INTRADEPARTMENTAL CORRESPONDENCE March 8, 2019 1.0 BPC #19-0072 TO: The Honorable Board of Police Commissioners FROM: Inspector General, Police Commission SUBJECT: REVIEW OF SELECTED LOS ANGELES POLICE DEPARTMENT DATA-DRIVEN POLICING STRATEGIES RECOMMENDED ACTION REVIEW and APPROVE the Office of the Inspector General’s Review of Selected Los Angeles Police Department (LAPD) Data-Driven Policing Strategies. DISCUSSION At the direction of the Police Commission, the Office of the Inspector General (OIG) analyzed two specific data-driven policing strategies being utilized by the LAPD: the Los Angeles Strategic Extraction and Restoration (LASER) Program; and PredPol, which is short for Predictive Policing. Also at the Commission’s direction, the OIG conducted a review of a community survey program called ELUCD. I am available to provide any further information the Board may require. E-Copy – Original Signature on File with the Police Commission MARK P. SMITH Inspector General Police Commission Attachment LOS ANGELES POLICE COMMISSION REVIEW OF SELECTED LOS ANGELES POLICE DEPARTMENT DATA-DRIVEN POLICING STRATEGIES Conducted by the OFFICE OF THE INSPECTOR GENERAL MARK P. SMITH Inspector General March 12, 2019 TABLE OF CONTENTS I. INTRODUCTION AND EXECUTIVE SUMMARY...............................................................1 II. REVIEW OF THE LASER PROGRAM...................................................................................3 A. Program Overview.............................................................................................................3 B. Past Evaluations of the LASER Program ..........................................................................9 C. LASER Program Funding ...............................................................................................10 D. OIG Review of the LASER Program ..............................................................................10 III. REVIEW OF PREDPOL .........................................................................................................25 A. Program Overview...........................................................................................................25 B. Past Evaluations of PredPol.............................................................................................26 C. PredPol Program Funding ...............................................................................................27 D. OIG Review of PredPol ...................................................................................................27 IV. REVIEW OF ELUCD..............................................................................................................30 A. Program Overview...........................................................................................................30 B. ELUCD Funding .............................................................................................................31 C. OIG Review of ELUCD ..................................................................................................31 V. RETENTION, SHARING, AND REPORTING OF DATA ...................................................31 VI. RECOMMENDATIONS .........................................................................................................32 A. Offender-Based Programs ...............................................................................................32 B. Location-Based Programs ...............................................................................................33 C. Reporting and Evaluation ................................................................................................33 VII. APPENDIX .............................................................................................................................. I A. Chronic Offender Criteria.................................................................................................. i B. Example of a Letter Sent to a Chronic Offender ............................................................. iii C. Chronic Offender General Information .............................................................................v D. Chronic Offenders – Points Recorded in the Database ................................................... vi E. LASER Dosage by Vehicle Status Code ......................................................................... ix F. Changes in LASER Dosage and Violent Crime by Quarter ..............................................x G. LASER Dosage and Violent Crime by Area ................................................................... xi H. Anchor Points ................................................................................................................. xii I. PredPol Location Data ................................................................................................... xiii J. Example of an ELUCD Weekly Sentiment Report ....................................................... xiv OFFICE OF THE INSPECTOR GENERAL REVIEW OF SELECTED LOS ANGELES POLICE DEPARTMENT DATA-DRIVEN POLICING STRATEGIES I. INTRODUCTION AND EXECUTIVE SUMMARY On July 24, 2018, a special meeting was held by the Board of Police Commissioners (BOPC or “Commission”) to discuss Data-Driven Policing strategies used by the Los Angeles Police Department (LAPD or “Department”). Verbal presentations were made by representatives of the Office of the Inspector General (OIG), the American Civil Liberties Union of Southern California, and the Stop LAPD Spying Coalition, as well as by Department command staff. Several types of technologies, programs, and strategies were discussed at the meeting, including automated license plate readers, video recording systems, and data-driven policing strategies. Community concerns and issues related to these programs were also expressed during the meeting. On August 14, 2018, the BOPC adopted a motion directing the OIG to conduct a review of the LAPD’s use of three programs that utilize data to inform and drive policing strategies: • • • Operation LASER, also known as the Los Angeles Strategic Extraction and Restoration (LASER) Program, which contains both a person-based and a location-based component; PredPol, a predictive policing system that is location-based; and The ELUCD survey platform, which is designed to inform police departments about public sentiment on a variety of relevant topics. The OIG reviewed the goals and strategies of each program and assessed any available data about how the program was actually operating, including its potential impact on people and communities. Chronic Offender Program A major focus of the OIG’s review was an assessment of the Chronic Offender Program, which is the person-based component of Operation LASER. This assessment involved the analysis of a database designed by the Department to track information and updates about each person designated as a Chronic Offender. Based on this data, as well as information collected through site visits, the OIG identified significant inconsistencies in how the Chronic Offender program was being administered, particularly with regard to selection and documentation practices from Area to Area. These inconsistencies appeared to be related to a lack of centralized oversight, as well as a lack of formalized and detailed protocols and procedures. The OIG also found that the format of the available data made it difficult, in some cases, to determine which activities were being conducted as the result of the program, and to assess the program’s overall impact. Notwithstanding these data issues, however, the OIG found that the majority of people identified as Chronic Offenders had few, if any, actual contacts with the police, who often reported that they attempted to locate the designated person but could not find them. Although the database did list a number of arrests and stops of people designated as Chronic Offenders, most of these could not clearly be connected with Operation LASER based on the information provided. To the extent the Department continues to deploy a person-based Review of Selected LAPD Data-Driven Policing Strategies Page 2 strategy, more rigorous parameters about the selection of people, as well as the tracking of data, should allow for a better assessment of these issues. Location-Based Strategies: Operation LASER and PredPol With respect to its location-based programs, the OIG found that the Department has developed a comprehensive infrastructure for capturing and tracking data related to both Operation LASER and PredPol. This includes well-designed dashboards for tracking the amount of time being spent by officers in designated locations via the use of automated GPS systems or status codes manually entered by the officers; the dashboards also allow users to drill down into crime trends. The OIG’s review found that, overall, targeted crime has decreased as police presence has increased; however, results broken down by quarter and Area were more mixed. In general, given the difficulty of isolating the impact of these programs, as opposed to other factors that may impact crime, the OIG cautions against drawing strong conclusions from the available statistics. In an effort to assess the impact of these programs on locations and communities, the OIG looked at the frequency and duration of reported officer presence in LASER and PredPol locations and found that, in most cases, the amount of time spent in these areas appeared to be relatively limited. For both programs, much of the time reported appeared to be contributed by vehicles that were not in service, or by officers driving through or past the location. Based on a review of the data, the instances involving officer-initiated activity in those areas appeared to be minimal. That said, however, the OIG did note a small proportion of events involving long durations or repeated visits. Based on the available information, it was generally not clear whether these visits were driven by the underlying program, or whether they were the result of other Department activities or strategies. In looking more closely at the information collected by Department systems, the OIG also noted some data anomalies, including discrepancies between automated and manually-entered data and high levels of not-in-service hours. As the Department is still developing and refining these tracking systems, these issues will need to be addressed to ensure that the data collected maintains the level of precision needed to meaningfully evaluate each program’s effectiveness and ensure that Department resources are being appropriately deployed. ELUCD In its review of LAPD’s use of ELUCD’s survey platform for measuring public sentiment, the OIG learned that the Department currently does not have a contract with ELUCD. At present, the company does provide the Department with some general data gleaned from its own survey work, but that data is limited in detail and scope. The OIG will continue to track the Department’s efforts to broaden its strategy for collecting feedback from the public. Recommendations and Next Steps In the course of preparing this report, the OIG met with the Department on several occasions to discuss each of the selected programs and to go over the OIG’s general findings. The OIG found the Department to be very open and receptive in discussing the issues identified. In Review of Selected LAPD Data-Driven Policing Strategies Page 3 acknowledging these issues, the Department noted that both the LASER and PredPol Programs, as well as their associated tools for tracking and visualizing data, were provided to Areas as part of an overall toolkit to identify possible strategies for reduction of crime in their Areas. Due to varying needs across the Department, many Areas appear to have adapted these tools for their own use, leading to some of the differences and inconsistencies the OIG identified. The Department also indicated that it has been working to develop and test different methods to more accurately track and measure dosage in the relevant areas. The Department has already begun a process to overhaul some program components, particularly the Chronic Offender Program, in order to address issues that have been identified through community and OIG feedback, as well as through its own review. As described briefly in the report, some of the proposed changes for a revised offender-based program include more narrowly constraining the selection process, incorporating disclosure and appeal processes, and developing a centralized oversight component. The Department also expects to implement additional technology to assist in more accurately tracking data related to officers’ activities in the field, including those related to data-driven policing strategies. Furthermore, the Department has indicated its intention to implement a “precision policing” framework moving forward. Precision policing refers to an emerging approach that combines intensive crime analysis – and a focused response that values precision over high levels of enforcement – with neighborhood engagement and collaboration. The OIG is encouraged by the Department’s proactive approach in improving its use of data to inform its policing strategies. As part of its review of the Department’s current strategies, the OIG has also developed a series of recommendations focused on improving consistency, increasing transparency, and strengthening oversight and analysis of these programs as they move into the next phase. II. REVIEW OF THE LASER PROGRAM A. Program Overview Operation LASER refers to the Los Angeles Strategic Extraction and Restoration Program. It stems from the federally-funded Smart Policing Initiative (SPI), which is a project of the U.S. Department of Justice, Bureau of Justice Assistance that sought to “either build on the concepts of offender-based and location-based (‘hotspot’) policing by replicating evidence-based practices or to encourage exploration of new, unique solutions.”1 The initiative provided funding to ten law enforcement agencies – each of which was required to select a research partner – to assist them in identifying “effective, efficient, and economical” strategies for addressing crime. Operation LASER was designed by the Los Angeles SPI team, made up of Department personnel and the Department’s designated research partner, Justice & Security Strategies, Inc. (JSS). The research phase of the LASER program began in 2009, and the program was first “Smart Policing Initiative,” BJA Fact Sheet, Bureau of Justice Assistance, Office of Justice Programs, U.S. Department of Justice, September 2009. 