i ANNUAL REPORT 20 20 RACIAL & IDENTITY PROFILING ADVISORY BOARD

1 RACIAL AND IDENTITY PROFILING ADVISORY (RIPA) BOARD SAHAR DURALI, Board Co-Chair, Associate Director of Litigation and Policy, Neighborhood Legal Services of Los Angeles; Appointed by the Attorney General of California (Board Co-Chair as of September 2019) SHERIFF DAVID ROBINSON, Board Co-Chair, Sheriff, Kings County Sheriff’s Office; Designee of the President of California State Sheriffs’ Association MICAH ALI, Vice President, Compton Unified School District Board of Trustees; Appointed by the President Pro Tempore of the Senate OSCAR BOBROW, Chief Deputy Public Defender, Solano County; Designee of the California Public Defenders Association PASTOR J.

EDGAR BOYD, Pastor, First African Methodist Episcopal Church of Los Angeles (FAME); Appointed by the Attorney General of California SANDRA C.

BROWN, Lieutenant (ret.), Palo Alto Police Department; Appointed by the President Pro Tempore of the Senate ANDREA GUERRERO, Executive Director, Alliance San Diego; Appointed by the Speaker of the Assembly (Board Co-Chair through September 2019) LAWANDA HAWKINS, Founder, Justice for Murdered Children; Appointed by the Governor of California DAMON KURTZ, Vice President, Peace Officers Research Association of California (PORAC); Designee of President of PORAC REVEREND BEN MCBRIDE, Co-Director, PICO California; Founder, Empower Initiative; Appointed by the Attorney General of California EDWARD MEDRANO, Chief, California Department of Justice, Division of Law Enforcement; Designee of the Attorney General of California DOUGLAS ODEN, Senior Litigation Attorney, Law Offices of Oden & Greene; Appointed by the Speaker of the Assembly STEVEN RAPHAEL, Professor of Public Policy, University of California, Berkeley; Appointed by the Governor of California TIMOTHY P.

SILARD, President, Rosenberg Foundation; Appointed by the Attorney General of California COMMISSIONER WARREN STANLEY, Commissioner, California Highway Patrol; Designated by The Racial and Identity Profiling Act of 2015 CHIEF DAVID SWING, President, California Police Chiefs Association; Chief, Morgan Hill Police Department, Designated by The Racial and Identity Profiling Act of 2015 TIMOTHY WALKER, Senior, San Francisco State University; Mentor, Community Coalition, Los Angeles; Appointed by the Attorney General of California

2 The Racial and Identity Profiling Advisory (RIPA) Board thanks the following staff from the California Department of Justice for their assistance and contributions to this report: Editors California Department of Justice, Civil Rights Enforcement

Section (CRES) Allison S. Elgart, Deputy Attorney General Catherine Z. Ysrael, Deputy Attorney General Domonique C. Alcaraz, Deputy Attorney General Anna Rick, Associate Governmental Program Analyst Aisha Martin-Walton, Retired Annuitant California Department of Justice, California Justice Information Services Division (CJIS) Kevin Walker, Research Associate I, Research Center Trent Simmons, Ph.D., Research Analyst II, Research Center Tiana Osborne, Research Analyst I, Research Center Evelyn Reynoso, Research Analyst I, Research Center Project Supervisors Nancy A.

Beninati, Supervising Deputy Attorney General, CRES Randie C.

Chance, Ph.D., Director, Research Center, CJIS Jenny Reich, Director, Justice Data and Investigative Services Bureau, CJIS Additional Editors and Contributors Amanda Burke, Ph.D., Research Associate I, Research Center Erin Choi, Program Manager, Client Services Program, CJIS Charles Hwu, Data Processing Manager, Application Development Bureau, CJIS Tiffany Jantz, Ph.D., Research Associate I, Research Center Tanya Koshy, Deputy Attorney General, CRES Audra Opdyke, Assistant Director, Justice Data and Investigative Services Bureau, CJIS Jannie Scott, Ph.D., Research Associate I, Research Center Christine Sun, Special Assistant to the Attorney General, Executive Office

3 The RIPA Board thanks former Associate Governmental Program Analyst Kelsey Geiser for her significant contributions and dedication to this initiative over the years. The RIPA Board thanks Alfred Palma from the California Department of Justice for his diligence and for serving as the travel coordinator to the Board. The RIPA Board thanks the Commission on Peace Officer Standards and Training (POST) for its partnership and looks forward to continuing to build on this collaboration.

The RIPA Board thanks Magnus Lofstrom, Ph.D., Brandon Martin, MA, and Justin Goss, MPP, of The Public Policy Institute of California (PPIC). PPIC researchers provided technical research assistance but are not responsible for the accuracy of the raw source data and/or any conclusions extrapolated from the technical research assistance provided and contained in the RIPA report.

The RIPA Board thanks Jack Glaser, Ph.D., Professor, Goldman School of Public Policy, University of California, Berkeley; and Emily Owens, Ph.D., Professor, Department of Criminology, Law and Society and Department of Economics, University of California, Irvine, for their assistance in preparing this year’s report. The RIPA Board appreciates the participation of community members, members of law enforcement, advocates, researchers, and other stakeholders.

Public participation is essential to this process, and the RIPA Board thanks all Californians who have attended meetings, submitted letters, and otherwise engaged with the work of the Board. We look forward to continuing input from the public.

4 Table of Contents EXECUTIVE

SUMMARY 5 OPENING LETTER FROM THE RIPA BOARD CO-CHAIRS 13 INTRODUCTION 15 ANALYSIS OF WAVE 1 STOP DATA: JULY 1, 2018 - DECEMBER 31, 2018 20 RACIAL AND IDENTITY PROFILING POLICIES AND ACCOUNTABILITY 43 CALLS FOR SERVICE AND BIAS BY PROXY 54 CIVILIAN COMPLAINTS: POLICIES AND DATA ANALYSES 58 REVIEW OF WAVE 1 AGENCY COMPLAINT FORMS 81 POST TRAINING RELATED TO RACIAL AND IDENTITY PROFILING 91 RELEVANT LEGISLATION ENACTED IN 2019 95 CONCLUSION 96

5 EXECUTIVE

SUMMARY California’s Racial and Identity Profiling Advisory Board (Board) is pleased to release its Third Annual Report. The Board was created by the Racial and Identity Profiling Act of 2015 (RIPA) to shepherd data collection and provide public reports with the ultimate objective to eliminate racial and identity profiling and improve and understand diversity in law enforcement through training, education, and outreach.

For the first time, the Board’s report includes an analysis of the stop data collected under RIPA, which requires nearly all California law enforcement agencies to submit demographic data on all detentions and searches. This report also provides recommendations that law enforcement can incorporate to enhance their policies, procedures, and trainings on topics that intersect with bias and racial and identity profiling.

This report provides the Board’s recommendations for next steps for all stakeholders – advocacy groups, community members, law enforcement, and policymakers – who can collectively advance the goals of RIPA. In rendering these recommendations, the Board hopes to further carry out its mission to eliminate racial and identity profiling and improve law enforcement and community relations.

Recommendations for Law Enforcement Agencies The Board has engaged in an extensive review of best practices to provide law enforcement with concrete recommendations focused on improving bias-free policing and civilian complaint policies and procedures. The Board recommends that law enforcement engage with their communities as they develop and improve policies and practices that are strong and effective while also enhancing transparency, building trust, and promoting the safety and, well-being of all parties.

Below we provide an overview of the recommendations included in this year’s report, and we strongly encourage stakeholders to review the detailed policies set forth later in this report and in the attached Appendix. Policies: This report contains model language for the following: a clear, written bias-free policing policy;

definitions related to bias; the limited circumstances when personal characteristics of an individual may be considered; training; data collection and analysis; encounters with the community; accountability and adherence to the policy; and supervisory review. The Board recommends that all agency personnel, both sworn and civilian, receive training on their bias-free policing policies. Agencies are further encouraged to develop policies and training on how to prevent bias by proxy when responding to a call for service.

In addition to including model language, the Board conducted a policy review to assist Wave 1 agencies in identifying areas of opportunity to incorporate the best practices and model language presented in this report and the 2019 RIPA Annual Report. For the purposes of this report, Wave 1 agencies refers to the eight largest law enforcement agencies in the state that began collecting stop data on July 1, 2018, and reported it to the department on April 1, 2019.

Civilian Complaints: Law enforcement agencies should evaluate their civilian complaint process and align their complaint forms, where practical, with the best practices laid out in this report. The Board conducted a review of the complaint forms of the Wave 1 agencies to identify areas of opportunity to adopt additional best practices.

The report examines the civilian complaint data, including data on reported racial and identity profiling allegations submitted to the Department of Justice by all RIPA reporting agencies in 2018; the report then highlights the factors that impact the disparities in the number of reported complaints by each agency.

6 Recommendations for Community Members The 2020 Annual Report contains recommendations that advocates and community members can use to engage with law enforcement to improve policies, accountability, and enforcement measures. The Board hopes community members can take the model language and best practices delineated in the report to push law enforcement agencies to improve their policies and procedures. The Board also thanks members of the community for attending Board and subcommittee meetings and providing public comment.

The Board hopes community members will continue to engage with the Board regarding its work. Recommendations for Policymakers The Board hopes the California Legislature and local governments can increase funding to law enforcement agencies to implement RIPA by supporting not only the data collection itself, but also in supporting law enforcement’s evaluation of the collected data as well as the development of anti-bias training and policies.

To effectively fulfill their mandate under RIPA, law enforcement agencies must develop and further refine their data collection systems for stops, review and revise their policies and practices, and make other changes to personnel, supervision, and training. They cannot do so without additional funding and support. With respect to civilian complaints, the Board recommends that the Legislature amend Penal Code

section 148.6 by striking the language imposing criminal sanctions for filing a false complaint. By doing so, the Board hopes to resolve a conflict between state and federal law, as well as remove cautionary language that is potentially chilling to the filing of a civilian complaint. Findings Regarding RIPA Stop Data • Between July 1, 2018 and December 31, 2018, the eight largest agencies in California, referred to as Wave 1 agencies in this report, collected data on vehicle and pedestrian stops.

RIPA defines a stop as a detention and/or search by a peace officer. • Reporting agencies stopped over 1.8 million individuals during the stop data collection period. The California Highway Patrol conducted the most stops of all reporting agencies, which is unsurprising given the size and geographic jurisdiction of the agency and its primary mission with respect to highway safety.

7 • 95.3 percent of stops were officer-initiated, while 4.7 percent of stops were in response to a call for service, radio call, or dispatch. • Individuals perceived to be Hispanic (39.8%), White (33.2%), or Black (15.2%) comprised the majority of stopped individuals. 1,033,421, 57% 336,681, 19% 136,635, 8% 89,455, 5% 62,433, 3% 56,409, 3% 44,505, 3% 40,515, 2% Number of Stops by Agency California Highway Patrol Los Angeles Police Department Los Angeles County SheriffÕs Department San Diego Police Department San Bernardino County SheriffÕs Department San Francisco Police Department Riverside County SheriffÕs Department San Diego County SheriffÕs Department 39.8% 33.2% 15.2% 5.5% 4.4% 1.2% 0.6% 0.2% Percent of Stopped Individuals Hispanic White Black Asian Middle Eastern/South Asian Multiracial Pacific Islander Native American

8 2017 American Community Survey (ACS) 39.8% 33.2% 15.1% 5.5% 4.4% 1.2% 0.6% 0.2% 0.0% 41.4% 34.7% 6.3% 11.9% 1.8% 3.0% 0.3% 0.3% 0.3% Hispanic White Black Asian Middle Eastern/ South Asian Multiracial Pacific Islander Native American Other Stop Data ASC • The most commonly reported reason for a stop across all racial/ethnic groups was traffic violations, followed by reasonable suspicion.

A higher percentage of Black individuals were stopped for reasonable suspicion than any other racial identity group. • To provide context for the racial distribution of stopped individuals, the Board compared the distribution to two benchmark data sources: 1) the American Community Survey (ACS) and 2) the Statewide Integrated Traffic Records System (SWITRS).

