The URL can be used to link to this page

Home My WebLink AboutStaff Report 2512-5589CITY OF PALO ALTO CITY COUNCIL Special Meeting Monday, March 09, 2026 Council Chambers & Hybrid 5:30 PM Agenda Item A.Informational Report: Report from the National Policing Institute Regarding Palo Alto's RIPA Data (Racial and Identity Profiling Act of 2015); CEQA – Not a Project City Council Staff Report From: City Manager Report Type: INFORMATION REPORTS Lead Department: City Manager Meeting Date: March 9, 2026 Report #:2512-5589 TITLE Informational Report: Report from the National Policing Institute Regarding Palo Alto's RIPA Data (Racial and Identity Profiling Act of 2015); CEQA – Not a Project RECOMMENDATION Staff recommends that the City Council accept this informational report containing the report from the National Policing Institute regarding Palo Alto’s Racial and Identity Profiling Act of 2015 (RIPA) data as well as information from City staff as context related to the report. BACKGROUND Assembly Bill 953, the Racial and Identity Profiling Act of 2015 (RIPA), requires all California law enforcement agencies to collect specific information on certain types of police contacts (detentions and contacts during which a person is searched). It also requires that this data be reported to the California Department of Justice (DOJ) and publicly shared. RIPA also requires the Attorney General to issue regulations for the collection and reporting of this stop data (statewide). (Gov. Code, § 12525.5, subd. (e)). The implementation timeline for law enforcement agencies varied based on the number of sworn officers employed by the agencies. The first wave of agencies required to start data collection were those with 1,000 or more officers; this data collection began mid-2018. The second wave of agencies, with 500-999 officers started data collection in January 2020. The third wave of agencies, with 250-499 officers started data collection in January 2021. The last wave of agencies, with fewer than 250 officers, started data collection in January 2022. Palo Alto was in the fourth wave of agencies for implementation and began the first year of RIPA data collection, per state law, in 2022. Thus, the data shared with the National Policing Institute (NPI) for analysis was the full year data of 2022 and 2023, both of which were also publicly available on the City’s RIPA data dashboard. The City’s RIPA dashboard published the first round of data in 2023. The information in the dashboard has been available for public review through an interactive RIPA data dashboard that display that same data by applying user-defined filters. In addition to the data dashboard, the public can also download the raw data. The dashboard is one of the only RIPA data dashboards among nearby law enforcement agencies. Cultivating Experiences and Appreciation Fostering a Welcoming Environment Applying an Equity Lens: Strategic Review of the Organization through an Equity Lens ANALYSIS Palo Alto’s RIPA data is entered by officers at the completion of their interaction with the person/contact. The data required to be collected includes: The date, time, duration, and location of the stop Information about the stopped person, as perceived by the officer: Race or ethnicity Gender Age If the officer perceived the person to be LGBT[QIA+] If the officer perceived the person as having limited or no English fluency If the officer perceived the person as having a disability The reason for the stop If the stop was made in response to a call for service Actions taken by the officer during the stop The basis for any search and if property was seized If any contraband or evidence was discovered The result of the stop The officer's unique identification number, years of experience at the time of the stop, and assignment at the time of the stop At the end of each calendar year, Palo Alto staff compile the data and submit it for review to the California Department of Justice for certification. Upon certification of the data, the City uploads the data to the City’s website and makes it available on the City’s RIPA Dashboard. The City asked NPI to work collaboratively with the City to assist in understanding the context of police encounters with members of the public during stops and to advise the City on the interpretation and policy implications of RIPA data from analysis findings. The City also asked NPI to examine the public dashboard and to recommend any enhancements. The key findings in the report, found on report pages 34-36 in more detail, are: - Expand and improve data collection - Implement ongoing trend monitoring - Improve the utility of the RIPA Dashboard - Integrate officer and community perspectives Regarding the Dashboard, the key findings are: - Clean and pre-process data before publication - Enhance visualizations and functionality - Expand methodological documentation The recommendations related to the Dashboard are intended to improve the Dashboard usefulness and functionality. As stated in the Police Department responses (Attachment B), the Police Department will seek to implement some aspects of these recommendations. It is noteworthy that the NPI report references the Dashboard as a compliance tool for State law. However, the State RIPA requirement is for an agency to collect the data, report it to DOJ, and then make the data publicly available. The Palo Alto Dashboard goes beyond State RIPA data requirements by having a publicly available interactive data dashboard. Of all the cities in the State, Palo Alto is one of the few cities with a dashboard and is one of the smallest cities with a dashboard. The Dashboard recommendations are suggestions to make the existing dashboard more user friendly and better able to assist individuals that are seeking to interact with the data differently. The Police Department will continue to improve the Dashboard as resources allow. Attachment A and responses from the Palo Alto Police Department regarding the recommendations and key findings are in Attachment B. In addition to those responses, staff also wanted to provide some additional information below as considerations when reviewing the report. Next Steps: FISCAL/RESOURCE IMPACT STAKEHOLDER ENGAGEMENT ENVIRONMENTAL REVIEW ATTACHMENTS APPROVED BY: AN ANALYSIS OF POLICE STOP DATA Prepared for the City of Palo Alto, California January 2026 This project was supported by Grant No. 15 PBJA - 21 - GK - 04011 - JAGT awarded by the Bureau of Justice Assistance. The Bureau of Justice Assistance is a component of the U.S. Department of Justice’ s Office of Justice Programs, which also includes the Bureau of Justice Statistics, the National Institute of Justice, the Office of Juvenile Justice and Delinquency Prevention, the Office for Victims of Crime, and the SMART Office. Points of view or opinions in this document are those of the author and do not necessarily represent the official position or policies of the U. S. Department of Justice. ABOUT THE KNOWLEDGE LAB The Knowledge Lab is a trusted resource for law enforcement and the communities they serve. It supports public safety by identifying and disseminating fair and effective policing practices and technical assistance. The Knowledge Lab collaborates with stakeholder organizations, researchers, and agencies to leverage and share resources and guidance already developed within these professional networks and works to create resources where they are needed. Learn more about the Knowledge Lab at www.leknowledgelab.org. ABOUT NPI The National Policing Institute (NPI) is an independent, nonpartisan, nonprofit research and training institute committed to addressing policing’s most complex challenges through evidence-informed, innovative solutions. By translating research and lessons learned into practice, NPI assists law enforcement organizations in adopting and adapting the most effective programs, resources, and tools available to serve their communities. Learn more at www.policinginstitute.org. ABOUT PALO ALTO POLICE DEPARTMENT The Palo Alto Police Department (PAPD) serves a diverse and engaged community of approximately 68,000 residents covering 24 square miles in the heart of Silicon Valley. The department’s jurisdiction includes residential neighborhoods, a dynamic downtown business district, Stanford University, and several major transit corridors. PAPD has 150 employees and provides a full range of law enforcement services, including patrol, investigations, traffic enforcement, and community policing initiatives. AUTHOR ACKNOWLEDGEMENT This report was written by Colby Dolly, Ph.D., Director of Science & Innovation; Ryan Fisher, Ph.D., Senior Research Manager; Sarah Lawrence, Senior Program Manager; Jie Gao, Research Data Scientist; and Murat Yildirim, M.S., Research Data Manager, at the National Policing Institute. TABLE OF CONTENTS ABOUT THE KNOWLEDGE LAB 2 ABOUT NPI 2 ABOUT PALO ALTO POLICE DEPARTMENT 2 AUTHOR ACKNOWLEDGEMENT 2 INTRODUCTION 1 DATA AND METHODOLOGY 1 RIPA DATA 1 PAPD DATA ANALYSIS RESULTS 7 PAPD STOP PATTERNS 7 SUMMARY OF FINDINGS 33 RECOMMENDATIONS 34 REFERENCES 37 APPENDIX A: RE-CLASSIFYING PAPD RIPA CODES 38 A N A N A L Y SI S O F P OL I C E STO P D A TA / 1 INTRODUCTION The City of Palo Alto requested assistance from the Law Enforcement Knowledge Lab (the Lab) to analyze their Racial and Identity Profiling Act (RIPA) data and public dashboard, including analyzing stop patterns throughout the city. Data collected by the Palo Alto Police Department (PAPD) from 2022 to 2023 under California’s state-mandated RIPA were analyzed by the Lab’s research team. The PAPD serves a diverse community of approximately 68,000 residents, encompassing residential neighborhoods, Stanford University, and several major commercial corridors. As part of statewide efforts to promote transparency and accountability in policing, the department routinely records detailed information on stops, including demographic details, reasons for stops, searches conducted, and stop outcomes. The City was interested in questions related to when, where, and how stops were occurring. The primary goal of this work was to support the PAPD in assessing and strengthening its collection, analysis, and presentation of stop data reported under RIPA. PAPD asked the Knowledge Lab for a detailed audit of the department’s stop records, a comprehensive analysis of stop patterns, and an evaluation of the public-facing dashboard. The project aimed to help the City, including the Police Department and its stakeholders, better understand how this data can be