Policing and Risk: Identifying Low Risk Calls

A society wants order and the police were developed as the government vehicle to achieve that goal.  There is often tension between the goal of order and how it is achieved.  The tension often results from errors on the part of the responding officer.  This explains why alternative methods for dealing with lower-level social disorders have been explored.  Nevertheless, alternative systems to deal with order maintenance and peace keeping may carry their own risks and unknown costs.  How do we know precisely the risk level of different call types?  Should regular police officers continue to respond to low-level problems?  Joining the podcast is Dr. Loren Atherley, the Sr Director of Performance Analytics & Research and the Senior Research Scientist for the Seattle Police Department.  We discuss his recent study titled, Risk Managed Demand: Operational Risk Management in Police Response to Calls for Service.

Main Topics

• An “all-hazards response” is the traditional approach to social problems that require police intervention.
• Risk Management literature tells us that risk can be minimized but not eliminated.   So, how much risk are we prepared to accept with a non-officer response?
• The research identified 4 tiers of social response.
• There is no clearly identified “call type” that is low hazard; rather a model of call triage is being developed.

Don't forget to like, FOLLOW, and share.  Sharing this podcast or an episode is one of the best complements I can receive, which will help grow the show.

And don't forget to provide a review.  Giving five stars is never a bad idea.

Feel free to email me your comments using the "send us a text" option (above), or at the following email address: policeinservicetrainingpodcast@gmail.com 

You can also contact me at: Bluesky: @policeinservice.bsky.social

Transcript

SPEAKER_01 0:12

Welcome to the Police in Service Training Podcast. This podcast is dedicated to providing research evidence to street-level police officers and command staff alike. The program is intended to help the police and law enforcement community create better programs, understand challenging policies, and dispel the myths of police officer behavior. I'm your host, Scott Phillips. The history of the police, their origins and development is far too long to discuss in a short podcast. Suffice it to say that society wants order, and they sometimes grudgingly accept the police as the government vehicle to achieve that goal. Still, there can be tension between the goal of order and how it is achieved. The tension between the goals and police behavior often result from errors on the part of the responding officer. This explains why alternative methods for dealing with lower level social disorder have been explored. One alternative is the co-responder model for dealing with a person experiencing a mental health crisis, which I discussed on a recent podcast. Like the cost to society associated with having an enforcement mechanism in the form of a police officer and the possibility of errors and mistakes, alternative systems to deal with order maintenance and peacekeeping may carry their own risks and unknown costs. But are the costs really that high? Should regular police officers continue to respond to lower level problems? Joining me on the podcast today is Dr. Lauren Atherley, and we're going to discuss his recent study titled Risk Managed Demand, Operational Risk Management in Police Response to Calls for Service. Lauren is the Senior Director of Performance Analytics and Research and a Senior Research Scientist for the Seattle Police Department. He holds a PhD in criminology from the Institute of Criminology at the University of Cambridge and is an adjunct professor at the Seattle and is an adjunct professor at Seattle University. Almost made it all the way. Thanks for joining the podcast, Lauren.

SPEAKER_00 2:11

Yeah, thanks for having me.

SPEAKER_01 2:12

Great. I appreciate you coming on. You're out in Seattle and you're willing to talk to me at the three-hour time difference, so that's great. Now, I have to imagine that you've had access or you have access to mountains of policing data ensconced as you are in the Seattle PD. Why study risk?