Black individuals represented a higher proportion of stopped individuals than their relative proportion of the population in both benchmark datasets. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Total White Pacific Islander Native American Multiracial Middle Eastern/South Asian Hispanic Black Asian Traffic Violation Reasonable Suspicion Other Statewide Integrated Traffic Records System (SWITRS) 40.0% 33.6% 13.4% 6.1% 6.9% 38.4% 33.9% 9.7% 9.3% 8.8% Hispanic White Black Asian Other Stop Data (Traffic Violations Only) SWITRS

9 • The veil of darkness (VOD) method is a third benchmarking method used this year. The VOD analysis compares the proportion of individuals stopped during daylight hours to the proportion of individuals stopped when it is dark outside during the intertwilight period, i.e., the time of day that is dark during Standard Time, but light during Daylight Savings Time. Having a higher proportion of stops of individuals of a particular racial or ethnic group occur in the light, compared to White individuals, may be considered evidence of bias towards that group.

The VOD analysis of this year’s data indicated disparities in stops during light hours vs. dark hours for some racial and ethnic identity groups.

For example, individuals perceived to be Pacific Islander or Multiracial had a higher proportion of their stops occur during light hours than individuals perceived to be White. • Overall, 9.9 percent of stopped individuals were subject to a person or property search. • Officers searched Black individuals at a rate 2.9 times the rate at which they searched White individuals (18.7% vs. 6.5%). • Middle Eastern/South Asian individuals had the lowest search rate (2.8%). • Search yield rate analyses showed that, when officers searched individuals, contraband or evidence was generally found on White individuals at higher rates than individuals from all other groups. 50.3% 51.2% 49.5% 50.5% 46.5% 45.5% 43.0% 50.0% 49.7% 48.8% 50.5% 49.5% 53.5% 54.5% 57.0% 50.0% 0% 50% 100% Asian (9,428) Black (25,045) Hispanic (66,682) Middle Eastern/ South Asian (8,333) Multiracial (2,170) Native American

(389) Pacific Islander

(931) White (54,546) Inter-Twilight Stop Frequencies by Race/Ethnicity Light Dark 90.1% 93.5% 92.1% 88.9% 88.4% 97.2% 88.8% 81.3% 96.9% 9.9% 6.5% 7.9% 11.1% 11.6% 2.8% 11.2% 18.7% 3.1% Total White Pacific Islander Native American Multiracial Middle Eastern/South Asian Hispanic Black Asian Not Searched Searched 21.8% 22.5% 21.0% 18.8% 22.4% 19.3% 21.1% 24.3% Search Yield Rate Percent of Searched Individuals Asian Black Hispanic Middle Eastern/South Asian Multiracial Native American Pacific Islander White

10 • When examining search yield rates by the presumed level of discretion available to the officer in deciding to conduct a search, yield rates for racial/ethnic groups of color were lower than for White individuals for higher-discretion searches, i.e., searches for which the only basis for search was “consent given.” This was also true for most racial/ethnic groups of color when only examining lower discretion searches (searches in which the basis for search was incident to arrest, vehicle inventory, or search warrant), with the exception of Black and Multiracial individuals, who had higher yield rates than White individuals for lower discretion searches. • 60.3 percent of all individuals stopped were issued a citation and/or arrested.

Native American and Black individuals had the highest arrest rates and the lowest rates of citation. Middle Eastern/South Asian and Asian individuals had the highest citation rates and the lowest arrest rates. 57.6% 36.8% 48.7% 61.4% 45.9% 38.0% 48.5% 49.4% 11.4% 15.2% 13.6% 7.0% 13.5% 16.0% 13.9% 11.3% Asian Black Hispanic Middle Eastern/South Asian Multiracial Native American Pacific Islander White Percent of Individuals Citations Arrests

11 Findings Regarding Civilian Complaint Data There were 1,081 allegations of racial or identity profiling filed in 2018 with the 134 law enforcement agencies subject to RIPA. Of these, 78 percent of the complaints included allegations of racial or identity profiling. The following table shows the total number of civilian complaints reported in 2018 by Wave 1 agencies, the number of allegations of racial or identity profiling, and the number of sworn personnel each agency employed in 2018. There were notable disparities in the total complaints and racial and identity profiling allegations reported by agency.

The reasons for these disparities likely include: 1) lack of uniformity regarding what constitutes a “civilian complaint” and how to quantify and document complaints; 2) lack of uniformity regarding how to process civilian complaints; 3) varying accessibility and knowledge of an agency’s complaint process; 4) disparate accessibility for people with disabilities; and 5) the potential deterrent impact of Penal Code

section 148.6. Nationality, 37, 4% Gender, 51, 5% Religion, 15, 1% Age, 21, 2% Gender Identity Expression, 25, 2% Mental Disability, 30, 3% Sexual Orientation, 24, 2% Physical Disability , 34, 3% Race and Ethnicity, 844, 78% Total Racial and Identity Profiling Allegations Reported

12 Wave 1 Agency Complaints Reported and Number of Sworn Personnel Employed in 2018 Agency Total Complaints Reported Profiling Complaints Reported Sworn Personnel Los Angeles Police Department 1,907 274 (14%) 9,974 Los Angeles County Sheriff’s Department 986 67 (6.7%) 9,426 California Highway Patrol 287 35 (12%) 7,286 San Diego County Sheriff’s Department 9 1 (11%) 2,572 San Francisco Police Department 678 21 (3%) 2,306 San Bernardino County Sheriff’s Department 104 35 (33%) 2,018 Riverside County Sheriff’s Department 46 4 (9%) 1,795 San Diego Police Department 74 15 (20%) 1,731

13 Opening Letter from RIPA Board Co-Chairs Last year marked a major milestone for the Racial and Identity Profiling Act of 2015 (RIPA), the Racial and Identity Profiling Advisory Board (Board), and the State of California. In 2019, the California Department of Justice (Department) received its first set of stop data from the eight largest law enforcement agencies in the state (Wave 1 agencies). The Board has analyzed this data and incorporated the results into this year’s report.

Specifically, the Board reviewed comprehensive demographic data on all stops and searches reported by the California Highway Patrol, Los Angeles Police Department, Los Angeles County Sheriff’s Department, Riverside County Sheriff's Department, San Bernardino County Sheriff's Department, San Diego County Sheriff's Department, San Diego Police Department, and San Francisco Police Department. This first wave of data documented approximately 1.8 million police detentions and searches across California. This is only the beginning.

All California law enforcement agencies will begin reporting data on a rolling basis through 2023, generating public data on statewide stops and searches on an unprecedented scale. To understand the momentousness of this accomplishment, we must reflect on how this began and the work ahead needed to eliminate racial and identity profiling. In 2015, the California Legislature passed RIPA, groundbreaking legislation that requires all law enforcement agencies statewide to uniformly collect and report demographic data on all police stops and searches.

RIPA also mandated the creation of the Board, with the bold intention of eliminating racial and identity profiling in policing. In 2016, its inaugural year, the Board made recommendations to the Attorney General’s Office on its drafting of regulations to implement RIPA. Under this stop data program, reporting officers must collect data on the reason for each detention or search, as well as detailed demographic data, including the perceived race or ethnicity, gender, age, LGBT identity, disability, and limited English fluency of the person detained or searched.

Since its inception, the Board has engaged in a thorough study and examination of several civilian - facing aspects of law enforcement that relate to racial and identity profiling, including law enforcement training, civilian complaint processes, policies regarding racial and identity profiling and accountability, and policies regarding calls for service. In this year’s report as well as in previous ones, the Board has compiled comprehensive, evidence-based best practice recommendations and model policies.

Now that RIPA and the Board have been in effect for four years, what does the future hold and what are the next steps? The Board urges all law enforcement agencies to compare their own policies to the best practice recommendations offered by the Board. However, the Board’s recommendations are only a starting point; we encourage agencies to think about how they can strive to go beyond the Board’s recommendations.

We urge law enforcement agencies to work with and engage their home communities to develop policies and practices that advance equity and root out bias and harmful practices of racial profiling in all aspects of operations. Additionally, we urge law enforcement, advocates, and community members to reflect on and make use of the stop data reported for their home communities. We are hopeful that the stop data can serve as a starting point for meaningful collaboration and change, and look forward to supporting the community and law enforcement agencies in these endeavors.

14 We also strongly support increased funding for the implementation of RIPA. To date, the Board is unaware of any state funding allocated to local law enforcement agencies to implement these sweeping changes. The future will depend on fully funding the implementation of this important legislation which left unfunded, may soon hinder much of the work. Many agencies, especially small ones, are struggling from the lack of sufficient funding. We cannot let this legislation fail. Funding for this legislation must be a priority to ensure that this important work is done right.

Finally, we extend our sincere appreciation and gratitude to everyone who has been on this journey with the Board throughout the last several years. The work of the Board to help identify and eliminate racial and identity profiling cannot be done without the continued engagement of the community and the commitment of law enforcement. We would especially like to recognize members of the public, particularly individuals who have shared their experiences of racial profiling, who have been indispensable participants in the Board’s work.

We thank you for sharing your expertise, your time, your stories, and your pain with us over the years. We also thank law enforcement agencies around the state for embracing RIPA, sharing your implementation of this law, and ensuring complete and comprehensive data collection and reporting. We know this was no small feat and look forward to continued partnership with you in coming years. -Co-Chairs Sahar Durali and David Robinson

15 Introduction The Racial and Identity Profiling Act of 2015 (RIPA) created the Racial and Identity Profiling Advisory Board (Board), which is tasked with the ambitious charge of improving racial and identity sensitivity in law enforcement with the hope of eliminating bias in policing. 1 The Board is composed of 19 members representing a wide range of sectors and expertise, including civil and human rights, law enforcement, and academia.

The Board’s work is enhanced by the diverse perspectives and backgrounds of its members, as well as by the vibrant discourse brought to Board and subcommittee meetings by advocates, individuals impacted by racial profiling issues, members of the law enforcement community, and members of the public at large. Together, the Board and its stakeholders share the common goals of improving law enforcement-community relations, building trust, making policing more equitable, and striving to make all Californians feel respected and safe. These goals can be achieved through collaboration, transparency, and accountability.

Background Since its inception, the Board has engaged with diverse stakeholders who share the goal of eliminating racial and identity profiling. The Board has heard from the community at Board and subcommittee meetings, consulted with the Department, and collaborated with the Commission on Peace Officer Standards and Training (POST) on its trainings related to racial and identity profiling. The Board also produced and released two annual reports describing the ongoing efforts to assess and prevent racial and identity profiling in California.

These annual reports give the Board an opportunity to share detailed findings on the impact that race and identity may have in shaping law enforcement activities in California, as well as identifying best practices and policy recommendations to identify and eliminate racial and identity profiling. 2 To that end, RIPA requires each annual report to include: • An analysis of law enforcement data regarding stops made by officers and civilian complaints; • An analysis of law enforcement training on racial and identity differences discussed in Penal Code

section 13519.4; • A review and analysis of racial and identity profiling policies and practices across geographic areas in California; and • Evidence-based research on intentional and implicit biases that affect law enforcement stop, search, and seizure tactics. 3 1 Pen. Code, § 13519.4, subd. (j)(1). 2 Pen. Code, § 13519.4, subd. (j)(3)(E). 3 Pen. Code, § 13519.4, subd. (j)(3).

16 RIPA also requires POST to consult with the Board in developing its trainings on racial and identity differences to better educate law enforcement about unlawful profiling and bias.

In addition, RIPA mandates that: 4 • The majority of California’s law enforcement agencies (LEAs) collect information on stops made by their officers, and report this information to the Department; RIPA also tasked the Department with writing the regulations to implement this data collection, in consultation with the Board and other stakeholders; 5 • The stop data collected be made publicly available, except for the personal information of the person stopped and the unique identifying information of the reporting officer, which shall be protected from disclosure; and • Several changes to the civilian complaint data be reported to and published by the Department. 6 Type of Data Collected for Each Stop The data collected about each stop includes three categories of information: 1) information about the stop itself, 2) information perceived by the officer about the person stopped, and 3) information about the officer making the stop.

