Youth Electoral Engagement in Canada
3. Contrasting young voters and non-voters
We now focus our attention exclusively on youth and on the differences among youth voters and abstainers. For this part of the analysis we pool together the Canadian Election Studies conducted in 1997, 2000, 2004, 2006, and 2008.Footnote 7 We need to pool these five studies in order to get a sufficient number of both voters and non-voters within each of our two youth groups. The CES data, like all election surveys, overestimate turnout mostly because those who are not interested in politics (and who are less inclined to vote) are less prone to respond to election surveys (Brehm 1993). We have thus weighted the data so that self-reported turnout in the survey corresponds to the official turnout as reported by Elections Canada. In this part of the analysis, we use the same age categories we used in an earlier version of this report, namely 18-24 and 25-30 .
In contrasting young voters and non-voters, we first consider socio-demographic factors. We then consider political engagement factors. In each case, we consider simple bivariate relationships, i.e. relationships between voting and one other factor. Following this, we consider multiple factors at the same time to determine the independent effect of each.
As we shall see, not every factor matters much for the decision to vote (e.g., gender), and some others appear to matter differently for different age groups (e.g., marriage). Finally, political factors seem to exert more influence than socio-demographic factors.
Our analysis begins with socio-demographic characteristics. Table 2 shows the estimated proportion of voters among different groups. Several important patterns emerge. The most apparent pattern is that regardless of the variable, turnout in every category increases as people get older. For example, consider those whose household income is less than $40,000 per annum. On average, 34% of those aged 18-24 vote. However, among those aged 25-30 in the same income group, 41% report voting. This increase with age can be seen across every one of our categories (we also demonstrate this general trend in the next section).
In comparing voters and non-voters, no clear differences occur according to gender. The difference between men and women never exceeds 1.5 percentage points. By contrast, we see that income appears to exert an influence over the decision to vote or not to vote. Indeed, the gap in voter participation between those with a household income below $40,000 and those above is nearly 6 percentage points among those aged 18-24 and 11 percentage points among those aged 25-30 . The effect of some postsecondary education is similar, though the size of the effect is larger. The participation gap among those aged 18-24 is 9 percentage points. It grows to 17 percentage points among those aged 25-30 .
What of the effects of being a student? Our results suggest that being a student has the effect of increasing participation among those aged 18-24 (the gap is 9 percentage points) while there is really no difference among those aged 25-30 .
|Age||No Postsecondary||Some Postsecondary|
|Age||Not a student||Student|
|Age||Not born in Canada||Born in Canada|
Data are drawn from the 1997, 2000, 2004, 2006, and 2008 Canadian Election Studies.
We find that where one lives has an effect on the likelihood of voting. Among those aged 18-24 , those who live in rural settings outvote those in urban environments by some 7 percentage points. However, among those aged 25-30 , the pattern appears to be the reverse as the gap grows to 12 percentage points in the direction of those living in urban centres. This is a puzzling pattern and we cannot determine its cause at this time. The most plausible hypothesis, however, is that this is due to differences in the rural population between two age groups attributable to mobility. There is a general migration among individuals from rural to urban settings. However, this migration is most likely to occur among those who are better educated and better off and seeking more lucrative employment or higher education. Accordingly, those who select out of a rural environment by the time they reach our second age group are more likely to vote than those who opt to remain in a rural setting.
Unfortunately, the CES does not include a measure of mobility, though we suspect that mobility is an important factor in lower turnout rates among youth. It should certainly be explored further in the future.
Marriage appears to have similarly contradictory effects. In our younger group, single individuals are more likely to vote than married citizens. Indeed, the gap is some 5 percentage points. However, this gap reverses in favour of the married among those aged 25-30 . Again, we do not have the data at hand to decisively explain this pattern;Footnote 8 however we do note that marriage does not play a role in our later multivariate analysis.
Religiosity has been known to influence the decision to vote for some time (see, for instance, Abramson et al. 2006, 90, Table 4.4). We find support for this proposition in our data, particularly among the youngest voters. For those citizens aged 18-24 , religious adherents participate more than non-adherents to the tune of 4 percentage points. This gap declines to below 2 percentage points among those aged 25-30 .
Finally, we find that whether citizens are born in Canada has important effects on the decision to vote or not to vote, but that this effect declines with age. For our younger citizens, those who are Canadian-born are much more likely to vote. Indeed, the gap is 12 percentage points, the largest observed among all of our socio-demographic factors. However, this gap declines to just 4 percentage point among voters aged 25-30 . This suggests that those who are born outside of Canada take slightly longer than their Canadian-born counterparts to come to socialize into Canadian politics. However, these results also suggest that this socialization and resulting participation do occur given time. It remains to be seen whether this influence is robust after controlling for other political factors. To anticipate our results, we do find that even after controlling for political factors, individuals born outside of Canada have a markedly lower probability of voting. This suggests that something in the experience of being an immigrant makes voting either more difficult or less attractive, even among those who are informed about and engaged in Canadian politics.