SPEAKER_00 2:29

Yeah, uh, you know, I I think I'm incredibly fortunate uh with the access to the data, but also the system that I have, uh, access to study here. I like to say that the Seattle Police Department is the greatest social science laboratory in the world. Uh, and I'm I made a conscious choice uh when I left graduate school to spend a little bit of time working in practice. Uh I think a lot of people who get involved in the study of criminology, whether it's penology or policing or victimology, um, tend to dig really deep into the academic literature, but taking a look at practically how that applies and affects uh people's lives is really a rare and unique opportunity afforded to just a small number of people who are fortunate to work in those spaces. And so I took a position in the Seattle Police Department specifically for uh our consent decree. So Seattle uh was uh the subject of a federal consent decree from uh 2012 until 2005, just last year. And uh as a part of that uh effort uh most of the focus of reform was really on forming an evidence basis and then using that evidence basis to rationalize whether whether reform or meaningful change should be met. Um I think for better or worse, uh in 2020 we were kind of closing out that consent decree process when um the murder of George Floyd gave rise to very serious consideration of uh alternatives to an armed response. And uh so I was specifically asked to take a look at what types of calls could we send a non-police responder to safely. And, you know, fundamentally, that question I think is approached from uh a variety of uh perspectives, uh logistical, um uh you know, sort of conceptual perspectives, but fundamentally what we're dealing with there is a matter of risk management. And risk management is uh actually a concept that I think there's really a good foundation for in a lot of other high-risk uh um institutions and and functions. So um it wasn't so much that I specifically made a point to go or was directed to look at risk management as the sort of topic of the day, uh, and that sort of critical decision point was best responded to or answered from the perspective of uh a risk manager. And that's what that's what brought me to this uh area of study.

SPEAKER_01 5:03

Okay, very good. Now, uh obviously I always have my thoughts when I read these articles, but can you explain to the listeners why this particular area of inquiry is of importance or relevance to the police?

SPEAKER_00 5:14

Yeah, so I think mostly uh it's important because when the police get this wrong, uh we see a lot of progressive public policy run the other direction. Um everybody wants that right size response. Uh, everybody wants that just right application or allocation of government resources. When um I think your average person calls 911 and they're reporting, you know, difficulty with a neighbor or uh problems uh with a family member who may be struggling uh with mental health. I think they assume that they're getting specialist resources uh provided for um you know by government funding and infrastructure. And that's really that's really not the case. Um for better or worse, uh most American municipalities and government have evolved this notion of an all-hazard responder to deal with that initial emergent need for order in a civil society. Uh, and that really is a police officer. And it's a police officer because we require that which uh Bittner really sort of uses to define the police officer, which is the ability to use coercive authority uh in the interest of uh greater society. Uh police are uniquely capable of not only rendering uh, you know, sort of bridge services to bring somebody with um, you know, sort of acute risk to themselves or others to a facility that can house them safely, that kind of emergent detention function, but they also can uh mitigate a dispute between neighbors, uh, they can arrest somebody who's engaged in a crime, and I think most importantly, they can defend their lives. And that's really the stumbling block that we run into whenever we try something new in policing. Uh we put uh new resources, new responders, uh people who are specialists in these specific areas of service delivery in positions of risk that is really obscure and hidden. And most of the time, in fact, um the vast majority of the time, it's a perfectly safe thing to do. Um but uh it's one of those areas where the public is very intolerant of the types of failure that can occur. Uh a you know, social worker who dies at the hands of uh a person um on the street who's in need of mental health care services uh ends up being a really big deal. In fact, it usually ends up making national news because our propensity to catastrophize and um uh certainly social media and commercialized news media takes that one occurrence, which is tragic at an individual level, and replays it over and over and over again. And now all of a sudden, the public policy that, you know, from a utilitarian perspective makes good sense most of the time uh ends up being colored by that one catastrophic occurrence. And so it makes it very difficult for us to advance more progressive approaches to service delivery.

SPEAKER_01 8:19

Right, which makes it difficult. And as you say, one one one incident in one location in a country of 18,000, that's just the United States, you've got thousands of other police agencies. I have listeners in Canada, which you know I'm happy about. I don't have a clue how many agencies, but the point is that one of those agencies gets something that happens, it makes the national news, and suddenly all police agencies are tainted with those kinds of things. Now, in in your actual research in the article, and you could you kind of jumped ahead on this, which is which is fine, you mentioned that there are alternatives to the all-hazard responder. In other words, you you don't always have to send a police officer. However, again, you kind of touched on this because society and the police in general are risk-averse, and nobody likes it. We're kind of stuck in that inescapable cycle. Can you explain that part of this?

