open Secondary menu

Estimation of Voter Turnout by Age Group at the 38th Federal General Election (June 28, 2004)


The results suggest caution in concluding that more young people participated in the 2004 general election than in the 2000 general election. Given the very different methodologies employed, it is not appropriate to compare the results presented in this study to the Pammett-LeDuc findings, or indeed to any survey-based study. Public opinion surveys are invaluable research tools that enable researchers to better understand various aspects of the vote and related attitudes and behaviours. But when it comes to measuring voter turnout across age groups, surveys consistently tend to produce inflated figures. In this regard, the estimates presented in this study provide a new benchmark against which future participation rates should be measured. Regrettably, it is not possible to look back for comparisons, as the election-day lists used for previous elections were destroyed one year after the event, as prescribed by the law.

Perhaps one of the most striking findings of this study is that participation by first-time voters (18–21˝ year-olds) appears to have been higher than that of second-time voters (21˝–24 year-olds). This pattern, which was observed across the country in many provinces and one territory, could signal the beginning of a reversal of the decline among young voters witnessed in recent years. While the cause of this pattern is not yet evident, it is possible to suppose that the increased efforts by Elections Canada and others to educate, inform, and make the electoral process more accessible to youth, had a positive impact. In this context, the February 2004 mail-out to potential first-time voters is particularly noteworthy.

These new results should not lead to the conclusion that the problem of low youth turnout is resolved. The fact remains that young people continue to be significantly less likely to exercise their right to vote than older generations. We know that in the 2004 general election younger cohorts constituted a larger proportion of voters than in 2000. And while it is possible that first-time voters participated in somewhat greater numbers in 2004, they are still voting at a rate of some 35% less than the oldest age group. Further research, including more qualitative studies of young electors, is required to better understand the reasons for this trend.

The Chief Electoral Officer of Canada has a mandate to make the electoral process better known to Canadians, particularly those who may have difficulties exercising their right to vote. For this purpose, and given the results of this study, Elections Canada is committed to maintaining the momentum of its initiatives in this area both during and between elections, and will continue pursuing activities and partnerships to inform young people about their democratic rights and the importance of participating in the electoral process.

Appendix: Methodology


The goal of this study was to estimate the voter turnout rates at the 38th general election held on June 28, 2004, for specific age groups of interest. To this end, turnout was determined by dividing the number of people who voted by the number of eligible electors (Canadian citizens 18 years of age and older). Typically, studies of youth turnout have compared the number of youth voters to the estimated number of youth in the electoral population. However, in Canada, the official voting results are traditionally calculated by taking the number of registered electors.

For the purpose of this study, the estimated electoral population was based on Statistics Canada's 2001 Census. Adjustments were made to the Census population counts to remove non-citizens and citizens under 18 years of age, since they were not eligible to vote. Additional adjustments were made to take into account population growth during the time elapsed between Census Day and polling day, using demographic information provided by Statistics Canada.

In order to calculate turnout by age, two pieces of information were needed: evidence of the act of casting a ballot and date of birth. Date of birth information is available in the National Register of Electors and is copied onto the final list of electors produced after the election. The voting method (on polling day, by special ballot, or at advance polls), as set out in the Canada Elections Act, was also accounted for in the sampling strategy.

The study sample consisted of three different components: electors who voted at advance polls or by special ballot, electors who registered on the spot on polling day, and electors already registered who simply showed up and voted on polling day. The first two components required no sampling procedures, as the relevant information was available for all electors using those voting mechanisms. The third component consisted of a sample.

Administrative controls ensure that Elections Canada can keep records of electors who vote at advance polls or by special ballot; therefore, those who voted by these means could be identified on the final list of electors. This first component comprised 1,454,700 electors.

Electors who registered on polling day could easily be identified from the operational system, and all were assumed to have voted. Date of birth information for this group was taken from the registration certificates. This second component included 896,900 voters.

When electors show up at their polling station to vote, their name is physically struck off the list of electors for that polling division as a control measure. These lists, which are returned to Elections Canada with the ballots, are the only evidence that electors who were not required to register on polling day actually voted that day. Originally, a sample of some 90,000 such voters was selected to produce national turnout estimates. Some 20,000 voters were subsequently added to increase the reliability of turnout estimates at the provincial level. This third component was therefore the only one that was based on sampling errors. In this case, date of birth information came from the National Register of Electors.

The micro-data used for this study contain private information and therefore cannot be made publicly available.

Age Groups

Age groups were determined on the basis of an elector's age on polling day (June 28, 2004), and in such a way as to isolate those who had turned 18 since the previous general election (after November 27, 2000). This group, composed of 18–21˝ year-olds, represented young people who were eligible to vote for the first time federally in the 2004 general election. The second group, comprising 21˝–24 year-olds, represented those who had been eligible to vote federally for the first time in the 2000 general election, and therefore eligible to vote in a federal general election for the second time in 2004. These two youngest groups formed "youth" for the purpose of this study.

The following table illustrates the age groups based on dates of birth.

