uniteTOlove.ca

Community Safety Perception Gap — Playbook

Crime stats are falling but fear isn't — why Torontonians feel less safe than the numbers say they should.

DRAFTThe playbookThe evidence file

What Toronto can do about a 76-point gap between what residents believe about crime and what the City's own numbers show — starting with getting its own numbers straight.

The honest bottom line

Seventy-two percent of Torontonians think homicides are rising. They're not — by most accounts, they fell sharply in 2025. But ask exactly how sharply, and you get two different answers depending which report you read. Toronto's own website, meanwhile, is still citing a safety ranking from 2017. If the city wants residents to believe the good news, it should probably start by getting its own number straight. A December 2025 poll from Liaison Strategies — a firm with real, checkable election-prediction credentials, not an anonymous online survey — found that 76% of Toronto residents believe crime in Canada rose over the past year, and 72% believe homicides in Toronto specifically are rising. The same poll found the identical pattern next door: in Peel Region, most Brampton and Mississauga residents believe homicides are increasing, even though Peel Regional Police reported just 10 homicides through October 2025 versus 19 for the whole of 2024. In York Region, homicides were reportedly down 67% — and still, over half of Vaughan and Markham residents think they're going up. Here's the frustrating part for anyone trying to check the other side of that gap: the actual 2025 numbers aren't as clean as the perception-gap headline suggests. Toronto's homicide count for the year gets reported as either 39 or 42, depending on the source. The auto-theft decline gets reported as anywhere from about 24% to 34%. And when this research went looking for Toronto's current standing on the Economist's well-known Safe Cities Index — the kind of ranking that should be able to settle "is Toronto actually safe" pretty definitively — the City's own website turned out to be citing 2017 numbers. Not 2024. Not even 2021. Nine-year-old rankings, sitting on a live City of Toronto page, being used as if current. There's a real success story buried under all this — the Toronto Community Crisis Service, a non-police response for mental-health crisis calls, diverted 78% of relevant 911 calls away from police entirely in its original, independently-evaluated 2022 pilot. That's a genuinely strong number from a genuinely credible evaluator. The problem is the program has since grown roughly tenfold in reported budget and is expanding onto subway platforms — environments the original evaluation never tested — and the same 78% figure keeps getting cited as if it still applies at that scale, without this research finding evidence anyone's gone back to check. Two things are worth doing, at minimal cost. First: build one authoritative, current public page for Toronto's actual crime trends, and retire the stale one. The City already has the underlying data — the Toronto Police Service runs its own public data portal. This costs next to nothing; it's a consolidation job, not a data-collection one. It won't single-handedly close a 76-point perception gap — nothing this cheap could — but it removes the excuse that the real numbers are hard to find. Second: go back and re-evaluate the Crisis Service at its actual current size. The original third-party evaluation was rigorous and independent — that's exactly why it's worth repeating now that the program is ten times bigger and operating somewhere new. If the 78% diversion rate still holds at scale and on transit platforms, that's a genuinely strong result worth restating with fresh evidence instead of a three-year-old pilot number. If it doesn't hold, that's exactly the kind of thing a city should want to know before scaling further, not after. This doesn't explain why the perception gap exists — media coverage, social media, individual high-profile incidents, or something else entirely. Nobody in this research pass found a study that actually answers that question for Toronto specifically, and guessing wouldn't be honest. What's proposed here is cheap, boring, and exactly the kind of unglamorous data hygiene that a perception gap this large deserves before anyone reaches for a bigger, more expensive answer.

---

a recommendation card — Standing, Public Reconciliation of Crime-Trend Data and Perception Polling

Card id: a recommendation card · Issue: community-safety-perception-gap · Backgrounder: our research file for that page · Trust: New load-bearing findings (problem framing) / NEW (evidence anchors)

Problem

A directly and independently fetched December 2025 Liaison Strategies survey found that 76% of Torontonians believe crime in Canada has increased over the past year, and 72% believe homicides specifically are rising, even as the same period's homicide and auto-theft figures moved sharply downward by multiple accounts [NEW-CG01–CG05]. This page’s own discovery could not resolve the exact 2025 homicide count (sources disagree: 39 vs. 42) or auto-theft decline percentage (sources disagree: roughly 24%–34%) [NEW-CG06, NEW-CG07] — meaning even a resident or journalist motivated to check the "actual" numbers against the perception-gap finding encounters inconsistent public figures rather than one clear, authoritative source. Separately, the City's own Safe Cities Index citation page (toronto.ca) was found, on direct fetch, to display stale 2017 data rather than any current ranking [NEW-CG08] — a small but concrete example of the City's own public-facing safety-data communication lagging behind the safety trends it could be citing. This card addresses the data-clarity gap underlying the perception gap, not the perception gap's psychological or media-driven causes, which this page’s sources do not analyze.

