PPD Findings Final 121922

Decembert 2022
Assessing Arrest & Traffic Stop

Patterns in Portland, ME
AN ANALYSIS OF PORTLAND POLICE DEPARTMENT DATA
Acknowledgments
Authors
Jack McDevitt, PhD, Northeastern University
George Shaler, MPH, University of Southern Maine
Sarah Goan, MPP, University of Southern Maine
Stephen Abeyta, PhD Candidate, Northeastern University
Carlos Cuevas, PhD, Northeastern University
Layout and Design
Becky Wurwarg, University of Southern Maine
Special Thanks
The project team would like to extend our gratitude to the Portland and the South Portland
police departments for their support, funding, and collaboration on this project.
Thank you to Roux Institute for their generous support in funding this work.
Thank you to Abdulkadir Ali for your contributions to the project design, careful review of
analyses, thoughtful questions, and continued engagement throughout this project.
Thank you to our Community Advisory Group for your thoughtful questions and feedback.
We would like to thank Reverend Kenneth Lewis, Reggie Parsons, and Pedro Vazquez for
your engagement throughout the project and review of the analysis and this report.
Assessing Arrest & Traffic Stop Patterns ii

Contents
Introduction1
Background
About This Report
Summary of Key Findings
Methodology 5
Data Extraction & Analysis
Samples & Population Estimates
Data & Analysis Limitations
Detailed Findings: Portland Police Data Analysis 9

Portland Police Department Arrests, 2018-2020
Portland Police Department Traffic Stops, 2018-2020
Conclusion & Recommendations 29

Conclusion
Recommendations for Portland Police Department
Recommendations for the Portland Police Department, City of Portland, and

Strategic Partners
Appendices35
Appendix A: Population Data Tables (Adults, Juveniles, Drivers)
Appendix B: Portland Police Department Patrol Areas (Beat Map)
Appendix C: Portland PD Arrests Chi-Square Results Tables
Appendix D: Portland Residents Adult Arrest Rates
Appendix E: Portland PD Traffic Data Chi-Square Results Tables
Appendix F: Portland PD Logistic Models for Multiple Regression Analysis
Assessing Arrest & Traffic Stop Patterns iii

Introduction
Background
With an authorized strength of 161 sworn officers and 59 civilian included alternative and community-based responses, as well as
employees, the Portland Police Department is the largest munic- changes to internal police policies and practices (e.g., screening,
ipal law enforcement agency in the state of Maine. In addition training, oversight).2
to uniformed operations, the department houses numerous units,
teams and programs that respond to a wide range of community In 2021, Portland’s Police Chief commissioned an analysis of the
needs, including the following: department’s data to better understand whether racial and ethnic
disparities exist in the decision to arrest individuals and to issue
• Crisis Negotiators Team, an eight-member team trained to
traffic citations and, if so, the need for any internal policy changes
respond to incidents involving suicidal behaviors, domestic
to reduce these disparities. This was an unusual step as most
crises, high-risk warrants, or barricaded suspects;
studies of racial disparities within policing are not requested
• Behavioral Health Response Program in partnership with by law enforcement agencies themselves, but rather are
the Opportunity Alliance, including a Behavioral Health conducted at the behest of state or federal oversight agencies.
Coordinator, Substance Use Disorder Liaison, Alternative The Portland PD subsequently invited the South Portland Police
Response Liaison, and Crisis Intervention Training (CIT); Department to take part in the analysis. Both departments hired
• Community Policing Unit, a unit that includes community the Catherine Cutler Institute at the University of Southern Maine
policing coordinators, community police officers, a and the Institute on Race and Justice at Northeastern University.
neighborhood prosecutor and a youth services officer; The Roux Institute of Northeastern University’s Roux Institute,
• Victim and Witness Advocate Program which employs a which has a large presence in Portland, provided additional funding
full-time advocate to provide emotional support, guidance to support this project as part of their ongoing commitment to
with investigative and court processes, paperwork, and improve the quality of life for local Maine communities.
referrals to social service agencies.
The Portland PD, like many departments across the country, faced
community outrage in the aftermath of George Floyd's killing in
About This Report
May of 2020. Following the protests held in Portland, the city
hired an outside firm to investigate the Portland PD’s response It is important to firmly place our local inquiry against the back-
to incidents which occurred at the protests in June 2020.1 That drop of national trends and to acknowledge the systematic factors
review, conducted by an independent firm with a law enforce- which contribute to myriad disparities experienced by historically
ment background, determined that Portland’s police officers had marginalized communities nationwide, even if many of them are
demonstrated a “facilitation mindset” rather than using disorder outside the scope of this targeted examination. Structural racism
control tactics and methods (which rely on the use of mass arrests and discrimination in America are widespread and deeply rooted in
and shows of force). In addition, the city established the Racial our criminal legal systems, public health systems, education systems
Equity Steering Committee which released a report with recom- and economic opportunities. Nationwide data show that BIPOC3
mendations to improve the city’s approach to public safety, which people in the U.S. have higher rates of poverty,4 higher rates of
1 For more information and access to the full text of the report, see this article
2 See the Portland city website for more information about the Racial Equity Steering Committee and to read their full report.
3 BIPOC is used throughout this report as a category that includes individuals who are Black, Indigenous, and Persons of Color.
4 Creamer, J. (September 15, 2020). Inequalities Persist Despite Decline in Poverty for All Major Race and Hispanic Origin Groups. Washington D.C.: U.S. Census Bureau, Poverty
Statistics Branch. https://www.census.gov/library/stories/2020/09/poverty-rates-for-blacks-and-hispanics-reached-historic-lows-in-2019.html
Assessing Arrest & Traffic Stop Patterns 1

INTRODUCTION
Portland
school discipline5 and disconnection,6 higher rates of experiencing and makes recommendations for next steps. The South Portland
homelessness7 and housing instability,8 and health disparities.9,10 data analysis and findings can be found in a separate report. Specif-
In Maine, racial, socioeconomic, education, and criminal justice ically, this report looks at racial, ethnic, age, gender and geograph-
disparities persist similar to national trends. For example, in 2020, ical factors which may be associated with Portland PD arrests and
39.7% of Black or African American children in Maine lived in fami- traffic citations among adults and juveniles between 2018 and 2020,
lies whose income was below the federal poverty level compared a period inclusive of both the early part of the pandemic and the
to 12.5% of White children.11 Furthermore, BIPOC, particularly Black protests.
or African American, individuals are overrepresented in every facet
of the U.S. criminal justice system.12 Previous research in Maine has The following research questions guided the analysis:
found that BIPOC people are overrepresented in Maine’s prisons
• Are there differences in arrests and traffic citations by race,
and jails.13,14 BIPOC individuals also experience higher instances of
ethnicity, gender, or age?
individual bias, discrimination, and hate crimes15 and a recent report
found that anti-Black or African American hate or bias crimes were • To what extent do associated factors, crime types,
the most frequently reported in Maine between 2008 and 2017.16 demographics, time of day, and location impact the
likelihood of getting arrested or receiving a traffic citation?
Within this context, this project examined the available arrest and • What does this data tell us about individual officer discretion
traffic citation data from the Portland and South Portland Police and how it may affect disparities in arrests?
Departments to determine if there is any evidence of dispropor-
tionate enforcement activities in either city. This report outlines
the findings of the analysis for the Portland Police Department
5 U.S. Department of Education Office for Civil Rights. (March 2014). Civil Rights Data Collection, Data Snapshot: School Discipline, Issue Brief No. 1. Washington, D.C. https://
ocrdata.ed.gov/assets/downloads/CRDC-School-Discipline-Snapshot.pdf
6 Lewis, K. (2021). A Decade Undone: 2021 Update. New York: Measure of America, Social Science Research Council. https://ssrc-static.s3.amazonaws.com/moa/
ADecadeUndone2021Update.pdf
7 Moses, J. (2018). Demographic Data Project, Part III: Race, Ethnicity, and Homelessness. National Alliance to End Homelessness. https://endhomelessness.org/demographic-data-
project-race/
8 Greene, S. & McCargo, A. (May 29, 2020). New Data Suggests COVID-19 is Widening Housing Disparities by Race and Income. Washington D.C.: Urban Institute. https://www.
urban.org/urban-wire/new-data-suggest-covid-19-widening-housing-disparities-race-and-income
9 Carratala, S., & Maxwell, C. (May 7, 2020). Health Disparities by Race and Ethnicity. Center for American Progress. https://www.americanprogress.org/article/health-disparities-
race-ethnicity/
10 Center for Disease Control and Prevention. (May 18, 2022). Health Disparities: Provisional Death Counts for Coronavirus Disease 2019 (COVID-19). https://www.cdc.gov/nchs/
nvss/vsrr/covid19/health_disparities.htm#RaceHispanicOrigin
11 The Annie E. Casey Foundation. (2022). Kids Count Data Center, Children in poverty by race and ethnicity (5-yr ACS) in Maine. https://datacenter.kidscount.org/data/tables/9738-
children-in-poverty-by-race-and-ethnicity-5-yr-acs#detailed/2/any/false/574,1729,37,871,870,573,869/10,172,9,12,1,13,185/19003
12 Sawyer, W., & Wagner, P. (March 14, 2022). Mass Incarceration: The Whole Pie 2022. Prison Policy Initiative. https://www.prisonpolicy.org/reports/pie2022.html#community
13 Vera Institute of Justice. (2019, December). Incarceration Trends in Maine. https://www.vera.org/downloads/pdfdownloads/state-incarceration-trends-maine.pdf
14 Prison Policy Initiative. (2021). Maine profile. https://www.prisonpolicy.org/profiles/ME.html
15 U.S. Department of Justice. (2020). FBI Releases 2020 Hate Crime Statistics. https://www.justice.gov/hatecrimes/hate-crime-statistics
16 Brintlinger, H., Shaler, G., & McDevitt, J. (2022). Bias and Hate Crimes in Maine: Reconciling Reported and Investigated Crimes. Maine Statistical Analysis Center. Portland, ME:
University of Southern Maine. https://cpb-us-w2.wpmucdn.com/wpsites.maine.edu/dist/2/115/files/2017/10/2022-Bias-and-Hate-Crimes-in-Maine-Report.pdf

INTRODUCTION
Portland, Maine
Portland
Summary of Key Findings
This analysis did not find statistical evidence of biased-based policing by members of
the Portland Police Department, that is, instances where an officer made a decision or
took action based on the individual’s race or ethnicity rather than the individual’s behavior.
This does not mean such incidents do not happen, but rather no patterns emerge that
demonstrate severe and persistent occurrences.
However, despite limitations in the scope of this study, this analysis does provide evidence
of some racial and ethnic disparities in arrests and traffic stops particularly among
Black or African American individuals. Further research is needed to determine to what
extent the observed disparities are the result of factors that were outside the scope of this
analysis such as specific policing practices (e.g., patrol patterns, officer-level arrest trends,
and officer attitudes), and other social circumstances (e.g., socio-economic conditions,
mental health and wellbeing, etc.).
The following key findings emerged from the analysis of the Portland Police Department
data:
BLACK OR AFRICAN AMERICAN INDIVIDUALS EXPERIENCE DISPARITIES IN ARRESTS

IN PORTLAND. Although direct comparisons between police data and the population are
limited because they are measured differently in each data source, our study found that
Black or African American individuals accounted for 17% of all arrests among Maine resi-
dents in Portland, and only 5% of the population estimate. Among Portland residents, Black
or African American individuals were arrested at an average annual rate that is higher than
White residents. Many factors which could be contributing to these disparities were outside
the scope of this study and therefore more research is recommended.
THE HIGHEST NUMBER OF ARRESTS OCCURRED IN THE WEST BAYSIDE NEIGHBOR-

HOOD (Portland Patrol Area 4) which accounted for one-third (33%) of all arrests during
the study period. However, Black or Latinx individuals were more likely to be arrested in the
East Bayside (Patrol Area 2) and Parkside (Patrol Area 6) neighborhoods compared to White
individuals.
THE ODDS OF BLACK OR LATINX INDIVIDUALS BEING ARRESTED INCREASED WHEN

THE INCIDENT WAS INITIATED BY THE OFFICER RATHER THAN INITIATED BY A
911 CALL. Overall, 49% of all arrests did not include a warrant, and 55% of arrests were
officer-initiated. Regression analysis found a 17% increase in the odds of a Black or Latinx
individual being arrested when the incident was initiated by the officer. Arrests that are
officer-initiated, with no warrant present are typically more discretionary on the part of the
arresting officer.