1 Review of Selected LAPD Data-Driven Policing Strategies Page 4 deployed in Newton Area in September 2011. Beginning in 2015, the program was expanded to other Areas in phases. At the time of the OIG’s review, the LASER concept had been expanded to a total of 16 of the Department’s 21 geographic areas.2 As stated in its written materials, the overall goal of LASER is to reduce violent and gun-related crime. According to a report published by the SPI team, LASER has five primary objectives in furtherance of the overall goal: • • • • • Extract offenders from specific neighborhoods in the areas. Restore peace to neighborhoods and communities. Remove the anonymity of gun offenders. Remove the anonymity of gang members. Reduce gun and gang-related crime. The report states: “The basic premise is to target with laser-like precision the violent repeat offenders and gang members who commit crimes in the specific target areas. The program is analogous to laser surgery, where a trained medical doctor uses modern technology to remove tumors or improve eyesight. First, the area is carefully diagnosed: Who are the offenders, and where and when are they involved in criminal activity? Plans are then developed to remove offenders from an area with minimal invasiveness and minimal harm to the people and areas around them. Extraction of offenders takes place in a ‘non-invasive’ manner (no task forces or saturation patrol activities), and the result produces less disruption in neighborhoods. Continuing with the medical analogy, by extracting offenders surgically, recovery time of the neighborhood is faster.”3 Operation LASER has two major components, each of which was reviewed by the OIG. The first is a person-based strategy referred to as the Chronic Offender Program. The second is a location-based strategy, which focuses on identifying and increasing police presence in hotspots referred to as LASER Zones. This strategy also includes the identification of “Anchor Points,” or locations that are connected to certain crimes occurring in that area. A general overview of both components, as designed by the SPI team and described in materials provided to the OIG, is provided below. 2 In 2015, the program was expanded to 77 th Street, Southwest, and Southeast Areas to address the high number of violent crimes that had occurred in the prior year. In 2016, the program was expanded to Rampart, Hollenbeck, Northeast, and Harbor Areas. In 2017, the program was expanded to Foothill, Hollywood, Mission, and Olympic Areas. In 2018, Pacific, Wilshire, West Los Angeles, and Central Areas began using LASER. The remaining Areas of Van Nuys, North Hollywood, Devonshire, West Valley, and Topanga were slated to begin using LASER in 2019. “Smart Policing: Los Angeles, California Smart Policing Initiative: Reducing Gun-Related Violence Through Operation LASER,” Smart Policing Initiative: Site Spotlight. Bureau of Justice Assistance, U.S. Department of Justice, October 2012. 3 Review of Selected LAPD Data-Driven Policing Strategies Page 5 1. LASER Program's Person-Based Strategy: The Chronic Offender Program4 According to SPI materials, the overall goal of the Chronic Offender Program was initially to identify persons who were committing violent crimes in a target area and to remove them from the area, presumably by arresting them. This goal appears to have evolved over time, with more recent documentation about the program suggesting engagement strategies that appear designed to deter future crime, such as by notifying identified Chronic Offenders5 that the police are aware of their criminal activity. Once a Chronic Offender is selected, using pre-determined criteria, a Chronic Offender Bulletin is generated and disseminated to field personnel. These bulletins are intended to “assist officers in identifying crime trends and solving current investigations, and to give officers a tool for proactive police work (e.g., a list of offenders to proactively seek out).”6 Selection of Chronic Offenders The process of identifying a Chronic Offender is referred to as conducting a “work up.” As described by Department leadership and SPI materials, this process involves personnel from an Area’s Crime Intelligence Detail (CID), which may be a combination of civilian and sworn analysts. These analysts are responsible for reviewing Arrest Reports, Investigative Reports, and Field Interview cards on a daily basis, looking for anything related to violent crime and/or incidents that involved a gun. Once someone has been selected for a work-up, their criminal history undergoes a review and vetting process with the use of Palantir and other Department systems.7 Each person selected for a work-up is assigned a point total based on their criminal history and the other factors listed below. A ranking system is then used to determine the 12 people with the highest number of points, who are ultimately placed on an Area’s list of Chronic Offenders. According to Department materials, each Area using LASER should maintain a list of at least 12 Chronic Offenders at any given time, in addition to other offenders who would replace those 4 Following the July 2018 Commission meeting on this topic, Areas were instructed to suspend the entry of data into the Chronic Offender database. The Department has since been working on a revised version of the program, which has not yet been approved by the Commission. As such, this report focuses primarily on a review of the current program – and its related data – rather than on any revised version. For the purposes of this report, the term “Chronic Offender” refers to a person who has been designated as such, as part of Operation LASER. 5 “Smart Policing: Los Angeles, California Smart Policing Initiative: Reducing Gun-Related Violence Through Operation LASER,” Smart Policing Initiative: Site Spotlight. Bureau of Justice Assistance, U.S. Department of Justice, October 2012. 6 7 The Palantir platform provides a single access point to multiple sources of law enforcement data for relatively easy use and advanced visualization. Current LAPD users have access to information derived from several existing databases, ranging from national and statewide criminal history systems to county statistics and the Department’s own crime, arrest, and field interview data. Review of Selected LAPD Data-Driven Policing Strategies Page 6 from the list of 12 who have been inactivated from the program due to being taken in custody or for other reasons. The point system used for the Chronic Offender Program has changed somewhat since it was first implemented. At the inception of the program, each person who was the subject of a workup received the following: • • • • • 5 points if the individual is a gang member. 5 points if the individual is on parole or probation. 5 points if the individual had any prior arrests with a handgun. 5 points if the individual had any violent crimes on his or her rap sheet. 1 point for every “quality police contact” in the last two years.8 In 2017, two criteria in the point system above were modified to include the following considerations: • • Identify the number of violent crime arrests the individual had over the last two years. Apply 5 points for each violent crime arrest. Determine whether the individual has used a gun in the course of his/her activities. Apply 5 points for each incident involving a gun over the last two years. Also in 2017, Areas were directed to identify 5-10 Chronic Offenders as “back-ups” in addition to identifying those with the 12 highest point totals.9 (Please see Appendix A). Creation of the Chronic Offender Bulletin Using the Palantir system, an analyst can generate a Chronic Offender Bulletin that provides information on the identified person, including the booking photograph, name, address, date of birth, moniker, description, arrest history, gang affiliation, probation/parole status, vehicle(s) driven, outstanding warrants, and most recent police contacts. This bulletin is intended to be presented during roll calls and posted on the roll call room dashboard. Officers can also access the bulletins on their vehicle's Mobile Data Computer (MDC). The template for Chronic Offender Bulletins contains the following advisory language, which was approved by the City Attorney’s Office in 2011 (according to representatives of that office): “The below listed individual is not wanted at this time. This publication is designed to provide information on prominent known offenders, career criminals, etc. and its contents may not, without additional specific facts, be used as reasonable suspicion to detain, nor probable cause to arrest the individual. If you become aware of an individual that matches the suspect description on a crime Although “quality police contact” was not defined in any of the documents the Department provided to the OIG, it was explained by Department personnel that these were Field Interview cards where the narrative of the contact indicated that a gun was involved in the underlying incident. 8 The updated criteria were outlined in a two-page document titled, “Chronic Offenders: Purpose and Check List. Los Angeles Police Department. October 2017.” 9 Review of Selected LAPD Data-Driven Policing Strategies Page 7 report, prior to any further investigation you shall contact the appropriate detective coordinator for that crime.” Strategies for Intervention Once developed, an Area’s list of 12 Chronic Offenders is presented to the Area Commanding Officer for approval. The Area Commanding Officer then determines which field personnel (Patrol Unit, Gang Enforcement Detail, Parole Compliance Unit, etc.) to assign to a given Chronic Offender for the purposes of conducting follow-up with that individual. Based on Department materials provided to the OIG, the Department's recommended follow-up activities included: 1) sending a letter to the offender; 2) conducting warrant checks; 3) conducting parole/probation compliance checks; and 4) conducting door knocks and advising the offender of available programs and services designed to reduce the risk of recidivism. Personnel who are assigned an offender are to provide a status update to their Commanding Officer every two weeks regarding what actions have been taken with that offender. This information is also entered into a database. 2. LASER Program's Location-Based Strategy: LASER Zones The second component of the LASER program involves identifying specific locations for intervention by officers in the field. These locations, also called LASER Zones or hotspot corridors, are selected based on a historical analysis of gun-related crime data, and they are meant to be maintained for a period of at least nine months. Each LASER Zone is entered into the Palantir data analytics platform, which then allows the Department to conduct detailed tracking of crimes occurring in each zone as well as the amount of time officers spend there.10 Selection of LASER Zones Department personnel utilize ArcMap and the Crime Analysis Mapping System (CAMS) software to analyze crime data. This software allows an analyst to view a map that shows the location of violent crime and gun-related incidents. A “heatmap” layer is used to display density levels of incidents, referred to as hotspots, in shades of color ranging from white (representing an isolated incident), to yellow, orange, red, and scarlet red (representing a large cluster of incidents). The analyst then creates a box on the map – a LASER Zone – that covers where hotspots have the highest level of density, focusing on high-traffic commercial areas. LASER Zone maps for each Area are presented to the respective Commanding Officer, who has the discretion to either reject, accept, or modify the size of the identified LASER Zones. The 10 As noted in a previous footnote, the Palantir platform provides access to information deriving from various data sources, which includes criminal history systems as well as Department-generated crime, arrest, and field interview data. The platform also includes a series of user-friendly dashboards that allow Department members to visualize and analyze data regarding both criminal activity and officer time spent in areas designated as LASER Zones or other targeted geographic areas with recent crime activity, which are generally known as “missions.” Review of Selected LAPD Data-Driven Policing Strategies Page 8 table below provides a breakdown per Area of the Department's 40 LASER Zones as of December 1, 2018, according to the Palantir system.11 CENTRAL BUREAU Number of Laser Zones WEST BUREAU Central Area Rampart Area Hollenbeck Area Northeast Area Newton Area 0 1 3 3 3 Hollywood Area Wilshire Area West Los Angeles Area Pacific Area Olympic Area SOUTH BUREAU Number of Laser Zones VALLEY BUREAU Southwest Area Harbor Area 77th Street Area Southeast Area 5 4 3 3 Foothill Area Mission Area Number of Laser Zones 3 3 0 1 3 Number of Laser Zones 2 3 Strategies for Intervention Once a LASER Zone is approved, field personnel are directed to assess what causes crimes to be concentrated in that zone. Additionally, officers (generally Senior Lead Officers and/or Detectives) are tasked with identifying specific locations within a LASER Zone that possibly have a nexus to the crimes committed