Why Police-Homeless Outreach Teams Don’t Always Work

Estimated reading time: 8 minutes

Homeless Outreach and Co-Responder Teams: What You Need to Know

Homelessness is growing issue within the United States. And when there’s no plan in place, the job of figuring out what to do often lands on police.

Most officers didn’t sign up to shuffle homeless folks from one location to another. But without a system, that’s the default. Maybe that’s where you are now. Maybe you’re being told to come up with a plan. Or you’re hearing from city leaders who want to try a co-responder model, like CAHOOTS in Eugene or STAR in Denver.

You’ve probably heard of co-responder models. The idea is simple. Pair a cop with a civilian (someone with mental health or outreach training) and respond together. Two sets of expertise, One goal: get vulnerable people connected to services, not just moved along.

Some of these programs work. Some don’t. And a new study out of Philadelphia helps explain why.

Testing Co-Responder Teams in Homeless Outreach: Philadelphia Study

From June 2022 to June 2023, SEPTA transit police in Philadelphia teamed up with researchers Jerry Ratcliffe and Hayley Wight to test a co-responder model in Philadelphia’s transit system.

The setup: pair a SAVE officer (a transit cop tasked with homeless outreach) with a civilian outreach specialist. The mission? Get people into shelter or treatment.

Officers worked some shifts solo, some with a specialist. The question: Does having a co-responder boost treatment acceptance?

The answer: it’s complicated.

Co-response and homelessness: the SEPTA transit police SAVE experiment – Journal of Experimental Criminology

Objectives We test the benefit of adding an outreach specialist to a dedicated police team tasked with helping the vulnerable community in the transit system move to treatment or shelter. Methods For a year, officer shifts were randomized to determine when they were accompanied by an outreach specialist. One hundred and fifty-eight in-depth treatment conversations regarding treatment or shelter with 165 vulnerable people were assessed for whether they were subsequently transported to a suitable facility. Results Likelihood of an individual in a treatment conversation with a specialist and a police officer being transported to a facility was 29% greater than the likelihood for an individual talking with only a police officer; however, this finding was not statistically significant. Conclusions With the outcome of getting vulnerable people (mainly people experiencing homelessness) to accept transportation to a shelter or treatment facility, the co-responder model did not significantly outperform the effect of specially trained police officers working independently of the outreach specialist.

Do Co-Responder Teams Work?

The short version: officers paired with outreach specialists got more people into treatment than officers working alone – 41% vs. 32%. But the difference wasn’t statistically significant.

What does that actually mean?

It means the gap wasn’t big enough to rule out random chance. In research, a finding is only called “statistically significant” if you’re confident that, if you ran the same experiment 100 times, you’d get a similar result most of the time. This is figured out by using statistical models.

That didn’t happen here.

Why not? Probably because of the small sample size and uneven group sizes. Officers working solo had 104 conversations, while officer-specialist teams only had 61. When the numbers are small and lopsided like that, even a 9-point difference (33/104 vs. 25/61) isn’t enough to be sure it wasn’t just a fluke and won’t swing the other way if this study was run again.

Some might look at this and say, “Then forget the outreach specialists. Cops can do this just as well.”

But here’s the thing: this didn’t fail because of the model. It failed because the outreach specialists were not prepared.

Let’s break that down.

A Closer Look at the Officers

The officers in this study weren’t just regular patrol cops. They were a part of the SAVE unit, a police-only team designed to get those sheltering in the transit system into shelters. They received:

• Basic and advanced Crisis Intervention Training (CIT)
• De-escalation and mental health training
• Trained in recognizing and responding to those who have special needs
• Built relationships with shelters and treatment providers

One even had prior experience as a social worker.

They weren’t juggling this work with 20 other calls for service. They had a focused mission and the tools to carry it out. That mattered.

Now Look at the Outreach Specialists

On paper, the outreach specialists should have made a big difference. But in practice, the role wasn’t well supported:

• Turnover was high. Some left after just a few days.
• One specialist was found to be illegally carrying mace and a handgun, and was fired.
• The job was physically demanding, poorly defined, and often unappealing.
• Officers reported working with at least ten different specialists over the study year.

