Friday, July 31, 2009

Dashboard indicators

I watch my speedometer, gas gauge, temperature gauge, warning lights and other instruments to monitor the vital functions of my car. Similarly, I watch the key metrics of the police department in order to stay informed on how we are doing. I have a lot of data that I monitor on a regular basis, pertaining to many aspects of the department’s operations: the monthly statistical summary, personnel projections, year-to-date crime data, bi-weekly salary report, budget status report, and so forth. It’s sort of my personal “dashboard” of the organization.

From a broader perspective, the police department benchmarks several key progress indicators that are contained in the City’s outcome-based budget. We are contributing primarily to two of the City’s eight outcomes: Safety & Security and Livable Neighborhoods. For each of these outcomes, we have identified the specific goals we contribute to, and progress indicators that monitor how we are doing on those goals.

The compilation of the eight outcomes, the goals under each, and the progress indicators composes the Budget Outcome chapter of the City of Lincoln’s budget. In order to monitor how we are doing, I created some simple graphics that are meant to provide the key data at a glance—just like your instrument panel. You can watch, too. A collaboration yesterday morning between the Excel-nerd chief and the web guru Officer Katie Flood resulted in the dashboard indicators on our public webpage . Just follow the Dashboard link in the contents menu, or click this thumbnail.


Anonymous said...

Wouldn't the rate of violent crime (excepting murder, of course) be dependent upon reporting rates? Wouldn't burglary be similarly colored, since not all residential burglaries are reported (for reasons previously mentioned in this blog)?

How about just doing one with murder and business robbery as the crimes that make up "violent crime", and one for business burglary. I'm not busting your chops, but I'm interested in the crime stats with the smallest difference between actual and reported rates.

Tom Casady said...

No offense taken at all. You and I have discussed this on several past occasions in these pages, and we both agree that comparative crime stats have to be taken with a grain of salt, and that murder and business robberies are probably the two most reliable numbers.

Problem is, there are so few of both in Lincoln, that you just can't discern any trend. If you want to watch how things are going over a period of years, you have to pick something that occurs with enough frequency that a trend will be apparent if it exists.

Since our reporting practices here at LPD are reasonably stable, the year-to-year data ought to be fairly indicative of what's going on--unless for some strange reason the public willingness to report these crimes rises or falls dramatically. I don't think that's likely, and any variation in people's reporting habit would not be so significant as to render these comparisons fruitless.

Regarding burglary, we selected that as a measure of property crime for the same reason: stable reporting practices and large enough numbers.

None of these are perfect, but you've got to measure something if you want to gauge your performance. I've tried to select indicators that are least subject to error and misinterpretation, at least somewhat influenced by what the police do, simple to explain to people, and for which data is readily available.

Anonymous said...

I was curious about how you come up with your numbers for crimes that you give out to the media and public. Do you use the UCR code? or do you use other data.
The UCR codes would lead to very false numbers in my opinion. I have worked cases in the past where records changed the UCR code. For instance, I worked a case of a domestic assault where ex live in boyfriend punched his women in the face because he was cheating, then on the way out the door to flee before the police came he grabbed the cell phone off the table which was shared mutually by both of them. Records changed this to a robbery in the UCR code. So now we have a violent crime of robbery which it wasn't. The case was cleared with the arrest of this person and he was charged with domestic assault. But the UCR shows we cleared a robbery. Can you please explain this? How can a records person change our reports and elements to fit something else that it isn't? This was brought up to the supervisor level and nothing could be done.

Tom Casady said...

8:50 -

Statistics I provide are usually the FBI Uniform Crime Report (UCR) Part 1 Crimes. The definitions of these are not always the same as Nebraska Statutes. The FBI publishes a 164 page handbook on UCR coding. It is not for the feint of heart, but definitions are illustrated with good examples.

The UCR definitions of aggravated assault and burglary, for instance, have some quite distinct differences from the definition of a felony assault and a burglary in Nebraska State Statutes. Here's as common example: someone walks through an open garage door and takes a bicycle. This is clearly a UCR burglary (see page 30.) But this crime does not fulfill the elements of the crime of burglary in Nebraska Statue 28-507, which requires forcible entry.

If you want to collect nationwide date, you have to use some definitional guidelines, since laws vary from state to state. Hence, a Lincoln police officer may investigate that theft of a bicycle, submitting a report for larceny, and the Records Unit will code the crime as a UCR Burglary.

For statistical purposes and reporting, UCR rules. For criminal prosecution, Nebraska State Statutes rule. Trying to get 317 police officers to understand the intricacies of UCR defintions and coding guidelines would be next to to impossible. This is why many departments have problems with their UCR reporting. It's a pretty specialized task, and requires a high level of consistency.

In the robbery/assault example you gave, the coding call is close. Under the FBI definition of robbery, ownership is not a deciding factor, rather "care, custody, or control":

"...take anything of value from the care, custody, or control of a person or persons by force or threat of force or violence and/or by putting the victim in fear."