there. These identified locations are then labeled as Anchor Points. The identification of Anchor Points relies on an officer's experience and expertise from working in the field. Areas are then tasked with preparing strategies to address the issues identified in the LASER Zones and Anchor Points. Some examples of possible strategies listed in Department materials include abatement, eviction, licensing/conditional use permits, or changes in environmental design. Field officers are encouraged to spend time in LASER Zones to provide high police visibility when they are not occupied with radio calls. The specific times spent in a LASER Zone should approximate the specific times when analysis showed that the crimes in that zone were occurring. Most patrol vehicles are equipped with an Automated Vehicle Locator (AVL) device that automatically uploads data pertaining to how much time the vehicle is spending inside a LASER Zone. If a vehicle is not equipped with an AVL device, officers are required to log their LASER Zone time manually using their vehicle's MDC. The amount of time an officer spends in a LASER Zone is referred to by the Department as “dosage,” and reports comparing dosage and 11 Although Central and West Los Angeles Areas deployed the LASER Program in 2018, Palantir advised that they are still awaiting LASER Zone information from these two Areas. Operation LASER has not yet been implemented at Van Nuys, West Valley, North Hollywood, Devonshire, and Topanga Areas, all of which are located within Valley Bureau. Review of Selected LAPD Data-Driven Policing Strategies Page 9 crime statistics over a specified period of time can be viewed on one of several Palantir dashboards. The progress of a LASER Zone is monitored for a period of nine to 12 months and is tracked in a database that was created for the LASER Program.12 After nine to 12 months, a LASER Zone may be reevaluated. If a high volume of crimes continued to occur in a particular LASER Zone, that zone might remain in place. If, however, the volume of crime in that zone dropped to a low level, a new LASER Zone might be identified to replace it. B. Past Evaluations of the LASER Program With regard to the measurement of the LASER Program's success, JSS has produced a series of materials highlighting the accomplishments of the pilot program in Newton Area. The most recent report is a summary of findings produced in January 2018, which was provided to the OIG by the Department. The OIG notes that although the summary was dated January 2018, it referred to two evaluations of the LASER Program studying Newton Area results between 2011 and 2012. The researchers compared violent crime trends in Newton Area to those in 18 other Areas, and found that, while Part I violent crimes decreased significantly during the study period in Newton, they did not do so in other Areas.13 The report provided to the OIG ultimately concluded the following with regard to violent crime in Newton Area in 2012: “Simply put, Operation LASER succeeded in reducing homicides in Newton by 56% compared to 2011 (36 vs 16) and 59% compared to 2010 (39 vs 16). Newton ended 2012 with an all-time low of 16 homicides. In addition, overall violent crime dropped 19% in Newton (from 2011 to 2012) and Newton ranked number one in violent crime reduction in the entire LAPD for 2012.”14 The OIG is not aware of additional studies of the program at Newton or other Areas that have deployed LASER since its inception.15 12 The database was created sometime in 2017. Prior to the creation of the database, Areas maintained LASER data independently. The database was created so that the Department could have a uniform and centralized system of records. Based on the OIG's review of the records in the database, and as confirmed by Department personnel, not all data collected by the Areas prior to 2017 was transferred into the database. 13 Part I violent crimes encompass reported crimes classified as Homicides, Robberies, Rapes, and Aggravated Assaults. For details about the underlying studies, please see: “Smart Policing: Los Angeles, California Smart Policing Initiative: Reducing Gun-Related Violence Through Operation LASER,” Smart Policing Initiative: Site Spotlight. Bureau of Justice Assistance, U.S. Department of Justice, October 2012, and “Operation LASER and the Effectiveness of Hotspot Patrol: A Panel Analysis,” Justice & Security Strategies, Inc., 2013. “Smart Policing in the Los Angeles Police Department - Los Angeles' Strategic Extraction and Restoration Program (LASER)”, Justice & Security Strategies, Inc. January 2018. 14 15 The Department also provided the OIG with PowerPoint presentations that included an overview of its Community Safety Operations Center (CSOC) results. CSOC was used to target a rise in violent crime across four geographic Areas – 77th Street, Southeast, Southwest, and Newton Areas. Since the CSOC model encompasses a number of additional strategies beyond Operation LASER, however, the OIG was not able to determine the extent to which these results should be attributed to that program. Review of Selected LAPD Data-Driven Policing Strategies Page 10 C. LASER Program Funding Operation LASER has been primarily funded by grants from the Bureau of Justice Assistance that were awarded to the Department in 2009 ($499,959) and 2014 ($400,000). The table below provides a summarized breakdown of the expenditures that occurred between August 2010 and September 2018 that were associated with LASER.16 Expenditure $163,901.00 $ 60,642.97 $ 15,881.03 $ 84,230.95 $104,804.49 $ 8,498.83 $413,141.57 $ 48,449.70 $899,550.54 Cost Category Personnel Salary Fringe Benefits Travel Equipment Supplies Indirect Costs Contract with JSS Other Costs (LAN Installation, Sprint Invoices) D. OIG Review of the LASER Program The OIG’s review of the LASER Program included the following components: • • • • • Review of available materials regarding the operation of the program Literature review and review of community feedback provided to BOPC Meetings with Department command staff Site visits and meetings with selected Area personnel (two Areas per Bureau) Analysis of available data, including data collected in the Chronic Offender database, Chronic Offender Bulletins, and relevant Palantir dashboards In reviewing the program, the OIG’s first objective was to illuminate the stated goals and design of the program, and to evaluate the extent to which LASER practices and outcomes appeared to be aligned with those goals. The second objective was to assess – to the extent possible given the available data – the impacts or consequences of these practices on those people and places selected as part of the program. Finally, the review was focused on identifying areas for improvement or revision, and on developing recommendations for the Commission’s consideration. 16 This information was provided by Fiscal Operations Division. The OIG noted that there are likely indirect costs associated with Operation LASER as well. For example, much of the work of tracking LASER dosage, as well as crime rates in LASER Zones, is conducted through the LASER dashboard on the Palantir “Mission Control” platform, which provides a series of methods for visualizing and analyzing data geographically and over time. The annual total cost of these applications is estimated at approximately $1.8 million per year, but the program includes a great deal of functionality beyond the LASER dashboard. Review of Selected LAPD Data-Driven Policing Strategies Page 11 1. Program Materials In order to conduct its review, the OIG submitted a request to the Department for any documents related to the programs being reviewed. In response, the Department provided two documents which: 1) provided a basic overview and set of protocols for selecting Chronic Offenders; and 2) suggested possible activities in following up with a person designated as such.17 The OIG noted that these documents provided limited specific information about several areas of the program, including: • What parameters should be used in assessing whether an arrest was for a “violent crime.” For example, does this include only Part I violent crimes or would it include other crimes as well? • What parameters should be used in determining whether a person has “used a gun” in the course of their activities. For example, would an arrest for possession of a gun receive points? • Examples of the types of Field Interview cards and arrests that might be “most relevant” in selecting a person for a work-up. • What constitutes a “quality” police contact for the purposes of a work-up. • What parameters should be used to determine whether a person is a member of a gang. • How to handle arrests that did not result in a conviction or were rejected for prosecution. The OIG also noted that the documents provided only general information about how strategies for engagement should be selected, how often a Chronic Offender was to be contacted or otherwise engaged, how the Chronic Offender Database was to be used, and whether and how people should be removed from the program (other than when they are found to be in custody). For example, the documents suggested as one engagement strategy the use of a letter to inform a person that they were on the Chronic Offender List. However, this did not appear to be a mandatory activity, and it is therefore an illustration of the lack of consistency that exists in the Chronic Offender Program. The OIG also found that the language related to making stops of Chronic Offenders lacked precision. After suggesting that officers who see designated Chronic Offenders “may stop them, do a field interview, and let them go, if appropriate,” the document also states that “[i]n many situations, however, as with all stops, [the stops] should be constitutional and legal.” This language should be clarified to unequivocally state that stops must have legal justification in all situations, and – as clearly laid forth in the advisory language approved by the City Attorney’s 17 See Appendix A for a copy of the most recent Chronic Offender Program protocol. As noted previously, the OIG was also provided with Microsoft PowerPoint documents that included overviews and histories of the Community Safety Operations Centers (CSOC), including the CSOC’s use of the LASER Program. Review of Selected LAPD Data-Driven Policing Strategies Page 12 Office – that a person’s status as a designated Chronic Offender should not be used as the basis for any detention or arrest. As there did not appear to be any more formal documents – such as Special Orders, Department Notices/Correspondence, or Manuals – relating to Operation LASER, and given the sensitivity and complexity of this type of program, the OIG recommends that the Department develop formal guidelines to ensure consistency and accountability on an ongoing basis for any such program. 2. Training Protocols The OIG found that training practices related to Operation LASER also appeared, in many cases, to be informal. Training on the program is currently provided by Operations-South Bureau CSOC personnel, who conduct an on-site visit to each Area that is slated to begin using the LASER program. During its own site visits, the OIG was advised that incoming CID personnel assigned to replace an outgoing LASER coordinator may not always receive training from CSOC personnel, but might rather receive general information handed down from the previous coordinator. This practice, along with the lack of detailed written materials, may have contributed to some of the inconsistencies identified by the OIG in its review. The OIG recommends that, going forward, training for incoming staff on Operation LASER or similar programs be formalized to ensure a clear understanding of the goals and principles of the program as well as more consistent adherence to the program’s guidelines. 