INTRODUCTION
Portland, Maine
Portland
• There is no evidence that race and ethnicity were related to the decision to request
multiple charges, which prior research suggests can be an indicator of biased
decision-making.17
• Use of force was reported in less than two percent of arrests across the three-year
time period. Of those, 24% involved BIPOC individuals, who made 20.4% of the total
arrested population, a difference that was not statistically significant.
PEOPLE EXPERIENCING HOMELESSNESS REPRESENTED OVER ONE-THIRD OF ALL

PORTLAND PD ARRESTS. Individuals who were unhoused at the time of their arrest were
more likely to be White, men, between the ages of 40 and 59, and also more likely to be
arrested multiple times throughout the study period. In addition, they were more likely to
be arrested in the West Bayside neighborhood (Portland Patrol Area 4) which accounted for
52% of all arrests among people who were unhoused.
BLACK OR AFRICAN AMERICAN DRIVERS EXPERIENCE DISPARITIES IN TRAFFIC

STOPS OVERALL, BUT WHITE DIVERS ARE MORE LIKELY TO RECEIVE A CITATION.
While Black or African American drivers represented only 4% of the driving age population,
they accounted for 14% of all traffic stops in Portland among Maine residents. However,
they had lower rates of citations compared to White drivers (26% vs. 37% received citations
as opposed to a warning).
• Black or African American individuals were stopped more frequently during the
nighttime (19%) than during the daylight hours (11%). This finding contradicts the
“Veil of Darkness” theory18, 19 which suggests Black or African American drivers would
be more likely stopped during the day in cases of racial profiling or discrimination.
THE MOST COMMON LOCATION FOR TRAFFIC CITATIONS WAS THE WEST BAYSIDE
NEIGHBORHOOD (16% OF TOTAL). Black or African American drivers were more likely
to receive a traffic citation in this neighborhood compared to White drivers (25% vs. 15%).
The odds of the driver being Black or African American increased by 38% when the stop
occurred in this neighborhood.
17 Roh, S. & Robinson, M. (2009). A Geographic Approach to Racial Profiling: The Microanalysis and Macroanalysis
of Racial Disparity in Traffic Stops. Police Quarterly, 12(2), 137-169.
18 Grogger, J., & Ridgeway, G. (2006). Testing for racial profiling in traffic stops from behind a veil of darkness.
Journal of the American Statistical Association, 101(475), 878-887
19 Pierson, E., Simoiu, C., Overgoor, J. et al. A large-scale analysis of racial disparities in police stops across the
United States. Nat Hum Behav 4, 736–745 (2020).

Methodology
Data Extraction & Analysis
In order to answer the research questions, the research teams from University of Southern
Maine’s (USM) Catherine Cutler Institute and the Institute on Race and Justice at North-
eastern University worked with the Portland and South Portland police departments to
develop appropriate data extracts. It is important to note that the Portland and South
Portland Police Departments provided all the data for these analyses. The staff within each
department designed and used queries to pull the data from their respective management
information systems. The data files were sent to the USM team using secure data transfer
protocols. All data files were cleaned and de-identified prior to analysis to ensure confiden-
tiality.
The arrest data included records for all adults and juveniles arrested over a three-year
period (2018-2020). As the data files were pulled in multiple extracts, the research team first
matched the files using a combination of arrest/incident ID numbers, individual names, and
dates of birth. The research team then cleaned the data files to remove test records and
standardize fields and terminology that were inconsistently recorded (e.g., some individuals
who were unhoused listed the shelter address, while others were marked as transient). In
some cases, the research team also created new variables and categories from various fields
to aid the final analysis (see definitions on the next page). Location data were also cleaned
and recoded to assign regional groups and geo-tagged for mapping purposes.
The traffic stop data included records of citations from the same period (2018-2020) for
both cities, and warnings (verbal and written) from Portland only. Data on warnings were
not available for South Portland for the period studied. These files were similarly cleaned
and combined using record ID numbers. Duplicate and test/error records were removed,
and fields were recoded to standardize terminology like the arrest data. Traffic citations
were also coded into groups to categorize the types of citations.
The research teams used statistical software to conduct analysis (descriptive statistics and
multivariate analyses). A chi-square goodness of fit test was used to compare the popu-
lation characteristics to that of the various samples. Tests of column proportions and/or
chi-square tests were used to compare groups within the samples as noted throughout the
report. In both cases, a statistically significant result indicates that the observed differences
are greater than we might expect by chance alone. In addition, to further test the association
of variables the research team conducted multiple logistic regression models. The results of
these models identify where statistical disparities may exist which may or may not indicate
incidents of bias.

METHODOLOGY
Portland, Maine
Portland
Definitions & Coding
Race Portland PD Call Source

Includes a recoded version of the original race data from the data includes three primary call-sources which define how the inci-
extract. In most cases, the data extract did not allow for multiple dent was initially reported/initiated: Administrative, 911 Calls, and
race selections and therefore the multiracial data is unavailable in Officer-initiated incidents. Officer-initiated calls are incidents that
the police data unless otherwise noted. Categories include Asian occurred because the officer was on scene and made the decision
or Pacific Islander, Black or African American, Native American to intervene. 911 calls are incidents where the officer responds
or Indigenous, or White. In the police department data, race is as a result of a call to the 911 emergency line. Administrative calls
recorded as perceived by the officer and is not self-identified by include any call that came in through the non-emergency police
the individual. department phone line.
Ethnicity Registration Violations

Includes records which were identified as being Latinx or Hispanic include failure to display or produce a registration, inspection, or
origin. This category is not exclusive with race. For example, an insurance information.
individual may be identified as both Latinx and Black or African
American. In the police department data, ethnicity is recorded as Moving Violations
perceived by the officer and is not self-identified by the individual. include but are not limited to speeding, failure to stop at a stop sign/
red light, driving the wrong way, failure to yield, distracted driving,
BIPOC driving without a valid license, and safety belt or car seat violations.
A grouped category identifying individuals as either BIPOC or
White. The BIPOC group includes all individuals who were identi- Equipment Violations
fied as Asian or Pacific Islander, Black or African American, Native include any defective vehicle violation such as a broken light, inap-
American or Indigenous, or Latinx. propriate or unsafe tires, excessive exhaust noise, tinted windows,
etc.
Unhoused
The adult arrest data extracts included a flag which identified indi-
viduals as being “homeless” or “transient” at the time of their arrest.
This information was combined with other variables to identify
all records where the individual was unhoused at the time of the
arrest. Address information was also examined and anyone with a
shelter address or motel address was also identified as unhoused.
The Portland-South Portland Metro Area

U.S. Census Metro designation which includes towns and cities
surrounding the two cities (Durham, Baldwin, Cape Elizabeth,
Casco, Chebeague Island, Cumberland, Falmouth, Freeport, Frye
Island, Gorham, Gray, Long Island, Naples, New Gloucester, North
Yarmouth, Portland, Pownal, Raymond, Scarborough, Sebago,
South Portland, Standish, Westbrook, Windham, Yarmouth,
Arundel, Biddeford, Buxton, Cornish, Dayton, Hollis, Kennebunk,
Kennebunkport, Limerick, Limington, Lyman, Old Orchard Beach,
Saco, and Waterboro).
Assessing Disproportionality
Arrest & Traffic Stop
in Patterns
Police Activities 6
METHODOLOGY
Portland
Samples & Population Estimates
The various data extracts resulted in multiple datasets (Table Table 1

1), including: Total Records by Dataset
• Adult Arrest Records for both Portland and South Portland PORTLAND PD SOUTH PORTLAND PD
• Juvenile Arrest Records for both Portland and South Adult Arrests 7,536 2,542
Portland Juvenile Arrests 166 289
• Traffic Citations (Portland and South Portland) and Warnings Traffic Citations 5,056 4,020
(Portland only).
Traffic Warnings 8,965 N/A
The Portland datasets and analysis are the focus of this report. Table 2
Population Estimates for Various Age Groups, U.S.
The research team downloaded population totals, and demo- Census 2020 Data20–28
graphics on gender, age, and race/ethnicity from various popula-
SOUTH
tion tables on the U.S. Census Bureau (2022) website. Race and
AGE GROUP MAINE PORTLAND PORTLAND
ethnicity data by age was pulled from single race data tables and
Total 1,340,825 66,706 25,665
combined. All population level data are 2020 American Community
Survey (ACS) 5-year estimates.,20,21,22,23,24,25,26,27,28 Adults (18+) 1,089,858 56,442 21,417
Juveniles (Under 18) 250,967 10,264 4,248

Population data were pulled for specific age groups: adults (18+), Drivers (15+) 1,135,578 57,948 22,311
juveniles (under 18), and drivers (15+). Table 2 shows the total popu-
lation for the three age groups.20-28 Both cities have many visitors
and commuters coming into their cities each day which changes Demographic characteristics were then calculated for each of the
the demographics of the population with whom each department sub-population groups to establish baseline population compari-
interacts. To account for this, the research team adjusted the sons for the various samples (Adult Arrests, Juvenile Arrests, and
population estimates based on the residence distribution of the Traffic Stops). While the adjusted population estimates may not
arrest and traffic stop samples. For example, Portland residents capture the true population of individuals who come into contact
accounted for 71% of the arrests made by Portland PD, and there- with police, they represent a closer approximation for comparing
fore we used Portland demographics to account for 71% of our the data given the methodological limitations. Full population
population estimates. Although out-of-state and unknown records demographic estimates for each group can be found in Appendix A.
are included in the remainder of the report, we excluded them
from the residence distribution used in this calculation because we
could not adjust the demographics appropriately.
20 U.S. Census Bureau. (2022a). 2020 American Community Survey 5-Year Estimates [Data Table: S0101].
21 U.S. Census Bureau. (2022b). 2020 American Community Survey 5-Year Estimates [Data Table: B01001B].
22 U.S. Census Bureau. (2022c). 2020 American Community Survey 5-Year Estimates [Data Table: B01001G].
23 U.S. Census Bureau. (2022d). 2020 American Community Survey 5-Year Estimates [Data Table: B01001I].
24 U.S. Census Bureau. (2022e). 2020 American Community Survey 5-Year Estimates [Data Table: B01001D].
25 U.S. Census Bureau. (2022f). 2020 American Community Survey 5-Year Estimates [Data Table: B01001E].
26 U.S. Census Bureau. (2022g). 2020 American Community Survey 5-Year Estimates [Data Table: B01001C].
27 U.S. Census Bureau. (2022h). 2020 American Community Survey 5-Year Estimates [Data Table: B01001F].
28 U.S. Census Bureau. (2022i). 2020 American Community Survey 5-Year Estimates [Data Table: B01001H].