Table 1, below, spells out in more detail the information the officer must report in each of those three categories. 7 Table 1: Officer Reporting Requirements Information Regarding Stop 1. Date, Time, and Duration 2. Location 3. Reason for Stop 4. Was Stop in Response to Call for Service? 5. Actions Taken During Stop 6. Contraband or Evidence Discovered 7. Property Seized 8. Result of Stop 4 Assem. Bill No. 1518 (2017-2018 Reg. Sess.) § 1-2. 5 Gov. Code, § 12525.2, subds. (a), (e). 6 Pen.

Code, § 13012. 7 For more information on the specific data collected, please see State of California Department of Justice Office of the Attorney General. (2017). AB 953: Template Based on the Final Regulations. Available at https://oag.ca.gov/sites/all/files/agweb/pdfs/ripa/regs-template.pdf.

17 Information Regarding Officer’s Perception of Person Stopped 1. Perceived Race or Ethnicity 2. Perceived Age 3. Perceived Gender 4. Perceived to be LGBT 5. Limited or No English Fluency 6. Perceived or Known Disability Information Regarding Officer 1. Officer’s Identification Number 2. Years of Experience 3. Type of Assignment When reporting this information for each stop, the reporting officer selects from a standardized list of responses. These drop-down menus streamline the reporting process and, importantly, ensure that the data that is collected is uniform across all agencies.

Separate from and in addition to these drop - down menus, officers are further required to complete an explanatory field (of no more than 250 characters) providing in their own words the Reason for Stop and Basis for the Search (if one is conducted). Methods of Submitting Data to the Statewide Repository In the spirit of facilitating a large and diverse array of individual law enforcement agencies to successfully implement the stop data requirements, the size of an agency determines when it is required to begin collecting and submitting data to the Department.

Stop data collection for the eight largest agencies in the state began on July 1, 2018. These agencies have informally been termed the "Wave 1" agencies due to the rolling nature of the stop data collection time line. Accordingly, the next set of agencies to begin data collection are thus termed "Wave 2" and so on until the final group, "Wave 4" begins collecting the data (Table 2). Additionally, the data submission regulations provide agencies with three methods to submit data.

These three methods of submitting data to the statewide repository are: 1) a DOJ-hosted Web Application, 2) Web Services, and 3) Secure File Transfer Protocol. The Department developed these three submission methods to provide flexibility to meet the needs of an agency’s local infrastructure. Importantly, the data standards for each of these methods are the same; each method utilizes standard fields and validation checks, which will be discussed in the next

section of this chapter. Table 3 details the submission methods that Wave 1 agencies are currently using.

18 Table 2: Collection and Reporting Deadlines by “Wave” Reporting Wave Size of Agency Data Collection Begins Data Must be Reported to DOJ Approximate Number of Agencies 1 1,000+ July 1, 2018 April 1, 2019 8 2 667-999 Jan. 1, 2019 April 1, 2020 7 3 334-666 Jan. 1, 2021 April 1, 2022 10 4 1-333 Jan. 1, 2022 April 1, 2023 400+ Table 3: Wave 1 Agency Submission Methods Agency Type of Data Submission California Highway Patrol Web Services Los Angeles Police Department Secure File Transfer Protocol Los Angeles Sheriff’s Office Web Services Riverside Sheriff’s Office Secure File Transfer Protocol* San Bernardino Sheriff’s Office Web Services* San Diego Police Department Web Services* San Diego Sheriff’s Office Web Services* San Francisco Police Department DOJ-hosted Web Application *These agencies are using a locally installed copy of an application developed by the San Diego Sheriff’s Office and submitting data to the Department through Web Services or Secure File Transfer Protocol.

All records submitted to the Department are stored in a statewide repository called the Stop Data Collection System (SDCS). The SDCS uses a series of rules and user permissions to protect the quality and integrity of the data. Some of these rules are listed below. • Reported data must be complete and must follow uniform standards.

19 • Access to stop records is restricted. • A specified error resolution process must be followed. • Once submitted, perception data (i.e., perceived demographic data about the person stopped) is locked and cannot be changed by the officer or agency. • Transactions are stored in system audit logs.

20 Analysis of Wave 1 Stop Data: July 1, 2018 – December 31, 2018 In the first wave of reporting (Wave 1), the eight largest law enforcement agencies in California collected data about stops conducted from July 1, 2018 to December 31, 2018. Officers collected data on over 1.8 million stops. RIPA defines stops as a detention and/or search of an individual.

The records include data on the demographic information of the stopped individuals as perceived by the officer. 8 The demographic information includes race/ethnicity, gender, LGBT identity, age, disability status, and English fluency, as well as a range of descriptive information designed to provide context for the reason for the stop, what occurred during the stop, and the resolution of the stop. The purpose of collecting this data is to attempt to systematically document and analyze detentions and/or searches of all individuals to determine whether disparities occur depending on race and/or identity.

For this year’s Report, the Board presents stop data analyses focused on the race/ethnicity of the person stopped. 9 Addressing racial profiling was a driving force in enacting RIPA. The different types of analyses used in this year’s report were included after significant discussion in Board subcommittee meetings, full Board meetings, and input by members of the public. The analyses were conducted to answer the question of whether the perceived race/ethnicity of a stopped individual plays a role in whether they are stopped and/or in the actions an officer takes during a stop.

In future reports, the Board intends to focus its analyses on other demographic characteristics of this rich dataset. The decisions made, or actions taken, by the officer can be broken into two types: “pre-stop” and “post-stop.” “Pre-stop” decisions refer to an officer’s decision to stop an individual in the first place. Our pre-stop inquiry analyzes the number of stops of members of the various perceived racial and ethnic groups. This analysis is important because it gives us the ability to examine whether different groups are stopped at different rates, which might indicate that potential bias is present.

Because of the difficulty in establishing “benchmarks” – meaning how people would behave in an unbiased world – we have employed several established methodologies to analyze Wave 1 stop data and consider whether the data indicates evidence of racial bias in officers’ pre-stop decisions. First, we compared the demographics of persons stopped to two datasets intended to approximate the general population of residents and drivers, respectively, within the jurisdictions of the Wave 1 LEAs.

Specifically, the two datasets we used are (1) the weighted residential population data from the American Community Survey (ACS) (to obtain a resident population benchmark) and (2) the not-at - fault vehicle collision data from a database maintained by the California Highway Patrol (CHP) (to obtain a driver population benchmark). In addition to these population comparisons, we also analyzed the Wave 1 stop data using the veil of darkness methodology.

As discussed in prior Board Reports 10 , this methodology compares stop frequencies during daylight hours, when it could be more likely for an 8 RIPA requires that the demographic information be recorded based upon the officer’s perception, meaning that an officer should not use information from documents or ask individuals directly about their demographic information when completing the stop data form.

However, nothing in RIPA prohibits an officer from obtaining such information within the course and scope of their lawful duties. 9 Although the data collected contains officers’ perception of various identity groups and other demographics, this

chapter focuses only on their perceptions of race/ethnicity. See the Technical Report for analyses of the data from more identity groups, as well as disaggregated statistical information for each agency. 10 See page 23 of the 2019 RIPA Board report for an explanation of the Veil of Darkness methodology.

21 officer to perceive race, to stop frequencies at night, when it could be more difficult for an officer to perceive race before stopping someone. Another way to get around the issue of benchmarks is to examine post-stop decisions made by the officer. Conducting a search, for example, is conditional on already having stopped an individual. Thus, we can be more confident in comparing the rates at which different identity groups are searched because we know for certain in calculating these ratios what the denominator is: people who have already been stopped.

Searches are worth exploring for another reason – they come with their own outcome, namely whether or not the search resulted in, or yielded, the recovery of any contraband or evidence. The yield rate is a measure of the “efficacy of the search.” If the success of searches (i.e. the search yielding contraband) differs across different identity groups, it could be indicative of officers having higher or lower thresholds for searching some groups relative to others and it allows for a stronger case that bias may be a driving factor for searching an individual, as opposed to some other variable like crime rate.

We also examined the enforcement rates by race and ethnicity, meaning the rate by which an individual who was stopped is given a citation and/or arrested as a result of the stop. To introduce these methodologies, we first set forth the data regarding the perceived racial and ethnic identity demographics of individuals stopped by the Wave 1 agencies. We then present the results by race and ethnicity for the other elements of the stop, beginning with the reported conditions underlying an officer’s decision to initiate the stop, such as the primary reason for the stop and the circumstances leading to the stop.

We then apply the methods discussed above in an effort to see whether the data demonstrate evidence of potential bias in officer pre-stop and post-stop decisions.

Summary of Main Results The Board’s analysis of Wave 1 data suggests that officers from these agencies stopped each racial or ethnic group at frequencies that differed from both the weighted ACS residential population estimates and the CHP driver information. These differences were most pronounced for Black individuals, who composed a significantly larger proportion of the individuals who were stopped than they did in either of the two comparison datasets (i.e., the weighted residential population or the driver population).

The opposite was true for Asian individuals; Asian individuals represented a smaller proportion of the individuals officers stopped than they did in the comparison datasets. Using the veil of darkness method, the analysis of Wave 1 data shows that stop frequencies differed between racial or ethnic groups based on the level of presumed visibility given the time of day. Individuals perceived as Pacific Islander had the highest proportion of their stops occur in the light. Officers stopped White individuals almost equally in the light and dark.

A higher proportion of stops of Black individuals were in the dark hours as opposed to the light hours. As for post-stop outcomes, using the yield rate analysis, the data showed that certain groups of people of color may experience higher degrees of scrutiny by law enforcement compared to White individuals, particularly with respect to search activity. For example, officers searched Hispanic, Black, Native American, and Multiracial individuals at a higher rate than they “Perceived” Identity All racial and ethnic groups referenced in this

section are based on the reporting officer’s perception of the race or ethnicity of stopped individuals. Officers may perceive individuals differently than how the individuals self-identify.

22 searched White individuals, despite discovering contraband on members of these groups less frequently when searched. Finally, Wave 1 data shows that the outcome of an enforcement action varied by racial or ethnic group, with Native American and Black individuals having the highest arrest rates and the lowest rates of citation. Middle Eastern/South Asian and Asian individuals had the highest citation rates and among the lowest arrest rates.

As discussed in prior Board Reports, any one methodology that aims to evaluate bias suffers from some limitations, suggesting that it is often useful to employ multiple methodologies. Therefore, the use of certain methodologies this year should not be interpreted to mean that the Board will limit itself to these methodologies in future Board reports.

Indeed, to gain a fuller understanding of the issues underlying the Board’s goals to develop policy recommendations based upon fact-based evidence, the Board welcomes suggestions from all stakeholders – including academics, law enforcement and the community – about supplemental analysis or alternative methods to examine the stop data in the future. Stop Demographics Wave 1 agencies submitted data regarding stops of more than 1.8 million individuals. RIPA requires officers to record a person’s identity based upon the officer’s perception.

Officers may not ask individuals to self-identify their identity group when completing the stop data form. Because of this, the data reflects what the officer perceived the individual’s identity group to be. Of the approximately 1.8 million reported stops, individuals perceived by officers as Hispanic (39.8%) constituted the highest proportion of stopped individuals, followed by White (33.2%), Black (15.1 %), Asian (5.5%), Middle Eastern/South Asian (4.4%) and all other groups (2%; includes Pacific Islander, Native American, and Multiracial 11 individuals; see Figure 1). Figure 1.

Race/Ethnicity Distribution of Stopped Individuals 11 Officers can select multiple perceived identity categories per stopped individual, if appropriate. For example, an officer could perceive a person as being both White and Black. In our analyses, we categorize all such persons as Multiracial.

23 Decision to Stop Reason for stop: Across all racial and ethnic groups, the most common primary reason officers reported for initiating a stop was a traffic violation, which includes moving and non-moving violations and equipment violations (84.8 percent of all stops; see Figure 2). 12 Approximately 85 percent of stops of White and Hispanic individuals were stopped for traffic violations, while 76 percent of stops of Black individuals were for traffic violations.

