Local Campaign Communications and Personalization

Although election campaigns are increasingly centralized, local actors generally maintain autonomy to make important strategic choices about their public-facing activities (for example, Cross et al., Reference Cross, Malloy, Small and Stephenson2015; Cross et al., Reference Cross, Pruysers and Currie-Wood2022; Kam, Reference Kam2009; Stephenson et al., Reference Stephenson, Lawlor, Cross and Blais2019). The stratarchical form of party organization (Carty, Reference Carty2002), with a division of labour between national and local actors, allows candidates the autonomy to shape the identities they present to voters. In a dynamic campaign context, it can be difficult for central campaigns to supervise local activities and the centre does not necessarily aim to micromanage them (Robbins-Kanter, Reference Robbins-Kanter2022).

That local candidates may present curated identities to voters is consistent with the concept of political personalization. This signifies that individual politicians and their personal characteristics are increasingly emphasized at the expense of their parties or policies (Cross et al., Reference Cross, Katz, Pruysers, Cross, Katz and Pruysers2018). Balmas et al. (Reference Balmas, Rahat, Sheafer and Shenhav2014) distinguish between centralized and decentralized personalization, depending on whether the increased focus on politicians’ personal attributes occurs with national or local figures. Evidence of decentralized personalism can be found in the personal attributes that candidates highlight or omit in their campaign communications. Although we lack provincial data, Cross and Young (Reference Cross and Young2015) and Cross et al. (Reference Cross, Currie-Wood and Pruysers2020) find relatively high levels of decentralized campaign personalism among federal candidates.

Existing scholarship shows that, in curating their identities, candidates may adapt to real or perceived biases within the electorate. This is especially true for candidates from traditionally underrepresented groups, including women (Everitt and Raney, Reference Everitt, Raney and Tremblay2019; Wagner and Everitt, Reference Wagner, Everitt, Wagner and Everitt2019). Focusing on racial background, Wagner et al. (Reference Wagner, Karen Bird and Lalancette2023) show that most Black candidates in the 2021 federal election “de-racialized” their campaign communications or chose to subtly and strategically direct messages to Black voters. In the case of social class, we expect candidates to emphasize or deemphasize class characteristics based on personal preference and whether they see these attributes as electoral assets or liabilities.

Political Candidates and Class Identity

Descriptive representation considers the extent to which elected officials reflect politically relevant characteristics of the general population. It has been argued that descriptive representation facilitates substantive representation, as legislators from diverse backgrounds are more likely to understand and advocate for the concerns of a diverse population (Sevi et al., Reference Sevi, Blais and Mayer2020: 219). While one should not overstate the influence of individual MPs in Canadian legislatures due to high levels of party discipline (Marland, Reference Marland2020), the demographic composition of legislatures shapes the pool from which leaders choose cabinet members and pressures that caucuses exert on party leadership.

Class status is a politically salient aspect of candidate identity that has received minimal scholarly attention. Unlike other aspects of identity, such as gender or ethnoracial identity, class background can rarely be assumed simply from looking at a candidate's picture.Footnote ² This provides greater possibility for candidates to use their campaign biographies to either emphasize or downplay particular class signifiers. While the influence of class over partisan politics has been weaker in Canada than in other countries (Johnston, Reference Johnston2017), class matters to Canadian politics, whether viewed in terms of voting behaviour (Gidengil, Reference Gidengil and Baer2002; Polacko et al., Reference Polako, Kiss and Graefe2022) or in terms of its broader (and intersectional) effects on political outcomes and representation (Livingstone and Weinfeld, Reference Livingstone and Weinfeld2015; Pilon, Reference Pilon2015; Reference Pilon and Laycock2019).

Social class is a contested concept that is composed of and shaped by a number of factors. Our analysis treats class status as a descriptive category with a range of indicators that include occupation, education and union membership.Footnote ³ This approach reflects the dominant state of Canadian political science scholarship about class, which frequently adopts a positivist stance (Borwein, Reference Borwein2022; Gidengil et al., Reference Gidengil, Nevitte, Blais, Everitt and Fournier2012; Johnston, Reference Johnston2017: 46; Pammet, Reference Pammett and Wearing1991; Polacko et al., Reference Polako, Kiss and Graefe2022; Robbins-Kanter and Sevi, Reference Robbins-Kanter, Sevi, Robbins-Kanter, Koop and Troup2025; Sevi et al., Reference Sevi, Blais and Mayer2020). Occupation is a strong indicator of material conditions, social status and life chances (Connelly et al., Reference Connelly, Gayle and Lambert2016; Parkin, Reference Parkin1971). In terms of education, we view college certifications as indicating a connection to working-class identity and university degrees a connection to white-collar or professional status.