SPEAKER_00 9:05

Yeah, uh I think um, you know, it's something that I actually uh coined a phrase for uh in response to some work done by uh Dr. Jerry Radcliffe, uh, who was really sort of surfacing this paradox that that we have, which is there there are lots of apparent opportunities for a specialist response, uh, but it does often require an all-hazard responder to be able to triage and right-size that response. So as much as we want to uh send uh a mental health professional to that person on the street who um is uh you know an obvious threat to themselves or others, uh maybe walking in and out of traffic or may have a um, you know, apparently be under some acute delusion or psychosis. As much as it seems intuitive to send a mental health responder to that call, for every one of those, there are um many more uh where it is something that appears to be not right and and needs to be addressed. It is um it is uh a condition being uh described by a member of the public that's not even fully described and usually ends up classified uh by uh 911 uh call takers in the 911 center as disturbance other. And for us to really be able to differentiate between that disturbance other that constitutes uh you know somebody who has specific suicidal ideation but isn't an immediate, emergent hazard to themselves, and tell the difference between that person who is uh you know wandering in and out of traffic and may have a concealed weapon, uh, that uh little bit of nuance between those two situations uh really does require somebody who can uh on very short notice defend themselves. And so um what I term the rat cliff paradox is uh the apparent need uh and ability to send a specialist response, but the requirement to send an all-hazard responder to be able to identify that kind of thing. And when you don't identify it correctly, uh the consequences are often hazardous and dire.

SPEAKER_01 11:16

Yeah, I think we're gonna have to uh you know credit Jerry on this podcast. Uh he would like that. But this actually, you're getting to something that happened. I'm reminded now, Buffalo about two weeks ago, uh there was a guy on the street, he was act acting erratically, he had, I think it was a pair of scissors, and he was trying to stab himself or cut himself, and the police showed up and they ended up shooting him because he actually attacked the officers. But again, this gets back to the well, if we want to be risk averse, isn't the all-hazard responder model still good?

SPEAKER_00 11:47

Yeah, and and I think that's one of those situations where um you know people who are looking to be critical of the way that things always have been done, critical of the status quo, say, well, if you hadn't sent a police officer to that situation, you wouldn't have shot them. Uh I think we uh we demand a form of perfectionism from police that is fairly unreasonable. And while it's it's it's wholly unpopular and and not a great use of statistics to really sort of do this, uh if you're looking at the 270, 280 million calls for service that are received by 911 every year and the roughly thousand or 1,200 fatal encounters uh that result from the police response, you're talking about a very slim margin uh for uh for failure in that system. And while any one of those failures is unacceptable, that's that is a pretty successful um you know, sort of safety rate that we're dealing with. And uh I I think when we um start to think critically about the alternatives, we lose sight of the fact that um police are very well trained and equipped to deal with all sorts of hazardous conditions. And so that person with scissors uh who is uh behaving erratically uh doesn't necessarily mean that the police are going to show up and shoot them. Usually what it means is uh with the appropriate amount of situational awareness, uh police knowing that a hazard exists will uh show up, establish a perimeter that prevents police from being essentially provoked into using force because they can prevent an innocent bystander from coming into contact with that person and that person threatening their life, or uh a uh a hasty uh contact with a social worker or a police officer, they can establish a perimeter and then they can provide that time distance and shielding that allows them to de-escalate that situation. With an established perimeter uh with safety risk managers in place, you can you can slow that situation down, you can negotiate with that person, you can uh provide uh some of those mental health provider services in the field, and you do have at least an opportunity to de-escalate and and prevent a violent situation. Um if we sort of assume that that situation is best handled by somebody who can't shoot that person, we run the risk of uh people being taken hostage or that person using that weapon against themselves and others. And so um I think we lose sight of the fact that the all-hazard responder not only comes with the ability to use deadly force and defense of another, but they also come with the experience tactics and support to be able to create that time distance and shielding that allows us to change that situation from an emergent, you know, sort of physical hazard to people, people, life, health, and safety into a situation where you can have a conversation with that person. You can de-escalate it and ideally resolve that. And actually, most of the time, the statistics would suggest resolve that situation without anybody being harmed. Uh, those are the kinds of opportunities that we create with an all-hazard responder that we lose uh when we potentially expose somebody who isn't as well equipped to manage that situation.

