KROLL Presentations/Power Point to PS Commissioners 3-7-2022 — original pdf

Backup

Kroll Phase B Report Evaluation of Austin Police Department: Use of Force / Public Interactions / Recruitment, Selection, and Promotions Presentation to Austin Public Safety Commission March 7, 2022 Introduction / Scope of Work Scope of Report Kroll’s evaluation addressed four distinct areas 2 3 1 Analysis of APD use-of-force incidents / Jan. 1, 2017 - Dec. 31, 2020 (48 months) Review of 1,321 APD use of force incidents / June – November 2019 (6 months) Analysis of public interactions with civilians (e.g., traffic stops, arrests, citations, and searches) / 2020 (12 months) 4 Evaluation of recruitment, selection, and promotion policies and practices 3 Report Overview Section 3 Section 4 Provides a 48-month analysis (2017-2020) and contextualized understanding of how, when, and against whom the APD uses force. Are there disparate impacts based on race, ethnicity, or gender / geographical sectors / other factors? Provides a qualitative analysis and review of 1,321 use-of-force incidents from June to November 2019. Is force appropriately applied? Does APD unnecessarily escalate encounters? Is their sufficient supervisory review? Section 5 Documents patterns and trends observed for APD motor vehicle stops during 2020 (1 year) and arrests from 2017-2020 (4 years) and examines racial/ethnic disparities in the outcomes. Section 6 Reviews and analyzes APD’s recruitment, selection, and promotion processes and potential impact on racial, ethnic, and gender diversity. Section 7 Kroll recommendations. 4 Section 3: Review and Data Analysis of APD Use of Force (2017-2020) Definitions • Disproportionality • Disparity • Bias o A difference in outcomes within a single racial/ethnic group (e.g., use of force against Black individuals) compared to that group’s representation in a selected comparison population (e.g., Black residential population) o A difference in outcomes across groups (e.g., racial/ethnic groups, gender, etc.) in policing o Prejudice in favor of or against one thing, person, or group compared with another, usually in a way considered to be unfair • Racially biased policing o Occurs when law enforcement inappropriately considers race or ethnicity in their decisions to intervene in a law enforcement capacity 6 If you find disparity what does that mean? How much disparity is too much? • Statistical analyses measure disparity or disproportionality, not bias o Cannot be reliably used to determine the reasons for differences o Cannot conclude that disparity, even high levels of disparity, is proof of bias – bright line does not exist • Why do the analyses then? o Identifying disparity allows you to examine patterns and trends more closely o Identify the questions to ask to determine whether there are legitimate explanations for disparities o Develop more appropriate corrective measures (e.g., training, supervision, policy) 7 2017-2020 Trends: APD arrests declined 51% Use of force incidents increased 58% Total Number of APD Arrests Number of Known and Unknown Individuals Who Experienced Use of Force h t n o M y b s t s e r r A f o # 4500 4000 3500 3000 2500 2000 1500 1000 500 0 3109 2864 1962 2091 1000 1868 2008 2776 3500 3000 2500 2000 1500 s l a u d i v i d n I f o # 500 0 2389 720 2020 8 7 1 - n a J 7 1 - r p A 7 1 - l u J 7 1 - t c O 8 1 - n a J 8 1 - r p A 8 1 - l u J 8 1 - t c O 9 1 - n a J 9 1 - r p A 9 1 - l u J 9 1 - t c O 0 2 - n a J 0 2 - r p A 0 2 - l u J 0 2 - t c O 94 2017 83 2018 88 2019 Unknown Individuals Known Individuals APD Use of Force 2017-2020 Gender 26.1% 73.9% Female (n=2,362) Male (n=6,679) 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Severity Level of Use of Force 49.3% 44.7% Race/Ethnicity 1.8% 33.1% 31.4% 33.6% Black (n=2,836) Hispanic (n=3,037) White (n=2,994) Other (n=174) 5.3% 0.5% Level 1 (n=26) Level 2 (n=275) Level 3 (n=2,547) Level 4 (n=2,311) • Most severe = Level 1; Least severe = Level 4 • No significant racial/ethnic differences in force severity • Significant gender differences in force severity • Females 1.6 times more likely than males to experience least severe level of force • Males