For others, reducing class identity to relatively static categories, such as income levels, risks losing the fundamental sense of the phenomenon (Brodie and Jenson, Reference Brodie and Jenson1988; Mahon, Reference Mahon and Panitch1977; McCormack and Workman, Reference McCormack and Workman2015; Panitch, Reference Panitch1977; Pilon, Reference Pilon2015). Such indicators problematically ignore the systematic, subjective, or relational aspects of class. These critical perspectives instead emphasize that class distinctions exist within a social system that must be viewed in relational terms, for example, workers are defined as those forced to sell their labour to capitalist classes. That said, empirical research typically requires observable indicators to specify variables of interest. Occupation, education, and union membership may provide an incomplete picture of one's class position but these indicators nonetheless offer insight into politically salient elements of social stratification. As such, we can ask, how are the class identities that candidates emphasize affected by party affiliation, riding demographics, competitiveness and provincial socio-political context? Moreover, are there discrepancies between the identities that candidates emphasize and those that can be established by other publicly verifiable sources?

We focus our analysis on Ontario and Quebec elections for several reasons. First, while some prior analysis of class discourse identity in Canada has examined federal political candidates (Robbins-Kanter and Sevi, Reference Robbins-Kanter, Sevi, Robbins-Kanter, Koop and Troup2025; Sevi et al., Reference Sevi, Blais and Mayer2020), there is no research that includes provincial data. Our analysis allows us to compare candidate self-presentation in two distinct subnational party systems, and their large population share provides an appropriate sample size of nearly 1,000 candidates. Second, our analysis includes examination of how Quebec's cultural distinctiveness may affect how political candidates address issues of class. In particular, we explain how issues of language and cultural identity may “crowd out” class identity. Third, our data analysis required access to candidates’ online biographies, which are typically removed after an election campaign has ended. During data collection in 2022, Ontario and Quebec were the only two provinces that held elections. Moreover, since these elections took place in the same year, this limits the need to control for intervening factors that may influence candidates’ incentives to mention different class identities. Lastly, the importance of social class should be stronger in provincial than federal elections, because the regional differences which characterize the latter are diminished (Lambert et al., Reference Lambert, Curtis, Brown and Kay1987; Ornstein, Reference Ornstein2004).

In provincial politics, Ontario has traditionally shown modest class divides in vote choice. The PCs have tended to perform better with higher income voters while the NDP historically overperforms with modest-income and unionized voters (Malloy, Reference Malloy, Collier and Malloy2018). The class basis of Liberal support is more complex, featuring an apparent middle-class, educated base, as seen in their advantage with university-educated voters (Graefe, Reference Graefe2018). The varying importance of class voting is tied to parties’ decisions to make class appeals (Ornstein, Reference Ornstein2004). For example, class polarization was clearly apparent during the 1999 election, when all parties made class appeals centered on social programs and public spending, and the neoliberal Harris Conservatives received much weaker vote shares from union members, lower-income voters and those in blue-collar occupations (Ornstein, 95).

Conversely, provincial politics in Quebec is multidimensional, with the federalist/sovereigntist dimension at times overshadowing the left/right dimension (for example, Nadeau and Bélanger, Reference Nadeau, Bélanger, Kanji, Bilodeau and Scotto2012). While defending Canadian federalism, the Parti Libéral du Québec (PLQ) historically sits on the economic centre-right of the spectrum (Bélanger et al., Reference Bélanger, Daoust, Mahéo and Nadeau2022: 91). In the contemporary party system, the Parti Québecois (PQ) has been its main adversary, standing in favour of independence and ideologically left of centre (Bélanger et al., Reference Bélanger, Daoust, Mahéo and Nadeau2022). Right-of-centre parties (the Action Démocratique du Québec from 1994–2012 and after 2012, the Coalition Avenir Québec) have also disturbed this two-party system, articulating an “autonomist” position that is neither federalist nor sovereigntist. Data shows that higher annual income is at times associated with greater propensity to support the PLQ or CAQ (Nadeau and Bélanger, Reference Nadeau and Bélanger2013: 231). This is more likely due to economic policies beyond the sovereignty question. Indeed, from 1968–2015 there has been no consistent relationship between class status and support for Quebec independence (Guay, Reference Guay2017: 17).