For Asian and Middle Eastern/South Asian individuals, traffic violation was the reason given for initiating 93.6 percent and 94.9 percent of the stops, respectively. The second most common reported reason for stop was reasonable suspicion of criminal activity (11.4 percent of all stops), referred to as “reasonable suspicion” hereafter (see Figure 2). 13 Black individuals were stopped for reasonable suspicion in 19.5 percent of their stops, while 10.8 percent of stops of White individuals and 10.6 percent of stops of Hispanic individuals were for reasonable suspicion.

Only 3.6 percent of Middle Eastern/South Asian individuals were stopped for reasonable suspicion. All other reasons for stop constituted less than 4 percent of the data. 14 Figure 2. Primary Reason for Stop by Race/Ethnicity 12 See Technical Report

Section 1 Table 2.3.2 for the racial/ethnic breakdown by traffic violation subtype. 13 The Board understands that an officer may initiate contact with a person as part of his/her community caretaking function without suspecting that the person is engaged in criminal activity. However, officers currently must record community caretaking stops under the “reasonable suspicion” reason for stop. Officers indicated that 3.5 percent of stops initiated due to reasonable suspicion were for community caretaking purposes. This constituted only 0.4 percent of stops overall.

Since a percentage this small would not be viewable in Figure 2, community caretaking stops were not separated out from the reasonable suspicion stops. 14 Other reasons for stop included mandatory supervision (0.6 %), warrants (0.7 %), truancy (0.3 %), possible violations of the Education Code (<0.1 %), to determine whether student violated school policy (>0.1 %), or consensual encounters that resulted in a search (2.2 %). We aggregated these reasons for stop into the category labeled “Other” in Figure 2.

24 Stop circumstance: Stops take place within a broader context. Stops can be initiated either by an officer (“officer-initiated stop”) or in response to a call for service, radio call, or dispatch (“call for service”). 15 A call for service is not a reason for a stop. Whether or not a person was stopped in response to a call for service provides additional information that is helpful to contextualize stop data. Approximately 5 percent of all stopped individuals were reportedly stopped in response to a call for service, as opposed to a stop initiated by an officer (see Table 4).

This percentage varied by race/ethnicity, but no more than 8 percent of stopped individuals from any racial or ethnic group were stopped in response to calls for service. The Wave 1 data also shows that individuals of different racial or ethnic groups varied in their stop rates for officer-initiated stops and calls for service (see Figure 3). 16 Table 4.

Stop Circumstance by Race/Ethnicity Stop Circumstance Race/Ethnicity of Stopped Individual White Hispanic Black Asian Middle Eastern/ South Asian Pacific Islander Native American Multiracial Officer Initiated 568,900 95.2 % 686,017 95.8 % 251,291 92.7 % 96,734 97.3 % 78,171 97.8 % 10,047 94.9 % 3,665 94.3 % 19,851 93.5 % Call for Service 28,865 4.8 % 30,012 4.2 % 19,897 7.3 % 2,713 2.7 % 1,746 2.2 % 537 5.1 % 220 5.7 % 1,388 6.5 % Total 597,765 100 % 716,029 100 % 271,188 100 % 99,447 100 % 79,917 100 % 10,584 100 % 3,885 100 % 21,239 100 % 15 Officers are required to indicate if a stop was in response to a call for service (also known as a radio call or dispatch).

An interaction that occurs when an officer responds to a call for service is only reportable if the interaction meets the definition of “stop” for data collection purposes, meaning any detention by a peace officer of a person or any peace officer interaction with a person in which the officer conducts a search. 11 CCR § 999.224. This information is collected independently from the reason for a stop. 16 “Officer-initiated stops” are defined as any stop where an officer did not indicate that the stop of an individual was made in response to a call for service, radio call, or dispatch.

Key Terms: Stop Circumstance Call for service: when an officer indicates that the stop of an individual was made in response to a call for service, radio call, or dispatch. Officer-initiated stop: when an officer does not indicate that the stop of an individual was made in response to a call for service, radio call, or dispatch.

25 Figure 3. Stop Circumstance by Race/Ethnicity Comparisons to Reference Data As noted above, several methodologies can assist researchers in analyzing stop data to determine the existence of racial bias. As will be discussed below, there are notable concerns with relying entirely on one comparison method. Accordingly, our analysis instead presents the results of three separate methods designed to provide reference points from which to compare the stop frequencies by racial/ethnic group in these data.

These methods contextualize stop frequencies using: (1) residential population data; (2) vehicle collision data; and (3) light condition data. Residential population data: We used residential population estimate data from the 2017 American Community Survey (ACS) to provide a contextual residential benchmark for the race/ethnicity of individuals stopped by Wave 1 agencies during the data collection period. 17 The United States Census Bureau administers the ACS annually.

Our weighting methodology made the ACS data more reflective of the areas within the jurisdictions of Wave 1 agencies, rather than the state or country as a whole. 18 Figure 4 displays the racial/ethnic distribution of: (1) stopped individuals from the 2018 data; and (2) estimated residential population of the areas within the jurisdiction of Wave 1 agencies. 19 Because the CHP conducted more than half of the stops during the data collection period, we also provide the residential population data table excluding CHP data in the Technical Report. 20 17 At the time we sourced the ACS data (October 2019), 2017 was the most recent year available. 18 For a description of the weighting scheme, see the ACS table notes in the Technical Report

Section 1 Subsection 2. 19 See Table 2.13.2 in the Technical Report for a weighted ACS breakdown by race for all agencies without California Highway Patrol data. 20 See Table 2.13.2 in the Technical Report for a weighted ACS breakdown by race for all agencies without CHP data. In general, when exempting CHP from analysis, the disparities between stop frequencies and residential population representation increase for Asian, Black, Hispanic, and White individuals. Of these racial/ethnic groups, Black individuals represented a larger proportion of stopped individuals than their share of the residential population data. The opposite was true for Asian, Hispanic, and White individuals.

26 Figure 4. Residential Population Comparison to Stop Data Considerations for and limitations of residential population data: Like all approaches for examining law enforcement stop data, there are important considerations and limitations to recognize when using residential population data within this context. To start, RIPA stop data regulations and the ACS categorize racial/ethnic groups differently (e.g., RIPA regulations explicitly include Israeli individuals in the Middle Eastern/South Asian group, but the ACS does not have an Israeli ethnic category).

ACS data also have a category for “Other,” which we could not map to any RIPA race/ethnicity group. Additionally, race/ethnicity information collected for RIPA is based on officer perception, while ACS respondents self-identify their own race/ethnicity. This distinction reflects a difference in purpose between the two databases. The objective of the stop data is to approach the problem of racial and identity profiling, which is why the agencies collect the officer’s perception of race/ethnicity.

The ACS, on the other hand, is to provide an accurate representation of information regarding community residents (i.e. social, economic, housing, and demographic characteristics). The RIPA and ACS data collection also occurred during different years (the second half of 2018, and 2017, respectively). The ACS data comparison has other limitations. ACS contains information collected from residents within particular areas.

However, officers often stop individuals who are not residents of the areas where the stops take place, but rather are in those areas for other reasons (e.g., going to work, going shopping, visiting friends/family, etc.). Jurisdictions likely vary in the proportion of non-residents they stop, but the stop data does not contain information regarding a person’s residence. 21 Moreover, 21 Missouri is an example of a state that is collecting this information, to an extent.

The Missouri Attorney General’s Office added data collection procedures to collect information on the residency of stopped individuals for vehicle stops in 2018. This information is available in Appendix C of the 2018 Vehicle Stops Report, available

27 some locations tend to have large-scale events (e.g., concerts, parades, conferences, etc.), are tourist destinations, or have large populations of individuals experiencing homelessness, all of which may present considerations that are even more difficult to account for. Furthermore, officers may concentrate their patrol efforts in certain areas and thus may not have equal probabilities of encountering residents of all areas in their jurisdiction. Additionally, ACS data may not accurately count certain groups that may be less inclined to respond to surveys (e.g. homeless or undocumented individuals).

For all of these reasons, the demographics (perceived or actual) of the population of people stopped by law enforcement may not always match the self-reported demographics of residential populations at the city, county, or state level. 22 Vehicle collision data: Another type of data that some studies have employed to provide context to stop data is vehicle collision data. Accordingly, as an alternative set of comparison data to ACS, we also provide vehicle collision data as context for the RIPA stop data. California law enforcement agencies submit data gathered from collision scenes to the CHP.

The CHP stores these data in a database called the Statewide Integrated Traffic Records System (SWITRS). 23 We obtained a dataset containing all reported collision records from reporting agencies for calendar year 2018.

We limited the data from SWITRS to not-at-fault parties from collisions reported by Wave 1 agencies, with the idea that this group of drivers is selected somewhat randomly because another driver struck them with their vehicle. 24 This is important because the purpose of the data we selected is to serve as a benchmark of drivers in general, not just the less-skilled or inattentive drivers that may tend to be at fault more frequently. We then employed a similar method used for the ACS data to make the SWITRS data more reflective of stop activity that occurred in the jurisdictions of Wave 1 agencies.

Figure 5 displays the distribution of the perceived race/ethnicity of (1) individuals stopped for traffic violations from the 2018 RIPA data and (2) the weighted not-at-fault party SWITRS data reported by Wave 1 agencies in 2018. This figure is specific to traffic violations, which constitute a majority (84.8%) of stops in the RIPA stop data (see Figure 2). As we did with the residential population data, we also provide these data with CHP excluded in the Technical Report. 25 at https://ago.mo.gov/docs/default-source/public-safety/2018appendixc.pdf?sfvrsn=2.

The experiences of Missouri law enforcement agencies may not directly compare to those of California law enforcement agencies, however. 22 For more information on this issue, see previous RIPA reports or the following publication by the United States Community Oriented Policing Services, available at https://ric-zai-inc.com/ric.php?page=detail&id=COPS-P044. 23 See https://www.chp.ca.gov/programs-services/services-information/switrs-internet-statewide-integrated - traffic-records-system for more information on SWITRS. 24 Not all studies that employ vehicle collision data utilize only the not-at-fault party data (e.g., Withrow, Brian L., and Howard Williams. “Proposing a Benchmark Based on Vehicle Collision Data in Racial Profiling Research.” Criminal Justice Review 40, no. 4 (2015): 449–69. https://doi.org/10.1177/0734016815591819.) 25 See Table 2.13.4 in the Technical Report for a weighted SWITRS breakdown for race/ethnicity without CHP data.

In general, when CHP data is excluded from analysis, the disparity between the stop data and the vehicle collision data increased for Black and White individuals, as well as the group categorized as “Other” for this analysis (Middle Eastern/South Asian, Multiracial, Native American, and Pacific Islander individuals). Of the three groups where disparities increased by exempting CHP, Black individuals comprised a greater proportion of those stopped relative to their representation in the collision data. The opposite was true of White and “Other” individuals.

28 Figure 5. Vehicle Collision Data Comparison to Stop Data Considerations for and limitations of vehicle collision data: As with residential population data, there are important caveats about making comparisons between RIPA and SWITRS data. First, SWITRS collects race/ethnicity information for fewer groups than are present in the RIPA regulations.

As a result, some RIPA race/ethnic groups were aggregated into an “Other” category for Figure 5. 26 Second, officers may collect race/ethnicity information differently between the two datasets; RIPA relies solely on officer-perception data, while officers may enter the race/ethnicity data in SWITRS after examining documentation or having an individual self-identify. Third, there is no specific data element to differentiate motorists from pedestrians in the RIPA dataset; the closest within the RIPA data is to examine stops that officers indicated they initiated for traffic violations.

However, several Vehicle Codes regulate pedestrian behavior; this means that some individuals stopped for traffic (e.g. Vehicle Code) violations could be pedestrians. Fourth, although there is a variable that indicates what party was at fault in the SWITRS database, it is possible that officers are incorrect in determining which party 26 In this analysis, the “Other” category consists of Middle Eastern/South Asian, Multiracial, Native American, and Pacific Islander individuals. SWITRS Quick -Reference Limitations 1. RIPA stop data collection and SWITRS categorize racial/ethnic groups differently. 2.