METHODOLOGY
Portland
Data & Analysis Limitations
This analysis provides insights into the arrests and traffic stops for individual self-identifies, and thus certain groups may be under-
the cities of Portland and South Portland over a three-year period, or over-represented. In addition, the U.S. Census data includes
however, it is limited in scope. There are many factors that lead to self-reported race and ethnicity information and has its own
and influence an arrest or traffic stop interaction and outcomes, limitations in the ability to accurately count all groups; there-
and many of these factors are external or unmeasured in this anal- fore, comparisons of population data and police department
ysis. This analysis looked solely at the available records from the data must be viewed with caution.
two departments and did not consider any systemic policies or
practices, or any community or individual experience information. ɕ This study did not include input from the point of view of
In many instances, these data can tell us that there is a differ- the directly impacted individual. These events and the expe-
ence between various groups, but it does not always tell us riences of the directly impacted individual(s) are important in
why that difference exists. understanding the full context of an incident. This analysis did
not interview or survey the individuals involved in any incident
When interpreting the data and making conclusions, it is important (police officer or community member) to learn their percep-
to keep these limitations in mind. tion of the incident.
ɕ The data includes only records that lead to an arrest or ɕ This analysis did not examine policy or system level issues
warning/citation. It does not include every interaction with (at the local, state, or national level) or context. Procedures
the departments where an arrest was not made or a warning/ and policies such as staffing decisions, COVID-19 policy deci-
citation was not given. A sample that included the data on sions, mandatory arrest rules, etc. may impact these results
every interaction with the departments would offer more but are out of the scope of this study.
opportunities for analyzing the relationship between the
various characteristics and the likelihood of getting arrested or ɕ Due to the unknown influence of factors not included in
receiving a warning or citation during a traffic stop, especially this study, a statistical disparity does not necessarily point
at it pertains to race and ethnicity. to officer discrimination or racial profiling. The methods
used in this study measure the patterns and trends that may
ɕ This analysis only includes incidents where the local police indicate racial disparities and point to potential biased deci-
departments were the primary agency. This does not sion-making. However, there are many unmeasured external
include arrests or traffic incidents in either location where the factors (the individual’s economic situation, the officer’s
primary agency was a state or federal department such as the thought process, institutional or systemic factors, patrolling
Drug Enforcement Agency (DEA), Immigration and Customs patterns, etc.) that were out of the scope of this study to be
Enforcement (ICE), or the Maine State Police. In addition, given able to conclude whether an incident was one of bias. In some
the proximity of several other large towns and cities, there cases, missing data led the research team to remove some
may be cases where Portland police department was involved cases from the analysis.
in an incident, but again not the primary responding agency as
local departments sometimes collaborate. Even with these limitations, this study provides an overview of
police department data and identifies many patterns and trends
ɕ The data are reported and entered by the officers and for further exploration. The insights from this analysis also point to
identities and characteristics are not self-reported by the specific areas where the Police Department and the city can work
individuals themselves. Therefore, in some cases, the racial together to identify solutions to further improve safety and well-
and ethnic identities of the individuals may not reflect how an being of the local communities.

Detailed Findings: Portland Police Data Analysis
Portland Police Department Arrests,

2018-2020
In total, the Portland PD made 7,536 arrests between 2018 and 2020 Table 3
which involved 5,706 individual adults (18+). (Detailed arrest figures Demographics of the Full Sample of Adult Arrests by
for South Portland appear in a separate report). Of those, 952 Portland PD, 2018-2020 (n=7,536)
(17%) people were arrested multiple times during the study period,
ranging from two to 17 individual arrest incidents and accounting GENDER # OF ARRESTS %
for 37% of all adult arrests during the study period.
Men 5,773 76.6%
The majority of the arrests (77%) were individuals from the Port- Women 1,761 23.4%
land-South Portland Metro Area.29 Only a small proportion of Missing 2 0.0%
arrests were people from other areas of Maine (7%), or from out
RACE/ETHNICITY
of state (4%).30
BIPOC 1,540 20.4%
Black or African American 1,273 16.9%
Among adults (including all residences) the majority of those
Latinx/Hispanic 189 2.5%
arrested were White (79%) men (77%). One-fifth (20%) of the
Asian/Pacific Islander 77 1.0%
arrest records were individuals who were identified by the arresting
Native American/Indigenous 13 0.2%
officers as BIPOC, with 17% identified as Black or African American,
White 5,982 79.4%
and 3% identified as Latinx or Hispanic. Table 3 depicts these data
Missing 14 0.2%
in detail. The sample demographics are compared to the overall
population in detail later in this report. AGE
18–24 1,146 15.2%
25–29 1,141 15.1%
30–39 2,386 31.7%
40–49 1,485 19.7%
50–59 1,120 14.9%
60 or older 255 3.4%
Missing 3 0.0%
29 The Portland-South Portland Metro Area includes towns and cities surrounding the two cities. See definitions for the full list of towns/cities included.
30 Residence information was missing from 12% (n=903) of the arrest records.

D E TA I L E D F I N D I N G S : P O R T L A N D P O L I C E D ATA A N A LY S I S
Portland
Arrest Location
The research team examined the location of the incident to identify patterns of arrests
around the city. The Portland PD uses Patrol Areas31 as a means for dividing up the city into
11 separate areas for officers to patrol. As Table 4 shows, more than half (56%) of all arrests
occurred in three Patrol Areas: West Bayside (Patrol Area 4), Downtown (Patrol Area 3), and
the West End (Patrol Area 5). The highest number of arrests occurred in West Bayside
(Patrol Area 4) which represented one-third (33%) of all arrests.
Table 4
Number of Arrests by Location, 2018-2020, (n=7,536)
PORTLAND PD PATROL AREA NUMBER OF ARRESTS PERCENT

4 West Bayside 2,460 32.6% 33% of all arrests occurred in
3 Downtown 910 12.10% West Bayside (Patrol Area 4)
5 West End 848 11.3%
11 Oakdale/Back Cove/East Deering 564 7.5%
8 Riverton/Deering Center/Rosemont 553 7.3%
6 Parkside 533 7.1%
2 Upper East Bayside/Munjoy Hill 392 5.2%
10 North Deering/Back Cove 391 5.2%
7 Stroudwater/Libbytown/Nasons Corner 381 5.1%
1 Lower East Bayside/Munjoy Hill 239 3.2%
9 Riverton/North Deering 223 3.0%
Other (Peaks, Jetport, Out of Town) 42 0.6%
Furthermore, the research team mapped the arrests by Census Tract to identify areas with a
high density of arrests. The map on the following page displays these areas which is consis-
tent with the data in Table 4. However, there are some other areas where a high number of
arrests occurred, such as in the Back Cove neighborhood, Valley Street/Western Promenade
area, and near interstate 95 on the Westbrook-Portland line.32
31 See Appendix B for a map of the Portland PD Patrol Areas (also known as Police Beats). This includes 11 different
designations. In the table above neighborhood designations are approximate as they do not align exactly with the
Police Patrol Areas. See appendix for full list and street details.
32 The map shows Portland PD adult arrests from 2018-2020 focusing on mainland Portland. As the table shows a
small number of arrests were also made on islands or in other towns (not all of these are shown).

Portland
Figure 1
Map of Total Number of Arrests in Portland by Census Tract, 2018-2020
1-11 = Portland PD Patrol Area
Number of Arrests
0–0
0–2
2–4
4–7
7–11
11–16
16–21
21–27
9
27–33
33–40
40–52
52–61
61–80
10
80–97
8
For
97–127
e
11
st
127–145
145–161
161–183
Brig
s
183–234
es
hto
gr
n
n
Co
234–473
2
7 1
4 Fr
an
6 kl
3 in

Portland
The research team also examined arrest locations to determine if any patterns emerged
among different areas of the city and the various racial and ethnic groups. As shown in Table
5, Black and Latinx individuals were statistically more likely to have been arrested
in Patrol Areas 2 and 6, whereas White individuals were more likely to have been
arrested in Patrol Area 5. Both groups were equally likely to have been arrested in Patrol
Area 4.
Arrest locations were also mapped to compare the locations of arrests among individuals
who were Black or Latinx to those who were White. The map on the next page shows the
number of arrests by Census Tract and reveals areas of high concentrations of arrests in
certain areas of the city for Black or Latinx Individuals.
Table 5
Arrest Location of Black/Latinx Individuals Compared to White Individuals
(n=7,432)
PORTLAND PD PATROL AREA BLACK OR LATINX WHITE

6 Parkside 129 8.9%^ 402 6.70% Black and Latinx individuals were
2 Upper East Bayside/Munjoy Hill 109 7.5%^ 272 4.50%
statistically more likely to have
4 West Bayside 486 33.50% 1946 32.50%
been arrested in Patrol Areas 6
3 Downtown 165 11.40% 730 12.20% and 2
11 Oakdale/Back Cove/East Deering 122 8.40% 433 7.20%
5 West End 103 7.10% 738 12.3%^
8 Riverton/Deering Center/Rosemont 98 6.80% 449 7.50% ^ Indicates a statistically significant difference

(z-test for column proportions) between
7 Stroudwater/Libbytown/Nasons Corner 66 4.60% 308 5.10%
Black/Latinx individuals and White individuals
10 North Deering/Back Cove 67 4.60% 318 5.30% (p<.05).
1 Lower East Bayside/Munjoy Hill 51 3.50% 183 3.10% Other race/ethnicity groups (Native Amer-
9 Riverton/North Deering 47 3.20% 169 2.80% ican/Indigenous and Asian/Pacific Islander)
are not included in the table or maps due to
Other (Peaks, Jetport, Out of Town) 7 0.50% 34 0.60% small sample sizes.
Use of Force
Use of force was reported in less than two percent of arrests (146 incidents) across the
three-year time period. The most common reasons officers reported using force were
resisting officers’ requests (42%), not responding to commands (18%), or attacking an
officer (17%). When analyzed further, 24% of those instances of use of force involved BIPOC
individuals, who made 20.4% of the arrested population. However, because there were so
few instances overall, this difference was not statistically significant but may warrant further
investigation by the Portland PD.

Portland
Figure 2
Map of Total Number of Arrests Among Black/Latinx Individuals in Portland by Census Tract, 2018-2020
1-11 = Portland PD Patrol Area
Number of Arrests
0–0
0–1
1–2
2–3
3–4
4–5
5–6
6–7
7–9
9–11
9 11–13
13–15
15–17
17–21
21–25
10 25–27
8 27–31
For
11 31–33
ste
33–48
48–102
s
es
Brig
gr
hto
n
Co
n
2
7 1
4
6 Fr
an
3 kl
in

Portland
Circumstances Surrounding the Arrest
Call Source and Type of Arrest Table 6

The Portland PD data includes three primary call sources by which an Type of Arrest, Portland PD
incident is initiated: officer-initiated incidents, 911 calls, and admin-
istrative calls.33 More than half (55%) of all arrests were officer INCIDENT TYPE PORTLAND PD
-initiated incidents. Of the remaining arrests, 23% were initiated as % N
administrative calls, and 22% were initiated as 911 calls. On-view Arrest (no warrant) 49% 3,690
Summons 31% 2,332

The Portland PD arrest records were grouped into three categories
Warrant Arrest 20% 1,514
based on the outcome of the incident: summons,34 on-view arrests
(non-warrant), and warrant arrests. Nearly half of all incidents
were on-view arrests without a warrant (49%). Summons Table 7
accounted for 31% of all incidents and 20% of arrests were made Top Charges, Portland PD Adult Arrests (n= 11,339)
because the individual had an outstanding warrant. In the case of
warrants, an officer is required to make an arrest. While dependent INCIDENT TYPE PORTLAND PD
on the severity and type of offense, arrests that are officer-initiated % N
and no warrant is present tend to be more discretionary on the Warrant 17.1% 1942
part of the arresting officer.
Violation of Conditional Release 13.9% 1579
Types of Charges Criminal Traffic Violations 11.7% 1332
Charges were grouped into categories to compare trends. Most Assault 9.0% 1025
people received one (69%) or two charges (20%), with an average
Drug Possession 8.0% 909
number of 1.5 charges per incident. As Table 7 shows, top charges
included warrants, violations of conditional release, criminal traffic Criminal Trespassing 6.5% 742
violations (not including OUIs), assault, and drug possession.35 Theft 5.2% 593
Public Drinking 5.1% 583

Charges were further examined by race/ethnicity group to iden-
tify any patterns. Overall, White individuals were more likely than OUI 4.4% 496
BIPOC individuals to be arrested for a warrant (18% vs. 14%), drug
Disorderly Conduct 2.9% 329
possession (9% vs. 4%), or theft (6% vs. 4%). BIPOC individuals
were more likely than White individuals to be arrested for traffic Criminal Mischief 2.4% 274
violations (18% vs. 10%).36 Refusing Arrest 2.1% 240
Criminal Threatening 1.7% 188
Burglary 1.0% 118
Other 8.7% 989
33 Call source is the reason the incident was initiated. See the definitions in the methods section for more information about call sources.
34 Summons are an order directing the individual to appear in court and answer the charges later.
35 Total n=11,339 due to multiple charges per incident. Percentages are calculated out of the total number of charges.
36 Differences are statistically significant (z-test for column proportions, p<.05).

