How Do Canadian Public Sector Employees Vote? A Look at the Evidence

Bottom line up front

There is systematic Canadian evidence that public sector employees lean to the left of private sector workers — primarily toward the NDP and Liberals, and away from the Conservatives — but the effect is more modest than American partisan rhetoric about a "deep state" or a "public class" would suggest. The single best academic source is a 2026 paper by Matthew Polacko, Simon Kiss and Peter Graefe in the Canadian Review of Sociology ("Revisiting the Sectoral Cleavage in Canada"), which uses pooled Canadian Election Study (CES) data from 1968–2019. There is, however, no publicly released Canadian poll that breaks down decided‑voter support by employment sector with the kind of "X% Liberal / Y% Conservative / Z% NDP" topline that exists for, say, gender or age. Most evidence comes from academic regression analyses, riding‑level returns in Ottawa, union endorsement patterns, and the federal Public Service Commission's biennial Non-Partisanship Survey — which by design does not ask employees how they vote. The discussion below pulls these strands together and is explicit about what the data can and cannot tell us.

1. The core academic finding: a real but moderate sectoral cleavage

Polacko, Kiss & Graefe (2026), Canadian Review of Sociology

This is the most current and comprehensive Canadian study. Using every Canadian Election Study from 1968 to 2019, the authors find:

A statistically significant "sectoral cleavage": public sector workers are less likely to vote Conservative than private sector workers, controlling for demographics.
The largest positive effect is on NDP support, not Liberal support — public sector employees are more likely than private sector workers to vote NDP.
Most of the public sector NDP/Liberal advantage disappears once union membership is controlled for: i.e., the public sector skews left in large part because it is heavily unionized, not because government work itself produces a partisan identity.
Within the public sector, the divide is concentrated among managers and professionals, who are markedly more left-wing on economic issues (redistribution, welfare spending, government job creation, public child care) than their private-sector counterparts. Working-class and routine non-manual public sector employees (which would include the parks workers, sanitation workers, transit operators and groundskeepers the user is curious about) show almost no policy difference from their private-sector working-class counterparts.
On social and cultural issues (immigration, moral traditionalism), public and private sector workers look nearly identical.
Public and private sector employees vote at roughly the same rate — undermining the "rational budget‑maximizer" theory that would predict higher turnout among public servants.

The authors do not publish a single "X% of public servants vote Liberal" topline; their analysis is in the form of regression coefficients and predicted probabilities. But the thrust is clear: a small but real left-tilt, driven mainly by unionized professionals.

Earlier baseline: Blais, Blake & Dion (1990)

The classic Canadian study (Comparative Political Studies, 23(3): 381–403), using data through the mid‑1980s, found only a slight tendency for public sector workers to vote NDP and against the Progressive Conservatives — weaker than in the United States. Together with Polacko et al., this implies the cleavage has grown somewhat over time, particularly since 2004 as the Conservative Party consolidated and the public service unionized further.

Rounce & Levasseur (2015), Canadian Political Science Review

Using the SSHRC-funded Canadian Provincial Election Project (CPEP) covering eight provinces in the 2011-2012 provincial elections, this study examined public, private and non-profit sector employees. It found "diverging orientations" across sectors but considerable diversity within each — i.e., the public sector is not monolithic. It is one of the few Canadian studies to look at the provincial level specifically.

Polacko, Kiss & Graefe (2022), Canadian Journal of Political Science

This earlier paper on class voting (1965–2019) found a related and important countertrend: the working class outside Quebec abandoned the NDP for the Conservatives starting around 2004, driven mainly by cultural issues (immigration, moral traditionalism). This is relevant because it suggests blue-collar municipal workers — sanitation, parks, transit — may not lean as decisively left as one might assume from the union endorsement pattern of their employers.

2. What does not exist

To answer the user's question honestly: there are no publicly released datasets from Elections Canada, Statistics Canada, Abacus, Léger, EKOS, Angus Reid, Nanos or Ipsos that publish a "vote intention by employment sector" crosstab on a regular basis.