3. Site Visits To get a sense of how Operation LASER was actually being used at the Areas, the OIG conducted site visits to eight Areas – two from each Bureau – that had been utilizing the program for more than two years. The OIG interviewed personnel from Foothill, Northeast, Southwest, Hollywood, Mission, Newton, Southeast, and Olympic Areas to determine how Chronic Offenders and Anchor Points were being selected, how this information was being disseminated, and what strategies were being used. Overall, the OIG found CID and other personnel to be helpful and open about their use and analysis of crime data, and they provided detailed information about their strategies and practices as well as the reasons behind them. In general, there appeared to be a great deal of thought and precision in the analysts’ approaches to tracking crime on a continuous basis and in digesting and summarizing complex data for their respective Area’s use in addressing crime. As summarized below, the OIG also found that each Area had adapted the LASER program – and particularly the Chronic Offender Program – to meet their specific needs, leading to variability in how the program was being administered: • Differences in how Anchor Points were selected. Three Areas identified Anchor Points that were outside of its LASER Zones, which resulted in LASER dosage data not being captured for these Anchor Points. • Differences in how Chronic Offenders were selected. Five Areas selected Chronic Offenders based on verbal or other informal referrals from field personnel. Although Review of Selected LAPD Data-Driven Policing Strategies Page 13 these referrals were not based on the point system, three Areas reported that they nonetheless calculated the points in order to enter these Chronic Offenders into the database. Two Areas did not use the point system at all. The OIG found that three Areas included offenders with a history of only property crime arrests rather than violent or gun-related arrests. Two Areas described conducting an Area-wide query of all arrests rather than identifying offenders via crime reports. Three Areas had identified more than the specified protocol of 10 to 12 people to serve as back-ups for their Chronic Offender List. One Area did not identify any back-ups at all. • Differences in how Chronic Offender Bulletins were created and disseminated. Five Areas did not generate Chronic Offender Bulletins for each of its identified offenders. Three of these Areas combined all of their Chronic Offenders onto one sheet, while the other two did not create any Chronic Offender Bulletins at all. • Differences in what actions were taken. Four Areas reported having a practice of sending a letter to Chronic Offenders advising them that the Department had identified them as having repeated arrests. (Please refer to Appendix B for an example of the letter.) The letters, which appeared to differ in content from Area to Area, also encouraged the recipient not to engage in further criminal activity and provided them with information on service providers/organizations that were available to assist them. The other four Areas did not provide Chronic Offenders with a letter informing them of their selection for the Chronic Offender Program.18 Two Areas did not conduct any follow-up activities on their offenders, while another Area provided direction for patrol officers to attempt to contact offenders on a daily basis. Yet another Area directed GED officers to conduct compliance checks on a sporadic basis. In analyzing these differences, the OIG notes that, in many cases, Area personnel provided compelling reasons for diverging from generally accepted LASER parameters. These adaptations, where driven by local data and Area concerns, may in fact yield strategies that are more effective or better suited to the needs of the particular problems or communities being addressed. The OIG also notes that some of these adaptations – such as not disseminating Chronic Offender Bulletins to officers – appeared to be in response to concerns voiced by the community and may mitigate those concerns to some degree. These inconsistencies also raise two relevant issues, however. First, variations among Areas must be taken into consideration when trying to draw conclusions about the effectiveness of a given program. For example, while it may be appropriate for a particular Area to use certain elements of the LASER Program to focus on property crime rather than violent crime, it would not necessarily then be appropriate to draw conclusions about the program’s effect on violent crime, either at the Area or Department level. Second, these variations indicate a lack of 18 Note that the sending of the letter was not a required component of the program. Review of Selected LAPD Data-Driven Policing Strategies Page 14 centralized oversight and may undercut safeguards related to the selection of people and places designated for intervention, as well as in applying the appropriate interventions. This is particularly true for any person-based strategy, which must balance the seriousness of the risk of further criminal activity with the intrusion into an affected person’s life. The OIG identified, for example, concerns with practices such as using informal referrals to select Chronic Offenders, which may not have been subjected to a documented vetting process; a focus on lessserious property crimes rather than more-serious violent crimes; and the use of bulletins without the appropriate level of detail and/or advisory language. The OIG recommends that, for any intervention strategy that identifies particular people and places, the parameters for these selections be carefully drawn out and strictly overseen. 4. Review of Chronic Offender Program Data As noted earlier in this report, the Department suspended its use of the Chronic Offender Program, as well its use of the associated tracking database, in August 2018. Since that time, it has been working to revise the program to ensure that it addresses concerns identified by the OIG and others. The below review focuses primarily on data from the existing program that was available as of the date that the program was suspended. The Department’s proposed revisions are briefly discussed in a later section of the report. The Department maintains a LASER database which tracks basic information about each Chronic Offender. The database also provides a variety of tools and reports that allow Areas to track each person on the list, including what actions have been taken and what updates have been entered. The OIG used this database to conduct a general analysis of the characteristics of those individuals designated as Chronic Offenders, as well as any enforcement or engagement activities tracked for each person. At the time the database was suspended in August 2018, it was populated with a total of 637 individuals. Of those, 234 (37%) were marked as active, while the remaining 403 (63%) were marked inactive. The OIG found that this was done through the use of a checkbox that could be checked and unchecked as needed. This format, combined with the lack of dates that a person was made active or inactive, made it difficult to determine how many of the people listed in the database had been considered Chronic Offenders and for how long. For a detailed breakdown of Chronic Offenders by Area and status, please see Appendix C. One area of particular interest to the OIG was the demographic makeup of those on the Chronic Offender List. The OIG found that this information was not included in the data entry form included in the database. To provide this information to the Commission, the OIG looked up the race/ethnicity and gender of each “active” person in the database, the breakdown of which is included in the chart below. Review of Selected LAPD Data-Driven Policing Strategies Page 15 “Active” Chronic Offenders by Race/Ethnicity and Gender Race/Ethnicity Male # Male % Female # Female % Hispanic/Latino 116 49.8% 8 3.4% Black/African American 70 30.0% 2 0.9% White 28 12.0% 6 2.6% Other 3 1.3% 0 0.0% Total 124 72 34 3 % 53.2% 30.9% 14.6% 1.3% Total 23319 100.0% 217 93.1% 16 6.9% The OIG reviewed violent crime arrest data to compare the demographic makeup of the Department's active Chronic Offenders to persons who had been arrested by the LAPD for violent crimes during the existence of the LASER program.20 As shown below, the overall racial/ethnic makeup of Chronic Offenders roughly approximates the makeup of those arrested for Part I violent crimes. The OIG noted, however, that male Chronic Offenders appear to be over-represented, and female Chronic Offenders appear to be under-represented when comparing the two figures. Part I Violent Crime Arrests by Race/Ethnicity and Gender 2012-2018 Race/Ethnicity Male Female Total Hispanic/Latino 40% 9% 49% Black/African American 26% 8% 34% White 9% 3% 12% Other 4% 1% 5% Total 79% 21% 100% Chronic Offender Points Applied The OIG conducted an analysis of the number of Chronic Offender Points associated with each person in the Chronic Offender database in order to spot any issues. Most notably, and as described in the section on Site Visits, it appeared that some Areas were not assigning points at all when selecting offenders, relying instead on referrals from detectives or patrol personnel. Apparently as a result, 37 people listed as “Active,” as well as 75 people listed as “Inactive,” were added to the database with a total of zero points. Overall, the assigned points per offender ranged from zero to 101. While the OIG found a broad range of points assigned, the majority of people in the database – about 59 percent – had 25 points or less. A full breakdown of the point totals, as reflected in the database, is included in Appendix D. 19 20 One person was entered into the database twice but is counted only once here. The OIG obtained Homicide, Rape, Robbery, and Aggravated Assault data covering LAPD arrests from January 1, 2012, to December 31, 2018. The percentages in the table were calculated from a total of 82,808 arrests for violent crime. Review of Selected LAPD Data-Driven Policing Strategies Page 16 Due to the Chronic Offender Program’s focus on individuals who are most actively involved in violent and/or gun-related crime, the OIG also reviewed the points assigned for these categories, where available, and found the following:21 • While some Chronic Offenders were listed as having a large number of arrests for violent crimes, nearly half – 44 percent – of those with detailed point calculations were listed as having either zero or one such arrest. • While about half of Chronic Offenders were listed as having one or more reported arrests for gun-related crimes, about half were listed as having no such arrests. • Nearly 10 percent of the Chronic Offenders in the database did not have any “quality police contacts” recorded, and the majority had less than five such contacts. Alternatively, several Chronic Offenders were listed as having been contacted by the police anywhere from 20 to 45 times. The points assigned for these categories are further detailed in Appendix D. To verify the arrest points assigned in the database, the OIG selected a sample of up to five active Chronic Offenders per Area for closer review. For each of the Department’s 21 geographic Areas, the OIG conducted a detailed review of the Chronic Offender Bulletins and related data for the two people with the highest listed number of Chronic Offender Points, as well as the three with the lowest number of points.22 Based on that review, it appeared that there were significant inconsistencies in terms of how Chronic Offenders had been selected or retained in the program, as well as how Chronic Offender Points were being calculated and tracked. For example, the database included people who were in custody, who had been arrested for only non-violent crimes, and whose points were either not entered or appeared to be over- or under-stated. The OIG also noted that, although it had requested all available Chronic Offender Bulletins, no bulletins had been submitted to the OIG for 42 of the 101 Chronic Offenders selected for closer review. Format and Retention of Chronic Offender Bulletins The OIG requested copies of Chronic Offender Bulletins that had been created since the inception of the LASER Program to determine whether the format used, and information being provided, was consistent with the format approved for the program. Overall, the number of 21 Due to changes in how the system was designed, some records included only a total combined score, while others included points broken down by category (e.g., number of arrests for violent crime). Only 255 (168 active and 87 inactive) of the 637 people in the database had been assigned detailed point calculations by category. The number of arrests was calculated by dividing the number of relevant points by 5. 22 These 101 cases made up 43 percent of the total active cases, but they were not necessarily representative of all cases as they were selected specifically based on their high or low rank. As noted earlier in this report, Devonshire, North Hollywood, Topanga, Van Nuys, and West Valley were Areas that had not yet been slated to use the LASER program; however, these Areas did have Chronic Offenders in the database. Two Areas – Pacific and North Hollywood – had fewer than five active Chronic Offenders in the database. Review of Selected LAPD Data-Driven Policing Strategies Page 17 bulletins submitted to the OIG from each of the Areas using Operation LASER ranged from 153 to zero (three of the Areas submitted no bulletins).23 Newton Area, which had originally piloted the LASER Program, submitted only four bulletins and reported that none of them had been disseminated to officers there. The OIG found that, of the 16 Areas submitting bulletins, nine used the approved format, including the advisory language approved by the City Attorney’s Office addressing the bulletin’s use as justification to detain or arrest the person listed on it. Other Areas had modified the format, including some that combined multiple Chronic Offenders onto one sheet. These combined sheets were a concern because they did not include the level of detail that was originally included in the initial bulletin format. Missing details included, for example, information about the types of arrests that had resulted in the person being placed on the Chronic Offender List. The OIG also identified several instances in which the bulletin indicated that a person was on “formal” or “summary” probation without clearly indicating whether they had search conditions. The OIG also noted that one Area’s bulletins included language advising officers to stop the person on the bulletin or – in later bulletin versions – to “develop reasonable suspicion to do so,” without clearly stating that the officers would need an independent legal basis in order to justify such a stop. To the extent that the Department continues to use bulletins such as those that are part of the LASER Program, the OIG recommends that Areas be required to retain all such bulletins for the purposes of accountability and oversight, and that they be required to use only the approved format. Contacts and Other Actions Taken with Chronic Offenders The Chronic Offender database is designed to track update entries made for each Chronic Offender. These may include, for instance, LASER-driven activities conducted by officers (such as compliance checks or other attempts to contact a Chronic Offender), as well as more general updates gleaned from a review of Department databases (such as contacts with law enforcement, arrests, releases from custody, or deaths). The OIG determined that almost 30 percent of the people in the database had no updates listed, with an additional 18 percent having just one such entry. In contrast, about eight percent of Chronic Offenders in the database had more than 10 update entries, with a small number of people (14) having between 20 and 34 such entries.24 23 The three Areas using LASER that did not submit bulletins were Mission, Pacific, and West Los Angeles Areas. The OIG also noted that there were three Areas – Devonshire, Van Nuys, and West Valley – not currently slated to use the LASER Program that had generated Chronic Offender Bulletins anyway. 