In recent elections, the prospects for Quebec independence have weakened, and the 2014, 2018 and 2022 elections saw new parties elevate the importance of left/right and class politics in the province: the Parti Conservateur du Québec on the right and Québec Solidaire (QS) on the left (Daoust and Jabbour, Reference Daoust and Jabbour2020). That said, the continued importance of cultural and linguistic identity may lessen the importance of class appeals relative to other provinces (Lambert et al., Reference Lambert, Curtis, Brown and Kay1987). We examine three main factors that shape candidates’ class identities and how they are presented to voters: party ideology, riding characteristics and competitiveness. First, we consider how party ideology shapes both the recruitment of individuals from different class backgrounds and the extent to which they choose to highlight diverse class identities. Second, we consider two aspects of district context: the distinct politicization of class in Ontario versus Quebec ridings and the degree to which candidates present identities to match the class composition of their district. Finally, we consider riding competitiveness and how stereotypes around what makes a candidate “electable” shape candidate recruitment and self-presentation.

Hypotheses

We break down our expectations regarding candidate class identity into three stages. For party ideology, we expect that left party candidates will be the most likely to reference working-class identities and white-collar public-sector occupations. The disproportionate strength of left parties with both sets of voters should make it more likely that they recruit candidates from such backgrounds and for candidates to emphasize these identities to mobilize potential supporters. Since white-collar public-sector workers benefit from an expanded state, we would expect them to be more supportive of left parties that advocate for larger government. While our analysis will further compare between left parties, we do not anticipate major differences between them in terms of expressions of candidate class identity.Footnote ⁴

H1a: Left party candidates will be more likely to reference blue-collar occupations, white-collar public-sector occupations and union membership.

We expect that right party candidates are more likely to reference white-collar private-sector backgrounds. This is because such parties generally advocate for lower taxes and smaller government programs, which often aligns with the views or interests of voters from these backgrounds. It bears noting, however, that centre-right parties have been making efforts to court working-class voters (Speer, Reference Speer2021), and the Ontario PCs received notable private-sector union endorsements during the 2022 election (Bieman, Reference Bieman2022).

H1b: Right party candidates will reference white-collar private-sector identities more than others and working-class identities more than liberal candidates.

For their part, liberal parties have increasingly made efforts to win over middle-class progressive voters (Arceneaux and Wielen, Reference Arceneaux and Vander Wielen2023; Westlake, Reference Westlake, Robbins-Kanter, Koop and Troup2025). Higher levels of education tend to correlate with progressive or liberal views as well as higher incomes. Progressive views should make highly educated candidates and voters less sympathetic to right parties, while higher incomes should make them less supportive of left parties. This should leave liberal parties with a pool of more educated candidates who are more likely to reference higher education.

H1c: Liberal party candidates will make more references to university degrees than candidates for the other parties.

At our second stage, we expect political context to shape the recruitment of candidates in three ways. First, we expect that parties will be more likely to recruit candidates that match the demographics of their districts. Related to this, since candidates often run in districts where they reside, the candidate recruitment pool will be affected by the proportion of individuals in each riding from different class backgrounds. Therefore, candidate emergence also entails self-selection that is independent from party recruitment efforts. Due to self-selection, it should be more likely for a working-class candidate to emerge in a riding with more working-class residents. Third, there is a substantial literature that suggests social connections and group membership shape vote choice (for example, Berelson et al., Reference Berelson, Lazarsfeld and McPhee1954; Gidengil et al., Reference Gidengil, Nevitte, Blais, Everitt and Fournier2012). This gives parties and candidates an incentive to highlight common identities when campaigning.

H2a: Candidates will be more likely to reference class identities that match the demographics of the ridings in which they are running.

Comparing provinces, the politics surrounding language and culture in Quebec means that arts and culture are politicized to a greater degree (Desjardins, Reference Desjardins2008; Ellis and Woolstencroft, Reference Ellis, Woolstencroft, Pammett and Dornan2009). We expect this to result in more candidates from arts-and-culture backgrounds becoming involved in politics and for those candidates to place a greater emphasis on these occupational backgrounds when campaigning.

H2b: Candidates in Quebec will be more likely to reference arts-and-culture class identities and more likely to declare that they are students.

There is also greater student mobilization and student union involvement in Quebec politics. This activism has a distinctive left-wing character (Lacoursiere, Reference Lacoursière2007), and thus should largely affect left parties, leading them to recruit more students and for those candidates to place greater emphasis on such identities.

Finally, we expect that riding competitiveness shapes candidate recruitment and the identities that are emphasized. Parties and voters have stereotypes about who an “electable” candidate is, and we expect these stereotypes to play a greater role in shaping who parties nominate in more competitive ridings. Candidates with experience in business, law or in the civil service typically conform with expectations about who is qualified as a politician (Carnes, Reference Carnes2018; Miller, Reference Miller1995; Sevi et al., Reference Sevi, Blais and Mayer2020). In addition, competitive races with multiple nomination candidates are more financially costly. These realities should make it less likely that working-class candidates are nominated in competitive ridings.Footnote ⁵

H3: The more competitive a riding, the less likely candidates will be to reference working-class identities and the more likely they will be to reference white-collar identities.

