The 2021 Canadian federal election was a “pandemic election” during which policies to address the COVID-19 crisis were a major focus of the campaign. The snap election was called two years before the next scheduled federal election. It was widely viewed as a plebiscite on the Liberal government's handling of the pandemic and as a gambit by Justin Trudeau to regain the parliamentary majority he lost in 2019. However, halfway through the campaign, Maxime Bernier and his People's Party of Canada (PPC) began attracting media attention. Founded in 2018, the PPC fared poorly in its first federal election in 2019. Bernier's campaign messaging in 2021 centred on individual freedom, taking aim at public health restrictions: lockdowns, masking requirements and vaccine mandates. PPC election rallies drew large, enthusiastic crowds, and the party experienced a rise in voting intentions, as reported by campaign-period election polls. The COVID-19 pandemic had seemingly made the PPC a party that could no longer be ignored.
The PPC's opposition to pandemic-related policies is consistent with other (mostly far-right) populist movements that sought to oppose restrictive public health policies (Brubaker, Reference Brubaker2021; Vieten, Reference Vieten2020; Wondreys and Mudde, Reference Wondreys and Mudde2022). Yet, as has been the case for most populist parties during the pandemic (Wondreys and Mudde, Reference Wondreys and Mudde2022), the PPC did not gain a windfall of seats. While it was able to improve on its showing in 2019 by increasing its vote share from 1.6 per cent to 4.9 per cent, it did not manage to win a single seat. Nevertheless, the PPC did see a sharp rise in its electoral support, which likely cost the Conservative Party of Canada (CPC) some seats in closely contested races (Grenier, Reference Grenier2021). It is therefore important to understand the factors that are related to support for the PPC in the 2021 Canadian federal election.
In light of the PPC's brief history, there is relatively little published research on the party and its supporters (but see Budd, Reference Budd2021). More broadly, there is currently limited research on the relationship between populist parties and pandemic elections. Given these gaps in the research literature, the current study aims to assess the relative importance of different factors on support for the PPC, while also contributing to the study of Canadian electoral behaviour more broadly and to our understanding of populist politics and the COVID-19 crisis.
To do so, we use data from an original post-election survey (n = 18,950) and focus on how Canadians’ vote choice was shaped by populism and opposition to COVID-19 measures. Results from regression models and structural equation models (SEMs) indicate that both populist attitudes and opposition to COVID-19 measures were associated with support for the PPC. However, the findings demonstrate that opposition to COVID-19-related public health restrictions is a much stronger factor than populism in shaping support for the PPC. Also, the results of the SEMs show that the effect of populism on PPC support was partly mediated by opposition to public health restrictions, supporting the contention that populists are strongly likely to be against COVID-19 health measures.
Populism, COVID-19 and the People's Party of Canada
Canada was certainly not spared from the COVID-19 pandemic. Different levels of government responded by imposing a range of public health restrictions to curb the spread of the virus (Hale et al., Reference Hale, Angrist, Goldszmidt, Kira, Petherick, Phillips and Samuel Webster2021). Canada also deployed a relatively quick vaccination campaign in 2021. COVID-19-related measures initially enjoyed broad support from citizens, and trust in the government was among the highest in the world (Gallichan-Lowe, Reference Gallichan-Lowe2020; Lazarus et al., Reference Lazarus, Ratzan, Palayew, Billari, Binagwaho, Kimball and Heidi J. Larson2020). These events set the context for the snap election called for September 20, 2021, by the governing Liberal Party of Canada (LPC) led by Prime Minister Justin Trudeau, who arguably sought to use the popular goodwill generated by the government's response to the pandemic as an opportunity to turn the LPC's minority government into a majority. The Liberals sought to use their success in procuring and distributing COVID-19 vaccines to their electoral advantage while employing it as a wedge issue against the CPC by promising to implement a COVID-19 vaccine mandate for federal public servants and international transportation workers, as well as for air and rail travellers across Canada (Gilmore, Reference Gilmore2021).
The duration of the COVID-19 crisis, though, generated “pandemic fatigue” among the Canadian mass public, eroding popular support for COVID-19-related measures (Bricker, Reference Bricker2021; Quirion, Reference Quirion2022). Protests against public health measures became more common across the country. This movement arguably culminated with the so-called Freedom Convoy protest of February 2022, which saw a three-week occupation of Ottawa and of Canada–US border crossings in Alberta and Ontario by angry protesters.
Over the course of the 2021 federal election campaign, Maxime Bernier and his upstart PPC focused heavily on the issue of COVID-19-related measures. The PPC was able to use the growing popular discontent over mask and vaccine mandates, as well as lockdowns, to attract large crowds to campaign rallies and increase its poll numbers (Dawson, Reference Dawson2021; Debusmann Jr., Reference Debusmann2021).
