Understanding and Improving The 2009 NPMSRP Reporting Maps

Note: A new map page under construction, until then you can view our 2009 maps here and see our latest maps in our 2010 midyear statistical report here, thank you!

As part of the National Police Misconduct Statistics and Reporting Project’s 2009 Police Misconduct Statistical Report we generated a few maps to show not only where police misconduct occurred, but also how that translated into relative per capita rates of misconduct. However, it may help to go into a bit more detail about what these maps really show and why they were included in that report… and why we should have included even more.

First, The 2009 Police Misconduct Push-Pin Map:

Despite the labels on some of the other maps that were made, this is the only true incident-based map since each push-pin represents a single report of police misconduct that occurred during the 8.5 month tracking period that the NPMSRP was active last year. The other maps show either per capita rates or the number of officers involved with reported incidents of misconduct.

This map isn’t that valuable except as an eye-catching way to show how much police misconduct occurs in the US within that period of time. It would have been more valuable if it were interactive and a visitor could zoom in and out of the map and click on each pin to see information about each incident. Unfortunately, the type of software needed to do that costs more than the NPMSRP, an unfunded independent project, can afford.

Next, The NPMSRP 2009 Localized Police Misconduct Incident Density Map:

As mentioned above, this map isn’t actually an incident map, it is a density map that shows how many officers were associated with reports of police misconduct within the reporting period at a county-level. That is to say, the number of officers working in agencies, city and county, located within a given county that were referenced in a report of police misconduct.

It should be noted that when reports mention incidents where unspecified multiple officers were involved, the NPMSRP makes a conservative estimate as to what that number might be. For example, the general rule is that a traffic stop where a report mentions more than one officer was involved will be recorded as involving 2 officers, a wrong door drug raid will be recorded as involving 3 officers, crowd control 4… As you can see, these are very conservative estimates.

The other problems with this map involve the color scale. First, the colors used for the scale were a mistake since the darker colors for higher number of reported officers obscures some locations where misconduct was prevalent. Namely places like smaller counties near state borders and independent cities like Washington DC because the dark purple is hard to distinguish from the state borders. Second, by topping the scale at 30+ it also obscures the fact that some localities had rates far exceeding 30 reports. In fact, some more than tripled that number. Unfortunately, these problems weren’t realized until the mapping was half-way completed.

This map was still under development when the 2009 preliminary report was released but may be included in the final report when that is published. The value of this map is debatable really, it would have been more valuable as a heat map that showed per capita misconduct density on a localized level but we just don’t have that capability since something like that would require specialized software. This map, just zooming into county level for number of officers involved, took 50 hours to create manually. So, as you can see, doing anything more complex than that would just be too time-prohibitive without software to aid the process. (and you can tell why I didn’t want to redo the scale after I was half-way through)

Next, The NPMSRP 2009 Police Misconduct Incident Map (State View)

This map is similar to the localized map discussed previously in that it shows the number of officers cited in reports of police misconduct recorded during the active project period, April 2009 – mid-December 2009, last year.  By itself, this map doesn’t really have any value and, in fact, can be pretty misleading since it’s hard to tell whether the states that are above average are above average because of their population size or if there is something else at play.

However, this map was included because it is valuable in demonstrating how the number of officers cited in each area translates into an actual per capita rate and can help people visualize why a state that doesn’t look like it has a problem on this map can still show an above average per capita misconduct rate. So, while not useful on its own, it can be useful if viewed side-by-side with the next map.

The NPMSRP 2009 Police Misconduct Density Map (per capita state level)

This map shows the per capita police misconduct rates at a state-level for the recording period. To specify, it shows the per capita ratio of how many officers per state were cited in reports of police misconduct as compared with the number of sworn law enforcement officers reportedly employed at all law enforcement agencies within that state.

While fairly valuable on it’s own, this map still runs into problems with people’s expectations. For example, a lot of people expect states like Illinois or California to rank high, especially when they live there and constantly hear about reports of police misconduct on a weekly basis. (Yes, weekly basis… some cities had more reports than the number of weeks that the reports were recorded). This is why the incident map was included, to help address those questions… however, there’s a problem with that, which is why the following maps that weren’t included are needed too.

The Police Officers Per State Map

This map is fairly self-explanatory yet, in and of itself, doesn’t really have any value if we can’t get any context to go with what we’re looking at. But, paired with our maps above, it tells us how a few reports in a state like Vermont can send the per capita misconduct rate for that state up to the top while the same number of reports in a state like California would barely register.

But, does that really mean anything? Well, statistically, only if the sample time frame is so short that an aberration can creep into the statistics. Otherwise, the per capita rate is what it is, that one state with a lower police population is showing a higher rate of misconduct when compared to the average, then that state is above average for a reason. Our reporting period for 2009 was 8.5 months, at the mid-point Vermont was at the top of the misconduct rate chart. Yet, a few months later, it dropped to third place. So, this tells us that our sample time frame was too short… which is why we hope to get a full 12 months done in 2010.

But… there’s one more map that will help complete the puzzle, because it can help tell us if the police per capita rate for a state is abnormal and, if so, how that might impact the statistics…

The Police Per Capita Map

This map shows the number of law enforcement officers per every 100,000 people in each state, in other words, it’s a police per capita map. By itself this map can tell us if a state is dealing with criminal activity with a police-state mentality by throwing more officers, and thus more money, at the problem or if a state is doing something differently, or just can’t afford to keep so many expensive cops on the payroll.

However, in combination with the other maps, this map can help us determine if having over 81,000 police officers employed in California is overkill or about average for the state’s population. It can also tell us if that deviation from the average is also having an effect on the misconduct rate in some way, whether it might mask what would otherwise be a higher misconduct per capita rate if that state’s officer’s per capita rate were normalized, or if the low number of law enforcement officers might have an effect on police misconduct rates.

For example, law enforcement union representatives often suggest that high misconduct rates are caused by overworked police officers. If this were true, the maps would show a correlation between states with low per capita police officer numbers and states with high per capita police misconduct rates. If we compare the relative maps we see that this hypothesis is inconclusive in that some states with low per capita police rates have a higher than average police misconduct rate, but that this is not true in all cases and that the states with the lowest per capita police rates don’t have the highest per capita misconduct rates.

To sum it all up…

As you can see, a single map or chart can’t show us the whole picture of police misconduct… It takes a lot of work to paint a more detailed picture but, in the end, it’s worthwhile to do so. Its also clear that it will take a lot more work in the coming year for us to develop better reports and better ways to give people more valuable information about police misconduct in the US in ways that are easy to understand and visualize. That includes maps that are more interactive, more fluid, and filled with more pertinent information that will help us better understand the nature of police misconduct in the US.

Information that, hopefully, can be used to help address and reduce the rates of police misconduct that we see today. We hope the NPMSRP will be able to do that if we can keep it going through 2010.