interpreted, identify appropriate benchmarks, and provide clear, recommendations to improve the quality, transparency, and utility of stop data moving forward. The analysis employed descriptive and inferential statistics to identify patterns in Palo Alto’s stops. Specifically, PAPD asked the Knowledge Lab to examine the frequency of stops by time of day, geographic mapping of stops by street and police beat, and evaluating differences in stop reasons and outcomes by race and ethnicity. A veil of darkness analysis—commonly used to detect potential racial disparities in policing—was conducted to explore whether stop likelihood varied by time of day. The approach is useful in identifying potential racial disparities but in and of itself does not conclusively explain why disparities are present. . By systematically examining PAPD stop patterns and characteristics, the report aims to highlight trends, inform policy discussions, and support ongoing efforts to evaluate equitable policing practices that emphasize public safety. While this analysis does not make causal determinations, it identifies areas where further review may be warranted based on observed disparities or enforcement patterns. DATA AND METHODOLOGY RIPA Data Under the Racial and Identity Profiling Act (RIPA) of 2015, California law enforcement agencies are required to collect information on certain types of police contacts, including detentions and contacts during which a person is searched. The data items collected include: A N A N A L Y SI S O F P OL I C E STO P D A TA / 2 • Date, time, duration, and location of the stop • Demographic Information about the stopped person, as perceived by the officer: • Race or ethnicity • Gender • Age • If the officer perceived the person to be LGBT • If the officer perceived the person as having limited or no English fluency • If the officer perceived the person as having a disability • Reason for the stop • If the stop was made in response to a call for service • Actions taken by the officer during the stop • Basis for any search and if property was seized • If any contraband or evidence was discovered • Result of the stop • Officer's unique identification number, years of experience at the time of the stop, and assignment at the time of the stop DATA CLEANING AND PREPARATION Stop Data The PAPD provided the Lab’s research team with two RIPA data files (submitted to and verified by the California Department of Justice) for 2022 and 2023 reported stops. This data includes traffic and pedestrian stops made by the PAPD, either as a result of officer-initiated activity or a call for service. The files were merged to create a master stops file that included 10,509 records with 170 fields. Of these 10,509, over 83% resulted from officer-initiated activity (8,686). The initial data files were provided at the stop level and included instances where multiple people were present for a single stop. The Lab’s research team created two files for the analysis: a stop- level file and an individual-level file. This allowed researchers to examine stops without double- counting incidents for which more than one person was present and examine the demographics of individuals involved in stops. As part of the data cleaning process, the perceived race and ethnicity of stopped individuals needed to be recoded since it was a multiple-selection field (allowing officers to record multiple race or ethnicity categories per individual). The Lab research team used the race and ethnicity coding structure that the California RIPA Board reports have been using in their reporting. Thus, when an individual was recorded with multiple race and ethnicity designations, that individual was coded as “multi-racial” (219 total records). All other race and ethnicity codes were single- selection, and this is reflected in the race and ethnicity analyses presented below (with the exception of six missing race and ethnicity records). Preliminary analyses indicated that several racial and ethnic groups (e.g., Multi-racial, Native American, Pacific Islander, Unknown) represented fewer than 5% of total stops. Due to these groups’ limited representation, and to maintain clarity and interpretability in visualizations and analyses throughout this report, these categories have not been included in detailed figures and tables. Their overall representation and counts are provided below for transparency. A N A N A L Y SI S O F P OL I C E STO P D A TA / 3 Table 1- Race and Ethnicity Stop Counts (2022 to 2023) Race and Ethnicity Number of Stops Percentage White 3423 32.57% Hispanic 3289 31.30% Black 1262 12.01% Asian 1232 11.72% Middle Eastern/South Asian 826 7.86% Other 1 252 2.40% Multi -racial 219 2.08% Missing 6 0.06% Additional data fields allowed for multiple selections, including the actions taken and stop results fields, which required cleaning and preparation for analysis. As the multiple selections were all included in a single field and separated by commas, recoding was done to make the data usable for analysis. The Lab research team separated each value to create two different variable types for each field: a hierarchical variable code and a series of dichotomous variables to capture each code associated with the recorded activity. The hierarchical classifications created by the Lab research team are in Appendix A. The recoding process involved classifying officer stop actions and results of stops by severity, with the most serious types superseding all others regardless of the number of codes present. For example, if both a chemical spray was used and a person was searched during the stop, then chemical spray was recorded as the most severe action taken during that stop, being “Use of Non-Lethal Force” for our analyses. For ease of analysis moving forward, the PAPD should explore the feasibility of fixing this multi-response issue by allowing 1 This includes individuals coded as Native American, Pacific Islander, or Unknown. A N A N A L Y SI S O F P OL I C E STO P D A TA / 4 D ES C R I P T I V E A N A LY S I S each response to be its own field. This will provide analytical clarity for any analyses moving forward and aid with interpretation and transparency in the public dashboard. These classifications were developed using a combination of legal reasoning, use-of-force policy frameworks, and empirical logic. Each action code was assigned to a broader category reflecting its relative severity, intrusiveness, or impact on individual liberty, informed by nationally recognized frameworks such as the National Institute of Justice (NIJ) use-of-force continuum and guiding principles from the Police Executive Research Forum (PERF). Actions involving physical control, weapon use, or evidence collection were assigned higher severity. The stop result hierarchy reflects the legal weight and practical consequences of each outcome for the stopped individuals (e.g., arrests involve the deprivation of liberty and are subject to the highest level of legal scrutiny vs. a warning or infraction). The classification also aligns with standard reporting and accountability frameworks used by state agencies and civil rights organizations to monitor law enforcement outcomes. These hierarchical frameworks are intended to support descriptive comparisons and statistical modeling by capturing the escalating legal, physical, or procedural implications of officer decisions. While the categorizations required subjective judgment, they were refined in consultation with agency partners to ensure they accurately reflect officer behavior. Spatial Data The PAPD provided beats, reporting districts, and neighborhood shapefiles. The RIPA stop data were geocoded using the PAPD-provided address location fields and spatially joined with beats, reporting districts, and neighborhood shapefiles. The address field in the RIPA data was typically represented at the hundred-block level, masking some location specificity. While valuable for identifying broad patterns, analyses using hundred-block address aggregation can create artificial clustering effects. Understandably, officers may need to identify the closest intersection or point of interest during some stops. However, this does lead to multiple stops being clustered at specific points and obscuring some of the micro-differences that more precise location reporting would provide. Recognizing this issue, the research team created additional spatial datasets to try and disentangle this issue and better represent the clustering of stop activity. This data collection issue and its implications for analysis are discussed further in the spatiotemporal analysis methodology section below. Methodology To better understand police enforcement actions and their impact on disparities, the first step is to describe the available data and examine its patterns and trends. Descriptive analysis involves summarizing outcome data to provide a clear and concise overview of key characteristics and patterns within a dataset, allowing for insights into the data's central tendencies, variability, and distribution (Witte and Witte, 2017). Descriptive analyses, including crosstabulations, help facilitate understanding of basic patterns and distributions in the data and cannot be used to explain or A N A N A L Y SI S O F P OL I C E STO P D A TA / 5 P A P D R I P A D A T A D A S H BOA R D R E V I E W V E IL O F D A R K N E S S A N A L Y S IS predict trends. Such analyses are typically a precursor to more complex statistical techniques (Shadish et al., 2002; Witte and Witte, 2017). In addition to the quantitative analysis of stop data, the project team conducted a focused review of the department’s public-facing RIPA dashboard. This review assessed the dashboard’s structure, content accuracy, user navigation, and visual presentation. The team evaluated how effectively the dashboard communicated stop data trends to different audiences and compared its design and functionality against similar tools used by peer agencies. Findings from this assessment informed a set of targeted recommendations aimed at enhancing the dashboard’s clarity, accessibility, and utility for both public transparency and internal decision-making. The veil of darkness technique, developed by Grogger and Ridgeway (2006), tests for potential racial and ethnic bias in vehicle stops by exploiting variations in natural lighting conditions. This method compares the racial distribution of vehicle stops made during daylight versus in darkness during the "inter-twilight period"—times of day that are light during certain parts