The proposal to expand vaccine mandates by the Liberals was quite controversial. The New Democratic Party of Canada (NDP) supported this proposal and even encouraged the Liberals to take it a step further by implementing it quickly and implementing disciplinary actions for federal employees who refused to get vaccinated. The Green Party of Canada (GPC) took an ambivalent position on this issue. Its then leader, Annamie Paul, repeatedly stated that the party was considering the proposal but needed more details. The Bloc Québécois (BQ) and the CPC encouraged Canadians to get vaccinated but opposed the extension of vaccine mandates to travellers and federal employees, preferring alternatives such as rapid testing for non-vaccinated individuals. The PPC, for its part, not only opposed vaccine mandates but also opposed rapid testing strategies and promised to fight “authoritarian” measures imposed by provincial governments. Though the PPC was not the only party to oppose the Liberals’ campaign promise to extend vaccine mandates, it was clearly the “get back to normal” party.
Though Canada has a long history of populist parties, an unapologetic anti-establishment party had been missing from the federal landscape since the days of the Reform Party and the Canadian Alliance. Nevertheless, a fair share of Canadians maintained populist attitudes, which correlated with vote choice, even in the absence of a manifestly populist party (Medeiros, Reference Medeiros2021). The emergence of the PPC in 2018 was a return of unremitting anti-elitism to the federal partisan scene. The PPC, however, deviates from the authoritarianism of populist radical-right parties by promoting libertarian policies in terms of individual freedoms and free market economics (Budd, Reference Budd2021). Therefore, the PPC can be characterized as a nationalist populist party that focuses on issues of identity (immigration) and anti-establishment / anti-elite rhetoric.
The PPC focused on this anti-establishment, libertarian and nativist messaging in 2019 in its first federal election. However, the party received only 1.6 per cent of the popular vote and failed to elect a single member of Parliament (MP). In 2021, the PPC tripled its vote share, winning 4.9 per cent of the vote, though it again failed to win a single seat despite having more than twice the vote share of the GPC (2.3 per cent), a party that elected two MPs.
As a relatively new party, the PPC has not been subject to much research. It is also not an inconsequential party, despite its level of electoral support: the PPC's presence on the ballot may have led to vote splitting on the political right, costing the CPC three to seven seats in the 2021 election (Grenier, Reference Grenier2021). While the PPC's increase in support likely did not alter the relative party standings in the House of Commons, it nevertheless did change the electoral landscape in Canada.
We therefore consider it important to understand the sources of PPC support among the Canadian electorate. Further, events such as the Freedom Convoy occupation of Ottawa demonstrate the salience of the kinds of anti-establishment notions proffered by the PPC to Canadian politics. These considerations lead us to ask, What factors were related to support for the PPC in the 2021 Canadian federal election? We focus on two main factors: populist attitudes and opposition to public health restrictions.
The study of populism has typically focused on populist radical-right parties and candidates, where populism is often defined as a combination of nativism, authoritarianism and populist ideology (Mudde, Reference Mudde2007). While these three attitudinal dimensions have been shown to be related yet distinct (Blanchet and Medeiros, Reference Blanchet and Medeiros2019; Medeiros, Reference Medeiros2021; Rooduijn, Reference Rooduijn2014a), nativism is not always a feature of populist politics (De Cleen et al., Reference De Cleen, Glynos and Mondon2018; Rooduijn, Reference Rooduijn2019). As Rooduijn (Reference Rooduijn2019) describes it, populism is a vertical antagonism between “the people” and “the elite,” whereas nativism is a horizontal antagonism between “the nation” and the “dangerous other.” Rather, populist movements universally target the establishment and the elite.
While scholarship has mostly focused on populism from the perspective of politicians and political parties (the supply side of populism), there has been a growing interest in the populist attitudes held by citizens (Hawkins and Rovira Kaltwasser, Reference Hawkins and Kaltwasser2017; Van Hauwaert and Van Kessel, Reference Van Hauwaert and Van Kessel2018). Populist attitudes have been associated with a discontent and distrust of the governance by elites and a desire for more direct citizen involvement in policy making (Erisen et al., Reference Erisen, Guidi, Martini, Toprakkiran, Isernia and Littvay2021; Kitschelt, Reference Kitschelt, Mény and Surel2002).Though some scholars have emphasized the need for a socio-economic crisis and/or a populist political party to activate the association between populist attitudes and vote choice (Hawkins et al., Reference Hawkins, Kaltwasser and Andreadis2020; Rhodes-Purdy et al., Reference Rhodes-Purdy, Navarre and Utych2021; Rooduijn et al., Reference Rooduijn, van der Brug and de Lange2016), others have called into question the need for specific conditions for populist attitudes to correlate with vote choice (Medeiros, Reference Medeiros2021; Santana-Pereira and Cancela, Reference Santana-Pereira and Cancela2020). Regardless of the salience of the demand side of populism, populist attitudes nevertheless correlate with vote choice. Furthermore, when populist parties and candidates are available to voters, research has found a link between populist attitudes and electoral support for populists (see, for example, Akkerman et al.., Reference Akkerman, Zaslove and Spruyt2017; Hawkins et al., Reference Hawkins, Kaltwasser and Andreadis2020). Specifically, the stronger the populist attitudes held by individuals, the more likely they are to vote for populist parties (Van Hauwaert and Van Kessel, Reference Van Hauwaert and Van Kessel2018).