significantly more likely than females to experience the higher force severity 9 Highest Levels of Resistance Displayed Toward Officers 26.1% 10.3% 0.6% 4.3% 0.8% 0% 10% 20% 30% 40% 50% 60% Deadly Resistance Aggressive Resistance Prepatory Resistance Defensive Resistance Passive Resistance Not Resistant • Average levels of resistance displayed toward officers were consistent across gender and racial/ethnic groups and stable each year • Unknown individuals displayed more serious levels of resistance than known individuals Individuals Displaying Aggressive or Deadly Resistance 57.8% 55.2% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 27.2% Unknown Individuals Known Individuals 10 Individuals’ Impairment by Race/Ethnicity Perceived Impairment • Individuals perceived to be under the influence of drugs/alcohol and/or with mental health issues were at greater risk for use of force Impairment of Individuals during Use of Force Events Race/Ethnicity Comparison • Black individuals were most likely to have force used against them when no impairment Impairment Type by Individuals’ Race/Ethnicity 80.0% 59.5% 37.8% 22.6% 80.0% 60.0% 40.0% 20.0% 0.0% 65.3% 62.4% 50.5% 46.6% 38.5% 28.5% 60.0% 40.0% 20.0% 0.0% 29.8% 24.0% 14.3% No Impairment Listed Drugs/Alcohol EDP/Mentally Unstable Note: Under the influence of drugs/alcohol and EDP/Mentally Unstable are not mutually exclusive % No Impairment Listed % Alcohol/Drugs % EDP/Mentally Unstable 11 Black Hispanic White Repeat Uses of Force Repeat Uses of Force (2017 – 2020) • 30% of those who had force used against them were involved in more than one use of force event • Individuals with perceived impairments were more likely to have multiple use of force encounters • Black individuals were more likely to have multiple use of force encounters Single vs. Multiple Use of Force Events, by Individuals’ Impairment Single vs. Multiple Use of Force Events, by Individuals’ Race/Ethnicity 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 58.7% 61.2% 54.2% 72.8% 72.6% 63.8% 30.9% 25.6% 15.6% 36.2% 27.2% 27.4% 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Single Event Multiple Events No Impairment Listed Drugs/Alcohol EDP/Mentally Unstable Single Event Multiple Events Black Hispanic White 12 Use of Force by APD Sector APD Sectors • The frequency of police use of force varies dramatically across APD sectors • George Sector accounted for 23% of use of force incidents Figure 10: Use of Force per 10,000 residents* 1711.9 2,000.0 1,600.0 1,200.0 800.0 400.0 0.0 32.9 42.7 133.0 51.1 99.3 58.0 138.1 155.6 ADAM (n=167,893) BAKER (n=136,062) CHARLIE (n=76,142) DAVID (n=155,240) EDWARD (n=113,937) FRANK (n=146,134) GEORGE (n=12,086) HENRY (n=70,884) IDA (n=64,284) *The airport has no population so APT sector is excluded from this graph. 13 Measuring Disparity: What are Benchmark Analyses? • Entered national conversation as part of traffic stop / profiling research in 1990s o Percent stopped by race had to be compared to some other data to determine level of disparity o Can apply to any police outcome (e.g., stops, arrests, force) • The numerator represents the individuals who experienced the actual outcomes • The denominator represents the expected rate of the outcomes based on a comparison data source (i.e., the benchmark) 14 Disproportionality v. Disparity • Within group comparison: Disproportionality Index = Proportion of racial/ethnic groups observed uses of force Proportion of racial/ethnic groups expected uses of force o DI = 1.0 indicates no disparity o DI > 1.0 indicates disparity (e.g., group UOF rate more than expected based on benchmark) o DI < 1.0 indicates reverse disparity (e.g., group UOF rate less than expected based on benchmark) o Larger the size of the DI, the greater the disproportion •Between group comparison: Disparity Ratio = Minority Group’s Disproportionality Index Majority Group’s Disproportionality Index o Interpreted as the likelihood of an individual in the minority group having force used compared to the majority group − For example: DR = 2.0 – the minority group is two times more likely to have force used against them in comparison to the majority group) 15 Limitations