open Secondary menu

The Burden of Voting in the 2019 Canadian Federal Election

Chapter 4: Who finds voting difficult?

We have seen in Chapter 3 that voting is perceived to be easy by the great majority of citizens. In this chapter we focus on the exceptions: we identify subgroups of the population for whom some aspects of voting are at least somewhat difficult, and we explore the reasons why it is so. Throughout the analyses, our dependent variable is the perceived burden of voting. We use the responses provided in wave 1, in which a large sample allows us to examine specific groups and subgroups. The four measures of perceived burden are measured on a four-point scale from 0 (very easy) to 3 (very difficult): registering, going to the polling station, voting, and deciding who to vote for. The means of all the burden indicators are low for all four actions: registration (mean 0.08), going to the polling station (mean: 0.36), voting (mean: 0.32) and deciding which candidate to vote for (mean: 0.94).

We proceed in two steps. We first examine the relationship between perceived burdens and a standard set of 11 socio-demographic characteristics: age, gender, race, 19 language, country of origin, region, education, income, employment, marital status, and mobility. We then proceed to a more in-depth analysis of four specific groups, which we have reasons to believe face greater obstacles: Indigenous people, persons with disabilities, those who had never voted before, and those between 18 and 34 years old who are not in education, employment or training (NEET).

Socio-demographic characteristics and their impact on the perceived burden of voting

Let us start with socio-demographic characteristics. We have retained 11 of them; most of them are dichotomized (men vs. women) but we distinguish low/medium/high education and income, 20 six regions (Atlantic, Quebec, Ontario, Prairies, BC and Territories), five occupational groups (unemployed, student, retired, caregiver, employed) 21 and six age categories (18–24, 25–34, 35–44, 45–54, 55–64, 65+). This gives us a total of 35 groups. Table 1 in Appendix C shows the mean score for each of the four burdens for each of the 35 groups.

Many of these characteristics are correlated with each other (for instance, the better educated tend to have a higher income). In order to ascertain the specific independent impact of each of them we perform multivariate regressions where the dependent variables are the four burdens and the independent variables are the socioeconomic characteristics listed above. In this case, for each characteristic, we define one group as the reference category, and we show how much higher or lower a given burden is in the other groups, relative to that reference category. For instance, for gender, the reference category is men. The coefficient associated with women tells us how much higher or lower a burden is for women compared to men, everything else being equal (controlling for all the other socio-demographic characteristics). The results of these regressions are presented in Figures 4.1 and 4.2 (Table 2, Appendix C displays the regression models).

Figure 4.1. The impact of socio-demographic characteristics on perceived burdens: registration and decision

Figure 4.1. The impact of socio-demographic characteristics on perceived burdens: registration and decision

Text version of "Figure 4.1. The impact of socio-demographic characteristics on perceived burdens: registration and decision"

This figure shows two plots representing the impact of socio-demographic characteristics on the levels of perceived burden associated to registering to vote and deciding who to vote for, respectively. The socio-demographics presented are: age, gender, racial minority, foreign language speakers, region, education level, income level, employment status, marital status and whether or not respondents had recently moved. The effects of each of these socio-demographic variables are described in the paragraph below.

Note: Reference categories are the following: Age: 45–54; Gender: Male; Racial Minority: White; Other Language: English/French; Foreign Born: Born in Canada; Regions: Ontario; Education: Medium Education; Income: Medium; Employment: Employed; Not Married: Married; Moved: Not Moved.

Figure 4.2. The impact of socio-demographic characteristics on perceived burdens: going to polling station and casting a ballot

Figure 4.2. The impact of socio-demographic characteristics on perceived burdens: going to polling station and casting a ballot

Text version of "Figure 4.2. The impact of socio-demographic characteristics on perceived burdens: going to polling station and casting a ballot"

This figure shows two plots representing the impact of socio-demographic characteristics on the levels of perceived burden associated to going to the polling station and casting a ballot, respectively. The socio-demographics presented are: age, gender, racial minority, foreign language speakers, region, education level, income level, employment status, marital status and whether or not respondents had recently moved. The effects of each of these socio-demographic variables are described in the paragraph below.

Note: Reference categories are the following: Age: 45–54; Gender: Male; Racial Minority: White; Other Language: English/French; Foreign Born: Born in Canada; Regions: Ontario; Education: Medium Education; Income: Medium; Employment: Employed; Not Married: Married; Moved: Not Moved.

Let us first consider age. We distinguish six age groups: those aged 18–24, 25–34, 35–44, 45–54, 55–65, and 65 and over. Table 1 in Appendix C shows that each of the burdens decreases more or less linearly as one gets older. The only exception to the linear pattern concerns registration, where the 18–24 group is clearly distinct. It is only in that group that registration is a meaningful obstacle. The patterns are very much the same when we control for the other socio-demographic characteristics in the multivariate setup (Figures 4.1 and 4.2). Being young is the most powerful predictor of facing a major obstacle to voting.