Still, existing research also highlights the roles played by personal frustration and societal-level crises in explaining voters’ attraction toward populist parties. Dissatisfaction with one's social standing or personal circumstances tends to draw voters to populist parties (Guth and Nelsen, Reference Guth and Nelsen2021; Lubbers and Scheepers, Reference Lubbers and Scheepers2000; Mayer and Perrineau, Reference Mayer and Perrineau1992; Schumacher and Rooduijn, Reference Schumacher and Rooduijn2013). At the same time, the electoral success of populist movements has often been linked to economic and demographic changes in society (in particular, immigration) that give rise to popular grievances (Betz, Reference Betz, Betz and Immerfall1998; Mudde, Reference Mudde2010). What follows from this is that when societal crises—like a global pandemic—occur, popular frustrations are likely to arise, and populist parties should find fertile electoral ground.
The 2021 Canadian federal election was held during a major societal crisis. Moreover, the PPC is an unapologetic populist party and previous studies offer ample evidence that voters who hold populist attitudes are more likely to vote for such parties. Our first hypothesis is therefore:
H1: Populist attitudes increase the likelihood of voting for the PPC.
Populist parties also fought against COVID-19 health measures, as these policies are associated with “elite overprotectiveness” (Brubaker, Reference Brubaker2021). The origins of populist opposition to pandemic health measures are sufficiently clear. Populism research has consistently highlighted how populist politicians incite conflict between the (virtuous) “people” and (corrupt) “elites” (Mudde, Reference Mudde2007). The anti-elitist antagonism of populism, in which distrust of elites combines with a desire to have policy align with the perceived will of the majority of the people, applies beyond political leaders and has also been shown to target intellectuals and scientific experts (Hawkins and Rovira Kaltwasser, Reference Hawkins, Kaltwasser and A2018; Mede and Schäfer, Reference Mede and Schäfer2020; Merkley, Reference Merkley2020). Such anti-science and anti-expert thinking extends readily to public health experts (Lasco and Curato, Reference Lasco and Curato2019). Bellolio (Reference Bellolio2022: 1) identifies three ways in which populists attacked public health experts during the COVID-19 crisis: “(1) they raise a moral objection against scientists who have been allegedly corrupted by foreign interests, turning them into enemies of the people; (2) they present a democratic objection against the technocratic claim that scientific experts should rule regardless of the popular will; and (3) they employ an epistemic argument against scientific reasoning, which is said to be inferior to common-sense and folk wisdom, and antithetical to the immediateness of political action.” According to Bellolio (Reference Bellolio2022: 3), all three types of objections align with the “core feature of populism”: the conflict between the people and elites. Essentially, populists would be antagonistic toward many COVID-19 health measures because those measures would be promoted by distrusted political and scientific elites, and they would perceive those policies to not be aligned with the will of the majority of the people. However, while some populist politicians tried to benefit from the COVID-19 crisis, as they have often done in times of turmoil, they had a difficult time taking advantage of this crisis in the European context, where they found it challenging to advance alternative policies to those implemented by incumbent governments (Bobba and Hubé, Reference Bobba, Hubé, Bobba and Hubé2021). Further, Wondreys and Mudde (Reference Wondreys and Mudde2022) highlight that populist radical-right parties have not consistently opposed COVID-19-related public health measures. Rather, they have done so when they have been outside power: opposition parties are likely to be strong—indeed, reflexive—critics of COVID-19 health measures proposed by the incumbent government.
When we look at whether pandemic elections have aided the electoral fortunes of populist parties, the results so far are not conclusive. The 2021 pandemic election in Czechia saw anti-populist parties use the COVID-19 crisis to criticize the performance of the ruling populist party, Akce nespokojených občanů (ANO), and ultimately unseat it (Havlík and Kluknavská, Reference Havlík and Kluknavská2022). In Germany, the 2021 federal election saw the populist party Alternative für Deutschland (AfD), a stern opponent to COVID-19 measures, lose votes compared to the previous election (Bayerlein and Metten, Reference Bayerlein and Metten2022). Still, pandemic election results have not been all negative for populist parties. The populist Chega party saw an important breakthrough in the 2022 Portuguese legislative elections. Yet Chega and the other parties chose not to politicize the COVID-19 crisis, focusing rather on other topics (Lopes, Reference Lopes2023). Marine Le Pen's Rassemblement National (RN) tried taking ownership of the crisis by criticizing elites, ruling politicians and health experts and by proposing alternatives that matched its populist leanings during the 2022 French presidential and legislative elections (Froio, Reference Froio2022). Although Le Pen and RN saw their electoral support increase to historic heights, the exact impact of their opposition to COVID-19 measures is difficult to determine. Consequently, the relationship between the COVID-19 pandemic and support for populist parties remains unclear.