Table 4: Details on age groups
Age groups Born on or after And before Electors in population (estimated) Percentage
18–21˝ November 28, 1982 June 29, 1986 1,416,638 6.2
21˝–24 June 29, 1979 November 28, 1982 1,327,382 5.8
25–29 June 29, 1974 June 29, 1979 1,793,493 7.9
30–37 June 29, 1966 June 29, 1974 3,081,459 13.5
38–47 June 29, 1956 June 29, 1966 4,907,862 21.6
48–57 June 29, 1946 June 29, 1956 4,231,475 18.6
58–67 June 29, 1936 June 29, 1946 2,704,571 11.9
68+   June 29, 1936 3,303,620 14.5
Total     22,766,499 100.0

Sample Design

As previously explained, a sample was selected among previously registered electors who voted at their polling stations on June 28, 2004 (third component). The sampling unit was the polling division, and the sample itself was selected from the 59,514 polling divisions across the country.

For practical considerations, the sample was established in two steps. First, a number of federal electoral districts were randomly selected within each province. This selection yielded 28 electoral districts. Since the three territories each have only one electoral district, they were included by default, for a total of 31 districts overall. The number of electoral districts selected was determined to ensure adequate coverage at the national level.

The second step was to identify 20 polling divisions within each of the selected electoral districts for the original sample used to produce national estimates. The augmented sample, designed to improve the statistical reliability of provincial estimates, included 139 additional polling divisions. To make sure the sample represented the population as much as possible in terms of turnout, the polling divisions were ordered by turnout and selected using systematic sampling (selection of polling divisions from the list at regular intervals).

Margins of Error

At the national level, the sample produced estimates with a margin of error of ±4.3%, 19 times out of 20. In other words, if the same sampling strategy were replicated 20 times, we would expect that, 19 times out of 20 (95% of the time), the estimated turnout rate for each age group would not differ from the real value in the population by more than 4.3%. At first glance, the national margin of error may seem inconsistent with the large number of electors captured in the sample (over 90,000). From a methodological perspective, the overall margin of error has to be based on the number of polling divisions selected (sample unit) rather than the number of individual voters it contains.

The following table provides the margins of error for each age group at the national level as well as for each province, territory and aggregated region. Estimated turnout rates provided in this study may vary from the real value in the population by plus or minus their corresponding margins of error (in percentage points).

Due to the sampling procedures, the data does not allow reliable estimates of voter turnout for a geographical territory smaller than the province.

Table 5: Margins of error
18–21˝ 21˝–24 25–29 30–37 38–47 48–57 58–67 68+
Canada 3.6 3.5 4.0 4.0 5.2 5.4 6.4 9.4
Newfoundland and
4.2 4.8 5.9 4.2 3.1 3.7 4.9 6.2
Prince Edward Island 7.3 5.0 10.7 2.8 2.9 2.1 11.3 4.0
Nova Scotia 4.6 20.2 14.4 2.4 13.8 15.6 9.1 33.5
New Brunswick 2.1 2.8 11.0 4.5 2.0 5.1 12.2 4.5
Quebec 3.2 6.6 6.5 7.8 6.6 4.7 5.9 12.4
Ontario 7.5 3.1 4.9 7.0 7.5 6.8 6.1 13.9
Manitoba 3.6 4.9 11.2 16.2 6.6 5.5 5.7 34.4
Saskatchewan 5.1 3.0 4.7 5.1 8.2 11.2 7.7 25.4
Alberta 6.6 3.1 3.9 9.6 6.0 4.1 6.9 20.6
British Columbia 8.2 7.4 11.7 9.6 6.8 3.7 10.5 16.6
Yukon 2.8 4.5 8.2 3.6 3.9 4.2 8.0 13.1
Northwest Territories 5.5 4.1 5.8 3.4 4.1 7.5 7.4 7.6
Nunavut 3.7 2.6 3.5 2.2 3.1 5.2 4.3 8.8
Atlantic (NL, PE, NS, NB) 2.2 8.0 6.8 2.0 5.6 6.3 5.4 13.9
Prairies (MB, SK, AB) 4.1 2.2 3.4 6.9 4.2 3.4 4.4 14.9
Territories (YT, NT, NU) 2.5 2.2 3.4 1.9 2.4 3.6 4.6 6.8

Note: Margins of error vary considerably because polls were selected randomly with no control for age groups.

1 In absolute numbers, more people voted in the 2004 general election than in that of 2000. However, the number of electors in the National Register of Electors increased more than the electoral population did, hence the lower turnout rate. Official voting results can be consulted at The official report to Parliament on the 2004 general election is also available at > General Information > Official Reports > Elections Canada's Official Reports.

2 With the exception of the 1972 federal general election.

3 This trend appears to be confirmed by the 2004 CES; see Elisabeth Gidengil et al., "Missing the Message: Young Adults and the Election Issues," Electoral Insight, January 2005. For additional analysis of youth voting in the 2004 election, see André Turcotte, "Different Strokes: Why Young Canadians Don't Vote," in the same issue.

4 Report and database available at > Electoral Law and Policy > "Explaining the Turnout Decline in Canadian Federal Elections."

5 For the purpose of this study, turnout is calculated on the basis of the estimated population of eligible voters (or the voting age population), as opposed to the number of registered electors.

[ Previous ]