In short, outreach specialists weren’t set up to succeed. Many weren’t told what the job really involved. Others weren’t trained or equipped to handle the demands. And some didn’t stick around long enough to make a difference.

Bottom line: the officers were set up to succeed. The specialists weren’t.

So What’s the Takeaway?

Don’t read this study as a knock on co-responder models. Read it as a reminder that implementation matters more than theory. A program lives or dies by who you hire, how you train them, and what support you give them.

Here’s what the research and experience tell us:

1. A Well-Trained Officer Can Do the Job

Officers in this study were trained, connected, and focused. That allowed them to handle these encounters well even without a specialist. If your agency builds something similar, you might see success without needing a co-responder on every shift. But keep in mind: this only works when officers are dedicated to this work, not juggling it alongside patrol.

2. Outreach Specialists Are Not Plug-and-Play

You can’t just throw a civilian into this job and expect it to work. Outreach specialists need:

• A clear, honest job description
• Training in homelessness, addiction, mental health, and local resources
• Physical readiness for the work including walking long shifts and working underground
• Support, safety, and pay that match the demands

Lived experience helps, but structure and preparation matter more.

3. Prior Models Rely On Strong Partnerships

Other co-responder and alternative response models show promise but only when the structure is solid. For example:

• Denver’s STAR program(clinician + paramedic) reduced over 1,300 criminal offenses in six precincts.
• Eugene’s CAHOOTS program has been running for 30+ years with success in behavioral health crises. Unfortunately, it has recently ended due to funding cuts.
• CIT-trained officers are more likely to use verbal de-escalation and link people to care, rather than default to enforcement.

But here’s the key: These programs are staffed with people trained and prepared to handle these calls. That’s the consistent thread.

A Few Words on the Research Limitations

This study used a randomized design to determine whether each shift would include just an officer or an officer paired with an outreach specialist. Randomization helps reduce the chance that one group ends up with better or worse conditions than the other. For example, day shifts and night shifts can bring different challenges when it comes to engaging people in treatment. Randomizing helps ensure that both groups experience those differences equally.

But the real world is messy.

The research team did their best to stick to the randomized design, but they couldn’t control for last-minute staffing issues.  As a result, the number of treatment conversations was unbalanced. Officers working alone had 104 conversations, while officer-specialist teams had only 61.

For anyone doing field research, this is a good reminder that things rarely go exactly as planned. Randomized designs are ideal, but flexibility is essential.

What Agencies Can Take From This

If you’re thinking about launching or adjusting a co-responder program, here’s what to keep in mind:

• Specialized units beat split attention. Don’t ask regular patrol officers to tack this work onto their shift. Make space for a dedicated team.
• Train for the mission. Crisis response, de-escalation, and familiarity with local resources are non-negotiables.
• Support your civilians. Make sure they know what they’re signing up for and have the tools to succeed.
• Measure the right outcomes. Track not just treatment acceptance, but follow-up, repeat contacts, and impact on public space or call volume.

Final Word

Co-responder models aren’t a magic fix. They’re a tool. If you put the right people in the right roles, they can work. But skip the training, pick the wrong specialists, or tack the job onto an already-stretched officer and you’ll get results that reflect that.

Enjoyed this post?

Subscribe below and let me know which topics you want to hear more about. It only takes a minute.

If you found this helpful, share it with your team or on your socials. It helps more people make sense of research that matters.

Carlee Ruiz

Carlee Ruiz is the founder of PoliceResearchHub, a platform that turns academic research into practical insights for law enforcement. She brings seven years of experience as a crime analyst and justice researcher, with a focus on evidence based policing, gun violence prevention, and public safety strategy. Carlee currently works as a researcher at National Policing Institute, where she supports evaluations of law enforcement programs and criminal justice initiatives across the country. Through PoliceResearchHub, she aims to make research easier to understand and more useful for the people doing the work on the ground.