of the year and dark during others due to seasonal variation in sunset times. The underlying logic is that if officers use race in stop decisions, this behavior would be more difficult during darkness when drivers' race is less visible, leading to different racial distributions between daylight and darkness stops made during the same clock times. The implementation of the veil of darkness test involves several key analytical steps. First, the appropriate inter-twilight window based on local sunset times throughout the year is identified. This typically includes those evening hours transitioning between light and dark as seasons change. Next, stops within this window are coded for lighting conditions based on precise sunset times for each date. Finally, a statistical analysis (typically logistic regression) compares the odds of non-white drivers being stopped during daylight versus darkness while controlling for location, day of week, and seasonal patterns. The primary output of the analysis is an odds ratio that quantifies the relationship between lighting conditions and the racial distribution of stops. An odds ratio greater than 1.0 indicates that non-white drivers are likelier to be stopped during daylight hours, potentially suggesting bias in stop decisions. Conversely, a ratio of less than 1.0 indicates that non-white drivers are more likely to be stopped during darkness. The statistical significance of these ratios helps determine whether observed differences reflect systematic patterns rather than random variation. While the veil of darkness test represents a methodologically rigorous approach to analyzing potential bias, it relies on several key assumptions to consider when interpreting results. The method assumes that driving patterns by race and ethnicity remain consistent between daylight and darkness during comparable clock times and that the visibility of the driver by officers is the primary factor that changes between lighting conditions. Similarly, this analysis assumes that officers do not consider the condition or type of vehicle as a proxy for race (Baumgartner et al., A N A N A L Y SI S O F P OL I C E STO P D A TA / 6 S P A T IO T E MP O R A L A N A L Y S IS AN OVERVIEW OF THE RESEARCH 2017). Additionally, the analysis is limited to vehicle stops during inter-twilight, which may not represent patterns and practices in all enforcement activity. Furthermore, the increased presence of artificial lighting in some areas may reduce the distinction between daylight and nighttime conditions. Despite these limitations, the veil of darkness technique provides valuable insights by controlling for many confounding factors that complicate other approaches to measuring enforcement disparities. A series of geographic and time-of-day analyses examined the geographic and hourly patterns in officially recorded PAPD stops. First, all PAPD-provided official data sources that included addresses or latitude and longitude coordinates were geocoded to their nearest street segment using ArcGIS Pro, a geographical analysis software system. The geocoding results consistently demonstrated a hit rate of over 90%, well above the established standard of 85% (Chainey and Ratcliffe, 2005). Spatial data was analyzed using ArcGIS Pro to create density and heat maps to examine associations among PAPD activity datasets and the underlying environment.2 Administrative datasets that included incident time information were rounded to the nearest hour in a 24-hour period and visualized using radar plots to examine what times of day had the highest number of events to examine differences across police activity and racial and ethnic groups. 2 Due to the PAPD's reporting methods for location data, addresses are typically recorded at the hundred-block level (e.g., 1200 Main St rather than 1234 Main St.), which can create artificial hot spots of data as slightly different addresses may all be classified into a single point. Thus, an additional analytical step was taken to join all geocoded event points to Palo Alto street segments and then calculate the density of police activity per the street segment length to better approximate and visualize official police activity at the street segment level. A N A N A L Y SI S O F P OL I C E STO P D A TA / 7 PAPD DATA ANALYSIS RESULTS The Lab research team’s findings are presented below. They begin with descriptive statistics and then present the more sophisticated approaches discussed above. PAPD Stop Patterns PAPD’s monthly stops from January 2022 through December 2023 were examined, as shown below in Figure 1, revealing fluctuations in enforcement activity over time. During this time period, the number of monthly stops averaged 438 per month. The number of stops remained relatively stable for much of 2022 before increasing sharply in late 2022 and early 2023. The highest number of stops occurred in March 2023, with 686 recorded stops. After this peak, stop frequency declined throughout the remainder of 2023. Ridgeway, 2006; Smith et al., 2017; 2021). However, these advanced techniques often A N A N A L Y SI S O F P OL I C E STO P D A TA / 8 Figure 1- PAPD Monthly Stops (January 2022 to December 2023) Figure 2 shows stop patterns by racial and ethnic groups. The trends for these groups followed a general overall pattern, with an apparent increase in stops leading up to early 2023, peaking in March/April of 2023 and followed by a decline to roughly 100 monthly stops. Stops of individuals perceived as Black, Asian, and Middle Eastern/South Asian remained lower in absolute numbers and exhibited less pronounced fluctuations over time. The data suggest that the overall volume of stops and their distribution by race changed over time in ways that may be relevant for further examination. The period of increased stops in early 2023 and the subsequent decline may warrant a closer review to understand what factors coincided with these shifts. The department might consider examining whether the increased stop activity in early 2023 corresponds with specific enforcement initiatives, shifts in policing priorities, seasonal variations in community activity, or changes in staffing and resource deployment. Additional data, such as calls-for-service, crime incident reports, community events calendars, or officer deployment schedules, could be integrated into analyses to better contextualize these patterns. Linking stop data to other operational datasets would enable the department to determine whether these trends align with specific operational decisions or community dynamics, providing a deeper understanding of the observed fluctuations in enforcement activity. In addition, qualitative research via interviews or focus groups with PAPD staff and community members may reveal other critical contextual factors that may have impacted these patterns. A N A N A L Y SI S O F P OL I C E STO P D A TA / 9 Figure 2- PAPD Monthly Stops by Race/Ethnicity (January 2022 to December 2023) Figure 3 presents stops by police beat for 2022 and 2023. Beat 2 recorded the highest number of stops in both years and increased from 1,908 in 2022 to 2,636 in 2023, a 38% increase. Beat 3 also experienced an increase between 2022 and 2023, with stops increasing from 753 to 1,378, an 83% increase year over year. Beats 1 and 4 were relatively more stable, with Beat 1 increasing 9% from 860 to 936 stops, and Beat 4 decreasing 6% from 548 to 515. These differences suggest enhanced enforcement activity between 2022 and 2023 in Beats 2 and 3 but not in Beats 1 and 4. A N A N A L Y SI S O F P OL I C E STO P D A TA / 10 Figure 3- PAPD Stops by Police Beat (2022 vs. 2023) Figure 4 examines stop patterns by police beat and perceived race and ethnicity3. Across all beats and both years, individuals perceived as white or Hispanic experienced the highest number of stops. In Beat 2, the number of stops increased notably for nearly all racial and ethnic groups: stops of white individuals rose from 616 in 2022 to 832 in 2023 (a 35% increase); Hispanic individuals increased 40% from 604 to 844; Asian individuals from 149 to 326 (119% increase); and Middle Eastern/South Asian individuals from 130 to 251 (93% increase). Interestingly, stops of Black individuals in Beat 2 remained stable (290 to 292) across both years. In Beat 3, there were substantial increases in stops across all racial and ethnic groups, particularly for Asian and Middle Eastern/South Asians, which more than doubled with a 177% and 105% increase, respectively. In contrast, white (72%), Hispanic (87%), and Black (58% ) stops rose more moderately. The sharp increases in stop volume in Beats 2 and 3 may benefit from further exploration. In contrast, the relative stability or decline in Beats 1 and 4 suggests that there may be important contextual differences between the increasing and decreasing beats. Further examining these geographic differences could provide additional insight into the factors influencing stop distribution across Palo Alto (e.g., specific land uses, crime and call-for-service data, or PAPD enforcement priorities). It should be noted that these could also result from changes in reporting 3 Use caution when comparing across racial and ethnic groups, as each has a slightly different maximum value on the y-axis. A N A N A L Y SI S O F P OL I C E STO P D A TA / 11 practices between the two years and that additional years of data may display more stable trends moving forward. Figure 4- PAPD Stops by Police Beat and Race/Ethnicity (2022 vs. 2023) Figure 5 presents the reasons for stops and reveals that traffic violations accounted for the majority of stops, with 7,712 recorded instances (74%). The second most common reason was reasonable suspicion that an individual was engaged in criminal activity, which accounted for 1,848 stops. Stops for traffic violations occurred at more than four times the rate of the next highest category—stops based on reasonable suspicion of criminal activity (a ratio of just over 4:1). Stops related to outstanding arrest warrants was the next most common type at 489 stops, followed by consensual encounters (331 stops). A N A N A L Y SI S O F P OL I C E STO P D A TA / 12 Figure 5- PAPD Stops by Reason Provided (2022 to 2023) Figure 6 shows the two most frequent stop reasons by perceived race and ethnicity, revealing notable differences in the distribution of stops across demographic groups. The chart on the left shows that stops for traffic violations were most frequently associated with individuals perceived as Hispanic (34.2%) and white (31.3%), followed