Collections
Topics
Six curated collections. Each gathers the indicators, reports and graphics that belong to one subject, opens with a plain-language guide to how the measures relate to one another, and states the traps that catch people reading this material for the first time.
Environment
Air, canopy, waste, water and household energy: conditions measured at fixed points, and behaviour measured at the household.
Environmental measures divide neatly into two groups that are often mixed together. The first group is measured by instruments at fixed locations — air quality stations, water sampling points, weather stations. These readings are precise about the place they were taken and say nothing reliable about conditions a few kilometres away. The second group is assembled from records of what people and organisations did: tonnes of waste collected, homes retrofitted, litres billed. These cover everyone in a jurisdiction but depend entirely on how carefully the records were kept.
Reading the two groups the same way produces most of the confusion in environmental reporting. A city can have excellent air quality at its monitoring station and a serious local problem near a highway. A city can raise its diversion rate by changing what counts as diverted rather than by changing what residents do. Neither is dishonest; both are easy to misread.
What is in this collection
- Seven indicators: low-risk air quality days, fine particulate concentration, tree canopy cover, residential waste diversion, organics capture, residential water use, and programme-supported retrofits.
- Two reports: urban tree canopy measurement, and household energy retrofits with measured outcomes.
- Four graphics: the annual air quality dot matrix, waste destination stacked series, the provincial diversion tile map, and a retrofit before-and-after comparison.
Common traps
- Canopy percentages are not comparable between cities unless the imagery season, minimum patch size and boundary all match. They usually do not.
- A diversion rate is a ratio. It rises when diverted tonnage grows or when total waste falls, and those two situations mean different things.
- Counting completed retrofits is not the same as measuring energy saved. Both appear here, labelled separately.
Index in the low band.
Leaf-on imagery.
Kept from landfill.
Programme supported.
The headline diversion rate combines these. Splitting them shows that most of the recent movement comes from organics programmes rather than from recycling.
Transportation
Service supplied, trips taken, networks built and journeys made.
Transportation data has an unusual property: the supply side is measured very well and the demand side is measured badly. Agencies know precisely how many vehicle hours they scheduled, on which routes, at what times. What they know about trips depends on their fare technology, their transfer rules and their sampling method — and those differ so much between agencies that a national ridership total is always an assembly of incompatible parts.
The same asymmetry applies to cycling and walking. Municipalities can state their protected lane kilometres exactly. Trip counts come from a handful of automatic counters extrapolated across a whole city, and the extrapolation assumptions are rarely published.
What is in this collection
- Six indicators: transit boardings, scheduled service hours, trips per resident, protected cycling lane kilometres, average commute duration, and commuting mode share.
- Two reports: transit ridership since 2019, and cycling network connectivity.
- Five graphics: ridership against service, the mode share slope, the network facility breakdown, a peak versus off-peak service comparison, and the trip-counting explainer.
Common traps
- Ridership expressed as a percentage of 2019 assumes 2019 was a normal year for that system. For several systems it was a record year, and for a few it was a strike year.
- Kilometres of infrastructure says nothing about whether the pieces connect.
- Average commute duration is calculated only for people who commute, so it falls when people stop commuting rather than when journeys get faster.
Linked trips.
Scheduled.
Separated facilities.
One way, all modes.
Peak service has been fully restored while peak ridership has not. Off-peak ridership has recovered further than the service supporting it. Reading only the total conceals both facts.
Education
Enrolment, facilities, pathways after school, and learning that happens outside institutions.
Enrolment is one of the most reliably counted figures in public administration, because funding depends on it. That reliability makes it tempting to use enrolment as a proxy for things it does not measure: neighbourhood age structure, family formation, or the effect of new housing. It is a poor proxy for all three, mainly because catchment boundaries change, school choice policies vary, and a meaningful share of students attend a school outside the catchment they live in.
Facility capacity is measured less consistently. A school’s stated capacity depends on how portable classrooms are treated, whether specialist rooms count, and what class-size assumption the board applies. Two boards can describe identical buildings with capacity figures that differ by fifteen per cent.
What is in this collection
- Five indicators: enrolment by level, utilisation against stated capacity, students per educator, post-secondary participation, and adult learning participation.
- One report: school enrolment and residential construction.
- Three graphics: enrolment by level over time, a utilisation distribution, and the housing type to school-age children comparison.
