open Secondary menu

Youth participation and cynicism during the Canadian federal election of 2015

Note to the Reader

This report was presented at the conference “Youth Political Participation: On the Diverse Roads to Democracy,” June 16–17, 2016, Montreal, Quebec.


A Report Commissioned
by Elections Canada

Ioana Alexandra Manoliu and
Katherine V. R. Sullivan

Department of Political Science
University of Montreal

June 2016

As was shown in a previous report (Gélineau, 2013), Canadian youth participation is affected by many factors, such as political interest and political participation, but also by one's level of political cynicism, which can be defined as "a mistrust generalized from particular leaders or political groups to the political process as a whole – a process perceived to corrupt the persons who participate in it and that draws corrupt persons as participants" (Cappella & Jamieson, 1997: 166). Some have already pointed to the potential importance of a "culture of cynicism" in our contemporary democratic societies (Chaloupka, 1999; Goldfarb, 1991), and it is reasonable to expect that it may also have an impact in Canada, especially among youth. Exploring the potential impact of cynicism in Canada will also help us broaden Gélineau's (2013) previous work.

We use data from the National Youth Survey conducted in 2015 to analyse the potential effects of cynicism on youth participation during the last federal election. Our report is divided in two sections. Firstly, we aim to identify explanatory factors of cynicism by looking at socio-demographic characteristics, general attitudes toward the political system, and other predictors associated with cynicism, and we focus our attention on the impact of cynicism on electoral participation. Secondly, we examine potential consequences of cynicism on general democratic attitudes.

In this report, cynicism is assessed using a variety of questions tapping respondents' general feelings towards democracy, political parties, politicians and voting. Cynicism was operationalized following Rubenson et al. (2004) and was measured by combining four questions: a measure of trust in the government "I do not think government cares much about what people like me think"; a measure of trust in political parties "All federal parties are the same"; and two scales from 0 to 100 rating how respondents feel about politicians in general and political parties.Footnote 1

Part 1 - Explanatory factors of cynicism

In the following section, we explore various factors that could explain cynicism among Canadian youth. We first look at the impact of socio-demographic characteristics, and then move to the impact of general attitudes toward the political system.

a. Socio-demographic characteristics

In Table 1, we first look at the mean levels of cynicism according to a variety of socio-demographic characteristics. Respondents between the ages of 18 and 19 have a mean level of cynicism of 40.69 on the scale that ranges from 0 to 100. Cynicism is slightly higher among respondents from 20 to 24, as well as those between the ages of 25 and 29, which respectively have means of 43.00 and 43.44. Respondents of 30 to 34 years of age are the most cynical, with an average of 50.40, while cynicism appears to be somewhat lower among older respondents. Additionally, men (43.45) are slightly less cynical than women (45.40), but the difference is not statistically significant. Income does not seem to have much impact, as all income groups exhibit similar levels of cynicism. In line with the results obtained by Agger et al. (1961), the level of cynicism lowers with the level of education completed. Namely, respondents with post-secondary education (43.73) and full-time students (42.13) show lower levels than employed youth. Alternatively, unemployed respondents (50.46) and those having not completed post-secondary education (46.38) show greater levels of cynicism.

Table 1 Footnote 2 : Cynicism by Socio-Demographic Characteristics Footnote 3
18–19 40.69
20–24 43.00
25–29 43.44
30–34 50.40
35–older 41.07
Men 43.45
Women 45.40
<$40,000 45.11
>$40,000 43.19
High school 46.38
Post-secondary 43.73
Employed or other 44.03
Unemployed 50.46
Student 42.13
English 44.52
French 45.65
Other 40.83
Not Aboriginal 44.15
Aboriginal 48.55
Visible minority 45.60
Not visible minority 43.59
Born in Canada 44.61
Born abroad 42.93
Not disabled 43.97
Disabled 49.36
Atlantic  39.11
Quebec 45.30
Ontario 45.97
Prairies 41.34
Alberta 44.38
British Columbia/Territories 42.47
Living Environment
Rural 45.82
Urban 43.88

In terms of language and ethnocultural differences, both suggest that respondents whose first language is neither French nor English are least cynical (40.83), as well as respondents born abroad, with a mean of 42.93. Additionally, disabled respondents appear to be more cynical (49.36) than Canadians who are not disabled (43.97). Furthermore, Aboriginal respondents display a high level of cynicism (48.55), which is higher than non-Aboriginal respondents (44.15).