Nevertheless, the level of vaccine hesitancy in European countries has been shown to correlate with the level of electoral support for populist parties (Kennedy, Reference Kennedy2020). Some research has even found an association between scientific skepticism and populism (Mede and Schäfer, Reference Mede and Schäfer2020), as well as an association between vaccine hesitancy and voting for populist parties (Kennedy, Reference Kennedy2019). At the same time, other studies point to left–right ideology and partisanship as stronger correlates of attitudes toward COVID-19-related measures (Galanopoulos and Venizelos, Reference Galanopoulos and Venizelos2022; Gravelle et al., Reference Gravelle, Phillips, Reifler and Scotto2022; Shino and Smith, Reference Shino and Smith2021). Even so, recent research has indeed shown that vaccine hesitancy during the COVID-19 crisis does increase the likelihood of individuals to support populist parties (Serani, Reference Serani2023).
The 2021 Canadian federal election campaign was very much a “pandemic election” in which policy responses to COVID-19 played an important role in the different parties’ platforms and campaign rhetoric. While the Liberal campaign promised vaccine mandates for the federal public service and in federally regulated industries, the PPC focused heavily on the need for freedom and personal choice in fighting against lockdowns as well as against mask and vaccine mandates imposed by public health authorities. Based on previous research into vaccine hesitancy and support for populist parties, we put forth our second hypothesis:
H2: Opposition to COVID-19-related public health restrictions increases the likelihood of voting for the PPC.
At the same time, the anti-elitism and anti-intellectualism inherent in populist thinking points to a link between populism and opposition to public health restrictions. Indeed, resistance to COVID-related mask and vaccine mandates and lockdowns—as well as criticism of public health experts—was strongest among populist-leaning politicians, at both the federal and provincial levels. Also, research has shown that distrust toward elites and experts is strong in both the vaccine hesitant and populists (Carvalho Bivar et al., Reference Bivar, de Aguiar, Cavalcanti Santos and Gualberto Cardoso2021), helping to explain why individuals with stronger populist attitudes tend to be more hesitant about COVID-19 vaccines (Edwards et al., Reference Edwards, Biddle, Gray and Sollis2021). These findings align with previous research that has found that vaccine hesitancy is anchored in a broader anti-elite and anti-establishment worldview (Stoeckel et al., Reference Stoeckel, Carter, Lyons and Reifler2022). Interestingly, populists tend to resist politicized vaccines (for example, MMR and COVID-19) and not other types (for example, seasonal influenza, human papillomavirus, and so on), in part because of a lack of confidence in these vaccines that politicized debates engender (Kohler and Koinig, Reference Kohler and Koinig2023).
Given the link between populist attitudes and vaccine hesitancy, as well as the increased likelihood of the vaccine hesitant to vote for populists (Serani, Reference Serani2023), opposition to public health restrictions may act as the more proximate factor related to support for the PPC but could be itself shaped by populist sentiment. Put another way, the effect of populism on PPC support may be mediated by opposition to public health restrictions. This informs our third hypothesis:
H3: The effect of populism on PPC support is mediated by opposition to public health restrictions.
Data and Methods
We test these theoretical expectations by drawing on an original post-election survey collected by Momentive (the makers of the SurveyMonkey online survey data collection platform). Data were collected from September 21, 2021 to December 12, 2021 using SurveyMonkey's End Page methodology. This involves recruiting survey respondents online from the millions of survey-takers who complete one of the thousands of user-created surveys on SurveyMonkey's platform every day. After completing a survey on an unrelated topic, selected respondents from Canada (ascertained from their internet protocol [IP] addresses) were presented with a survey completion web page (End Page) inviting them to complete another voluntary, uncompensated research survey. Respondents clicking on a link titled “Take the Survey” were then directed to the post-election survey. This sampling method yields demographically representative samples that have been used in previous studies of consumer behaviour, political behaviour and public opinion in Canada and other Western democracies (Gravelle, Reference Gravelle2021; Gravelle et al., Reference Gravelle, Phillips, Reifler and Scotto2022; Williams et al., Reference Williams, Gravelle and Klar2022). The extended time period of survey data collection also allowed for an unusually large sample of 27,720, including 20,835 self-reported voters. Of these, 18,950 respondents reside in the 312 federal electoral districts (FEDs, or ridings) contested by the PPC; this subsample includes 1,469 self-reported PPC voters.Footnote 1 Riding-level sample sizes range from a minimum of 5 to a maximum of 148, with a median of 60.
Though our survey data constitute an online nonprobability sample, we contend that they are well suited to our research aims. Our primary focus is on the attitudinal constructs associated with PPC support and not sample point estimates for party support. Further, previous research on Canadian public opinion finds few consequential differences between online nonprobability and probability-based samples (Breton et al., Reference Breton, Cutler, Lachance and Mierke-Zatwarnicki2017; Stephenson and Crête, Reference Stephenson and Crête2011). Our data are thus fit for purpose (Baker et al., Reference Baker, Brick, Bates, Battaglia, Couper, Dever and Roger Tourangeau2013).
The post-election survey instrument asked all respondents how closely they followed news about the federal election. The survey then asked respondents whether they had voted in the federal election; those reporting they had cast a ballot were then asked about the manner in which they had voted (for example: at a polling station on election day, at an advance poll, by mail-in ballot) and which party they had voted for. Following the questions on their self-reported vote, respondents were presented with questions capturing their populist attitudes. These included four items representing a subset of the widely used populist attitudes scale presented by Akkerman and colleagues (Reference Akkerman, Mudde and Zaslove2014). The items gauged respondents’ agreement or disagreement with the following statements: “Members of Parliament need to follow the will of the people,” “The people, and not politicians, should make our most important policy decisions,” “The political differences between the elite and the people are larger than the differences among the people,” and “What people call ‘compromise’ in politics is really just selling out on one’s principles.” As a composite scale, these items obtain a minimally acceptable level of internal reliability (Cronbach's α = 0.66), which can be interpreted as indicating an acceptable level of internal reliability.
