Science1 publisher3 min readPublished
University of Oregon researchers estimate CAHOOTS took 18% of calls bound for Eugene police
University of Oregon researchers estimate Eugene's CAHOOTS teams directly took 18% of the calls that would otherwise have gone to police. Their three diversion measures give cities one shared definition, though each measure divides by a different pool of calls.
The Scientist · Science desk
Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction
What happened
- Counting calls that CAHOOTS may have kept out of the emergency system altogether raises the estimated diversion to 23%.
- Among calls that fell within both police and CAHOOTS responsibilities, the crisis teams handled nearly half.
- The Science Advances paper by Rori Rohlfs, Claire Herbert and Nathan Burton reports an impact substantially greater than some earlier estimates had suggested.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- decision A council weighing a crisis team's budget has to pick a measure first, because the same Eugene program scores 18% or nearly 50% depending on the question asked.
- constraint With no benchmark attached, the rates place a program within its own system but cannot by themselves tell a city whether 18% justifies the spending.
- exposure The 5-point prevention component rests on calls that never happened, so it is the part of any funding case built on these rates that is most open to challenge.
- capability Cities can now run one set of definitions on their own dispatch data, so a claimed diversion rate can be checked against a shared standard.
Each of the three figures divides by a different pool of calls. The Substitution Diversion Rate counts CAHOOTS responses that directly replaced a police response [3]. The Prevention-adjusted Diversion Rate adds the drop in police responses that the authors attribute to early intervention by the crisis team [4]. The Overlapping-mandate Diversion Rate is limited to calls that either police or CAHOOTS could have handled [5]. The near-50% figure is a share of that overlapping set [8]. A reader who takes it as half of Eugene's police workload has misread it.
The design addresses a real flaw in how these programs have been judged. Police and crisis teams do not take the same kinds of calls, and programs differ in staffing, duties and the calls they are equipped for, so comparing raw call counts misleads [10]. "Until now, there has been a lot of talk about programs' 'diversion rates' without a clear, consistent definition," said Rori Rohlfs, an associate professor of data science at the University of Oregon [2][11].
The gap between the first two rates is 5 percentage points [1]. That slice is the prevention effect: calls that, on the authors' estimate, never entered the emergency system because CAHOOTS was available [7]. Counting events that did not happen requires a counterfactual. It is the part of the result a skeptic will push on first. The phys.org summary of the paper does not give the study period, the call volumes, the method behind the prevention estimate, or which earlier estimates the authors say the program exceeded [9].
The researchers set no target for what a successful program looks like [12]. "Mobile crisis programs across the country operate very differently, including who responds and what kinds of calls they are equipped to handle," said Claire Herbert, an associate professor of sociology and a co-author [2][13].
The data come from the oldest program of its kind in the United States. CAHOOTS has sent two-person teams, a crisis worker paired with a medic, to Eugene calls since 1989 [14]. The city stopped funding it in 2025, and it still runs in neighboring Springfield [15]. Herbert said the measures would have been valuable for Eugene to have years earlier [16]. She is a former CAHOOTS worker who, in her words, "experienced the loss of the service in a very visceral way" [17].
I think the standardisation is the most useful part of the work. Communities have had no consistent way to measure diversion [18], and a city that applies these definitions to its own dispatch data gets a figure it can defend in a budget hearing [1]. All three rates count police demand moved or avoided. A budget also turns on what each response costs and how callers fared, and those questions sit outside this framework.
What to watch
- The full Science Advances methods on how prevented calls were estimated, since that counterfactual accounts for the 5 points between the 18% and 23% figures.
- Whether other cities publish Substitution, Prevention-adjusted and Overlapping-mandate rates for their own programs, which would allow the first comparison across programs on shared definitions.
- Whether Eugene revisits funding for community crisis response now that a standard estimate of CAHOOTS's diversion exists.