Data and Methods

Measuring class identity

We require measures of the class identities that candidates emphasize when campaigning as well as their separately verified class identities. To measure the former, research assistants coded the biographies that candidates posted to their campaigning websites during the course of the 2022 election campaigns in Ontario and Quebec. While we do not expect that all or most voters encounter candidates’ websites, we anticipate that the biographies will reflect the identities candidates wish to emphasize. For instance, the biographical information on these websites typically also appears on campaign literature that is widely distributed. Unlike campaign literature, data from these biographies is easy to gather. Unlike media reports on candidates, they are unmediated by journalists who may have their own views about which candidate identities are important. Biographies thus present easily codable textual data that allow us to pinpoint the identities candidates view as important.

We collected screenshots of all candidate biographies linked from the major parties’ websites during the final week of the campaign: May 26, 2022, for Ontario and between September 28–30, 2022, for Quebec. Screen captures were hand-coded using NVivo qualitative analysis software by a research assistant for mentions of occupation, education and union affiliation. We did not attempt to interpret the tone or context behind a class identifier. We allowed for candidates to be coded in multiple categories such that they could express multiple occupations (including past occupations) and educational qualifications. This was done because we expect that candidates will draw on various life experiences to build affinity with voters. Candidates that dropped out or were suspended by their parties were excluded from the analysis.Footnote ⁶

One research assistant coded biographies for Ontario candidates and a second research assistant coded Quebec candidates. Research assistants were asked to code each specific career experience mentioned in a candidate biography by noting the exact job title. Coders were not asked to determine how specific occupations fit into any of the broader categories used in the analysis, limiting the need for subjective interpretation. Coders did the same for references of educational credentials and union membership. The authors then classified each specific occupation mentioned into broader categories using the descriptions of NACIS categories included in Table 1 provided by Statistics Canada (Statistics Canada, 2024c). As with occupation, the authors combined codes for the specific university qualifications into the broader categories used in the analysis (for example, mentions of a PhD, medical degree, law degree and master’s were combined into a category for advanced degree).

Table 1.

Aggregations of Occupation Categories

Our analysis features three indicators of class identity: occupation, education and union membership. We created seven occupational categories based on aggregations in the North American Industry Classification System (NAICS) used in the Canadian census, as shown in Table 1. Educational background contains four possible categories: student, college, bachelor's degree and advanced degree. Union membership captures whether a candidate mentioned a union affiliation.

It also bears examining whether candidates ignore or embellish certain aspects of their class identities within their biographies. To measure this, research assistants gathered data on past occupations and education for as many candidates as possible, from the social networking platform LinkedIn, as well as publicly available news sources.Footnote ⁷ We obtained independent occupational information for most of our candidates: 359 of 369 total candidates in Ontario and 597 of 622 in Quebec. Educational information was more challenging to find. We obtained this for 292 candidates in Ontario and 474 in Quebec. Research assistants recorded the exact job mentioned in a candidate's LinkedIn or publicly available news stories. The authors then determined how to classify the job into broader categories using the same NACIS descriptions as were used when coding candidate biographies.

Results and Analysis

Candidates’ expression of class identities

Figure 1 shows the proportion of candidates that reference different types of occupations (because candidates can mention multiple or no occupation, bars do not add up to 1). The most referenced types of occupations are white-collar public sector and white-collar private sector. In Ontario, 47 per cent of candidates referenced a white-collar private-sector occupation while 44 per cent referenced having held a white-collar public-sector occupation. The respective percentages for Quebec are 43 per cent and 45 per cent. References to working-class occupations are rare in both provinces. In Ontario, just 2 per cent of candidates referenced having held a blue-collar occupation and 2 per cent referenced a service-sector occupation. In Quebec, those numbers were 5 and less than 1 per cent, respectively (2 candidates out of 624). As expected, differences arise in the arts-and-culture sector. Roughly 12 per cent of Quebec candidates referenced arts and culture–sector occupations versus 7 per cent in Ontario.

Figure 1.

References to Occupation by Province

Bars show the total proportion of candidates referencing each category of occupation.

Figure 2 shows that the bulk of candidates referencing education referred to either having an undergraduate university degree or an advanced degree (as with occupation, candidates could reference multiple levels of education or no education). Provincial differences are minimal, although Quebec candidates were more likely to reference having college degrees (a result of Quebec's CEGEP system) and more likely to reference being students, at 7 per cent versus 2 per cent of Ontario candidates. An equal proportion of candidates in each province mention being members of unions, at 7 per cent.