Measures of opposition to (or, conversely, support for) COVID-related public health restrictions make up five items taken from Gravelle and colleagues (Reference Gravelle, Phillips, Reifler and Scotto2022), including opposition to mandatory vaccinations required to access public places and public events and to employment-related vaccination requirements. The scale formed from these items has excellent internal reliability (α = 0.96). The analyses also adjust for important attitudinal and demographic controls. The respondents’ nativist attitudes were also measured. Nativism was measured using five items adapted from existing studies of nativist, anti-immigrant attitudes in English-speaking societies (Gravelle, Reference Gravelle2018, Reference Gravelle2019) that also resemble items tested in Europe (Rooduijn, Reference Rooduijn2014b). This scale obtains good internal reliability (α = 0.89). Results from a confirmatory factor analysis (CFA) model also confirms that opposition to COVID-19 health measures, nativism and populism are captured well by the items used and that they measure three distinct underlying concepts (see Table A3 in the online appendix). To capture respondents’ political orientations, the surveys contained measures of self-rated ideological placement and party identification. Respondents were asked to place themselves on a 0–10 scale ranging from left to right, reflecting conventional language in discussing political orientations in Canadian politics (Cochrane, Reference Cochrane2015). The surveys also asked respondents about their demographic characteristics, including sex, age, educational attainment, and province of residence. The full survey instrument is included in the online appendix.
We further augment our survey data with key contextual variables. Because Canadian federal elections are properly 338 distinct contests for single-member seats, we use respondents’ self-reported postal codes to append the FED using the Postal Code Conversion File (PCCF) and FED map shapefiles from Statistics Canada. Doing so also allows us to identify voters in the 312 ridings out of 338 contested by the PPC. Furthermore, it permits us to determine which ridings are safe seats for a given party, as a voter in a noncompetitive riding (someone whose ballot has no prospect of determining the outcome of the election) may be more inclined to cast a protest vote for a minor party (such as the PPC) than a voter would in a competitive riding (Schimpf, Reference Schimpf2019). We operationalize a noncompetitive, or safe, seat as a riding where a single party won by a margin of 10 percentage points or more in the 2021, 2019 and 2015 federal elections.
The local context of the COVID-19 pandemic might also plausibly shape vote choice. In the 2021 German federal election, regions that were hit especially hard by the COVID-19 crisis were less likely to support the AfD, which strongly opposed COVID-19 health measures (Bayerlein and Metten, Reference Bayerlein and Metten2022). Accordingly, we locate respondents in their provincial health region using map shapefiles from Statistics Canada and provincial governments. These health region codes are then linked to the logged cumulative count of COVID-19 cases in a respondent's health region as of election day, September 20, 2021, obtained from the Health Infobase of the Government of Canada (https://health-infobase.canada.ca).
We report our results below using binary logit models where the modelled outcome is voting for the PPC candidate in one's riding (versus voting for any other candidate). Canada is a multiparty system and one in which ballot composition differs between Quebec and the rest of Canada due to the BQ. For this reason, we also report multinomial logit models in Table A4 (rest of Canada) and Table A5 (Quebec) in the online appendix, though we note that these more complex models do not yield substantively different results from the simpler binary logit models. It is also reasonable to model voters as nested within ridings, as well as a public health context (proxied by provincial health regions) and a socio-demographic context. These considerations suggest that a multilevel modelling approach is appropriate (Hox, Reference Hox2010; Raudenbush and Bryk, Reference Raudenbush and Bryk2002). We therefore also report generalized linear mixed models (GLMMs) in Table A6 in the online appendix. These show minimal variance for random intercepts specified by riding, health region and census subdivision, as well as fixed effects that again yield the same substantive results as the simpler binary logit models.
We endeavour to facilitate the interpretation of the results by scaling all continuous independent variables 0–1 and then centring them at their means. To test the mediation of populism on voting for the PPC via opposition to public health restrictions as advanced in H3, we also test our expectations using a SEM, which allows us to model our constructs of interest as latent variables and also to decompose the effects of particular constructs on PPC support into direct and indirect (that is, mediated) paths (Kline, Reference Kline2016). These results are reported in Table 2.
Results
The binary logit models yield several noteworthy findings (see Table 1). Considering first some of the demographic covariates, both Model 1.1 and Model 1.2 indicate that older voters are less likely to report voting for the PPC, though there is no consistent difference between women and men. The results also do not point to a robust relationship between educational attainment and PPC support. Considering regional differences, the results for Model 1.2 notably indicate that residents of Saskatchewan and Alberta are less (not more) likely to report voting for the PPC. These results are more intuitive than they might seem at first, since they reflect Saskatchewan and Alberta residents’ propensity to support the CPC over casting a protest vote. Despite Maxime Bernier's electoral history in Quebec, voters in the province are not significantly more likely to support the PPC. To wit, francophones are less likely than anglophones to report voting for the PPC. The logit models also indicate little in the way of a safe-seat effect on casting a putative protest vote for the PPC. There is also little evidence of differences in support between regions with comparatively lower or higher COVID-19 case counts or in areas with higher immigrant concentration or greater increases in immigrant concentration.
