Skip to main content

Table 2 Treatment vs. synthetic control: stop profile, crash outcome, and crime outcomes

From: Re-prioritizing traffic stops to reduce motor vehicle crash outcomes and racial disparities

 

Fayetteville Police Department

Synthetic Control

Difference between Fayetteville and Synthetic Control

Pre-intervention annual average

Post-intervention annual average

Post-intervention annual average

Annual Difference

Percent Change and 95% CI (%)

Linear test p-value

Permutation test p-value

Traffic Stop Profile

 Total Safety Stops

13,968 (100%)

34,930 (100%)

15,786 (100%)

+ 19,144

+ 121.3

(+ 17.1, + 318.1)

0.0055

< 0.0001

 % Safety Stops

6119 (43.8%)

23,786 (68.1%)

7296 (46.2%)

+ 21.9%

+ 47.3

(+ 20.0, + 80.9)

0.0001

< 0.0001

 % Regulatory & Equip. Stops

6073 (43.5%)

9583 (27.4%)

6951 (44%)

−16.6%

−37.7

(−54.6, − 14.5)

0.0012

< 0.0001

 % Discretionary

1776 (12.7%)

1562 (4.5%)

1367 (8.7%)

−4.2%

−48.4

(−55.5, − 40.1)

< 0.0001

< 0.0001

Measures of Traffic Stop Disparity

 % Black non-Hispanic Stops

56.8%

54.7%

58.8%

−4.1%

−7.0

(−8.9, −5.0)

< 0.0001

0.250

 Black non-Hispanic TSRR

2.5

2.2

2.8

n/a

−21.3

(−28.5, −13.3)

< 0.0001

0.125

Motor Vehicle Crash Outcomes

 Crashes (all)

5298 (100%)

5160 (100%)

5925 (100%)

−765.0

−12.9

(−37.5, + 21.3)

0.4439

0.125

 Crashes (w/ injuries)

1886 (35.6%)

1639 (31.8%)

2118 (41%)

−479.3

−22.6

(−48.5, + 16.3)

0.2763

0.125

 Traffic Fatalities

62.3

48.8

68.0

−19.3

−28.3

(−64.1, + 43.2)

0.4146

0.125

Crime Outcomes

 Violent Crimes

1223.6

1233.5

1257.3

−23.8

−1.9

(−32.8, + 43.2)

0.9218

> 0.99

 Violent Crime Rate (per 1000)

730.5

596.9

582.4

+ 14.5

+ 2.5

(−14.0, + 22.2)

0.7815

0.750

 Index Crimes

13,367.4

11,658.0

12,896.4

− 1238.4

−9.6

(−24.5, + 8.2)

0.2923

0.500

 Index Crime Rate (per 1000)

7848.1

5637.3

5933.4

−296.1

−5.0

(−12.8, + 3.5)

0.2482

0.750

  1. Table includes both annual averages pre-intervention (2002–2012) and post-intervention (2013–2016). Note: confidence intervals are not symmetrical around point estimates because different methods were used to produce each and small numbers further limited convergence