Canadian attitudes towards voting during the COVID-19 pandemic – Wave 4
Report prepared for Elections Canada
Peter John Loewen & Eric Merkley
PEARL (Policy, Elections, and Representation Lab)
Munk School of Global Affairs and Public Policy
University of Toronto
January 27, 2021
This report is the fourth in a series on the views of Canadians towards electoral participation during the COVID-19 pandemic. With the emergence of new strains of the Sars-CoV2 virus, the slow roll out of vaccines, and the limited efficacy of current restrictions on citizen activities, it seems likely that the pandemic will continue for at least several more months, likely beyond the end of 2021. In light of this, understanding how citizens' views are changing towards otherwise normal and everyday actions is important. Chief among these is voting in a federal election.
As with previous papers, we report here the results from a survey of a large representative sample of Canadian eligible voters. We probe their views on voting safety, on preferred voting methods, and on their willingness to work during elections.
At a topline, we find the following:
- Voting intention is holding steady from earlier waves. While the share of Canadians likely to vote were an election called tomorrow has not increased measurably since the summer of 2020, it also has not declined.
- While turnout likelihood has not increased, a likely leading indicator of turnout has. We find marked increases in the perceived safety of voting in person compared to earlier waves of our study. Key to these increased perceptions is confidence that masking will be observed at polling places.
- When we explore the willingness of Canadians to work as poll workers, we find that there is a drop compared to the summer of 2019, especially among older Canadians. This demographic is key in polling place staffing. However, we do find that this decline can be addressed by stressing the importance of poll work.
- People are generally more inclined to work at the polls on weekends, and this is particularly true for younger respondents.
We note that throughout our report, we do explore whether turnout would be enhanced by expanding voting to a two- or three-day window. We find little evidence for this.
In what follows, we introduce our data, present our topline results, present more rigorous statistical tests, and then conclude.
Methodology – Survey Design
Our study relies on a survey of 2,898 adult Canadians conducted between January 11 and January 24, 2021. Our sample was provided by Dynata Inc, an international survey sample provider who control a large number of proprietary survey panels. Our sample was constructed with national level quotas for age, gender, region, and language. We then weight the data using an iterative proportional fitting package (ipfweight) in STATA16. note 1 Respondents completed the survey on the Qualtrics platform.
Our survey is principally focused on COVID-19. The survey also contains a battery of questions on past political participation, leader evaluations, and the like. The survey is non-partisan and no elements concerning partisan preferences have been analyzed for this report or shared with Elections Canada. In addition to COVID-19 related questions, respondents were also asked to answer a number of questions designed in consultation with Elections Canada.
In the first section of the survey, respondents were asked the following question where some respondents were asked to consider their answer in light of either a two-day or three-day election period. This question was repeated towards the end of the survey as well:
If a federal election was held tomorrow, how likely or unlikely would you be to vote?
For the next federal election, polling day might be held over three days (Saturday, Sunday, Monday) instead of the usual one day (Monday). If a federal election was held tomorrow, how likely or unlikely would you be to vote?
For the next federal election, polling day might be held over two days (Saturday and Sunday) instead of the usual one day (Monday). If a federal election was held tomorrow, how likely or unlikely would you be to vote?
- Certain to vote
- Likely to vote
- Unlikely to vote
- Certain not to vote
In the final section of the survey, respondents were asked 5 questions in addition to the second iteration of the vote intention question from above. Respondents were randomized to one of the following three questions based on the randomization from question 1:
- Now imagine a federal election was to be held in the coming weeks. If you were to vote, and considering the current situation with COVID-19, what voting method do you think you would use? [select only one]
Now imagine a federal election was to be held in the coming weeks, with three polling days instead of one. If you were to vote, and considering the current situation with COVID-19, what voting method do you think you would use? [select only one]
Now imagine a federal election was to be held in the coming weeks, with two polling days instead of one. If you were to vote, and considering the current situation with COVID-19, what voting method do you think you would use? [select only one]
- At the polling station on election day ([None/Saturday, Sunday, Monday/Saturday, Sunday])
- At the advance polling station one week before the [election/election days]
- At a local Elections Canada office before [election day/election days]
- By mail
- Other (please specify)
- Don't know yet
Respondents were then asked:
- Given the current situation with COVID-19, how unsafe or safe do you think it is to vote in person at a polling station?