Table 1. Explaining Support for the People's Party of Canada (binary logit)
Note: Models are fit by binary logistic regression with standard errors adjusted for the complex sample design (stratification by province and clustering by federal electoral district, with post-stratification weights). All continuous independent variables are mean-centred.
* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001
Turning to the main attitudinal factors of interest, Model 1.1 confirms our expectations that populist attitudes are associated with a heightened propensity for supporting the PPC over other parties, supporting H1. The coefficient is positive and substantively important. Unsurprisingly, nativism and right-wing ideology are also associated with a greater propensity for voting for the PPC. The inclusion of opposition to public health restrictions in Model 1.2, however, introduces a variable with an even stronger association with PPC support, thus also supporting H2. Controlling for opposition to public health restrictions also reduces the coefficients for populism, nativism and right-wing ideology with voting for the PPC, though they nevertheless remain statistically significant.
To gauge the absolute magnitude of these relationships, we estimated the predicted probabilities for PPC support for a given voter profile. Doing so allows us to show that the substantive effects of populism, nativism and left–right ideology are in fact quite modest in comparison to the effect of opposition to public health restrictions. For example, a woman voter from Ontario aged 49 with a high school education identifying with the far left has a probability of 0.005 of supporting the PPC; a voter on the far right has a probability of 0.02. A voter with the same demographic profile with the lowest score on populism has a probability of 0.003 of supporting the PPC; a voter with the highest score has a probability of 0.02. A voter with the lowest score on nativism has a probability of 0.006 of supporting the PPC, while a voter with the highest score has a probability of 0.02. By contrast, a voter with the demographic profile above expressing strong support for COVID-related public health restrictions has a probability of 0.003 of supporting the PPC, while a voter expressing strong opposition has a probability of 0.19. These results suggest that the PPC was in some measure successful in capturing the anti-lockdown, anti-vaccine constituency among the Canadian electorate and that attitudinal factors typically associated with support for radical-right parties were less relevant. Similarly, contextual factors such as the electoral competitiveness of one's riding, the public health context (specifically related to the prevalence of COVID-19) and the demographic context (both immigrant concentration and change in immigrant concentration over time) do not appear relevant.Footnote 2
Additionally, results from the multinomial logit specification (presented in Table A4 in the online appendix) indicate that both populism and opposition to public health restrictions separate voters between the mainstream right (namely the CPC) and the PPC. Models A3.1 and A3.2 (in Table A4 in the online appendix) present results from all provinces except Quebec and set LPC support as the reference category. These models show—unsurprisingly—that self-placement on the political right is associated with a higher likelihood of PPC and Conservative support. Heightened nativist sentiment is also associated with greater PPC and Conservative support. Interestingly, populism is positively associated with support for all opposition parties—that is, not just the PPC and Conservatives but also the NDP and Green Party. This result might suggest a link between populism and anti-incumbent voting, though existing research has highlighted the negative association between populist attitudes and voting for the LPC even when the latter is not the incumbent (Medeiros, Reference Medeiros2021). At the same time, heightened populist attitudes are associated with a higher likelihood of supporting the PPC over any other party, including the Conservatives. Models A3.3 and A3.4 present the same multinomial logit models for all provinces except Quebec, but with the PPC as the reference category. Here, we see negative and significant coefficients for populism for Liberal, Conservative, NDP and Green support. Similarly, the negative and significant coefficients for the opposition to public health restrictions scale indicate that greater opposition to public health restrictions is associated with a higher likelihood of supporting the PPC over any of the main political parties, including the Conservatives.
Multinomial logit results for Quebec (presented in Models A4.1–A4.4 in Table A5 of the online appendix) are broadly similar. Models A4.1 and A4.2 (with LPC support as the reference category) again show that voters on the political right have a higher likelihood supporting either the PPC or the Conservatives. Nativism is also associated with greater PPC, Conservative and BQ support. Quebec voters expressing strong opposition to public health restrictions are similarly more likely to support either the PPC or Conservatives over the Liberals. Re-specifying the models with the PPC as the reference category once more (in Models A4.3 and A4.4), we again find the negative and significant coefficients for the opposition to public health restrictions scale on LPC, CPC, NDP and BQ support. This again highlights the outsized role played by hostility to public health measures in shaping support for the PPC and how such hostility differentiates PPC and Conservative supporters.
In sum, we find that many of the factors related to PPC support are also connected with Conservative support—in particular, identification with the political right, nativism, populism and opposition to public health restrictions. The magnitude of the relations of both populism and opposition with public health restrictions on PPC support, however, is appreciably stronger than the corresponding ones on Conservative support.