Finally, our results suggest some regional differences, as respondents living in the Atlantic region are the least cynical, with a score of 39.11, whereas Ontarians (45.97), Quebeckers (45.30) and Albertans (44.38) are the most cynical. Also, as per statistics listed above, the difference between rural (45.82) and urban (43.88) living environments is rather small and not statistically significant, suggesting that Canadians living in a city are not a lot more cynical than their rural counterparts.

According to our findings as presented in Table 1, the socio-demographic categories of the population that seem most prone to showing higher levels of cynicism are women, Canadians between the ages of 30 and 34, those having an annual income of less than $40,000, respondents having not completed post-secondary studies, and those who are unemployed. Additionally, our results suggest that categories of the population that speak French as a first language, are Aboriginal, are part of a visible minority, were born outside of Canada, are disabled, or who live in a rural area or in the provinces of Quebec, Ontario or Alberta are also more cynical.

b. General attitudes toward the political system

In an attempt to better understand youth participation in relation to cynicism, we focus on general attitudes toward the political system by looking at the respondents' sense of civic duty, interest in the election, general political interest, political knowledge, self-efficacy, socialization, political discussion and civic education. Table 2 reports the mean levels of cynicism according to these different attitudes for all respondents. Since all the variables but the sense of civic duty are continuous, only the low and high categories are reported.Footnote 4

Table 2 Footnote 5 : Cynicism by General Attitudes toward the Political System
Sense of civic duty
Choice 48.19
Duty 40.43
Interest in the election
Interested 41.62
Not interested 63.54
General political interest
Interested 41.02
Not interested 58.56
Political knowledge
Low 54.10
High 37.39
Low 60.70
High 39.23
Never 51.59
Socialized 41.64
Political discussion
Low 66.64
High 40.63
Civic education
Low 48.19
High 39.93

Results as shown in Table 2 suggest that those who believe voting is a civic duty (40.43) rather than a choice (48.19) are less cynical. Additionally, respondents with higher levels of political knowledge, having correctly answered questions pertaining to Canadian politics, show lower levels of cynicism (37.39) compared to their less-informed counterparts (54.10). This is in line with respondents' perception of self-efficacy, in the sense that those with low levels of self-efficacy (60.70) are nearly twice as cynical as respondents who believe their vote will make a difference (39.23).

In terms of the influence of the social environment when growing up, respondents having been more exposed to political topics during discussions at home, with friends or at school all show lower levels of cynicism. Namely, those having discussed political topics when growing up (40.63) are much less cynical than those having not (66.64). These results are similar to those pertaining to civic education. Respondents having followed a course on civic education are much less cynical (39.93) than their counterparts (48.19). Overall, members of the electorate who are less interested in political matters and have lower levels of political knowledge and self-efficacy tended to be more cynical. Furthermore, those having a less politically stimulating social and educational environment also tend to be more politically cynical.

In general, the differences in cynicism with respect to behavioural variables cited above are much stronger than results observed with socio-demographic variables in Table 1. Hence, our results suggest that cynicism is highly related to attitudinal factors. Since these attitudes are known to be related and to share similar predictors, multivariate analyses are required to parse out the impact of these variables on cynicism.

c. Predictors of cynicism

In order to better explain the factors that are connected more or less strongly with cynicism, we ran regression models in which cynicism is the dependent variable. Model 1 includes all the socio-demographic factors already mentioned, while Model 2 takes into account the attitudinal variables seen above. These models allow us to examine the specific impact of each independent variable on cynicism, while taking into account all the other variables in the model. All independent variables were coded to range from 0 to 1 so as to report the maximum impact of each variable on cynicism. Hence, coefficients represent how much more or less cynical respondents are when they have the maximum value on the independent variable. A coefficient that can confidently be distinguished from 0 is said to be statistically significant, meaning that we can generalise the effect of this variable from the dataset to the general population.

