The previous three pages covered what official Canadian energy data contains. This one covers how it goes wrong in use — which is a different problem, and a more common one.
Worth stating at the outset: almost none of this involves fabrication. The overwhelming majority of misleading energy claims in circulation are built from accurate official figures, assembled in a way that supports a conclusion the figures do not support. That is harder to detect than invention, because every individual component checks out.
Failure one: the unstated boundary
Covered at length elsewhere in this atlas because it is the largest single source of apparent contradiction. A figure for emissions, cost or employment is meaningless without knowing what is inside the box.
The tell: a claim that sounds startling given what you already know. If Canadian oil appears either far cleaner or far dirtier than seems plausible, someone has changed the boundary, usually without noticing.
Failure two: the convenient base year
Percentage changes require a base year, and the choice is often made to flatter. A sector whose emissions peaked in 2005 will show impressive reductions from 2005 and modest ones from 2015. Both are accurate. Only one is informative about recent performance.
The tell: a percentage reduction with no year attached, or a base year that differs from the one used in the surrounding discussion.
Failure three: units doing hidden work
Three unit confusions recur often enough to name.
Capacity versus generation. Megawatts are capacity; megawatt-hours are energy delivered. A gigawatt of solar and a gigawatt of gas do not produce comparable annual output. Reporting capacity additions as if they were output is the most frequent error in electricity coverage.
Intensity versus total. Emissions per barrel and total emissions move independently. Both can improve; one can improve while the other worsens. Which is quoted tells you the author's purpose.
Gross versus net. In land use and in offsets particularly, gross and net figures can differ by a large margin and are rarely labelled clearly.
Failure four: mismatched vintages
Because inventories recalculate their full history, a figure from an older edition does not match the same year in the current one. Comparing across vintages produces an apparent change that is entirely an artefact of restatement.
The tell: two figures cited with different publication dates and no note that they come from the same release.
Failure five: chosen endpoints
Energy series are volatile. Production responds to commodity cycles, electricity emissions to weather and hydro conditions, and everything to economic shocks. Choose a start year at a trough and an end year at a peak and you can demonstrate almost any trend you like from unimpeachable data.
The 2020 pandemic year is the current favourite, in both directions. Comparisons starting or ending there should be treated as suspect until the endpoints are justified.
The defence, in four habits
Get the whole series. Not the two numbers someone has selected for you. Looking at a full series usually makes cherry-picked endpoints obvious within seconds, and it is a free download.
State the reference period, not the publication date. "2024 data published in 2026" is precise. "2026 figures" is usually wrong.
Reproduce one number yourself. Occasionally take a figure from an analysis and find it in the primary source. This is the single best calibration exercise available, and it usually takes a few minutes. Some organisations survive it consistently; some do not.
Note what the data cannot answer. Official statistics tell you what happened. They cannot tell you what caused it, what would have happened otherwise, or what will happen next. Causal and counterfactual claims are always analysis layered on top of data, and the analysis is where the disagreement actually lives.
The hardest case: correct and misleading
The failure modes above all have a tell. There is one pattern that does not, and it is worth naming because it is the most effective.
Take a true statement, from official data, correctly bounded, correctly dated — and present it without the context that makes it interpretable. For example: a province's electricity emissions fell sharply in a given year. True, verifiable, properly sourced. Unstated: it was a strong hydro year and the following year they rose again.
Nothing in that claim fails any of the five checks. The only defence is the habit of asking what a figure should be compared against, and whether the comparison offered is the natural one. If a single year is quoted where a multi-year average would be the obvious choice, that is usually deliberate — and it is the one case where asking who benefits from the framing does real work.
A closing note on good faith
It would be easy to read this page as a guide to catching people out. That is not the intent, and the assumption behind it would be wrong most of the time.
Boundary errors, base-year slips and unit confusion are overwhelmingly honest. Energy statistics are genuinely complicated, the conventions are counter-intuitive, and skilled analysts get them wrong routinely. The point of learning the failure modes is not to accuse anyone. It is to read more accurately yourself — and to notice when your own preferred source has made one, which is the harder and more valuable half of the exercise.
Sources
- Environment and Climate Change Canada, National Inventory Report recalculation notes.
- Canadian Centre for Energy Information, series documentation and units.
- Statistics Canada, revision policy for energy statistics.
Organisational descriptions reflect each body’s own published material and independent reporting as of September 2026. Funding arrangements and mandates change; check the primary source before relying on anything here.