In addition to socio-demographic factors, the decision of individuals to vote or not to vote is likely affected by their engagement in politics. We begin by considering individuals' interest in politics.Footnote 9 Unsurprisingly, those who profess a high interest in politics are more likely to vote. Indeed, among those aged 18-24 the gap is some 22 percentage points between those who express a medium or high general interest in politics and those who express a low interest. This gap grows further to 28 percentage points among those aged 25-30 ! (Table 3)
A similar pattern emerges according to political information, which is either measured as a share of correct answers to a series of factual questions or is evaluated by an interviewer (see Blais et al. 2009).Footnote 10 Among those aged 18-24 , those who are regarded as having medium or high political information vote at a rate 23 percentage points higher than those with low information. This gap grows to 27 percentage points among those aged 25-30 .Footnote 11 As with interest in politics, information about politics plays a very important role in the decision to participate in federal elections.
Campaign events may also play a role in motivating voters to go to the polls. Principal among these events are the leaders' debates. Watching debates can help clarify the differences between parties and give voters of all stripes more concrete reasons to go the polls and cast a ballot. The CES data suggest that among those aged 18-24 , the participation gap between those who watch a debate and those who do not is some 24 percentage points. This gap is similar (24 points) among those aged 25-30 .
|Age||Low interest in politics||Med/high interest in politics|
|Age||Low political information||Med/high political information|
|Age||Did not watch debate||Watched debate|
|Age||No other political activity||Other political activities|
|Age||No use of internet for news||Use of internet for news|
Data are drawn from the 1997, 2000, 2004, 2006, and 2008 Canadian Election Studies.
Until now, we have considered factors of engagement with formal electoral politics. But what is the relationship between engagement in other social or political activities and the decision to vote in elections? For example, what is the relationship between engaging in protest or other unconventional political action and the decision to vote. It is sometimes suggested that youth are not voting because they have found more meaningful political activities in which to engage. If this is the case, then we should find that turnout is at least slightly lower among those citizens who engage in other political activities, such as signing a petition or attending a protest. In three elections (2000, 2004, and 2008), the CES asked respondents to indicate how many of various political activities they have engaged in, namely signing a petition, participating in a boycott, attending a lawful demonstration, joining an illegal protest, or occupying a building or factory. As it turns out, those who participate in other political activities appear more likely to vote in federal elections than those who do not participate in other activities. Among those aged 18-24 , the turnout gap between those who engaged in no other activity and those who did is 16 percentage points. This gap is attenuated among those aged 25-30 , but it still exists at 11 percentage points. Other political activities thus appear to pull citizens into voting at a much younger rate. Among those who do not protest, this difference is only made up later on by other positive factors, such as age, increasing income, increased education, etc.
Finally, we consider whether accessing information about politics over the internet is related to voting among young people. In four elections (2000, 2004, 2006, and 2008), the CES has asked Canadians if they have used the internet to access information about the respective federal election. As can be seen, those who use the internet for this information acquisition are more likely to vote. Among our younger group, participation for those who access information on the internet is 16 percentage points higher. This difference increases to 26 percentage points among those aged 25-30 . While causation could run in either direction, we do think it likely that access to the internet makes the information acquisition required to vote in an election easier and is thus logically associated with higher participation.Footnote 12
These results suggest that there are several factors which distinguish voters and non-voters among Canadian youth. Some of these factors are related to an individual's station in life. But others are related to an individual's interest and engagement in politics. To understand which factors are most important, we perform a logistic regression that considers first all socio-demographic factors and then adds in political factors. The results of this regression tell us which factors are the most important determinants of youth voting and which do not play an independent role.Footnote 13
Table A1 (in the Appendix) indicates the independent effects of these different factors and Table 4 summarizes the impact of the most influential factors. The first set of results focuses on socio-demographic characteristics. The results show that the two most important factors are education and place of birth. The better educated have odds of voting 52% higher than those who do not have post-secondary education while those born in Canada have odds of voting over nonvoting 61% greater than those born outside the country. Then there are four other factors, age, income, gender, and residence, with some modest effect. Indeed, we find that those with an income greater than $40,000 are more likely to vote. Similarly, those who are aged 25-30 are more likely to vote than those aged 18-24 . We should note, however, that much larger age effects likely exist between those aged over 30 and those considered here. We find that women appear less likely to vote. Likewise, those in rural settings appear less likely to vote, on average. However, these last two effects do not persist when we control for other political factors.