by Asian (14.6%), Black (10.3%), and Middle Eastern/South Asian (9.7%) individuals. The chart on the right shows a different distribution for stops based on reasonable suspicion of criminal activity. White individuals comprised the largest share (42.1%) of these stops, followed by Hispanic (28.4%) and Black (18.0%) individuals. Asian (7.1%) and Middle Eastern/South Asian (4.4%) individuals represented smaller proportions. These comparisons highlight differences in the racial and ethnic makeup of stops depending on the stated reason for the stop. To better understand what may be driving these differences, further analysis could examine whether geographic areas, types of encounters, or enforcement assignments are associated with each stop reason. An in-depth examination of these types of stops, specifically via case reports, interviews with officers, or analysis of additional administrative datasets (e.g., calls for service, suspect and victimization data), may provide additional insight into the causes of these patterns. A N A N A L Y SI S O F P OL I C E STO P D A TA / 13 Figure 6- PAPD Stops by Reason by Race/Ethnicity (2022 to 2023) PAPD Stop Outcomes STOP ACTIONS TAKEN BY RACE/ ETHNICITY 2022 -2024 An analysis of the most serious action undertaken during a stop is presented in Table 2. As mentioned in the methodology section, multiple actions could be taken within a stop, and actions were coded into a hierarchy of seriousness, with the most serious action being used (See Appendix A). The stop actions in Table 2 reveal that most stops resulted in no action being recorded, accounting for 7,525 of all stops across racial groups (excluding those in “Other” and “Multi-racial” categories). Notably, Asian (83.2%) and Middle Eastern (80.5%) individuals had the highest proportion of stops resulting in no action, while Black individuals had the lowest proportion at 60.3%. Among stops that did result in further action, detention and restraint was the most common type of action taken, occurring in 1,968 stops. Black individuals experienced the highest rate of detention and restraint (26.7%), compared to other groups, followed by white (19.7%), Hispanic (18.3%), Middle Eastern (14.0%), and Asian (13.4%) individuals. Search and seizure occurred in 682 stops, again disproportionately impacting Black individuals (9.6%) compared to white (7.6%), Hispanic (6.9%), Middle Eastern (2.7%), and Asian (1.9%) individuals. Other actions—including non- physical actions, firearm pointing, physical contact without force, and use of non-lethal force— occurred less frequently, with firearm pointing and use of non-lethal force being particularly rare. The observed differences in the proportion of stops leading to more intrusive outcomes, particularly detention and search activity among Black individuals, highlight areas where further exploration into enforcement practices and community interactions may be beneficial. A N A N A L Y SI S O F P OL I C E STO P D A TA / 14 Table 2- PAPD Actions Taken during Stop by Race/Ethnicity (2022 to 2023) Action Taken Stops Asian Black Hispanic Eastern White None 7,525 1,025 760 2,354 665 2,365 1,968 Search and Seizure 682 276 Firearm Pointing 18 15 7 Percentages denote the number of stops relative to the total number of stops in each group. STOP RESULTS BY RACE/ ETHNICITY More than half of the stops across these racial and ethnic groups (~57%4 during 2022 and 2023) resulted in no action or minimal intervention, including written and verbal warnings (see Table 3). Approximately 28% resulted in administrative actions, and 12% resulted in an arrest. Table 3 indicates that the distribution of stop results also varied by race. Again, stop results were re- classified into higher-order categories to provide clarity and improve analysis and interpretation5. Hispanic individuals (59.9%) and Middle Eastern individuals (58.7%) had the highest proportion of stops resulting in no action or minimal intervention, followed closely by Asian (55.4%), white (54.5%), and Black (53.6%) individuals. Administrative actions, which include citations and other formal processes, were more common among Asian individuals (36% of their stops) compared to different groups, followed by Middle Eastern (32%), white (27.2%), Hispanic (26.1%), and Black (20.8%) individuals. Arrests also occurred disproportionately among Black individuals (21.6%), compared to white (12.7%), Hispanic (12%), Middle Eastern (6.1%), and Asian (5.5%) individuals. The disparities in stops, particularly in relation to disparities seen in the actions taken during stops (e.g., firearm pointing, detention, and restraint) and the stop outcomes (e.g., arrests), highlight notable differences in the results of enforcement actions across demographic groups. The PAPD 4 Total stops for these five racial and ethnic groups equals 10,029, as it excludes the “Other” and “Multi-Racial” groups. Including those groups produced similar percentages. 5 212 stops were re-classified using the stop result hierarchy in Appendix A. A N A N A L Y SI S O F P OL I C E STO P D A TA / 15 may want to further investigate, monitor, and report these demographic trends to examine changes and patterns. Qualitative research, including interviews or focus groups with officers and community members, coupled with policy review, may provide additional context to these patterns. Additional administrative datasets could allow for more in-depth and complex statistical analysis, especially if the department could link stop data with calls for service, arrest, and crime data to include additional contextual information about the circumstances surrounding stops. Table 3- PAPD Stop Result by Race/Ethnicity (2022 to 2023) Stop Result Stops Asian Black Hispanic Eastern White No Action or Minimal Intervention 5682 683 (55.4%) 677 (53.6%) 1971 (59.9%) 485 (58.7%) 1866 (54.5%) Administrative Actions 2757 Arrests 1220 Interventions 370 35 (2.8%) 50 (4%) (2%) (3.3%) (5.6%) number of stops relative to the total stops in each group. PERCENTAGE OF MONTHLY STOPS WITH SEARCH CONDUCTED An examination of search activity across stops, presented in Figure 7, reveals fluctuations in the proportion of stops that resulted in a search. Analyzing monthly trends in stops involving searches is crucial for understanding variations in police practices over time, identifying potential disparities, and evaluating the consistency and effectiveness of search decisions. Examining data across a two-year period provides sufficient detail to detect meaningful patterns and trends, ensuring a more comprehensive analysis than shorter time frames. On average, 16.8% of stops involved a search, though this percentage varied over time. The highest search rate occurred in August 2022, when 27.3% of stops included a search. While the search rate showed periods of increase and decline, it generally remained within a range of 10% to 20% in most months. When disaggregated by race and ethnicity, as in Figure 8, Black individuals consistently experienced the highest monthly percentage of stops involving searches, with frequent peaks approaching or exceeding 30%. In comparison, search rates for Hispanic and white individuals were lower but often mirrored each other closely, typically between 10% and 25%. Asian and Middle Eastern/South Asian individuals consistently had the lowest search rates, generally below 15%. A N A N A L Y SI S O F P OL I C E STO P D A TA / 16 Supporting these findings, the average monthly search rate over the two years varied significantly by race and ethnicity. Black individuals had the highest average monthly search rate of 27.1%, substantially exceeding those of other groups. White and Hispanic individuals experienced searches on ~18% of their average monthly stops, while Asian and Middle Eastern/South Asian individuals were subject to searches on roughly 8% of their monthly stops. For Black, white, and Hispanic individuals, their monthly search percentage exceeded their average at least 33% of the months during the two-year period, while Asian and Middle Eastern/South Asian individuals only exceeded their monthly average 25% of the time. These fluctuations indicate that PAPD may want to continuously monitor search and hit rates over time to ensure that spikes and troughs in these patterns reflect strategic public safety needs and priorities. Figure 7- PAPD Percentage of Monthly Stops with Search Conducted (January 2022 to December 2023) A N A N A L Y SI S O F P OL I C E STO P D A TA / 17 Figure 8- PAPD Percentage of Monthly Stops with Search Conducted by Race/Ethnicity (January 2022 to December 2023) HIT RATE ANALYSIS BY RACE/ ETHNICITY The search hit rate, or the percentage of searches that resulted in the discovery of contraband or evidence, varied across racial groups and is presented in Table 4. Searches of white individuals had the highest hit rate at 12.4%, followed by Hispanic individuals at 10.3%, Black individuals at 8.3%, Asian individuals at 5.7%, and Middle Eastern/South Asian individuals at 4.9%. Searches of white individuals were more likely to yield contraband or evidence compared to searches of other racial and ethnic groups. Figure 9 provides additional insights by comparing search effectiveness, or "hit rates," for non- white groups relative to white individuals across police beats. The diagonal dashed line represents equal hit rates between white individuals and another racial/ethnic group. Points above the diagonal line indicate that the non-white group had a higher search hit rate than white individuals in that police beat; points below the line indicate that white individuals had a higher hit rate than the non-white group6. For Asian individuals, only police Beat 1 showed higher hit rates compared to white individuals during 2022 and 2023. For Black individuals, three out of four beats (Beats 1, 2, and 3) showed 6 Where points fall along the X-axis for the non-white groups, this indicates that no searches were conducted for that group in that beat. A N A N A L Y SI S O F P OL I C E STO P D A TA / 18 points clearly below the parity line, indicating consistently lower search hit rates than white individuals. The pattern for Hispanic individuals showed two beats (Beats 1 and 2) below the diagonal line, indicating lower hit rates than white individuals, with Beats 3 and 4 slightly above parity. Finally, the pattern for Middle Eastern/South Asian individuals had all hit rates falling below the parity line, although 3 of these were the result of zero searches in those beats. Taken collectively, Figure 9 indicates