open Secondary menu

Electoral Insight – New Ways of Building Democracy

Electoral Insight – November 1999

Voter Turnout in CanadaFootnote 1 – Findings from the 1997 Canadian Election Study

Tony Coulson
Former Research Project Officer, Elections Canada

This paper analyzes data from the 1997 Canadian Election Study, which was primarily funded by the Social Sciences and Humanities Research Council, with an additional financial contribution from Elections Canada. Elections Canada suggested specific questions about voter participation and electoral administration. The paper was prepared in 1998 when the author was Research Project Officer at Elections Canada.


In the 36 federal general elections and three referendums held in Canada since 1867, an average of 71 percent of registered electors has turned out to vote.Footnote 2 Voter turnout ranged from a low of 44 percent in the prohibition plebiscite of 1898 to a high of 79 percent at the general election of 1958.Footnote 3

In each decade between the 1940s and the 1980s, the average voter turnout in Canadian federal elections has been in the range of 73 percent to 78 percent. Despite the appearance of lower turnout rates in the 1990s official turnout figures of 69.6 percent in 1993 and 67 percent in 1997 there is evidence that the real decline occurred in 1997. Due to the inclusion of duplicate names and the names of deceased persons on the 1993 final list, the official turnout was artificially low.Footnote 4 On the basis of the growth of the electoral lists over time, Blais et al. have estimated the voter turnout rate in 1993 at approximately 73 percent.Footnote 5 The turnout rate of 67 percent in 1997, however, is the lowest since 1925.

The 1997 decline in Canadian voter turnout can be viewed as something of a paradox for two reasons. First, Canadians like the citizens of other industrialized nations are better educated and have easier access to larger amounts of information than ever before.Footnote 6 According to traditional expectations, these better-educated and more-informed citizens should be more likely to vote.Footnote 7

Second, in the period since 1992, numerous legislative amendments have been made to reduce "administrative disenfranchisement."Footnote 8 Numerous provisions in Canada's electoral legislation now facilitate the exercise of the franchise.Footnote 9

Multivariate Analysis

Table 1

Indicator variables Coefficient
Atlantic    .02 (.03)
Quebec    .06 (.02)**
Prairies – .02 (.02)
British Columbia    .03 (.02)
Indicator variables Coefficient
young – .11 (.02)***
low income – .02 (.02) a
unemployed – .03 (.03)
new immigrant – .06 (.03)*
non-Christian – .06 (.02)**
university graduate    .07 (.02)***
retired person    .07 (.02)**
Indicator variables Coefficient
low political interest – .04 (.02)*
politics too complicated – .03 (.02)
high information    .09 (.03)***
Indicator variables Coefficient
no party identification – .10 (.02)***
cynicism – .04 (.03)
Indicator variables Coefficient
unaware of voting options – .07 (.02)***
no important issue – .12 (.02)***
electoral district competitiveness – .01 (.03)
intercept    .87 (.03)***
adjusted R2    .11
F 16.86***
‡ n = 2412;ap < .10; * p < .05; ** p < .01; *** p < .001

Through a multivariate analysis of data from the 1997 Canadian Election Study (CES), several determinants of voter turnout have been identified. The 1997 Canadian Election Study is a three-wave survey of the Canadian electorate. The first survey was conducted during the election campaign period and 3 949 interviews were completed by telephone. The second survey was conducted in the period following the election and 3 170 of the campaign-period respondents were re-interviewed, again by telephone. The third component of the study was a self-administered mail-back survey which was completed by 1 857 of the post-election respondents.Footnote 10 Data analyzed here are drawn exclusively from the campaign wave and post-election telephone surveys.

The analysis focused on four explanations: demographic, psychological, political, and contextual. First, the demographic explanation relates to the personal resources of voters and non-voters. It is argued that individuals' demographic attributes give them the resources to participate in politics. "Those with limited personal resources little formal education, low income, no job, or marginal occupational status are considered less likely to vote than others."Footnote 11

Second, the psychological explanation suggests that individuals' general involvement in politics affects their turnout. General political interest, possession of politically relevant information, and an individual belief that means of political influence are available are believed to influence elector participation.

Third, political factors such as affiliation with a political party, or a high level of political cynicism should also influence voter turnout.