of Benchmarks • Reliable benchmarks are proxy measures for people who are “similarly situated” or “at risk” of experiencing the same outcome, assuming no bias exists • No benchmarks examine all risk factors that might explain racial/ethnic differences in outcomes • Results vary widely by benchmark – can lead to dramatically different conclusions o Residential census data is a particularly flawed benchmark in terms of ability to measure risk For Example: • For traffic stops, the risk of being stopped may be influenced by: Driving quantity, quality, location, and times Condition of vehicle o o o Motorist & passenger characteristics and behaviors (including offending behavior) • For use of force, an individual’s risk of having force used against them may be influenced by: o Frequency, nature, and location of contacts with the police o Known or suspected involvement in criminal activity o Individual characteristics and behaviors during the encounter (particularly resistance) 16 Racial/Ethnic Disproportionality and Disparity Analyses of Use of Force Rates: City-Wide Kroll examined five benchmarks (i.e., comparison groups): Disparity Ratios: 1) Residential Population • Blacks were 6.7 times and Hispanics 1.5 times more likely than whites to have force used against them compared to their representation in the residential population 2) Arrestee Population (all offenses) • When compared to arrest and suspect-based benchmarks, these 3) Arrestee Population (Part I Violent disparities are much reduced or eliminated offenses) offenses) 4) Criminal Suspect Population (all ─ All arrestees—Blacks were 1.2 times more likely and Hispanics less likely to experience force compared to whites ─ Part 1 Violent arrestees—Blacks and Hispanics were less likely than whites to experience force 5) Criminal Suspect Population (Part I Violent offenses) ─ All suspects—Blacks were equally likely and Hispanics less likely than whites to have force used against them ─ Part 1 Violence suspects—Hispanics were equally likely and Blacks less likely compared to whites to have force used against them 17 Citywide Disparity Ratios by Benchmark Figure 21: Citywide Disparity Ratios by Race with Various Benchmarks as Denominators 6.70 o i t a R y t i r a p s i D 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 1.20 1.03 1.50 1.04 1.04 0.55 0.83 0.97 0.91 0.71 Black % of Population % of All Suspects Crime Incidents % of All Arrests % of All Suspects UCR Part I Incidents Hispanic % of All UCR Part I Arrests 18 Racial/Ethnic Disproportionality and Disparity Analyses of Use of Force Rates: Sector Level • Major disparities in use of force for Blacks (across all APD sectors) and Hispanics (half of APD • Benchmark 1: Residential Population sectors) • Benchmark 2: Arrestee Population • Minor or no disparities in use of force for Blacks (all sectors) and Hispanics (most sectors) • Two sectors (George and Ida) showed Hispanics are 1.5 times more likely to have force used • Benchmark 3: Criminal Suspects • Minor or no disparities in use of force for Blacks and Hispanics (most sectors) • One sector (George) showed Blacks are 1.2 times more likely to have force used • Two sectors (Edwards and George) showed Hispanics are 1.2 and 1.6 times more likely to have force used 19 Multivariate Statistical Analyses • Statistical models that simultaneously control for multiple factors that predict stop outcomes (i.e., “holding all else constant”) • Officer decision making is complex – factors influencing police behavior o Suspects’ characteristics o Legal characteristics o Characteristics of the stop oOfficers’ characteristics oOrganizational influences oCommunity influences • But…statistical models do not include all possible and relevant variables – • Multivariate tests are most appropriate as descriptive tool to assess strengths model misspecification of relationships 20 Multivariate statistical analysis • Kroll examined 126,096 arrests from 2017-2020 to identify factors that predict whether force was used Findings: • As expected, strongest predictors of force within arrests are legal and incident characteristics (e.g., custodial, weekend, weapon seizures) • Note: Individual’s