Cynicism is coded to range from 0 to 100; hence all coefficients represent the maximum impact of the dependent variables on this scale. In Model 1, we see that, taking all other variables constant, women score on average 2.07 points higher on the cynicism scale than men. This difference is statistically significant. The most educated respondents are, on average, 3.71 points less cynical than the least educated, and this coefficient is significant. Similarly, respondents with income over $40,000 exhibit 2.20-points less cynicism, and this is statistically significant. Looking at the age variables, we see that in Model 1, only respondents between the ages of 30 and 34 have a significantly different level of cynicism than respondents aged 35 or older. Respondents aged 30 to 34 exhibit 9.49 points more cynicism on our scale ranging from 0 to 100. Hence, without taking the attitudinal variables into account (this is done in Model 2), it appears that only those between the ages of 30 and 34 are more cynical than their older counterparts. Furthermore, Francophones are not significantly more or less cynical than Anglophones, but respondents who speak a language other than English or French at home are significantly less cynical than Anglophones (2.98 points). Finally, when all other independent variables are taken into account, respondents living in rural areas are significantly less cynical than urban respondents (2.09). However, respondents having a disability are significantly more cynical (4.89 points).

Model 2 takes into account attitudinal variables, which allows us to examine whether the effects of socio-demographic variables in Model 1 are spurious – meaning that they are, in fact, the product of other attitudinal variables – or hidden by these attitudes that are also linked to socio-demographic characteristics. In Model 2, women do not exhibit significantly higher levels of cynicism (0.66 points) than men. This suggests that women are not really more cynical than men when we take the attitudes included in the model into account. Education is no longer significant in Model 2, while it was in Model 1, which means that the more educated are not more cynical when the attitudes fostered by education are taken into account. The impact of income also becomes insignificant, which means that income is not related to cynicism when the attitudinal variables are accounted for. With respect to the age of the respondents, the results in Model 2 are somewhat different than in Model 1, in which attitudes were not accounted for. In Model 2, respondents who are 18 or 19 years old are less cynical (2.98 points) than those aged 35 and older. Respondents between the ages of 20 and 24, as well as those of 25 to 29 years old, are not more or less cynical than respondents of 35 years of age or older. Yet respondents between the ages of 30 and 34 are on average significantly (5.93 points) more cynical than those of 35 or older. Results regarding language are somewhat similar to what they were in Model 1, meaning that respondents who speak a language other than English or French exhibit less cynicism than Anglophones, and the difference is rather small (2.43 points), without reaching statistical significance. In Model 2, those living in rural areas do not exhibit more or less cynicism than urban respondents, while in Model 1 they were significantly less cynical. Finally, respondents having a disability or who identify as Aboriginal are significantly more cynical, by 3.61 and 3.48 points, respectively.

Table 3: Predictors of Cynicism – Linear Regression Models
Model 1   Model 2  
Female 2.07 * (0.91) 0.66 (0.86)
Education -3.71 * (1.29) 0.04 (1.16)
Income -2.20 * (1.04) -1.15 (0.95)
18–19 -2.51 (1.73) -2.98 (1.62)
20–24 0.85 (1.49) -1.36 (1.39)
25–29 1.82 (1.52) 0.18 (1.39)
30–34 9.49 * (1.50) 5.93 * (1.41)
First Language
French 1.99 (1.22) 1.17 (1.13)
Other -2.98 * (1.47) -2.43 (1.44)
Rural -2.09* (1.03) -1.32 (0.97)
Disability 4.89* (1.71) 3.61* (1.54)
Aboriginal 3.26 (1.72) 3.48* (1.56)
Duty -2.87 * (0.90)
Interest in election -11.38 * (1.71)
General interest -7.60 * (1.36)
Political knowledge -5.79 * (1.73)
Political discussion -0.96 * (0.21)
Constant 45.14 * (1.99) 72.62 * (2.57)
Observations 2563 2517
R2 0.054 0.197

Standard error in parenthesis.