RIPA data does not expressly identify drivers; rather, it identifies persons stopped for traffic violations. 3. Officers may be incorrect in determining which party was at fault, in some cases. 4. Identity groups could differ in their likelihood of being captured in the SWITRS data. 5. SWITRS data collection policies are not uniform across the entire state.

29 was at fault when entering the data in some cases. Fifth, the likelihood of becoming a not-at-fault party to a vehicle collision could differ amongst identity groups in some areas. Sixth, not all agencies in the state respond to collisions where there were no injuries and not all agencies determine which party was at fault for the collision, so not all collisions are reflected in the dataset. 27 Lastly, fewer empirical studies employ this type of data than residential population data; in part, this is because it is generally harder to access than residential population data made readily available by the U.S.

Census Bureau, meaning that there is less known regarding other potential issues about this benchmark. Light condition data: The proportion of stops represented by different racial or ethnic groups varied by time of day (see Figure 6). 28 White individuals composed a higher percentage of stops during daylight hours, as compared to evening hours when there was less light out. Conversely, Black and Hispanic individuals composed a higher relative percentage of stops during evening hours than daylight hours. 29 Hourly stop shares for Asian persons were relatively consistent over time.

These data could indicate that light conditions may affect the likelihood of being stopped differently by race/ethnicity. Figure 6. Stop Distribution by Race/Ethnicity by Hour of the Day To more directly test whether light conditions affect stop frequencies, the Board adopted a method introduced by two researchers working for the RAND Corporation on a study of Oakland Police 27 All Wave 1 agencies reported some parties in their 2018 SWITRS data to be at-fault.

This limitation of SWITRS data may be more relevant in future years when more agencies are included in the analyses. 28 Middle Eastern/South Asian, Multiracial, Native American, and Pacific Islander individuals are grouped into the “Other” category in this figure. 29 Hispanic and White individuals were stopped in the highest proportions at all hours of the day.

30 Department vehicle stop data. 30 These researchers suggested that differences in stop frequencies by race/ethnicity could be contextualized using civil twilight data. 31 This approach, often referred to as the “veil of darkness” (VOD), hypothesizes that if officers target some individuals for stops more than others based on their race, evidence of profiling should be most apparent during daylight when the race of drivers is presumably most visible.

Conversely, if race were more difficult to see in darkness, then officers would be less able to rely on race as a factor in making decisions about whom they stop during the night. Since the original study that established the VOD approach, many other studies have adopted variations of this framework to analyze stop data. RIPA Board’s decision to include VOD methodology: The inclusion of the VOD test was a topic of robust discussion at the November 20, 2019 Board meeting.

Some members of the RIPA Board expressed concerns about the VOD methodology while other members believed it was beneficial to include this analysis. Some Board Members presented the following arguments for the exclusion of the VOD analysis: • VOD is based only on traffic stops during a certain period in the day. • The CHP stop data makes up more than half of the stops analyzed in the VOD test and the nature of their stops are categorically different than those of other agencies.

First, the number of traffic violation stops varied widely across Wave 1 agencies: for example, they made up 98.5 percent of the CHP’s stops, but only 42.5 percent for San Diego Police Department’s stops. Second, the CHP noted that it was more difficult to perceive the identity of people stopped on the highway with or without daylight.

In response, some Board members believed the test was an unfavorable method for use on data collected for stops on highways. • The Board believed that this methodology did not adequately address other limitations such as lighting from street lights in urban areas; this was of additional concern given that the agencies that submitted data in 2018 were primarily ones that police urban areas. • Several published studies have shown that with the loss of light during daylight savings time there is an increase in crime; this might alter the behavior of law enforcement officers and it may interact with race in a complex way. • Compared to other methods utilized in this report, the Board believed that the VOD is excessively technical and, therefore, requires a disproportionate amount of explanation to communicate how the analysis was performed.

The report gives the residential data one page of analysis, the collision data one page of analysis, and the VOD five pages of analysis. 30 Grogger & Ridgeway, Testing for Racial Profiling in Traffic Stops from Behind a Veil of Darkness

(2006) RAND Corporation. 31 Civil twilight is defined as the illumination level sufficient for most ordinary outdoor activities to be done without artificial lighting before sunrise or after sunset. Therefore, it is dark outside when civil twilight ends; civil twilight ends when the sun is six degrees below the horizon.

31 • Given the complicated framework underlying the VOD analysis, the subtleties of results produced by these methods are difficult to interpret and may lead to confusion. The Board was concerned that it may seem that it was providing conflicting results to the public.

Other Board members made the following arguments for inclusion of the VOD analysis: • In the 2019 Report, the Board identified VOD as one of several methodologies that might be used in analyzing the data and excluding the methodology now that the analysis had been completed might signal a lack of transparency to some stakeholders. • This methodology has a research base, including articles published in academic publications, such as the Journal of the American Statistical Association. • There is a desire to present the results from multiple analytical methods.

This will allow for judgments to be made about the appropriateness of each methodology for agencies to analyze their data. • There is an interest in seeing if it will be possible to draw comparisons between the VOD analyses in this year’s report to those in the future when a larger dataset will be available. After the discussion, a motion was made to exclude the VOD analysis pending further review by the Stop Data Subcommittee, given the concerns with whether the VOD test had validity. The Board vote was evenly divided (five ayes, five nays, one abstention) and thus the motion to remove the VOD test did not pass.

The Board is including the VOD analysis in this year’s report with the hope that it will receive feedback from the community, academics, and law enforcement with respect to the efficacy of using this type of analysis in the future. Certainly, the Board has a strong interest in continuing to pursue multiple different analytical methods that will be useful to both the public and law enforcement moving forward.

Accordingly, the Board requested that the Stop Data Subcommittee continue to review VOD and any other methods of which it becomes aware and to make recommendations to the full Board with respect to methodologies to include in future reports. Although the VOD methodology has its own limitations, it avoids issues that surround population - based benchmarking. Instead, it compares the proportion of stopped individuals of a given race during daylight to the group’s proportion during dark hours. Thus, we employ the VOD approach as one of the multiple comparative approaches in this report to analyze the stop data.

Veil of Darkness methodology: The VOD technique examines stops that occur during a standardized inter-twilight period, or the time of day that is dark during Standard Time but light during Daylight Savings Time. By limiting the analysis to only those stops that occurred during this period, frequency comparisons are less susceptible to factors that vary by time of day (e.g ... commuting patterns).

To identify the inter-twilight period for the 2018 data, we sourced civil twilight times for each stop date and location using the United States Naval Observatory database. 32 We bounded the inter-twilight period using the earliest and latest instances of civil twilight for each location across the entire reporting period (approximately 4:54 pm to 9:30 pm). As shown in Figure 7, stops that occurred between the earliest end of civil twilight and the latest end of civil twilight would be included in the analyses. The blue line represents the end of civil twilight for a given day.

Stops that occurred with 32 This information is sensitive to location. Civil twilight can vary by over an hour on the same day across the state.

32 sunlight fall under the blue line, while those without sunlight occurred above the blue line. The large dip in the trajectory of the blue line on November 4 th is when the time switched from Daylight Savings Time back to Standard Time. Figure 7: Inter-Twilight Period Example Using 2018 Data for San Francisco, CA Only officer-initiated stops for traffic violations were included in this analysis for several reasons. First, many studies that employ a VOD framework utilize vehicle stop data only; traffic violations are the closest proxy to vehicle stops found in RIPA stop data.

Second, the assumptions underlying VOD are most likely to hold true for stops made outdoors, for people who are obscured by their vehicle, and for stops where officers are not called to the scene; these criteria are truer of stops made for traffic violations than those made for other reasons, including reasonable suspicion.

It is important to note that stops made for reasonable suspicion may often be more discretionary than those made for traffic violations, and may therefore be more likely to reveal instances of racial profiling; however, these stops are more likely to introduce additional confounding factors that violate the assumptions of VOD. Accordingly, reasonable suspicion stops are included in the analyses provided in other sections of this report.

33 Considerations and limitations of the VOD: The VOD approach was developed to address limitations of benchmarking comparisons; however, this does not mean that the VOD is without limitations of its own. To start, even under dark outdoor conditions with no artificial light, it is likely that some officers are able to perceive the race of individuals from close distances. Additionally, many patrol areas have some artificial light (e.g. streetlights, store signage, porch lights, etc.) that reduces the degree to which darkness may hinder their ability to perceive race.

There may also be certain types of violations (e.g. equipment violations) that some racial groups may have different propensities to commit due to economic or other reasons, which can be differently visible depending on whether it is light or dark outside. 33 Drivers belonging to some identity groups may also change their driving behavior based on the perceived likelihood of officers being able to correctly perceive their identity group membership. 34 Separate from the issue of lighting conditions is the potential issue that seasonal differences in driving patterns of certain groups could also influence the racial composition of drivers on roadways.

The VOD test also only examines data from within the inter - twilight period, meaning that obtaining large sample sizes for smaller racial groups (e.g. Native American persons) requires many reporting agencies or a dataset that contains more historical data than the RIPA dataset does currently. The VOD is also a test best fit for vehicle stop data, but RIPA data do not explicitly differentiate vehicle stops from pedestrian stops; therefore, analysts must narrow the data using an approximate method by examining traffic violations.

Lastly, there may be observable proxies for race (e.g., the make and model of the vehicle, the location of the stop, etc.) that officers could utilize to guess the race of drivers that could affect the assumptions of the test. Stop frequencies by race and sunlight availability: Across the reporting period, there was a near 50/50 split between the proportion of individuals stopped during the inter-twilight period under light (50.1%) and dark conditions (49.9%). White persons had the closest stop distribution to a 50/50 split.

Asian, Middle Eastern/South Asian, and Black individuals had slightly more of their members stopped under dark conditions than light within the inter-twilight period (50.3% - 51.2%). Pacific Islander individuals 33 Ritter, Joseph A. “How Do Police Use Race in Traffic Stops and Searches?

Tests Based on Observability of Race.” Journal of Economic Behavior & Organization 135 (2017): 82–98. https://doi.org/10.1016/j.jebo.2017.02.005. 34 Kalinowski, Jesse, Ross, Stephen L. & Ross, Matthew B. “Endogenous Driving Behavior in Veil of Darkness Tests for Racial Profiling.” Working Paper, Human Capital and Economic Opportunity Global Working Group, The University of Chicago, February 2017. VOD Quick -Reference Limitations 1. Reduced visibility under darker conditions does not mean no visibility, so officers may still be able to perceive race prior to initiating stops. 2.

The likelihood of some identity groups to commit certain offenses or be stopped for certain offenses could differ across lighting conditions. 3. Seasonal differences in driving patterns of certain identity groups could also influence the identity group composition of drivers. 4. The method only examines data from a set period of time and for a single type of stop (traffic violations). 5. Officers could use observable proxies to guess the race of drivers.

34 had the highest proportion of their members stopped in the light (57.0%), followed by Native American, Multiracial, and Hispanic individuals. Under the assumptions of VOD, having a higher proportion of a group stopped under light conditions may be considered as evidence of bias towards that group. Figure 8 displays the proportion of each race/ethnicity group stopped under each condition.

Compared to White individuals, Multiracial, and Pacific Islander individuals were more likely to be stopped in the light, while Black individuals were more likely to be stopped in the dark. 35 Given that CHP made over half the stops during the data collection period, and that most of the stops that CHP made were for traffic violations, we also conducted this analysis without CHP data and provide the table in the Technical Report. 36 Figure 8: Inter-Twilight Stop Frequencies by Race/Ethnicity Post-Stop Outcomes Search rates: Conducting a search of a person or their property was the most common reportable action officers took during a stop.

