Community Violence Intervention is good

We also need to keep rigorously testing the programs we fund

Governor Pritzker with leaders from the Institute for Non-Violence Chicago, Westside Health Authority, and Together Chicago. Source: Governor’s Office.

We’ve spent the last few months looking at the efficacy of a range of violence reduction strategies. So it seems appropriate to talk about Community Violence Intervention: a tool that Chicago has invested heavily in over the last few years. That support has come from the public sector, including City, State and County, as well as the business and philanthropic community.¹

It’s a lot harder to evaluate a group of different community-based interventions than it is to test more standardized approaches. And because CVI is a less established intervention than other tools, there’s a less robust literature base evaluating its effectiveness. That makes this a little harder, and more important. Let’s dive in.

What is community violence intervention?

Community violence intervention is a set of services designed to change the behavior of individuals most likely to shoot others or be shot themselves. Programs can vary in their design, but they generally aim to engage high risk individuals (such as individuals getting out of prison, or who were arrested alongside recent shooters), with a mix of proactive outreach and social support. CRED, one of the largest and most prominent groups in Chicago, outlines 5 key tools to do so: street outreach, therapy (cognitive behavioral interventions), life coaching, education (i.e. helping participants get a GED or vocational training), and employment.

But because different organizations vary in the individuals that they target, the services that they offer, and the efficacy with which they operate, there’s not one highly standardized approach to community violence intervention. You really have to evaluate the efficacy of individual components or organizations (with systematic approaches). Then you have to try to assess how generalizable those learnings and tools are.

Cognitive behavioral therapy is really good

Individuals engaging in criminal violence often have a pretty distorted view of the world. They tend to see themselves as victims, respond disproportionately to slights, and generally have less executive function over their emotions. But these beliefs are not set in stone. Cognitive Behavioral Therapy (CBT) is a specific type of structured intervention that pushes individuals to take accountability for their decisions and reflect on the choices that spurred them to engage in violence.

Done correctly, this can change how individuals react in risky situations. A Campbell Systematic Review of 58 high quality studies finds strong evidence that CBT reduces rates of recidivism by offenders by 25%. Here in Chicago, a randomized evaluation of a CBT-based tool for high-risk individuals in Chicago Public Schools found reductions in arrests for violent crime of 45-50%.

That’s great news. CBT is a relatively standardized, replicable tool that we can use in a range of interventions, and helps explain why CVI programs work. It also means that we should prioritize CVI programming that includes high quality Cognitive Behavioral Therapy programming.

There is solid evidence for CVI programming in Chicago

We also have positive evidence for the effectiveness of other CVI programming in Chicago. In particular, the University of Chicago Crime lab ran a randomized study of the Heartland Alliance’s READI program (Rapid Employment and Development Initiative). High risk individuals were identified by administrative data, recruited by READI outreach workers, or because they were leaving jail or on parole. Then they offered half of those across each pathway various CVI services.

That evaluation didn’t find a statistically significant result for its pre-registered index of violence outcomes. But it did find a 65% decline in shootings and homicide arrests. It also saw bigger declines in violence for those individuals referred by outreach workers (79% reduction in arrests).² As the authors write, that likely speaks to the on-the-ground knowledge of the outreach workers, who are better able to identify high-risk individuals in their communities.

There are also studies of other CVI programs in Chicago like CRED and Communities Partnering 4 Peace (CP4P) that compare individuals receiving CVI programming to similar individuals in the community. Those include a recent study finding that CRED alumni were 73% less likely to be arrested for violent crime than individuals with similar demographics and arrest histories.

I can understand why evaluators do that, but it’s really hard to reliably compare someone who chooses to enter a CVI program to someone who doesn’t. Even if an individual looks like someone else at high risk of violence, their choice to participate in CVI programming likely says something about their future risk profile independent of the CVI programming itself. That’s why it’s generally preferable to look at evaluations like the Crime Lab Study, which look at differences between folks who are *randomly offered* CVI programming compared to those not given that opportunity.

