Nine questions, answered with public data
Each study started as a question about housing or city services and became a full analysis: data anyone can download, scripts anyone can rerun, limits stated up front. Each study lives on its own page and comes in two parts: a plain-language story for readers who never took a statistics class, then technical notes with the full specification, estimate tables, and robustness checks for those who want them.
Is high rent bad for your health?
2.3× the oddsOne in four renters rates their own mental health as fair or poor. Two survey cycles, 37,000 renter households in all, show the trouble tracks missed payments more than the size of the bill, and that the link predates the pandemic.
Logistic regression · survey-weighted Data: survey cycles 2018 + 2022 Read the study → 02 · Homelessness servicesDid the 2023 refugee crisis change Toronto's shelter numbers?
A lasting +21% shiftThe numbers had been climbing 6% a month long before the crisis made news. Fifty-four months of the city's own counts show what changed, when, and for how long.
Interrupted time series · comparison group Data: monthly through June 2026 · pipeline auto-refreshes Read the study → 03 · Housing qualityAre apartments worse in poorer neighbourhoods?
84 in 100 buildingshave no air conditioning, in rich and poor neighbourhoods alike. Meanwhile the income gradient everyone expects in audit scores turns out to be nearly flat.
Descriptive + OLS regression Data: evaluations to July 2026 · history 2017–2023 Read the study → 04 · Short-term rentalsWhat does Airbnb's 28-day loophole rent for?
58.7% past the lineMost entire-home listings now sit outside Toronto's short-term rental rules, priced about half of short stays. A one-bedroom still works out to $3,671 a month.
Hedonic regression · fixed effects Data: snapshot June 15, 2026 · CMHC October 2025 Read the study → 05 · Urban planningHow long do Toronto building applications wait?
Median age: 4.1 yearsToronto's queue holds 5,065 open applications, and half of everything submitted since 2022 is still open. The slowest files are stuck at the appeals stage, which points at a different fix.
Descriptive survival + logistic regression Data: snapshot July 29, 2026 Read the study → 06 · Public healthWhat did the pandemic do to opioid deaths?
+85%, now recedingDeaths jumped 85% when the pandemic hit and peaked in late 2021. Ten years of federal data show a wave now receding, at very different speeds across provinces.
Interrupted time series · single series Data: deaths reported to May 12, 2026 Read the study → Across all sixWhat the data says about Toronto's housing squeeze
Five levers, five ownersThe studies read as one system: where homes leak out of the market, where supply waits, where the harm lands, and which government holds each lever.
Synthesis · plain-language brief Read the synthesis → 07 · Real estate & developmentWhat does Toronto build, and who ends up holding it?
58% investor-ownedA quarter-million permitted homes, a two-year permit clock for the towers that carry most of them, a generation without rental construction, and condos that grow more investor-owned with every vintage.
Descriptive pipeline · duration analysis Data: permits to July 2026 · ownership records 2022 Read the study → 08 · Housing policyDid legalizing multiplexes work?
+27% above trendToronto legalized fourplexes citywide in May 2023. Applications jumped 27% above an already-rising path, projects got bigger, and ground-up builds tripled their share.
Interrupted time series · single series Data: applications to May 2026 Read the study → 09 · Recreation servicesWho gets a spot in Toronto's recreation programs?
2.4 per 10,000 seniorsHeld against the census, the City's own weekly course feed shows North York running thinnest per resident, children's supply varying twofold by district, and older adults barely programmed for anywhere.
Descriptive · per-capita rates · data-quality audit Data: program feed snapshot August 17, 2026 · Census 2021 Read the study → The originalDoes Airbnb make rent more expensive?
My MA thesisThe research that started all of this: a full study of Airbnb and Toronto rents, told chapter by chapter, with the award-winning methodology behind it.
Instrumental variables (2SLS) Data: September 2024 snapshot Read the research →What the nine studies have in common
Data to model to chart
Each study is three scripts: fetch the data, run the analysis, draw the charts. Every number on these pages comes straight from the saved model output, never retyped by hand. How I choose which method →
Bugs found and kept
A hidden-whitespace bug that flipped a key result. A straight-line model that missed the real curve. A name mismatch that silently dropped 239 buildings. Each study's page tells the mistake, the catch, and the fix. The pre-flight checklist they add up to →
Honest by design
Every page has a "what this does not prove" section, an academic-style reference list for its data and methods, and charts drawn to keep small differences looking small. Trust in an analyst is built there, in the fine print.
Readable by every reader
Every chart carries a full written description for screen readers, the interactive pieces work by keyboard, animations respect the reduced-motion setting, the pages still work with JavaScript switched off, and each study prints to a clean brief. Public-sector standards, applied to my own site first.