Wednesday, July 28, 2021

What if any changes to police deployment patterns might reduce violent crime? Hotspot policing vs. ↑ resources for detectives

A friend emailed to ask my opinion of "hotspot" policing tactics being promoted by Houston PD. Here's how I responded:
There are SO many studies on this topic, many of them very micro-focused and not particularly useful, let me give you a big-picture, 50-year overview of the research findings on this.

One of the most robust findings in criminology is that patrol doesn't reduce crime overall or make people feel safer (going back to a major field study in Kansas City in the '70s), and police staffing levels appear to have no relationship either way to crime going up or down.

However, this result didn't sit well with police or their advocates, and in the 1980s, criminologists began to revisit the question, this time shrinking both the geographic areas examined and the time periods considered. Finding a negative result wasn't considered a failure of the tactic, just evidence that the geographic and temporal constraints hadn't been sufficiently narrowed. Eventually, they were able to demonstrate that flooding a neighborhood with police to perform stop and frisks and/or pretext stops correlated to reduced reports of serious crime IN THAT GEOGRAPHIC AREA for whatever period of time they kept it up. There are a bunch of studies out there like that.

However, few of the hotspot studies I've ever seen claim this is anything more than a short-term effect that goes away as soon as police leave.  And most researchers will admit it's likely crime just bleeds into other geographic areas, the way air moves to the sides when you squeeze a balloon.

N.b., generally, what you see when these studies are portrayed in the policy arena is a bait and switch. Cops say "hot spot policing works" then use that to call for increased staffing. But we KNOW increased staffing doesn't correlate to greater safety. The hotspot research is about deployment of EXISTING officers, not an argument for hiring more overall.

Finally, if I were making public-safety recommendations for Houston based on the current data, I wouldn't be focused on patrol or hotspot policing, but beefing up the detective ranks, maybe even AT THE EXPENSE of patrol. Again: The real issues are how officers are deployed, not how many there are. 
There are 200 Narcotics Division detectives at HPD - far more than in homicide. I've argued Narcotics should be entirely disbanded, and those detectives should be moved to investigate 1) homicides and 2) shootings that do NOT result in death. The latter are almost completely ignored but are essentially similar to the murder cases; whether the victim lives or dies has more to do with the EMTs and doctors than the intentions of the shooter. (I'm not generally a fan of the Manhattan Institute, but they recently published a report reaching the same conclusion.) 
So that's the redeployment I think we should be pushing for if the goals are to reduce racial disparities (they're TERRIBLE in Narcotics) while reducing violent crime: Expansion of detective resources to investigate non-fatal shootings. That'd do FAR more to improve safety than anyone would ever claim for hotspot policing.

If you ask what police are actually DOING to reduce crime in hotspot areas, criminologists have no answer. It boils down to what I've dubbed the "Scarecrow Theory" of policing: Their mere, occasional presence wards off potential criminals. But cops aren't deployed theoretically, and as a practical matter, what they do while they're there (if they're deployed to a hotspot and not responding to IRL crime reports) are traffic stops and stop-and-frisks of pedestrians. And most of the people with whom they engage are not and never will be shooters; there's a disconnect between the strategy and the desired results.

I don't consider it some radical position to say homicides and non-fatal shootings should be better investigated: Clearance rates for murder in Houston have declined from 89% in 2011 to 49% last year. And "hotspot" policing would do nothing to change that dynamic.

If the problem you want to solve is violent crime, focus on violent crime. Don't engage in generalized harassment in black and brown neighborhoods then assume reduced murders will somehow be a secondary effect.

13 comments:

  1. Totally agree on reallocating resources to homicides and shooting investigations, with a special emphasis on some of the civilian and non-police aspects: making sure the crime lab technicians have resources and are well-trained (a particularly heavy lift in Houston these days, I know), and making sure that there's money for things like witness protection.

    ReplyDelete
  2. So while your end suggestion (doesn't make sense to hire extra people to do hot spots) is correct, the evidence you use to get there is a bit tortuous.

    We now know the Kansas experiment was poorly designed (low power to detect an effect) -- typical of any scientific study in social sciences 50 years old. The hot spots studies yes were in response to this, since we know random patrol (driving around aimlessly, which is still unfortunately the norm for most PDs) is likely not effective at all and is a waste of gas. Having officers focus attention on specific small areas that have consistently high crime logically makes more sense, and has been shown to be effective when analyzing the effect across multiple studies. See https://link.springer.com/article/10.1007/s11292-019-09372-3, which shows effects averaged across 60+ studies and counting.