24 The OIG reviewed the recorded entries for the 14 Chronic Offenders who had 20 or more updates or actions taken and found that 44 percent described unsuccessful attempts to surveil or locate the Chronic Offender. Another 12 percent of the entries were information-only updates. The remaining entries primarily described arrests or vehicle/pedestrian stops, and it was generally not clear whether such activities were LASER-driven. In all, only 12 actual in-person contacts with this group of 14 Chronic Offenders could be directly attributed to LASER-driven activities, based on the information provided. Review of Selected LAPD Data-Driven Policing Strategies Page 18 Chronic Offenders by Number of Update Entries Number of Chronic Offenders 250 200 192 150 116 87 100 54 50 42 19 16 26 17 14 10 6 5 6 4 2 1 1 3 2 3 3 1 2 1 1 1 1 1 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 23 25 26 27 28 30 31 34 Number of Entries in Database In an attempt to better understand how the Chronic Offender Program was actually being implemented, as well as its impact on Chronic Offenders themselves, the OIG reviewed and classified all of the update entries in the database and found the following:25 • 34 percent of entries described arrests of Chronic Offenders. In most cases, however, the OIG could not determine whether the arrest was LASER-related, or whether it was simply recorded as a tracking entry. • 20 percent of entries described officers’ attempts to conduct parole or probation compliance checks, or to otherwise locate and make contact with Chronic Offenders. In about 70 percent of these cases, the officers reported that they were unable to locate the person they were seeking.26 • 14 percent of entries described traffic or pedestrian stops that had been conducted by officers. As with the arrest entries, it was not clear in most cases whether these were the product of LASER-driven activity or were simply being tracked as part of the program. • 10 percent of entries provided information or status updates about Chronic Offenders, such as updates about their criminal case, pending release, or death. 25 All percentages are approximate. The OIG looked at a total of 2,250 update entries in total; approximately 10 percent included limited information on the type of activity being performed, the results of the activity, or the description of what occurred during the contact. 26 The OIG also noted that 79 entries appeared to use canned language to describe attempts to locate or contact the person they were seeking, all of which indicated that the officers were unable to find that person. Review of Selected LAPD Data-Driven Policing Strategies Page 19 • 9 percent of entries described officers attempting to surveil or observe Chronic Offenders to see if they were committing violations. In about 71 percent of these instances, the officers reported being unable to locate the person they were seeking. • 7 percent of updates described an official letter being issued to Chronic Offenders. In considering this analysis, the direct impact of the Chronic Offender program on most of the individuals selected for it appears somewhat limited in scope. It is important to note, however, that the OIG was unable to determine whether about half of all the entries it researched were driven by Operation LASER. Better and more consistent collection of data would help to provide insight about what actions are being taken as the result of the program, as well as their overall impact and effectiveness. In further assessing the impact of Chronic Offender status on individuals, the OIG reviewed Department records to determine whether any of the 637 persons in the Chronic Offender database had filed a complaint of misconduct or had been involved in a use of force related to their Chronic Offender status. The OIG found that one complaint appeared to arise from a person’s selection as a Chronic Offender;27 one Non-Categorical Use of Force incident appeared to be connected to a LASER-related activity; and no Categorical Uses of Force appeared to be connected to a LASER-related activity. 5. Review of LASER Zone and Anchor Point Data LASER Zone Dosage As discussed in previous sections, a second component of the LASER program involves the identification of specific LASER Zones, as well as the subsequent tracking of the time spent by LAPD vehicles in each of these locations and the number of crimes occurring there over a given period of time. Visible police presence in a location is known as “dosage,” which is captured and measured primarily using data from a police vehicle’s AVL device whenever it is inside of a pre-programmed location. City-wide and for each of the 14 applicable geographic Areas,28 the OIG reviewed both the annual and quarterly changes in LASER dosage and any accompanying changes in the number of violent crimes occurring within LASER Zones.29 Specifically, the review compared annual and quarterly data from 2017 to 2018, which was gathered using the relevant Palantir dashboard. It is important to note that there may be many other factors affecting crime rates that the OIG was not able to control for, and that the LASER Program itself has other components which may or 27 In seven complaints involving four Chronic Offenders, it could not be determined whether the complaints were related to the LASER Program. 28 Central and West Los Angeles Areas did not have LASER Zones programmed into the system and are therefore not included in this analysis. In this context, the term “violent crime” refers to the four types of crimes categorized as Part I violent crime by the Uniform Crime Reporting system – criminal homicide, rape, robbery, and aggravated assault. 29 Review of Selected LAPD Data-Driven Policing Strategies Page 20 may not impact crime rates. As such, the numbers provided here are intended primarily to provide general information about time spent by officers in LASER Zones, as well as about overall changes in violent crime rates in each Area. City-wide, the reported annual LASER dosage for 2018 was 138,498 hours (379.4 hours/day). The annual dosage per Area in the LASER Zones ranged from 2,237 hours (6.1 hours/day) for Foothill to 53,841 hours (147.5 hours/day) for Hollenbeck.30 2018 Dosage by Area and Vehicle Status Reporting Area Hollenbeck Southwest Hollywood Newton Rampart Olympic Mission 77th Street Wilshire Southeast Northeast Harbor Foothill Pacific Total Total Hours 53,841 22,856 15,057 7,207 5,598 5,593 5,041 5,041 4,859 4,031 3,318 3,295 2,237 524 Hours/ Day 147.51 62.62 41.25 19.75 15.34 15.32 13.81 13.81 13.31 11.04 9.09 9.03 6.13 1.44 Not-In-Service Hours 31,766 14,589 7,048 2,607 1,563 2,314 2,151 2,464 1,680 1,115 1,709 924 906 140 Remaining Hours 22,075 8,268 8,009 4,600 4,035 3,279 2,890 2,577 3,179 2,916 1,610 2,371 1,331 384 % NotIn-Service 59% 64% 47% 36% 28% 41% 43% 49% 35% 28% 51% 28% 41% 27% 138,498 379.45 70,975 67,523 51% The OIG noted that the City-wide dosage includes 70,975 hours (51%) for which the identified vehicle’s status was classified as “Not In Service.” According to the Department, not-in-service hours occur when unmanned police cars are parked in LASER Zones. Department personnel have indicated that these hours are included in the LASER dosage due to the fact that even a parked LAPD vehicle can add to the Department’s visibility and deterrent effect on crime. In reviewing the 2018 data, however, the OIG noted three Areas with unusually high levels of not-in-service hours: Hollenbeck, Southwest, and Hollywood. The OIG discovered that, during the relevant period, each of these Areas had a LASER Zone that was programmed to include an LAPD facility. Based on the OIG’s review and on conversations with Palantir personnel, it appears that, when police cars equipped with the AVL device were parked at or in these LAPD facilities, their hours were credited to the respective LASER Zones and likely impacted the relevance of these numbers. As such, up to about one-third of the Department’s annual LASER dosage for the period reviewed by the OIG appears to have been contributed by such vehicles. 30 Pacific Area, which had the lowest amount of dosage at 524 hours, was excluded from all range analyses since it did not implement the LASER Program until August 2018. Review of Selected LAPD Data-Driven Policing Strategies Page 21 Because of concerns about the appropriateness of including these hours, all not-in-service hours from these LASER Zones were excluded from further analysis of the data. Types of Activity Conducted in LASER Zones The OIG reviewed LASER Zone data to determine the types of activities being conducted by officers, as well as their duration. Other than not-in-service hours, the OIG found that time spent in LASER Zones – as captured by the AVL system – was relatively limited. Overall, these hours totaled about 67,529, which comes out to a daily average of 4.6 hours per each of the 40 reported LASER Zones. (See Appendix E for details.) The largest proportion of this time, about 39 percent of it, was coded by officers as “At Scene,” which generally indicates that the unit has responded to a radio call or other dispatch request. Code Six time, which primarily encompasses officer-initiated activities such as pedestrian and vehicle stops or extra patrol, was of particular interest to the OIG, as these events reflect much of the discretionary activity occurring within each LASER Zone.31 Based on reported data, Code Six activities made up about 14 percent of all in-service LASER hours, with an average of a little more than one event, and 30 minutes spent, by officers per LASER Zone per day. In all, threequarters of these events lasted for a half-hour or less, with about 42 percent listed as lasting about 0.1 hours (six minutes).32 Changes in Dosage and Violent Crime over Time The OIG sought to assess whether changes in dosage were accompanied by a corresponding increase or decrease in violent crime in the associated LASER Zones. Department-wide, the OIG found that there was a 5 percent annual increase in LASER dosage from 2017 to 2018. This was accompanied by an overall 5 percent decrease in violent crime inside the Department’s LASER Zones, compared to a 4 percent decrease in non-LASER Zone violent crime over the same period.33 When disaggregating the changes by quarter, it was difficult to identify any particular trend. The OIG noted that increases in dosage were not necessarily accompanied by decreases in violent crime, nor were decreases in dosage accompanied by increases in violent crime. For instance, when comparing the first quarter of 2018 to the first quarter of 2017, a 23-percent increase in dosage was accompanied by a 2-percent decrease in violent crime. In contrast, when comparing the fourth quarter of 2018 to the fourth quarter of 2017, a 17-percent decrease in dosage was accompanied by a 9-percent decrease in violent crime. The charts in Appendix F show the quarterly Department-wide changes in dosage and violent crime: (a) from quarter to quarter over the two-year period reviewed by OIG; and (b) from each quarter of 2018 to the same quarter of 2017. 31 There were also a small number of hours (613) coded as Traffic Stops within the LASER Zones. 32 As 0.1 hours appears to be the minimum amount of time that is reported by the Palantir system, the actual duration may have been even shorter in some instances. 33 Violent crime in non-LASER Zones was calculated by subtracting crimes in LASER Zones from all violent crime in the Area. This calculation includes only those Areas that had LASER Zones programmed into the system. Review of Selected LAPD Data-Driven Policing Strategies Page 22 Similarly, the OIG looked at the relationship between changes in dosage and changes in violent crime by Area from 2017 to 2018. As shown in the table in Appendix G, the results were also mixed. Although there were violent crime decreases in several Areas, there did not appear to be a clear pattern between the size or change in dosage and changes in violent crime. For example, in Newton Area a 30-percent increase in dosage was accompanied by a 23-percent decrease in violent crime. In contrast, Olympic Area saw a five-percent increase in dosage, which was accompanied by a five percent increase in violent crime. The OIG did note that, in all but one Area that increased dosage by more than 10 percent, violent crime decreased by a minimum of three percent. Finally, the OIG compared changes in violent crime between LASER and non-LASER Zones from 2017 to 2018 within the 13 Areas using the LASER program:34 • Nine Areas reported a decrease in violent crime in their LASER Zones, ranging from one percent to 23 percent. In eight of these cases, non-LASER zones also saw a decrease in violent crime, ranging from one to nine percent. • One Area reported no change in violent crime in its LASER Zones. In that case, nonLASER zones saw an increase in violent crime of four percent. • Three Areas reported an increase in violent crime in their LASER Zones, ranging from three percent to 13 percent. In all three of those Areas, non-LASER zones conversely reported an overall decrease in violent crime, ranging from four to ten percent. Overall, the OIG found that in seven Areas, LASER Zones showed better results than nonLASER Zones. In the remaining six Areas, results for LASER Zones were the same as, or worse than, those for non-LASER Zones. Anchor Point Analysis As discussed earlier in this report, LASER Zones and Anchor Points are identified by field personnel and approved by the Area Commanding Officer. The OIG reviewed the Chronic Offender database, which is also used to capture data related to LASER Zones and Anchor