Elections Canada publishes turnout analyses (e.g., the 2010 working paper on turnout by age) but is statutorily prohibited from collecting partisan preference data.
Statistics Canada does not link political views to employment sector. Its work on voting (e.g., the 2012 Factors associated with voting analysis) covers turnout, not party choice. Notably, that StatCan analysis dropped union membership from its model because it was so highly collinear with public-sector employment.
The major commercial pollsters do publish union-member crosstabs occasionally, but public-sector employment as a separate breakdown is not standard. Researchers generally have to use the Canadian Election Study (an academic SSHRC-funded post-election survey) to get at this question.

3. The Public Service Commission's Non-Partisanship Survey: what it does and doesn't measure

The federal Public Service Commission runs the biennial Staffing and Non-Partisanship Survey (SNPS) — most recent waves in 2018, 2021 and 2023, with a 2025 wave underway. The 2023 survey went to all employees in 77 federal departments under the Public Service Employment Act; 84,741 responded for a 31.7% response rate, representing 273,186 employees.

The SNPS asks about behaviour and awareness, not vote choice. Because federal public servants have a constitutional right to political activity but a statutory duty of impartiality, the survey simply asks whether respondents engaged in any political activity beyond voting. Reporting in Policy Options (April 2025, Kathryn May) summarized the trend:

2018: ~98% of federal employees reported they had not engaged in any political activity beyond voting.
2023: 94% — a small but notable softening.
73% in 2018 felt their work units operated impartially; 91% in 2023.

So the survey gives no direct read on partisanship, but it does suggest a slow drift toward more political engagement among federal public servants. Crucially, the Values and Ethics Code for the Public Sector and the political-activity provisions of the PSEA require federal employees to obtain permission from the PSC before seeking nomination or candidacy. This regime arguably suppresses observable partisan behaviour — measured donations, public endorsements, candidacy — relative to what employees actually believe.

The 2025 candidacy data point

The PSC confirmed in 2025 that 65 federal public servants sought permission to run in the April 2025 election, of whom 24 ended up on the ballot for nine different parties. The Conservatives had the largest share with nine federal-public-servant candidates — to the surprise of many observers, since the older stereotype had been of a Liberal-leaning bureaucracy. Reporters covering the story noted that this complicates, rather than confirms, the simple "Liberal-dominated bureaucracy" myth. But 24 candidates out of ~273,000 federal employees is too small a number to be representative.

4. Ottawa-area ridings as a real-world proxy

Because the National Capital Region has the highest concentration of federal public servants in the country, Ottawa ridings are the most-studied indirect indicator. The pattern over decades:

Ottawa ridings with the highest civil servant density — historically Ottawa—Vanier and Ottawa South — have been some of the safest Liberal seats in the country. Ottawa—Vanier, with high francophone and civil servant populations, was held by the Liberals continuously from its creation. Ottawa South has been held by Liberal David McGuinty continuously since 2004.
Ottawa Centre, which contains Parliament Hill itself and many public-sector workers, has flipped Liberal/NDP repeatedly. Ed Broadbent (NDP) won it in 2004; Paul Dewar (NDP) held it 2006–2015; Catherine McKenna (Liberal) won it 2015 and 2019; the seat had the highest turnout in Canada in 2015. Provincially, the same boundaries have gone NDP ten times and Liberal eight times since 1968 — and never Conservative.
The suburban Ottawa ridings (Kanata, Nepean, Orléans) are swing/Liberal-leaning rather than Conservative strongholds even in years of strong national Conservative performance (e.g., 2011).
The most rural/exurban riding, Carleton, was the major Conservative seat — held by Pierre Poilievre from 2004 onward.

The 2025 election: a natural experiment

Conservative leader Pierre Poilievre lost his own Carleton seat to Liberal Bruce Fanjoy by roughly 4,300–4,500 votes on April 28, 2025, after holding it for two decades. Poilievre's own post-election explanation, given to CBC's The House, was that the loss came down to public-sector unions:

"It's an Ottawa riding with a lot of federal public servants who disagreed with that approach [his pledge to cut public service jobs]."