Figure 2.

References to Education by Province

Bars show the total proportion of candidates referencing each education category.

Determinants of references

Figure 3 shows average marginal effects for variables of interest on mentions of different occupations. Party ideology matters most across occupation categories. Candidates for left parties are about 5 percentage points more likely to reference having an arts-and-culture occupation than candidates for the two Liberal parties and 18 percentage points less likely to mention having a white-collar private-sector occupation. They are also about 16 percentage points more likely to mention having a white-collar public-sector occupation than candidates for right parties.Footnote ¹¹ Candidates for the three right parties by contrast are 18 percentage points more likely to reference a white-collar private-sector occupation than candidates for left parties.Footnote ¹²

Figure 3.

Effects on Mentions of Occupation

This figure shows average marginal effects. Lines show ranges for a 95 per cent confidence level. * p<0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001. Full regression models are in table A1 in the appendix.

The figure shows little evidence that the contextual or competitiveness variables matter consistently across each occupation type. Riding competitiveness only appears to matter to mentions of arts-and-culture occupations, with a 10 per cent increase in the margin of victory in a riding associated with a modest 1 percentage point increase in the likelihood that a candidate mentions such occupations.

Interestingly, whether a riding is in Quebec matters minimally to the mention of occupation categories. This suggests that party ideology overwhelms provincial differences. We find important differences between parties of similar ideologies across different provincial contexts. Figure 4 shows clear differences in references to occupation between left parties. As expected, PQ and QS candidates are more likely to reference having arts-and-culture sector occupations. This is demonstrated and discussed in more detail in Appendix C. The PQ also stands out with respect to mentions of white-collar public-sector occupations.

Figure 4.

Effects on References to Occupation (Left Party Candidates)

This figure shows average marginal effects. Lines show ranges for a 95 per cent confidence level. * p < 0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001. Full regression models are found in table A2 in the appendix.

We find key differences across the three right parties. Figure 5 shows these are related to mentions of both white-collar private- and public-sector occupations. For example, Quebec Conservative candidates are about 29 percentage points more likely than Ontario PC candidates to mention white-collar private-sector occupations and 22 percentage points more likely than candidates for the CAQ.

Figure 5.

Effects on References to Occupation (Right Parties)

This figure shows average marginal effects. Lines show ranges for a 95 per cent confidence level. * p < 0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001. Full regression models are found in Table A4 in the appendix.

Figure 6 shows that, like with occupation, party ideology stands out as most consistently affecting mentions of education. Candidates for the two liberal parties are least likely to mention being a student, and left party candidates are also about 4 percentage points more likely to mention being students than right party candidates. In terms of degree type, right party candidates are almost 9 percentage points more likely to reference having a college credential than liberal party candidates, while left party candidates are only 3 percentage points more likely. Right party candidates are also about 11 percentage points less likely than liberal candidates to reference holding an advanced degree.

Figure 6.

Effects on References to Education

This figure shows average marginal effects. Lines show ranges for a 95 per cent confidence level. * p < 0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001. Full regression models are in table A5 in the appendix.

We find little evidence that the demographic make-up of ridings or their competitiveness shapes references to education. None of the measures for the proportion of credential holders in a riding correlates with the likelihood of a candidate referencing such a credential. Comparing between provinces also reveals minimal differences. For example, Quebec candidates are just under 2-percentage points less likely to reference having advanced degrees than Ontario candidates. As with occupation though, there may be differences within party ideologies that can be disaggregated further. Figure 7 shows minimal evidence that comparing left parties captures distinctions between Ontario and Quebec candidate references to educational background.Footnote ¹³ However, QS candidates were more likely to reference having university degrees than Ontario NDP candidates: 26 percentage points more likely to reference bachelor's degrees and 27 points more likely to reference an advanced degree. They were also more likely to refence having a university degree than PQ candidates (16 per cent for bachelor's degrees and 17 for advanced).

Figure 7.

Effects on References to Education (Left Parties)

This figure shows average marginal effects. Lines show ranges for a 95 per cent confidence level. * p < 0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001. Full regression models are in table A6 in the appendix.

The dynamics between the liberal parties are the inverse of those between QS and the Ontario NDP. Figure 8 shows that PLQ candidates were about 28 percentage points less likely than Ontario Liberals to reference having a bachelor's degree and about 16 percentage points less likely to reference having an advanced degree.

Figure 8.

Effects on References to Education (Liberal Parties)

This figure shows average marginal effects. Lines show ranges for a 95 per cent confidence level. * p < 0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001. Full regression models are in Table A7 in the appendix.