Points, to determine what types of locations are currently, or had been previously, identified as Anchor Points by the Areas. As shown in the chart in Appendix H, the vast majority of Anchor Points were identified as commercial businesses or shopping/commercial areas, along with a small number of parks, homeless encampments, and other areas. The OIG reviewed the types of engagement or enforcement activities recorded by officers related to those Anchor Points and found that the most common types included directed patrols and foot beats in the area, followed by contacts with the manager or owner of a business at that location. Approximately 14 percent of recorded activities included an arrest, although it was generally not clear whether the arrests were made 34 This analysis does not include Pacific Area due to the LASER Program having been implemented there in 2018. Review of Selected LAPD Data-Driven Policing Strategies Page 23 due to LASER-generated activity or whether they simply happened to occur at the location of the Anchor Point. Anchor Point Activities by Type Reported Activity Directed Patrol Footbeat in Area Contacted Manager/Owner Arrest Flyer Distribution Other Issued Citation Spoke with Citizens/Community Meeting Recap Decoy Vehicle Field Roll Call Traffic Stops Bus Stop Patrol Transient Enforcement Total Entries Number of Entries 295 268 105 82 39 38 31 27 26 19 19 13 9 4 766 Percent of Activities 39% 35% 14% 11% 5% 5% 4% 4% 3% 2% 2% 2% 1% 1% -35 Although Department materials indicate that Anchor Point enforcement might include strategies such as evictions or changes in permitting or environmental design, these activities generally did not appear in the database. The OIG also reviewed the detailed narrative for each instance where the activity was listed as involving a contact with a manager or owner of a business. Based on this review, it appears that these contacts were largely advisory or for the purposes of offering assistance or obtaining information. The Department’s activities at these locations were generally conducted in cooperation with the business owners and managers, and contacts did not appear to include enforcement action related to the business itself. 6. Conclusions and Next Steps Chronic Offender Program In considering the available data, the OIG identified significant barriers in evaluating the Chronic Offender Program. The most significant of these was a lack of clear, reliable data that could be used to measure both the inputs and outcomes related to these efforts. Another barrier was the significant variation in how the program was being used across different Areas. These variations also led to concerns about the extent to which the Chronic Offender Program practices are aligned with a clear set of goals, and whether these activities are properly balancing the 35 Some entries included multiple activities. As such, these percentages do not add up to 100 percent. Review of Selected LAPD Data-Driven Policing Strategies Page 24 potentially intrusive nature of the program with the risk to public safety posed by the people included in it. One of the primary areas that lacked clarity was the overall goal of the program itself. As stated in its materials, Operation LASER was initially designed to arrest and remove people who had a disproportionate impact on violent and gun-related crime in their community. These goals appear to have evolved into something closer to the deterrence of crime, as evidenced by the use of a letter and other types of engagement. The OIG also observed that some Areas appear to simply use the program as a way to track designated offenders and provide general awareness for officers. Each of these objectives would necessarily involve different activities and – importantly – different measures of success. While the overall goal might be the general reduction of violent crime, a program focused on extraction may naturally count an arrest of a particular person as a measure of success, while one focused on deterrence might ostensibly look for the absence of a crime and/or an arrest involving the person. The OIG supports the Department’s intention to move away from the concept of extraction and toward a program focused on support, deterrence, and accountability. The OIG recommends that, to the extent it continues with its person-based strategy, the Department develop parameters that carefully constrain the selection process as well as the type and frequency of contacts that are permitted by the program. One possible approach would be to focus exclusively on high-risk parolees or probationers with search conditions, and to focus on ways to provide support and follow-up to assist them in avoiding criminal activity. Any personbased system must also incorporate strong controls and oversight to ensure that rights are protected, and that the strategy balances criminal risk with safeguards against unwarranted intrusions into a person’s life. The OIG also recommends that the Department modify its database to ensure that it is capturing the information it needs to conduct ongoing evaluations of the program. This should include, for example, any complaints filed against the Department related to a person who is a subject of the program, as well as any uses of force involving such a person. The Department has already taken several steps to revise its person-based strategy in response to these and other concerns from the OIG and the community. Some of the proposed changes focus on narrowly constraining the selection process to further reduce discretion and the possible impact of bias, focusing exclusively on people with a history of Part I violent crimes, removing the categories of gang involvement and quality contacts from the selection process, and eliminating the requirement to have a certain number of people on an associated list. Other changes will focus on implementing disclosure, appeal, and removal protocols and on creating a centralized oversight process. LASER Zones and Anchor Points The OIG’s review of the LASER dosage revealed a significant possible issue with the data – the large proportion of not-in-service hours. The inclusion of three LAPD facilities within the LASER Zone boundaries led to as much as one-third of all total LASER dosage being submitted by parked vehicles at or inside those facilities. Due to these data characteristics, along with the Review of Selected LAPD Data-Driven Policing Strategies Page 25 difficulty of isolating LASER-based activities from other activities, it was difficult to draw conclusions about the effect of Operation LASER on violent crime. Studies by the Los Angeles SPI team have identified some effects of the program on various types of crime, but these studies have not been recently updated and do not take into account the detailed LASER dosage data currently available. The OIG’s more basic review of the data revealed somewhat mixed relationships between amounts of dosage and rates of violent crime. With respect to Anchor Points, the OIG found that these locations were primarily commercial businesses and that, based on the data provided, the Department’s activities in these areas generally appeared to be conducted in cooperation with the business owners or managers. While there was most likely some impact on people who were present in these areas, the data did not appear to raise concerns such as residences being targeted or enforcement actions being taken against the commercial establishments identified as Anchor Points. The OIG was not able to evaluate the overall effectiveness – in terms of reducing crime – of the selection of, and intervention with, Anchor Points. While a stated goal of identifying these points is to focus on locations that disproportionally contribute to crime in the designated LASER Zone, it was difficult to draw correlations with violent crime data. Moreover, the OIG noted that at least some of the Anchor Points were located outside of a LASER Zone, and the database did not have a mechanism for tracking crime outcomes specifically at those locations. The OIG recommends that the Department work to review and better understand its locationbased LASER data, with particular attention to not-in-service data, to ensure that it captures the types of activity that will allow it to properly measure both inputs and outcomes related to the goals of the LASER Program. It also recommends that the Department ensure that LAPD facilities are not included inside LASER Zones. III. REVIEW OF PREDPOL A. Program Overview PredPol, which is short for predictive policing, is a software program that is designed to “predict” where and when crimes will most likely occur over the next 12 hours. The PredPol software uses an algorithm that analyzes 10 years of crime data, including the types of crimes, as well as their locations, dates, and times. PredPol results are generated by the software's algorithm, and its data does not include information about specific individuals. PredPol marks a location on a map with a red box, referred to as a PredPol hotspot, which represents a 500 square foot area identified as high-risk. The PredPol system generates reports on a daily basis that display PredPol hotspot maps for each geographic Area. The reports generated cover two 12-hour time periods for each day. When not occupied with radio calls or other police-related duties, patrol officers are given “missions” to respond to a PredPol hotspot to provide high police visibility. Similar to LASER, the amount of time an officer spends in a PredPol hotspot is referred to as dosage, which can be measured in either minutes or hours. According to the Department, there are three methods for measuring PredPol dosage: 1) The PredPol software provides analytical Review of Selected LAPD Data-Driven Policing Strategies Page 26 tools that automatically calculate dosage time based on a police vehicle's AVL device; 2) The Palantir software measures dosage time using a combination of a police vehicle's AVL data and status codes provided by field personnel; or 3) An officer can manually enter their status via the Computer Aided Dispatch (CAD) System to indicate they are performing PredPol duties. The CAD system calculates the time from when an officer indicates they are in a PredPol hotspot to when they provide an update to indicate they have left a PredPol hotspot. Note: The OIG was advised that prior to the Department acquiring AVL devices for its police vehicles, dosage hours were only measured via the CAD system. Given that this is a self-reporting method, instances where a unit forgets to provide a status update when it leaves a PredPol hotspot may result in dosage hours being over-inflated. PredPol was first deployed in Foothill Area in 2011 and is currently available Department-wide. The Department uses PredPol for two categories of vehicle-related crimes – Motor Vehicle Theft and Burglary/Theft from a Vehicle. The objective of the program is to predict when and where vehicle-related crimes are most likely to occur and to direct limited police resources to specific target areas (PredPol hotspot boxes) to help reduce crime rates and victimization. As such, a reduction of vehicle-related crime rates is used as a measurement of the program's success. B. Past Evaluations of PredPol In conducting its review, the OIG noted two published studies relating to a completed predictive policing experiment which involved three LAPD geographic Areas and two districts of Kent, England.36 The objective of this experiment was to compare crime analysts’ success at predicting where certain types of crime would occur to the PredPol algorithm’s success at predicting where the same types of crime would occur, and to compare the impact of patrolling those areas on selected crimes. For LAPD, the crime types included: (1) theft of motor vehicles, (2) burglary from motor vehicles, and (3) all other burglary. Per the results of the first study, crime was almost twice as likely to occur in the locations selected by the algorithm than in the locations selected by the crime analysts, who primarily used COMPSTAT and historical crime mapping to make their predictions. This study also found that patrols using locations selected by the program reduced expected crime by 7.4 percent, which was twice as high as the reduction when officers patrolled the areas selected by the analysts. The second study of this experiment evaluated the extent to which overall LAPD arrest rates were impacted by the use of predictive policing, as well as whether there was an impact on the proportion of minority individuals arrested. The researchers reported that, while overall arrests were higher in predictive policing locations, this appeared to be explained by higher crime rates These Areas were Foothill, North Hollywood, and Southwest Areas. For more detail, see “Randomized Controlled Field Trials of Predictive Policing.” The Journal of the American Statistical Association, 2015, and “Does Predictive Policing Lead to Biased Arrests? Results From a Randomized Controlled Trial,” Statistics and Public Policy, 2018. 