More interesting results appear when we look at the attitudinal variables included in Model 2. All these variables yield negative and significant coefficients. Respondents who believe that voting is a civic duty are on average 2.87 points less cynical than those who believe that voting is a choice. The most knowledgeable respondents are also 5.79 points less cynical than the least informed. Looking at political interest, we see that the most interested respondents are on average 7.60 points less cynical than the least interested, and those more interested in the election are 11.38 points less cynical than their less-interested counterparts. Finally, the R-square of 0.05 in Model 1 means that the model explains about 5% (0.05) of the variation in the dependent variable, whereas Model 2 explains about 19% (0.19) of the variation. Hence, comparing the R-square between models 1 and 2 clearly suggests that the attitudinal variables have much more explanatory power than the socio-demographic variables.

All in all, interest in the election is the most important and substantive predictor of cynicism, but other variables also have meaningful influence, although arguably not of the same magnitude. Nonetheless, our results indicate that factors that are usually related to what are considered to be positive civic outcomes are also linked to cynicism. Generally speaking, more politicized Canadians tend to be less cynical.

In order to further explore the link between cynicism and political knowledge and interest, we examine the main source of information about the elections, ranging from Elections Canada (advertising, social media), the VIC, printed media, television, radio, social media, and the government to family or friends. Hence, we analyse the potential effect of media consumption on respondents' level of cynicism.

A first look at the results suggests that respondents having consulted any media – with the exception of the voter information card – are for the most part significantly less cynical than those who did not seek political information. We also use a few control variables, such as gender, education, income and age.

Table 4: Cynicism and Main Source of Information – Linear Regression Model
Elections Canada 2.72 (2.19)
Voter information card 8.75* (2.69)
Newspaper/magazine -5.74* (1.79)
Website of a media, blog or other web source -3.19* (1.29)
Social media (Facebook, Twitter, Youtube) -1.96 (1.38)
Radio -0.55 (2.59)
Family or friends 2.61 (1.96)
Other -0.85 (2.09)
Women 1.06 (0.90)
Education (post-secondary) -2.31 (1.27)
Unemployed 5.93* (2.02)
Income -2.29* (1.01)
18–19 years old -3.69* (1.81)
20–24 years old -0.07 (1.56)
25–29 years old 0.67 (1.55)
30–34 years old 7.91* (1.52)
Constant 44.75* (2.11)
N 2493
R2 0.061

Standard errors in parentheses.
*p< 0.05

In our sample, 22% of respondents reported using television as their main source of campaign information. Hence, television is used as the reference category. Our results show that respondents who report using newspapers or magazines as their main information source are about 5.74 points less cynical than those who report television as being their main source of campaign information. However, respondents who report having used the VIC as their main source of information are 8.75 points more cynical than those having consulted television for campaign information. It is important to mention that the VIC only provides very specific information about the voting process, rather than about the election in general. Respondents who report using websites as their main source of information are 3.19 points less cynical than those who mainly used television.

Furthermore, our results suggest that information sources that require respondents to actively seek campaign information – newspapers or magazines, as well as a media website – are negatively associated with cynicism, as these respondents are less cynical than the large majority of respondents who mainly used television. On the other hand, respondents who report using their VIC sent by Elections Canada as a main information source – that is, respondents who do not actively seek information – are much more cynical.

Part 2 - Consequences of cynicism on participation

In this second section, we examine the impact of cynicism on both electoral and more general political participation. In this report, electoral participation is assessed by vote frequency, meaning the number of elections during which the respondent voted, including the federal election in 2015. The level of political participation is assessed with questions pertaining to levels of involvement in current issues, such as contacting a politician to express their views on an issue, participating in a demonstration or protest march, signing a petition, displaying a sign for a party or a candidate during the election, or even participating in an event organized by a party or a candidate during the election.Footnote 6

Results as shown in Table 5 suggest that respondents having voted in the last federal election prove to be much less cynical (40.63) than those having chosen to abstain (55.14), as per statistics below. This difference is statistically significant. Furthermore, the difference between those having shown high levels of political involvement (38.60) and those having refrained from participating (45.19) is small and not statistically significant. Finally, the 17-point difference between those having voted in previous elections (38.35) and those having voted in none (55.20) is eloquent and significant.

