Instrumental
Variance
Alberta separatism polling · September 2025 – April 2026 · A methodology record.
Eight polling firms. Eight months. The same electorate.
Every dedicated Alberta survey produced a result between 23% and 31%.
The numbers above 35% that dominated coverage were not from those surveys.
They were products of the instrument — the sample used, the question asked,
the scenario constructed. This is the forensic record.
How a Number Travels
The 72-hour propagation chain · May 23–25, 2025
On May 16–18, 2025, Léger polled 1,537 Canadians nationally. The Alberta subsample was 133 people — unweighted. Watch what happened to that number over the next 72 hours.
Alberta separatism · n=1,537 ✓
separation · n=133 ⚠
split" on independence ✕
researcher. 57%. ✕✕
Four outlets ran the same Canadian Press wire. The headline finding — 55% of Canadians understand Alberta separatism — is correctly sourced to a national survey of 1,537. The story is methodologically sound at the national level. The subsample problem is what comes next.
The Toronto Sun editorial board takes the same Léger survey and makes a claim the wire stories did not. The 47% Alberta subsample — 133 unweighted respondents — is presented as evidence that Albertans are "almost evenly split." The subsample size is not disclosed. The margin of error is not disclosed. The dedicated Alberta poll from the same firm one week earlier (n=1,000, 29%) is not mentioned.
behind the 47% claim
not disclosed
same firm, 7 days prior
The same 72-hour period produced US right-wing media coverage of Alberta separatism that cited no polling numbers at all — but built directly on the "moving toward secession" narrative the inflated figures had established. No methodology. No sample size. No caveat. The number had already done its work upstream.
The endpoint of the distortion chain. CO24 cites a poll from "the Canadian Institute for Public Opinion (CIPO)" — a name with real historical roots as the defunct Canadian arm of the Gallup organization, last active decades ago. It did not conduct this survey. The lead researcher quoted, "Dr. Elaine Chong," does not appear in any polling registry or academic database. The survey is described as n=3,200 Canadians, April 28–May 15 — a sample size and field period that matches no published poll. The topline: 57%. The actual Léger figure from the same week: 55%, n=1,537. Whether this is deliberate fabrication or AI-generated content that hallucinated a plausible institutional name, the result is the same: a fake poll, published, indexed, and potentially cited downstream.
Canada arm —
did not run this poll
professional
registry
published
survey
Not every outlet amplified. Trevor Harrison's opinion piece in the Calgary Herald interrogated the political use of the data rather than the data itself — and arrived at the same conclusion the polling architecture makes visible: the numbers reflect a political instrument, not a constitutional preference.
Signal vs. Noise
If the number is inside the band, it is a measurement. If it is outside the band, examine the instrument. Every dedicated Alberta survey conducted between September 2025 and April 2026 falls within the 23%–31% signal band. Every number above 35% is traceable to a specific methodological departure — a subsample, a modeled projection, a scenario construct. That departure is the variance. The instrument produced it.
Two Trackers. Same Population. Different Story.
Léger and Pollara both ran dedicated Alberta surveys in the same January–April 2026 window. They used different question wording, different field periods, and produced systematically different numbers. Placed side by side against their own prior waves, the two series tell divergent stories about the same electorate — and the divergence is larger than any single poll's margin of error.
