the most interested report voting up to 30 points more than the least interested; among those aged 55 and over, 11 at mostReported turnout of the most interested minus that of the least interested, within each age group, in points, with 95% confidence intervals

Election	Study	Age	Interest	Reported turnout, % [95% CI]	n	Weighting
2007	QES 2007	18 to 34	Low interest	75.7 [65.4, 83.7]	117	weighted
2007	QES 2007	18 to 34	Some interest	87.0 [80.8, 91.5]	246	weighted
2007	QES 2007	18 to 34	High interest	87.0 [78.9, 92.2]	197	weighted
2007	QES 2007	35 to 54	Low interest	82.2 [72.8, 88.8]	127	weighted
2007	QES 2007	35 to 54	Some interest	92.9 [88.2, 95.8]	369	weighted
2007	QES 2007	35 to 54	High interest	94.4 [89.8, 97.0]	238	weighted
2007	QES 2007	55 and over	Low interest	91.3 [83.4, 95.6]	110	weighted
2007	QES 2007	55 and over	Some interest	97.1 [94.1, 98.6]	310	weighted
2007	QES 2007	55 and over	High interest	95.9 [92.9, 97.7]	403	weighted
2012	QES 2012	18 to 34	Low interest	77.1 [69.6, 83.2]	174	weighted
2012	QES 2012	18 to 34	Some interest	95.2 [91.6, 97.3]	236	weighted
2012	QES 2012	18 to 34	High interest	96.2 [90.0, 98.6]	119	weighted
2012	QES 2012	35 to 54	Low interest	89.3 [84.4, 92.8]	211	weighted
2012	QES 2012	35 to 54	Some interest	94.9 [91.2, 97.1]	292	weighted
2012	QES 2012	35 to 54	High interest	97.9 [93.6, 99.3]	126	weighted
2012	QES 2012	55 and over	Low interest	97.4 [89.8, 99.4]	64	weighted
2012	QES 2012	55 and over	Some interest	94.6 [90.0, 97.2]	162	weighted
2012	QES 2012	55 and over	High interest	99.1 [93.8, 99.9]	88	weighted
2014	QES 2014	18 to 34	Low interest	66.4 [56.4, 75.1]	138	weighted
2014	QES 2014	18 to 34	Some interest	86.0 [78.4, 91.3]	167	weighted
2014	QES 2014	18 to 34	High interest	96.3 [90.1, 98.6]	84	weighted
2014	QES 2014	35 to 54	Low interest	81.4 [73.8, 87.1]	171	weighted
2014	QES 2014	35 to 54	Some interest	96.2 [92.4, 98.1]	279	weighted
2014	QES 2014	35 to 54	High interest	95.2 [87.9, 98.2]	126	weighted
2014	QES 2014	55 and over	Low interest	89.6 [80.0, 94.9]	86	weighted
2014	QES 2014	55 and over	Some interest	94.3 [87.9, 97.4]	269	weighted
2014	QES 2014	55 and over	High interest	96.3 [90.6, 98.6]	169	weighted
2018	QES 2018	18 to 34	Low interest	60.6 [53.8, 67.0]	269	weighted
2018	QES 2018	18 to 34	Some interest	75.2 [69.6, 80.1]	317	weighted
2018	QES 2018	18 to 34	High interest	80.2 [71.4, 86.8]	120	weighted
2018	QES 2018	35 to 54	Low interest	73.4 [65.5, 80.0]	159	weighted
2018	QES 2018	35 to 54	Some interest	80.8 [75.1, 85.5]	241	weighted
2018	QES 2018	35 to 54	High interest	94.8 [88.8, 97.7]	114	weighted
2018	QES 2018	55 and over	Low interest	83.8 [78.0, 88.2]	226	weighted
2018	QES 2018	55 and over	Some interest	92.2 [89.6, 94.2]	697	weighted
2018	QES 2018	55 and over	High interest	94.8 [92.0, 96.6]	456	weighted
2022	QES 2022	18 to 34	Low interest	73.9 [61.9, 83.1]	95	weighted
2022	QES 2022	18 to 34	Some interest	94.0 [81.8, 98.2]	105	weighted
2022	QES 2022	18 to 34	High interest	91.9 [78.9, 97.2]	67	weighted
2022	QES 2022	35 to 54	Low interest	69.5 [50.9, 83.4]	80	weighted
2022	QES 2022	35 to 54	Some interest	84.7 [76.8, 90.3]	150	weighted
2022	QES 2022	35 to 54	High interest	97.3 [93.1, 99.0]	164	weighted
2022	QES 2022	55 and over	Low interest	91.1 [80.5, 96.2]	67	weighted
2022	QES 2022	55 and over	Some interest	94.0 [87.5, 97.2]	194	weighted
2022	QES 2022	55 and over	High interest	96.5 [92.7, 98.3]	284	weighted