Next is gender. With and without controls, there is little gender gap. The only meaningful difference is that women find it more difficult to decide whom to vote for. This may be related to the fact that women are more likely to recognize that they are not very well informed (Thomas 2012) or are less likely to be very interested in politics. 22

We then find that minority groups, those born abroad, whose language spoken at home is neither English or French, and who are non-white, tend to perceive greater obstacles. Among these factors race seems to matter most: specifically, non-white electors perceive going to a polling station and casting a ballot as more difficult than white electors. The impact of race, however, is substantially reduced when the other socio-economic variables are taken into account. Interestingly, linguistic minorities and immigrants find it as easy as other Canadians to go to the polling station and to cast their vote.

Regional differences are muted. For some reason, most of the burdens are perceived to be slightly higher in Quebec (except for registration).

The various burdens systematically decrease with education and income. Interestingly, the impact of income is bigger than that of education. It is particularly interesting to note that the better educated (in the multivariate setup) do not find it easier to decide whom to vote for.

With respect to employment, the two groups who tend to perceive the greatest obstacles are the unemployed and students. At the other end, those who are retired report minimal difficulty. The patterns regarding students and retirees simply reflect the fact that these groups are very young or very old. When age is controlled for, these groups do not differ from the employed. Things are different for the unemployed. They still report higher burdens, even after controlling for variables such as age, education, income and race.

Regarding marital status, we find that those who are married (including having a common-law partner) find the various burdens relatively light. However, marriage does not make much difference once we take into account other socio-demographic characteristics, especially age.

Finally, those who moved recently are more likely to report some difficulty. The impact of mobility is reduced in the multivariate estimation, but it remains significant.

Among all the socio-demographic characteristics, age appears to be most strongly associated with perceived burdens; these burdens are clearly higher in the 18–24 group. Burdens also tend to be higher among the poor, racial minorities, and those who recently moved.

Groups who face specific barriers

We now turn our attention to four specific groups which we have reasons to believe face greater obstacles: Indigenous electors, including First Nations, Métis and Inuit electors, persons with disabilities, those who never voted before either by choice or because they were not eligible, and young people aged 18–34 who are not in education, employment or training (NEET). For each group, we show whether their mean burden is higher than in the rest of the sample, and we determine whether their higher burden can be explained by other socio-demographic characteristics such as age, education or income.

Let us start with Indigenous people. Table 3 in Appendix C presents the relevant findings for this group, including those within it who live on and off reserve (separately). We have examined whether burdens were perceived differently among First Nations (off reserve), Métis and Inuit; we observed small differences. We can see that burdens are for the most part higher among Indigenous people, especially among those who live on reserves. There is one interesting exception, and it concerns the "decision" burden. The obstacles that Indigenous people face concern administrative procedures and not the challenge of deciding which party/candidate to support.

The question is whether the obstacles that the Indigenous people have to overcome are specific to their group or whether they are the consequence of their socio-demographic characteristics, that is, of their being younger, less educated and poorer. We address this question by performing multivariate regressions for each of the four burdens in which we introduce a dummy for the Indigenous group or subgroups, while also including controls for all the socio-demographic characteristics that we have considered (Tables 4-8 in Appendix C). 23 The results of these regressions are presented in Figures 4.3 and 4.4. As before, the black dots indicate how much higher or lower a given burden is among a given group (let us say, Indigenous people), compared with the reference group (not Indigenous). The hollow circle indicates how much higher or lower a given burden is within a given subgroup (let us say Indigenous on reserve), compared with the reference group (Others).

Overall, we see that Indigenous people find it more difficult to register, go to a polling station and cast their vote. They do not find it difficult to decide whom to vote for. We then examine Indigenous people living on and off reserve. We find that it is the former who face bigger hurdles, notably with respect to going to a polling station and casting their vote.

The second group consists of persons with disabilities. The survey included questions asking respondents how much difficulty they have (none, some, a lot or "unable to do this at all") with the following actions: hearing, seeing, using stairs, or grasping objects (see Q7, Wave 1). This allows us to create a functional disability scale and to distinguish people with no disability, those with mild or moderate disability, and those with a severe disability. The latter two groups correspond to 24% and 8% of the sample, respectively. Table 3 in the Appendix shows that the burden of voting is indeed higher in this group, especially among those with a severe disability with respect to going to the polling station. Figures 4.3 and 4.4 show that the pattern is the same when we control for age, education, income, and other socio-demographic characteristics. The biggest expected difficulty is felt by those with severe disability and it concerns going to the polling station.

The third group corresponds to those individuals who never voted before. We distinguish three subgroups. The first consists of those who were too young to vote at the time of the previous federal election in October 2015. 24 They make up 3% of the sample. The second is those who were not Canadian citizens at the time of the previous election and who acquired citizenship before the 2019 election. They represent 0.2% of our respondents. The third is made up of those who indicated (see Q15, Wave 1) that they had abstained in all previous elections they were eligible to vote in; they constitute 5% of the sample.