resistance is not measured in APD’s arrest data and offense severity is not reliable in APD’s data – neither predictor is included in the statistical models • Overall, there were small racial disparities in whether force was used in arrest situations ─ Blacks were slightly more likely than whites to be involved in arrests that resulted in force ─ Arrests within areas with higher violent crime rates had a greater likelihood of use of force • In communities with higher violent crime rates, no racial/ethnic differences in the likelihood of force • Black individuals arrested in communities with low violent crime rates slightly more likely to have force used against them 21 Use of Force Analysis: Summary • Force has significantly increased during 4-year period examined, while arrests have sharply decreased o Trends may be partially explained by: Changes in use of force reporting, increased use of alternatives to arrest, and changing pattern in use of force experienced in 2020 (possible result of civil unrest, changes in crime patterns, pandemic response, etc.) • Trends noted in use of force (consider for continuous improvement): o Impaired individuals (drugs/alcohol) o Individuals experiencing mental / behavioral health crisis o Changes in frequency and severity of resistance shown o Consistent problems in data collection that limits detailed analyses • Some racial/ethnic disparities found across statistical techniques o Majority of disparities reported are substantively small and may be result of unmeasured factors o Consistent findings across statistical techniques for racial/ethnic disparities in George Sector 22 Section 4: Qualitative Use of Force Analysis – June to November 2019 (6 Months) Use of Force (June – November 2019) Kroll evaluated 1,321 incidents involving 2,960 uses of force from June 1, 2019 to November 30, 2019 112 incidents (8.5%) contained issues of concern • 82 incidents (88 individuals) involved inappropriate force or unnecessary escalation of the encounter • 30 cases involved additional issues of concern • In all cases, supervisors were notified The racial/ethnic breakdown of the 88 individuals: • Black – 28.4% • White – 21.6% • Hispanic – 47.7% • Asian/Other – 2.3%. 24 Problem Areas Stop and Frisk without Reasonable Suspicion Unnecessary Escalation Resisting Detention/ Search Charges Supervisory Issues Taser Usage Pointing of Firearms / Actively Targeting Neck Restraints/ Chokeholds Head Strikes Mental Health Related Body Worn Cameras 25 Section 5: Analysis of Traffic Stops, Arrests, Citations, and Searches 2020 (Total = 68,330) Analysis of All Arrests 2017-2020 (Total=128,213) APD Vehicle Stop Data Limitations • Initial scope of work requested an analysis of stops, citations, charges, arrests, and searches for a six- month period from June to November 2019 • APD does not have a comprehensive motor vehicle stop database o Instead combines three separate databases (warnings, citations, and arrests) that are not mutually exclusive o Traffic stop data collected prior to January 1, 2020 has known errors and could not be reliably used − Stops counted multiple times if a single stop resulted in multiple outcomes (e.g., citation and arrest) – also resulted in double counting of searches − No automated method available for identifying and removing duplicates – must be manually cleaned − Current findings should not be directly compared to previous reports on APD traffic stops because these reports were produced using data we now know to be invalid • Data collected after January 1, 2020 has undergone extensive cleaning efforts by APD staff 27 Motor Vehicle Stops: January 1 to December 31, 2020 Race/Ethnicity and Gender of Individuals Involved in APD Motor Vehicle Stops Reason for Stop for APD Motor Vehicle Stops 74.8% 5.3% 14.9% 44.8% 35.0% Black Hispanic White Other 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 17.3% 7.7% 0.2% Moving Traffic Violation Vehicle Traffic Violation Violation of Law Other Than Traffic Pre-existing knowledge (i.e. warrant) 35.9% 64.1% Male Female 28 Outcomes of Motor Vehicle Stops: Jan 1 - Dec 31, 2020 Arrest, 5.4% Verbal Warning, 11.2% Citation, 31.7% Written Warning, 