Action

Require the City of Toronto to publish and maintain a single, standing, plain-language "safety trends dashboard" page — sourced directly and only from the Toronto Police Service's own Public Safety Data Portal and Statistics Canada — updated at least quarterly, that states the current year-to-date homicide, shooting, and auto-theft trend in one authoritative place, replacing or supplementing the currently stale Safe Cities Index citation page this review found displaying nine-year-old data.

Jurisdiction split

Cost

Order-of-magnitude: low tens of thousands of dollars CAD for the dashboard build and ongoing maintenance, anchored to the comparator that this consolidates data the City and Toronto Police Service already collect and publish separately (the Public Safety Data Portal already exists) — this is a presentation/consolidation cost, not a new data-collection program.

Funding path

Existing City of Toronto communications/open-data operating budget; no new funding mechanism required, since the underlying data infrastructure (the Toronto Police Service's own Public Safety Data Portal) already exists.

Who benefits, and how

Toronto residents generally, via one clear, current, authoritative answer to "is crime actually up or down" rather than the inconsistent figures this page’s own discovery encountered across multiple otherwise-credible sources; journalists and researchers (including future passes on this page), via a stable citation point instead of having to reconcile conflicting secondary reporting.

Who bears the cost, and how

City taxpayers city-wide, via the existing communications/open-data budget; no disproportionate cost to any named group.

Who benefits from the status quo

No beneficiary identified — the backgrounder's Cui Bono table is empty; the backgrounder notes media-coverage economics as a plausible general structural factor in perception gaps but does not name a specific entity, so no row is populated here either.

Financial ROI

Not separately modeled — this is a transparency/communication measure. No source in this page quantifies a financial return from clearer public safety-data communication specifically.

Economic ROI

Not yet estimable. A live-discovery search this review (economic impact of perceived-safety improvements on local business/tourism activity) found general commentary but no rigorously quantified, Toronto-specific study connecting perception-gap closure specifically (as distinct from actual crime reduction, a different question) to measurable economic activity. Confidence: low — genuine gap; this card does not borrow general "fear reduces economic activity" intuitions without a specific citable source.

Social ROI

Directional: if a persistent perception gap drives avoidance of public space, transit, or the night economy (a mechanism this document does not independently verify for Toronto but which is a commonly discussed general pattern), then closing the gap with clearer public data is plausibly protective of civic/social participation — but no source in this page quantifies this effect for Toronto specifically. Confidence: low-medium — directional only.

Environmental ROI

Genuinely environmentally neutral — a data-dashboard consolidation has no emissions, land-use, water, or waste mechanism. Confidence: high.

Evidence

Confidence & uncertainties

Medium confidence on the problem statement (the data-inconsistency and stale-page findings are directly, independently confirmed this review); this card does not claim that better data communication alone would close the perception gap — general public-opinion research (cited generally, not from a specific study this review verified) suggests media coverage and salience effects likely play a role independent of data availability, a limitation this card does not overstate.

Status

DRAFT — blocked on: identifying the City's specific current data-publication workflow in more detail; fairness and legal review.

---

a recommendation card — Recurring, Independent Re-Evaluation of TCCS as It Scales

Card id: a recommendation card · Issue: community-safety-perception-gap · Backgrounder: our research file for that page · Trust: New load-bearing findings (problem framing) / NEW (evidence anchors)

Problem

The Toronto Community Crisis Service's foundational evaluation — a 78% no-police-involvement diversion rate, independently conducted by CAMH-affiliated third-party evaluators — covers a specific six-month pilot window (March–September 2022) across four pilot areas, independently and directly confirmed from the City's own primary release this review [NEW-CG10]. Search-surfaced (not independently fetched) figures suggest TCCS has since grown substantially — a reported $31.7 million 2025 budget rising to a proposed $35.3 million in 2026, over 200 staff, and planned TTC subway-platform expansion — while continuing to cite the same 78% figure [NEW-CG11]. This card addresses the risk that a headline metric from a small, specifically-evaluated pilot is being carried forward as if automatically still accurate at roughly 10x the original program's scale and into new operating environments (e.g., transit platforms) the original evaluation did not cover, not a claim that the figure is wrong.