Portland
Arrests Compared to Population-level
Characteristics
Table 8
A population estimate was created using 2020 5-year ACS esti- Portland PD Arrests Compared to Weighted Population
Characteristics, All Maine Residents
mates20–28 to be more representative of the population with whom
the Portland PD is interacting.37 As outlined previously, comparing POPULATION SAMPLE ALL
arrests to U.S. Census population data is difficult, and observed GENDER (n= 6368) ESTIMATE YEARS
population disparities do not necessarily mean that police discrim- Men 48.4% 75.3%^
ination or biased-based policing are the cause of the disparities. Women 51.6%^ 24.7%
Nonetheless, the comparison provides a useful starting point by
which to identify areas and patterns in need of further exploration, RACE/ETHNICITY (n= 6355)
and possible solutions to mitigate existing disparities. BIPOC 12.8% 20.9%^
Black or African American 4.9% 17.2%^
As shown in Table 8, when the sample of arrested individuals is Asian/Pacific Islander 3.2%^ 1.1%
compared to the population estimate,38 some notable differences Native American/Indigenous 0.4%^ 0.1%
emerge. Similar to national trends, men are arrested more often Latinx/Hispanic 2.3% 2.6%
than women. In addition, one-fifth (21%) of the adults who were Two or More Races 2.3% 0.o%
arrested by Portland PD were identified as BIPOC, while only White, Not Hispanic 87.2%^ 79.1%
13% of the Portland PD population estimate is BIPOC. This

AGE (n= 6366)
difference is made up mostly of Black or African American individ-
18–24 11.4% 16.2%^
uals who represent only 5% of the population but account for 17%
25–29 12.8% 15.4%^
of the arrests by Portland PD during the study period. The arrest
30–39 18.7% 31.6%^
sample also includes a higher proportion of individuals under the
40–49 14.2% 19.2%^
age of 50 compared to the overall population. When we limit the
50–59 15.7%^ 14.3%
sample to only arrests of Portland residents (60% of all arrests),
60 or older 27.2%^ 3.3%
the same patterns hold.
^ Indicates a statistically significant difference between the Portland PD Popula-

tion and the sample of Portland PD arrests (where p<.05). Full chi-square test
results can be found in the appendix.
All population estimates are based on 2020 5-year ACS estimates from the U.S.
Census Bureau. See methods section for more information.
Race and ethnicity categories are not exclusive and may add to more than 100%.
37 See the methods section for more information.

38 The Portland PD interacts with individuals from locations outside of Portland. Therefore, in order to compare to population level characteristics, the research team created a
weighted average population baseline that is representative of the residences of the arrestees in the Portland PD data. This comparison excludes anyone from out of state and those
who were missing residence information. See the methods section for more information.

Portland
Arrest Rates Among Portland Residents
In 2020, the city of Portland had a population of 66,706 people with 56,442 over the age
of 18.39 Overall, 60% (4,514) Portland PD arrests were people from the city of Portland. The
arrests among Portland residents were examined in comparison to city-level population
data.27-35 This data was used to create average annual rates of arrest out of 100 Portland
residents in each subgroup.40
As shown in Figure 3, Portland residents who were Black or African American were
arrested at an average annual rate 3 times higher than White residents (8.1 vs. 2.4 out of
100). As previously noted, while this comparison to U.S. Census data is limited,41 this analysis
points to a disparity which should be used as a guideline for the Portland Police Department
to look deeper into the arrests of BIPOC individuals, particularly those who are Black or
African American.
Figure 3
Average Annual Rate of Arrest (Out of 100 People) by Portland PD, Portland
Residents Only
Portland Residents Overall, 2.7
Black/African
American, 8.1
Latinx/Hispanic, 2.9
White, Not Hispanic, 2.4
Native American/Indigenous, 1.2
Asian/Pacific Islander, 0.7
40 See the appendix for more details on this calculation.
41 The comparison of the Portland PD arrest data to U.S. Census data is limited due to differences in the ways the
data is collected (self-reported vs. officer reported) and U.S. Census limitations in representing all groups accurately.
See the methods and limitations sections for more information. Even with these limitations the U.S. Census data
provides the best comparable data set to help identify differences between the population and the sample of arrests.

Portland
Portland PD Juvenile Arrests
In 2020, Maine had an estimated 250,957 young people under the
Table 9
age of 18, with 10,264 of those youth living in the city of Portland.42
From 2018 to 2020, the Portland PD made 166 arrests of 123 Portland PD Juvenile Arrests Compared to Population
Estimates, 2018-2020 (n=165)
individual youth (under the age of 18). During the study period,
22% (n=27) of the individuals had multiple records of arrest and
POPULATION
accounted for 42% of all juvenile arrests. The majority of these
GENDER ESTIMATE (UNDER 18) PPD ARRESTS
young people were from the Portland-South Portland Metro Area
Boys 50.5% 69.1%^
(95%), including 84% of those from Portland. A small number were
Girls 49.5%^ 30.9%
youth from out of state (<1%) or other areas of Maine (4%).
RACE/ETHNICITY
Researchers calculated a weighted population estimate for the BIPOC 31.2% 48.8%^
Portland PD juvenile population using population-level character- Black or African American 18.6% 38.4%^
istics and adjusting for residence.43 As shown in Table 9, compared Asian/Pacific Islander 2.9% 0.6%
to the population estimate, the juvenile arrests during the study Native American/Indigenous 0.2% 0.0%
period were more likely44 to be boys (69%), and more likely to be Latinx/Hispanic 3.3% 9.7%^
BIPOC (49%). Black or African American youth represented Two or More Races 6.5% 0.6%
38% of all Portland juvenile arrests, but only 19% of the popu- White, Not Hispanic 68.8%^ 51.2%
lation estimate. Even with the small sample sizes, these data
demonstrate disparity in arrests among BIPOC youth in Portland AGE
which is similar to the trends in adult arrests found in this analysis 5–9 27.3% 0.6%
as well as nationwide. While some previous research has been done 10–14 27.2% 31.3%
statewide,45 additional research such as qualitative exploration and 15–17 15.3% 67.5%^
community conversations could help explain some of these trends 18 — 0.6%
among the juvenile population in Portland. The similar patterns of

disproportionate arrests of Black or African-American youth and ^ Indicates a statistically significant difference between the two subgroups
adults should spur the Portland Police Department to look deeper (p<.05).
into these arrests patterns. Race and ethnicity categories are not exclusive and may add to more than 100%.
43 This does not represent the population of the city of Portland but rather a weighted average based on the residences in the sample to account for other areas around Maine.
For more information about this population estimate see the appendix and methods sections.
44 The differences between the PPD juvenile arrests sample and the population characteristics were tested for statistical significance where p<.05. All significant differences in the
table between the arrests sample and the corresponding subgroup are noted with a “^”. Full chi-square test results can be found in the appendix.
45 See the Place Matters report series which includes several reports examining youth pathways and opportunities in Maine.

Portland
People Experiencing Homelessness in Portland
Research shows housing instability has long been associated with
Table 10
justice system involvement.46,47 In January 2020, an estimated
Portland PD Arrests by Housing Status at the Time of
2,097 people were unhoused in Maine,48 up from 1,714 in 2018,49 an
Arrest, 2018-2020 (n=165)
18% increase in just two years. This increase was likely impacted
by a combination of the economic effects from the rising housing UNHOUSED HOUSED
crisis in Maine and the influx of asylum seekers during this same GENDER (n=2,685) (n=4,851)
period. In 2021, Portland city officials reported serving 850 individ- Men 80.4%^ 74.6%
uals in city-run shelters and area hotels.50 Women 19.6% 25.4%^
From 2018 to 2020, people experiencing homelessness repre- RACE/ETHNICITY
sented 36% (2,685) of all Portland PD arrests.51 Individuals who BIPOC 17.2% 22.3%^
were unhoused at the time of their arrest were more likely to be White, Not Hispanic 82.8%^ 77.7%
White, men, and between the ages of 40 and 59 compared to those

AGE
who were housed at the time of their arrest.52
18–24 8.4% 19.0%^
25–29 11.4% 17.2%^
In addition, people experiencing homelessness were more likely
30–39 31.2% 31.9%
to be arrested multiple times throughout the study period
40–49 24.3%^ 17.2%
compared to people who were housed (1.6 arrests on average
50–59 21.2%^ 11.4%
vs. 1.1).53
60 or older 3.5% 3.3%
^ Indicates a statistically significant difference between the two subgroups

(p<.05).
46 Chapin Hall, Voices of Youth Count. (2017). Missed Opportunities: Youth Homelessness in America, National Estimates. Retrieved from https://voicesofyouthcount.org/wp-
content/uploads/2017/11/VoYC-National-Estimates-Brief-Chapin-Hall-2017.pdf
47 Gillespie, S., & Batko, S. (September 16, 2020). Feature: Five Charts That Explain the Homelessness- Jail Cycle and How to Break It. Washington D.C.: Urban Institute. Retrieved
from https://www.urban.org/features/five-charts-explain-homelessness-jail-cycle-and-how-break-it
48 U.S. Department of Housing and Urban Development. (2021). HUD 2020 Continuum of Care Homeless Assistance Programs Homeless Populations and Subpopulations Report-
Maine. Retrieved from https://files.hudexchange.info/reports/published/CoC_PopSub_State_ME_2020.pdf
49 U.S. Department of Housing and Urban Development. (2019). HUD 2018 Continuum of Care Homeless Assistance Programs Homeless Populations and Subpopulations Report-
Maine. Retrieved from https://files.hudexchange.info/reports/published/CoC_PopSub_State_ME_2018.pdf
50 Billings, R. (October 26, 2021). Portland’s unsheltered homeless population is on the rise as temperatures drop. Portland: Portland Press Herald. Retrieved from https://www.
pressherald.com/2021/10/26/number-of-unsheltered-homeless-rises-in-portland-as-temperature-drops/
51 Several variables were used to identify records where the individual was unhoused at the time of their arrest. See the methods section for more information. This includes all
records which may have been people from out of state or from other areas in Maine, as well as those with missing residence information who were also identified as unhoused in the
Portland PD data system.
52 Unhoused individuals were more likely to be men, X2(1)=32.401, p<.001; White, X2(1)=27.119, p<.001; and between the ages of 40-59, X2(5)=324.372, p=.000
53 This difference is statistically significant, (t-test of means, p<.001).