51.7% • Categories are mutually exclusive • APD collects data in three different datasets; only the most serious outcome resulting from a traffic stop is measured • Cannot determine what percentage of stops result in any warning or citation, only percentage of stops that had warning or citation as most serious outcome recorded • Cannot determine if multiple warnings or citations were issued to a single individual during a stop • Cannot determine who (driver or passenger) was warned, cited, or arrested 29 Racial/Ethnic Differences in Vehicle Stops • Examining racial/ethnic differences Racial/Ethnic Differences in Reason for Stop in who gets stopped depends on a 100.0% 80% 72% 68% benchmark analysis • But…benchmark comparisons are only appropriate for discretionary stops; APD stop data does not distinguish officer-initiated from dispatched stops • Kroll examined racial/ethnic differences in reason for stop and stop outcomes 80.0% 60.0% 40.0% 20.0% 0.0% 11% 9% 5% 21% 19% 15% Moving Traffic Violation Vehicle Traffic Violation Violation of Law Other Than Traffic Black (n=10,156) Hispanic (n=23,933) White (30,638) Note: Black individuals were also more likely than all other racial/ethnic groups to be stopped based on pre-existing knowledge (0.5% Blacks, compared to 0.2% Hispanics, 0.1% whites, and 0.1% others) 30 Racial/Ethnic Differences in Vehicle Stops Outcomes Racial/Ethnic Differences in Vehicle Stop Outcomes 70.0% 60.0% 50.0% 40.0% 30.0% 10.0% 0.0% 20.0% 16.1% 10.4% 10.4% 59.4% 45.7% 43.2% 39.3% 30.6% 26.3% Verbal Warning Written Warning Citation Only Arrest Black (n=10,156) Hispanic (n=23,933) White (n=30,638) 7.5% 7.1% 3.9% 31 Multivariate statistical analyses Kroll examined 68,330 vehicle stops in 2020 to identify factors that predict warnings, citations, and arrests Findings: • Strongest predictors of outcomes are legal and incident characteristics (e.g., reason for stop, whether contraband was seized, city street vs. highway) • Note: missing important predictor variables (e.g., whether the stop was officer-initiated or dispatched, offense severity, an individuals’ age, and community characteristics including the location of the stop, neighborhood crime rate, SES, and racial composition) • After controlling for some legal & incident characteristics, significant racial/ethnic differences in outcomes remain: − Warnings: Blacks and Hispanics were significantly less likely than whites to be issued warnings − Citations: Hispanics and those of other race/ethnicity were 1.5 and 1.3 times more likely than whites to be issued citations − Arrests: Blacks and Hispanics were 1.7 and 1.5 times more likely to be arrested, while individuals of other race/ethnicity were 2.4 times less likely to be arrested compared to whites 32 Searches during Vehicle Traffic Stops Searches were conducted in 7.6% (5,224) of 68,330 vehicle stops in 2020 Racial/Ethnic Differences in Searches Reasons for Search 0.5% 1.0% 11.3% 10.0% 5.1% 2.8% 54.0% Gender Differences in Searches Black (n=1,143) Hispanic (n=2,405) White (n=1,575) Other (n=101) 37.0% 7.6% Consent Contraband/Evidence in Plain View Incidental to Arrest Inventory of Towed Vehicle Probable Cause 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% 9.0% 5.3% Female (n=1,291) Male (n=3,933) 33 Racial/Ethnic Differences in Search Reasons during Traffic Stops 62.1% 54.1% 43.2% • Only examine racial/ethnic differences in three most common reasons for search: probable cause, incident to arrest, and inventory of towed vehicle • Black and Hispanic individuals were more likely than whites to be searched based on probable cause • White individuals were more likely than all other racial/ethnic groups to be searched incidental to arrest 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 47.9% 36.5% 29.9% 7.1% 7.8% 6.9% Incidental to Arrest Probable Cause Inventory of Towed Vehicle Black (n=1,143) Hispanic (n=2,405) White (n=1,575) 34 • Almost one quarter of all searches (23.9%) Racial/Ethnic Differences Seizures during Discretionary Searches (n=1,982) Contraband Seizures during Searches resulted in seizures of contraband • Mandatory searches: o 21.3% have contraband seizures • Discretionary