Overall, officers conducted a search of a person or their property in 9.9 percent (n = 178,975) of the stops reported. 37 Figure 9 shows the percentage by race/ethnicity of all individuals who were subjected to a search of their person and/or property. 38 The racial/ethnic 35 We used logistic regression and the same model specification as Grogger & Ridgeway, 2006. For detailed information regarding the model specifications and results, see Table 2.14.3 of

Section 1, Subsection 2 in the Technical Report. Tables 2.14.5 through 2.14.6 also display alternative VOD analyses without California Highway Patrol and for the change in stop frequency before and after daylight savings. 36 See Table 2.14.4 in the Technical Report for VOD regression results excluding California Highway Patrol data. Compared to analyzing all agencies together, excluding CHP in subsequent analyses produced contrasting results. With the exception of Hispanic persons, the strength of the disparity reversed for all other groups.

Specifically, if the disparity in stop probability at night was significant for a group in the full analysis, significance was lost with the exclusion of CHP. But, if the disparity in stop probability was not significant in the full analysis, significance was gained with the exclusion of CHP. For example, the disparity between Black and White individuals was significant when all data was included, but was no longer significant when CHP data was excluded. 37 This includes both searches of the person (9.2 percent of individuals) and searches of their property (4 percent of individuals).

Officers could conduct both a person and property search of the same person, which is why both these search types taken together amount to 9.9 percent of individuals, rather than 13.2 percent.

These figures do not include canine searches (0.1 % of individuals). 38 Middle Eastern/South Asian (2.8 %) and Asian (3.1 %) persons had lower search rates than White persons. 50.3% 51.2% 49.5% 50.5% 46.5% 45.5% 43.0% 50.0% 49.7% 48.8% 50.5% 49.5% 53.5% 54.5% 57.0% 50.0% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Asian (9,428) Black (25,045) Hispanic (66,682) Middle Eastern/ South Asian (8,333) Multiracial (2,170) Native American

(389) Pacific Islander

(931) White (54,546) Light Dark

35 group with the highest percentage of stops where a search occurred was Black individuals; stops of Black individuals involved a search 18.7 percent of the time, while the racial/ethnic group with the next closest search rate (Multiracial individuals) had a search rate less than two thirds as high as Black individuals. Officers searched Black individuals whom they stopped at a rate that was 2.9 times the rate they searched White individuals. Figure 9.

Search Frequency by Race/Ethnicity Basis for search: We created search discretion categories in our data, adapting what previous studies have done to explore the issue of officer discretion for searches.39 We examined searches in two categories: “higher discretion” and “lower discretion” (Figure 10).40 Administrative, or “lower discretion,” searches are most often required under department policy and include those performed following an arrest, pursuant to a warrant, or after impounding a vehicle. 41 On the contrary, “higher discretion” searches are those where officers have the most flexibility in determining who to search, and include only those occurrences where consent is the only basis provided. 42 Individuals for whom 39 See Chanin, J., Welsh, M., & Nurge, D. (2018).

Criminal Justice Policy Review, 29(6–7), 561–583 or Mosher, C., & Pickerill, J. (2011). Seattle University Law Review, 35(3), 769. 40 For a more thorough review on the distinctions between lower and higher discretion searches, see Chanin et al. (2018). For the purposes of this report, searches conducted with a warrant were also included in the “low” discretion category. 41 Corresponding bases for search found in the RIPA Stop Data include incident to arrest, search warrant, and vehicle inventory. 42 Also of note, some studies include “Terry” searches or frisks (see Terry v.

Ohio) in the higher discretion search category as well. Terry Searches include those justified as a protective search (pat search) for weapons based on reasonable belief that the person is dangerous or carrying a weapon. Terry searches do not have a direct analog in the RIPA regulations. However, the Board has received public comments about proxies for Terry searches in the stop data.

In response, an additional version of the yield rate analysis using an alternate higher-discretion categorization was included in the Technical Report (Table 2.15.8); the alternate higher-discretion scheme includes searches based on consent, officer safety, or suspected weapons and excludes all other potential search bases. See footnote 45 for a synopsis of how this alternative categorization scheme affected results.

36 officers provided other search bases (e.g. canine detection, officer safety) are not included in either of the two discretion categories. Thus, these individuals were not included in the discretion level analyses. Figure 10 displays the racial and ethnic distribution of individuals searched by officers in higher and lower discretion searches. Figure 10. Search Discretion by Race/Ethnicity Search efficacy: There are a number of factors that an officer may use when deciding to undertake a discretionary search.

A central factor is the strength of an officer’s suspicion that the stopped individual has contraband and that a search will reveal that contraband. If an officer’s suspicion is a primary factor and the officer is not using race as part of their decision to search, then we would expect individuals would have to exhibit roughly the same level of suspicious behavior (e.g. the frequency of furtive movements) for an officer to decide to conduct a search. We also would expect that the more suspicious a person appears, the more likely it is that they have contraband.

Combining these two assumptions creates a statistical test for whether or not officers apply different standards to people from different identity groups. If officers are less likely to find contraband after searching people of a particular identity group, then we assume this means that the searched individuals in that identity group are objectively less suspicious, and thus subject to search because of their identity rather than any suspicious behavior.

Alternatively, if searches yield comparable rates of contraband and evidence across all racial groups, this would suggest officers’ thresholds of suspicion justifying a search are similar across race. The following sections employ various analyses to explore this possibility. Ke y Terms Yield rate: proportion of searched individuals found in possession of contraband or evidence.

Officer-discretion level: level of discretion available to the officer in deciding to conduct a search. • Higher: includes searches where the only listed basis for search was “consent given”. • Lower: incident to arrest, vehicle inventory, and search warrants.

37 We examine search yield rates in the following sections. A search yield rate is the proportion of individuals that were subject to a search that officers found to be in possession of contraband or evidence. 43 Yield rates are calculated in the following manner: !"#$%& () *%+&,ℎ%. /0.121."+34 516ℎ 7(06&+$+0. (& 821.%0,% 9(6+3 !"#$%& *%+&,ℎ%. /0.121."+34 ∗ 100 Search yield rate is a measure of search efficacy. Thus, higher rates indicate that searches were successful and resulted in finding contraband or evidence (a “hit”) more often.

Understanding the efficacy of searches can help reveal whether certain identity groups are under a greater degree of unwarranted scrutiny during stops. Before discussing yield rates, it is important to note that conducting searches is not the only way officers discover contraband or evidence. RIPA data collection allows officers to report that they discovered contraband or evidence regardless of whether or not they searched an individual. Wave 1 officers discovered contraband or evidence on 3.4 percent (60,792) of individuals they stopped.

Of the individuals who had contraband or evidence discovered during their stop, 65.3 percent (39,676) of individuals were searched, while the remaining 34.7 percent (21,115) were not searched. 44 Only searched individuals are included in yield rate analyses. Search yield rates: Figure 11 displays the search yield rates of the racial/ethnic groups collected under RIPA. The search yield rate for White individuals was 24.3 percent. Yield rates were lower for all racial groups of color compared to White individuals (1.8 to 5.6 percentage points lower).

This shows that officers were less successful at finding contraband or evidence of wrongdoing when searching individuals of color than White individuals. 43 RIPA regulations do not differentiate between cases where contraband is found in plain view prior to conducting a search versus cases where no contraband or evidence is viewed prior to the search. 44 While 72.7 percent of yields from searches came from drug-related contraband, the most frequently discovered contraband or evidence type from stops without a search was alcohol (43.1%).

Black individuals (2.17%) had the highest intra-group rates of contraband discovered in the absence of a search while Middle Eastern/South Asian individuals (0.33%) had the lowest. In this section, we did not perform further analyses surrounding contraband or evidence discovered in cases where individuals were not searched. Future analyses may examine these circumstances.

38 Figure 11. Search Yield Rates by Race/Ethnicity Yield rates by discretion level: In addition to examining overall search yield rates, we examine yield rates based on the level of discretion the officer had in deciding to conduct the search (see Figure 12). 45 When officers conducted highly discretionary searches of individuals (only basis was consent), officers had higher yield rates for White persons than for all other racial/ethnic groups.

Most racial or ethnic groups also had lower yield rates than White persons for searches where officers are presumed to have less discretion (lower-discretion searches); however, yield rates for Black persons were 1.3 percentage points higher for lower-discretion searches—despite experiencing a smaller proportion of searches with this level of discretion (see Figure 10). 46 45 Searches that were not categorized as lower or higher discretion constituted 48.3 percent of stops with searches. See Table 2.15.7 in

Section 1, Subsection 2 of the Technical Report for a breakdown of search yield rates by each individual basis for search. 46 Under the alternative higher discretion categorization scheme (see footnote 41), individuals of color, except persons perceived to be Pacific Islander, had lower yield rates than White persons. With the exception of Middle Eastern/South Asian and Native American individuals, yield rate differences between White individuals and other racial or ethnic groups decreased when including searches for officer safety and suspected weapons in the higher discretion category.

39 Figure 12. Search Yield Rates by Search Discretion by Race/Ethnicity Stop circumstance: The stops for 95.3 percent of all individuals were officer-initiated, while 4.7 percent of all stops were in response to a call for service. 47 Thus, the findings in the yield rate analysis overall are largely driven by officer-initiated stops. All individuals of color had lower yield rates compared to White individuals overall (see Figure 11).

When analyzed calls for service separately to better understand the issue, the difference in yield rates between White individuals and many persons of color was less pronounced. 48 It is worth noting that stops made in response to a call for service may or may not be of the subject of the call. Considerations and limitations of search yield rates: Search yield rate tests avoid some of the issues of other tests because yield rates do not require the stop data to be matched with, or compared to, another set of data.

However, one consideration when examining yield rates is that there can be observable factors that influence officers’ decisions to search individuals related to the identity of the stopped individual that RIPA stop data collection may not capture. If this were the case, then we could incorrectly attribute this identity-neutral reason for differences in search frequency to identity. 47 CHP conducted the fewest stops in response to a call for service. To ensure the results of the overall yield rate analysis were not driven solely by CHP, we also analyzed the data without their records (Appendix D, Table 7).

When we exclude CHP data, the direction of all significant disparities between White and Non-White groups matched the results of analyses from all reporting agencies together (e.g. Asian persons had lower overall rates than White persons regardless of whether CHP data were included).

For this reason, the CHP data was included in all results discussed in the main report body. 48 The percentage point difference in yield rates between Multiracial, Pacific Islander, Native American, Hispanic, and Black individuals and White individuals was less when examining only individuals stopped in response to calls for services than when we examined all searched individuals, regardless of the stop circumstance.

40 Further, since the analysis is based on all discoveries of contraband, differences in the frequency with which people in one identity group are very suspicious and would always be searched, can mask racial differences in the frequency with which people who are only slightly suspicious are searched. 49 Enforcement Rates for Stops with Searches: To understand how frequently officers searched individuals and then decided to take an enforcement action afterwards, we examined enforcement rates for searched individuals.

For the purpose of this report, we define enforcement rates as the proportion of a group of stopped individuals who were arrested or received a citation. We excluded stops where officers listed “incident to arrest” or “vehicle inventory” as a basis for the search. 50 After excluding these stops, we learned that officers took enforcement action with 26 percent of the individuals they searched. The proportion of the searched individuals that were subject to an enforcement action varied by race/ethnicity, with Black individuals having the lowest rate (21.2%) and White individuals having the highest rate (33%).

Officer-initiated stops with searches appear to drive the overall enforcement rates of searched individuals. When examining only officer-initiated stops with searches, White individuals (35.8%) had higher enforcement rates than all other racial or ethnic groups; Black individuals had the lowest enforcement rates (20.9%). The distribution of enforcement rates for the small proportion of searched individuals who were stopped in response to a call for service was different from the overall and officer-initiated enforcement rates.

Enforcement Rates for all Stops: Officers took enforcement action on 60.3 percent of all individuals stopped during the reporting period, ranging from 51.6 percent of Black individuals to 68.7 percent of Asian individuals (Figure 13). 51 These trends were driven by officer-initiated stops where Black individuals (52.2%) continued to have the lowest enforcement rates overall and Asian individuals (69.4%) the highest.