that searches involving Black, Hispanic, and particularly Middle Eastern/South Asian individuals typically yielded lower hit rates compared to searches of white individuals across most police beats. This pattern raises questions about the effectiveness of searches across different racial groups and whether search criteria result in different success rates in discovering contraband or evidence. Again, qualitative research or the inclusion of additional quantitative administrative datasets could provide important insights into the context surrounding stop and search activity by PAPD. Table 4- Search Hit Rates by Race/Ethnicity (2022 to 2023) Race/Ethnicity Search Hit Rate Black 8.3% Middle Eastern/South Asian 4.9% A N A N A L Y SI S O F P OL I C E STO P D A TA / 19 Figure 9- PAPD Search Hit Rates by Police Beat and Race/Ethnicity (2022 to 2023) VEIL OF DARKNESS ANALYSIS The Veil of Darkness (VoD) analysis is a statistical method used to assess whether drivers of different racial groups are stopped at different rates depending on whether it is daylight or nighttime. The underlying idea is that during low-light conditions, such as after sunset, it becomes more difficult for officers to visually identify a driver’s race before initiating a stop. If racial disparities in stop rates are driven by visual identification of race, we would expect non-white drivers to be stopped more frequently during daylight (when race is visible) and less frequently after dark (when race identification is difficult). This difference in stop rates between daylight and darkness can serve as evidence of racial bias in traffic enforcement. Important limitations of VoD analysis should be noted. The method assumes that driver behavior, traffic patterns, and the likelihood of committing traffic violations are similar across racial and ethnic groups during both daylight and nighttime hours. If different groups have systematically different driving patterns, locations, or behaviors during these time periods, this could confound the results. Additionally, VoD analysis cannot establish direct causation and should be interpreted alongside other evidence when assessing potential bias in traffic enforcement. A N A N A L Y SI S O F P OL I C E STO P D A TA / 20 The VoD analysis results presented below use an odds ratio to compare the likelihood of a traffic stop occurring under different lighting conditions for drivers of different racial groups. The odds ratio represents how much more or less likely a driver is to be stopped during darkness compared to daylight. An odds ratio of 1.0 means no difference between daylight and darkness; an odds ratio below 1.0 indicates lower likelihood of being stopped during darkness (suggesting higher stop rates during daylight when race is visible), while an odds ratio above 1.0 indicates higher likelihood of being stopped during darkness. When the analysis of non-white driver stops show odds ratios significantly below 1.0, this pattern suggests that racial cues visible during daylight may be influencing stop decisions. Table 5 below displays the results of the Veil of Darkness tests for all race and ethnicity groups. For Black drivers (compared to white drivers), the odds ratios of 0.62 (without controlling for police beat) and 0.54 (when controlling for police beat) are both statistically significant (p < .05) as shown in Table 5. This means that Black drivers were significantly less likely to be stopped at nighttime than during daytime, with the likelihood of being stopped during darkness being about 38% lower (0.62) to 46% lower (0.54) than during daylight. The odds ratio of 0.54 after controlling for police beat suggests an even more substantial effect. Put differently, this indicates that Black drivers experience higher stop rates during daylight hours when their race is more easily identifiable to officers. This difference in stop rates between lighting conditions warrants further investigation to understand the underlying factors contributing to this pattern. In contrast, the odds ratios for Hispanic and white drivers were 0.86 in both models, and these results were not statistically significant. Similarly, Asian drivers showed odds ratios close to 1.0 (0.96 without control, 0.84 with control), neither statistically significant, indicating no strong relationship between lighting condition and stops. Middle Eastern/South Asian individuals also had odds ratios close to 1.0 (1.02 in both models), demonstrating no significant relationship between darkness conditions and their likelihood of being stopped. This suggests that there was no meaningful difference in the likelihood of stops for other non-white and white drivers based on whether the stop occurred during daylight or nighttime hours. The significant difference in odds ratios observed for Black drivers but not for Hispanic drivers may warrant further examination. While the VoD analysis does not establish direct causation, the significantly lower odds of Black drivers being stopped during darkness (when race identification is difficult) compared to daylight represents a notable pattern that differs from other racial groups analyzed. Palo Alto PD staff and city officials have noted that the city experiences significant commuter traffic that substantially changes the population dynamics throughout the day, which may contribute to differences in stop patterns between daylight and nighttime hours. This commuter presence could influence the demographic composition of drivers on the road during different time periods, potentially affecting the interpretation of VoD results. Additional review of stop activity via case reports and qualitative interviews or focus groups with officers may help to understand what factors contribute to these observed differences. A N A N A L Y SI S O F P OL I C E STO P D A TA / 21 Table 5- Veil of Darkness Results (2022 to 2023) Race Control for Police Beat Sample Size Odds Ratio Black and White Drivers No 1488 0.62* Black and White Drivers Yes 1488 0.54* Hispanic and White Drivers No 1093 0.85N.S. Hispanic and White Drivers Yes 1093 0.86N.S. Asian and White Drivers No 669 0.96N.S. Asian and White Drivers Yes 669 0.84N.S. No 663 1.02N.S. 663 1.02N.S. Spatiotemporal Analysis STOPS BY TIME OF DAY The circular bar chart in Figure 10 visualizes the distribution of stops across different hours of the day. The radial design means that each bar extends outward from the center, representing the total number of stops during each hour. The labels along the outer edge of the chart indicate the hour of the day, ranging from 0:00 (midnight) to 23:00 (11 PM). Longer bars correspond to more stops during that hour, while shorter bars indicate fewer stops. This format allows for an intuitive comparison of stop frequency across different times of the day. Figure 10 shows a peak in stop activity between 22:00 (10 PM) and 0:00 (midnight), with the highest number of stops occurring around 23:00 (11 PM). Stops declined after midnight through the early morning hours. The number of stops rose during the morning and increased steadily through the afternoon. A more pronounced increase occurred in the evening, leading to the late-night peak. Several factors could help understand this time-of-day pattern, such as enforcement strategies, changes in officer deployment schedules, or variations in traffic and pedestrian activity and behavior by time-of-day. The sharp increase in stops in the late evening, followed by a steep A N A N A L Y SI S O F P OL I C E STO P D A TA / 22 decline after midnight, may be worth further exploration to assess whether this is a result of enforcement priorities, shift schedules, policy decisions, or calls for service that contribute to this observed pattern. Figure 10- PAPD Stops by Time of Day (2022 to 2023) The set of circular bar charts in Figure 11 visualizes the distribution of stops across different hours of the day, broken down by race and ethnicity. While all groups experience a notable concentration of stops in the late evening, primarily between 9:00 PM and midnight, the shape and intensity of these peaks differ across racial and ethnic categories. For instance, Hispanic individuals experience not only the highest total number of stops but also the most pronounced late-night concentration. White individuals similarly experience a notable late-evening peak, but their stop distribution also extends further into early morning hours compared to other groups. The distributions for Black and Middle Eastern/South Asian individuals reflect similar late-evening peaks, but these groups experience comparatively fewer stops overall. Asian individuals display a distinct pattern with fewer total stops and a narrower late-night peak. The lowest stop volumes for all racial and ethnic groups occurred in the early morning hours, particularly between 5:00 AM and 10:00 AM, before increasing throughout the afternoon and peaking again in the late evening. These differing temporal patterns raise questions about the underlying operational decisions, community behaviors, or geographic distributions that drive stop activity at different hours for these groups. Exploring these hourly stop distributions further by integrating additional data such as calls-for-service patterns, officer shift assignments, local event calendars, and geographic deployment strategies could shed valuable light on the observed disparities and similarities. Understanding these nuances may assist the department in evaluating enforcement strategies to ensure alignment with both community safety priorities and equitable policing objectives. A N A N A L Y SI S O F P OL I C E STO P D A TA / 23 Figure 11- PAPD Stops by Time of Day and Race/Ethnicity (2022 to 2023) PAPD STOP DENSITY HEATMAPS Figure 12 displays the geographic distribution of stops conducted by the PAPD across police beats from January 2022 through December 2023. The heat maps in this report are color-coded to represent ranges of stop density, with red and yellow areas representing the higher concentration of stops and blue and green areas indicating areas with lower stop density. The visualization reveals that stops were heavily concentrated in Beat 1 and Beat 2. The most prominent hotspots were in Beat 2 near the border with East Palo Alto and Menlo Park, the College Terrace area in Beat 1, and locations within Beat 2 closer to the border with Beat 3. These areas showed the highest densities of stops, suggesting significant enforcement activity in these regions. Beats 3 and 4 exhibited notably lower stop densities, with