Given the different levels of construct validity implied by the values of Cronbach's alpha for our measures of populism, nativism and opposition to public health measures (above), direct comparisons of their effects on PPC support in the binary logit and multinomial logit models should be approached with caution. It is well known that measurement error in the independent variables implies potential bias (either upward or downward) in parameter estimates in linear and generalized linear models. In other words, the propagation of measurement error from scales used as independent variables may bias their coefficients (Kline, Reference Kline2016). Given our data, the main concern is with the validity of our populism scale, and thus with the coefficients for populism. A long-standing solution to such measurement issues is to turn to SEM. In a SEM framework, the concepts of interest are typically represented as latent variables measured using multiple (observed) items, and measurement error is modelled explicitly. Accounting for measurement reliability and validity in the measurement model of a SEM in turn provides estimates for the effects of the independent variables on the dependent variable(s)—that is, the structural model—that are assumed to be free of measurement error and thus any bias in the structural coefficients (Hayduk, Reference Hayduk1987; Kline, Reference Kline2016).
The results from the binary logit and multinomial logit models also suggest that the relationship of populism with voting for the PPC is (at least partly) mediated by opposition to public health restrictions: the magnitude of populism's effect on PPC support is substantially reduced when opposition to public health restrictions is included in the model. SEM is again well suited to testing this mediation by testing specific indirect paths of theoretical interest—in our case, the indirect relationship between populism and PPC support via opposition to public health restrictions. We thus specify a structural equation model where PPC voting is regressed on populism, nativism, left–right ideology, and opposition to public health restrictions represented as latent variables (see Figure 1). Opposition to public health restrictions is regressed on populism and left–right ideology in the same model, implying both a direct effect of populism on voting for the PPC, as well as an indirect effect via opposition to public health restrictions.
Figure 1. Structural Equation Model (SEM): Explaining PPC Voting.
Before examining the structural component of our SEM (that is, the relations between the different factors), we first assess overall model fit and the measurement component of the model. The results reported in Table 2 indicate good overall model fit by current standards used in the SEM literature (Kline, Reference Kline2016). The values for the root mean square error of approximation (RMSEA) of 0.048, the comparative fit index (CFI) of 0.996 and the standardized root mean square residual (SRMR) of 0.045 indicate a SEM with good fit. Turning to the model's measurement component, the standardized factor loadings for the items measuring nativism and opposition to public health restrictions are consistently high at 0.76 or above. With only a single indicator of ideology, the standardized factor loading of 0.94 simply reflects that its error variance (and factor loading) is specified by the model instead of estimated from the data. The factor loadings for the populism factor are more variable, with standardized loadings between 0.44 and 0.76. This suggests some measurement shortcomings with the Akkerman and colleagues (Reference Akkerman, Mudde and Zaslove2014) items that have been noted in other research (Hameleers and de Vreese, Reference Hameleers and de Vreese2020; Schulz et al., Reference Schulz, Müller, Schemer, Wirz, Wettstein and Wirth2018).
Table 2. Explaining Support for the People's Party of Canada (SEM)
Note: Model fit by robust weighted least squares (WLSMV) estimation with standard errors adjusted for the complex sample design (stratification by province and clustering by federal electoral district, with post-stratification weights); n = 18,950; ‡ constrained to 1 for model identification.
*** p ≤ 0.001
Model fit: χ 2SB = 4,301.76, df = 95, p < 0.001; RMSEA = 0.048 (0.047–0.050), p(close fit) = 0.986; CFI = 0.996; SRMR = 0.045
Turning to the structural component of our SEM, the results confirm our expectations. Opposition to public health restrictions exerts a strong direct effect on voting for the PPC, while the direct effect of populism is modest (again supporting H1 and H2). At the same time, populism exerts a substantively large effect on opposition to public health restrictions. These results imply that the total effect of populism on voting for the PPC is therefore greater (and substantively larger) than what is implied by its direct effect. Expressed as predicted probabilities, the direct effect of populism obtained from the SEM differs little from Model 1.2 using binary logit. A voter scoring the minimum on the populism factor has a predicted probability of supporting the PPC of less than 0.001; a voter scoring the maximum on the populism factor has a predicted probability of supporting the PPC of 0.02. This, however, understates populism's importance, since it also exerts an effect on opposition to public health restrictions. Examining the total effect of populism—its direct effect plus its indirect effect via opposition to public health restrictions—points to greater substantive importance, and to the mediation of populism by opposition to public health restrictions, as expected by H3. Accounting for populism's total effect indicates that a voter scoring the minimum on the populism factor has a predicted probability of supporting the PPC of less than 0.001; a voter scoring the maximum on the populism factor has a predicted probability of supporting the PPC of 0.09 (see Figure 2). Still, opposition to COVID-related public health restrictions remains the strongest and most proximate factor shaping support for the PPC. Based on the SEM results, a voter expressing weak support for COVID-related public health restrictions has a probability of less than 0.001 of supporting the PPC, while a voter expressing strong opposition has a probability of 0.29 (see Figure 3).
Figure 2. Direct and Total Effects of Populism on PPC Voting (SEM).