He blamed an "aggressive" campaign by public sector unions. While the riding was redrawn ahead of 2025 (transposed 2021 results would have still given Poilievre a win) and Trump-related sentiment also boosted the Liberals, the Carleton result is the strongest recent practical illustration of a public-sector backlash against a Conservative leader who openly campaigned on cutting federal jobs (he had pledged 17,000 cuts/year). Across the National Capital Region, Liberals also retained Ottawa—Vanier, Ottawa South, Ottawa West-Nepean, Ottawa Centre, Orléans and Kanata, and Mark Carney himself won Nepean. The only public-sector-heavy Ottawa riding the Conservatives had recently been competitive in (Carleton) was lost; the urban core stayed Liberal/NDP.

This is consistent with the academic finding: a moderate left tilt that becomes electorally decisive when a Conservative leader explicitly threatens public-sector employment.

5. Public sector unions and political endorsements

Direct vote breakdowns by union are hard to find, but endorsement and political-activity patterns are well documented:

CUPE (Canadian Union of Public Employees, ~750,000 members — the largest union in Canada, covering most municipal workers, school boards, hospitals and many provincial public services) is a long-standing, formal "founding partner" of the NDP and endorsed Jagmeet Singh and the NDP in the 2025 federal election.
PSAC (Public Service Alliance of Canada, ~245,000 federal public service members) declined to issue a formal endorsement in 2025 but ran a "For You, Canada" campaign explicitly warning voters about Conservative cuts and posted ads stating "Pierre Poilievre is not for you, Canada."
PIPSC (Professional Institute of the Public Service of Canada), CAPE (Canadian Association of Professional Employees), NUPGE (National Union of Public and General Employees), and the Canadian Federation of Nurses Unions were all listed by Canadian Dimension (2025) among the 13 national unions that ran third-party election campaigns; most were anti-Conservative rather than formally pro-NDP.
The Amalgamated Transit Union and United Steelworkers also formally endorsed the NDP in 2025.
On the Conservative side at the federal level in 2025, the Boilermakers were the only national union to endorse. Ontario provincial Conservative Premier Doug Ford has been more successful at peeling off building-trades, firefighter, police and some Unifor locals — but these are mostly private-sector or quasi-public safety unions rather than the parks/sanitation/transit workers in the user's question.

A relevant caveat from research published in The Conversation (2025): survey work on the actual influence of union endorsements on members' votes shows the effect is modest, and union type does not significantly change responsiveness — public-sector union members are no more likely to follow their union's endorsement than private-sector union members. So the leadership's NDP affiliation does not translate one-for-one into NDP votes among rank-and-file members.

Financial donations are constrained by Canada's strict election-finance regime: corporate and union political donations to federal parties have been banned since 2007 (after a phased reduction starting in 2003), so there is no equivalent to the U.S. literature on federal-employee donation patterns (e.g., Bonica et al.). What unions can do is run third-party advertising, which is registered with Elections Canada — and the public-sector unions' third-party spending has consistently targeted Conservatives.

6. Municipal-level workers (parks, sanitation, transit)

This is the hardest layer to measure, and the answer is that there is essentially no Canadian survey research that breaks out vote choice for blue-collar municipal employees specifically. What we have:

Most municipal workers in major Canadian cities (Toronto, Vancouver, Montreal, Calgary, Ottawa) are CUPE members or members of CUPE-affiliated locals (e.g., CUPE 416 — Toronto outside workers; CUPE 79 — Toronto inside workers; CUPE Local 15 — Vancouver). Municipal elections in Canada are non-partisan in most provinces (Vancouver and Montreal use civic parties; Toronto and Calgary do not have formal party labels), which makes union endorsements muddier than at the provincial/federal level.
The Polacko et al. finding that the public-sector NDP advantage among routine non-manual and working-class employees is statistically indistinguishable from their private-sector counterparts is the single most important data point here. In other words: the Polish-Canadian groundskeeper at a Toronto park is, on average, very close to the warehouse worker at a private logistics firm in their voting patterns. The public-sector left tilt is concentrated in professional and managerial ranks (analysts, policy advisors, scientists, lawyers, executives), not in frontline operations.
A 2022 Canadian Journal of Political Science analysis (Polacko, Kiss & Graefe) showed the working class outside Quebec has shifted noticeably toward the Conservatives since 2004, with cultural issues driving the shift. Canadian Dimension (May 2025) noted that several high union-density constituencies — including Windsor West with a 23-point Conservative swing — flipped or moved sharply right in 2025.