Table 5: Cynicism and Electoral Participation – Bivariate Relation
Vote in federal election 2015
Abstained 55.14
Voted 40.63
Political participation
Low 45.19
High 38.60
Vote frequency
Low 55.20
High 38.35

Furthermore, results presented in Table 5 confirm the conventional wisdom that cynicism generally has a negative impact on the electorate. Indeed, the most politically involved and assiduous respondents appear to also be the least cynical. However, electoral and general political participation are known to be influenced by various other important factors that need to be taken into account before coming to tangible conclusions regarding the impact of cynicism. Hence, multivariate analyses are required.

Table 6 displays the results of two regression models predicting electoral participation in Model 1, and general political participation in Model 2. Since voting is a dichotomous variable – one can either vote or abstain – Model 1 displays the results of a logistic regression model. Alternatively, Model 2 displays the results of a usual OLS regression model in which the dependent variable is an additive scale ranging from 0 to 5, capturing participation in various democratic acts. In both cases, our interest lies in the impact of cynicism on both types of participation when accounting for other known important predictors.

In Model 1, cynicism has a negative impact on the propensity to vote in 2015, which is consistent with Gélineau (2013), who found a significant relationship between cynicism and youth voting levels. With respect to other known important variables, education is surprisingly not related to the probability to vote. Interest in the election, political knowledge, political interest in general, political discussion and the belief that voting is a duty rather than a choice (Blais, 2000) are all significant and have an impact on the probability to vote. In all cases, respondents under the age of 35 are less likely to vote than their older counterparts, which is rather unsurprising considering that it has now been well-established that younger citizens tend to vote less (Blais, 2000; Blais et al., 2002; Gélineau, 2013). Respondents who report speaking a first language other than English or French are also less inclined to vote.

Table 6: Cynicism and Political Participation – Logistic and OLS Regressions
Model 1 Model 2
Cynicism -1.63 * (0.33) 0.05 (0.12)
Women (men as reference) 0.06 (0.14) 0.04 (0.05)
Education 0.27 (0.15) 0.13* (0.05)
Income 0.03 (0.14) -0.09 (0.05)
Age (35+ as reference)
18–19 -1.39* (0.29) 0.15 (0.09)
20–24 -1.19* (0.25) 0.03 (0.07)
25–29 -0.91* (0.26) 0.03 (0.07)
30–34 -1.23* (0.24) 0.01 (0.07)
Language (English as reference)
French 0.02 (0.18) -0.09 (0.05)
Other language -0.47* (0.23) -0.10 (0.07)
Rural 0.11 (0.15) 0.11* (0.05)
Disability 0.27 (0.25) 0.38* (0.09)
Aboriginal 0.11 (0.25) 0.10 (0.09)
Duty 1.71* (0.14) 0.04 (0.05)
Interest in election 1.29* (0.22) 0.01 (0.07)
Political interest in general 0.43* (0.19) 0.17* (0.06)
Political knowledge 1.50* (0.25) 0.40* (0.09)
Political discussion 0.14* (0.03) 0.11* (0.01)
Constant -1.40* (0.43) -0.71* (0.14)
Observations 2493 2421
R2 0.32 0.122

Standard errors in parentheses.

The dependent variable in Model 1 is whether or not one has voted in 2015, whereas the Model 2 the dependent variable is one's general participation. Hence, Model 1 is a logistic regression model, while Model 2 is a linear regression.

p< 0.05

Model 2 displays the results of a common linear regression model predicting general political participation using the same predictors as in Model 1. Unsurprisingly, political knowledge, political interest, political discussion and education are all positively and significantly associated with the propensity to engage in a political act. Living in a rural area, as well as having a disability, also seem to be associated with general political participation. Most youth are not significantly less involved in more political acts compared to those aged 35 and older. Yet our main interest lies in the impact of cynicism, which exhibits a positive effect. Highly cynical respondents do not seem to exhibit less engagement in political acts than the least cynical ones. Hence, cynicism has a negative impact on the propensity to vote (Model 1), but does not affect the frequency with which one engages in other political acts.