searches o 28.2% have contraband seizures • Racial/ethnic differences in only discretionary searches should be used to examine disparities 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 31.6% 28.3% 24.4% Black Hispanic White 35 2020 Vehicle Stop Data Findings: Summary • Vehicle Stops: • Searches during Vehicle Stops: o Serious data limitations; no benchmark analyses conducted o ¾ of vehicle stops for moving traffic violations; some racial/ethnic differences in reasons for the stop o Majority of stops (73%) result in warnings (verbal or written); arrest is infrequent event (5.4% of stops) o Racial/ethnic differences in vehicle stop outcomes observed even after controlling for some other factors − Hispanic individuals 1.5 times more likely than whites to be issued citations − Black and Hispanic individuals 1.7 and 1.5 times more likely to be arrested compared to whites o Searches during vehicle stops occur infrequently (7.6% of all stops) o Searches are most likely for mandatory reasons (61.6%) – incident to arrest and inventory o Racial/ethnic differences in searches − 11.3% of Black individuals stopped, 10.0% of Hispanic individuals, 5.1% of white individuals, 2.7% other o Racial/ethnic differences in reasons for a search − Black and Hispanic individuals are more likely searched for discretionary reasons; white individuals are more likely searched for mandatory reasons o Higher percentage of Black and Hispanic discretionary searches result in contraband seizures − 31.6% of Black individuals searched, 28.3% of Hispanic individuals searched; 24.4% of white individuals searched 36 APD Arrests: January 1, 2017 – December 31, 2020 Kroll examined 128,213 total arrests from 2017 – 2020 Race/Ethnicity and Gender of Individuals Arrested Number of Arrests by Race/Ethnicity 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 2017 (n=42,098) 2018 (n=35,036) 2019 (n=30,280) 2020 (n=20,799) Black Hispanic White 1.60% 75.90% 24.10% 36.60% 33.90% Male Female 27.70% White Black Hispanic Other 37 Racial / Ethnic Differences: Custodial vs. Non-custodial Arrests • 74.3% of arrests involved physical custody 75.2% 74.9% 72.5% • Minor racial/ethnic differences in arrest type o Black individuals were slightly less likely than Hispanic and white individuals to be taken into custody when arrested, and slightly more likely to be cited and released with a court summons 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% 27.5% 24.8% 25.1% Custodial Non-Custodial Black Hispanic White 38 Arrest Rates across APD Sectors Arrest Rates per 10,000 residents by APD Sector 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 • Arrest rates differed across APD sectors, but George Sector was a clear outlier – arrest rate five times higher than the next closest sector 12,155 • George is the least populated and geographically smallest sector, but home to Austin’s entertainment district • Arrest rates cannot be explained by violent crime rate George Sector – Percentages Relative to Remainder of City 23.1% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 1.3% % Population % Offenses % Violent % Arrests % Force Crime 39 2,016 1,574 914 949 615 749 2,409 2,030 11.6% 7.2% 5.9% Arrests that include Searches: 2017 – 2020 • 79.5% of arrests involve searches • Of searches conducted, 82.6% incident to arrest • Data limitations: o APD only includes one reason for search o Temporal ordering of searches and arrests is difficult to determine (search conducted during an arrest can be the reason for the arrest or the result of an arrest) • No racial/ethnic differences in search during arrests o 80.0% of Black arrestees searched, 79.2% of both Hispanic and white arrestees • 30.3% of searches during arrests resulted in contraband seizure (8.1% in seizure of a weapon) • Overall seizure rate highest for Black arrestees • Weapon seizures: o Blacks and Hispanics = 8.5%, Whites = 7.3% Arrests with Searches: Contraband Seizures by Race/Ethnicity 34.5% 29.1% 28.5% 40% 35% 30% 25% 20% 15% 10% 5% 0% 8.5% 8.5% 7.3% Any contraband seizure Weapon Seizure Black Hispanic White 40 2017 - 2020 Arrest Data Findings: Summary • Arrests: • Searches during Arrests: o Decline in