Canadian attitudes towards voting during the COVID-19 pandemic – Wave 3
Research Memo: Do Age and Indigenous Identity Moderate the Effects of COVID-19 on Prospective Voter Turnout?
Peter Loewen & Eric Merkley
PEARL (Policy, Elections, and Representation Lab)
Munk School of Global Affairs and Public Policy
University of Toronto
October 26, 2020
In reports 1 and 2 we provided experimental evidence that priming survey respondents to think about COVID-19 reduced their prospective intention to vote in a federal election.Footnote 1 We find these effects to be strongest among those who perceived a high level of risk from COVID-19. Here, we reanalyze data from our within-subjects experiments, conducted in waves 9 through 13 of the Media Ecosystem Observatory project, to shed light on whether effects differ by respondents' age and Indigenous identity.
We examine how much individuals reduce their turnout intention once they are exposed to large amounts of information on COVID-19. We find that younger respondents are more likely than older respondents to reduce turnout as a result of thinking about COVID-19. We fail to find differences between Indigenous and non-Indigenous respondents, though our study is underpowered, with few Indigenous respondents. We do find that less educated respondents are expected to reduce turnout more, as are urban respondents.
We conducted five nationally representative surveys with the following fielding dates:
- June 15-June 18, 2020 (N = 2,552)
- June 22-June 29, 2020 (N = 2,548)
- June 29-July 6, 2020 (N = 2,495)
- July 7-July 13, 2020 (N = 2,539)
- July 14-July 21, 2020 (N = 2,526)
Each survey had national level quotas set on age, gender, language, and region to ensure each sample's representativeness of the Canadian population. We then weight the data using an iterative proportional fitting package (ipfweight) in STATA 16 within region by age and gender. Respondents completed each survey on the Qualtrics platform.
Each survey runs between 20 and 25 minutes long and is principally about attitudes and behaviours related to the COVID-19 pandemic. We treat the survey itself as a mechanism to get people thinking about COVID-19. At the start of the survey we ask the following question: "If an election was held tomorrow, how likely or unlikely would you be to vote? (response categories: certain, likely, unlikely, certain not to).
Towards the end of the survey we repeat the question with a prime about COVID-19: "Considering the current situation with COVID-19, if an election was held tomorrow, how likely or unlikely would you be to vote? In report 1, based only on the first wave's data, we find that respondents reduced their intention to vote between the beginning and end of the survey and that this was especially true among those who perceive a high level of risk from COVID-19.
In this memo, we analyze the effects of age and Indigenous identity on the reduction in turnout intention, controlling for other demographic characteristics. For more statistical power, we pool across all five in which we included our within-subjects experiment. We estimate a model predicting change in turnout intention with age (grouped as 18-24, 25-34, 35-44, 45-54, 55-64, and 65 or older), education, income, Indigenous identity, rural/urban residence, gender, and region.
The estimates are provided in Table 1. We graphically display the estimated effects for age in Figure 1. There is a significant reduction in turnout across the entire age range (p<0.001), but the effect is over twice as large for those between the ages of 18 and 24 (-0.19) as for those aged 65 and over (-0.08). Note that only 8% of our sample is under the age of 24. Consequently, our confidence intervals are wide at that end of the age range. We cannot statistically distinguish the effects of being between the ages of 18 and 24 and between 25-34 or 35-44.
Figure 1. Predicted reduction in turnout intention by age group.
Note: 95% confidence interval.
Text version of "Figure 1. Predicted reduction in turnout intention by age group."
This line graph shows the relationship between the COVID-19 prime on prospective voter turnout (y-axis) and voter age groups of electors (x-axis) on a scale of (-.25 to .05) with a 95% confidence interval:
|Age group||Predicted effect of COVID-19 prime on perspective voter turnout|
We see no effect of Indigenous identity on changes in turnout intention. It is worth noting that only 2.6% of our sample identifies as such, so we are statistically underpowered to detect mild to modest effects. These results should be treated with considerable caution.
Education and rural/urban residence are also significant predictors of changes in intention. An individual with secondary education is expected to reduce their intention more (-0.14) than someone with a bachelor's degree (-0.11), but these effects are small. An individual living in a large city is also expected to reduce their turnout more (-0.14) than someone living in a rural area (-0.09).
* p<0.1, ** p<0.05, *** p<0.01
Return to source of Footnote 1 Report 1 was released in June, 2020, while report 2 was released in August, 2020.