Hotspots: The rare occasion bigger is not better
Throwing police officers at crime hot spots has long been a preferred tactic of police commanders. It’s not a bad idea. Nearly 20 years ago, the National Research Council’s Committee to Law and Justice noted "studies that focused police resources on crime hot spots provide the strongest collective evidence of police effectiveness that is now available." (1) So it might be easy to fall victim to the notion that designating more and more of a precinct a crime hotspot would be beneficial. This is a mistake.
I raise this, because I’ve started noticing that police commanders (in a number of departments I am working with) tend to get a bit carried away and expand their saturation patrol areas without adding additional resources. It is understandable. If a tactic can help, surely the benefits will increase when expanding the target area, right? And if under pressure from the police chief to do something about a problem in a neighboring area, why not just add to the existing hotspot?
Here is a quick simulation to demonstrate why focused activity is preferable to spreading your limited resources more widely.
A quick demonstration
This image shows 20 artificial police beats in a single police precinct with random numbers generated in each cell (range 0-10) representing fictitious crime counts. They are colored blue to red. The total amount of crime is 106.8, shown in green. Lousy number, but I did use a random number generator.
Now imagine a police commander has enough additional resources to have a 10% improvement in every one of the 20 beats. Cells in bold show where this change has been made. In each cell, the (rounded) cell value has been multiplied by 0.9 to represent a 10% reduction.
As you can see, the new crime total is 96.1 and we have reduced crime by 10.7. A 10% overall reduction. Well done commander.
But... what if we focus the same amount of hotspot police activity into ten beats, rather than 20? So let's hypothesize we can have a 20% reduction in 10 beats, rather than a 10% reduction in 20 beats. It is the same benefit, but just more focused.
Now - even though we have not added ANY extra resources to half the precinct - we have actually reduced the overall precinct crime amount to 90.4, reducing it overall by 16.4. This is a 15.3% crime reduction. The reason, is we are focusing resources on the places that generate the most return on that investment.
Let's go further. What if we focus resources in only a QUARTER of the precinct's beats? Now by concentrating a .4 effect in 5 cells, we get down to 87.8 with a monster crime reduction of 19 (nearly 18% crime reduction).
More? A .5 reduction across only *FOUR* of our 20 beats generates an overall reduction of 22.2%, getting down to just 83.1. Remember, we haven't used any additional resources, but we have squeezed more effect from those same officers.
So can we go too far? Apparently we can. Focusing on just two hotspots within the precinct - even if we reduce those two cells to zero - does not generate the same level of crime reduction. So there is a limit beyond which we lose effect.
So what lessons can be learned?
First, diluting resources as widely as possible is a bad idea in a (fairly standard) environment where there are hot and cold spots within a precinct. This is the concept of dosage. Police commanders, think of yourself as a surgeon with a scalpel, focusing your work where the cancer is most prevalent.
Second, in most areas, there is probably a level of diminishing returns where throwing more and more resources into a tiny area becomes ineffective. However, the overall lesson should be clear. Concentrate, concentrate, concentrate. Resist the urge to expand when you do not have additional resources. A strong effect in the most important places is better than a weak effect overall.
A strong effect in the most important places is better than a weak effect overall.
Third, modeling your own crime situation with an exercise such as this - or finding a good crime analyst to help you with it - might produce some insights into the distribution of crime in your area. After all, your mileage may differ. You might find it helpful in making deployment decisions. And that's good for the community.
(1) Page 250 of National Research Council. (2004). Fairness and Effectiveness in Policing: The Evidence (Editors: Wesley Skogan and Kathleen Frydl). Washington, DC: Committee to Law and Justice, Division of Behavioral and Social Sciences and Education.