Portland
As shown in Table 11, more than half (52%) of the arrests of people who were unhoused
occurred in Portland Patrol Area 4 (West Bayside). People experiencing homelessness
were significantly more likely to have been arrested in this area compared to those who were
housed at the time of their arrest. Both groups, regardless of housing status were equally
likely to have been arrested in Patrol Areas 1, 3, and 5. However, people experiencing home-
lessness were less likely to have been arrested in every other area in the city (compared to
those who were housed).
Table 11
Arrest Location by Housing Status (n=7,536)
PORTLAND PD PATROL AREA UNHOUSED HOUSED

4 West Bayside 51.9%^ 1,394 22.0% 1,066 52% the arrests of people who
5 West End 11.4% 307 11.2% 541 were unhoused occurred in West
3 Downtown 11.1% 299 12.6% 611 Bayside (Patrol Area 4)
6 Parkside 6.1% 164 7.6%^ 369
11 Oakdale/Back Cove/East Deering 5.4% 144 8.7%^ 420
2 Upper East Bayside/Munjoy Hill 4.2% 114 5.7%^ 278
8 Riverton/Deering Center/Rosemont 3.2% 87 9.6%^ 466
1 Lower East Bayside/Munjoy Hill 2.8% 75 3.4% 164
7 Stroudwater/Libbytown/Nasons Corner 1.8% 48 6.9%^ 333
10 North Deering/Back Cove 1.0% 27 7.5%^ 364
9 Riverton/North Deering 0.7% 19 4.2%^ 204
Other (Peaks, Jetport, Out of Town) 0.3% 7 0.7%^ 35
^ Indicates a statistically significant difference (z-test for column proportions, p<.05) between people who were
unhoused and people who were housed at the time of their arrest.
The most common charges among people who were experiencing homelessness at
the time of their arrest included warrants (20%), violations of conditional release
(16%), criminal trespassing (12%), public drinking (11%), drug possession (9%), and
theft (7%). Unhoused individuals were more likely to be charged with nearly all of these
(warrants, violations of conditional release, criminal trespassing, public drinking, and theft)
compared to people who were not experiencing homelessness. Meanwhile, both groups
experienced similar rates of drug possession charges.54
54 Data not shown. Differences are statistically significant (z-test of column proportions, p<.05).

Portland
Factors Influencing the Race of the Arrested
Individual
Based on the findings from the descriptive analyses, the researchers identified several
factors which they further examined using multiple logistic regression. This approach helps
to isolate the extent to which a variable is associated with an outcome, while controlling
for the other factors in the model that might influence that outcome. The model assessed
the effect of age, gender, housing status, call source, time of day, and Patrol Area on the
likelihood of an arrested individual being Black or Latinx (dependent variable). Because an
officer in Portland is required to make an arrest if an individual has an outstanding warrant,
the researchers included outstanding warrants in the model as well.
Figure 4
Percent Increase/Decrease in Odds of the Arrested Individual Being Black/Latinx
Age Under 40 125%

Male 99%
Patrol Area 2 51%
Patrol Area 6 28%
Time 12AM to 3AM 21%
Time 9PM to 12AM 21%
Officer-Initiated Incident 17%
Unhoused -15%
Warrant -26%
Patrol Area 5 -40%
Overall, Figure 4 shows that the odds of the arrestee being Black or Latinx increased
when the individual was under the age of 40 and male, when the incident occurred
in Patrol Areas 2 or 6, during the hours of 9:00 pm to 3:00 am, and when the inci-
dent was officer-initiated. The odds of an individual being Black or Latinx decreased when
the incident occurred in Patrol Area 5, the individual was unhoused, and when they had
an outstanding warrant.55 Although these variables accounted for only 7% of the variance,
which suggests that many factors which influence arrests by race/ethnicity are external to
this analysis, the analysis upholds the findings of disparities shared in the previous sections.
In addition, a key finding from this model is that officer-initiated incidents are more likely
to result in the arrest of a Black or Latinx individual. In fact, the model shows a 17%
increase in the odds of a Black or Latinx individual being arrested when the incident was
initiated by the officer.
55 This logistic regression model showed that the above variables were all significantly associated with the arrested
individual being Black or Latinx (X2(10)=296.673, p<.001). These variables explained 6.8% (R2) of the variance in the
race/ethnicity and correctly predicted 80.4% of the cases. See Appendix F for the full model results.

Portland
Multiple Charges Analysis
The decision to request multiple charges from the District Attorney
is typically a discretionary decision by a police officer making the
arrest. Prior research has suggested that race or ethnicity may have
an impact on the number and severity of charges which police offi-
cers request, and has found that officers request multiple charges
in cases they believe are a high risk to the community.56 For this
study, the researchers conducted multiple logistic regression anal-
ysis to examine the association between gender, race, age, location,
housing status, call source, and time of day with the decision to
request multiple charges.
Overall, the results of the model57 showed that the odds of an

individual receiving more than one charge increased when the
incident was a 911 call, when the individual was unhoused, when
the individual was between the ages of 25 and 39, when the inci-
dent occurred in Patrol Areas 6, 7, or 8, and when the incident
occurred between 3:00 am and 6:00 am. It is important to note
that in this analysis the race or ethnicity and the gender of the
person arrested were not significantly related to the decision
to request multiple charges. The model supports the findings
in the previous sections of this report showing that people who
were experiencing homelessness at the time of their arrest were
more likely to receive multiple charges. In fact, people who were
unhoused had a 27% increase in the odds of receiving multiple
charges. In addition, 911 calls resulted in a 56% increase in the odds
of receiving multiple charges. This model only accounted for 2% of
the variance in multiple charges and therefore additional research
is needed to determine other factors which may be influencing
these requests that are external to this analysis.
56 Roh, S. & Robinson, M. (2009). A Geographic Approach to Racial Profiling: The Microanalysis and Macroanalysis of Racial Disparity in Traffic Stops. Police Quarterly, 12(2), 137-
169.
57 This logistic regression model showed that the noted variables were all significantly associated with multiple charges being requested (X2(8)=101.408, p<.001). These variables
explained 2.1%(R2) of the variance and correctly predicted 68.9% of the cases. See Appendix F for the full results.

Portland
Portland Police Department Traffic Stops, 2018-2020
From 2018 to 2020 the Portland PD conducted 14,021 traffic stops which resulted in either
a warning or citation. In total, Portland PD gave out 5,056 citations and 8,965 warnings over
the three-year period. As seen in Table 12, citations represented 36% of the total sample.
Reviewing citations is particularly important since they represent a financial burden to those
who receive them both in the short term, as a result of the fine imposed by the citation, and
in the long term, as a result of possible increases in insurance premiums due to the citation.
It is important to understand if the financial burden imposed by citations falls disproportion-
ately on one group or another.
Table 12
Traffic Stops Resulting in a Warning or Citation, Portland PD 2018–2020
PORTLAND PD
TRAFFIC STOPS 2018 2019 2020 TOTAL % OF TOTAL
Warnings 3,902 3,653 1,410 8,965 64%
Citations 2,047 2,110 899 5,056 36%
Total 5,949 5,763 2,309 14,021 —
In both 2018 and 2019 the total stops were just under 6,000, however, in 2020 there was a
large drop in stops overall (down 60%) likely due to the COVID-19 pandemic with only 2,309
stops in total for 2020. Nonetheless, the number of citations as a proportion of all stops
remained about the same across all three years (34.1% to 38.9%).
Just under half (49%) of all citations were individuals from the city of Portland. About
one third (34%) of the citations were individuals from other Portland-South Portland Metro
Area locations,58 and another 10% were individuals from other areas of Maine. Only 3% of all
citations were people from out of state. 59
58 The Portland-South Portland Metro Area includes towns and cities surrounding the two cities. See definitions for the full list
of towns/cities included.
59 There were 185 (3.7%) citation records with missing residence data. The warnings records did not include any information
on the individual’s residence.

Portland
Overall, the majority of drivers stopped were men or boys (60%). Table 13
However, as shown in Table 13 women and girls received citations Portland PD Traffic Stops Resulting in a Warning or
(vs. warnings) at a slightly higher statistically significant rate (37% Citation, 2018–2020
vs. 34%) compared to men and boys. In addition, while the majority
of drivers stopped were White (78%), BIPOC drivers represented Total Stops Total Issued
17% of all stops which is significantly higher than population esti- GENDER N % Citatons (%)
mates (12% BIPOC). More specifically, Black or African American Men/Boys 8,472 60.4% 34.0%
drivers represented 14% of all stops, while only making up Women/Girls 5,266 37.6% 37.3%*
4% of the population estimate.60 However, on average, drivers

RACE/ETHNICITY
who were BIPOC received citations at significantly lower rates
BIPOC 2,361 16.8% 27.4%
compared to drivers who were White (27% vs. 37%).61
Black or African American 2,008 14.3% 26.4%
Asian/Pacific Islander 314 2.2% 27.7%
Native American/Indigenous 11 0.1% 18.2%
Time of Day Latinx/Hispanic 28 0.2% 100.0%
White, Not Hispanic 11,176 79.7% 36.6%*
Previous research has suggested one way to examine the presence
of bias in traffic stops is to examine the time of day and whether AGE
the stop occurred at night or during daylight hours.62,63 This is 15–17 89 0.6% 100.0%
commonly referred to as the “veil of darkness” hypothesis, which 18–24 2,426 17.3% 29.0%*
assumes that a police officer’s bias in initial stops may be reduced 25–29 2,287 16.3% 34.6%
at night because the visibility of the driver (and their race) is limited. 30–39 3,294 23.5% 35.9%
Time associated factors were analyzed, and all traffic stops (warn- 40–49 2,220 15.8% 36.8%
ings and citations) were coded based on the incident time to esti- 50–59 1,796 12.8% 39.4%
mate whether the incident took place during daylight hours or 60 or older 1,474 10.5% 37.0%
nighttime. In total, 49% (n=6,861) of the traffic stops which resulted
in a warning or citation took place during nighttime hours. ^Indicates a statistically significant difference (z-test of column proportions,
p<.05) between % of warnings and % of citations for the group highlighted.
* The last column includes a calculation of the total proportion of individuals

within each group who received a citation rather than a warning. For example,
while men/boys represented 60% of all traffic stops, only 34% of those men/
boys who were stopped received a citation and the remaining 66% received a
warning. Women and girls were had significantly higher (p<.05) rates of citations
compared to men and boys. In addition, White drivers had significantly higher
rates of citations compared to BIPOC drivers, while 18- to 24-year-old drivers
had significantly lower citation rates compared to those 25 or older.
60 Population data (Appendix A) and chi-square results (Appendix C) can be found in appendix. In addition, the differences between the population estimate and traffic citations are explored in more
detail later in this report.
61 BIPOC drivers overall were found to have statistically significantly lower rates of citations compared to White drivers, (X2(1)=72.539, p=.000).
62 Grogger, J., & Ridgeway, G. (2006). Testing for racial profiling in traffic stops from behind a veil of darkness. Journal of the American Statistical Association, 101(475), 878-887
63 Pierson, E., Simoiu, C., Overgoor, J. et al. A large-scale analysis of racial disparities in police stops across the United States. Nat Hum Behav 4, 736–745 (2020).

Portland
Table 14
Portland PD Traffic Stops by Time of Day and Race/Ethnicity, 2018–2020
Daylight (n=7,023) Darkness (n=6514)
RACE/ETHNICITY N % N %
BIPOC 916 13.0% 1,445 22.2%^
Black or African American 750 10.7% 1,258 19.3%^ Contrary to the “veil of darkness”
Asian/Pacific Islander 136 1.9% 178 2.7%^ theory, Black or African American
Latinx/Hispanic 25 0.4% 3 0.0% individuals were stopped more
Native American/Indigenous 5 0.1% 6 0.1%
frequently during the nighttime
White, Not Hispanic 6,107 87.%^ 5,069 77.8%
than the daylight.
^ Indicates a statistically significant difference between the daylight and darkness subgroups (p<.05).
The Veil of Darkness theory suggests that in cases of discrimination, Black or African Amer-
ican drivers would be more likely to be stopped during the day since it would be easier for
officers with conscious or unconscious biases to determine the race of the driver. As shown
in Table 14, contrary to the theory, Black or African American individuals were stopped
more frequently during the nighttime (19%) than during the daylight hours (11%).64
More specifically, Black or African American drivers were more likely to be stopped between
the hours of 9:00 PM and 2:59 AM. In fact, among Black or African American drivers, 47%
of all stops occurred during those hours compared to 30% of stops for all other race and
ethnic groups.65 While there are a number of citicisms of the “Veil of Darkness” approach,
according to this theory, this data does not indicate widespread profiling in traffic stops
based solely on the driver’s race or ethnicity during this study period.
64 Black or African American individuals were more likely to be stopped at night than during the daylight hours
and White individuals were more likely to be stopped during the day, X2(4)=229.105, p=.000. While sample sizes for
other groups were small, BIPOC individuals as a group were more likely to be stopped at night compared to the day,
X2(1)=196.072, p=.000
65 Data not shown. Black or African American individuals were more likely to have been stopped compared to all
other race/ethnicity groups between 9:00 PM and 11:59 PM (28% of stops among Black drivers compared to 20% for
all other groups) and between 12:00 AM and 2:59 AM (20% of all stops among Black drivers compared to 10% for all
other groups), X2(7)=306.701, p=.000.