SPEAKER_01 15:20

Yeah, the um the in the in the perfect world, as you were saying, you you establish the time and distance, those kinds of things. But uh again, sometimes the perfect world doesn't exist. But we're gonna be good talking about the the world where those more serious kinds of in incidents are actually a low risk. But uh, before we get there, I wanted to talk to you about have you explain the the different other other industries that you mentioned in your paper that help us help us at least at least okay, how could how do we understand risk management? What were the other industries that you talked about that uh can help us understand what that's all about?

SPEAKER_00 15:53

Yeah, uh you know, I think this is one of those areas where um the policing world's um desire to build something policing specific uh rears its kind of ugly head. Uh when it comes to organizational development or um, you know, any number of kind of standard industry practices, it seems like policing wants to build their own specialized version. But when we start to think about other industries that have very uh limited tolerance for failure um and you know, sort of the potential for really catastrophic failure that occurs very rarely, that risk profile does appear in a variety of places uh that have actually dealt with it pretty effectively. And um, you know, in the in the paper that I wrote, and um, you know, just sort of as I contemplate this uh kind of routinely, airlines uh are a really good example of uh you know situations where um failure doesn't occur very often, but when it does occur, it's often catastrophic, not only for the people who are on that aircraft, but also for the airline itself and for that industry. Um but that doesn't prevent airlines from doing business. And I think police are in a similar situation. Um failure obviously doesn't occur very often uh in this environment. Uh there are literally millions of interactions with a police officer that take place every day uh and uh even with varying levels of pre-existing risk and exposure in those events resolve without injury uh or harm to anybody associated with them. But um, as we've you know sort of noticed uh recently under the second Great Awakening, uh, you know, really since um uh Michael Brown, uh the Michael Brown and the murder of George Floyd and the like, uh, one or two uh sort of uh fringe events can really generate a powerful national movement to upset and change the way that we do things. And so uh a similar form of catastrophe, not only at the individual level, but organizationally, exists for policing that we see for airlines, that we also see in chemical manufacturing and in the nuclear industry. And uh especially in the nuclear industry uh and in uh commercial aviation, those organizations have existed, have uh continued to exist because they have evolved the system for managing those really catastrophic consequences, even though they don't occur very often. Um that that system of risk management that not only considers the worst case scenario, but also how often that might occur, or what the probability or likelihood that that might occur helps to contextualize risk so that they can continue to operate. Um before I I uh got a permanent job um uh shortly after I was out of my my master's degree, uh, I took a job at the airport as a safety manager. I uh I got my pilot's license in high school, and it was uh it was a nice opportunity to do something that was uh different than policing research. So I took a job uh here at Boeing Field in Seattle, uh working for a small uh corporate operator doing uh safety risk management. And um we always used to joke that you know if we wanted to have and achieve uh ultimate safety in this environment, we would um shoot all the airplanes and burn all the pilots. Uh you know, if if we didn't do anything, we could have the safest operation ever. Right. But you're not gonna make any money that way. Uh and I mean we have uh we have a similar situation of policing. If we just parked all the cars and we kept everybody inside the precinct, nobody would ever get hurt. But people might get hurt because the police didn't respond. There is an opportunity cost to every one of these decisions, and that requires some sort of a systematic calculus that you can apply to making decisions about risk that allow you to continue to operate. They either allow you to continue to fly airplanes or they allow you to continue to deliver police service safely and effectively in the way that our communities demand.

SPEAKER_01 20:04

Now, as I was reading your paper, you're you're about to steal a little bit of my thunder, but I'll say this anyway. No, no, it's a that's okay. Um I saw the section the section at how much risk is acceptable. And it's it's that's that's more of a you know one of those flip philosophical questions. You know, what do you what are you prepared to do, kind of a thing. I was reminded of a quote, and then I don't remember who who it was. It's been attributed to several, so I'm not gonna give it to anybody, but uh, as you were saying, if you park all the cars, if you park all the airplanes, and the the quote is a ship is safe in harbor, but that's not what ships are for. Right. So we're still we do have to at least ask a philosophical question of what are you prepared to do? How much risk are you willing to accept?