36 Review of Selected LAPD Data-Driven Policing Strategies Page 27 in those areas. They found no statistically significant difference in the proportion of minority individuals arrested in each Area when predictive policing strategies were in use. C. PredPol Program Funding PredPol software is currently provided free to the Department.37 D. OIG Review of PredPol 1. Analysis of Dosage Hours and Vehicle-Related Crimes The OIG attempted to conduct its own analysis of PredPol dosage and its relationship to changes in the number of vehicle-related crime incidents. As with Operation LASER, the OIG reviewed dosage reports in conjunction with vehicle-related crime rates for calendar years 2017 and 2018 using data obtained from the Palantir PredPol dashboard.38 The OIG found that, City-wide, as reported on the dashboard, the dosage for 2018 was 28,264 hours (77.4 hours/day). This represented a 41-percent increase over the 2017 dosage of 19,978 hours, which was accompanied by a reported 3-percent decrease in vehicle-related crimes. When the OIG compared these hours to those reported by the proprietary PredPol site, however, it found that total dosage for each Area on that site varied significantly from dosage reported by the Palantir system. For example: • While Devonshire Area’s annual Palantir dosage totaled 3,877 hours, its dosage in the PredPol system totaled 901 hours. • Conversely, while Harbor Area’s Palantir dosage totaled 9 hours, its PredPol system dosage totaled 373 hours. The differences in the data reported by these two systems are likely due to the fact that, for the Palantir system, officers must enter a designated status code into their mobile terminals when they enter a PredPol location or zone. This status must then be closed out when they leave. This system may therefore be vulnerable to both under- and over-reporting when units neglect to either “log in” to, or “log out” from, a PredPol zone. Alternatively, the PredPol system tracks dosage using automated AVL data, which is similar to that captured by the LASER program. As described below, this system appeared to collect data from some locations that included LAPD facilities, which might also have affected the data. The OIG additionally noted other discrepancies between dosage amounts and officer activity logs that, due to the automated nature of the AVL system, could not be explained without further analysis. In discussing this issue with the Department, it indicated that it is already exploring technology that would more reliably capture dosage data. 37 As with Operation LASER, there are additional indirect costs related to the development and maintenance of the PredPol Palantir dashboard, but this cost was not specifically broken out from the overall contract. 38 According to Department representatives, although there are two systems that track PredPol dosage in different ways, Palantir is the primary interface used to track PredPol dosage and crime statistics. Review of Selected LAPD Data-Driven Policing Strategies Page 28 Due to differences between the two sets of data, the OIG did not conduct additional analysis of changes in dosage or crime by quarter or Area. 2. Potential Impact of PredPol-Related Activities on Designated Locations As described earlier, designated PredPol boxes or zones change daily and are based on the system’s analysis of available incident data. In order to assess the potential impact of PredPol activities, the OIG looked at one month of AVL data to determine how many PredPol locations received visits during that period, and how many visits were made per location. The PredPol site provides information about the address, time, duration, and unit designation of each visit to a PredPol location by an LAPD vehicle. To complete its analysis, the OIG downloaded all patrol dosage data for the month of January 2019 from the PredPol site. In doing so, it found that approximately 1,359 individual PredPol boxes had been selected and visited across that time period, for a total of about 1,293 hours spent across 24,247 visits. The number of individual PredPol locations visited ranged from 108, in Olympic Area, to 32, in Van Nuys Area. A breakdown of PredPol locations by Area is found in Appendix I. The OIG also looked to see how many times each location was visited and found that, during the month of January 2019, about a third of locations had been visited three times or fewer, and twothirds had been visited 10 times or fewer. Number of Visits per PredPol Location, January 2019 Number of Visits 1 time 2 times 3 times 4-10 times 11-20 times 21-30 times 30-100 times 101-200 times More than 200 times Total Number Locations Percent of Locations39 229 158 116 363 201 96 151 37 8 1359 17% 12% 9% 27% 15% 7% 11% 3% 1% 100% The OIG also noted a small number of locations that had been visited a large number of times, one of which had 458 logged visits and another of which had 189 visits. These two locations also reported the highest and second-highest amounts of time spent by officers, respectively. In looking more closely at these locations, the OIG found that they both encompassed LAPD facilities. Other locations with a large number of visits appeared to include busy intersections or commercial areas such as malls or other shopping centers. The OIG also noted some high39 Due to rounding, percentages shown may not add up to 100. Review of Selected LAPD Data-Driven Policing Strategies Page 29 volume locations that were very close to LAPD facilities (even if not encompassing them) and may have therefore been on a frequent route to or from those facilities. Finally, the OIG reviewed the duration of each visit and found that the majority – about 74 percent – lasted less than a minute, and more than half lasted less than 30 seconds. Based on these numbers, it appears that the vast majority of PredPol visits may have consisted of officers driving through or past the designated location. While there is no way to know whether this was done for the specific purpose of providing visibility at a PredPol location, these short visits seem likely to have provided some police visibility without a significant law enforcement impact on people in those areas. Some of the longer visits may represent stops, responses to radio calls, or other enforcement activity at a specific location, but they made up a fairly small proportion of overall PredPol visits.40 Duration of PredPol Visits Duration of Visits 30 seconds or less 31 seconds - 1 minute 1 - 5 minutes 5 - 10 minutes 10 minutes - 1 hour More than an hour Total Number of Visits Percent of Total 13,645 4,206 4,419 656 1,126 195 56% 17% 18% 3% 5% 1% 24,247 100% The OIG did identify a smaller number of visits with long durations, including five reported visits that lasted over 10 hours. The OIG reviewed the associated unit logs for each of these visits and found that the officers did not appear to be in the specific location for one long visit, but were rather responding to radio calls or conducting other enforcement activity in surrounding areas during the specified time period. Given that most of these locations appeared to be some distance from the designated PredPol location, it was not clear from the available information why this was logged as PredPol dosage. The OIG considered, for instance, the possibility that the AVL GPS system was not functioning properly, that the visit was inadvertently associated with the wrong unit, or that there was another unidentified technical issue at play. In the end, however, the OIG was unable to reconcile the differences among the various data sets. 3. Conclusions and Next Steps As with LASER, the OIG’s review of PredPol dosage revealed potential discrepancies with how dosage data is being collected that made it difficult to draw conclusions about the effectiveness of the system in reducing vehicle or other crime. While the Department’s tools for tracking and 40 Visits to two locations that included LAPD facilities were found to make up 14 percent of visits longer than 10 minutes, and 30 percent of visits longer than one hour. Review of Selected LAPD Data-Driven Policing Strategies Page 30 visualizing dosage and crime are well designed and very user-friendly, questions about the underlying data need to be resolved in order for these tools to reach their full potential. As the objective of PredPol is to provide high visibility in hotspots or locations where crimes are predicted to occur, one potential impact on the community may be an increase of police presence or enforcement in the hotspot areas. The available data appears to indicate that these hotspots are distributed throughout the Department and that the highest-volume locations are business areas (or LAPD facilities) rather than primarily residential areas. Notwithstanding data issues, the OIG found that the impact of PredPol on the community seems to be limited by the fact that the majority of PredPol visits to a given location appeared to be very short and, in most cases, occur only a few times per month. The OIG did note some areas, however, that were subject to many visits or, in some cases, relatively long visits. The collection of more precise data – particularly data that is able to tie PredPol locations to the types of enforcement activities occurring there – would assist in determining the overall impact on the community. The OIG recommends that the Department work to clarify any discrepancies in the data collected, and to ensure that systems for capturing dosage focus on gathering precise and relevant data about officers’ activities in PredPol hotspots. IV. REVIEW OF ELUCD41 A. Program Overview ELUCD, derived from the word “elucidate,” is a technology company that pushes out survey questions via advertisements to a smartphone, tablet, or computer based on a person's location and the application the person is viewing on their device. For the survey questions to be pushed out to a smartphone, location services (GPS) must be enabled on the device. Each survey asks three main questions: 1. Do you feel safe in your neighborhood? 2. Do you trust the police? 3. Are you confident in your police department? The answers to these questions form the basis of a calculated score that ranges from 100 to 900, similar to a credit score rating. Additional survey questions can be added and can include an open-ended question such as, “What is the number one issue or problem on your block or in your neighborhood that you would like the police to deal with?” This allows participants to provide a detailed response. The program allows reports to be generated for a variety of time periods (i.e., weekly, monthly, annually) and for various levels (i.e., Area, Bureau, City-wide). The objective of the ELUCD program is to provide a police department with a real-time “sentiment meter” 41 Although the services that ELUCD provides do not necessarily fall into the category of data-driven policing programs, the BOPC requested that the OIG include a review of ELUCD in this report prior to the Department making any contractual commitments with the company. Review of Selected LAPD Data-Driven Policing Strategies Page 31 based on the survey results. Additionally, the program provides a police department with information about current community issues and with community feedback about those issues. Based on the information received from the survey data, police departments can potentially develop strategies to improve community sentiment, and they can address current issues identified through community feedback. B. ELUCD Funding The Department has no prior or existing contracts with ELUCD. In the Department's 2018-19 Fiscal Year Proposed Budget, a request of $500,000 was submitted to fund ELUCD, but it was not approved. As such, the Department is not using ELUCD as part of any formal arrangement, although it continues to explore this option as part of a larger public-sentiment survey strategy. C. OIG Review of ELUCD ELUCD has been taking surveys of the Los Angeles area at the company's own expense. Although the datasets collected from these surveys are not shared with the Department, ELUCD occasionally provides the Department with a Weekly Sentiment Report that displays current survey scores along with community concerns. ELUCD provided the OIG with a sample weekly report, which is attached in Appendix J. According to ELUCD, the scores displayed on the report were derived from approximately 20,000 community surveys. The three main questions asked by ELUCD are used to compare the results of other agencies across the nation. When asked about what type of personal information is collected when a survey is completed, ELUCD indicated that participants are asked to share their age, gender, race, and whether they live in the area where they are completing the survey. Additionally, at the end of the survey, the participant has the option of providing their email address. In some instances, ELUCD will follow-up with the participant in the future via email, informing them that they had completed a survey and asking if their sentiments have changed. Additionally, participants are always given the option of submitting their surveys anonymously. According to ELUCD, the company maintains a detailed privacy policy. Representatives have stated that the company does not collect a person's name, telephone number, or address. The device a participant uses to complete the survey is assigned a unique ID that allows ELUCD to determine whether a survey is being completed again and again from the same device, which could result in survey data being skewed. Without having access to ELUCD's datasets, the OIG did not examine this program further. V. RETENTION, SHARING, AND REPORTING OF DATA To complete this report, the OIG obtained access to, and reviewed each database related to, the selected data-driven programs. The OIG also requested additional data in the form of any existing Chronic Offender Bulletins, which are not currently stored in the database and are maintained as standalone documents. In evaluating these systems, the OIG also looked at who has access to the relevant records, as well as how those records are being retained. While some of the information used to select and track Chronic Offenders is taken from shared databases, the OIG found that data about each program’s operations is maintained separately and is generally Review of Selected LAPD Data-Driven Policing Strategies Page 32 not shared outside the LAPD, other than with research partners. The OIG also found that, while the majority of this data is maintained on an ongoing basis, there was no specific retention policy for the Chronic Offender Bulletin documents. As a result, these documents were not retained or available for the OIG’s review in many cases. The OIG found that, although the Department captures a variety of data about both PredPol and Operation LASER, it does not appear to compile regular reports on these programs. Given the public interest in these programs and in data-driven policing strategies in general, the OIG recommends that the Department develop a retention policy, as well as a system for regular reporting of basic usage and outcome data to the Commission and the public. Information to be tracked might include the types of data contained in this report, including dosage and crime data, general statistical information about the people and locations targeted for intervention, and information about activities and outcomes related to the Department’s data-driven programs. VI. RECOMMENDATIONS Based on the review and findings detailed in this Report, the OIG has several recommendations for Department action in furtherance of an improved approach to its data-driven strategies. A. Offender-Based Programs To the extent that the Department continues with any data-driven, offender-based policing strategies, the OIG recommends that it: 1. Establish formal written guidelines, to be approved by the BOPC, which: a. clearly articulate the goals and expected results of the program; b. provide clear direction of the selection process including: time parameters, procedures for conducting a work-up, and specific crimes the program is intended to target; c. avoid designating a required minimum number of people to be selected; d. provide disclosure and appeal processes for each person selected for the program; e. provide direction on how and when a person is to be removed from the program; f. clearly define any aspects of the strategy that may be adapted to meet the needs of individual Areas; g. include mandatory program activities (such as providing an offender a letter); and, h. specify prohibited program activities or limits (such as the frequency with which a person may be contacted). 