When performing operations analysis on patrol, one necessary ingredient is an accurate workload assessment. The best work schedule depends on how closely your available patrol units align with the demand for service at any given time. This demand for service is most accurately measured by how much time is dedicated to handling calls. Since the number of calls handled is an easy metric to track, this number is sometimes used to assume a resulting amount of work. Fortunately there are now tools available that show the inadequacy of relying on the quantity of calls for service. The unique circumstances of every response causes significant variance in time required. Using the number of calls can be misleading and should not be the primary measurement.
Fortunately, Computer-aided Dispatch (CAD) Systems provide the capability to accurately track all phases of a response to a call: dispatch time, en-route time, arrive time, and clear time. Using those four time stamps, we are able to asses the two most important pieces of data for analysis- travel time and on-scene time. Knowing how much time elapsed in each of these categories allows us to apply an advanced mathematical relation called a queueing model. This mathematical model asserts that a given quantity of available resources can handle a certain amount of work. Therefore, with a known quantity of work, we can determine how many resources are needed to adequately staff for it.
A problem I have regularly seen when examining data from CAD systems is a lack of arrive and clear time stamps. In an emergency, the officer’s focus is primarily on responding to the scene as quickly as possible to render assistance. Enforcing the requirement that they press a key on their mobile computer or relay to dispatch their arrival is often thought of as a nuisance, and habits may form to avoid it completely. When an arrive or clear time stamp is missing from the record, it is unknown how much time was allocated to that call. Because of this, there are only two options for analysis: assume an amount of time or ignore the time completely.
To highlight this problem, I analyzed a sample of CAD data from 34 police and sheriffs departments. I have not shared the names of the agencies, but they cover agencies with a dozen officers to those with over a thousand. I gathered all events that were classified as citizen-generated and had a significant likelihood that time was actually spent on the event but was missing. This method excluded those where the unit was preempted or cancelled before arriving on scene.
The graph in Figure 1 shows the percentage of missing arrive or clear time stamps. Events may have had an arrival but no clear, or had no arrival but a clear at least 10 minutes later than the en-route time stamp.
Agency 6 shows nearly 25% of their unit responses did not log an arrive time. This means that up to one quarter of their on-scene time is unknown. We know when the unit cleared from the call, but we don’t know how long it took to arrive on the scene and therefore the amount of service time spent handling the event.
Agency 32 shows a different kind of problem. There was an arrival time logged, however there was no clear time. The only choice we have here is to assume a certain amount of time spent on-scene. We have two options here. We guess the amount of time spent on-scene – which could be wildly inaccurate; or, we choose to not include the call in analysis. In order to avoid including guessed data in mathematical analysis, 25% of their unit responses must be completely ignored. The result is a significant under-counting of demand for service. If this was a problem that was evenly distributed at all times, one could just inflate the workload by 25% (my inner-analyst cringes while typing this). However, this type of problem tends to be more prevalent when certain officers or groups of officers are on duty. This is not a technical problem, it is a policy problem.
I would encourage every agency to do a review of their own data to identify if a policy needs to be amended, enforced, or implemented. The accuracy of these time stamps may determine if an officer will have backup or not when in danger.
Written by Dan Harris, President and CTO of Corona Solutions