Longer police response times in New Orleans have become a well-established fact since last year’s media revelations on the subject. But what role have increased response times played in artificially depressing crime totals?
Analysis suggests nearly 11,000 crime incidents since 2011 and over 7,000 since the start of 2014 were identified as either ‘Unfounded’ (UNF) or ‘Gone on Arrival’ (GOA) rather than reported as potential crimes because of the length of time it took for an officer to be dispatched.
Long dispatch times have likely deflated official crime totals between roughly 11 and 17 percent in each year since 2014 and reviewing this phenomenon in 2016 shows the problem getting worse as the year goes on.
NOPD has clearly taken steps to remedy this problem, piloting a number of new initiatives to reduce response times and increase the speed with which the department reaches calls. What this analysis shows, however, is the gigantic scale of the problem and the degree to which lower priority calls are being left by the wayside.
Crime has risen by nearly 25 percent in New Orleans since 2011, far exceeding the city’s growth in population. The drop in murders and shootings over the last five years is a very positive step, but the unresolved issues with response times remains a major hurdle that has eluded an effective solution thus far.
A note on methodology: this analysis used Calls for Service data from 2011 through the end of May 2016. Dispatch times are defined as the difference between a call’s creation and the dispatch of an NOPD officer.
Dispatch times were used rather than arrival times in order to be more inclusive as more incidents are missing arrival times than missing dispatch times. A call’s arrival time was used if no dispatch time was available. Only calls marked with a Uniform Crime Report (UCR) Part I crime type (theft, auto theft, burglary, robbery, assault, arson, homicide, or rape) and a disposition of either UNF, GOA or ‘Report to Follow’ (RTF) were considered.
Previous analysis have shown that there is a near perfect correlation between number of UCR Part I Calls for Service in a given time period and the number of officially reported UCR Part I crimes. Crimes that were unreported in Calls for Service, therefore, are almost certainly unreported to the FBI in UCR.
Analysis shows there is a large drop off when it takes 2 or more hours for a police officer to be dispatched to property crimes (theft, auto theft and burglary) and 30 minutes or more to person crimes (arson, aggravated assault, robbery, homicide).
Last year, for example, when police were dispatched to a theft (signal 67) in under 2 hours it was given a disposition of ‘RTF’ 80.4 percent of the time. When it took over two hours the percent of thefts marked RTF plummeted to 39.6 percent. Hence longer dispatch times led to substantially fewer crimes being marked ‘RTF’ and a lower-than-expected number reported to the FBI.
The percent of UCR Part I crimes that have been artificially deflated can be calculated by finding the number of Calls for Service each year with a GOA, RTF or UNF disposition in one of several categories. These include: theft, auto theft, vehicle burglary, residence burglary, simple burglary, bicycle theft, business burglary, theft from exterior, shoplifting, armed robbery, aggravated assault and battery, and simple robbery.
Longer dispatch times do not appear to have caused any artificial deflation in the reported numbers of homicide, rape, aggravated battery by shooting, or aggravated arson.
The next step in the analysis is to calculate the percent of calls marked RTF that receive a dispatch in under 2 hours to property crimes or 30 minutes to person crimes. Multiply that percentage by the total number of calls for each crime type gives the expected number of crimes that would be marked RTF if the dispatch times were faster. The difference between the expected number of crimes marked RTF and the observed number telling you how many crimes have been “deflated” by longer dispatch times.
This process can be repeated for all incident types to determine the level of deflation citywide for each year since 2011. Doing so shows the degree to which dispatch time has a major effect on whether or not a call is marked RTF.
For example, aggravated batteries are usually dispatched at an emergency Priority 2 level, but there were 107 Priority 1 aggravated battery Calls for Service in 2015. Those calls received an average dispatch time of 6 hours and 12 minutes, and only 14 of those calls (14.3 percent) were deemed RTF compared to 50.7 percent of Priority 2 aggravated battery calls in 2015.
The progression of this phenomenon over time can be seen in the below chart of theft incidents since 2011. Reports are actually being written more often when police are dispatched in under 2 hours, but the percentage of thefts being reached in under 2 hours has fallen from around 80 percent in 2011 and 2012 to the mid-50s in 2015 and through the end of May 2016.
Applying this methodology to theft incidents creates an estimated 1,031 theft incidents in 2015 that were marked GOA or UNF but may have been marked RTF if all incidents were reported at the percent they were marked for incidents with a 2 hour dispatch time. Theft incidents, therefore, were deflated an estimated 24 percent below their expected number if all such incidents received dispatches of under 2 hours.
There were roughly 18,000 UCR Part I Calls for Service marked RTF in New Orleans in 2015, but this process suggests the expected number of crime incidents was closer to 21,600. In other words, the observed number was about 17 percent lower than expected in 2015 if all property crimes were dispatched in 2 hours and all person crimes in 30 minutes.
This table shows over 9,000 UCR Part I crimes since 2013 have been marked unfounded or gone on arrival that likely would have resulted in crime reports with quicker responses. The official crime statistics reported to the FBI showed UCR Part I crime was down 6 percent from 2014 to 2015, but examining the data in this manner suggests that nearly the entire difference was due to longer dispatch times in 2015 as the number of calls was virtually identical.
This effect is even more clearly seen by breaking down each crime type through May 2016 and comparing those results to 2012, a year with what could be considered a relatively normal level of crime reporting.
So far this year only business burglaries get a dispatch in under two hours over 70 percent of the time for property crimes, and six of the nine property crime types get two hour dispatches in under 60 percent of calls. In 2012, by contrast, all property crime call types received dispatches in under two hours over 80 percent of the time.
Longer dispatch times are decreasing the percentage of lower priority calls that get relatively quick dispatches and increasing the likelihood that the victims will be gone on arrival. As a result, more crimes have been missed in the first five months of 2016 than were missed in each of 2011 and 2012.
The problem is less egregious for person crimes, though there have been an estimated 100 robberies and aggravated assaults/batteries that have been marked GOA due to longer response times.
Moreover, breaking down the deflation of each crime type by month shows that the problem is getting worse as the year goes on. As shown in the chart below, deflation in April and May have largely matched 2015 levels.
National police response time data is notoriously difficult to come by, but all indications are that the response time problem in New Orleans is uniquely (and possibly historically) bad. A Bureau of Justice Statistics report on response times in 2008 showed police responded to 80.8 percent of property crimes in under 1 hour. Some cities show response times, and that those that do badly dwarf New Orleans.
Response times have improved in 2016 from 2015 levels, but the below breakdown of dispatch times by UCR Part I crime type shows most crimes receive slower dispatches than they did in 2014.
Detroit is a city that has been identified as one of the worst in the nation, but a review of Detroit 911 call data for 2016 shows the average response takes between 10 and 38 minutes depending on the call’s priority. The average response for a burglary or auto theft in Detroit in 2016 was 42 minutes as of the end of May according to the city’s data, in New Orleans it was 2 hours and 49 minutes.
Crime deflation is a major issue because it prevents victims of crime from receiving the response and investigation that they deserve. The issue also raises doubts about the city’s official crime statistics and whether they are accurately capturing crime trends in New Orleans. Steps have been taken to rectify the issue, but the analysis strongly suggests that there is a long way to go before the problem is resolved.