SPEAKER_00 20:47

Yeah. And you know, that's unique to every organization, and there is no right answer. Um, the the core of my my risk-to-manage risk managed demand work is really about providing a systematic framework to help communities think about what they're willing to accept. And that changes, you know, probably not day to day, but it changes a little bit month to month, and it certainly changes year to year, and it absolutely changes with political cycles. Uh what can be enduring, however, uh is a system that allows you to remember where you came from, uh, to remember what's important, and to make consistent decisions so that uh you can achieve that systematic management of risk in one of those environments.

SPEAKER_01 21:33

Now let's jump into your study, let's talk about that. Uh again, you're sitting inside the Seattle PD and you you got access to like in three years, 700,000 calls for service. Can you tell us just a little bit more about what you did with the data before you analyzed it?

SPEAKER_00 21:47

Yeah. Uh so this was uh this was good fun. We actually took uh large tranche of data uh and uh whittled it down just to those calls for service uh coming from the public. So, you know. Trying to control for some of those recursive effects that come from police responding or electing to respond or from the risk assessment reflecting change in public safety priorities. We took, you know, just those calls for service coming from our community, those community-generated calls for service. And then we also had to do a little bit of fun work to make sure that we were matching up two different uh emergency management systems so that we could have an objective perspective on the outcomes of those calls. So like a lot of police departments, like a lot of municipalities, uh police and fire are actually separate from each other uh in the city of Seattle. And so the computer-aided dispatch system that's used to manage police responses is uh completely siloed and separate. It's uh it's completely gapped from the uh system that's used to manage uh emergency response and emergency medical response in particular. But we were looking specifically for very objective indications of the severe outcomes of these responses, including a police officer. And so utilizing a medical response uh that is uh the result of uh a responder making an assessment that is completely separate from the police department gives us uh some very clean objectivity uh and a nice uh uh coding model that really is applicable across systems. Um laceration is the same here as it will be in another region of the country, uh, and a head injury is assessed the same way you know by an EMT as it would be uh by an emergency room physician. These are objective outcomes. So uh we had to actually take the police call uh and the resulting uh fire emergency medical service call and match them up in space and time. Uh so we looked for uh overlapping geography. We used a search buffer and then also overlapping periods of time. We took the the uh 60-minute period of an hour, we broke it up into 10-minute sections, and we looked for not just overlapping space, uh, so uh responses in proximity to each other, but also overlapping uh responses in proximity that occur at the same time. Uh and then from there we looked uh at fuzzy name match to make sure that we had the right person. So uh combining the police response with a person that was involved in that police response and what the outcome uh was to that person uh allowed us to score the severity of outcome associated with that call for service. Uh, and then taking a look at a large number of those responses, uh, we're able to put together uh some indication of likelihood uh based on you know prior performance, how often that severe outcome occurs associated with that call type. And essentially from there, you plot it on a matrix, the same matrix that uh you know you would use in uh commercial aviation or uh nuclear safety to create that relativistic measure of uh risk in the form of what we call a credible worst-case scenario.

SPEAKER_01 25:17

Okay, so bottom line, can can you the police increase their efficiency? In other words, were there particular types of calls that might be considered low risk that you might not need to send an officer?

SPEAKER_00 25:29

Yeah, uh, you know, there are a whole lot of calls, and I think this will resonate with anybody who's worked in this environment. Um there are about 50% of call types that are um eligible for either not an immediate police response. So these would be the types of things that uh like the differential police response experiments from the 1970s and 80s uh actually were trying to identify and leverage as an opportunity for um uh making an appointment or a walk-up counter or uh calling in, because it it's a it is a request for police service, but it doesn't require police right away. Right. Uh as well as those requests for service that actually don't require a police officer at all, or those types of service that may require uh a police officer to be available for security purposes, for safety, uh, but can actually be primarily handled by a non-police responder. So out of the um sort of volume of 700,000 or so calls that we evaluated, there were really four tiers of service. Uh those that required uh uh a police response, necessarily because of what the risk that they represented, uh, those that could be attended by a police and co-responder together in a in a uh collaborative, uh cooperative uh response, uh those that a non-police responder could handle with police nearby to intervene quickly, and those that didn't require an immediate police response. And the actual efficiency there is about 25% of the total service time that was evaluated during the study period were police responses that actually could be efficiencies. Uh, they could be time that's given back to the police responders. Were we able to triage more effectively and apply that right-sized response? So about 50% of call types resulting in about uh 25% efficiency or return on that police service that didn't need to be spent on those calls.