only a few scattered points of moderate activity. This suggests that enforcement activity was concentrated in specific zones. The spatial distribution of stops may warrant further examination to determine whether stop patterns correlate with factors such as traffic volume, business districts, residential areas, or specific law enforcement initiatives in these beats (as the geographic context varies by beat observed). Additionally, the concentration of stops in particular A N A N A L Y SI S O F P OL I C E STO P D A TA / 24 locations could be relevant for discussions on resource allocation and community engagement efforts within Palo Alto. Figure 12- PAPD Stop Density Heat Map In contrast, Figure 13 offers a more detailed, zoomed-in perspective on the specific hotspots identified within these beats, providing greater clarity on the intensity and location of police enforcement activities. The detailed heatmap reveals highly concentrated enforcement in specific intersections and blocks, notably near downtown Palo Alto in Beat 2, and a significant cluster at a major intersection in the southern portion of Beat 1. Moderate clusters in Beat 2 also emerge distinctly, highlighting locations that received consistent but less intense enforcement attention. Conversely, enforcement activities within Beats 3 and 4 appear relatively sparse and dispersed, suggesting that stops in these areas may be related more to routine patrols rather than targeted enforcement initiatives. This detailed visualization reinforces the key insight that stop patterns are not evenly distributed throughout Palo Alto, prompting important questions regarding the factors behind concentrated enforcement in specific areas. However, interpreting these patterns requires caution due to PAPD data recording practices. Addresses often reflect approximate or commonly reported locations (e.g., intersections or landmarks), potentially inflating apparent stop densities in these mapped A N A N A L Y SI S O F P OL I C E STO P D A TA / 25 hotspots. To enhance reliability and interpretability, stops were also aggregated at the street- segment level, allowing for a more nuanced understanding of enforcement concentrations along specific streets rather than isolated hotspots. The street-level analysis presented in Figures 14 and 15 can further inform discussions around resource allocation, strategic deployment, and community engagement by clarifying precisely which areas experience the most frequent and sustained police attention. Figure 13- PAPD Stop Density Detailed Heat Map PAPD STOP DENSITY BY STREET SEGMENT Figure 14 visualizes the density of stops per 1,000 feet of roadway in Palo Alto from 2022 to 2023, highlighting which streets experienced the highest concentration of stops. Aggregating the count of stops to specific street segments minimizes reporting issues and provides a more accurate representation of where stops occur at the street level. Streets are color-coded to reflect different levels of stop density, with darker red lines indicating areas with the highest concentration of stops, moderate red lines representing streets with mid-level stop density, and lighter-colored streets showing areas with lower stop activity. The data reveal that stops were most heavily concentrated along major corridors, such as University Avenue, Embarcadero Road, and El Camino Real. These thoroughfares, known for their high traffic volume and commercial activity (according to discussions with PAPD representatives), accounted for a significant portion of the stops recorded in the city. The central and northern parts of Palo Alto, particularly near downtown and the Stanford area, also exhibited high stop densities, A N A N A L Y SI S O F P OL I C E STO P D A TA / 26 suggesting that enforcement efforts were focused along these busy streets. In contrast, the southern and more residential areas of Palo Alto saw noticeably lower stop densities. Many neighborhood streets experienced minimal stop activity compared to the main arterial roads. This pattern suggests most stops occurred along primary traffic routes rather than within residential zones. This map provides valuable insight into the distribution of enforcement activity across Palo Alto by illustrating the geographic concentration of stops. The precise concentration of stops along major roads, while residential streets saw fewer stops, highlights potential areas for PAPD to ensure that they align their stop activity with those areas that maximize public safety and community needs. In addition, coupling this analysis with geographic information about calls for service and crime would allow the PAPD to ensure that their stop activity is aligned with the underlying public safety needs of the community. A N A N A L Y SI S O F P OL I C E STO P D A TA / 27 Figure 14- PAPD Stop Density by Street Segment Finally, the Lab research team examined whether the concentration of nightly stops seen in the time-of-day analyses resulted in different spatial patterns along street segments. Figure 15 provides a breakdown of stop density by street for only those stops occurring between 9 PM and 12 AM from January 2022 to December 2023. Stop activity during this period was concentrated along Embarcadero Road, University Avenue, and sections of El Camino Real, which were indicated by PAPD staff to be among the busiest sections of roadway in the jurisdiction. Additional areas with moderate stop densities appeared along Hamilton Avenue, Channing Avenue, and Churchill Avenue. These roads likely represent high-traffic areas during nighttime hours, with greater pedestrian and vehicle activity. Compared to the broader stop density map covering all hours, this time-specific analysis highlights how late- A N A N A L Y SI S O F P OL I C E STO P D A TA / 28 night enforcement efforts were concentrated along major thoroughfares but with a somewhat different distribution. While stops remained clustered along key roads, certain streets with high stop densities in the overall analysis showed reduced activity at night, suggesting variations in enforcement or travel patterns throughout the day. The concentration of stops along specific corridors between 9 PM and 12 AM may warrant further exploration into the factors contributing to this pattern, including nightlife activity, traffic enforcement priorities, and the relationship between stop frequency and areas with high vehicle or pedestrian presence during these hours. Additional data sources from other city and state agencies may highlight the types of ambient population movement that likely increase the likelihood of stops in these areas. Other types of PAPD administrative data may also provide helpful context around the general public safety environment in these areas during these times that may impact the probability of a stop being made. A N A N A L Y SI S O F P OL I C E STO P D A TA / 29 Figure 15- PAPD Stop Density by Street Segment (9PM to Midnight) A N A N A L Y SI S O F P OL I C E STO P D A TA / 30 S T R A T E G I C R EC O M M E N D A T I O N S F O R D A S H B O A R D EN H A N C EM EN T PAPD RIPA Dashboard Analysis In August 2024, PAPD also asked the Knowledge Lab to provide an assessment of its public RIPA data dashboard (https://www.paloalto.gov/Departments/Police/Public-Information-Portal/Racial- and-Identity-Profiling-Act-RIPA-Data ). The Lab’s examination found that while the dashboard effectively presents required RIPA data, significant opportunities exist to transform it from a basic reporting tool into a valuable resource for both community members and department personnel. Specifically, improvements in usability, audience clarity, and contextual analysis would be beneficial. The current dashboard is primarily a data repository rather than an analytical tool, with minimal contextual information to guide interpretation. The following section outlines recommendations to enhance the dashboard’s effectiveness. DEFINE PURPOSE AND AUDIENCE As of the writing of this report, the RIPA dashboard lacks clarity regarding its intended audience and purpose, increasing the risk of misinterpretation of complex policing data. The absence of explicit interpretive guidance leaves stakeholders uncertain about how to appropriately use and understand the data. Recommendations: • Clearly articulate the dashboard's purpose on the landing page, specifying whether it is intended primarily for public transparency, community education, internal analysis, or a combination thereof. • Develop differentiated user pathways that guide stakeholders, such as community members, researchers, and department personnel, to visualizations and contexts most relevant to their needs. • Create separate, secure interfaces for internal operational analyses distinct from public transparency dashboards. • Provide interpretive text that clearly explains both what the data show and its inherent limitations, particularly in interpreting disparities or enforcement outcomes. • Clarify how dashboard insights could support strategic departmental goals and operational decision-making, clearly communicating this alignment within the dashboard's stated purpose. ENHANCE DATA COMPREHENSIVENESS AND CONTEXT While the dashboard effectively presents raw stop counts, the absence of comparative benchmarks or analytical context limits users' ability to interpret the data meaningfully. Additionally, certain data fields, such as "Action Taken" and "Multiple Basis" for searches, present ambiguity or redundancy, further complicating interpretation. A N A N A L Y SI S O F P OL I C E STO P D A TA / 31 Recommendations: • Incorporate carefully selected benchmarks to contextualize stop data, such as traffic patterns, calls-for-service data, or at least population demographics, accompanied by clear caveats explaining the limitations of these benchmarks. • Undertake thorough data cleaning and pre-processing, specifically addressing issues such as inconsistent "Action Taken" fields and the overly broad "Multiple Basis" category. • Incorporate dynamic, trend-focused visualizations that clearly illustrate changes over time rather than static annual snapshots. • Add geographic visualizations to display spatial patterns of enforcement clearly. • Include detailed methodological explanations describing how RIPA data are collected and processed, and the associated limitations for interpretation. • Allow users to download raw data for independent analysis and consider automating more frequent data updates to ensure timeliness and relevance. • Integrate performance metrics that directly