    In terms of what police do in hot spots -- it is true that there is quite a large variance in what police departments do in these hot spot studies. You can break them down into quite regular bins though -- problem oriented policing, increased pedestrian stops/traffic stops, or foot patrol cover the vast majority (with some studies having multiple strategies at once). There is mixed evidence across these three what is the most effective, and I would say foot patrol has the most points in its favor at this point. (If you search up the work of Anthony Braga he has several meta-analysis and subset analysis all laying this out.) If the PD is specifically concerned about negative externalities for overpolicing, they can do a strategy that specifically does not promote traditional invasive strategies like stops. Note one of the most popular hot spots strategies is simply to have officers hang out in hot spots for less than 30 minutes a day, and this has been to be effective in multiple experimental designs/cities.

    The point about spatial displacement is quite wrong -- it has been specifically measured and quite a few of these studies find diffusion of benefits (crime goes down in areas nearby the hotspots that weren't themselves treated), the vast majority of these studies do not find evidence of crime displacement. Again whole meta-analysis pooling results from over 50+ studies have shown this -- it isn't really a point up for debate.

    Your final sentence is quite confusing -- if you are interested in reducing violent crime then of course hot spots policing is a reasonable tactic to employ. The clearance rate and crime rates are two different measures though. Hot spots of course won't increase clearance rates -- nor will increasing the clearance rate by default make the crime rate go down.

    ReplyDelete
  3. Andy, I'm glad you recognize that the patrol methods which are "still unfortunately the norm for most PDs" are of little crime fighting use. At least we can agree on that.

    In terms of what police do in hotspots, "foot patrol" correlates to what I've dubbed the Scarecrow Theory; otherwise, it's traffic stops and stop-and-frisk. I agree the hotspot advocates have framed their theories so vaguely they can promise to be all things to all people; that's part of what makes me distrust their claims. In practice, I think the more invasive methods are what police in Texas mean when they advocate "hotspot" methods.

    As for displacement, the studies you're referencing measure whether the crime bleeds into immediately surrounding neighborhoods in the short-term period they study. But what they don't do is tell us that crime overall in a city was reduced via hotspot methods. In every case I've seen that's found a positive result, they look at a short-term period and ignore the long-term.

    Hotspot policing has been all the rage among law enforcement and criminologists and has been implemented all over the country in recent years, yet violent crime is going up. But because they narrow the scope of inquiry by time and geography, you get to pretend it works. E.g., Chicago's been using hotspot policing for years and their violent crime rate is out of control.

    Not sure what you found confusing about my final sentence, which was, "Don't engage in generalized harassment in black and brown neighborhoods then assume reduced murders will somehow be a secondary effect." It seems pretty self explanatory. Perhaps you should read the pull-quote text more carefully, that line references my recommendation at the end of that email.

    ReplyDelete
  4. Here's the cite on Chicago using hotspot policing. Are they the model we're supposed to emulate?

    ReplyDelete
  5. Sorry, my post should have said *the last two sentences*. In your post (both the blockquote and in the final sentences) you clearly conflate clearances with crime rates. Hot spots, which are all about preventing crime, of course will not increase clearance rates. A bit of a straw man there.

    LOL at Chicago implementing hot spots for years -- your comments about long term are correct. Very few departments have any long standing practice of consistent intervention at hot spots (tends to change every few years, the people spearheading certain practices fizzle out and/or the chief turns over).

    The macro level rising crime and hot spots being popular is a bit of a red herring. Most PDs again don't have a rigorous hot spot strategy. For example Dallas PD had official on the map hot spots for quite some time, they didn't actually have a strategy to address them though, only putting a spot on the map clearly won't make crime magically go down.

    I wasn't familiar with that Chicago program offhand. Again there are many different types of hot spots interventions. For example POP programs are neither scare crow as you put it nor encourage invasive tactics like stops, https://onlinelibrary.wiley.com/doi/full/10.1002/cl2.1089.

    Academics are not in a position to do long term assessments, so here is how I think that issue should be done. Locally at the PD they should create metrics to consistently monitor their own hot spots strategy over time (even though on average hot spots work, of course locally it can be done in a poor way and not achieve any of its objectives). This local monitoring is true for other CJ interventions as well, such as bail reform. Included metrics can be stops/arrests or whatever, so if they want to make sure those are going needlessly up it can be addressed as well (New Orleans does this as part of their consent decree).

