Findings

Five years of Berkeley Police Department data, analyzed across five independent methods, converge on a consistent finding: motel conversions to interim housing on University Avenue and San Pablo Avenue were followed by measurable increases in police activity near the sites — increases that substantially exceeded what comparison corridors experienced over the same period. The effect is both volumetric and compositional. It is corroborated by officer-initiated stop data independent of citizen calls.

That finding has been available in the data since at least 2023. It has not previously been measured because the institutions positioned to measure it had no interest in doing so, and because journalism — however accurate in its event-level coverage — is not a tool for producing multi-site, longitudinal, comparison-controlled analysis. Berkeleyside and the Berkeley Scanner have reported these conversions and these corridors with skill and care; their work is part of the record this analysis draws on. The policy history is documented there in depth. What local reporting cannot produce is a pre/post study across five sites with indexed comparison corridors and five years of call data. This analysis provides that.

Scope. This study covers two treatment geographies: the University Avenue core cluster (1461, 1619, and 1761 University Ave) and the San Pablo node (1620 San Pablo Ave, Golden Bear Inn). North Shattuck — adjacent to the Hope Center at 2012 Berkeley Way — appears as corroborating evidence but is not analyzed at the same depth. Comparable analysis could be conducted on the Grayson Street facility, the Rodeway Inn corridor, or other interim housing sites citywide; those sites would likely show variation in degree, not kind. The fiscal, public health, and revenue costs documented in the context page are outside the scope of the police data analyzed here.

Two distinct treatment geographies: the University Avenue core cluster (1461, 1619, and 1761 University) and the San Pablo node (1620 San Pablo Ave, Golden Bear Inn). Each is analyzed independently. All results use non-traffic calls for service from the Berkeley Police Department's public dataset.

1. Concentration of Calls Near Study Sites

Non-traffic police calls are concentrated within 100 m of the studied properties relative to the surrounding study area.

Share of non-traffic calls within the combined study area (university cluster + San Pablo node) that fall within each site's 100 m zone, split by pre- and post-opening periods. Download data

2. Site-by-Site Pre/Post Changes

Monthly non-traffic calls in the 600 m zone around each site, 12 months before and after opening. Sites with insufficient post-period data are flagged.

1761 University has less than 6 months of post-opening data and is not shown as a definitive result. Download pre/post data

3. University Core Cluster Trend

Monthly non-traffic calls within the University Avenue core cluster, with each University site opening marked.

Bold line = 3-month rolling average. Dashed vertical lines = site opening dates (University Ave sites only).

4. San Pablo Node Trend

Monthly non-traffic calls within the San Pablo node (250 m buffer around 1620 San Pablo Ave).

Dashed vertical line = Golden Bear Inn opening date (January 2023).

5. Comparison Corridors

University core cluster and San Pablo node indexed to 100 during the pre-period baseline, alongside two comparison corridors.

North Shattuck and South Telegraph are comparison corridors with similar street character and no interim housing conversions in the study period. Divergence from comparison corridors is consistent with a local effect but does not prove causation.

6. Adjacent-Block Spillover

Zone-level comparison for each site: immediate 100 m zone vs. 100–300 m adjacent ring vs. 300–600 m wider area.

7. Berkeley Corridor Rankings

Non-traffic call trajectory for all named Berkeley primary and secondary corridors, 2021–2025 (index: 2021 monthly average = 100). Shelter corridors highlighted. Shattuck Avenue is excluded due to heterogeneous character zones unrelated to shelter siting.

Index above 100 = more calls in 2025 than 2021. Corridors with fewer than 30 annual calls in 2021 excluded as too quiet for stable rates.

8. Call Type Shift: What Changed, Not Just How Much

Pre- and post-opening rates per month for the most-changed call categories at each site (100m zone). Shows whether the neighborhood impact is volumetric or compositional.

Rates are calls per month within 100m of each site. Pre = 12 months before opening; post = all available months after opening. 1761 University Ave post-period is less than 3 months and shown for reference only.

9. The North Shattuck Cascade

Monthly non-traffic calls on North Shattuck (north of University Ave, 75m corridor buffer). The Hope Center at 2012 Berkeley Way — one block from Shattuck — opened September 2022.

Bold line = 3-month rolling average. The step-change in May–June 2022 precedes the official opening date, consistent with the facility becoming a destination during pre-opening construction.

10. Police Stops Corroborate Call Increases

Pedestrian stop rates (stops per month within 100m) before and after each site opened, from RIPA stop data. Stops are officer-initiated actions, independent of citizen calls.

Pre-period for 1461 University Ave covers October 2020–June 2021 (COVID period; low baseline should be interpreted cautiously). 1761 University Ave has near-zero post data and is omitted.

11. Citywide Context: Where These Sites Rank

To assess whether the study sites are outliers or simply reflect the character of active Berkeley corridors, we compared each site's 100 m zone against every 100 m grid point on seven named comparison corridors: South Telegraph, Shattuck south of University, Adeline, Ashby Avenue, College Avenue (Elmwood), Fourth Street (retail district), and Solano Avenue. The question: among all individual 100 m zones on those corridors, where do the study sites fall?

Calls. Against the five corridors with no major service concentrations — Ashby, Adeline, Elmwood, Fourth Street, and Solano — 1461 University and 1761 University are at the 92nd–100th percentile of every 100 m zone on every corridor. Golden Bear Inn is at the 75th–100th percentile. Even 1619 University, the quietest study site, is at the 64th–89th percentile across the same comparisons.

Stops. The officer-initiated stop data is sharper. Against those same five corridors, 1461 University, 1761 University, and the Golden Bear Inn are at or tied for the maximum stop count of any 100 m zone on every corridor. 1619 University is at the 88th–94th percentile. These are not outlier readings on noisy data — the study sites are consistently at the ceiling of what comparable Berkeley corridors produce.

Two corridors — South Telegraph and Shattuck south of University — have individual 100 m zones with higher call and stop counts than some study sites. Both have specific explanations. The peak Telegraph zones are in the former People's Park area, a decades-old street community that constituted its own de facto service concentration; that site has since been closed and the population has likely dispersed, partly into the downtown core. The peak Shattuck zone is the Downtown BART plaza — an area Berkeley residents know independently as a high-intensity distress concentration, and one that sits a short distance from the Hope Center on Berkeley Way. Neither location is a clean counterexample; both are instances of the same pattern the study sites represent.

The pre/post step-up distinguishes these sites from comparably active locations. High call and stop counts at downtown BART or the former People's Park predate the shelter conversions studied here. The study sites on University and San Pablo Avenues show a documented increase after opening — measured against their own pre-opening baselines, against indexed comparison corridors, and corroborated by independent stop data. The sites are not simply located in active areas. They became active in response to a specific event.

12. Interpretation Notes

• Concentration: 100 m site zones contain a disproportionate share of study-area calls, especially at 1461 University and 1761 University.
• Mixed pre/post results: 1461 University showed a substantial increase after opening; 1619 University and 1620 San Pablo showed modest declines; 1761 University has insufficient post data.
• Cluster vs. node: The University cluster and the San Pablo site show different patterns — grouping them into one corridor average would obscure that difference.
• Association, not causation: Changes in call volume after openings do not prove the sites caused the change. Comparison with control corridors provides context for broader citywide trends.
• Data limitations: Calls data required geocoding from block addresses; approximately 15% of records could not be geocoded. Opening dates are approximate for some sites.