The first part of the conventional wisdom holds. At every election, the young are less interested in politics than their elders. In 2018, 37% of respondents aged 18 to 34 said they were not at all or hardly interested, against 19% of those aged 55 and over. Interest is also linked to turnout, but mostly before 55. Under 55, the most interested report voting up to 30 points more than the least interested; among those aged 55 and over, the difference is 11 points at most. Unsurprisingly, older respondents are close to the ceiling: most report voting whatever their interest. In other words, the age gap is widest among the least interested. In 2018, it reaches 23 points between older and younger respondents with low interest, against 15 among those with high interest.

At equal interest, most of the gap remains

If the young vote less because they are less interested, comparing younger and older respondents with the same interest should shrink the gap considerably. We turn to this test now. For each election, we compare the age gap among all respondents with the gap among respondents who gave the same answer to the interest question.

Dumbbell chart, one row per Quebec Election Study from 2007 to 2022: the gap in reported turnout between respondents aged 55 and over and those aged 18 to 34, among all respondents as a tick, and among respondents with the same interest in politics as a dot with its confidence interval. At 5 of the 6 elections, holding interest equal narrows the gap by less than a third; in 2022, it narrows it by more than half, from 9 to 4 points. Values in the table view. — At 5 of 6 elections, holding interest equal narrows the age gap by less than a third; in 2022, by more than halfGap in reported turnout, 55 and over minus 18 to 34, among all respondents (tick) and among respondents with the same interest in politics (dot, 95% CI); at right, the difference

Election	Study	Age gap, all respondents [95% CI]	Age gap, same interest [95% CI]	Difference	Low interest, 18 to 34, %	Low interest, 55 and over, %	n	Interest in	Weighting
2007	QES 2007	+11.0 pts [6.9, 15.1]	+10.5 pts [6.5, 14.5]	−0.5 pts	19.9	13.4	2117	politics, 0 to 10	weighted
2008	QES 2008	+14.5 pts [9.0, 20.0]	+12.3 pts [7.1, 17.5]	−2.2 pts	50.2	34.8	1129	the election, 0 to 10	unweighted: no validated weight
2012	QES 2012	+6.7 pts [3.3, 10.1]	+4.9 pts [1.7, 8.1]	−1.8 pts	31.9	20.5	1472	politics, four points	weighted
2014	QES 2014	+12.9 pts [7.4, 18.4]	+10.0 pts [4.7, 15.3]	−2.9 pts	36.3	20.1	1489	politics, four points	weighted
2018	QES 2018	+20.6 pts [16.5, 24.7]	+17.6 pts [13.6, 21.6]	−3.0 pts	36.7	18.9	2599	politics, four points	weighted
2022	QES 2022	+8.5 pts [2.7, 14.4]	+3.8 pts [-2.1, 9.8]	−4.7 pts	35.3	11.9	1206	politics, 0 to 10	weighted

For the most part, the gap does not shrink. At 5 of the 6 elections, holding interest equal narrows the age gap by less than a third. In 2018, the gap goes from 21 to 18 points; in 2007, it barely moves (by 0.5 point). The exception is 2022. Among the respondents who also answered the interest question, the gap that year is about 9 points; holding interest equal cuts it by more than half, to about 4 points, and it can no longer be told apart from zero. That year, the gap in interest between age groups is much wider than in 2007, the other study that asked the same 0-10 question: 35% of those aged 18 to 34 say they are not at all or hardly interested, against 12% of those aged 55 and over (20% and 13% in 2007). In short, interest accounts for a small part of the age gap at most Quebec elections, and for about half of it in 2022.

Taking stock

In sum, the age gap holds, but the usual account of it holds only in part. The gap is real, present at every election since 1998, and the young are less interested in politics than their elders. However, interest accounts for only a few points of a gap that reached 21 in 2018. The exception is 2022, when it accounts for about half of a much smaller gap. Most of the gap, then, lies elsewhere. Blais and colleagues (2004) point to two other sources: the life cycle, as turnout rises with age within each generation, and a weaker sense among recent generations that voting is a duty. The first is part of the age gap itself, since at a single election the effect of age cannot be told apart from that of generation; the second is asked only in the 2022 study, so it cannot be followed from one election to the next, and we leave it aside here. Two limits also apply to what we find. First, these are reports, and the surveys overstate turnout: the real gap between younger and older citizens may be larger or smaller than the gap between their reports. Second, as with the life cycle, comparing age groups at each election cannot separate age from generation. Whether the young of 2018 will vote like their elders as they grow older, or keep a lower turnout throughout their lives, is a question that future Quebec Election Studies will help answer.