The sample for new Canadians is very small. Table 3 in the Appendix indicates that this group does not report higher burdens, but it is difficult to generalize because we have only 103 respondents who became Canadians after 2015. Things are different with respect those aged 18–21 who became eligible to vote for the first time since the last federal election; and even more so for those who were eligible but had always abstained before. Figures 4.3 and 4.4 show that it is only the latter group which systematically expects voting to be more difficult, and this applies to each burden. These results suggest that there's a link between regular abstention and the belief that voting may not be an easy process.

The last group is made up of young people (aged 18–34) who are not in education, employment, or training; they constitute 2% of the sample. As seen in Table 3 in the Appendix, this group systematically perceives higher burdens. But when we control for age and employment, the only meaningful difference is that they find it more difficult to register to vote (Figures 4.3 and 4.4).

The goal of this chapter has been to determine whether the perceived burden of voting is higher among some subgroups of the population. When we look at the various socio-demographic characteristics, we find that the most important of all is age; the perceived burden is highest among those aged 18–24. We also considered various groups for whom we had reasons to believe that voting could constitute a bigger hurdle. We find this to be the case only for specific subgroups, namely, Indigenous electors living on reserves and electors with severe disabilities; and only for some burdens, namely, going to the polling station and casting a vote. Finally, there is a strong link between prior abstention and the expectation that the voting process will be difficult, although the direction of causality is not absolutely clear.

Figure 4.3. How much higher (or lower) are the perceived burdens in various groups, controlling for socio-demographic characteristics: registration and decision

Figure 4.3. How much higher (or lower) are the perceived burdens in various groups, controlling for socio-demographic characteristics: registration and decision

Text version of "Figure 4.3. How much higher (or lower) are the perceived burdens in various groups, controlling for socio-demographic characteristics: registration and decision"

This figure shows two plots representing the levels of perceived burdens, among various groups of electors, associated to registering to vote and deciding who to vote for, respectively. The groups of electors shown in the figure are: Indigenous, Indigenous living on reserve, Indigenous living off reserve, disability, moderate disability, severe disability, never voted before, new Canadians, first time electors (aged 18-21), abstainers and NEET youth. The relative perceived burden levels for each of these groups is thoroughly described in the paragraphs above this figure.

Note: Reference categories are the following: Indigenous: Not Indigenous; Indigenous on reserve, Indigenous off reserve: Others; Disability: No Disability; Moderate Disability, Severe Disability: No Disability; Never voted before: Everyone Else; New Canadians, First‑time electors (aged 18–21), Abstainer: Everyone Else; NEET: Everyone Else.

Figure 4.4. How much higher (or lower) are the perceived burdens in various groups, controlling for socio-demographic characteristics: going to the polling station and casting a ballot

Figure 4.4. How much higher (or lower) are the perceived burdens in various groups, controlling for socio-demographic characteristics: going to the polling station and casting a ballot

Text version of "Figure 4.4. How much higher (or lower) are the perceived burdens in various groups, controlling for socio-demographic characteristics: going to the polling station and casting a ballot"

This figure shows two plots representing the levels of perceived burdens, among various groups of electors, associated to going to the polling station and casting a ballot, respectively. The groups of electors shown in the figure are: Indigenous, Indigenous living on reserve, Indigenous living off reserve, disability, moderate disability, severe disability, never voted before, new Canadians, first time electors (aged 18-21), abstainers and NEET youth. The relative perceived burden levels for each of these groups is thoroughly described in the paragraphs above this figure.

Note: Reference categories are the following: Indigenous: Not Indigenous; Indigenous on reserve, Indigenous off reserve: Others; Disability: No Disability; Moderate Disability, Severe Disability: No Disability; Never voted before: Everyone Else; New Canadians, First‑time electors (aged 18–21), Abstainer: Everyone Else; NEET: Everyone Else.

Footnotes

19 Race is measured with the following question: "What is your ethnic or cultural background?" White is coded as 0 and everything else is coded as 1. Question Q94_from Wave 1 in the 2019 National Electors Study.

20 For education, this corresponds to those who completed high school or less, those with post-secondary education such as a technical diploma but no completed university degree, and those with a completed bachelor's degree. For income, this corresponds to an annual household income below $30,000 to an income above $30,000 but below $110,000, and to an income above $110,000. For income, we chose the highest and lowest response categories.

21 There is a sixth occupational group, "other," for which we do not report the results. Unemployed is coded as individuals who are unemployed but looking for work; Student is coded as a student attending school full-time; Retired is coded as individuals that are retired; Caregiver is coded as a caregiver or homemaker; Employed is coded as individuals who are working full-time, part-time and self-employed; Other is coded as other (we do not show Other in the figures, but it is in the appendix in the regression tables).

22 https://www.elections.ca/content.aspx?section=res&dir=rec/eval/pes2019/nes/nesve&document=p3&lang=e#fig1

23 In the case of Indigenous people, we could not control for race since, by definition, they all belong to the same ethnic group.

24 There is of course the possibility that some of them were eligible (and voted) in a provincial or local election. Unfortunately, we do not have that information.