Tuesday, October 2, 2012

Evaluating times

The City of Lincoln has converted to a biennial budget, relieving those of us in managerial and accounting roles of the annual budget cycle, which begins in December, and ends in August.  Hallelujah.  The administration has decided, in the absence of the work normally associated with budget preparation, that the next year would be a good time to reassess and hone the performance indicators departments are using to gauge their progress towards the City's eight goals.

All three of the public safety agencies are in fairly good shape, but we do have a few measures that need to be tweaked, to make sure they are clearly stated, relevant, understandable by lay people, and backed by a reliable source of data.

LPD, LF&R and the Emergency Communications Center all have one or more performance measure that revolves around time:  response time to crimes in progress, dispatch time for Echo medical emergencies, response time to rescue alarms, and so forth.  For many years, we all set goals for call processing, dispatching, turnout time, response time and so forth as averages.  For example, one of police department's goals is to average less than 5 minutes response time to priority 1 and priority 2 dispatches.

A few years ago, Lincoln Fire and Rescue abandoned the use of averages, and began reporting response time in fractiles. Thus, two of LFR's goals are to achieve an emergency response to life-threatening medical emergencies within 6 minutes, 90% of the time, and the arrival of an advanced life support ambulance within 8 minutes, 90% of the time.  I am convinced that this is a far better way of analyzing response time than using an average.  An average can be skewed by a large number of very short response times, which counterbalance a smaller but significant subset of responses that are unacceptably long.

We are in the process of converting all of our time-based performance indicators that are still averages to fractiles, and setting benchmarks.  This will give us a more informative way of assessing whether we are doing a good job of responding quickly to time-critical incidents, whether police or fire, and of processing and dispatching emergent calls in the Emergency Communications Center.


Steve said...

You better check the "s" on your keyboard. It appears that it's not working all the time. Either that, or the third finger on your left hand (unless you're a hunter/pecker).

CoolStoryBro said...

You could also calculate the standard deviation, remove the outliers and produce an average of those within one or two standard deviations. Although that would skew your results into the positive side, the real work would be determining the factors that caused the others to fall outside of one and two standard deviations. As well, further determining if those factors are within your control.

Anonymous said...

It seems like almost everybody has a cell phone with both video and audio recording capability. Do you have a procedure in place that allows members of the General Public to send in videos of suspicious activity?

Gun Nut

Clean said...

The standard deviation suggestion makes a lot of sense when describing the results in a scientific way. It would also give the Director a lot of practice with his hunt/peck describing what a standard deviation is, every time he reports the results. The fractile description is a lot more intuitive for people who read the LJS or blogs.

Anonymous said...

This information is all good, but everyone involved has to keep in mind the human factor in this time keeping. Dispatchers are busy and at times have forgotten or delayed entry of arrival times. This often times seems to be overlooked by most supervisors and administrators, as these times are looked upon as gospel.