Site MenuThe Burden of Voting in the 2019 Canadian Federal Election
Chapter 2: The Data
Our analysis is based on the 2019 National Electors Study (NES), which was conducted by Elections Canada. The survey was conducted by telephone and online with eligible electors (Canadian citizens at least 18 years of age on election day) in the context of the 43rd federal general election held on October 21, 2019. footnote 3 It consists of a longitudinal component that followed the same respondents over three waves, as well as an additional telephone survey with a fresh sample that was administered immediately after the election. In this report, we only use the longitudinal component. We were consulted and provided comments and suggestions on initial versions of the questionnaire.
The longitudinal survey consists of three waves: the first before the election period between June 12 and July 14, 2019; the second during the election period between September 3 and October 20, 2019; and the third after the election between October 23 and December 9, 2019 (see Table 2.1). It is a three-wave panel, which means that participants in wave 1 are invited to respond to the same questions (as well as new ones) during the campaign and again after the election. This is important for us. As explained in Chapter 1, we are interested in citizens' perceptions of the burden of voting, and it is thus crucial to tap these perceptions early on. But we are also keen to determine whether these perceptions change over time and, especially, whether people are prone to over- or under-estimate the obstacles to voting.
Most of our analyses rely on the data collected in wave 1, which are based on a large sample that allows us to examine the burden of voting in many specific subgroups of the electorate. We use wave 2 and wave 3 occasionally to complement these analyses, especially to explore changes in citizens' perceptions, specific burdens that were not covered in wave 1, and to relate the perceived burden to actual turnout. We also do this to show that the results are robust. The sample size is about 50,000 respondents in wave 1, about 24,000 in wave 2, and about 19,000 in wave 3.
| Wave | Sample | Mode of Data Collection | Field Period | Sample Size |
|---|---|---|---|---|
| W1 | Longitudinal | Online, telephone | Pre-election: June 12 to July 14, 2019 | 49,993 |
| W2 | Longitudinal | Online | Election period: September 3 to October 20, 2019 | 23,880 |
| W3 | Longitudinal | Online, telephone | Post-election: October 23 to December 9, 2019 | 19,435 |
The participants were recruited in proportion to the population by province, age, and gender. Two-thirds of initial respondents were obtained via probability sampling and one-third from an online panel. Respondents from waves 2 and 3 were drawn solely from the initial wave 1 sample. Out of the 49,993 completed questionnaires (in wave 1), 46,930 were online and 3,063 were phone (both landline and cellphone) interviews.
The questionnaires include a great variety of questions designed to measure citizens' perceptions of the various burdens associated with voting. There are specific questions about how easy or difficult respondents expected (ex-ante) and found (ex-post) it to be to register (for those who were not registered), to go to the polling station, to cast their vote once arrived at the polling station, and to decide whom to vote for. footnote 4 The questionnaire also includes a detailed set of socio-demographic questions that allow us (given the large size of the sample) to examine perceived burdens among specific subgroups of the electorate for which we had reasons to believe that these burdens might be higher. There are also questions about level of political interest and sense of civic duty, the motivational sources of electoral participation, which will allow us to sort out the independent role of burdens in the decision to abstain. Finally, the respondents were asked in wave 1 whether they expected to vote or not and in wave 3 whether they actually voted. In short, the NES questionnaires provide a unique set of data to examine the magnitude of the various potential obstacles to voting in the whole electorate as well as in many subgroups and to ascertain their impact on citizens' decision to vote or not to vote.
The survey firm provided weights for each wave. The weights were constructed in two stages: first, to adjust for aspects of the study design (probability of selection, in-scope, non-response, and household size adjustments); and, second, to align the composition of the sample with known population characteristics. We compared the weighted and the unweighted data with respect to the four perceived burdens and all the independent variables we are interested in; the means are almost identical. Hence, we decided to use the unweighted data, which have the advantage that the numbers that we report are the raw numbers, without any transformation. footnote 5 Respondents who gave "don't know" as a response or did not provide a response are excluded.
Let us now proceed with the analysis of the survey. We start with an overall descriptive presentation of how easy or difficult Canadians perceive voting to be.
Footnotes
Return to source 3 For more information see the methodology report prepared by Phoenix Strategic Perspectives (June 2020).
Return to source 4 The questionnaire also included questions about the easiness or difficulty of nine specific steps that could be taken before people cast their vote. The problem is that respondents were asked to indicate which steps applied to them, and it is unclear what led people to answer that a given step did or did not apply to them. We use some of these questions to explore specific issues such as the difficulty associated with the identification card.
Return to source 5 There is the additional advantage that we do not have to use different weights when examining data from waves 2 or 3. However, the data reported here may differ from the data reported when weighted data are used (for example, in official NES reports).