Action

Require a fresh, independently-conducted (third-party, CAMH-affiliated-or-equivalent) outcome evaluation of TCCS at its current scale and current service areas — including any new transit-platform deployment specifically, since that is an operating context the original 2022 pilot evaluation did not cover — published on a recurring basis (at minimum, every time the program undergoes a major scale-up) rather than relying on the original pilot's figures indefinitely.

Jurisdiction split

Cost

Order-of-magnitude: low hundreds of thousands of dollars CAD for a rigorous, current-scale re-evaluation, anchored to the comparator that the original six-month pilot evaluation was itself commissioned and delivered by an established third-party evaluator (Provincial System Support Program/Shkaabe Makwa at CAMH) at a much smaller program scale — a comparable or somewhat larger evaluation budget is a reasonable estimate for a ~10x-larger current program, though no specific cost figure for the original evaluation was located to anchor a precise multiple.

Funding path

Existing TCCS program budget (reported at $31.7–35.3 million, “still being checked”) could reasonably absorb an evaluation cost that is a small fraction of that total; alternatively, the City's existing Social Development evaluation infrastructure (per the gun-gang-youth-violence-prevention backgrounder's own documented $1.176M "Advanced Statistical Analytics" and $1.946M "Program Staffing, Evaluation and Support" CPAF lines) could potentially be leveraged, though TCCS itself is a different program stream than that page’s CPAF/BSCF-funded portfolio and this card does not assume automatic budget transferability between them.

Who benefits, and how

Toronto residents and Council, via confidence that TCCS's headline effectiveness metric reflects its current scale and operating contexts rather than a legacy pilot figure; TCCS itself and its delivery partners, via updated evidence that could strengthen continued funding cases if the program continues performing well at scale.

Who bears the cost, and how

City taxpayers city-wide (or the TCCS program budget specifically), via existing or modestly expanded program/evaluation budget lines; no disproportionate cost to any named group.

Who benefits from the status quo

No beneficiary identified — the backgrounder's Cui Bono table is empty.

Financial ROI

Not separately modeled — this is an evaluation/governance measure, not a cost-avoidance one directly.

Economic ROI

Not yet estimable. No third-party study quantifies the economic value of re-evaluating a scaled crisis-response program specifically. Confidence: low — genuine gap.

Social ROI

Directional: if TCCS's diversion rate has held up at scale, confirming this strengthens the case for a program this page’s sources show already reduces police involvement in mental-health calls by a substantial majority — a plausible social benefit (reduced over-policing exposure for people in crisis) documented directionally in the original 2023 evaluation's own quoted partner statements (e.g., CMHA Toronto's quoted statement that the model "contributes to reducing the over policing of racialized and other equity seeking communities"), though this document treats that as the partner organization's own stated view, not an independently modeled outcome. Confidence: low-medium — directional, resting on a 2022-era evaluation whose current-scale validity is exactly what this card proposes to re-test.

Environmental ROI

Genuinely environmentally neutral — an evaluation-commissioning action has no emissions, land-use, water, or waste mechanism. Confidence: high.

Evidence

Confidence & uncertainties

Medium confidence on the problem statement (the multi-year gap between the original evaluation and the program's current scale is directly evidenced by comparing a confirmed 2022 pilot-scope evaluation against unverified but consistently-reported much-larger current figures); this card does not claim TCCS is failing at scale — only that the evidence to confirm or disconfirm that has not been independently re-established in this page’s sources since the original pilot.

Status

DRAFT — blocked on: confirming whether a more recent independent (non-original-pilot) TCCS evaluation already exists and was simply not located in this review's discovery; fairness and legal review.

---

Production record

Drafting record

Status: DRAFT · v1.0 · Date: 2026-07-14 · Provenance: mixed — problem statements trace to NEW discovery (independently fetched polling data and City evaluation reports); the recommendations themselves are this document's own L-PLATFORM position, per the L6 template's firewall discipline.

Playbook conversion (2026-08-11, Lane L3a): opened with "The honest bottom line" adapted from archive/dayone/community-safety-perception-gap.md (a recorded standing decision retired day-one memo, kept as history in archive/); ROI sections tightened, repeated "not yet estimable / not applicable" boilerplate collapsed to one honest line each, matching that page's recommendation cards's playbook shape. No a formally registered claim tokens present in this file; all NEW citations preserved unchanged. Note: this file was outside Lane L3a's originally-scanned batch order (its slug begins "community-" and was initially grouped with the c-range sweep that missed it); caught and converted in a final owned-set audit pass.

About this file