- Very unsafe
- Somewhat unsafe
- Somewhat safe
- Very safe
Respondents who did not respond "very safe" were then asked:
- Which of the following protective measures would make voting at a polling station safer? [select all that apply]
- Controls on the number of people allowed into the polling station at a time
- Social distancing measures to keep people at least 2 meters apart
- Requiring everyone to wear a mask while in the polling station
- Single-use pencils
- Provision of hand sanitizer at all polling stations
- Early voting hours for electors at risk
- Other (please specify)
- None of the above
Finally, respondents were also asked to rate their level of interest in working at the polls on election day. One group of respondents received information indicating they would get paid, another group was communicated the importance of the job without reference to pay, and another group received information indicating they would get paid and that Elections Canada would have safety measures in place to protect workers. A final group – the control condition – received none of the above pieces of information:
- During an election, Elections Canada hires and pays local people to work at the polls in their area. How interested would you be in working at the polls in an election?
During an election, Elections Canada needs local people to work at the polls in their area; without them it would be impossible to hold elections. How interested would you be in working at the polls in an election?
During an election, Elections Canada hires and pays local people to work at the polls in their area. How interested would you be in working at the polls in an election, knowing that Elections Canada would put all necessary measures in place to ensure the safety of its workers?
During an election, Elections Canada recruits local people to work at the polls in their area. How interested would you be in working at the polls in an election?
- Very interested
- Somewhat interested
- Not very interested
- Not at all interested
- Don't know
Those that indicated they were very, somewhat, or not very interested in working at the polls were asked which days they would prefer to work:
- If you were to work at the polls and were able to choose the days you work, when would you be interested and/or available? [Select all that apply]
- I'm not interested in working any of these days
- Don't know
We begin our analysis by considering the "topline" responses to each of our six questions. Table 1 presents our respondents' self-estimated likelihood of voting in the next federal election if it were held tomorrow, which was asked at the beginning of our survey. Some respondents were either asked to consider a three-day election period, a two-day period, or were given a standard vote intention question. We include the control conditions for the previous three waves as a point of comparison.
|Wave 1||Wave 2||Wave 3||Wave 4|
|Baseline||Baseline||Baseline||One-day election||Two-day election||Three-day election|
|Certain to vote||73.4||75.1||75.2||74.9||74.2||73.6|
|Likely to vote||19.1||17.2||17.7||17.4||18.6||18.5|
|Unlikely to vote||4.6||4.7||5.0||5.3||4.3||5.7|
|Certain not to vote||2.9||3.1||2.1||2.5||2.8||2.3|
As the data show, there has been stability in vote intention over all four waves. There is virtually no change. There also appears to be little difference in vote intention for those told about a possible two- or three-day voting periods. We will conduct a formal test of whether there is a difference across conditions in the next section.
|Start, one-day election||End, one-day election||Start, two-day election||End, two-day election||Start, three-day election||End, three-day election|
|Certain to vote||74.9||69.4||74.2||70.8||73.6||70.9|
|Likely to vote||17.4||21.2||18.6||20.5||18.5||19.4|
|Unlikely to vote||5.3||6.2||4.3||5.6||5.7||7.1|
|Certain not to vote||2.5||3.1||2.8||3.1||2.3||2.6|
We re-asked our vote intention question towards the end of our survey in order to force our respondents to think carefully about the COVID-19 pandemic. Table 2 plots the side-by-side of vote intention at the beginning and towards the end of our survey. There is a decline in vote intention – consistent with previous waves – across all three treatment groups. In the next section we will conduct a formal test as to whether there are statistically significant differences in the decline in turnout intention between conditions.
|Wave 1||Wave 2||Wave 3||Wave 4|
|Elections Canada office||4.7||5.1||3.9||4.2|
Note: Wave 4 results omits those primed to consider two-day or three-day election periods to maintain comparability with previous waves.
When we asked half of respondents about a preferred voting method, we find substantial preferences for non-in person voting methods. These results are shown in Table 3. While the majority of individuals indicate that they would vote in person, either at a polling station (28.7%), at an advance polling station (24.1%), or at an Elections Canada office before election day (4.2%), a substantial share of individuals (29%) indicate that they would prefer to vote by mail, which has seen a modest increase from prior waves. Just one-in-seven respondents (13.2%) indicate that they do not know how they would vote. We examine whether there are statistically significant differences in preferred voting method across 1- 2- and 3-day election periods in the next section.