Portland
Traffic Citations Compared to Population-level
Table 15
Characteristics Portland PD Traffic Citations Among Maine Residents
Compared to the Population Estimate, 2018-2020
A population estimate for drivers (age 15+) was created using 2020
ACS 5-year estimates and adjusted to make it comparable to the POPULATION CITATIONS
GENDER (n= 4671) ESTIMATE SAMPLE (n=4,695)
sample of traffic citations. 27-35 While the previous “Veil of Darkness”
Men 48.5% 59.3%^
analysis did not find evidence of racial profiling in traffic citations,
Women 51.5%^ 40.7%
compared to the population estimate men, BIPOC individuals,
and people ages 18-39 are overrepresented in the sample of RACE/ETHNICITY (n=4573)
Portland PD citations.66 In particular, people who were identified BIPOC 11.6% 13.9%^
as Black or African American are overrepresented compared to the Black or African American 4.2% 11.4%^
population of driving aged individuals (11% of citations vs. 4% of Asian/Pacific Islander 2.8%^ 1.8%
the population). These findings are consistent with the trends and Native American/Indigenous 0.5%^ 0.0%
disparities identified in the arrest records analysis and may indicate Latinx/Hispanic 2.2%^ 0.6%
other factors involved that are outside the scope of this study. Two or More Races 2.4% 0.0%
White, Not Hispanic 88.4%^ 86.1%
AGE (n= 4664)

Under 15 0.1%
15–17 3.30% 1.9%
18–24 10.80% 14.5%^
25–29 11.10% 16.1%^
30–39 17.10% 24.5%^
40–49 13.90% 16.9%
50–59 15.80% 14.8%
60 or older 28.0%^ 11.2%
^ Indicates a statistically significant difference between the Portland PD Popula-

tion and the sample of Portland PD citations (where p<.05). Full chi-square test
results can be found in Appendix C, Table 25.
66 The population estimate and sample compared exclude people from out of state and those with missing residences. The population estimate is a weighted average of Maine
and city-level data adjusted to be comparable to the residence of the individuals in the citations sample. See the methods section for more information and Appendix A for more
details on the population data.

Portland
Table 16
Type of Citation Violation Issued, Portland PD
TOTAL BLACK DRIVERS WHITE DRIVERS

N % N % N %
Registration Violations 2,986 57.7% 290 51.7% 2,560 58.9%^
Moving Violations 2150 41.5% 266 47.4%^ 1753 40.4%
Speeding 974 18.8% 79 14.1% 833 19.2%^
Failure to Stop 688 13.3% 91 16.2% 556 12.8%
Other Moving Violation 133 2.6% 17 3.0% 105 2.4%
Operating Without a Valid License 121 2.3% 35 6.2%^ 80 1.8%
Distracted Driving 108 2.1% 6 1.1% 96 2.2%
Seat belt or Child Car seat Violation 88 1.7% 11 2.0% 74 1.7%
Permit Violation 38 0.7% 27 4.8%^ 9 0.2%
Equipment Violations 39 0.8% 3 0.5% 29 0.7%
Excessive Noise 20 0.4% 2 0.4% 14 0.3%
Defective Vehicle Violation 19 0.4% 1 0.2% 15 0.3%
Other 4 0.1% 2 0.4% 2 0.0%
Type of Citation
Citations were grouped into four categories: registration violations, The most common type of moving violation was speeding which
moving violations, equipment violations or other.67 As Table 16 accounted for 19% of all citations. Additionally, failing to stop at
shows, 68 the most common type of citations violations were a red light/stop sign/etc. accounted for 13% of all citations. While
registration violations which represented 58% of all citations Black or African American drivers were more likely to receive
between 2018 and 2020. Another 42% of citations were moving moving violations overall, White drivers were more likely to receive
violations, and a small number (1%) were equipment violations. a violation for speeding compared to Black or African American
Drivers who were identified as Black or African American were more drivers (19% vs. 14%). Black or African American drivers were more
likely to have received a moving violation compared to drivers who likely to have received a citation for operating without a valid/with
were White (47% vs. 40%). Meanwhile, drivers who were White a suspended license compared to White drivers (6% vs. 2%). While
were more likely than Black or African American drivers to have violations related to learner’s permits (driving with non-family
received a registration violation (59% vs. 52%).69 passengers, driving after curfew, etc.) were only 1% of citations
overall, Black or African American drivers were also more likely
than White drivers to have received a citation for a permit viola-
tion (5% vs. 0.2%).
67 Registration violations include failures to display/produce registration, insurance, or inspection information. See the methods section for more details on what is included in
each category.
68 Other race/ethnicity groups (Latinx, Native American/Indigenous, Asian/Pacific Islander) are not shown in the table due to small sample sizes.
69 Black drivers were more likely to receive moving violations and White drivers were more likely to receive registration violations, X2(6)=21.949,p=.001.

Portland
Table 17
Type of Citation Violation Issued, Portland PD
TOTAL BLACK DRIVERS WHITE DRIVERS

PORTLAND PD PATROL AREA N % N % N %
4 West Bayside 2,986 57.7% 290 51.7% 2,560 58.9%^
11 Oakdale/Back Cove/East Deering 2150 41.5% 266 47.4%^ 1753 40.4%
10 North Deering/Back Cove 974 18.8% 79 14.1% 833 19.2%^
9 Riverton/North Deering 688 13.3% 91 16.2% 556 12.8%
7 Stroudwater/Libbytown/Nasons Corner 133 2.6% 17 3.0% 105 2.4%
6 Parkside 121 2.3% 35 6.2%^ 80 1.8%
8 Riverton/Deering Center/Rosemont 108 2.1% 6 1.1% 96 2.2%
2 Upper East Bayside/Munjoy Hill 88 1.7% 11 2.0% 74 1.7%
3 Downtown 38 0.7% 27 4.8%^ 9 0.2%
5 West End 39 0.8% 3 0.5% 29 0.7%
1 Lower East Bayside/Munjoy Hill 20 0.4% 2 0.4% 14 0.3%
Other (Peaks, Jetport, Out of Town) 19 0.4% 1 0.2% 15 0.3%
Location of Traffic Citations

Traffic citations were more dispersed around the city compared to
arrests. However, similar to arrests, the top location for traffic
citations was also the West Bayside neighborhood (Patrol
Area 4) where 16% of citations occurred. In addition, Patrol
Areas 10 and 11 (North Deering/Back Cove and Oakdale and East
Deering) were also top locations for traffic citations.
The research team further examined Patrol Areas by the drivers’

race/ethnicity to identify any patterns in specific areas of the
city. As shown in Table 17, Black or African American drivers were
more likely to receive a citation in the West Bayside neighborhood
(Patrol Area 4) compared to White drivers (25% vs. 15%). White
drivers were more likely to receive a citation in the Downtown area
(Patrol Area 3) and in the North Deering/Back Cove area (Patrol
Area 10) compared to Black or African American drivers (7% vs. 4%
and 13% vs. 9%).70
70 In the table above, ^ indicates a statistically significant difference between Black or African American drivers and White drivers (z-test for column proportions, p<.05). Other
race/ethnicity groups (Latinx, Native American/Indigenous, Asian/Pacific Islander) are not shown due to small sample sizes.

Portland
Factors Influencing the Race of the Driver Who Was Stopped
Based on the findings from the descriptive analysis, the researchers identified several factors
which they further examined using multiple logistic regression. As stated previously, this
approach helps to isolate the extent to which a variable is associated with an outcome, while
also controlling for other the other factors in the model that might influence that outcome.
The model assessed the effect of age, gender, time of day, and Patrol Area on the likelihood
of the driver stopped being Black or African American (dependent variable).71
As shown in Figure 5, the odds of the driver being stopped being Black or African Amer-
ican increased when the driver was male, aged 18 to 24, and the incident occurred in
Patrol Area 4 and at night between the hours of 9 PM and 3 AM. In addition, the odds of
the driver being Black or African American decreased when a citation was issued and when
the incident occurred in Patrol Area 3 (Downtown). These variables accounted for only
7% of the variance in the driver’s race, however, meaning that many crucial factors are not
captured by this analysis.72
Figure 5
Percent Increase/Decrease in Odds of the Stopped Individual Being Black or

African American
Male 106%
Time 12 AM to 3 AM 95%
Driver Age 18 to 24 60%
Time 9 PM to 12 AM 50%
Patrol Area 4 38%
Officer Age -2%
Citation Issued -22%
Patrol Area 3 -45%
71 While the arrest data regressions used Black/Latinx as a group for the dependent variable, the traffic records
included only a small number (n=28) identified as Latinx and thus these records were excluded from the regression
analysis.
72 This logistic regression model showed that the above variables were all significantly associated with the driver
being Black or African American (X2(8)=543.844, p<.001). These variables explained 7.0% (R2) of the variance in the
race/ethnicity and correctly predicted 85.2% of the cases. See Appendix F for the full regression results.

Conclusion & Recommendations
Conclusion
The analysis did not find statistical evidence of biased-based policing by members of the
Portland Police Department, that is, instances where an officer made a decision or took
action based on the individual’s race or ethnicity rather than the individual’s behavior. In
particular, our analysis does not find significant evidence that race and ethnicity were
related to the decision to request multiple charges, which prior research suggests can be
an indicator of biased decision-making.73 Additionally, in terms of traffic enforcement this
analysis found that for drivers stopped in Portland, White drivers were more likely to receive
a citation compared to BIPOC drivers. And contrary to the Veil of Darkness theory, Black or
African American individuals were not stopped more frequently during daylight hours. This
does not mean such incidents do not happen, but rather no patterns emerge that demon-
strate severe and persistent occurrences; indeed, identifying officer attitudes and individual
acts of bias-based policing by individual police officers was beyond the scope of this report.
However, despite limitations in the scope of this study, the analysis does provide some
evidence of racial and ethnic disparities in arrests and traffic stops, particularly among
Black or African American individuals, when compared to the proportion of Black or African
American individuals in the population estimates. Among Portland residents, Black or African
American individuals were arrested at an average annual rate that is three times higher than
White residents. Specific factors influencing these disparities include officer-initiated arrests
(where the odds of a of a Black or African American or Latinx individual being arrested
increased by 17%) and location (specifically, Patrol Areas 2 and 6). In addition, this analysis
found disparities involving people who were experiencing homelessness at the time of their
arrest, who make up more than one-third (36%) of all arrests in Portland. More than half of
the arrests among people experiencing homelessness occurred in the West Bayside neigh-
borhood which also includes many of the city shelters. Those individuals were more likely to
receive multiple charges.
The research team has identified many next steps which could help the Portland PD dig
more deeply into these particular areas, such as changes to data collection and further anal-
ysis, including at the officer level. Indeed, many factors which were external to this analysis
should be explored further in order to better understand the patterns observed in arrests
and traffic stops in Portland. The first set of recommendations below are for the Portland
PD, specifically, and are activities that the Department can undertake unilaterally to improve
their practice. The second set of recommendations will require the Portland PD and the
city of Portland to collaborate with local social service providers and community partners
to work towards these common goals. It is our firm belief that undertaking the best prac-
tices outlined below will help to reduce the disparities observed in this study, support local
communities, and bridge relations between the police department and the people they
serve.
73 Roh, S. & Robinson, M. (2009). A Geographic Approach to Racial Profiling: The Microanalysis and Macroanalysis
of Racial Disparity in Traffic Stops. Police Quarterly, 12(2), 137-169.