Turning to union membership, Figure 9 shows the effect of our variables of interest on the likelihood of a candidate mentioning a union affiliation. As expected, left party candidates are most likely to mention this, being about 11 percentage points more likely than both liberal parties and right parties to do so (only the comparison with liberal parties is shown in the figure).Footnote ¹⁴

Figure 9.

Effects on References to Union Membership

This figure shows average marginal effects. Lines show ranges for a 95 per cent confidence level. * p < 0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001. Full regression models are in Table A9 in the appendix.

Differences between provinces are more substantial when we compare parties within ideological groups. Of the left parties, Ontario NDP candidates are most likely to mention union membership. They are around 13 percentage points more likely than QS candidates and 9 percentage points more likely than PQ candidates to do so. This dynamic is reversed for the liberal parties, with PLQ candidates about 21 percentage points more likely than OLP candidates to mention being a union member.

Overall, the results show that competitiveness and riding context explain little with regards to references to class identity. The low number of candidates that reference working-class occupations makes it difficult to address the first part of H3. With respect to the white-collar components of H3, there does not appear to be a relationship between competitiveness and the likelihood of a candidate referencing a white-collar occupation (private- or public-sector). One could use education as a proxy for class here, with college certifications indicating a connection to the working-class and university degrees a connection to white-collar jobs. This provides no help to either component of H3 as we find no evidence of a negative relationship between party competitiveness and references to college credentials or a positive relationship between competitiveness and references to either bachelor's or advanced degrees (in fact, the relationship with bachelor's degrees is in the opposite direction). The evidence for the impact of riding context is also limited. Consistent with H2a, we find that candidates are more likely to reference white-collar private-sector occupations when running in ridings with larger numbers of such.Footnote ¹⁵

Of our two contextual hypotheses related to province, we find some support for H2b, that candidates in Quebec are more likely to reference arts-and-culture occupations as well as being students. Our descriptive statistics match both components of this hypothesis. However, both differences disappeared once we controlled for party ideology, riding demographics, competitiveness and candidate characteristics when looking at all parties. That said, left party candidates for the two Quebec left parties are more likely to reference arts-and-culture occupations than ONDP candidates.

The clearest differences are between parties. Despite the aforementioned difficulty of testing the blue-collar component of H1a, we find hints of support. Candidates for left parties were more likely to reference having college credentials than candidates for liberal parties (though interestingly not more likely than right party candidates), and more likely to reference union membership than candidates for any other parties. With respect to the white-collar public-sector component of H1a, candidates for left parties were more likely to reference white-collar public-sector jobs than candidates for right parties, though not more likely than liberals. In line with H1b, candidates for right parties were more likely to reference white-collar private-sector jobs than candidates for left parties, but we did not find evidence they were more likely to do so than liberals. Also consistent with H1b, we find that right party candidates were more likely than liberal candidates to reference a college credential. Lastly, as per H1c, we find some evidence that liberal candidates are more likely to reference having university degrees, but only for advanced degrees and only when compared to right parties.

Candidate Biographies and “Hidden Identities”

Finally, we turn to the question of whether elements of candidates’ class identities are omitted or embellished in campaign biographies. To address this, we compared what candidates said in their biographies to other, publicly available descriptions of candidate backgrounds. As such, we searched publicly available candidate profiles available on the platform LinkedIn, as well as news stories about candidates that contained biographical facts. We focus on education and occupation, as we were not able to find consistently available information verifying union membership. We found information on education for 766 candidates (292 in Ontario and 474 in Quebec) and occupation information for 956 candidates (359 in Ontario and 597 in Quebec). Candidates for whom we were unable to find information were dropped.

In terms of limitations to this analysis, the distribution of candidates without publicly available information is not random. Lower profile candidates in less competitive ridings are less likely to appear in online searches, especially if they are young candidates without substantial education or job experience. Candidates who have been in elected office for long periods at any level of government are also less likely to have occupation information available that predates their time in office. Moreover, just as candidates can choose what they include in their biographies, they can also curate social media sites such as LinkedIn to avoid sharing certain experiences. Inherent in working with publicly available data is being limited to working with candidates who were willing to post education and occupational information on public sites such as LinkedIn or to share that information with journalists.

We find some under-referencing of educational qualifications amongst candidates for all degree levels and in both provinces. Table 2 shows that, in Ontario, we found a college or equivalent qualification that was not referenced in the candidates’ biography for about 15 per cent of candidates, as well as bachelor's and advanced degrees not referenced for about 56 per cent and 25 per cent, respectively. There are similar numbers in Quebec, with 49 per cent of candidates having a bachelor's degree that they did not reference and 21 per cent having an advanced degree not referenced. There are more Quebec than Ontario candidates with unreferenced college degrees (about 45%), and we can speculate that this stems from Quebec's CEGEP system providing most candidates with a college credential. We can also speculate that unreferenced bachelor's degrees in both provinces may result from candidates who choose to mention holding an advanced degree but place less importance on referencing their bachelor's.