Fourth, the electoral context is thought to influence voter turnout. Included here are factors such as the level of campaign mobilization, the perceived competitiveness of the race, and the electoral laws. If an election does not capture the public's attention, is lopsided, or if procedural barriers to voting exist, not voting is expected to increase.Footnote 12

In general, it appears that each of the socio-demographic, psychological, political, and contextual explanations contributes something to our understanding of voter turnout in the 1997 Canadian general election.

A total of 19 indicator variables were regressed against the vote/not vote variable (4 regional, 7 socio-demographic, 3 psychological, 2 political and 3 contextual). The results of this analysis are presented in Table 1. Despite the theoretical reasonsFootnote 13 to expect each variable to affect turnout independently, only 10 indicator variables are found to be significant with the full set of controls in place.

The cell entries in Table 1 are unstandardized ordinary least squares regression coefficients (B) with standard errors in parentheses. Statistical significance is indicated by the symbols (see the notes at the bottom of the table). The regression coefficients are interpreted as the percentage change in the dependent variable resulting from a unit change in the independent variable. For example, the coefficient for "young" indicates that an elector aged 25 years or less in 1997 was 11 percent less likely to vote than an older elector.

Other socio-demographic factors also influence voter turnout. Recent immigrants to Canada and non-Christians are each about 6 percent less likely than others to vote, whereas university graduates and retired persons are about 7 percent more likely than others to vote.

In terms of psychological indicators, survey respondents with a low level of general political interest are about 4 percent less likely than others to vote, whereas highly informed electors are about 9 percent more likely than their less informed counterparts to vote.

The results for the political variables indicate that an elector who does not identify with one of the political parties is about 10 percent less likely to vote than someone who does have such an affiliation.

Finally, in terms of the electoral context, the data indicate that electors who are unaware that it is possible to vote if they are unable to go to the poll on election day are 7 percent less likely than others to vote. In addition, electors who indicated that there was no important election issue are about 12 percent less likely to vote than those who named an important issue.

This analysis suggests that individual resources, such as education, time and democratic experience (retirement, and length of time in Canada), do contribute to elector participation. The data also indicate that psychological factors, such as a general interest in politics and a high level of information also contribute. Partisanship, a political factor, is also relevant to voter turnout, as is the electoral context and its effect on elector mobilization and the perceived accessibility of the system.

Reasons for Not Voting

Responses to a question asking non-voters if there was a particular reason they did not vote were also analyzed. These reasons can be grouped into five categories (and others).

First, 38 percent of non-voters indicated that they did not want to vote, did not know who to vote for, or gave other cynical reasons for not voting. Respondents in this group were also significantly more likely than others to have no party identification and to have indicated that there was no important election issue.

Second, 27.3 percent of non-voters reported that they did not vote because they were busy or working. Students and working people are more likely than others (retirees, homemakers, and the unemployed) to have given this response.

Third, 9.5 percent of non-voters reported being away from their electoral district (either absent or on vacation). Not surprisingly, students who often attend university or college away from home are significantly more likely than others to have given this reason.

Fourth, 6.6 percent of reported non-voters were sick or busy with family or personal business and therefore did not vote. The proportion of respondents giving this reason increases significantly with the age of the respondent.

The fifth group is comprised of approximately 6.6 percent of non-voters, or about 1 percent of the post-election sample, who gave responses that can be considered administrative in nature. Seventeen respondents (3 percent of non-voters) stated that they did not vote because they were not enumerated, registered or otherwise on the list of electors. Seven respondents (1.3 percent of non-voters) stated that they had inadequate information. Five respondents (0.9 percent of non-voters) indicated that the polling station was too far from their homes, four (0.7 percent) reported arriving at the poll too late, one could not enter the poll due to a wheelchair accessibility problem, and one did not receive a ballot (presumably this respondent had applied to vote under the Special Voting Rules).

Thus, these findings are generally consistent with the results of the multivariate analysis presented in the previous section.

Awareness of Voting Methods

Respondents were also asked if it is possible for someone to vote if they cannot get to the poll on election day. The results indicate that the advance polls are the best-known alternative method of voting (named by 47 percent of all respondents to the survey). Fewer respondents appear to be aware of Special Voting Rules approximately 5 percent of all respondents gave this answer.

Many respondents incorrectly reported that voting by proxy (which has been repealed), and voting by phone are available alternatives at the federal level.