CO N C LU S I O N & R ECO M M E N DAT I O N S
Portland
Recommendations for Portland Police Department
The Portland PD should explore additional research and analysis to address the limitations
of this study and further explain some of the findings, particularly those involving racial and
ethnic disparities. In addition, making system-wide updates to data collection and imple-
menting ongoing monitoring processes would allow for better transparency, accountability,
and more effective analysis moving forward. These recommendations are also aligned with
statewide efforts and recent legislation74 to improve the data collection and procedures
associated with traffic stops in order to eliminate racial profiling, as well as recommendations
from the city of Portland’s Racial Equity Steering Committee.75 Moreover, these represent
areas where the Portland PD can lead by example in supporting anti-racism and anti-bias
policies, practices, and programs.
01. Promote transparency, engagement, and monitoring.
Portland PD leadership should initiate and participate in community meetings to share and
discuss these findings with community groups, city leadership, organizations concerned
about racial equity, and local human rights commissions. In addition, the department should
expand their current practices for sharing data via social media and with the City Council,
including periodic presentations, which will allow for ongoing reporting and monitoring of
key data points and transparent reporting of race and ethnicity statistics to the public.
02. Implement data collection improvements.
The first step to being able to identify patterns of racial profiling and bias is to collect
accurate and complete data on all interactions that police officers have with community
members, regardless of the outcome of that interaction. This study revealed data quality,
gaps, and tracking issues which should be addressed to improve data quality and efficiency
for future research and monitoring purposes. The data extraction required a complex, time
consuming process to gather data from multiple database tables. In many cases, information
was not available and there were a large number of errors, inconsistencies, and missing data.
As the department rolls out the new data management system, the following are specific
items that could improve the efficiency and quality of reporting:
• Inconsistencies in coding and naming conventions limit efficient and quality reporting
abilities. Standardized naming and procedures should be implemented to reduce this
in the future.
• Exploring officer-initiated incidents in greater depth (e.g., by officer, type of charges,
patrol patterns, time of day, etc.) could offer some explanation to disparities
74 For more information about this law see LD132/HP88 An Act to Implement the Attorney General’s
Recommendations on Data Collection in Order to Eliminate Racial Profiling in Maine.
75 Abdurraqib, S. (April 26, 2021). City of Portland Racial Equity Steering Committee, Full Report. https://content.
civicplus.com/api/assets/41c75af1-d867-4dc0-b4af-39af4a6470d8

Portland
observed in this study.
• The department should explore ways to capture data on all interactions that police
officers have with community members, including incidents where an arrest was not
made, or a warning/citation was not issued.
03. Conduct additional research.

The limitations in this study mean additional research is needed to further understand the
trends and patterns detected in the data and determine if there is evidence of individual
officer bias. Specifically, a high priority for the department should be an internal review of
officer-initiated arrests that examines the discretionary nature, individual officer level anal-
ysis, and any additional information that is available on the reason behind the initial inter-
action and events that occurred. Specific attention should be paid to arrests that include
highly discretionary activities such as drinking, drug use, or disorderly conduct. This new
analysis should compare (and control for) patterns of officers who work in similar areas
and on similar shifts to determine if any officers are outliers in terms of the groups they
arrest or cite. The department should also explore the patrol patterns in the neighborhoods
identified in this analysis specifically where Black or Latinx individuals are more likely to be
arrested (Patrol Areas 2 and 6). Because the numbers are low, but the disparities do exist,
some additional qualitative research to explore juvenile arrests could also be beneficial to
the department.
04. Continue to fund and implement ongoing anti-bias efforts for all
staff.
Our analysis found some evidence of disparities related to officer-initiated arrests, which
suggests that the Portland PD should continue to implement and/or expand measures to
reduce officer bias and increase accountability, such as annual training on systemic racism
and implicit bias, and adequately assessing applicants for bias. These strategies are strongly
supported by national research as effective means to reduce racial inequities in the crim-
inal justice system76 and align with the city of Portland’s Racial Equity Steering Commit-
tee’s (RESC) recommendation to work with consultants to design and implement regular
staff assessments of bias and provide ongoing anti-racism training for all city employees,
including law enforcement. The RESC made several recommendations for specific consul-
tants including the Mid Atlantic Equity Consortium (MAEC) or GARE.77 In addition, the Fair
and Impartial Policing Curriculum developed by the University of Southern Florida is another
that might be considered.
76 The Sentencing Project. (2008). Reducing Racial Disparity in the Criminal Justice System, A Manual for
Practitioners and Policymakers. Washington, D.C. https://www.ojp.gov/ncjrs/virtual-library/abstracts/reducing-racial-
disparity-criminal-justice-system-manual
77 Abdurraqib, S. (April 26, 2021). City of Portland Racial Equity Steering Committee, Full Report. https://content.
civicplus.com/api/assets/41c75af1-d867-4dc0-b4af-39af4a6470d8

Portland
05. Continue to develop recruitment and advancement pathways to
increase staff diversity.
Although not included elsewhere in our analysis because the numbers were so small, only
3% (n=5) of the officers listed in the arrest records were BIPOC. In keeping with best prac-
tices, the Portland PD should continue to develop recruitment and advancement pathways
that attract and retain a more diverse police force, drawing on the diversity of the local
community. The department should also review its current policies and procedures to see
what changes can be made to better support these efforts. The Portland Police Depart-
ment should consider a more decentralized approach to recruitment efforts that empowers
local patrol officers to identify and mentor youth, new Mainers, and other local community
members who may make excellent future police officers.

Portland
Recommendations for the Portland Police Department,
City of Portland, and Strategic Partners
National best practices recommend community-based models for crisis response and
community policing models as a way to reduce racial disparities. In addition, providing
supports and appropriate responses for individuals experiencing homelessness minimizes
the homelessness-jail cycle. The city of Portland and Portland PD currently fund and
implement a number of best practices in community-based policing, crisis response, and
responses to homelessness all of which should be expanded and leveraged to address racial
disparities. Many of these same recommendations were also outlined in detail in the City
of Portland’s Racial Equity Steering Committee report; this study provides more data to
support these needs in Portland to help reduce disparities.
06. Expand capacity for community-based crisis response.
In 2020, the Vera Institute of Justice published a comprehensive overview of alternative

models of crisis response which outlines the various options for internal police crisis
response teams, co-response options, and community-based models.78 Portland PD has a
behavioral health program, which includes a full-time coordinator, three full-time liaisons
who co-respond with officers, and Crisis Intervention Training (CIT) offered by NAMI Maine.
However, these approaches are police-led responses and police co-response options as
outlined by the Vera report but are not community-based models. The city of Portland
should consider ways to enhance the community-based models available in Portland by
expanding and prioritizing mobile crisis response teams that are external to the police, and
free up police resources to focus on issues of public safety.
07. Promote efforts that support and decriminalize homelessness.
With over one-third of arrests in Portland involving individuals who were unhoused, this
analysis echoes previously cited national research which demonstrates the cycle of home-
lessness, police interactions, and justice system involvement. The department currently
collaborates Homeless Outreach and Mobile Engagement Team (HOME Team) operated by
Milestone Recovery Services. Given the considerable number of arrests among people expe-
riencing homelessness in Portland, the city of Portland and the Portland PD should continue
to implement strategies that support these individuals and explicitly decriminalize common
activities such as loitering, sleeping in public places, and public urination. In addition, as
outlined previously, community-based response teams and mental health and substance use
programs can better support individuals experiencing homelessness in Portland. The city
should expand resources with local community-based organizations and the HOME Team
to support these efforts and provide more alternative responses to these issues, including
continuing to expand access to public restrooms.
78 Beck, J., Reuland, M., & Pope, L. (November 2020). Behavioral Health Crisis Alternatives, Shifting from Police to
Community Responses. New York: Vera Institute of Justice. https://www.vera.org/behavioral-health-crisis-alternatives

Portland
08. Pilot a community-based public safety model
Given the high concentration of BIPOC arrests found in some areas, a community-based
public safety model should be considered in Portland, particularly in those areas which are
predominantly BIPOC.79 Portland PD currently has a community policing program80 that
includes a community policing officer assigned to each city sector, community policing
centers with civilian community coordinators assigned to each, and a neighborhood pros-
ecutor program. The current community policing program aims to foster greater commu-
nication, understanding and trust between police and the communities in which they work.
However, in recent years, many advocacy groups and community leaders in Maine and
nationally have criticized community policing programs because they do not recognize the
historical trauma of over policing BIPOC communities and leave the power to arrest with
the police officer. In contrast, community-based public safety programs empower local citi-
zens and community-based organizations, which are external to police departments, with
the authority to address minor public safety issues and partner with or refer to police
departments as they see fit. In Portland, a partnership between the current community
policing centers and a new community-based, citizen-led program could help to amelio-
rate the observed racial disparities while also maintaining a partnership for larger issues.
Given the findings in this study, the city should focus on piloting these programs in areas
where there is a high concentration of arrests overall (Patrol Area 4) and among Black or
African American and Latinx individuals (Patrol Areas 2 and 6). The Urban Institute (2018)
published a guide81 on community-driven safety initiatives which outlines key considerations
and funding models, as well as examples of similar initiatives in other cities.
79 A community-based safety model was also recommended by the City of Portland’s Racial Equity Steering
Committee (Abdurraqib, 2021).
80 See www.portlandmaine.gov/464/Community-Policing-Officers for more information.
81 See the guide here.

Appendices
Appendix A: Population Data Tables (Adults,

Juveniles, Drivers)
Table 18
Adult (18+) Population Data and Weighted Average
PPD
MAINE PORTLAND WEIGHTED AVG.
Total 18+ Population (2020) 1,089,858 56,442
GENDER
Men 48.4% 48.5% 48.4%
Women 51.6% 51.5% 51.6%
RACE/ETHNICITY
BIPOC 6.3% 15.3% 12.8%
Black or African American 1.2% 6.3% 4.9%
Asian/Pacific Islander 1.1% 4.0% 3.2%
Native American/Indigenous 0.6% 0.4% 0.4%
Latinx/Hispanic 1.5% 2.6% 2.3%
Two or More Races 2.1% 2.3% 2.3%
Some Other Race 0.3% 0.6% 0.5%
White, Not Hispanic 93.7% 84.7% 87.2%
AGE
18–24 9.9% 11.8% 11.4%
25–29 7.4% 14.9% 12.8%
30–39 14.5% 20.4% 18.7%
40–49 14.8% 14.1% 14.2%
50–59 18.5% 14.5% 15.7%
60 or older 34.9% 24.4% 27.2%

APPENDICES
Portland
Table 19
Juvenile (Under 18) Population Data and Weighted Averages
PPD
Total Population Under 18

250,967 10,264
(2020)
GENDER
Boys 51% 50% 50.5%
Girls 49% 50% 49.5%
RACE/ETHNICITY
BIPOC 12% 34.4% 31.2%
Black or African American 2% 21.2% 18.6%
Asian/Pacific Islander 1% 3.2% 2.9%
Native American/Indigenous 1% 0.1% 0.2%
Latinx/Hispanic 3% 3.3% 3.3%
Two or More Races 5% 6.7% 6.5%
White, Not Hispanic 88% 65.6% 68.8%
AGE
Under 5 years 25% 31% 30.3%
5–9 years 28% 27% 27.3%
10–14 years 29% 27% 27.2%
15–17 years 18% 15% 15.3%

APPENDICES
Portland
Table 20
Driver (Age 15+) Population Data and Weighted Averages
PPD
Total 15+ Population (2020) 1,135,578 57,948
GENDER
Men/Boys 48.5% 48.6% 48.5%
Women/Girls 51.5% 51.4% 51.5%
RACE/ETHNICITY
BIPOC 6.4% 15.6% 11.6%
Black or African American 1.2% 6.5% 4.2%
Asian/Pacific Islander 1.2% 4.0% 2.8%
Native American/Indigenous 0.6% 0.3% 0.5%
Latinx/Hispanic 1.5% 2.6% 2.2%
Two or More Races 2.2% 2.5% 2.4%
White, Not Hispanic 93.6% 84.4% 88.4%
AGE
15–17 4.0% 2.6% 3.3%
18–24 9.5% 11.5% 10.8%
25–29 7.1% 14.5% 11.1%
30–39 13.9% 19.8% 17.1%
40–49 14.2% 13.7% 13.9%
50–59 17.7% 14.1% 15.8%
60 or older 33.5% 23.7% 28.0%

