Tuesday, November 20, 2007

Mmmmm... Research

Post #1: The Beginning

Orange County, FL has one of the strongest and most progressive vehicle pursuit policies in the nation. A vital part of a good policy is the ability to review incidents. Whenever someone fails to pull over when asked, deputies are required to fill out a Vehicle Apprehension (V.A.) Form. This form includes, among other things, location, time, and results of the attempt to pull over the suspect.

PursuitWatch.org has begun a informative study of the V.A. forms. They can tell us a lot about the state of police pursuits in a county with a large population and a progressive pursuit policy.

This blog post will be continually updated throughout the research process with:

- Methods of study
- Results of study
- Opinions of the research process
- And anything else I can think of

...Keep checking back.

Post #2: Problems

The Orange County Sheriff's Office has been wonderful in providing me the tools to do this research. There are, however, limitations to my access to the V.A. forms.

  • There are literally thousands of these forms.
  • Before June of 2006, V.A. forms were not kept in a central location. They were housed throughout the county in various offices. However, in June of last year, the decision was made that all V.A. forms should be kept together, so all of the forms dating back to 2003, were brought to headquarters. It would take endless hours to sort these forms out. So, the only way to gage proper sample size would be to draw from the forms, starting with July of 2006 to June of 2007, the time period after the forms begun to be organized. Of these files, they are separated by month.
  • There is sensitive information on the forms that I cannot view. For example, if the suspect that failed to stop is a juvenile, I cannot view any of the information. Also, things like names and social security numbers of the suspects are personal data that must be blacked out. Because of this I cannot myself sort through these forms. I can only view them once they have been reviewed. Will this spoil the results. No. I am taking every necessary step to ensure an accurate sample size. I will talk more about that as the process continues.
Next, I will be talking about the sample size that I have decided to go with, and why. Again, this study will be done correctly to ensure that the findings can be stated with confidence.

...More to come.

Post #3: Maps

An essential part of the research we are tackling here deals with the location of the incidents. We would like to be able to, for example, say that a "majority" or "1/3" of these incidents occurred in this part of the county. This, however, brings up a few questions. Since it is a quantitative study of the Vehicle Apprehension Forms, a way of coding must be established. Here are a few of the ideas that are being discussed.

We can divide the county up as the OCSO does, by sectors. Below is their county map divided by sectors.
A simple glance at the map brings up a glaring problem. How can this map be used to provide accurate numbers? For example, what if our study finds that there were four times as many incidents in zone 2 then there was in zone 6? Does that mean anything? Well, first, zone 2 is much, much larger than zone 6. But, at the same time, zone 6 might have a larger population, more roads, or more traffic.

Another way of approaching the geography issue is by dividing the county up in to areas with the same amount of square miles. This also has a few drawbacks. What if one sector has almost no roads, or traffic, thus no incidents?

So the question remains, can we accurately view the results using this method? The solution might involve using both methods. Also, as is the case with any research, the methods must be accurately explained so that, as a college professor of mine would always say, "you don't become dangerous."

Another method that will be done is simply dotting the point of the incident on a map of the county. That way one can understand that makeup of the counties incidents. This, however, is not quantitative.

1 comment:

over educated deputy said...

In many ways the vehicle apprehension forms are not quantitative in nature and would be more appropriately studied using qualitative measures since there are several factors at work here.

A more appropriate research methodology might be to use the existing sheriff's zones, and do a per capita comparison between them. Additionally you might compare this with the calls for service in the particular zone to determine if there is an apparent relationship between vehicle apprehension forms and reported crimes as measured by calls for service.
Unfortunately your attempts probably will be doomed by confounding variable which you have failed to take into account.

Sector 2 while large physically also has one zone (22) which had a higher call volume than all of Sector 1 last year.

A quick breakout of the OCSO sectors:
Sector 1- Apopka, Winter Park, Maitland, Eatonville, half of Pine Hills, and half of Ocoee. High call volume in some areas (South Apopka, Pine Hills), slow in others.
Sector 2- UCF area, Bithlo, Lake Nona. Exceedingly high call volume in some areas (UCF corridor) relatively lower in others
Sector 3- WIndermere, half of Ocoee, Half of Pine Hills, Oakland.
Exceedingly high call volume in some areas (Pine Hills) slow in Windermere, Oakland.
Sector 4- Orange blossom trail from sr 50 to Osceola county, Heavy call volume through the area, lower call volume elsewhere.
Sector 5 I drive tourist corridor Relatively low call volume, low violent crime
Sector 6- Disney Lake Buena Vista Low call volume- low crime

Comparing sector 6 with sectors 2 and 4 would be fallacious at best.