When respondents are asked to indicate how safe they think voting in person at a polling station will be given the situation with COVID-19, we find that the majority indicate that they regard it as either very safe (18.8%) or somewhat safe (45.4%). A little over a quarter (26.6%) regard voting as somewhat unsafe, while a little less than one-in-ten (9.2%) regard the practice as very unsafe. This is notable improvement from the last time we asked this question in wave 1 between June 15 and 18, 2020 when a majority of Canadians perceived in-person voting as unsafe (54.8%). These comparisons are shown in Table 4.
|Wave 1||Wave 4|
We asked individuals who felt that voting was somewhat safe, somewhat unsafe, or very unsafe, what protective measures would make voting at a polling station safer. Table 5 presents these frequencies for each item.
|Precaution||% indicating it would increase safety|
|Wave 1||Wave 4|
|Social distancing measures to keep people at least 2 meters apart||83%||82%|
|Provision of hand sanitizer at all polling stations||82%||78%|
|Controls on the number of people allowed into the polling station at a time||81%||79%|
|Requiring everyone to wear a mask while in the polling station||74%||82%|
|Early voting hours for electors at risk||64%||66%|
Note: Two percent of respondents selected "other" and offered an open end response. Six percent selected none of the above. Question was not asked for people who felt voting was "very safe".
Social distancing measures were seen as the most effective means to improve safety, chosen by 82% of respondents who did not feel in-person voting was very safe. Mandatory masks were similarly chosen by 82% of these respondents, followed closely by numerical limits (79%) and hand sanitizer provision (78%). Early voting hours (66%) and single-use pencils (62%) trailed on this measure. These numbers are extremely similar to those from wave 1 conducted in June of 2020 with one exception: we see an 8 point bump in the share of people thinking mandatory masks will make in-person voting safer.
Finally, we ask our respondents their level of interest in working at the polls on election day. Respondents were randomly assigned into four conditions: one emphasizing pay, another emphasizing the importance of poll workers, another noting pay and safety measures, and a control condition with none of these features. This question was asked to over 25,000 Canadians in a previous survey fielded from June 12 to July 14, 2019, where the sample was split between the pay and importance conditions we use here. We also asked this question in wave 3 fielded in October of 2020. We combine respondents in both conditions in each survey to provide a comparison.
|Base (All)||Wave 3 (All)||Wave 4 (All)||Base (18-34)||Wave 3 (18-34)||Wave 4 (18-34)||Base (35-54)||Wave 3 (35-54)||Wave 4 (35-54)||Base (55+)||Wave 3 (55+)||Wave 4 (55+)|
|Not at all||25.4||30.0||31.0||24.2||20.4||20.5||27.3||29.9||29.5||24.7||37.0||39.5|
Table 6 provides some evidence that interest in working at the polls has decreased since before the pandemic. Whereas in 2019 51% indicated some interest in working at the polls, now only 46% and 47% report the same inclination in waves 3 and 4, respectively. It is important to include a caveat to this direct comparison: the sample mode and provider are different between these two periods. We break this down by age group. The drop in interest is found primarily among older respondents. Individuals 55 and older used to exhibit the most interest in working at the polls. This is no longer the case. Indeed, this group demonstrates the largest drop in interest in working at a poll on election day. Previously, a majority were interested in working. A majority is now uninterested. We see no meaningful change in this dynamic since wave 3 conducted in October of 2020.
|Control||Pay||Importance||Pay & Safety Measures|
|Not very interested||16.5||16.2||15.7||16.3|
|Not at all interested||35.3||32.7||29.3||32.6|
Table 7 provides the breakdown of our Wave 4 results by treatment condition. 41% of respondents express some level of interest in working at the polls in the control condition, which is slightly higher in the pay (45%), pay & safety measures (46%), and the importance condition (48%). We return to this below with a formal test of statistical significance.
Among respondents who expressed some willingness to act as poll workers (i.e. any person who didn't answer "don't know" or "not at all interested"), 48% preferred to work on Saturday, a nearly identical 48% were willing to work on Sunday, while 40% said they would work on Monday. 15% selected none of these days and a further 8% said "don't know". 41% were willing to work either day on the weekend, while 28% were willing to work either of the three days.
|Interest =||All groups (Not very, Somewhat, Very interested)||Very interested|
|None of these||12.7||14.7||18.8||4.0||2.1||2.9|
Table 8 provides the breakdown by age groups. Younger respondents are more reluctant to choose all three days than those who are 55 or older, but they are comparatively more reluctant to choose Monday, while older respondents are more indifferent between the days of the week for working at the polls. These differences could be, in part, due to the fact that interest in working at the polls varies by age group. So the last three columns present responses only for those who report they were "very" interested in working at the polls. The same general finding holds. Younger individuals are less inclined to choose each of the three days in question but are comparatively more reluctant to select Monday.