relate dashboard findings to departmental objectives, facilitating the measurement of strategic policing goals. IMPROVE DASHBOARD DESIGN AND USER EXPERIENCE The existing two-tab structure (General and Individual) presents navigation challenges, and the current data presentation emphasizes dense tables over intuitive, interactive visualizations. Filtering capabilities are limited and do not persist across dashboard views, inhibiting comprehensive data exploration. Recommendations: • Restructure the dashboard around clearly defined, purpose-specific tabs, such as Overview, Stops, Searches, Outcomes, and Demographics. • Enhance filtering capabilities, ensuring user selections persist across multiple dashboard views to facilitate deeper data exploration. • Prioritize intuitive visualizations, such as interactive charts, graphs, and maps, complemented by interpretive context rather than relying solely on dense data tables. • Create interactive elements allowing users to explore targeted questions of interest. • Provide clear definitions of technical terms directly within the dashboard interface to support accurate interpretation. • Ensure the dashboard design is responsive and accessible across all device types, including mobile platforms. A N A N A L Y SI S O F P OL I C E STO P D A TA / 32 IMPLEMENTATION CONSIDERATIONS Effective implementation of these enhancements requires a commitment beyond technical development. They could be incorporated more broadly into PAPD's strategic priorities to further PAPD’s transparency and accountability efforts. Recommendations: • Engage community stakeholders throughout the dashboard redesign process to ensure the tool addresses public information needs as appropriate. • Develop a comprehensive communication strategy to educate users on the dashboard’s purpose, data limitations, and appropriate interpretation. • Establish regular review procedures to monitor and address data quality issues proactively. • Implement formal feedback mechanisms enabling users to suggest continuous improvements. • Clearly link dashboard data to broader departmental transparency and accountability initiatives. • Establish internal processes to routinely leverage dashboard data for operational decisions, ensuring alignment between enforcement practices and broader departmental goals. By implementing these recommendations, PAPD can significantly enhance its RIPA dashboard, transforming it into a dynamic resource that effectively communicates policing practices, fosters community trust through transparency, and provides actionable insights for continuous departmental improvement. A N A N A L Y SI S O F P OL I C E STO P D A TA / 33 G E N E R A L TR E N D S A N D P A T T E R N S SUMMARY OF FINDINGS An analysis of the Palo Alto Police Department's administrative data from January 2022 to December 2023 reveals several important patterns in the nature and distribution of PAPD activity. These findings, derived from various analytical approaches, provide insights into general enforcement patterns and traffic citations, including potential disparities that warrant attention. The analyses conducted by the Knowledge Lab can help inform the Palo Alto Police Department's strategic priorities and long-term planning as the department continues to collect this data (2024 data was not yet available at the time of analysis for this report). • After relative stability in 2022, the number of monthly stops significantly increased in early 2023, followed by a notable decrease in the second half of the year. An examination of stops by PAPD beats shows that the increase in early 2023 was driven by Beats 2 and 3. Beat 2 experienced a 38% increase in stops, and Beat 3 experienced an 83% increase between 2022 and 2023. Stops in Beats 1 and 4 were relatively more stable, increasing only by 9% and 6% during that period, suggesting enhanced enforcement activity in Beats 2 and 3 but not in Beats 1 and 4. • Of the 10,509 stops included in this analysis, 8,686 (over 83%) were recorded as resulting from officer-initiated activity rather than in response to a call. • When PAPD stops were examined by the reason for the stop, 74% were found to be for traffic violations during the two-year period. The second most frequent reason was reasonable suspicion that the person engaged in criminal activity at 18% of stops. • When PAPD stops were examined by the actions taken during the stop, most stops (over 70%) involved no action being recorded. Asian (83.2%) and Middle Eastern (80.5%) individuals had the highest proportion of stops resulting in no action, while Black individuals had the lowest proportion at 60.3%. • When PAPD stops were examined by the result of the stop, more than half of stops (~57%) resulted in no action or minimal intervention, including written and verbal warnings; 28% of stops resulted in administrative actions, and 12% resulted in an arrest. Disaggregating stop results by race/ethnicity shows that stops involving Black individuals were more likely to end with an arrest, and stops involving Asian individuals were more likely to end with a citation. • On average, 16.8% of stops involved a search during 2022 and 2023. Looking at monthly data, most months saw 10% to 20% of stops involving a search, with a peak occurring in August 2022, when 27.3% of stops included a search. • While monthly stops involving Black individuals generally had the highest percentage of searches (often over 30%) compared to other groups, stops involving Black individuals resulted in lower hit rates than white or Hispanic individuals. Searches of white individuals A N A N A L Y SI S O F P OL I C E STO P D A TA / 34 had the highest hit rate at 12.4%. Searches of white individuals were more likely to yield contraband or evidence compared to searches of other racial and ethnic groups. • Veil of darkness tests (as described on page 5) indicated that Black drivers were significantly less likely to be stopped during nighttime hours compared to daytime stops, relative to white drivers. No meaningful difference was found in the likelihood of stops for Hispanic and white drivers based on whether the stop occurred during daylight or nighttime hours. • The majority of stops occurred during the evening hours. Stop activity peaked between 10 PM and midnight, with the highest number of stops occurring around 11 PM. This pattern was maintained across all groups when looking at the time of stops by race and ethnicity. • The shifting concentration of stops along specific corridors between 9 PM and 12 AM may warrant further exploration into the factors contributing to this pattern, including nightlife activity, traffic enforcement priorities, and the relationship between stop frequency and areas with high vehicle or pedestrian presence during these hours. RECOMMENDATIONS The City asked the Knowledge Lab to provide recommendations based on the data analysis and findings of this work. The following recommendations are grounded in the findings from the stop data audit, analysis of enforcement patterns, and review of the department's public-facing dashboard. They reflect best practices in law enforcement transparency, data quality, and performance management. The goal is to help the Palo Alto Police Department strengthen its data infrastructure, improve public accountability, and align enforcement activities with community needs. DATA INFRASTRUCTURE AND COLLECTION Expand and Improve Data Collection • Consider recording all police-citizen encounters: It is recommended that PAPD establish a process for documenting all police-public contacts, including those that do not result in citations or searches, to provide a complete picture of PAPD activity. • Detailed traffic violation coding: It is recommended that PAPD consider recording specific traffic violation types rather than general categories to support more precise analysis of stop patterns and enforcement priorities. • Improve accuracy and completeness of key fields: It is recommended that PAPD consider enhancing data entry quality, particularly for stop reasons, outcomes, and officer actions, to reduce ambiguity and support clearer interpretation by recording each action or stop reason/outcome in their own fields. The current data structure allows for multiple selections but does not separate the selections from one another, resulting in added complexity to interpret or analyze the data. A N A N A L Y SI S O F P OL I C E STO P D A TA / 35 • Improve location data reporting: It is recommended that PAPD consider recording exact addresses or intersections where stops occur to improve spatial analysis and limiting the use of general landmarks or business names to avoid clustering unrelated stops. • Create integrated operational datasets: It is recommended that PAPD consider linking stop data with calls-for-service, crime reports, and arrest records. This will support more holistic and sophisticated analysis of officer activity and improve understanding of whether stops are aligned with crime prevention and community safety objectives. STRATEGIC MONITORING AND ANALYSIS Implement Ongoing Trend Monitoring • Track changes over time: It is recommended that PAPD consider establishing routine internal monitoring of stop volumes, reasons, outcomes, and demographic breakdowns to detect meaningful changes or emerging trends and utilizing the current dashboard and RIPA data for operational decision-making and monitor trends in stops alongside other key department metrics. • Identify outliers and anomalies: It is recommended that PAPD consider using the ongoing monitoring mechanisms (e.g., dashboards and regular internal meetings) and data analysis to flag significant shifts in stop activity (e.g., surges in searches, increases in nighttime stops, or demographic disparities in outcomes) that may require operational or policy review. • Evaluate alignment with community priorities: It is recommended that PAPD consider periodically assessing whether stop activity corresponds to known crime or traffic concerns, helping ensure that enforcement is strategic and responsive to current trends and community needs. PUBLIC TRANSPARENCY AND DASHBOARD ENHANCEMENTS Improve the Utility of the RIPA Dashboard • Clean and pre-process data before publication: It is recommended that PAPD consider standardizing values and resolving inconsistencies or multiple coding options in key fields before presenting data on public dashboards to minimize confusion and complexity. • Enhance visualizations and functionality: It is recommended that PAPD consider improving