| Firm | Field dates | n | Question type | Result | Movement vs. prior wave |
|---|---|---|---|---|---|
| · léger dedicated alberta series · | |||||
| Léger / Postmedia | May 9–12, 2025 | 1,000 | "Alberta becoming independent" — support/oppose | 29% | Baseline for series |
| Léger | Jan 23–26, 2026 | 1,003 | "Alberta should become an independent country" — 3-way choice | 18% | ↓ 11pp from May 2025 |
| Léger | Apr 3–6, 2026 | 1,003 | "Support for independence" — same 3-way choice | 23% | ↑ 5pp from Jan 2026 |
| · pollara dedicated alberta series · | |||||
| Pollara | Oct 2024 | Not published | "Vote for or against separating" — referendum binary | 20% | From Pollara tracker |
| Pollara (western) | Dec 5–20, 2025 | ~1,000 AB of 3,800 | "Vote for or against separating" — referendum binary | 19% | ↓ 1pp from Oct 2024 |
| Pollara | Mar 16–25, 2026 | 3,200 | "Vote for or against separating" — referendum binary | 27% | ↑ 8pp in ~13 weeks — described as "5-year high" |
| · cross-firm context · same period · | |||||
| Research Co. | Jan 4–6, 2026 | 703 | "Alberta becoming independent from Canada" — support/oppose | 31% | ↑ 9pp from June 2023 |
| Ipsos | Jan 9–14, 2026 | ~500 | "Vote for province to begin process of separating" — yes/lean | 29% | Drops to 16% after stress-test |
| Angus Reid | Feb 2–6, 2026 | 979 | 5-point spectrum including lean | 29% | Consistent with other dedicated surveys |
| Mainstreet | Feb 10–12, 2026 | 1,504 | "Independent country" — binary referendum, telephone IVR | 30% | Only telephone survey; consistent with online band |
| Abacus | Feb 24–Mar 1, 2026 | 1,000 | "Cease to be a province" — constitutional language | 26% | Lowest dedicated survey figure; constitutional framing |
Léger's January 2026 survey found 18% want independence — and 58% are concerned about separatist movements gaining influence. Mainstreet's February 2026 survey found 30% referendum intent — and 9% list independence as their top concern. Two firms, two methods, same finding: the population that would vote to separate is consistently larger than the population that considers it a priority issue. The gap between referendum intent and issue salience is not noise. It is the structural signature of a protest vote.
The Policy Lever
Separatism in Alberta is not an ideological position. It is a pressure instrument. The Mainstreet data (Feb 2026, n=1,504, telephone IVR) is the only survey that tested conditional scenarios empirically. The result: federal resource policy moves the number. American financial offers do not. Adjust the levers below to see what the data actually shows.
The Salience Stack
Mainstreet Research (Feb 10–12, 2026, n=1,504) asked Albertans to name their greatest issue of concern. The separation movement that polls at 30% referendum intent ranks ninth in issue salience at 9%. The ghost bar shows what "30% referendum intent" would look like if it were a real issue priority. It isn't. That gap is the piece.
The dashed outline is what 30% referendum intent would look like if it were a genuine top-priority issue. The solid bar is what Albertans actually reported when asked what concerns them most. The gap between those two bars is not apathy. It is the difference between a ballot posture and a lived priority. A movement where the majority of its own supporters rank it below crime, education, and immigration as a daily concern is not an independence movement. It is a pressure valve. The number exists to be cited, not acted on.
All Instruments · September 2025 – April 2026
Primary source record · Instrumental Variance
Primary source data for every poll reviewed. Field dates, sample sizes, and question wording confirmed from firm PDFs and press releases. Colour coding: green = dedicated Alberta survey, clean independence question; amber = conditional / scenario-based; red = subsample artifact or modeled projection.
| Firm | Field Dates | AB Sample | Question / Format | Result | Note |
|---|---|---|---|---|---|
| · dedicated alberta surveys · clean independence question · | |||||
| ThinkHQ Self-commissioned |
Sept 2025 n not published |
Albertans only | Clean "If a referendum held tomorrow, separate or remain?" Mock ballot, binary. |
23% | Baseline for current cycle. |
| Léger / Postmedia Postmedia |
May 9–12, 2025 n=1,000 AB |
Albertans only | Clean "Alberta becoming a country independent of Canada." Support/oppose scale. |
29% | Same firm asked national subsample 7 days later; produced 47%. CTV reported this correctly. |
| Research Co. Self-commissioned |
Jan 4–6, 2026 n=703 AB |