Table 2.

Mentioned vs. Verified Educational Qualifications

Percentages are in brackets (rows sum to 100% for each province).

There were few instances where candidates referenced a university qualification that we were unable to verify. For bachelor's and advanced degrees these categories each fell under 3 per cent. In Ontario, just over 3 per cent of candidates referenced a college qualification that we could not verify, while in Quebec the proportion fell under 3 per cent.

Our analysis reveals cases where occupations are substantially under-represented in candidate biographies, but not to the point where they would change our broad conclusions about the distribution of different occupations amongst candidates. It is first worth noting our two working-class occupation categories: blue-collar and service-sector occupations. Table 3 shows that just under 3 per cent of Ontario candidates had a blue-collar occupation that they did not reference in their biography while just over 3 per cent did the same for service sector occupations. In Quebec, these numbers are 1 per cent and just under 7 per cent, respectively. While these numbers are small relative to the total number of candidates, they are large relative to the number of candidates that reference blue-collar and service-sector occupations. Only 2 per cent of Ontario candidates referenced blue-collar occupations in their biographies. Including candidates that had but did not reference blue-collar occupations more than doubles the number of blue-collar candidates in our analysis. This is also true for service sector candidates in Ontario. In Quebec, a lower proportion of candidates had unmentioned blue-collar occupations but including them would still have increased that category from about 5 per cent of candidates to about 6 per cent. Most Quebec candidates who held service-sector occupations did not mention such occupations. Only 2 candidates (less than 0.5%) mentioned having such occupations, but we found that just under 7 per cent of candidates in Quebec had service-sector experience that they declined to mention.Footnote ¹⁶ For example, Benjamin Roy, the PLQ candidate in Saint-Jean is reported by local media as working at an automotive service centre but no occupation is mentioned in his online biography (Macfarlane, Reference Macfarlane2022).

Table 3.

Mentioned vs. Verified Occupations

Percentages are in brackets (rows sum to 100% for each province).

Despite the fact that a substantial proportion of candidates with blue-collar and service-sector occupations chose not to mention such occupations in their biographies, we are confident about our conclusion that the number of blue-collar and service-sector candidates running for office is relatively small. Even if we add together those who did and did not mention a blue-collar occupation, we end up with about 5 per cent of candidates in Ontario and about 6 per cent in Quebec. For service-sector occupations we find about 6 per cent in Ontario and about 7 per cent in Quebec. While there are some “shy” working-class candidates who may hide such experiences for strategic reasons, there are not so many so as to change our conclusions on the rarity of working-class candidates.

We found that larger numbers of candidates had white-collar (either private- or public-sector) occupations that they did not mention in their biographies: approximately 27 per cent of candidates in Ontario and 26 per cent of candidates in Quebec had unmentioned white-collar private-sector occupations. These numbers for white-collar public-sector occupations were about 21 per cent and 20 per cent in Ontario and Quebec respectively.Footnote ¹⁷ With these results, it is worth recalling that our initial analysis of campaign biographies allowed for candidates to reference multiple occupations. As such, these discrepancies between referenced and unreferenced occupations often result from those who chose to mention only a selection of their career experience, for example, by mentioning a white-collar private-sector job but not their white-collar public-sector experience. These findings suggest that there is interesting, further work to be done, both quantitative and qualitative, on what drives some candidates to mention particular identities and to hide or embellish others.

Discussion and Conclusion

Our analysis highlights the substantial underrepresentation of working-class identities among Ontario and Quebec candidates. First, we show that electors in both provinces were unlikely to encounter candidates that presented themselves as working-class. This is both because few candidates who ran for office had working-class backgrounds they could reference, and those that had such backgrounds sometimes chose not to reference them. Indeed, the difference between the content of campaign biographies and candidates’ actual career and educational experience suggests that working-class identities are underreported in campaign communications. This is somewhat surprising in a context where federal and provincial parties are making increased appeals to working-class voters (Budd, Reference Budd2020; Koop and Farney, Reference Koop, Farney, Robbins-Kanter, Koop and Troup2025; Speer, Reference Speer2021). We also specify conditions under which working-class markers of identity are more likely to appear. Union membership is more frequently referenced by left party candidates, while candidates for both left and right parties are more likely to emphasize college credentials versus candidates for liberal parties.

