Methodology

Purpose

This project examines whether four converted motel housing sites on University Avenue in Berkeley were followed by changes in nearby non-traffic police calls for service.

The goal is not to debate homelessness policy. The goal is to measure whether nearby public-safety demand changed after specific sites opened.

Sites Studied

Why Non-Traffic Calls

Traffic stops and collisions happen for many reasons unrelated to nearby housing.

Non-traffic calls are a better signal for neighborhood public-safety impact. These include:

Before / After Windows

For each site, we identified an approximate opening date.

We then compared:

Using the same location before and after helps control for pre-existing neighborhood characteristics — same street, same businesses, same transit access.

Geographic Zones

To measure whether effects spread beyond the site itself, we divided the analysis into three concentric zones:

Zone Description Approximate radius
Zone 1 Site block 100 m
Zone 2 Adjacent blocks 300 m
Zone 3 Wider nearby area 600 m

If only Zone 1 shows increases, impacts may be contained on-site. If Zone 2 also increases, nearby residents and businesses may be affected.

Control Corridors

We applied the same analysis to two busy Berkeley commercial corridors:

These serve as baselines for identifying citywide trends that would affect results regardless of shelter activity. Both have comparable pedestrian traffic, retail density, and transit access to the study corridors — they approximate what University Avenue and San Pablo Avenue might look like without shelter facilities.

Why not downtown / South Shattuck?

The blocks of Shattuck Avenue from University south to Channing are where many Berkeley residents most visibly encounter the effects of the city's homelessness infrastructure: daytime loitering, public intoxication, and the informal behavioral economy that clusters around service concentrations. The Men's Shelter is on Center Street, one block off Shattuck; the Hope Center is on Berkeley Way, also a block away. The effects radiate outward. An entire block face of Shattuck between Allston and the next block south sits vacant — a commercial strip that businesses have declined to occupy, which is its own indicator of how the market reads the corridor, and its own category of cost in foregone sales tax and property tax revenue. Real estate dynamics compound the picture. Two major developments have cleared existing retail from the corridor but stalled before delivering replacement activity. At 2190 Shattuck — the "Allston Pit," formerly a Walgreens — a fenced excavation has sat open while the developer repeatedly upsizes the project, now proposing 34 stories, waiting for interest rates and construction financing to align. At 2128 Oxford/Center Street, a block of long-standing retail including PIQ was cleared for a 26-story tower that has spent years in Landmarks Preservation Commission appeals over historic facade demolition; final permits were only recently reviewed in April 2026, with ground-floor retail still unbuilt. Both projects follow the same logic: construction costs and interest rates mean they only pencil out at extreme heights, so developers hold the land through repeated redesigns rather than build. The causal relationship runs in both directions — a more robust retail environment might have strengthened the economic case for completing these projects sooner, while the prolonged vacancies have further depressed the corridor. What is clear is that the net effect has been years of displaced tenants with no replacement, on a street already reading as inhospitable to retail investment.

This corridor is heavily used during the day by people connected to multiple shelter and service facilities across the area. Using it as a control would mean measuring a corridor already substantially affected by the same infrastructure we are studying. It appears in the analysis as corroborating evidence (see the North Shattuck Cascade, Finding 9), not as an independent baseline.

Why not Ashby BART or Adeline?

The Ashby BART area and the Adeline Street corridor have high pedestrian and transit volumes, but retail density is lower than University or San Pablo Avenues. One of our selection criteria for comparison corridors was that they share the commercial character that makes a corridor sensitive to changes in street activity. A transit-heavy corridor with limited retail doesn't serve as a meaningful comparison for a commercial avenue.

Why not Solano Avenue or Elmwood?

Solano and Elmwood are mixed retail corridors with comparable pedestrian activity and business density, but neither has a BART station or equivalent transit node. University and San Pablo both sit on high-frequency transit lines, which affects the population mix on the street and the nature of calls generated. Matching on retail without matching on transit would introduce a confound in the opposite direction.

That said, Solano, Elmwood, and Fourth Street — all retail districts with no shelter facilities and varying transit access — offer a useful potential baseline for what a commercial corridor without shelter pressure looks like over the same period. That comparison is not included in this analysis but would be a meaningful extension.

Rolling 3-Month Year-over-Year Comparisons

Some call types vary by season. To reduce seasonal bias, we also computed rolling 3-month windows compared year-over-year.

Example: February–April 2025 compared with February–April 2024.

This separates site-related patterns from annual seasonal cycles.

Measuring Change

We measured percentage change:

Percentage change = (post-period count − pre-period count) / pre-period count × 100

A result of +50% means 50% more calls in the 12 months after opening compared with the 12 months before.

Data Sources

See the Data Sources page for portal links and fetch dates.

Limits

This analysis does not prove that any individual call was caused by a housing site.

Other factors that can affect call counts:

Results show association and timing. They are not proof of causation.

The BPD call data is a floor, not a ceiling. The Specialized Care Unit (SCU), which operated from 2023 until it was handed to Alameda County in 2025, handled some incidents that would otherwise have generated police calls for service — welfare checks, mental health responses, and disturbance calls that were triaged to the co-response model instead. Those SCU-handled incidents do not appear in the BPD dataset analyzed here. To the extent SCU diverted calls near the study sites, the measured increases in police activity understate the true volume of public safety demand generated by these corridors. The data is conservative in the direction that matters for the core finding.

Why This Still Matters

When multiple sites show similar patterns — especially increases in adjacent blocks after opening — that consistency is useful evidence for community discussion, even without proof of causation.

The goal is to move from anecdote to measurable, reproducible findings.