Discussion and conclusion

Our report aims to shed light on both the predictors and the consequences of political cynicism on youth participation. Our results suggest that, in terms of socio-demographic characteristics, the older youth (30-to-34-year-olds) are the most cynical, as well as the least educated and least active – unemployed – respondents. Furthermore, many groups display high levels of cynicism – namely, Aboriginal respondents, respondents having a disability, respondents born abroad, and those who report speaking a language other than French or English as their first language. Furthermore, respondents who identify as a visible minority or who live in Ontario are the most cynical, followed closely by Quebec and Alberta.

Additionally, our results suggest that cynicism is highly related to attitudinal factors that are generally associated with negative democratic outcomes. In this sense, more interested and informed Canadians exhibit lower levels of cynicism, political interest being by far the most important and substantive predictor of cynicism. Voters also display less cynicism than non-voters. However, cynicism does not appear to hinder political participation in other democratic acts, such as signing a petition, contacting an elected official or protesting. From a participatory standpoint, this may be a good thing, as a healthy democracy also requires political participation that goes beyond the simple act of voting from its citizens.


Agger, R.E., Goldstein, M.N., & Pearl, S.A., 1961, Political cynicism: Measurement and meaning, Journal of Politics, 23 (3), 477–506.

Blais, A., 2000, To vote or not to vote: The merits and limits of rational choice theory, Pittsburgh: University of Pittsburg Press.

Blais, A., Gidengil, E., Nadeau, R., & Nevitte, N., 2002, Anatomy of Liberal Victory. Peterborough: Broadview Press.

Cappella, J.N., & Jamieson, K.H., 1997, Spiral of cynicism: The press and the public good, New York: Oxford University Press.

Chaloupka, W., 1999, Everybody know:. Cynicism in America, Minneapolis/London: University of Minnesota Press.

Gélineau, F., 2013, Who participates? A closer look at the results of the National Youth Survey, final report, prepared for Elections Canada.

Goldfarb, J., 1991, The cynical society: The culture of politics and the politics of culture in American Life, Chicago/London: The University of Chicago Press.

Rubenson, D., Blais, A., Fournier, P., Gidengil, E., & Nevitte, N., 2004, Accounting for the age gap in turnout, ActaPolitica, 39 (4): 407–421.


Political cynicism

The dependent variable is political cynicism, for which we have kept the same measure as Rubenson et al. (2004). Political cynicism is an index of 4 questions: Do you strongly agree, somewhat agree, somewhat disagree, or strongly disagree with the following statement? a) I do not think government cares much what people like me think (QF2c); b) federal parties are basically the same; there is not really a choice (QF2b): c) How do you feel about politicians in general? (QF3); d) How do you feel about political parties? (QF4) (the latter two being measured on a 100-point scale, 0 1/4 really dislike, 100 1/4 really like). The index is the sum of the 4 scores divided by 4 (taking values from 0 to 1).


Our age variable is divided into 5 age groups: 18–19, 20–24, 25–29, 30–34 and 35 and over. Respondents considered as youth are aged 34 and younger (as Gélineau, 2013).


Gender is a dichotomous variable where male=0 and female=1.


Income is a dichotomous variable with 2 categories: An annual income of less than $40,000 (coded 0) and an annual income of over $40,000 (coded 1).


Education is also a dichotomous variable, with 2 categories: respondents having completed up to high school education, and those having completed post-secondary education, which can be some college or commercial studies, college or commercial studies, some university studies, or a university diploma (BA, MA, PhD).


The occupation variable is divided into 3 categories: the employed respondents, the unemployed, and full-time students (those who report as currently studying or in training, as well as those who declared they are employed or self-employed and studying or training at the same time).

The following socio-demographic variables are all dichotomous: Aboriginal, visible minority, disabilities, living environment (rural and urban).


For the language variable, we have kept the original 3 categories: English, French, and those having another language as first language.


Self-efficacy (QF2D) is a dichotomous variable taking into consideration the opinions of people who "believe that by voting (they) can contribute to changing things" (coded 0 for those who answered "not at all agree" and "mostly not agree" and 1 for those "mostly agree" and "totally agree" with the fact their vote can change things).


Socialization (QG1) is a dichotomous variable (where 0 equals "never" and 1 "sometimes" or "often") derived from the question: "When you were young, to what frequency were you talking about politics or government at home?"