White individuals (38.9%) had lower enforcement rates than other racial or ethnic groups (41.4 – 49.5%) when analyzing calls for service independently. 49 See page 26 of the 2019 RIPA Board report for an example that illustrates infra-marginality. 50 By definition, these searches would come after the officer had already decided to take enforcement action and therefore do not follow the progression from search to enforcement that this analysis seeks to examine. 51 Enforcement action is a citation for infraction, in-field cite and release, custodial arrest without a warrant, custodial arrest with a warrant, or any combination of these four results of stop.

41 Figure 13. Citation and Arrest Rates by Race/Ethnicity 52 Analyzing enforcement by type of offense revealed a more nuanced pattern. Relative to other groups, a lower percentage of Black and Native American individuals were issued citations (36.8% – 38.0%), while Asian and Middle Eastern/South Asian individuals had higher citation rates (57.6% – 61.4%). Conversely, Black and Native American individuals were arrested at relatively high rates (15.2% – 16.0%) compared to Middle Eastern/South Asian and White individuals (7.0% – 11.3%), who had lower percentages of arrests overall (see Figure 13).

Ongoing Training to Ensure the Continued Integrity of Data Collection and Submission To gain insight into the specific needs of law enforcement agencies with respect to the technical aspects of data collection and submission to the Department, the Department’s Client Services Program (CSP) facilitated two Lessons Learned sessions during the fall of 2019.

The Department’s business, legal, technical, and research teams participated with law enforcement staff representing the fifteen agencies currently collecting stop data (the Wave 1 and Wave 2 agencies), as well as some from the Wave 3 agencies who are scheduled to begin collecting data on January 1, 2021.

The goal of the sessions was to elicit feedback on training, outreach, technology, timelines, annual close-out process, the designation and handling of persons’ personally identifiable information and officers’ unique identifying information, as well as responses to Public Records Act requests, data analysis, and future enhancements. The agencies were able to share their experiences and feedback, trade advice, and discuss gaps in training with the Department.

These sessions served as an open forum to share the lessons learned during the initial implementation process of the data collection and identified a need 52 The arrest category in Figure 13 includes custodial arrests (both with and without a warrant), as well as in-field cite and releases.

42 for more scenario-based training. The CSP will incorporate the feedback to improve the implementation process for the next group of agencies. Data Integrity Video In May 2019, the RIPA Board released a five-minute video in which six diverse stakeholders address data integrity for the RIPA stop data. The video outlines the role of law enforcement agencies and the Department in performing data integrity checks, as described by Dr.

Sharad Goel, Stanford University Assistant Professor and Founder and Executive Director of the Stanford Computational Policy Lab: “The integrity of the stop data is checked at several phases of the collection and analysis process ... If discrepancies are discovered anywhere in the [collection/reporting] pipeline, State officials can work with local jurisdictions to improve the quality of collected data.” Dr.

Jack Glaser, Professor at the Goldman School of Public Policy at the University of California, Berkeley, and recognized expert on racial profiling, explained what data integrity means and how it is achieved: “Data integrity, at its core, means that the numbers reflect reality. This happens when officers record all stops fully and forthrightly, and when these records are stored and shared consistently and transparently.

In order for people to be able to trust the data, it is crucial that reporting requirements and guidelines be consistent across and within departments.” The Data Integrity video is available on YouTube 53 and a link is provided on the RIPA Board webpage. 53 California Department of Justice. (2019, May 2). RIPA - Data Integrity [Video file]. Available at https://www.youtube.com/watch?v=F2evScIOFo0&t=3s.

43 Racial and Identity Profiling Policies and Accountability Both the United States and California Constitutions provide for equal protection under the law and the right to be free from unreasonable searches and seizures conducted by government. California law further guarantees these rights for all people, regardless of the actual or perceived race, color, ethnicity, national origin, age, religion, gender identity or expression, sexual orientation, or mental or physical disability of the individual. 54 Police action that is biased is illegal and violates these rights.

Biased-based policing, furthermore, alienates the public, fosters distrust of police, and undermines legitimate law enforcement efforts. 55 As stated by the California Legislature, racial and identity profiling is “abhorrent and cannot be tolerated.” 56 RIPA directs the Board to review and analyze “racial and identity profiling policies and practices across geographic areas in California, working in partnership with state and local law enforcement agencies.” 57 In its 2019 report, the Board surveyed all California law enforcement agencies subject to stop data reporting on their current policies and practices relevant to preventing racial and identity profiling and their efforts to enhance law enforcement-community relations and reduce bias in policing.

The Board found that while most agencies did have a specific policy or portion of a policy addressing racial and identity profiling, there was little consistency in the substance of the policies across agencies. 58 In light of this lack of consistency, this year’s report provides model language that law enforcement can include in their bias-free policing policies. This model language is based on existing evidence-based best practices provided in the Board’s last report.

The Board provides this language with the caveat that this model language is only a starting point for protecting the constitutional rights of Californians. Bias-free policing is constantly evolving, and thus policies will need frequent updating to track with the latest police practices. The Board encourages law enforcement agencies to collaborate with community members to develop their bias-free policing policies and to adapt the language of the recommended policies to fit the communities they serve.

Recommendations for Model Bias-Free Policing Policies A model bias-free policing policy is a stand-alone policy devoted to bias-free policing. It uses clear language, including

definitions of relevant terms, and expresses the agency or department’s responsibility to identify and eliminate racial and identity profiling. In addition to stating the agency or department’s core values and its commitment to bias-free policing, a model policy includes relevant federal and state law. A model policy is based on best practices, well researched, and regularly updated with changes in the law or best practices. A model bias-free policing policy includes cross references to other relevant agency policies on subjects such as civilian complaints, stops, use of force, training, and accountability.

It also includes references to relevant training that agency or department 54 Pen. Code, §13519.4. 55 Pen. Code, §13519.4, subd. (d)(1)-(4). 56 Pen. Code, §13519.4, subd. (d)(2). 57 Pen. Code, §13519.4, subds. (j)(3) & (A)-(E). 58 Of the 425 law enforcement agencies in the State that were sent the survey, 114 agencies participated, and thus the responses may not be representative of all agencies in the State. The current report focuses on the bias-free policing policies of the eight Wave 1 agencies that began collecting data on July 1, 2018.

44 personnel receive on subjects such as implicit bias, civilian complaint procedures, human and community relations, etc. A model stand-alone policy is easily accessible to both agency personnel and the public. All personnel, including dispatchers and non-sworn personnel, should receive training on the bias-free policing policy. Specific examples of behavior that violates the bias-free policing should be included in either the training or the policy itself. Below is model policy language and

definitions that LEAs can consider including in their bias - free policing policies. The Board notes that these recommendations are the floor, and not the ceiling, of best practice recommendations for bias-free policing policies. A.

Model Policy Language for Bias-Free Policing Policy • The [agency] expressly prohibits racial and identity profiling. • The [agency] is committed to providing services and enforcing laws in a professional, nondiscriminatory, fair, and equitable manner that keeps both the community and officers safe and protected. • The [agency] recognizes that explicit and implicit bias can occur at both an individual and an institutional level and is committed to addressing and eradicating both. • The intent of this policy is to increase the [agency’s] effectiveness as a law enforcement agency and to build mutual trust and respect with the [city, county or state’s] diverse groups and communities. • A fundamental right guaranteed by the Constitution of the United States is equal protection under the law guaranteed by the Fourteenth Amendment.

Along with this right to equal protection is the fundamental right to be free from unreasonable searches and seizures by government agents as guaranteed by the Fourth Amendment. • The [agency] is charged with protecting these rights.

Police action that is biased is unlawful and alienates the public, fosters distrust of police, and undermines legitimate law enforcement efforts. • All employees of [agency] are prohibited from taking actions based on actual or perceived personal characteristics, including but not limited to race, color, ethnicity, national origin, age, religion, gender identity or expression, sexual orientation, or mental or physical disability, except when engaging in the investigation of appropriate suspect-specific activity to identify a particular person or group. • [Agency] personnel must not delay or deny policing services based on an individual’s actual or perceived personally identifying characteristics.

45 B. Model Policy Language for

Definitions Related to Bias • Racial or Identity Profiling: the consideration of, or reliance on, to any degree, actual or perceived race, color, ethnicity, national origin, age, religion, gender identity or expression, sexual orientation, or mental or physical disability 59 in deciding which persons to subject to a stop or in deciding upon the scope or substance of law enforcement activities following a stop, except that an officer may consider or rely on characteristics listed in a specific suspect description.

Such activities include, but are not limited to, traffic or pedestrian stops, or actions taken during a stop, such as asking questions, frisks, consensual and nonconsensual searches of a person or any property, seizing any property, removing vehicle occupants during a traffic stop, issuing a citation, and making an arrest. 60 • Bias-Based Policing: conduct by peace officers motivated, implicitly or explicitly, by the officer’s beliefs about someone based on the person’s actual or perceived personal characteristics, i.e., race, color, ethnicity, national origin, age, religion, gender identity or expression, sexual orientation, or mental or physical disability. • Implicit Bias: the attitudes or stereotypes that affect a person’s understanding, actions, and decisions in an unconscious manner.

These biases, which encompass both favorable and unfavorable assessments, are activated involuntarily and without an individual’s awareness or intentional control. Implicit biases are different from known biases that individuals may choose to conceal. • Bias by Proxy: when an individual calls/contacts the police and makes false or ill-informed claims of misconduct about persons they dislike or are biased against based on explicit racial and identity profiling or implicit bias. 61 When the police act on a request for service based in unlawful bias, they risk perpetuating the caller’s bias.

Members should use their critical decision-making skills, drawing upon their training to assess whether there is criminal conduct. • Reasonable Suspicion to Detain: reasonable suspicion is a set of specific facts that would lead a reasonable person to believe that a crime is occurring, had occurred in the past, or is about to occur. Reasonable suspicion to detain is also established whenever there is any violation of law.

Reasonable suspicion cannot be based solely on a hunch or instinct. • Detention: a seizure of a person by an officer that results from physical restraint, unequivocal verbal commands, or words or conduct by an officer that would result in a reasonable person believing that he or she is not free to leave or otherwise disregard the officer. 62 • Reasonable Suspicion to Conduct a Pat Search: officers are justified in conducting a pat search if officers have a factual basis to suspect that a person is carrying a weapon, dangerous instrument, or an object that can be used as a weapon, or if the person poses a danger to the safety of the officer or others.

Officers must be able to articulate specific facts that support an 59 Some agencies include other personal characteristics in their racial or identity profiling policies, such as socioeconomic status or immigration status. 60 Cal. Pen. Code, § 13519.4, subd. (e). 61 Fridell, A. (2017). Comprehensive Program to Produce Fair and Impartial Policing. USA: Springer International Publishing, p. 90. 62 11 CCR § 999.224(a)(7).

46 objectively reasonable apprehension of danger under the circumstances and not base their decision to conduct a pat search on any perceived individual characteristics. Reasonable suspicion to conduct a pat search is different than reasonable suspicion to detain. The scope of the pat search is limited only to a cursory or pat down search of the outer clothing to locate possible weapons.

Once an officer realizes an object is not a weapon, or an object that can be used as a weapon, the officer must move on. • Probable Cause to Arrest: under the Fourth Amendment to the United States Constitution, arrests must be supported by probable cause. Probable cause to arrest is a set of specific facts that would lead a reasonable person to objectively believe and strongly suspect that a crime was committed by the person to be arrested. C.

Model Policy Language for Limited Circumstances in which Characteristics of an Individual May Be Considered • [Agency] members may only consider or rely on characteristics listed in a specific description of a suspect, victim, or witness based on trustworthy and relevant information that links a specific person to a particular unlawful incident. • Except as provided above, [agency] officers shall not consider personal characteristics in establishing either reasonable suspicion or probable cause. D.