Thus, these data suggest that many Canadian electors are not aware of the alternative voting methods available to them. The multivariate analysis, presented earlier, confirmed that this lack of awareness is a significant contributor to not voting.


As noted, the 1997 decline in Canadian voter turnout appears somewhat paradoxical, given the socio-structural situation in Canada on the one hand, and the legislative developments of recent years on the other.

On the legislative front, the findings are consistent; many Canadians are unaware of the range of alternative voting methods available to them and this is a significant determinant of not voting. Thus, if Canadians were made aware of the various voting methods available to them, turnout might increase somewhat.

In regard to the structural changes noted, it appears that the tendency for more educated and informed citizens to vote is being offset by one or more other factors. One well-documented trend in this regard is a general decline in political partisanship. Nevitte, for example, reviews an extensive literature and concludes that "citizen attachments to traditional political parties have been weakening in nearly every advanced industrial state during the last decade or so," and that publics are turning, in growing numbers, to forms of political participation other than voting.Footnote 14 On Canada specifically, Clarke et al. note that:

The overall percentage of nonidentifiers [those with no party identification] in our 1997 survey, 30 percent, is unprecedented for any election year since 1965. Prior to 1997, the largest percentage of nonidentifiers in any election year was 13 percent (in 1993), and the average was 12 percent.Footnote 15

Thus, given that non-identifiers are about 10 percent less likely than identifiers to vote, this decline in partisanship must have contributed to the reduced voter turnout in 1997.

It appears, then, that although Canadians are better educated, more informed, and more interested in politics, many are turning away from traditional party affiliations and toward other forms of political participation. This helps, in part, to explain the reduction in Canadian voter turnout.

It has also been argued that the 1997 Canadian general election failed to capture the attention of many citizens. Greenspon, for example, notes that the necessity of calling the election in June 1997 was questioned.Footnote 16 Pammett also notes that the number of Canadians who failed to name an important election issue is an indicator of the lack of public interest in the 1997 election.Footnote 17 As demonstrated, the view that there was no important election issue is a significant determinant of not voting. Thus, turnout would likely have been somewhat higher had an important issue emerged and captured the attention of Canadians in 1997.

Finally, comparative research demonstrates that institutional factors such as compulsory voting, the voting age, the proportionality of the electoral system, and other factors influence turnout rates.Footnote 18 Thus, reforms at this level could also have a positive effect on voter turnout in Canada.


Footnote 1 The author wishes to thank André Blais for useful comments on an earlier version of this analysis. Variable descriptions are available from the author.

Footnote 2 Canadian voter turnout is calculated as the percentage of those registered on the electoral lists who cast a vote.

Footnote 3 A History of the Vote in Canada (Ottawa: Minister of Public Works and Government Services for the Chief Electoral Officer of Canada, 1997), pp. xvii and Appendix.

Footnote 4 See "Note to the Reader", in Official Voting Results, Thirty-fifth General Election, 1993 (Ottawa: Chief Electoral Officer of Canada). Also note that the Canada Elections Act has since been amended to permit the removal of duplicate names and names of deceased persons.

Footnote 5 See A. Blais, A. Bilodeau, and C. Kam, "Déplacement des votes entre les élections de 1993 et 1997." (Montreal: typescript).

Footnote 6 On education, see N. Nevitte, The Decline of Deference: Canadian Value Change in Cross-National Perspective (Peterborough: Broadview Press, 1996), p. 24. On the "information explosion," see M. Kanji and K. Archer, "When and How Canadian Voters Decide: Searching for Systematic Trends in Canadian Election Campaigns," paper prepared for the Annual Canadian Political Science Association Meeting, Ottawa (May/June 1998), pp. 7, 11-13.

Footnote 7 See W. Mishler, Political Participation in Canada: Prospects for Democratic Citizenship (Toronto: Macmillan Co., 1979), pp. 68, 98; also Nevitte (1996), p. 50.

Footnote 8 J. Pammett, "Voting Turnout in Canada" in Bakvis, ed., Voter Turnout in Canada, Volume 15 of the research studies of the Royal Commission on Electoral Reform and Party Financing (Toronto: Dundurn Press, 1991), p. 35.