So the user's intuition watching municipal groundskeepers should be tempered: their union may endorse the NDP, but the workers themselves are likely more politically heterogeneous than any of Toronto's downtown federal-public-servant ridings.

7. Historical context

Pre-2000s: Blais, Blake & Dion (1990) found only a weak public-sector cleavage, smaller than in the U.S.
Harper era (2006–2015): Stephen Harper's government had a famously frosty relationship with the federal public service (Phoenix pay system rollout, contracting out, perceived muzzling of scientists). Polacko et al. show the sectoral cleavage grew in this period — public sector NDP/Liberal voting became more pronounced from 2004 onward. Ottawa-area Conservative gains in 2006 and 2011 were almost entirely in the rural/exurban portions of the National Capital Region; the urban civil-servant core stayed Liberal or NDP.
2015: Liberal sweep took every Ottawa-area riding except Carleton.
2019 and 2021: Liberals retained most Ottawa-area seats; Carleton stayed Conservative.
2025: The first election in which a Conservative leader directly campaigned on public service cuts — and the first in which the leader lost his own public-sector-adjacent riding. PSAC ran the most overtly political campaign in its history without a formal endorsement. CUPE formally endorsed the NDP. The pattern of public-sector unions running de facto anti-Conservative campaigns intensified.
Looking forward: The Carney government's November 2025 budget proposed cutting the federal public service by ~40,000 positions (≈10%). Polacko, Graefe and Kiss's December 2025 Conversation/McMaster commentary explicitly predicts this could push public-sector Liberal voters back to the NDP in the next election — a testable prediction.

8. Comparison to the United States and United Kingdom

For context, since direct Canadian numbers are scarce:

United States: Multiple studies (e.g., using FEC donation data — Bonica, Chilton, Rosenthal, Sen) find federal civilian employees donate disproportionately to Democrats, with the imbalance especially stark at agencies like the EPA, State Department and Department of Education. Bureau of Labor Statistics workers in 2016 donated overwhelmingly Democratic. The U.S. literature is much richer because campaign-finance disclosure is donor-level.
United Kingdom: British Election Study and YouGov data have repeatedly shown public-sector workers more Labour-leaning than private-sector workers, with the gap having grown since the Cameron-era austerity programme. Tepe (2012, Public Administration, 11 European countries) found a moderate but consistent public-sector left tilt across Western Europe.
Canada falls in between: a clear cleavage exists but is smaller than in the U.S. and somewhat smaller than in the U.K., consistent with the Blais, Blake and Dion (1990) framing that Canadian sectoral voting is comparatively muted.

9. Methodological caveats — important

Self-reported vote intention (the basis of CES, Léger, Abacus, etc.) suffers from social desirability and recall bias. Public-sector respondents may also be more cautious about disclosing partisan views given the Values and Ethics Code and the Public Service Employment Act political-activity provisions.
Sample sizes are small when the public sector is a subgroup of a national survey; that's why most rigorous Canadian work pools multiple CES waves.
"Public sector" is heterogeneous. Federal employees, provincial employees, healthcare workers, K-12 teachers, university professors, Crown corporation employees, and municipal sanitation workers are all "public sector" but have very different working conditions, unionization, and demographics. Most academic work bundles them, but the few studies that disaggregate (e.g., Rounce & Levasseur 2015) find substantial within-sector diversity.
No vote-choice data are collected by government surveys. The PSC's Non-Partisanship Survey deliberately avoids asking about vote choice. Statistics Canada surveys do not link political preference to employment sector. So all direct evidence on partisanship has to come from academic surveys (CES) or commercial polls that occasionally choose to publish a sector crosstab.
Riding-level data is contaminated by everything else about a riding (income, education, ethnicity, language, density). Ottawa Centre is not Liberal/NDP only because of public servants; it is also urban, highly educated, francophone-influenced, and renter-heavy.
Donation data is unavailable for individuals identified as public servants because Canadian federal donation limits are low ($1,775 per party per year in 2024) and disclosure does not include occupation in a structured field, unlike the U.S. FEC system.
Federal political-activity rules: Under the PSEA, federal public servants must obtain permission from the Public Service Commission before seeking a nomination or running as a candidate at any level. Lower-profile activities (donating, displaying signs, attending rallies, social-media expression) are permitted but must not impair perceived impartiality. This regulatory regime suppresses the visible footprint of partisan behaviour and is one reason direct measurement is hard.