arrests over 4-year period consistent across racial/ethnic groups o In ¾ of arrests, the individual was taken into custody o Black individuals were slightly more likely to have non-custodial arrests o Arrest rates differed across APD Sectors o George Sector was an outlier − Arrest and force rates were out of proportion to population, reported crimes, and violent crime o Serious data limitations o 80% of arrests involve searches; of searches conducted, 83% are incident to arrest − Unable to determine if search/seizure is reason for arrest or result of arrest o Unlike traffic stops, no racial/ethnic differences in search during arrests o 30% of searches result in contraband seizures o Black arrestees more likely to have contraband seizures than other racial/ethnic groups − 35% seizure rate for Black arrestee searches, compared to 29% for Hispanic and white arrestees 41 Section 6: Recruitment, Selection, and Promotions Figure 1. Race/Ethnicity Comparison of APD Personnel and City of Austin Population Statistics APD Diversity 100.0% 80.0% 60.0% 40.0% 20.0% 0.0% 66.7% 48.3% 33.9% 21.8% 7.5% 7.8% 7.7% 2.5% 0.7% 4.2% White Hispanic Asian Other Black or African American Austin Police Department Austin Census Population 43 APD Diversity (cont’d) Race/Ethnicity by Rank Figure 3. APD Sworn Personnel Race/Ethnicity by Rank, March 2021 100.0% 80.0% 60.0% 40.0% 20.0% 0.0% 64.6% 68.1% 70.3% 79.7% 84.2% 50.0% 33.3% 16.7% 0.5% 1.0% 1.7% 1.7% 1.4% 1.4% 5.3% 5.3% 15.1% 9.3% 11.6% 5.8% 24.1% 7.3% 3.0% 20.7% 8.0% 1.5% Police Officer (n=1,110) Police Corporal/Detective (n=401) Police Sergeant (n=172) Police Lieutenant (n=69) Police Commander (n=19) Assistant Police Chief (n=6) White Hispanic or Latino Black or African American Asian Other 44 APD Diversity (cont’d) Rank by Gender Figure 4. APD Sworn Personnel Gender by Rank, March 2021 100.0% 89.6% 87.8% 88.4% 88.4% 94.7% 80.0% 60.0% 40.0% 20.0% 0.0% 66.7% 33.3% 10.4% 12.2% 11.6% 11.6% 5.3% Police Officer (n=1,110) Police Corporal/Detective (n=401) Police Sergeant (n=172) Police Lieutenant (n=69) Police Commander (n=19) Assistant Police Chief (n=6) Male Female 45 Figure 6. Type of Recruiting Events for APD Recruits January 2016-March 2020 Figure 7. Demographic Characteristics of APD Recruits January 2016-March 2020 Recruitment 50.0% 49.0% 46.1% 60.0% 40.0% 30.0% 20.0% 10.0% 0.0% 28.2% 22.0% 28.6% 25.7% 36.2% 38.6% 33.2% 33.4% 80% 70% 60% 50% 40% 30% 20% 10% 0% 77.7% 67.1% 32.9% 22.1% 26.1% 18.7% 4.0% 3.3% 3.4% 2.9% Colleges & Universities Career & Job Fairs, Military-Related Info Sessions White Hispanic or Latino Black or African American Asian Other Male Female Overall Recruits Recruits who Applied Overall Recruits Recruits who Applied 46 Recruitment (cont’d) Most effective recruiting events? • Regardless of gender and race/ethnicity, recruits who applied were most likely to do so if they had been recruited at general career/job fairs or information sessions. • Data Collection: APD has had difficulty accurately matching information gathered from prospective applicants at recruiting events with the online applications that are later completed ─ The actual percentages of APD personnel that are women and people of color do not mirror the level of diversity in the recruitment pool 47 Hiring and Selection Hiring and Selection Process • APD hiring process is consistent with standard police department hiring practices in the US ─ APD frequently modifies selection practices to increase retention of diverse applicants Cadet Classes 130 - 143 • 6,601 total applicants  5,890 were disqualified at some point during the process  711 ultimately became cadets at the Academy 48 Hiring and Selection (cont’d) Findings: • The current written (i.e., cognitive ability) test for applicants continues to show racial/ethnic disparities Figure 14. NDRT vs NPST Written Test Results (Percent Who Pass by Race/Ethnicity) 100.0% 89.5% 94.4% 73.5% 77.1% 80.8% 72.0% 70.8% 74.8% 88.9% 86.8% 80.0% 60.0% 40.0% 20.0% 0.0% NDRT NPST Asian Black Hispanic Other White 49 Hiring and Selection (cont’d) Physical Ability Test • No significant racial/ethnic differences  Past gender differences in Physical Ability Test failures have