Civic education

Civic education is also a dichotomous variable created from an index combining 2 questions: "In secondary school did you attend classes on the government systems or politics?" (coded as"1" for yes and "2" for no) and "Have you participated in an election simulation in secondary or elementary school, for example Student Vote?" (coded "1" for yes and "2" for no).


Duty is a dichotomous variable created with an index combining 2 questions aiming to find out if the responded perceived voting as a duty or as a choice (coded "0"if it is a choice and "1" for duty).


In order to measure political interest, we have created two variables similar to those used by Gélineau (2013).

Firstly, interest in election is a dichotomous variable using this question: "In general, what was your level of interest for the last federal election?" Two categories were then created, the first being low interest ("not at all interested" and "not interested"), the second being high interest ("somewhat interested" and "very interested").

Secondly, interest in politics is similar to the first interest variable and was created by using this question: "To what degree are you interested in Canadian politics? Would you say you are 'not at all interested,' 'not very interested,' 'somewhat interested' or 'very interested'?" Two categories were once again created: low interest – those reporting having no or very little interest in politics, and high interest – those reporting being somewhat interested or very interested.

Media: main source of campaign information

The media variable was created by taking into account the main source of campaign information (creating an index of 8 options out of 13). The main declared sources were Elections Canada, the VIC, newspapers, television, media website, blog or other web source, as well as the radio and family or friends. Television was used as the reference category, as it was the majority's declared main source of campaign. The remaining options were included in the category "other sources," which refers to the government, general mail, candidates' flyers received by mail, etc.

Vote in 2015

Vote in 2015 is a dichotomous variable, coded 1 for those who have voted and 0 for those who have abstained in the last federal election of 2015.

Vote frequency

Vote frequency is a dichotomous variable, coded 0 for those who answered not having voted in any or some of the elections (federal, municipal, provincial) and 1 for those who report having voted in the majority or all of them. (QB1)

Political discussion

Political discussion is an index built by combining 2 questions (QG2B and QG2C): "Now, at what frequency do you talk about politics or government with the following persons or groups: friends/family?" As not all of the options were applicable to respondents, we have kept the friends and family options and not those mentioning the wife/husband, work colleagues or school colleagues.

Political participation

Political participation is an index combining 5 questions pertaining to political acts done in the last 12 months, including these: "communicated directly with a politician to give him your point of view on a matter"; "took part in a manifestation or a protest"; "signed a petition"; "displayed a sign-board for a party or candidate during the campaign"; "participated in an activity organized by a party or candidate during the electoral campaign" (all variables are dichotomous–coded "1" for yes and "2" for no).

Political knowledge

Political knowledge is a dichotomous variable created by using an index combining 5 questions: "Which party won the largest number of seats in the federal election of 19 October?" (QI1); "From which order of government is education issued – federal, provincial, municipal?"(QI2); "From which order of government is employment insurance issued – federal, provincial, municipal?" (QI3); "What is the name of the premier of your province/territory?" (QI4); "What job or political function does David Cameron occupy now?" (QI5).

The index was then dichotomized to create two categories: low political knowledge and high political knowledge.

Civic education

Civic education is an index combining 2 questions, one pertaining to having followed a course on government or politics in high school (dichotomous) and one to participation in an election simulation in their primary or secondary school (dichotomous).

Footnote 1 We would like to thank Alexandre Blanchet and Filip Kostelka for all their help, comments and suggestions.

Footnote 2 For all respondents.

Footnote 3 The dependent variable is political cynicism, for which we have kept the same measure as Rubenson et al. (2004). Political cynicism is an index of 4 questions: Do you strongly agree, somewhat agree, somewhat disagree, or strongly disagree with the following statement? a) I do not think government cares much what people like me think (QF2c); b) Federal parties are basically the same; there is not really a choice (QF2b); c) How do you feel about politicians in general? (QF3): d) How do you feel about political parties? (QF4) (the latter two being measured on a 100-point scale, 0 1/4 really dislike, 100 1/4 really like). The index is the sum of the 4 scores divided by 4 (taking values from 0 to 1).

Footnote 4 More details about the variables can be found in the Appendix.

Footnote 5 On a scale from 0 to 100.

Footnote 6 More details on variables are available in the Appendix.