2. Modify its Offender Database to capture: a. a description of why a person was selected for the program, and any specialized Department strategy related to that person, where relevant; b. the date a person is added to the database; c. the date a person becomes active or inactive; d. each person’s descent information for reporting purposes; e. detailed information about the nature and intent of any LASER-driven activity; Review of Selected LAPD Data-Driven Policing Strategies Page 33 f. results of any LASER-driven activity; and, g. the source of any status updates regarding a person in the database (e.g., a records search). 3. Specify a retention policy for any bulletins or related documents, and require that all Areas use a format that has been approved by the City Attorney’s Office. 4. Ensure that any revisions to the language used in the Offender Bulletin or Offender Letter are approved by the City Attorney. 5. Develop a consistent training process to be completed prior to use of the program. 6. Develop an oversight and audit structure to ensure the consistency of the data, as well as the consistent utilization of the program. As part of this process, centralize the maintenance and oversight of the Offender Database. B. Location-Based Programs With respect to the location-based components of Operation LASER and PredPol, the OIG recommends that the Department: 1. Establish formal written guidelines that specify how Areas are to identify LASER Zones and Anchor Points, when to conduct assessments of the Zones, and what strategies and activities are to be taken at these locations. 2. Ensure that LASER Zones and PredPol locations do not encompass LAPD facilities. 3. Reconcile and address inconsistent data or discrepancies between Palantir and PredPol datasets to ensure that dosage amounts are captured accurately. C. Reporting and Evaluation The OIG also recommends that the Department: 1. Develop a system for regular reporting of basic usage and outcome data to the Commission and the public. Information to be tracked might include the types of data contained in this report, including dosage and crime data, general statistical information about the people and locations targeted for intervention, and information about activities and outcomes related to the Department’s data-driven programs. 2. Look for opportunities to obtain independent evaluations of the efficacy and impact of each data-driven policing program. 3. Consider seeking community and Commission input prior to the implementation of any new data-driven policing strategies or any significant revisions to the current data-driven programs. VII. APPENDIX A. Chronic Offender Criteria (Page 1 0f 2) Chronic D?entlers: Purpose and Check List Los Aageles Pollce Department October 1011 A chronic offender bulletin is the result of worlr done by crime intelligence The purpose of the bulletin is to identify the most active violent chronic oll'enders in the Division and in the LASER. zones- These individuals are not suspects but persons of interest- The bulletins are for purposes only and for ollicer safety. To create a bulletin should t'oc us ?rst on the otl?ense and then the individual. Based on a daily review ofdosens of ?eld intewiew cards, arrests, and incident reports. identify a person who has been active and could be engaged in violent criminal beha vior. The analyst conducts a ?worlt up? to determine whether the person has been involved in a violent crime- has any prior arrests for a handgun, is a gang member, or is on probation or parole. Further. the analyst detewnines the number ot"quality' police contacts in the last two years. Check Review Field Interview Cards to dewrmine whether activities are associated with violent crime. Look on the back of the Fl lirst and then look at the front to consider the individual?s characteristics. Set aside the ones that are most relevant for a worlt up. Review Arrest reports that are violent crime or gun related. Set aside the ones that are most relevant. Using the abovementioned FIs and Arrests, use Palantir to do a worlt up ot'the individual- Do a search. resolve any con?icts, determine what the person has done over the last 2 years. ln Palantir- identify the number of violent crime arrests the individual has over the last 2 years. Apply 5 points for each violent crime arrest {this is a change from our original protocol}. Detewnine whether the individual has used a gun in the course ot'his'her activities. Apply 5 points for each incident involving a gun over the last 2 years. {this is a change from our original protocol}- Detemtine whether the individual is a member of a gang. Apply 5 points for gang membership- Detewnine whether the individual is on probation or parole. Apply 5 points for this status- Detewnine how many ?quality? police contacts the individual has had over the last 2 years- Apply 1 point for each contact. {Primarily Fls} Add up the number of points for this individual. This will give you a Chronic ?tl?ender Score. Review of Selected LAPD Data-Driven Policing Strategies Appendix Page ii Chronic Offender Criteria (Page 2 of 2) Review of Selected LAPD Data-Driven Policing Strategies Appendix Page iii B. Example of a Letter Sent to a Chronic Offender (Page 1 of 2) Review of Selected LAPD Data-Driven Policing Strategies Appendix Page iv Example of a Letter Sent to a Chronic Offender (Page 2 of 2) Review of Selected LAPD Data-Driven Policing Strategies Appendix Page v C. Chronic Offender General Information Assigned Area Number of Chronic Offenders by Area and Status 77th Street Central Devonshire Foothill Harbor Hollenbeck Hollywood Mission Newton North Hollywood Northeast Olympic Pacific Rampart Southeast Southwest Topanga Van Nuys West Los Angeles West Valley Wilshire Active Inactive 0 10 20 30 40 50 60 Number of Chronic Offenders in Database 70 80 90 Review of Selected LAPD Data-Driven Policing Strategies Appendix Page vi D. Chronic Offenders – Points Recorded in the Database Chronic Offender Total Assigned Points 66 or more points 12 Number of Assigned Points in Database 61-65 points 12 Active 4 2 56-60 points 16 51-55 points 17 6 46-50 points 17 5 6 41-45 points 32 36-40 points 10 20 21 31-35 points 27 26-30 points 26 21-25 points 15 19 32 16-20 points 28 11-15 points 27 38 27 20 6-10 points 5 points Inactive 67 12 3 4 No points 75 0 20 40 37 60 80 Number of Chronic Offenders 100 120 Review of Selected LAPD Data-Driven Policing Strategies Appendix Page vii Number of Violent Crime Arrests by Chronic Offender 60 Active 54 Inactive Number of Chronic Offenders 50 40 29 30 24 23 23 20 11 11 13 9 10 6 7 8 4 4 4 5 4 2 1 1 3 2 1 1 1 11 14 16 1 1 0 0 1 2 3 4 5 6 7 8 9 10 17 18 Number of Violent Crime Arrests Reported Number of Gun-Related Arrests by Chronic Offender 80 Number of Chronic Offenders 70 69 Active Inactive 67 60 60 50 40 30 23 20 16 10 5 5 4 1 1 1 1 1 0 0 1 2 3 4 5 6 Number of Gun-Related Arrests Reported 7 8 Review of Selected LAPD Data-Driven Policing Strategies Appendix Page viii Chronic Offender Points Assigned for Police Contacts 35 Active Inactive 30 Number of Chronic Offenders 25 21 20 24 14 15 13 15 7 11 10 10 12 5 10 11 9 7 3 7 7 4 5 5 3 0 3 1 2 5 4 2 2 2 1 2 1 2 1 2 1 2 1 1 3 3 2 1 2 2 2 1 2 1 1 1 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 26 29 30 34 35 40 45 Number of Points Assigned for "Quality Police Contacts" Review of Selected LAPD Data-Driven Policing Strategies Appendix Page ix E. LASER Dosage by Vehicle Status Code Vehicle Status Code At Scene, Not available Blank Code Six Follow-up At Scene Enroute Station Follow-up Enroute Extra Patrol PredPol Start of Watch Traffic Stop Dispatched Transport Complete End of Watch Back-up Pursuit At Scene, Available All Other Total 2018 In-Service Hours by Vehicle Status Code42 Annual Avg Daily Avg Daily Hrs Hours Hours per Zone 26,540 73 1.8 11,688 32 0.8 9,365 26 0.6 6,521 18 0.4 3,741 10 0.3 3,312 9 0.2 1,582 4 0.1 1,276 3 0.1 918 3 0.1 800 2 0.1 613 2 0.0 370 1 0.0 314 1 0.0 163 0 0.0 145 0 0.0 14 0 0.0 13 0 0.0 154 0 0.0 67,529 185 4.6 Percent of Hrs 39% 17% 14% 10% 6% 5% 2% 2% 1% 1% 1% 1% 0% 0% 0% 0% 0% 0% 100% Totals listed throughout the report may vary slightly due to rounding. Does not include “Not In Service” time, which made up approximately 51 percent of all time logged. 42 Review of Selected LAPD Data-Driven Policing Strategies Appendix Page x F. Changes in LASER Dosage and Violent Crime by Quarter Change from Prreceding Quarter Quarter-to-Quarter Changes in LASER Zones Comparison with Preceding Quarter Change in Dosage 30% 20% 20% 13% 14% 8% 5% 10% Change in Violent Crime 16% 3% 0% 0% -10% -10% -20% -5% -4% -5% -13% -30% -25% -40% (Q1 to) Q2 Q3 Q4 Q1 Q2 2017 Q3 Q4 2018 Quarter Percent Change from Prior Year Changes in LASER Zones 2017-2018 Same Quarter Comparison 30% 23% Change in Violent Crime 17% 20% Change in Dosage 10% 0% 0% 0% -2% -10% -9% -10% -20% -17% -30% Q1 Q2 Q3 Quarter Q4 Review of Selected LAPD Data-Driven Policing Strategies Appendix Page xi G. LASER Dosage and Violent Crime by Area Review of Selected LAPD Data-Driven Policing Strategies Appendix Page xii H. Anchor Points Anchor Points by Area and Type of Location Central Area Rampart Area # of Anchor Points 3 3 Hollenbeck Area 6 Northeast Area 5 Newton Area 6 CENTRAL BUREAU Location Type 1) Small Businesses, 2) Sidewalk, 3) Park 1) Store, 2) Hotel, 3) Liquor Store 1) Store, 2) Shopping, 3) Liquor Store, 4) Parking Lot, 5) Store, 6) Small Businesses 1) Store, 2) Store, 3) Small Businesses, 4) Homeless Encampment, 5) Homeless Encampment 1) Restaurant, 2) Park, 3) Shelter, 4) Park, 5) Sidewalk, 6) Shopping Center SOUTH BUREAU Southwest Area 7 Harbor Area 5 77th Street Area 5 Southeast Area 6 WEST BUREAU Hollywood Area Wilshire Area West Los Angeles Area Pacific Area 0 3 0 1 Olympic Area 8 VALLEY BUREAU Foothill Area Mission Area 5 5 1) Small Businesses, 2) Restaurant, 3) Restaurant, 4) Sidewalk, 5) Small Businesses, 6) Restaurant, 7) Parking Lot 1) Park, 2) Small Businesses, 3) Liquor Store, 4) Park, 5) Store 1) Restaurant, 2) Supermarket, 3) Indoor Swap Meet, 4) Supermarket, 5) Small Businesses 1) Small Businesses, 2) Small Businesses, 3) Small Businesses, 4) Restaurant, 5) Small Businesses, 6) Gas Station No Information in the Database 1) School, 2) Restaurant, 3) Shelter No Information in the Database 1) Park Metro Station, 2) Store, 3) Supermarket, 4) Metro Station, 5) Supermarket, 6) Small Businesses, 7) Small Businesses, 8) Metro Station 1) Homeless Shelter, 2) Park, 3) Store, 4) Park, 5) Park 1) Apartment, 2) Park, 3) Shopping Mall, 4) Park 5) Recreation Center 1) Shopping Mall, 2) University, Devonshire Area* 5 3) Shopping Mall, 4) Metro Station, 5) Park North Hollywood Area* 1 1) Motel 1) Shopping Center, 2) Shopping Center, Topanga Area* 3 3) Shopping Center *According to the Department, LASER has not yet been implemented at Van Nuys, West Valley, North Hollywood, Devonshire and Topanga Areas; however, as seen in the table above Devonshire, North Hollywood and Topanga Areas have anchor point information in the database. Review of Selected LAPD Data-Driven Policing Strategies Appendix Page xiii I. PredPol Location Data Area 77th Street Central Devonshire Foothill Harbor Hollenbeck Hollywood Mission Newton N Hollywood Northeast Olympic Pacific Rampart Southeast Southwest Topanga Van Nuys Wilshire West LA West Valley Total PredPol Locations Visited in January 2019 Locations Total Duration Average Duration Visited (Hours:Minutes) (Hours:Minutes) 76 62:29 0:03 59 229:18 0:04 71 79:33 0:04 39 21:27 0:03 56 31:49 0:02 55 130:13 0:07 57 71:25 0:02 60 45:38 0:02 62 86:40 0:03 48 33:09 0:04 54 32:57 0:04 108 71:10 0:02 94 45:33 0:02 70 33:37 0:01 81 107:26 0:02 92 85:45 0:02 44 29:09 0:03 32 10:36 0:02 100 51:20 0:03 61 25:03 0:05 40 9:28 0:01 1359 1293:55 0:03 Review of Selected LAPD Data-Driven Policing Strategies Appendix Page xiv J. Example of an ELUCD Weekly Sentiment Report (Page 1 of 2) Review of Selected LAPD Data-Driven Policing Strategies Appendix Page xv Example of an ELUCD Weekly Sentiment Report (Page 2 of 2)