interactivity and clarity of data displays and including filters that persist across views, incorporate more intuitive, engaging graphs and maps, and include explanatory text to guide interpretation. • Expand methodological documentation: It is recommended that PAPD consider providing clear explanations on the dashboard of how RIPA data is collected, processed, and where it is limited, helping community users understand what the data can and cannot say. COMMUNITY ENGAGEMENT AND QUALITATIVE RESEARCH A N A N A L Y SI S O F P OL I C E STO P D A TA / 36 Integrate Officer and Community Perspectives • Conduct interviews or focus groups: It is recommended that PAPD consider engaging officers and community members to explore perceptions of stop activity, contextual factors influencing enforcement decisions, and opportunities for improvement. These perspectives can complement quantitative analysis and provide deeper insight into the meaning behind the numbers. These recommendations will position PAPD to make its stop data more actionable, interpretable, and meaningful. Enhanced data infrastructure and monitoring can support more effective internal decision-making. Improving transparency and community engagement will build public trust and ensure that policing practices are equitable and evidence-based. A N A N A L Y SI S O F P OL I C E STO P D A TA / 37 REFERENCES Baumgartner, F. R., Christiani, L., Epp, D. A., Roach, K., & Shoub, K. (2017). Racial Disparities in Traffic Stop Outcomes. Duke Forum for Law & Social Change, 9(21). Chainey, S., & Ratcliffe, J. H. (2005). GIS and Crime Mapping. Chichester, West Sussex, England: Wiley. Geller, A., Goff, P. A., Lloyd, T., Haviland, A., Obermark, D., & Glaser, J. (2021). Measuring Racial Disparities in Police Use of Force: Methods Matter. Journal of Quantitative Criminology, 37(4), 1083–1113. https://doi.org/10.1007/s10940-020-09471-9 Grogger, J., & Ridgeway, G. (2006). Testing for Racial Profiling in Traffic Stops From Behind a Veil of Darkness. Journal of the American Statistical Association. https://doi.org/10.1198/016214506000000168 Knox, D., & Mummolo, J. (2020a). Making Inferences About Racial Disparities in Police Violence. Proceedings of the National Academy of Sciences, 117(3), 1261–1262. https://doi.org/10.1073/pnas.1919418117 Knox, D., & Mummolo, J. (2020b). Toward a General Causal Framework for the Study of Racial Bias in Policing. Journal of Political Institutions and Political Economy, 1(3), 341–378. Neil, R., & Winship, C. (2019). Methodological Challenges and Opportunities in Testing for Racial Discrimination in Policing. Annual Review of Criminology, 2(1), 73–98. https://doi.org/10.1146/annurev-criminol-011518-024731 Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and Quasi-Experimental Designs for Generalized Causal Inference. (pp. xxi, 623). Houghton, Mifflin and Company. Smith, M. R., Rojek, J. J., Petrocelli, M., & Withrow, B. (2017). Measuring Disparities in Police Activities: A State of the Art Review. Policing: An International Journal of Police Strategies & Management, 40(2), 166–183. https://doi.org/10.1108/PIJPSM-06-2016-0074 Smith, M. R., Tillyer, R., Lloyd, C., & Petrocelli, M. (2021). Benchmarking Disparities in Police Stops: A Comparative Application of 2nd and 3rd Generation Techniques. Justice Quarterly, 38(3), 513– 536. https://doi.org/10.1080/07418825.2019.1660395 Swencionis, J. K., & Goff, P. A. (2017). The Psychological Science of Racial Bias and Policing. Psychology, Public Policy, and Law, 23(4), 398–409. https://doi.org/10.1037/law0000130 Withrow, B. L. (2004). A Comparative Analysis of Commonly Used Benchmarks in Racial Profiling: A Research Note. Justice Research and Policy, 6(1), 71–92. https://doi.org/10.3818/JRP.6.1.2004.71 Witte, R. S., & Witte, J. S. (2017). Statistics. Wiley. A N A N A L Y SI S O F P OL I C E STO P D A TA / 38 C L A S S IF IC A T I O N S C HE M E FO R A C T I O NS T A K E N APPENDIX A: RE-CLASSIFYING PAPD RIPA CODES USE OF FORCE ( LETHAL) Code 9: Firearm discharged or used POINTING OF FIREARM Code 8: Firearm pointed at a person USE OF FORCE (NON- LETHAL) Code 7: Canine removed from vehicle or used to search Code 10: Electronic control device Code 11: Impact projectile discharged or used (e.g., blunt impact projectile, rubber bullets, or bean bag) Code 12: Canine bit or held person Code 13: Baton or other impact weapon used Code 14: Chemical spray used (e.g., pepper spray, mace, or other chemicals) DETENTION AND RESTRAINT Code 4: Curbside detention Code 5: Handcuffed or flex cuffed Code 6: Patrol car detention SEARCH AND SEIZURE Code 18: Search of person was conducted Code 20: Search of property was conducted Code 21: Property was seized Code 22: Vehicle impounded PHYSICAL CONTACT WITHOUT FORCE Code 2: Person removed from vehicle by physical contact Code 15: Other physical or vehicle contact A N A N A L Y SI S O F P OL I C E STO P D A TA / 39 C L A S S I FI C A T I O N S C HE M E FO R R E S U L T O F T HE S T O P NON -PHYSICAL ACTIONS Code 1: Person removed from vehicle by order Code 3: Field sobriety test conducted Code 16: Person photographed Code 17: Asked for consent to search person Code 19: Asked for consent to search property Code 23: Admission or written statement obtained from student Code 24: None ARRESTS Code 5: Custodial arrest pursuant to outstanding warrant Code 6: Custodial arrest without warrant FEDERAL INVOLVEMENT Code 11: Contacted U.S. Department of Homeland Security (e.g., ICE or CBP) NON - CRIMINAL INTERVENTIONS Code 10: Psychiatric hold (W&I Code 5150 or 5585.20) Code 8: Noncriminal transport or caretaking transport (including transport by officer, ambulance, or other agency) Code 9: Contacted parent/legal guardian or other person responsible for the minor Code 12: Referral to school administrator Code 13: Referral to school counselor or other support staff ADMINISTRATIVE ACTIONS Code 3: Citation for infraction (use for local ordinances only) Code 4: In-field cite and release Code 7: Field interview card completed NO ACTION OR MINIMAL INTERVENTION Code 1: No action Code 2: Warning (verbal or written) A N A N A L Y SI S O F P OL I C E STO P D A TA / 40 LEGAL AUTHORITY Code 3: Search warrant Code 10: Incident to arrest Code 4: Condition of parole/probation/PRCS/mandatory supervision SAFETY AND EXIGENT CIRCUMSTANCES Code 11: Exigent circumstances/emergency Code 2: Officer safety/safety of others EVIDENCE OF CRIME Code 9: Evidence of crime Code 5: Suspected weapons Code 6: Visible contraband Code 7: Odor of contraband Code 8: Canine detection CONSENT AND INVENTORY Code 1: Consent given Code 12: Vehicle inventory POLICY VIOLATIONS Code 13: Suspected violation of school policy C L A S S IF IC A T I O N S C H E ME O F B A S I S FO R S E A R C H C O DE S Palo Alto Police Department Responses to Recommendations from the National Policing Institute (NPI) Report February 2026 The following information is organized by recommendation as labeled by NPI. These are responses from the Palo Alto Police Department. Data Infrastructure and Collection The Palo Alto Police Department (“the Department”) collects data as legally required by the RIPA Act. Collecting RIPA-analogous demographic data for all police-public contacts is not practicable as it would be overly burdensome due to the many community interactions that officers have daily. Additionally, the brevity of many interactions (and the ubiquity of e-mail and telephonic interactions) makes collecting meaningful demographic data problematic. Consistent with the RIPA Act, for stops initiated due to a traffic violation, the Department already documents whether the violation is a moving violation, an equipment violation, or a non-moving violation, as well as the speciﬁc CA Vehicle Code section violated. The Department is exploring whether its RIPA Data Dashboard can be modiﬁed to allow the data set to be sorted at an even greater level of speciﬁcity. The RIPA Act sets forth a speciﬁc set of data points to be collected. The application that the Department uses to make RIPA data entries is an industry standard tool, used by a majority of CA agencies, that has been speciﬁcally designed to collect the prescribed data set. The Department cannot customize existing ﬁelds, and its ability to add non-mandatory ﬁelds is limited. Stop location data is captured via the computer-aided dispatch (CAD) system and derived from officer radio traffic or information provided by a reporting party (in the case of a dispatched call for service). When broadcasting the location of a stop, officers are trained to use landmarks, intersections, or 100 blocks for officer safety reasons (i.e., so that other officers can quickly locate the stop if help is needed). This practice does not diminish the precision of stop location data in a way that would meaningfully impact analysis. The Department’s CAD and report-writing systems are already integrated. The Department will explore whether further integration of these systems, and other disparate systems, is feasible, and whether future analysis could include this broader dataset. Strategic Monitoring and Analysis Going forward, the Department will review year-over-year RIPA data sets on an annual basis, in order to better track changes over time, identify any outliers and anomalies, and to evaluate alignment with community priorities. Public Transparency and Dashboard Enhancements The data points prescribed by the RIPA Act changed from 2023 to 2024, making year-over- year data comparison (and visual representation) challenging for those years. The Department is hopeful that, going forward, the standardized data ﬁelds will alleviate these issues. Of note, the Department was the ﬁrst (and continues to be the only) local law enforcement agency in the Bay Area to maintain a sortable RIPA data dashboard. The dashboard is not required by statute and goes beyond the state requirements to collect and post static data. The Department will continue to update the dashboard’s functionality annually, as resources permit, to improve the user interface and will use the recommendations from the report as suggestions. The Department will continue to maintain a dedicated RIPA webpage, which includes explanations of how RIPA data is collected and processed, and where it is limited. Community Engagement and Qualitative Research Per the recommendation in the NPI report to create a space for feedback, the Department has added to its dedicated RIPA webpage a “community feedback” function, which allows community members to submit questions, comments or concerns.