Footnote 9 Bill C-78 (1992) included level access, mobile polls, transfer certificates for electors with disabilities to vote at a poll with level access, and public education and information programs. Bill C-114 (1993) included the extension of polling day registration to urban areas, and the extension of the Special Voting Rules to Canadian citizens temporarily outside Canada, inmates, and electors in Canada who are unable to vote at the advance poll or on polling day in their electoral district. Bill C-63 (1997) provided for a longer period for revision of the lists of electors, some revisions by phone, and registration at advance polls; it also extended the deadline for the receipt of mail-in ballots and lengthened the voting day to 12 hours. The Royal Commission on Electoral Reform and Party Financing in its four-volume 1991 report, Reforming Electoral Democracy (Toronto: Dundurn Press), recommended many of these changes.

Footnote 10 The Canadian Election Study was primarily funded by the Social Sciences and Humanities Research Council, with an additional financial contribution from Elections Canada. The co-investigators are André Blais (Université de Montréal), Elisabeth Gidengil (McGill University), Richard Nadeau (Université de Montréal), and Neil Nevitte (University of Toronto). Fieldwork for this project was conducted by the Institute for Social Research at York University. Neither the co-investigators, nor the Social Sciences and Humanities Research Council, nor the Institute for Social Research bear any responsibility for the analyses and conclusions presented here.

Footnote 11 L. Ragsdale and J. G. Rusk, "Who are Nonvoters? Profiles from the 1990 Senate Elections," in American Journal of Political Science, Vol. 37, No. 3 (1993), p. 722. See also H. E. Brady, S. Verba, and K. L. Schlozman, "Beyond SES: A Resource Model of Political Participation," in American Political Science Review, Vol. 89, No. 2 (1995) pp. 271-294.

Footnote 12 L. Ragsdale and J. G. Rusk, "Who are Nonvoters?" (1993), p. 722; also M. Eagles, "The Franchise and Political Participation in Canada," in Tanguay and Gagnon, eds., Canadian Parties in Transition, 2nd ed. (Toronto: Nelson Canada, 1996), p. 315.

Footnote 13 On Canada, see M. Eagles, "The Franchise and Political Participation in Canada," in Tanguay and Gagnon, eds., Canadian Parties in Transition, 2nd ed. (Toronto: Nelson Canada, 1996); and Bakvis, ed., Voter Turnout in Canada, Volume 15 of the research studies of the Royal Commission on Electoral Reform and Party Financing (Toronto: Dundurn Press, 1991). General studies include: R. J. Dalton and M. P. Wattenberg, "The Not So Simple Act of Voting," in Ada W. Finifter, ed., Political Science: The State of the Discipline II (American Political Science Assoc., 1993); M. Franklin, "Electoral Participation", in LeDuc, Niemi and Norris, eds., Comparing Democracies: Elections and Voting in Global Perspective (Sage, 1996), pp. 216-235; and J. Blondel, R. Sinnott and P. Svensson, "Representation and Voter Participation," in European Journal of Political Research, Vol. 32 (1997), pp. 243-272.

Footnote 14 N. Nevitte, The Decline of Deference, pp. 50 & 76.

Footnote 15 H. D. Clarke, P. Wearing, A. Kornberg and M. C. Stewart, "The Contest that Nobody Won: The 1997 Canadian Federal Election and the National Party System" (unpublished paper, Department of Political Science, University of North Texas, Denton, Texas 76203, 1997). Clarke et al. note that "although there is no canonical version of a party identification question battery, the 1974-80 CNES and 1983-1997 PSC surveys have fully consistent question wording to facilitate over-time comparisons."

Footnote 16 E. Greenspon, "Following the Trail of Campaign 97," in Frizell and Pammett, eds., The Canadian General Election of 1997 (Toronto: Dundurn Press, 1997), p. 23; also L. LeDuc, "The Canadian Federal Election of 1997" in Electoral Studies, Vol. 17, No. 1 (1998), p. 134.

Footnote 17 J. H. Pammett, "The Voters Decide" in Frizell and Pammett, eds., The Canadian General Election of 1997 (Toronto: Dundurn Press, 1997), p. 236.

Footnote 18 A. Blais and A. Dobrzynska, "Turnout in Electoral Democracies," in European Journal of Political Research, Vol. 33 (1998), pp. 239-261.


The opinions expressed are those of the authors; they do not necessarily reflect those of the Chief Electoral Officer of Canada.