Albertans only | Clean "Alberta becoming a country independent from Canada." Support/oppose. ±3.7pp. |
31% | 42% among 18–34. Up 9pp from June 2023. |
| Ipsos Self-commissioned |
Jan 9–14, 2026 ~500 AB |
AB + QC (national study) | Clean (initial) "Vote for province to begin process of separating." Yes/lean/no/lean. Initial ask before stress-test. ±5.4pp. |
29% | Drops to 16% after cost stress-test. Committed 55%, conditional 25%, symbolic 20% of yes voters. |
| Angus Reid Self-commissioned |
Feb 2–6, 2026 n=979 AB |
Albertans only | Clean 5-point spectrum: Definitely leave / Lean leave / Lean stay / Definitely stay / Undecided. ±3pp. |
29% | Only 8% "definitely leave." 21% lean leave. |
| Mainstreet Self-commissioned |
Feb 10–12, 2026 n=1,504 AB |
Albertans only · telephone IVR | Clean "If a referendum held today on if Alberta should become an independent country, how would you vote?" Binary. ±2.5pp. |
30% | Only telephone survey in table. 9% list independence as top concern in same survey. |
| Abacus Data Self-commissioned |
Feb 20–25, 2026 n=1,000 AB |
Albertans only | Clean "The Province of Alberta shall become a sovereign country and cease to be a province of Canada." Agree/disagree. Constitutional language. |
26% | Constitutional framing produces lowest dedicated-survey figure. "Rock solid majority opposed." |
| Pollara Self-commissioned |
Mar 16–25, 2026 n=3,200 AB |
Albertans only · largest sample | Clean "If a referendum held tomorrow — vote for or against separating?" Binary, decided vote. ±1.7pp. |
27% | "5-year high." Tracker began Jan 2021. Not an all-time record. |
| · conditional / scenario-based · same surveys, different questions · | |||||
| Léger / Postmedia | May 9–12, 2025 n=1,000 AB |
Albertans only | Scenario "AB, SK, BC, MB forming an independent country." Multi-province western bloc scenario. Same survey as 29% row. |
35% | Different question, same survey. Western bloc — not Alberta independence. Often unreported separately. |
| Mainstreet | Feb 10–12, 2026 n=1,504 AB |
Albertans only | Scenario "If US promised economic union + dollar-parity exchange, how would you vote?" US annexation scenario. |
28% | −2pp from clean question. US financial offer suppresses support. APP's Washington pitch does not move Albertans. |
| Mainstreet | Feb 10–12, 2026 n=1,504 AB |
Albertans only | Scenario "If proposed pipeline agreement between Smith and Carney fell apart, how would you vote?" |
33% | +3pp from clean question. Federal resource policy failure is the lever. Highest conditional figure. |
| · numbers above 35% · instrument failures · | |||||
| Léger / Canadian Press | May 16–18, 2025 n=133 unweighted n=171 weighted |
National survey AB subsample only |
Subsample "Support or oppose Alberta becoming a country independent of Canada." Asked in national survey. True AB credibility interval: ±8.5pp. |
47% | Reported as "almost half of Albertans" — subsample size not disclosed. Same firm, 7 days earlier, n=1,000: 29%. |
| Pollara | Mar 16–25, 2026 n=2,576 (remain only) |
Non-separatists only | Projection "Even if I don't want to separate, I'd consider voting to separate to send a message to Ottawa." Agree/disagree. Projected to full sample assuming 100% follow-through. |
42% | Not a poll result. Modeled figure: 27% decided + 15% protest-vote agreement. Reported in key findings alongside 27% without methodological distinction. |
Variance
The instrument is in the record.
——
All data from primary polling PDFs and firm press releases.
Léger AB subsample: Special Report PDF, May 20, 2025 (n=133 unweighted).
Léger Jan 2026: Government of Alberta Report Card, fielded Jan 23–26, 2026.
Léger Apr 2026: Government of Alberta Report Card, fielded Apr 3–6, 2026.
Mainstreet: telephone IVR, Feb 10–12, 2026, ±2.5pp.
Ipsos: Confederation Stress Test, Jan 9–14, 2026.
——
thedelusioneers.substack.com · @the_delusioneers
No returns. No refunds. No records.
Primary sources: ThinkHQ Alberta Separation Survey, Sept 2025 · Léger/Postmedia Alberta Separatism PDF May 14 2025 (leger360.com) · Léger Special Report May 20 2025 — Trust in Government and Provincial Sovereignty (leger360.com) · Research Co. Alberta Separation Jan 8 2026 (researchco.ca) · Ipsos Confederation Stress Test Media Release Jan 25 2026 (ipsos.com) · Angus Reid Alberta Independence Feb 9 2026 (angusreid.org) · Mainstreet Research Alberta Feb 10–12 2026 (mainstreetresearch.ca) · Abacus Data Alberta Sovereignty Feb 2026 (abacusdata.ca) · Pollara Alberta Spotlight Apr 2026 (pollara.com) · AFP Fact Check "Invented graphic flips poll results on Alberta independence" Feb 2026