Summary table of what we know

| Question | Best available answer |
|---|---|
| Do Canadian public sector workers lean left? | Yes, modestly. They are less likely to vote Conservative and more likely to vote NDP/Liberal than private-sector workers (Polacko, Kiss & Graefe 2026, using CES 1968–2019). |
| Is the lean uniform across the public sector? | No. It is concentrated among professionals/managers. Working-class and routine non-manual public-sector employees vote similarly to their private-sector peers. |
| Is it because they're rationally protecting their jobs? | Mostly no. The data are more consistent with left-leaning attitudes on economic issues, particularly redistribution and welfare-state spending, and with union membership, than with budget-maximizing self-interest. |
| Hard percentages (e.g., "60% Liberal")? | Not available in any released poll. Only regression-based effect sizes from academic studies. |
| Do Ottawa ridings prove the point? | They are consistent with it but contaminated by other variables. Carleton in 2025 is the most striking single data point. |
| Did Harper's term shift the cleavage? | Yes — Polacko et al. show the sectoral cleavage strengthened from the mid-2000s onward, coinciding with the modern Conservative Party's rise. |
| What about municipal frontline workers (parks, sanitation, transit)? | Their union (mostly CUPE) is firmly NDP-aligned; the workers themselves probably are not, on average, dramatically different from comparable private-sector blue-collar workers. The 2022 working-class shift toward the Conservatives applies here. |
| Will the 2025 Liberal budget cuts move public-sector voters? | Polacko, Graefe and Kiss explicitly predict (December 2025) that some will move from Liberal to NDP in the next election. This is a forecast, not yet an observation. |

A short reading list for the curious

Polacko, Matthew, Simon Kiss and Peter Graefe. 2026. "Revisiting the Sectoral Cleavage in Canada: Evidence From the Canadian Election Studies." Canadian Review of Sociology 63 (early view). The single best current source.
Polacko, Matthew, Simon Kiss and Peter Graefe. 2022. "The Changing Nature of Class Voting in Canada, 1965–2019." Canadian Journal of Political Science 55(3): 663–686.
Blais, André, Donald E. Blake, and Stéphane Dion. 1990. "The Public/Private Sector Cleavage in North America." Comparative Political Studies 23(3): 381–403.
Rounce, Andrea D., and Karine Levasseur. 2015. "Public, Private, and Non-profit Sector Employees: Voting Behaviour and Ideology in the 2011-2012 Provincial Elections." Canadian Political Science Review 9(3).
Public Service Commission of Canada. 2023 Staffing and Non-Partisanship Survey: Report on the Results for the Federal Public Service.
Polacko, Graefe & Kiss. December 2025. "What Canada's public sector voting divide could mean for future elections." The Conversation (the plain-language summary of the Canadian Review of Sociology paper).
Kathryn May, "Public servants running in record numbers flip partisan assumptions," Policy Options, April 24, 2025.

In short: the evidence supports the user's intuition that Canadian public sector employees lean somewhat to the left of the country as a whole — but the lean is real mostly among unionized professionals and managers, the magnitude is moderate rather than dramatic, frontline blue-collar municipal workers are politically more heterogeneous than their union endorsements imply, and Canada simply does not collect (or has not released) the kind of "X% Liberal / Y% Conservative / Z% NDP" topline that the question implicitly asks for.