Common traps
- A board with falling enrolment and overcrowded schools is a normal situation, not a contradiction. Enrolment falls in some catchments and rises in others.
- Students per educator is not class size. It includes staff who do not lead a classroom.
All publicly funded levels.
Stated capacity basis.
All staff with teaching duties.
Any formal course in the year.
Growth at the secondary level reflects a cohort passing through rather than a change in participation. Cohort effects move slowly and predictably, which is why enrolment is one of the few series where the near future is genuinely legible.
Housing
Supply, tenure, affordability measures and the age of the building stock.
Housing generates more contested statistics than any other subject on this platform, and most of the disputes are definitional rather than empirical. A “housing unit” may or may not include a secondary suite. A “start” may be a foundation poured or a permit issued. An “affordable” unit may be priced against local median income, against a programme threshold, or simply cheaper than the units around it.
We publish the definitional detail alongside every figure because in this subject the definition is usually the disagreement. Where two credible sources produce different numbers for the same thing, both appear with an explanation of what each counts.
What is in this collection
- Eight indicators: starts, completions, units under construction, average build duration, rental vacancy, median rent-to-income, tenure mix, and period of construction of the occupied stock.
- Two reports: completions in mid-sized cities, and newcomer settlement patterns.
- Four graphics: the formation-and-completion series, the tenure donut, dwelling type by settlement size, and completions per thousand residents by province.
Common traps
- Starts and completions describe different moments in one pipeline. When build times lengthen, the two series can diverge for years without either being wrong.
- The vacancy rate covers purpose-built rental only. In several cities the excluded stock is larger than the surveyed stock.
- Median rent-to-income describes the median renter household. It says nothing about the households furthest from the median, which is usually the group in question.
Reported finished.
Foundations begun.
Purpose-built rental only.
Median renter household.
The lengthening of apartment build times is the mechanical reason starts and completions have moved apart. A start in 2023 now reaches completion, on average, in the second half of 2025.
Technology
Connectivity offered, connectivity delivered, and how people reach public services.
Connectivity statistics answer three different questions that are frequently confused. Availability asks whether a provider offers qualifying service in an area. Adoption asks whether a household has subscribed. Performance asks what speed the household actually receives. A jurisdiction can score well on availability and badly on the other two, and price is the usual reason.
Digital public service statistics carry a subtler problem. Use of an online service is easy to count; whether the interaction succeeded is not. A person who starts an online application, fails, and finishes by telephone appears in both channels and is counted as a success in the first.
What is in this collection
- Five indicators: availability at 50/10 Mbps, rural availability, median measured download speed, digital public service use, and channel preference.
- One report: rural broadband and what “available” counts.
- Three graphics: the availability series by area type, the provincial tile map, and digital service use by age.
Common traps
- Availability is measured for an area, not a household. In large rural service areas this gap is substantial and is quantified in the report.
- Median measured speed is drawn from voluntary tests, which over-represent households that suspect a problem.
Service offered in area.
Rural areas only.
Measured at household.
Last twelve months.
The distance between the two columns is the share of households that could buy qualifying service and have not. It widens as settlements get smaller, which points at price and at the quality of what is on offer rather than at coverage.
Communities
Libraries, public space, volunteering, civic participation and local services.
This is the collection where measurement is weakest and the subject matters most to people. Almost nothing here is counted by an agency with a statutory obligation to count it. Library returns are voluntary in several provinces. Volunteering is measured by survey. Civic participation is measured by whatever a given municipality decided to record.
We publish these measures with unusually heavy caveats, and we publish the response rate alongside the result wherever the source gives us one. A figure built on a 41 per cent response rate is still useful. It is not useful in the same way as one built on an administrative record, and the difference should be visible on the page rather than buried in a footnote.
What is in this collection
- Five indicators: library visits per resident, programme attendance, regulated childcare spaces, median service wait, and staffed service points.
- Two reports: library and community space use, and newcomer settlement patterns.
- Three graphics: library use in person and online, childcare spaces by settlement size, and the service point distribution.
Common traps
- Digital library sessions and in-person visits measure different behaviours and must not be added.
- Childcare spaces are counted where licensed, not where children live.
- A median wait time hides the tail, and the tail is what people complain about.
In person.
Sessions attended.
Ages 0 to 5.
Routine appointment.
Coverage for administrative sources, response rate for surveys. We publish this chart because the confidence you should place in a community indicator depends more on this figure than on anything else about it.