been eliminated Background History Statement • Black applicants were disqualified more often due to credit histories • White applicants were disqualified more often due to the polygraph, medical, or psychological exams • Drug Usage:  30% of white applicants were disqualified due to drug usage  vs. 16.2% (Black) and 21.8% (Hispanic) APD Hires • Ultimately, 66.0% of APD hires are white  vs. Hispanic - 21.5%, Black - 7.7%, Asian/Pacific Islander - 4.5% 50 APD Promotions The Promotional Process • From 2015 to 2020, Kroll found: ─ No significant gender differences in promotion outcomes ─ Asian and white promotion candidates are more likely to be promoted than Black and Hispanic candidates  The promotional written test may have an adverse impact on:  Candidates of color  Older candidates  Seniority bonus points have narrowed promotional score gaps for Black and Hispanic candidates 51 Promotions (cont’d) Assessment Center Scoring Hispanic candidates. Sergeant. Promotion Eligibility Lists  White - 60.1%  Black - 56.4%  Hispanic - 54.3% • Asian and white candidates score significantly better on the assessment centers than Black and ─ Assessment center scores have a disparate impact on Black and Hispanic promotion candidates for • Percent of candidates promoted of those who sought promotion:  Asian/Pacific Islander - 72.7% 52 Section 7: Recommendations Recommendations DATA COLLECTION RECOMMENDATIONS • Arrest Data • Use of Force Data • Traffic Stop Data • APD Policy • Training • Supervision • See Appendix to Section 7.5 (Data Fields for Traffic Stop Data) USE OF FORCE RECOMMENDATIONS 54 Recommendations ORGANIZATIONAL RECOMMENDATIONS • Examine Trends • Examine Racial/Ethnic Disparities • Monitor over time • Treat Statistical Findings as Diagnostic Tools • Adopt a Holistic Approach • Explore Other Data Sources re: Potential Factors Contributing to Racial/Ethnic Disparities • Understand Limitations of Data 55 Recommendations RECRUITMENT, SELECTION, AND PROMOTION RECOMMENDATIONS • Recruitment • Continue Intentional Efforts to Further Increase Diversity • Examine Recruiting Events – What works / What doesn’t? • Continue to Develop Community Partnerships • Improve Data Collection / Linking • Re-institute Explorer Program / Expand Internship Program • Consider Realistic Job Preview • Continue to Evaluate Disqualification Factors 56 Recommendations • Selection • Monitor Written Test Disparities • Retain Independent Consultant – Validation Study of Physical Fitness Requirements • Emphasize Necessity of Preparing for Physical Ability Test • Improve Record Keeping Process for Oral Interview Board • Promotions • Affirmatively Support Mentorship Programs • Analyze Promotional Score Data / Consider Other Assessment Centers • Reconsider Promotional Test Components and Weighting • Enhance Career Development / Training Opportunities 57 Questions? For more information, please contact: Mark J. Ehlers Managing Director 2000 Market Street, Suite 2700 Philadelphia, PA 19103 (215) 568-8305 mehlers@kroll.com Ab o u t Kr o l l © 2021 Kroll, LLC. All rights reserved. Kroll is the world’s premier provider of services and digital products related to valuation, governance, risk and transparency. We work with clients across diverse sectors in the areas of valuation, expert services, investigations, cyber security, corporate finance, restructuring, legal and business solutions, data analytics and regulatory compliance. Our firm has nearly 5,000 professionals in 30 countries and territories around the world. For more information, visit www.kroll.com. M&A advisory, capital raising and secondary market advisory services in the United States are provided by Duff & Phelps Securities, LLC. Member FINRA/SIPC. Pagemill Partners is a Division of Duff & Phelps Securities, LLC. M&A advisory, capital raising and secondary market advisory services in the United Kingdom are provided by Duff & Phelps Securities Ltd. (DPSL), which is authorized and regulated by the Financial Conduct Authority. Valuation Advisory Services in India are provided by Duff & Phelps India Private Limited under a category 1 merchant banker license issued by the Securities and Exchange Board of India.