SPEAKER_01 27:25

Okay, so could you bottom line it then? What are the couple of implications, two or three implications for the police, police departments, personnel leaders, what would they need to know that if you were to summarize this, would be important for them?

SPEAKER_00 27:38

So I think the most important thing here is skirting the issue that everybody tends to fixate on, which is what is the call type that you don't have to send police to anymore?

SPEAKER_01 27:46

Okay, fair enough.

SPEAKER_00 27:48

And that was immediately the problem that I was asked to solve. But the issue is that about 95-97% of the time, the call type that comes into the 911 center is different from how that resolves. So what we have is uh both a system for thinking about risk, thinking about what's acceptable and what isn't, uh, as well as, as I demonstrate um in a different paper uh on intelligent risk management, a model for call triage that makes use of natural language processing and listening in on the 911 call to be able to activate that call triage system. So it's less about deciding what specific call types a police officer no longer has to go to, and more about identifying the very diffuse uh and and varied context that helps give police and the responder ecosystem a perspective on risk that allows them to make those triage decisions in the moment. Um the takeaway here is that there is a system and the technology exists to be able to send diverse responders uh to decide what doesn't need a an immediate police response and to do so safely. But we're really on the uh the cusp, the very beginning of developing those technologies and deploying them to the field.

SPEAKER_01 29:04

Okay, well that's actually good to know because again, 18 uh 18,000 agencies, most of them, well, 89% of them are smaller, less than 50 officers, they're not going to have access to that technology. Now, they might be part of a larger 911 system in the county or something or in the region, but for larger agencies like Seattle, uh LA, uh New York, Chicago, they they would be able to tap into this, but it's like anything else, maybe this kind of technology then can diffuse to these other moderate-sized agencies.

unknown 29:36

Yeah.

SPEAKER_00 29:36

Uh and uh so we do actually have an NIJ grant to uh develop this real-time triage system. Uh we developed proof of concept. We're in the process of implementing that in practice and demonstrating that, yeah, that version one of that can work, it can be effective, it can assist people and what we call intelligent decision support uh in making that triage decision. Uh, and two things are going to result from that. Uh one, we're we're gonna release uh information technically about how that functions on our GitHub. So uh we've been fortunate here at Seattle Police to have the resources to do some really interesting technical things. And from time to time, we release computer code, uh, we release um you know code that people can leverage to replicate a lot of the analysis that we do here. So if you're an agency that has some technical capabilities, you'll be able to uh you know sort of leverage the code and the models that uh we produce, get them from our GitHub and deploy them in your environment. But I think one of the really important advancements here that we maybe don't fully appreciate is that in demonstrating this capability and in making it available, we make it available not only for police departments, which for the most part don't have the ability to do the development work to make it uh a reality, but we also make it available for industry. And um while we certainly don't do this to support any particular industry, we do recognize that most police departments don't have the ability to do you know, sort of really rigorous science uh or to develop computer code, but almost every police department knows how to buy things. And as long as somebody can build a better product that can be purchased, uh, we're happy to support that work.

SPEAKER_01 31:14

Well, this is this is really, like I say, for informative because even I didn't I didn't catch that in the article. I was looking for, okay, what kinds of what kinds of you know, you know, loud music disturbances, fireworks, those kinds of things, barking dog. And so you're telling us it's no, it's a little bit more complicated, and which is what I've always I've been telling students you know for for for decades. Policing is not hard, it's complicated, and you're giving us another example of how complex it can be.

SPEAKER_00 31:38

Yeah, in that uh in that sample of uh of call data between initial call type, final call type, and priority, there are about 42,000 flavors of call that we consider and and apply uh a risk management tool to.

SPEAKER_01 31:53

Excellent work. This is this is great information, uh Lauren. I really appreciate your time and thanks for coming on podcasts. You have a great day. It's been great. All right, thanks, you too. All the best.