This page provides insights and key themes from community forums and data stories from San Jose Police Department research studies. We will continue to track and update themes from community meetings and provide key takeaways from research conducted on policing.
IPA Community Forums
With the support of the Silicon Valley Community Foundation, The Office of the Independent Police Auditor (IPA) held a series of community outreach events addressing the topic of community trust in policing. The six events were held in communities throughout the City of San José and reached over 300 community members. These events were the inspiration for a text-message, social media, and web-based community feedback campaign that will run through 2020. Three priority area for this campaign include:
Experience and perception about San José Police Officers
Suggestions to improve the Office of the Independent Police Auditor
Ideas to help strengthen trust and relations between police and the community
A belief that vulnerable populations are treated more harshly than other members of the community by San Jose Police Officers
(AP Photo/Marcio Jose Sanchez, File)
A request that officers be given additional training to address community needs and be required to engage in positive community events
Want to make your voice heard? Click the button below to join the conversation.
Data Story: Study of San José Police Department Traffic & Pedestrian Stops
The City of San José hired a research team from the University of Texas at El Paso (UTEP) to conduct a study of traffic and pedestrian stops by San José Police Officers. The purpose of the project was to determine if race/ethnicity had an impact on police behavior (1) when initiating a stop, (2) when making decisions during the stop, and (3) the outcome at the end of the stop. The project looked at 83,831 stops that occurred between September 2013 and March 2016. For a detailed explanation of methods, please see pages 25-28 and 37-46 of the UTEP report.
- Statistical models indicated that there were some racial disparities in some stops, decisions and/or outcomes.
- When other statistical methods were used on the same data, there were no statistically significant differences based on race/ethnicity.
- Racial disparities were less prevalent in pedestrians stops compared to vehicle stops.
- University researchers provided recommendations to reduce actual or perceived racial/ethnic bias (see page 112 of the UTEP Report).
Initiating a vehicle stop
The main question that UTEP researchers wanted to answer was: do drivers of a specific race/ethnicity experience disproportionate vehicle stops in San Jose?
To answer this question, the first step researchers took was to build a model that took into account census data about neighborhoods. They then controlled for many of the factors about the neighborhoods that are being policed in order to better understand what things could be influencing stop rates. The statistical models took into account things like neighborhood, crime rates, unemployment rates, etc. But there are some limitations if you stop here and publish the results. For example, what if Hispanic people may drive more than Asian people in the Central Division? In that case, it might make sense to think that Asian drivers would be stopped less frequently, because there are fewer Asian drivers on the road. To handle this, researchers used two benchmarks to set an expectation against which they could compare the stop data. The benchmarks are used because they set an expectation of how drivers would be stopped if police initiation of stops is random and unbiased.
- Collision Data Benchmark: Racial composition of drivers involved in two-vehicle crashes from January 2013 – December 2015. Collisions where the drivers were both at-fault and not at-fault were considered.
- VOD Benchmark: The veil of darkness benchmark. This assumes that patrol officers are unable to identify the race of a driver at when it’s dark outside. So a comparison is made beteen daytime stops when an officer can identify race and nighttime stops when an officer cannot.
The researchers excluded 2,257 stops made by the Violent Crime Enforcement Team (VCET). An additional 1,241 additional cases were excluded due to unknown race/ethnicity information.
What happens during a vehicle stop?
The main question researchers wanted to answer was: once a driver is stopped, is the outcome of that stop likely to be different based on the race/ethnicity of the driver?
If a police officer takes certain procedural actions during a stop, those actions are recorded. These actions include any detention or restriction on a driver’s movement beyond the vehicle stop e.g. curb-sitting, handcuffing, arrest, etc. They also include other actions like field interviews and citations. Here’s how researchers approached the question:
- Step 1: Evaluate the recorded outcomes during the stop based exclusively on the race/ethnicity of drivers.
- Step 2: User multivariate analysis to control for and understand the potential effect that factors other than driver race/ethnicity had on recorded outcomes. Factors that the statistical model accounted for included things such as race/ethnicity of the officer, poverty level in the district where the stop occurred, and percentage of the district population between 15 and 24 years of age.
You can use the dropdown on the chart below to view analysis of the14 different outcomes.
Initiating a pedestrian stop
UTEP researchers then shifted their focus to pedestrian stops and asked the same questions. First, do people of a specific race/ethnicity experience disproportionate pedestrian stops in San José?
To answer this question, the researchers again established benchmarks to set an expectation of how pedestrians would be stopped if police initiations are random or unbiased. They then used statistical models to isolate the effect of race/ethnicity on the likelihood that a pedestrian is stopped.
- Violent Crime Benchmark: Violent crime suspects. This assumes that data for crime suspects is unbiased. If there are more violent crime suspects of a given race or in a given part of the city, then it is reasonable that patrol officers would stop people of that race and in that part of the city at higher rates.
- Service Calls Benchmark: 911 calls or calls for service to the SJPD. This makes the same assumption as the violent crime benchmark, but uses calls for select calls for service, i.e., drug-related, disturbances, prostitution, and suspicious persons.
The researchers excluded 1,505 stops made by the Violent Crime Enforcement Team (VCET). An additional 575 additional cases were excluded due to unknown race/ethnicity information.
What happens during a pedestrian stop?
Part two of the analysis of pedestrian stops was to explore the activities during the stop and the outcomes of the stop.
Researchers used the same approach for activities and outcomes during a pedestrian stop as they did when analyzing activities during a vehicle stops. You can use the dropdown on the chart below to see different outcomes.
For a more detailed explanation of methods, please see pages 25-28 and 37-46 of the UTEP report.
For additional information and resources about SJPD Use of Force and Crime data, please visit the SJPD Use of Force Analysis and the California Department of Justice Open Justice Portal. These resources are only provided as references and have not been validated by the IPA Office.