    I am not going to post more links due to spam, but a final point -- you seem to lump in "criminologists" or "hot spot advocates" like we are all a single cabal of schemers. I can go to the American Society of Criminology conference and find as many people arguing for defunding the police as I can for those who work with PDs. I am sure you can find some random person who says 'police should hire more bodies to do hot spots', but that isn't what myself (nor any of the contemporary big names in the area -- e.g. Sherman, Weisburd, Braga) advocate for. I have specifically written about reallocating on the intensive margin from reactive policing to proactive hot spots for example -- and shown how you can calculate whether an area has enough crime to justify that reallocation from purely a police labor cost perspective (so doing more with less). I have also specifically written about (as have many other people conducting hot spots policing) about mitigating secondary harms from these interventions. It isn't clear to me what criminologists nor hot spot advocates you have in mind -- you certainly aren't accurately representing my views nor the views of the majority in the field who actually do this research as far as I can tell.

    ReplyDelete
  6. Andy, I had suggested that I'd prioritize solving murders directly rather than attempting indirect means via hotspot policing to reduce them. I'm not sure why that confuses you, but it's simply a suggestion of a different approach, not a statement about hotspots and clearance rates. I think solving murders has more to do with, you know, reducing murders, than putting up scarecrows. Apparently you don't prioritize that as much. We must agree to disagree.

    That said, if you grant the "long term" critique, I'm not sure what exactly we're debating. Politicians aren't held to account for week to week stats in a neighborhood that only cops have access to, they're held to account for annual, total, citywide crime stats. Hotspot policing holds out promise of the latter but simply doesn't deliver IRL.

    Further, I find it a bit disingenuous for you to call it a "red herring" when I point out hotspot policing isn't working IRL. You seem to want to say the tactic works, but whenever it doesn't work it's because they didn't do it exactly like you might prefer. (It sounds like Marxists defending Communism during the Cold War, tbh: It'd work if they'd only do it correctly!) But that's not how policy making works. If it doesn't work when implemented on the ground, then it doesn't work. This stuff plays out on the streets and at the ballot box, not in theory.

    Finally, the hotspot research is used to promote police hiring CONSTANTLY - it's the main way it comes up in real-live policy making settings. Never once have I seen someone come forward to say "That's not what we meant." To pretend that academic hotspot proponents bear no responsibility for misuse of their research is an abdication. My depiction may not represent your views, but it accurately describes how this research is used by police chiefs and the unions in real-world debates, which is why I don't find it compelling compared to, you know, murder solving. The claims are overstated and it always turns out to be a bait and switch.

    Since we agree 1) that most police patrol is unhelpful and possibly even wasteful, 2) hotspot analyses can't promise sustained crime reductions over time, and 3) departments shouldn't hire more cops to patrol hotspots, then it sounds like we agree on more than we disagree and I suspect my tone may have offended you. If so, I'm sorry.

    ReplyDelete
  7. BTW, AW, it occurs to me part of the confusion may be that I was responding to an email, which I didn't publish, specifically asking about hotspots in the context of HPD saying it needed more officers to do it. So even if that framing feels foreign to you, it's the context of this analysis, and is in my experience how the issue generally comes up at the local government level.

    ReplyDelete
  8. What many do not mention, or know, about hot spots is that (it being software predicated) an area is determined to be a hot spot is based on spicific day and time determined input, so the output (ie; that particular area is determined to be a hot spot) is time-sensitive. In other words, a software designated hot spot is only that particular hot spot (ie; that one-time software output) for a certain, designated time period (eg; 48 hours, 72 hours, etc ). It is Not always a hot spot, nor is it a hot spot beyond the designated, software determined time frame. The question is, what software, what input and output parameters

    ReplyDelete
  9. Wish you would do a related story on Houston's mercenary police system. The city and county have programs for "additional patrol" (i.e., above zero) for a fee. The constables in particular supplement income with these programs and it is not insignificant. The contracts are with non-governmental entities (HOAs). They use taxpayer paid-for vehicles, gas, etc. Who is their master? The HOA? They are effectively mercenaries dressed to look like constables.

    Sure a peace officer license allows a constable to enforce the law like regular police and others. But this has become such a practice one has to wonder whether the focus is on harassing targets of the HOA for pay or really about enforcing any laws. The practice should be abolished - no "contract" neighborhood nonsense.