Going further

Blais, André, Elisabeth Gidengil, Neil Nevitte, and Richard Nadeau. 2004. “Where Does Turnout Decline Come From?” European Journal of Political Research 43 (2): 221–236. https://doi.org/10.1111/j.1475-6765.2004.00152.x
Rubenson, Daniel, André Blais, Patrick Fournier, Elisabeth Gidengil, and Neil Nevitte. 2004. “Accounting for the Age Gap in Turnout.” Acta Politica 39 (4): 407–421. https://doi.org/10.1057/palgrave.ap.5500079
Selb, Peter, and Simon Munzert. 2013. “Voter Overrepresentation, Vote Misreporting, and Turnout Bias in Postelection Surveys.” Electoral Studies 32 (1): 186–196. https://doi.org/10.1016/j.electstud.2012.11.004
Bélanger, Éric, Philippe Mongrain, Thomas Gareau-Paquette, and Valérie-Anne Mahéo. 2025. “A Party that Went Viral? The Drivers of Support for the Parti Conservateur du Québec in the 2022 Election.” Canadian Journal of Political Science 58 (2): 277–296. https://doi.org/10.1017/S0008423924000829

Bélanger and colleagues (2025) draw on the same 2022 study to explain the vote for the Parti Conservateur du Québec (PCQ), a party whose voters were younger than those of most other parties (only Québec solidaire’s were as young). How far the surveys are from the official results on vote choice, rather than turnout, is the subject of another example.

The data in R

The page rests on three pooled variables: turnout, the turnout reported after the election; age_group3; and pol_interest, interest in politics, rescaled from 0 to 1 whether the question had four points or ran from 0 to 10.

h <- qes_harmonize(
  studies = "all",
  targets = c("turnout", "age_group3", "pol_interest"),
  missing = "reasons", quiet = TRUE
)

The turnout of each age group in 2018, and the age gap at equal interest, with the survey package (the h of the call above):

library(survey)
d18 <- qes_design(h[h$study == "qes2018" & !h$eligible_voter %in% FALSE, ], weight = "weight_post")
d18 <- subset(d18, !is.na(turnout) & !is.na(age_group3))
svyby(~I(turnout == "Yes"), ~age_group3, d18, svyciprop, vartype = "ci", method = "logit")
#>              age_group3 I(turnout == "Yes")      ci_l      ci_u
#> 18-34             18-34           0.7026868 0.6641214 0.7385673
#> 35-54             35-54           0.8177600 0.7804389 0.8499568
#> 55 and over 55 and over           0.9142527 0.8964074 0.9292665

d18 <- update(subset(d18, !is.na(pol_interest)), voted = as.numeric(turnout == "Yes"),
              age = factor(as.character(age_group3)), answer = factor(pol_interest))
coef(summary(svyglm(voted ~ age + answer, d18)))[2:3, ]
#>                 Estimate Std. Error  t value     Pr(>|t|)
#> age35-54       0.1008001 0.02527438 3.988234 6.840858e-05
#> age55 and over 0.1762605 0.02045754 8.615921 1.185130e-17

About the data

Studies. Every study that asked, after the election, whether the respondent had voted; the CROP polls did not. Besides one Quebec Election Study per election, the table views include the panels led by Claire Durand at the 2007, 2012 and 2018 elections, which interviewed the same respondents before and after the vote. Official turnout is ballots cast over registered electors, from Élections Québec.
Interest in politics. The 2012 to 2018 studies asked a four-point question, scored 1, 0.7, 0.3 and 0; the 2007 and 2022 studies asked a 0-10 question, divided by 10. The scores of the four-point question are a coding decision, not a measurement, so the bands are best compared within an election. The 2008 study asked instead, after the vote, how interested respondents had been in the election (0-10), and the 2007 panel asked about interest in the campaign; the 2012 and 2018 panels have no comparable question.
Weights. Each study’s post-election weight, where it has a validated one; the others are left unweighted and drawn hollow (see the weights table).
Who was interviewed. The 1998 polls interviewed francophones only, and the respondents they interviewed again after the vote are not representative of all francophones. The 2018 study also interviewed 16- and 17-year-olds, who could not vote; they are left out.