Figure 3. Direct and Total Effects of Populism on Opposition to Public Health Restrictions.
Discussion and Conclusion
The 2021 Canadian federal election was made exceptional by the COVID-19 crisis. The main political parties debated policies to deal with the ongoing pandemic in a substantive though still arguably partisan manner. Maxime Bernier and the PPC diverged from the other parties with their unabashed opposition to COVD-19 measures. Resistance to lockdowns and to mandates for masks and vaccines took centre stage in the party's campaign. This strategy attracted enthusiastic crowds at PPC rallies and led to a surge in voting intentions. Given these developments, and given the scarcity of research on the PPC, our aim in this article was to better understand the factors shaping support for the PPC among the Canadian electorate.
The association between populist parties and opposition to COVID-19-induced public health measures has been widely observed, though scholarship on this topic is limited at present. The current study also sought to more broadly contribute to this lack of understanding by exploring what factors shape voting for the PPC.
Our analyses highlight three main findings. First, voters who held populist attitudes were more likely to vote for the PPC. This is rather unsurprising, as the PPC is an unapologetic anti-establishment and anti-elite party. This result also aligns with the scholarship from elsewhere that has demonstrated that populist attitudes lead to voting for populists (Akkerman et al., Reference Akkerman, Zaslove and Spruyt2017; Hawkins et al., Reference Hawkins, Kaltwasser and Andreadis2020; Van Hauwaert and Van Kessel, Reference Van Hauwaert and Van Kessel2018). Second, opposition to COVID-19 health measures was associated with a greater likelihood of voting for the PPC. The analyses demonstrate that opposition to COVID-19 health measures was the factor most strongly associated with voting for the PPC. Lastly, the SEMs allowed us to ascertain that a substantial portion of the total effect of populist attitudes on voting for the PPC was mediated by individuals’ opposition to public health restrictions. This finding also highlights the relationship between populism and opposition to COVD-19-related public health restrictions: the more populist an individual, the more likely that person was to oppose COVID-19 measures. This result provides important empirical evidence on the link between populism and opposition to public health measures among the Canadian electorate.
Still, it is important to note that recent research has highlighted theoretical and empirical distinctions between political populism and science populism (Eberl et al., Reference Eberl, Huber, Mede and Greussing2023). While similarities exist between both types of populism, they obviously differ in terms of whether the referent object is political elites or political power (on the one hand) or scientists or scientific knowledge (on the other). Our study examines the relationship between political populism and support for a political party. Our conceptualization and measurement of populism reflects this focus (Akkerman et al., Reference Akkerman, Mudde and Zaslove2014). Still, we readily concede that measures of populist attitudes beyond those available in our data might prove a better fit for the Canadian case (for example, Schulz et al., Reference Schulz, Müller, Schemer, Wirz, Wettstein and Wirth2018; Van Hauwaert et al., Reference Van Hauwaert, Schimpf and Azevedo2020). We also agree with Eberl and colleagues (Reference Eberl, Huber, Mede and Greussing2023) that populism is a multifaceted phenomenon whose remit extends beyond electoral politics, and so researchers should ensure congruence between their conceptualization of populism and the referent object to the greatest extent possible.
The pandemic allowed populist parties to mobilize supporters by focusing on resisting COVID-19 restrictions. As our results demonstrate, the association between populist attitudes and opposition to pandemic measures is robust. Still, populist parties have not yet benefited electorally from the pandemic (Wondreys and Mudde, Reference Wondreys and Mudde2022). The PPC's gains in the 2021 federal election are best described as modest. A sizable portion of Canadians had tired of the pandemic and no longer supported COVID-19 restrictions as they once did (Bricker, Reference Bricker2021; Quirion, Reference Quirion2022). Nevertheless, the PPC had only limited success in harnessing that discontent. Populist politicians should therefore reconsider the logic of resisting reasonable public health measures.
If we look specifically at the PPC, it is clear that the limited increase of the party's electoral fortunes came during an exceptional election. Unless the COVID-19 restrictions remain with us for years to come, the PPC might face more challenging electoral contests in the future, as the results indicate that opposition to pandemic measures played such an important role in voting for the party. Therefore, as Canadians return to a life without lockdowns and mask or vaccine mandates, the PPC might be returning to the fringes of electoral politics.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S000842392300015X.
Data availability statement
Replication materials related to this article have been placed in an Open Science Framework (OSF) repository, accessible at https://doi.org/10.17605/OSF.IO/T2CAJ.
Acknowledgments
This article was originally presented as a paper at the 2022 annual conference of the Canadian Political Science Association held virtually, and at the American Political Science Association annual meeting held in Montreal, Quebec. Thanks are due to Jean-Philippe Gauvin, Michael Martinez, Mathew Marques, Thomas Scotto and the anonymous reviewers at this journal for their helpful comments. Thanks are also due to Austin Pettis and Jon Cohen at Momentive for helping make the SurveyMonkey News Survey post-election survey data collection happen and to Michael Steelworthy at the Wilfrid Laurier University library for providing the spatial data needed to execute the analyses reported here.
Disclosure statement
The authors wish to disclose no conflict of interests in the research reported in the article.
Open access
