The basic arrest data for years 2009-20012 are shown below in Table 1. A Chi-square analysis of the proportion of arrests involving UOF (i.e. UOF Incidents/Total Arrests) showed that there was an overall significant difference across all four years ( χ2(3)=49.8, p < 0.001), and separate z-tests were subsequently carried out on selected pairs of proportions to determine the source of the difference. These analyses showed that there was a large decrease between 2010 and 2011 (0.030 vs. 0.015, z=5.056, p < 0.0001 one-tailed) and that decrease remained stable through 2012 (0.015 vs. 0.013, z=1.067, p > 0.05, two-tailed.) Moreover, of the two years under the new training system, the higher rate of arrests involving UOF in 2011 was also significantly lower than the smaller rate under the old system in 2009 (0.015 vs. 0.023, z = 2.984, p < 0.01, two-tailed.)
Table 1. Number and incidence of calls, arrests and UOF incidents by year
|Total Call Volume||42447||43041||46593||45920||43485||43706|
|Proportion of arrests involving UOF||0.023||0.030||0.015||0.013||0.013||0.013|
|Proportion of calls leading to arrest||0.123||0.122||0.109||0.120||.130||.120|
In evaluating these results, a critic might argue that the observed decrease in UOF could also be attributed to differences in the total number of arrests and/or the call volume that occurred during the years the new system was implemented. An examination of these values in Table 1 suggests that 2011 had the lowest total number of arrests of the four years under consideration, implying that the drop in UOF arrests in 2011 could simply be due to the fact that there were fewer incidents requiring an arrest. A Chi-square analysis showed that there was an overall difference in the proportion of calls leading to arrest (i.e. Total Arrests/Total Call Volume) between 2009-2012 ( χ2(3)=52.4, p < 0.001) and a follow-up z-test showed that the proportion of calls leading to arrest in 2011 was significantly lower than the next lowest rate, which occurred in 2012 (0.109 vs. 0.120, z = 5.151, p < 0.0002, two-tailed). However, the fact that the proportion of calls leading to arrest in 2012 recovered to the same level that it was in during 2009 – 2010 suggests that the unusually low number of arrests in 2011 may have been an anomaly, and that the stability of the decrease in the proportion of arrests involving UOF from 2011-2012 under the new training system, is the more important finding. The fact that 2011 had such a dramatic decline in arrests involving UOF even though it also had the largest call volume suggests that the new training system may have compensated for the added stress that the police were under in responding to incidents.
Additional support for the idea that the new training system influenced the decrease in UOF arrests can be found by examining all of the specific type of force events that had complete data over the four years under consideration, as shown below in Table 2. These results indicate that overcoming physical resistance was the most frequently used method across all four years, while personal weapon was the least common. However, these two methods were not subjected to further analysis because of two major statistical considerations: (1) Because police departments typically quantify use of force in terms of the total number of incidents in which force is needed to effect an arrest (without taking into account the number of officers and the number of different force events involved in a particular incident) overcoming physical resistance would produce several yearly proportions greater than one, and (2) personal weapon did not have enough data to support any kind of statistical test, even at an exploratory level. The resulting proportions for each force event relative to the total number of UOF incidents in that year is shown below in Table 3.
Table 2. Number of each force event used across all UOF incidents by year
|Overcame Physical Resistance||167||211||92||50||520||NA||NA||NA|
|Pursuit of flight (foot)||27||59||21||36||143||7||4||22|
|Total UOF Incidents
(More than 1 type of force may occur during an incident- Therefore this row accounts for Total Incidents- Not total Force Used)
** All Data in blue is NOT part of the original study, and was added after the fact using the same standards as the original study.**
In 2013, the use of force monitoring was further refined to include the category (minimal hands on) which better documents any minor use of force which required and officer to put his hands on a subject to gain compliance. An example would include grabbing a suspects arm/wrist to handcuff a subject that wasn’t complying with verbal commands.
Table 3. Proportion of each force event used relative to total UOF incidents by year
|OC Spray||0.13||0.06||0.03||0.03||Decrease *||analysis||pending|
|Compliance Hold||0.03||0.13||0.13||0.37||Increase *||analysis||pending|
|Pursuit of flight (foot)||0.22||0.37||0.27||0.51||Indeterminate *||analysis||pending|
* Indicates trend was significant using a two-tailed test
A Chi-square analysis was carried out on each of these five force events and the results showed a significant difference for OC Spray ( χ2(3)=11.431, p < 0.01), pursuit of flight ( χ2(3)=19.012, p < 0.001), and compliance hold ( χ2(3)=42.293, p < 0.001), but not for baton ( χ2(3)=5.809, p = .121) or taser ( χ2(3)=7.593, p = 0.05). Given the small number of counts in many of the cells, follow-up z-tests were not performed and so the resulting trends should be viewed as preliminary in nature. However, there is at least some evidence that the use of OC Spray has decreased as a result of the new system while the use of compliance hold has increased.
–end of report summary–