Vote Intention Experiment
Respondents were randomly assigned into three groups. One group received a standard question asking how certain they are to vote if an election were held tomorrow. Another group was asked this question in light of a possible two-day election period, while a final group was asked about their vote intention in the event of a three-day election period. Table 2 showed there to be little difference in vote intention across the three conditions in responses taken at the beginning of the survey. And this bears out in the formal test. There is no statistically significant difference between the control condition and either the two-day (0.00 on a 0-1 scale of vote intention, p=0.71) or the three-day election period (-0.00, p=0.71).
We also ask vote intention towards the end of the survey after respondents were exposed to a 20 minute survey principally about COVID-19. In previous reports, and in this wave, we see evidence of reduced turnout intention between the beginning and end of the surveys. Knowledge that there may be an extended window to vote in-person may mitigate the negative effects of COVID-19 on turnout intention. However, we see no evidence that this is the case. We see a significant reduction in turnout intention from the beginning to the end of the survey in the control condition (-0.02, p<0.01), but this reduction is not statistically significant from those experienced in either the two-day (0.01, p=0.28) or three-day election treatment conditions (0.01, p=0.14)
Vote Method Experiment
Our question asking respondents their preferred voting method was randomized in a way that was tied to the vote intention question. Some respondents were asked their preferred method if an election were held over the weekend or over the weekend and Monday, while others received the control question asking about a typical one-day election. We can see whether the addition of supplementary voting days has any bearing on the method respondents use to vote.
We estimate a multinomial logistic regression model where the categorical voting method item is regressed on the experimental treatment (i.e. one-, two-, or three-day election). We generate the model marginal effects in Figure 1. These can be interpreted as the effects of the two-day or three-day election period, compared to a one-day election on the probability of choosing any of the four voting methods. We adjust the confidence intervals for multiple comparisons. Two- and three-day election periods appear to reduce intention to vote by mail. However, election day voting doesn't notably increase in response to the treatments. It is not obvious why voting at Elections Canada offices or at advance polls would instead see marginal increases in response to the treatments.
Figure 1. Marginal effect of two-day and three-day election periods on the probability of voting on election day (top-left), voting in an advance poll (top-right), voting at an Elections Canada office (bottom-left), and voting by mail (bottom-right). Note: 95% confidence intervals adjusted for multiple comparisons using the Bonferroni method.
Text version of "Figure 1. Marginal effect of two-day and three-day election periods on the probability of voting on election day (top-left), voting in an advance poll (top-right), voting at an Elections Canada office (bottom-left), and voting by mail (bottom-right)."
These four difference-of-means graphs present the effect of treatments on the probability of electors voting (Y-axis, -.1 to .15 scale) using four different voting methods: Vote at a poll on election day (top left graph), vote at an advance poll (top right graph), vote at an Elections Canada office (bottom left graph) and vote by mail (bottom right graph). The first data point on the X-axis is "Three days" election period and the second data point on the X-axis is "Two day" election period. The breakdown of the respondents' answers are as follows:
|Three days||Two days|
|Upper level of confidence||.04||Upper level of confidence||.07|
|Lower level of confidence||-.05||Lower level of confidence||-.08|
|Three days||Two days|
|Upper level of confidence||.02||Upper level of confidence||.17|
|Lower level of confidence||-.02||Lower level of confidence||.07|
|Three days||Two days|
|Upper level of confidence||.05||Upper level of confidence||.06|
|Lower level of confidence||0||Lower level of confidence||-.02|
|Three days||Two days|
|Upper level of confidence||-.01||Upper level of confidence||-.01|
|Lower level of confidence||-.13||Lower level of confidence||-.12|
Determinants of Election Safety Perceptions
Our data allow us to see which groups of Canadians are more or less likely to perceive risk in in-person voting. We estimate two models. The first predicts perceptions of election day safety with demographic characteristics such as region, gender, age, location of residence, 2018 household income, education, and whether they (or someone in their household) are personally at risk of complications from COVID-19. The second model includes those same characteristics, while adding level of COVID-19 risk perceptions and whether or not the respondent voted in the 2019 federal election (1=Yes). We rescale all of our variables from 0-1 for ease of interpretation. The estimates are shown in Table 9. Descriptions of our variables can be found in Appendix A.
|COVID-19 risk perception||-0.27***||0.02|
|Voted in 2019||0.14***||0.02|
Note: * p<0.1, ** p<0.05, *** p<0.01. Safety perceptions scaled from 0=very unsafe to 1=very safe
According to model 1, respondents at risk of COVID-19 complications score 0.03 points lower in their assessment of the safety of in-person voting on its 0-1 scale (p=0.05). Respondents with a household income of over 250,000 dollars score 0.09 points higher than those earning no income in their safety perceptions (p<0.05). A respondent aged 65 is expected to score 0.03 points higher than a respondent aged 35 in their safety perceptions (p<0.01). Women score 4 points lower in their evaluations of the safety of in-person voting (p=0.01). In contrast to our estimates from Wave 1, urban residence is no longer negatively associated with safety perceptions.