On the other hand, was force, threat, violence or fear used to take the phone? There's a hierarchy rule to consider, as well, and a rule concerning separation in time and place. I think you might be able to make the case either way.

We have submitted a couple of times in recent years to UCR audits--at our request--to make sure we're doing it right. The FBI pulls a sample of cases, reads the investigative reports, and checks to make sure our coding matches the definitions. So, although there are no doubt errors made in a process that handles this kind of volume, I do not think we have any systematic shortcomings.

Anonymous said...

I was surprised to see that the response time is and has been under 5 minutes for a number of years. I've heard many people complain about response times being much higher. Is it possible that your response time data is inaccurate or are the comments I've heard unwarranted?

Anonymous said...

Interesting information. Hopefully no one takes this as an indicator that "things are fine as they are" and "we don't need more or as many police". Frankly, the statistics you've shown before comparing us to other cities our size and in NE are SCARY.

On a side note, has LPD changed their cruiser logo again? I saw one parked at the Highlands fire station the other day with a black/gray/silver logo on a white car instead of the familiar blue logo on white. Looks sharp...but was just curious.

Anonymous said...

I'm certainly glad that we don't have enough murders to discern a trend. In low-murder locales, even a single murder can send the rate up sharply.

For example, Sheridan, Wyoming. I believe that the only murder they've had during this century was in 2004, just one, but it rocked the rate up from zero to 6.2/100k. That was higher than Omaha's rate that year (a tame year for Omaha, only 20 murders for a 4.9/100k rate).

By the way, details on that single Sheridan murder weren't easy to find, and I've come up with nothing on it so far. If I was there, I'd just walk in to the Sheridan PD and ask them about it.

Anonymous said...

Hello Chief! Thanks for the blog, I've enjoyed it daily for a few months now.

I was wondering if you could comment on how exactly the police department can be held responsible for ensuring lower crime in Lincoln. I mean, it's not your fault if someone breaks into a home.

Another question that's related is how exactly do the police effect crime in Lincoln? Is it as simple as having a presence in certain parts of Lincoln? So the more officers in an area = less crime? Or perhaps the more that crime is caught and prosecuted = less crime?

Anonymous said...

Oh what a difference a letter makes!

I couldn't resist this correction of "feint of heart" in the same sentence as "definitions".

Faint of heart is defined as "timid, lacking conviction or courage".

Feint is defined as "a feigned attack designed to draw defensive action away from an intended target".

Thanks for the interesting blog topics and discussions - I read daily but have never commented before. I usually just overlook grammatical errors but couldn't resist this one.

Tom Casady said...

10:54 -
If you read the fine print at the bottom of the Dashboard page, you'll see that the 5 minute response time goal is for Priority 1 and Priority 2 events--such things as crimes in progress, alarms, incidents presenting an immediate threat to persons or property. In other words, the things where time really matters.

11:03 -

We are experimenting with some modifications to our patrol car graphics, so you may see some other variations in the next few months. We're striping up a single car, rather than just looking a computer mock-ups.

11:27 -

There are really only two performance indicators on this dashboard that the police control completely. For the remainder, we are only part of the production function. Nonetheless, there is a link between what we do and all of these indicators and I think our role in supressing and preventing crime is an important one.

Read more about what we do and why it matters in the five-part series I wrote back in November, beginning with Theory and Practice on Monday, and ending with Evidence-based policing on Thursday night.

Tom Casady said...

11:49 -

Typo. Sue me. ;-)

You can't edit comments, and it's too much effort to go back and retype the whole thing.

ARRRRG!!!! said...

Here's my dashboard.

Trevor Brass said...


Why do you think the "Safe & Secure" category is so volatile when compared with the other metrics? Does it have anything to do with a broad sweep of crimes?

Tom Casady said...

Trevor Brass-

It's really not that volatile, ranging from a low of 72% to a high of 79%. The scale of the Y axis (.68 to .85) exaggerates the year-to-year fluctuation. The same thing is true of professional performance. The minimum and maximum scale values in all these graphs were the defaults selected by Excel. The other five graph has a minimum value of 0. Excel must have picked a narrow range of scale values for these two due to the smaller variations in the values. I should manually set a lower minimum value for the y axis for both of those graphs, so the graph emphasizes the stability in these data, rather than the variation. I'll try to remember that when I update this with 2009. I should have noticed it like you did. It's perfectly accurate, though. I left the line and scale marks for the y axis off in these graphs just to keep them very simple and clean. The downside to that choice is that it is not immediately evident to a viewer that the scale starts at 68% for the Safe & Secure graph.

Anonymous said...

Giving Dogs Medicine - order xanax Xanax is preferable when anxiety needs to be treated short term, as it begins to work within the first week. [url=]order xanax online[/url]