For other occupational backgrounds encompassing substantial numbers of candidates (arts and culture, white-collar private sector, and white-collar public sector), we find that party ideology tends to most clearly shape the identities that candidates hold and emphasize. In terms of white-collar occupations, parties of the right stand out most clearly. Candidates for these parties are substantially less likely to emphasize white-collar public-sector occupations and more likely to emphasize white-collar private-sector occupations than their counterparts in other parties. These findings suggest that expressions of candidate identity, at least in these elections, tend to conform with the ideological commitments of their parties. Unsurprisingly, right parties, which tend to advocate for decreased public spending and smaller government, tend to feature fewer candidates who identify as public-sector workers. There is also a particular class orientation evident in the ideologically centrist liberal parties, as candidates for such parties are substantially more likely to reference holding advanced degrees. Given the impact of party affiliation, whether a voter is likely to have a candidate they feel comfortable supporting who shares their class background is constrained by their ideological positioning and those of the parties they are willing to support.

Where provincial context matters, as is the case for arts-and-culture occupations, the difference in provincial context tends to manifest within ideological groupings. In the case of arts and culture, it is candidates for left parties in Quebec that are more likely to emphasize arts-and-culture occupations than candidates for the Ontario NDP. Generally speaking, the differences between the class backgrounds held and articulated by candidates in Ontario versus Quebec are subtle and minor. This may be surprising given longstanding assumptions about socio-political differences between provinces, yet it suggests that similar ideological and strategic forces shape the class identities that candidates hold and choose to emphasize.

These findings have important implications for how Ontarians and Quebecers see themselves reflected in electoral politics. Working-class citizens are far less likely than other class groups to see people like themselves represented amongst politicians. While it can be challenging and contentious to identify the specific markers of working-class identity (Pilon, Reference Pilon and Laycock2019), the reality of substantial working-class underrepresentation is clear. For example, in 2023, 20.4 per cent of Canadians were employed in goods-producing sectors (Statistics Canada, 2024b) but only 2 per cent of Ontario candidates in 2022 mentioned blue-collar identities. In 2023, roughly 39 per cent of Quebecers and 26 per cent of Ontarians were union members. However, just 7 per cent of candidates in each province referenced a union affiliation in their campaign biographies. Conversely, as seen in federal politics, white-collar candidate identities are ubiquitous across political parties, including those on the left. Beyond the dearth of working-class politicians, we have shown that working-class markers of identity are not seen as relevant to political communications. Focusing overwhelmingly on white-collar qualifications may perpetuate the stereotypical view that politicians should come from these backgrounds.

This is notably despite parties’ increased overtures to working-class voters. For Ontario and Quebec parties, these efforts seemingly do not yet extend to recruiting working-class candidates who can emphasize a common class membership with large swaths of citizens. As mentioned, the continued emphasis on white-collar identities in campaign biographies may stem from the fact that these experiences more closely conform with stereotypical notions of who is qualified to be a politician. However, some evidence shows that citizens in advanced democracies are open to voting for politicians from a variety of class backgrounds (Campbell, and Cowley, Reference Campbell and Cowley2014a; Carnes and Sadin, Reference Carnes and Sadin2015; O'Grady, Reference O'Grady2019). For example, Campbell and Cowley (Reference Campbell and Cowley2014b) find that large segments of UK voters may actually prefer politicians from less affluent backgrounds. These findings suggest an opportunity for parties and candidates to “normalize” working-class life experience in their campaign communications.

Class identity is a neglected topic in Canadian politics. Scholars have argued that class consciousness is relatively weak and other social cleavages take precedence (Porter, Reference Porter1965; Gidengil, Reference Gidengil and Baer2002). Some note that class divides lack importance as a structuring force in the party system, linking this to a lack of class voting and class-based parties seen in other Western democracies (Johnston, Reference Johnston2017: 39; Ogmundson, Reference Ogmundson1975: 506). These realities undoubtedly weaken prospects for class-based representation in political discourse and the party system. Importantly, this only signifies that class divides are often overlooked, rather than nonexistent (Pilon, Reference Pilon, Robbins-Kanter, Koop and Troup2025). Canadian society is still characterized by sharp material differences, including economic inequality that is rising faster and at a more sustained level in Canada relative to other OECD countries (Polacko, Reference Polacko2020: 1325). The lack of class-based representation in other areas arguably highlights the importance of descriptive representation in terms of individual politicians who are willing to express a variety of class identities.

Most studies of descriptive representation consider the extent to which politicians reflect politically relevant traits of the electorate. This article adds an additional dimension to this topic by considering not only static markers of identity, but also the political choices made to highlight or downplay certain candidate characteristics. Future research might examine whether elected politicians who vocalize a particular identity or group membership in their campaign communications are more inclined to advocate for its substantive interests.