Model 2 includes COVID-19 risk perceptions and past voting patterns. People who voted in 2019 score 0.14 points higher in their election safety perceptions (p<0.01), while perceived threat from COVID-19 is powerfully associated with these perceptions. Respondents who perceive a very serious threat to both themselves and Canadians – some 24% of the sample – score 0.27 points lower in their evaluations of in-person voting safety compared to those who perceive no threat to either themselves or Canadians, who comprise only 3% of the sample (p<0.01). To put this result another way, if COVID-19 risk perceptions increased by 10 percentage points in the aggregate, we would expect perceptions of in-person voting safety to decrease by almost 3 percentage points all things being equal. Canadians who are very worried about COVID-19 are pessimistic about the safety of in-person voting.
Interest in Working at the Polls, Experimental Analysis
We asked our respondents their level of interest in working at the polls on election day. We randomly assigned our respondents into four groups: one receiving information that the position is paid, another emphasizing the importance of the role, another informing respondents both about pay and about safety measures that are likely to be in place, and a control condition with none of the above information. We rescale our outcome variable from 0-1 where 1 is for those who are very interested in working at the polls. 0.5 is for respondents who selected "don't know". We achieve similar results when instead excluding respondents who indicated "Don't know".
We find no difference between the control condition and treatment condition focused on pay (0.02, p=0.22) and no significant difference between the control and the pay and safety measures condition (0.03, p=0.18). The importance condition did, however, appear to increase willingness to work at the polls with a 0.05 improvement on the 0-1 scale compared to the control condition (p=0.01), which was not the case when this experiment was run in wave 3.
This report provides an important update to our previous reports. As in our earlier reported experiments and models, we find general support for the proposition that the COVID-19 pandemic is providing downward pressure on voter turnout, and that considering the pandemic and its safety risks are key factors in that pressure. But we have also found that implementing and informing citizens about the safety measures being taken by Elections Canada can reduce that depression. Our results are much the same here.
We note three encouraging findings. First, vote intention is holding steady, despite the COVID-19 pandemic arguably become more threatening and stretching out longer than expected. Second, most people view voting as safe, and this perception has increased since earlier waves. Third, we have found increased belief that masking will make voting safer. Given the near universal usage of masks, this may help in combatting downward pressure on turnout even more.
We also explored the willingness of individuals to work at polling places. We find that willingness to work at polling places has decreased since before the pandemic, but this decreased willingness has occurred almost entirely among older Canadians. However, we find that this can be addressed through stressing the importance of engagement as a poll worker – something we did not find in the previous wave.
Appendix A – Variable Descriptions
|COVID-19 risk||"Is anyone in your household in a high-risk group for which the annual seasonal influenza vaccine would usually be recommended by the Public Health Agency of Canada? (These conditions include, individuals who are pregnant or those with chronic respiratory disease, chronic heart disease, chronic kidney disease, chronic liver disease, chronic neurological disease, diabetes (all types), cancer, immunosuppression, dysfunction of the spleen, and/or BMI > 40)";
|Education||No schooling, some elementary, elementary, some secondary, secondary, some technical, technical, some university, bachelor's degree, master's degree, doctorate or professional, rescaled 0-1, don't knows coded as missing|
|Income||No income; 1-30,000; 30,001-60,000; 60,001-90,000; 90,001-110,000; 110,001-150,000; 150,001-200,000; 200,000+, rescaled 0-1, don't know/prefer not to answer coded as missing|
|Age||In years, rescaled 0-1 from minimum (18) to maximum (99)|
|Location||Rural area, small town, large town, mid-sized city, large city, rescaled 0-1|
|Region||1=Atlantic; 2=Quebec; 3=Ontario; 4=West|
|COVID-19 risk perception||
|Voted in 2019||1=Yes|
|In-person voting safety perceptions||Very unsafe, somewhat unsafe, somewhat safe, very safe rescaled 0-1|
|Voting turnout||Certain to not vote; unlikely, likely, certain to vote, rescaled 0-1|
Return to source of Footnote 1 Minimum and maximum untrimmed weights were 0.75 and 1.5. The average weight is 1 and the standard deviation is 0.10. All the results presented below replicate substantively in the absence of weights.