When we consider socio-demographic and political factors together, we also see that the impact of education is substantially reduced, such that the result becomes statistically insignificant. This indicates that the higher turnout of those with post-secondary education is due mostly to their higher level of attention to political affairs. The findings reported in Table 4 indicate that the decision to vote or not to vote is strongly shaped by one's degree of interest and information. Those who indicate a medium or high general interest in politics (about 47% of those aged 18-24 and 61% of those aged 25-30 ) are most likely to vote. Indeed, their odds of voting are 88% higher than those youth who do not express a high interest in politics. Information about politics has a similarly large effect. Among those individuals who are evaluated as having a medium or high amount of information (48% of those aged 18-24 and 59% of those aged 25-30 ), the odds of voting are 89% higher than those who have a low amount of information.Footnote 14 The two largest effects, then, are related to intellectual engagement in politics. In fact if we combine information and interest into an "engagement" factor, we find that the odds of voting are more than three times as high in the high engagement group (result not shown).
Our findings also suggest that the lower turnout rate observed among women is similarly related to their lower level of interest. Indeed when interest and information are taken into account, there is no gender or rural/urban gap. The situation is different, however, with respect to income and origin. In these cases, the initial relationship is maintained after the introduction of interest and information. It may well be that those who are relatively well-off and born in Canada are more integrated into their community. This is consistent with the view that the act of voting expresses in part one's social and psychological identification with society.
|Data are drawn from the 1997, 2000, 2004, 2006 and 2008 Canadian Election Studies. Each cell presents the change in probability of an individual voting given a characteristic. The estimates are drawn from the model reported in Table A1.|
|Socio-Demographic Factors||Odds Ratio Effect|
|Model 1 – Without Political Factors|
|Some post-secondary education||1.52|
|Income greater than $40,000||1.27|
|Model 2 – With Political Factors|
|Information about politics||1.89|
|Interest in politics||1.88|
|Income greater than $40,000||1.26|
Taking all of this information together, we can draw a profile of the typical average youth voter and non-voter. The average youth voter is both interested in and informed about politics. The average non-voter is not. Demographically, the average voter likely lives in a wealthier household and is more likely to be born in Canada. All other factors do not help us distinguish voters from non-voters.
Return to source of Footnote 7 By pooling our datasets, we make the assumption that socio-demographic and political factors have the same effect in each election. Because of the small number of young respondents in each individual study, it would be difficult to conduct election-specific analysis. However, we have examined regressions for each year separately and we do not find significant differences between elections. Accordingly, we rest with the pooled analysis.
Return to source of Footnote 8 Stoker and Jennings (1995) show that getting married initially depresses electoral participation but that its long term impact is positive. The Canadian data are consistent with such an interpretation.
Return to source of Footnote 9 Interest in politics is assessed by asking individuals how interested they are in politics generally. Individuals can give a response between 0 (not interested at all) and 10 (very interested). Those who answer four or higher are considered to have medium or high interest.
Return to source of Footnote 10 Political information is measured in two manners. In 2000, it is based on a respondents' general political knowledge, their ability to correctly identify party leaders, and their knowledge of parties' positions and promises, as well as their ability to provide an approximation of the federal surplus. In all other years, political knowledge was assessed by the interviewer. Bartels (1996) has argued convincingly that interviewer evaluations of knowledge are valid indicators of respondents' political knowledge and information.
Return to source of Footnote 11 Occasionally, the differences between voters and non-voters are significantly different than in the previous version of our report. This is due to two factors. First, there was a slightly larger number of respondents in 2008 than in previous years, meaning that responses in 2008 are given somewhat more weight. Second, not all measures are exact from year to year, and so will result in greater or lesser differences between voters and non-voters. Nonetheless, all of the reported differences are in the same direction as in the previous report, suggesting consistent patterns on our key variables. This likewise applies to our regression results presented in the next section.
Return to source of Footnote 12 Two further factors, party membership and cynicism should be considered. However, they are beyond the scope of the CES. First, there is likely a positive relationship between party membership and voting. However, the CES only once measures party membership, so there is an inadequate amount of data to test this relationship. We should note, however, that party membership is known to be comparatively low among young Canadians (Cross 2004) and so some of the decline in voter turnout could be caused by less engagement with politics. However, it is equally plausible that the relationship runs in the other direction and that party membership follows from engagement in politics. Second, there is a common argument that youth vote less because they are more cynical about politics than their older counterparts. While the CES does not measure cynicism in a consistent manner year over year, we do note that Blais et al. (2002, Table 3.1, pg 51) find no relationship between cynicism and voter turnout.
Return to source of Footnote 13 We consider only those factors that were included in every CES. Accordingly, we exclude other political activities and internet usage.
Return to source of Footnote 14 We note that age is not significant in the second set of regressions, contrary to our earlier report. We note two important caveats. First, the estimated effect for age in this model is statistically indistinguishable from 0, but also indistinguishable from our previous estimated effect for age. Second, most of the age effect is captured by increases in interest and information that occur as voters age.