Model Policy Language for Encounters with Community • To cultivate and foster transparency and trust with all communities, each [agency] member shall do the following when conducting pedestrian or vehicle stops or otherwise interacting with members of the public, unless circumstances indicate it would be unsafe to do so: o Be courteous, professional, and respectful. o Introduce themselves to the community member, providing name, agency affiliation, and badge number. [Agency] members should also provide this information in writing or on a business card. 63 o State the reason for the stop as soon as practicable, unless providing this information will compromise officer or public safety or a criminal investigation. o Answer questions that the individual may have about the stop. o Ensure that a detention is no longer than necessary to take appropriate action for the known or suspected offense and [agency] member convey the purpose of any reasonable delays. 63 President’s Task Force on 21 st Century Policing. (2015).

Final Report of the President’s Task Force on 21 st Century Policing. Washington, DC: Office of Community Oriented Policing Services, p. 27. Available at http://elearning-courses.net/iacp/html/webinarResources/170926/FinalReport21stCenturyPolicing.pdf (identified as recommendation 2.11, with accompanying Action Item 2.11.1 for promoting effective crime reduction while building public trust).

47 • All [agency] personnel, including dispatchers and non-sworn staff, shall not use harassing, intimidating, derogatory, or prejudiced language, including profanity or slurs, particularly when related to an individual’s actual or perceived individual characteristics. • Dispatchers and sworn personnel shall be aware of and take steps to curb the potential for bias by proxy in a call for service. • Officers should draw upon their training and use their critical decision-making skills to assess whether there is criminal conduct and to be aware of implicit bias and bias by proxy when carrying out their duties. • All [agency] personnel, including dispatchers and non-sworn personnel, shall aim to build community trust through all actions they take, especially in response to bias-based reports.

E. Model Policy Language for Training • The [agency] will ensure that, at a minimum, all officers and employees are compliant with requirements regarding bias-free policing training. • The [agency] will ensure that management includes a discussion of its bias-free policing policy with its officers and staff on an annual basis. • [Agency] officers should be mindful of their training on implicit bias and regularly reflect on specific ways their decision-making may be vulnerable to implicit bias. F.

Model Policy Language for Data Collection and Analysis • As required by the California Racial and Identity Profiling Act of 2015, [agency] is required to collect data on: (

a) civilian complaints that allege racial and identity profiling and (

b) perceived demographic and other detailed data regarding pedestrian and traffic stops. The data to be collected for stops includes, among other things, perceived race or ethnicity, approximate age, gender, LGBT identity, limited or no English fluency, or perceived or known disability, as well as other data such as the reason for the stop, whether a search was conducted, and the results of any such search.

All agencies must report this data to the California Department of Justice. • The [agency] should regularly analyze data, in consultation with [academics, police commissions, civilian review bodies, or advisory boards], to assist in identifying practices that may have a disparate impact on any group relative to the general population. G. Model Policy Language for Accountability and Adherence to the Policy • All [agency] personnel, including dispatchers and non-sworn personnel, are responsible for understanding and complying with this policy.

Any violation of this policy will subject the member to remedial action. o Types of remedial action should be outlined.

48 • All [agency] personnel, including dispatchers and non-sworn personnel, shall not retaliate against any person who complains of biased policing or expresses negative views about them or law enforcement in general. • All [agency] personnel, including dispatchers and non-sworn personnel, share the responsibility of preventing bias-based policing. Personnel shall report any violations of this policy they observe or of which they have knowledge. o Processes and procedures for reporting violations should be included. H.

Model Policy Language for Supervisory Review • Supervisors shall ensure that all personnel under their command, including dispatchers and non-sworn personnel, understand the content of this policy and comply with it at all times. o Supervisory processes and procedures for monitoring should be included. • Any employee who becomes aware of any instance of bias-based policing or any violation of this policy shall report it in accordance with established procedure. • Supervisors who fail to respond to, document, or review allegations of bias-based policing will be subject to remedial action. o Types of remedial action should be outlined. o Supervisor processes and procedures for review should be included.

Wave 1 Agency Bias-Free Policing Policy Review This year, the Board undertook a review of the bias-free policing or equivalent policies for all eight Wave 1 agencies. The matrix below summarizes the Board’s review of the most recent policies the Department obtained, based on the best practices outlined in the 2019 RIPA Board Report. Following the matrix is a more detailed review of each agency’s bias-free policing policy and related policies that contain relevant information.

In the 2019 Report, the Board recommended various best practices to assist agencies with having clear, thoughtful, and robust bias-free policing policies. To that end, the Board reviewed the factors below. First, the Board assessed whether the policy was clear about the agency’s prohibition against bias - based policing and whether that commitment was furthered by having a stand-alone policy. Additionally, the Board reviewed whether the policy defined bias-based policing and explained in what limited circumstances personal characteristics may be considered.

Next, the Board evaluated whether the policy was accessible to the public and whether the policy discussed guidelines according to which agency members should interact with the community. The Board also assessed whether the policy included a component on training related to racial and identity profiling. Lastly, the Board evaluated the accountability built into the policy by looking at whether the policy discussed analysis of data collected and supervisory review. In its review, the Board was not expecting each agency to exactly follow the above-mentioned model language.

Instead, the Board looked for instances where the concepts above were incorporated into the policies. These recommendations represent an accumulation of best practices identified by the United States Department of Justice (USDOJ) and other relevant empirical research conducted by well-regarded

49 organizations, including the Police Executive Research Forum (PERF), 64 the International Association of Chiefs of Police (IACP), 65 the Vera Institute, 66 Fair and Impartial Policing, 67 Stanford SPARQ, 68 and the Center for Policing Equity (CPE). 69 The Department shared this review with the subject LEAs to ensure accuracy before including this information in the report. The RIPA Board encourages all Wave 1 agencies to re-examine their policies.

The Policy Review that follows may assist agencies in identifying areas of opportunity to incorporate the best practices outlined in the Board’s 2019 report and the aforementioned model language. Wave 1 Agency Stand-Alone Bias-Free Policing Policy? Clearly Written? Easily Accessible? Uses Concrete

Definitions of Bias-Free Policing and/or Racial & Identity Profiling? Component on Limited Circumstances in which Characteristics of Individual May Be Considered? San Francisco PD ü ü ü ü ü CHP û ü û ü ü Los Angeles PD ü ü ü ü ü Riverside Sheriff ü ü û ü ü San Bernardino Sheriff ü ü û û û San Diego PD ü ü û ü û San Diego Sheriff ü ü ü û ü Los Angeles Sheriff û ü û û û 64 Police Executive Research Forum (PERF). Information available at https://www.policeforum.org/. 65 International Association of Chiefs of Police (IACP). Information available at https://www.theiacp.org/. 66 The Vera Institute of Justice.

Information available at https://www.vera.org/. 67 Fair and Impartial Policing. Information available at https://fipolicing.com/. 68 Stanford SPARQ. Information available at https://sparq.stanford.edu/. 69 Center for Policing Equity (CPE). Information available at http://policingequity.org/.

50 Wave 1 Agency Component on Encounters with Community? Component on Racial and Identity Profiling Training? Component on Data Analysis? Component Requiring Account - ability? Supervisory Review?

San Francisco PD ü û ü ü ü CHP ü ü ü ü ü Los Angeles PD ü û û ü ü Riverside Sheriff û ü û ü û San Bernardino Sheriff ü ü û û û San Diego PD û û û ü û San Diego Sheriff û û û û û Los Angeles Sheriff û û û ü û San Francisco PD: The San Francisco Police Department is in the process of revising Department General Order 5.17. 70 The information in the above chart is from a review of the current policy, which was revised in May 2011. The 2011 policy, which is available in English on the SFPD website 71 mentions equal protection and Fourth Amendment laws and contains a definition of biased policing.

In line with the Board’s best practice recommendations, it includes a component on the limited circumstances in which characteristics of individuals may be considered, as well as a component on communication with the community to prevent perceptions of biased policing. However, the policy does not contain a 70 In updating its anti-bias policy, the SFPD gathered various stakeholders from the community and local government, including the San Francisco Police Commission and the San Francisco Department of Police Accountability, to help draft the soon to be approved policy.

Other law enforcement agencies should consider a similar approach to improve community and law enforcement relations. 71 See San Francisco Police Department. (2011). General Order 5.17: Policy Prohibiting Biased Policing [PDF file]. Available at https://www.sanfranciscopolice.org/sites/default/files/2018 - 11/DGO5.17%20Policy%20Prohibiting%20Biased%20Policing.pdf.

51 component on racial and identity profiling training. It includes components for accountability and supervisory review. A separate policy, San Francisco Administration Code,

section 96A.3, mandates SFPD to conduct analysis and reporting of collected data. Quarterly reports with the data analysis, including an executive

summary, are available on the agency’s website. CHP: The California Highway Patrol does not have a stand-alone bias-free policing policy. Relevant content is integrated into the Enforcement Policy Manual and is additionally reflected in the Drug Programs Manual; neither of these manuals is available online. The Enforcement Policy Manual includes information on the requirements under current state and federal law. CHP policies define racial and identity profiling, as well as probable cause, consent, and reasonable suspicion.

They include a component on the limited circumstances in which characteristics of individuals may be considered, as well as a component on encounters with the community. Annual cultural awareness training is provided to all employees and includes training on racial profiling; an eight-hour classroom-training course is alternated with an online refresher course every odd-numbered calendar year. The policies include components for the analysis of the collected data, accountability, and supervisory review.

LAPD: The Los Angeles Police Department has a three-paragraph, stand-alone Policy Prohibiting Biased Policing that is clearly written and available in English on the LAPD website. 72 The policy was updated in November 2019, expanding protected classes to include immigration or employment status, language fluency, and homeless circumstance. The policy defines bias-free policing. It includes a component on the limited circumstances in which characteristics of individuals may be considered. Furthermore, it designates failure to comply as

an act of serious misconduct and requires employees to report violations of the policy. Related content is included in other policy sections, including encounters with the community. Supervisory review is addressed in a separate

section of the Department Manual.

Section 4/202.2 – Automated Field Data Reports (AFDR)/Completion and Tracking outlines officers’ responsibilities for completing AFDRs and describes supervisors’ responsibilities for: • reviewing AFDRs promptly to ensure that officers are properly completing the AFDR per the AFDR Completion Guide and Supervisor AFDR Completion Guide; • editing or directing the completing officer to revise the narrative portions of the AFDR, when appropriate; • ensuring that a legal basis for the detention and search (if applicable) is adequately articulated in the narrative; and, • ensuring that no identifying characteristics of the person(

s) being stopped or the officer(

s) involved are listed. Watch Commanders and Commanding Officers’ responsibilities related to AFDR are also specified. The LAPD policy does not include a component on racial and identity training. However, LAPD provided to the Board a ten-page Police Training and Education – 2019 Biased Policing Reduction 72 See Los Angeles Police Department. (2019). 2019 2 nd Quarter Manual. Available at http://lapdonline.org/lapd_manual/volume_1.htm#345.

52 Strategy document that includes detailed information about current training courses required of officers, supervisors, and command staff. The LAPD policy does not include a component on data analysis. LAPD did, however, share a document, Efforts to Reduce the Number of Biased Policing Complaints Report, which outlines the LAPD’s data analysis efforts. In a letter to the Department, dated December 2, 2019, the LAPD provided additional details about data analysis by a Steering Committee that meets every four weeks.

The letter also describes a Stop Data Dashboard that the LAPD is developing to provide commanding officers insight into the types of stops being conducted, reasons for stops, searches conducted, and actions taken by officers in the field. Riverside Sheriff: The Riverside County Sheriff’s Department has a clearly written stand-alone 73 policy that was last revised October 7, 2019. The policy is not available online. It defines bias-based policing and includes a component on the limited circumstances in which characteristics of individuals may be considered. There is no component on encounters with the community.

The policy includes a component on officer training and encourages members to familiarize themselves with racial and cultural differences if they have not yet received training. The policy does not include a component on data analysis; it does delineate, however, what data is collected for RIPA. The policy requires members to be responsible for reporting any biased-based policing they suspect or have knowledge of and encourages members to intervene whenever they see bias-based actions. The policy does not address supervisory review.

San Bernardino Sheriff: The San Bernardino County Sheriff’s Department has a clearly written two - sentence, stand-alone policy prohibiting biased policing. This policy is