Randomization isn’t perfect – it requires much higher sample sizes (and costs) in order to garner sufficient power to measure effects. It’s also very hard to entangle effects in networks; if you’re in a clique/gang/group that has two other members who get added to CVI programming, that may well reduce your risk of being shot or arrested. Paradoxically, that would make the CVI treatment look less effective - because if individuals tagged in the control group get safer as well, it’d make the difference with the treatment group smaller.³

CVI is a complement to policing

It’s also worth noting that effective CVI efforts are much more likely to succeed against a backdrop of high-quality policing. CVI offers a way out for individuals who are engaging in dangerous behavior. If there’s a high probability that that behavior will get them arrested, the CVI offramp becomes relatively more appealing.

That works both ways. One of the most impactful tools in the criminal justice toolbox is focused deterrence – a process by which police and community members target specific groups of individuals who are engaging in violence, pull every possible lever to improve enforcement, and inform the individuals involved that they can expect swift consequences if they harm others. Focused deterrence works: a Campbell Systematic Review of the available evidence found that it results in a 23% reduction, on average, in crime.⁴ It’s been a particularly important part of Baltimore’s recent record crime declines.

But as the Campbell reviewers noted, focused deterrence strategies were much more likely to succeed if they paired the stick of tough, focused enforcement with the carrot of outreach, education, and job training. As the authors wrote, “In focused deterrence, the value of street outreach workers is clear. Street outreach workers and their community-based allies play essential roles in the service delivery and communications components of focused deterrence.”

You should expect some things to go wrong

Finally, it’s also worth remembering that we’re talking about a challenging population. In the very high-quality CBT research cited above, the rates of violent reoffenders fell 25%. That’s a huge accomplishment – but because the base rate is so high (40% of non-treated participants re-offend), it still leaves a lot of program participants committing crimes. That will happen even in the most successful CVI programs.

And not all CVI programming is created equal. Different operators will have different levels of fidelity to high-quality programming. That doesn’t mean we should excuse poor performance – but bad operators shouldn’t discredit better performing ones. It’s also a reason to keep rigorously measuring outcomes at the operator level.

A fragile success story

Against this backdrop, it’s a genuine accomplishment that CVI programming has expanded so much in recent years. It has likely contributed to Chicago’s recent declines in violence. That success is a credit to funders at the City, County and State, as well as the business and philanthropic partners. Most importantly, it’s a credit to the organizations themselves – and in particular the outreach workers doing a dangerous job in difficult conditions.

But this progress is hard-won and easily lost. As the city’s budget challenges worsen, funding will become harder to find. And at some point – possibly this year – we’ll see violence trend in the wrong direction. When that happens, the pressure will be on.

Governments and donors should continue to fund and scale CVI efforts. They should also insist on a range of independent evaluations that assess specific programs and interventions. At least some of those interventions likely need to include randomization to account for the selection effects noted above.

None of that will be easy. The organizations doing this work are in a constant battle to protect the resources they have, and randomized evaluations will almost certainly cost more money to do correctly. But it will take both funding and high-quality evaluations to turn fragile progress into durable peace.

1. As a disclaimer, I helped support some of these efforts as part of a prior job. I haven’t been directly involved for a while now, but have have a lot of respect and admiration for the groups doing this work.

2. It’s worth noting that READI no longer exists as a standalone program, but some of the key programming and personnel have migrated over to similar efforts.

3. If you’re interested in reading more about these issues, I’d suggest starting here. Another concern with regard to randomization is that it’s potentially unethical to arbitrarily deny life-changing CVI services to individuals who you’ve identified as likely needing those services. But unfortunately, there are far more individuals in need of these services than we have resources for today – assuming relatively fixed budgets, randomization doesn’t change the total amount of services offered, just which individuals receive them.

4. Notably, the review found an only 16% reduction in crime for randomized evaluations.