    ReplyDelete
  10. > Academics are not in a position to do long term assessments, so here is how I think that issue should be done. Locally at the PD they should create metrics to consistently monitor their own hot spots strategy over time (even though on average hot spots work, of course locally it can be done in a poor way and not achieve any of its objectives). This local monitoring is true for other CJ interventions as well, such as bail reform. Included metrics can be stops/arrests or whatever, so if they want to make sure those are going needlessly up it can be addressed as well (New Orleans does this as part of their consent decree).


    Academics (and anybody else including the local health authority) could be in a position to do long-term assessments, if the local PD were transparent and open with their metrics and data.

    A reason the health authority is stronger is because then you stop talking about hot spot policing, and start talking about hot spot social working and hot spot harm reducing and hot spot ministering etc. Is a police intervention "better" (eg more effective, less downside risk, cheaper) than any other sort?

    ReplyDelete
  11. This paradigm goes round and round, but the overall result is the same: When law enforcement focus more effort on targeting crimes that might create victims more than than crimes that actually create victims, society suffers from more violent crime.

    I find it difficult to comprehend the obstinacy of interdiction advocates in holding an ultimately untenable position on this front. The thing that comes to the forefront of my mind in here involves laziness because targeting folks who just have mental health problems is easier than targeting individuals who victimize others.

    In the long-run, such enforcement practices cause the guardians to become that which they are intended to guard against: victimizers.

    At some point, one must ponder the motives of those touting the continuation of failed practices. That said, I am all for experimenting with different strategies in a quest to find more effective solutions; however, no one truly benefits from refusing that an experiment has shown that a strategy has failed. A maxim often attributed to Albert Einstein states: "The definition of insanity is doing the same thing over and over again and expecting a different result."

    ReplyDelete
  12. The term hot spots policing may be abused by police management when attempting to justify increased hiring but that is no different than when police departments said that they believed in "community policing" in an attempt to, again, justify their budget as well as to appease the public. Yet in that case, many small departments did nothing in attempting to institute community policing other than to relabel those activities they were already partial to as community policing initiatives. Don't blame those who are effective at using hot spots policing to address whatever they see as their current problem - just blame them for doing it inappropriately. We have seen the same thing happen with programs meant to reduce shooting incidents undercut by some departments who only labeled their traditional way of doing business as, for example, "Operation Ceasefire."

    And Grits, not all academics are policy oriented and willing to engage in a public forum so don't blame them when someone misuses their research.

    By the way, great discussion between you and Andy. Made your post even more interesting. My thanks to both of you.

    ReplyDelete
  13. IC...,

    The contract deputy program is not a mercenary system but nice attempt at mischaracterizing it. It's little different from any other department that allows its officers to work off-duty while in uniform, basically renting out their badge and gun to the contractor. Think shopping centers such as the River Oaks Shopping Center that hires off-duty police to patrol their property. It is a way to put deputy's into neighborhoods that otherwise would never see a deputy. The HOA's pay for a percentage of the deputy's time. It varies from a low of 70 percent to, I believe 90 percent (not sure if anyone is paying 100 percent). If they pay 70 percent, they get that deputy for 70 per cent of their time - reputedly. The deputies, if needed, respond outside of their HOAs for a major crime(part of the other 30 percent) but, generally, aren't responding to the low-level 911 calls.

    Who do they report to? The sheriff or constable but, yes. they also pay attention to the HOA coordinator requests....but not if it is a violation of policy, procedure or law. HOA asks them to work certain hours, those are the hours worked. Ask them to address folks running stop signs, that is what they will probably focus on. Bu it will be within the law.

    Why do HOAs hire them? Because they can't get enough deputy's to work their neighborhood as their aren't enough to cover the entire county. HOA money pays for extra deputies. Why does an HOA hire them? Because they like and want the visual presence of a deputy in their neighborhood, someone with a bit more smarts than private security and also with a bit more intimidation factor than private security. It gives warm fuzzes to see that unit drive by every so often, every shift, cruiser lights on at night, reminding knuckleheads not to work that neighborhood.

    Does it cost? Heck yes. I know of a small HOA that drops about $1,000 a day for their deputies. And they are more than happy to do so. Should they pay more? That is between the county and the HOAs but if they were to end the program (against the wishes of lots of voters and campaign contributors), there will be less deputies on the street because the county can't pay for 'em otherwise.

    As an aside, unlike the Kansas City patrol experiment, there are enough deputies present in the small area covered that it does, I believe, have a direct effect on potential crime.

    ReplyDelete