APPENDICES
Portland
Appendix B: Portland Police Department Patrol Areas (Beat Map)
PORTLAND POLICE DEPARTMENT BEAT MAP
/
en
Evergre
er
ock
ds
oo
W
rb
icke
Bro
Kn
ok
Gra
ss nce
y Pri
Maggi
ed
Re
Trefeth
en
Woodland
g
on
ray
str
n Sp
Ho
stal
Bel
Arm
pe ve Skillings
Cry
Oak de
Ocea
re
Tolm
law
5 Alic n
49 e
an
te Garsoe Sargent
rsta
r
In te
w
te
n
Town of Falmouth
n
vie
ro
gto
Car
5
49
Ca
in Beach
ng
ash Marriner
vie
Lo
te
et
rd
W Sunset
st
Cove
rsta
St rs
pboa
Farm
Cre
Spar
ela
on Ove
ec
In te
rest Ocean Ocean

Cla
Pa
vis
Ab
Lyndon
La
by
Da
Diamond
be
rick
rt
de
iet
Fre
Le
asan
ster
Qu
d
ia
ide hlan
nd
rg
Town of Falmouth
Ple
Hills Hig
l
Hil
Ha Geo
Isla
ys
Bal
dy
tis
ton
Arb
City Point
rt
st
lpar
Clay
Wen
Cur
ve
Fro
utu
k De
n
s
so e pw
ck oo Natick
Ja Ballfield Je
Roa
Hussey
r
d Park wel
nipe
Centennia
l
ring
rk rch
Ju
Cresfield it Pa e Bi
y
loch
Wile
Old
Allis
Bro
ck
e Church
mm Alpin Pine ther
W erry
leba
C
nd
Su
ok
Hea
D
on
Su
oir
l
yb
Ly
s Po
ha
Peaks Island
oo
m
ew
Ba
Reserv
nn
l ty
Ledg
Palmer
Chris
m
ta
rson
e
r
it
en th
tin ke c Elizabe
strial
ua Rusti
Mer
Pa
Rice Con
Bould
Q r
y Luthe g
Ma
a
to
d lod ac
Indu
rid
Me lin
ie
n
er Ster
rtin
Danie
n m
Flo
oo
er
Ca e
dd
Pa ph tum Su tt
Te
W She w
alc r Au
ble Herm
an Bracke
Ernest
l
ide
Fall
le
h ld Centra
l
Do fie tley m A
Rivers
ok Bar Bra Mea Uppe

r
Bro do
w
od Lower A
Eve
wo Willow
en
rg
se ue t Bo
Gre atho
re
Mill im
ro
ga
n Croq mpa use Epps
en
Ta
iken Pr Re ine
Palm
Rac
Cottage
ge
Nor
s
Auburn
Be olid ing ansa Welch

Co
er
th
ve m K
rn
Broo
yo
woo
rl bo W Alderbrook
Vete
y
e
Orch
rsid
ge San
9
Park
lark ian
d
er
k
In du olid Merid
ran Ve
y
Sk pip
Rive
stria Co
Adams
ard
e
Saugus nd
Lib
l why Spruc
Th
Sa
Milton
Mug
l
Vir
Winding
rbel
er
e
Island
Rand
da Onw
teran
nn
en
Ryefie
ty
Buca
Re
Ra
gin
Pe New
er
He
va
ard
Ne cent ay
Ta
sid
y
ck
ey
ia
Dov Maple
n
G
Cres
d
olp
Fis
rs 3rd
ne
Pon
Tu
Ly
e
er Je t
ld
nn er 3rd
nce
he
on
ss
City of Westbrook
h W
m Oak
wst ine
at
r
H
Ver 8th Ma
int
e
s
Nor
y
as bury
Cov
Gre
Bre ss
ile
Po
al
oo
ke
n
Pe
Oak
Torrington
Harm
pre
Be
to
Seashore
th
Ba
ll
lock
a
Long
nn
Cy
ew
rid
po
is
Rive
ell
N
Had
Flo
Interstate 295
n
rt
rt
arr
ony
ort ndo
ke
ne
ua
H
Lu gut
rton
Abb
sti
Be
ort
St
Ald Bra
th
Sh
Ca
a
a
in e
ure
ot
rr
de
ood Fa N ain
w
bes
ne
ru
Kn
M
ga
Waldro Pinew Fo Da
La
ert
g
Plym
ig
kota
Bro
rd
Pa
aw n
n Del rin
ha
Le
ht
w
ri
O
aw Lo
ad
Ea stla
xin
hio
n
rn
Id
ar
ou
w
a n
ping
e
ah
Fa
gto
on
e
ay
by to
th
W
stl
ad g Rid
o
id
M
tin
n
Rain Lib l
Sar
Ep
cl
te ge
mer
th ue n
es Six
Eu
mak
rt
Hu
M
W
To
m
a
er
lfo
ilk m Sa d
arl
Sum
Ho ar
ro
ie
Be
rn
all
Art
bo
th
nit gh
alth
d
ke
Fif Be
W
La
woo
ro
hu
Wild
be
Dra
l Northfi ge Gray
r
we
Leig eld Ed y stone

esle
Ca
wo
Town of Long Island

on
hton rry
rrill
Ch Ashley
Pri
rria
rita Be
d
ard
mm
od
Fiel
oo
nce
Ve
Ja Ruby
Hill
ase
ge
In ds
ew
rv
Co
Le
to
Co
ge to
ks
Ha
un
ne
m
dg
Dibi
sant
n
rs
nce
Hic
Ho
M
tr
oll Iffley Ha ew
Dor ag
y
ia
n
oo Wel
Allen
rvar
n
wre
g
ea
othy
to
Felic
Yale
alo
ie d
Nic
ot
d lsto
nce
Ph
lia
Gro
La
Good
Av
lilly
Presumpsc
ne
hol
ve
no
r
Pri
no
M
Ma
Ra
as
Woo
Mag
cA
er
True
Hemingwa
Feeney
Elea
n
ridge
ge Un dl
iver awn
ll
list
Schoon
Morri
sity Grafto
er
ors
Goodale
Co n Ebe
rnel
all
Fa
Newcomb
Maj
l Fa n
M
Provid
rm
ll Hill
10
Bro
Oly
w
ence
O
Ta Sle
y
bo
re
ld
Xivary
m ok
m
Saville
Bishop m
Trundy
fie
go
in
95
Warren arl
Fern
on
pia
Ra
n
By
an s
Whi wo
Ho
rrill
state
e
1
Mo Ver Be
lid
ord tho
ute
r
ttie rth
a
Morse
rk
de
Llo
ay
Ha adia gin
sh
s
Mu
Ro
Kid
In ter
yd ire
r
8
Arbor
w
Arc Ho
rra
de
n
lso
y
ess
Ju
Vic
Bell
est Ly
rn
Co
Pinecr to
St
rn
Sy nd
e
Colle
kle
co
r
ve
d
d
Bir ge
Polan lv a Fa
Stevens
Oa
Can
Ro Sh
rry
In
ch eo ia irfie
Fe
woo Bell se Sav erw

s
Campbell e ld
Kin W
Qua
da
rn
d Waverly in le ash oy oo
od
ald
ra
a
in d
Lo
nd
Ro ll gto
Penwo
Ha
Du
11
bin
ra
Woodva
nd
Elmwood Re n Pe
Ve
d
m
To
ad
Isla
le
bro
Kens
y
rr
Bir Tremaine Ills ke

ey
St Joseph Ba Broo Sh Le
ch y ksid Fr le nn
le
Larrab on y erw
ingt
va
y
e ox
Ba
ee Po le Gle t Jo
Dingle
Gle oo
co
e ck rd
ck
d
Su
n an
on
Can
Hav Wel ler

in
n
lin
Cov
ll
g
se
en
lv
es
da
Da k gton k
Pershin
Ca
Ga
t
yc
Bac
Bat
n
le rly
e
dlew
Ra
Great Diamond Island

Holly le
Out
an
Can
Bean Pot
Wal
m
Le ton
ell
w
s
Py
Holm
Ric ona
Ne
Gorge
st ed
Oro a
Taft
an
rd New
B
n
no h
Oce
as
Jossly
Bre ard Crosby

Ge
orge
Purch
Four Winds t ntw son

Be
M A
ble B a M lba Aust
Fore
art de ab oo Mayla
Popham am le line el d nd Ma in
ck
eside H ckwo
st
H tt
Grov Hartley
Warwi
Riggs ale Saw rth

Woo
Cabot
ond
yer
Lu
Irving
Essex
Parson
Kent
Clinton
James
x
r
t
lo
Ph s
lle
dfiel
Dorse
Wesse
Lom
e
Devon
w
tt
ip Bay
sid
Fu
te View
d
ne
ps
d
Hunt
ga
ncor
Matthews
d
lan
ay
Win Pleasant
d
Chenery
Den
on
Le
Merili
ym
on
Co
ne Concord Codman
Back Cove
ne
rm
Ra
Alde
st
Jean
Ha
n
Glenwood
Saunders
Ca
Hersey
re
Rand
Levia
th
Glen
b
lc
Lawn
er
East
pisi
Web
Berkeley
wor
Norwood
Hil
David Vannah ern

inst
Plowman
Grace
rid
c
May Prom
Rowe
ge
t
tm
Higgins enad
ge
er
mon
Ed
n
Was
H e
Wes
s arv gto n
is
re fto
Baxter
bia
Hill
se
ca H Revere Reve in
on ut
Cli
hing
e
Arl
ou al
Lu ale y d
Ro
aln
Woodfor
lum
elb tre
e
be rne
t
Terrac
t on
W
Lincoln Cove ld
ton
Montrose Rackleff Gou

Co
Pine
e
lm
orth Fellows
er
Highland
Be
Has gonn
g
Tree
rn
Ham
2
y
din
Coyle In
M
Belknap
ue
No
nd
Tu
gs
Indu
ld
dia
Anderson
Sa Fo
Q
M
Win
stria
Belfie
rt
hi
tin
Ro n rest
Starbird
orn
Beacon
mon
l
Mac
ckla Prospe Ashmont Be

err
in
La
Fre
Cutter
nd ct st ck g
ett Ves
fa
ill
d
em re
roft
ley
ye
Hollis
Longfellow Fo
hire
pe
Sh
nilw
an
ia
tt n
tt
Mot
Longfellow Atl r
nc
ere iso e an Mu
e
erid
lm ld
de
th O
on
vons
Iv Ev ad Fo
Ba
ou
Ke
tic njo bri

ie
y s ro
ea
rt
d
ye
rf
on
an
Way and m y
7
M on A
ee
x
Po
Brigh Orl No art am Fo M lle
De
ne y
D
m
Macy
Be
Fa D illi Ha ed Ke n
l
ott ero Ju
nia
irv ton
1
ne iew W t n nn llo
olc y
Cle
na
lo
Pit Ke
gg
en
Po
k
fo
Co
W Du
Sm
ev
Ban
Fra nce
De
rd
Preb
nd
ight rh al
re
e
Strou
ith
ar
M
e
Coachl nk
th
am
se
Sp in Mountfo
Fo
rin
ay
dwate
But
ou
Elizabe
iew
es
lin
le
ne arg rt
o
g
r th
F
lm
Bradley
Anson M
St
eri ld y
to
Bo
Riverv
l
Fa
Roundabout
rd Lo jo
et
reet
ra un
Caleb
ec r
nw
th
yd
cu de
Craigie
ie fo
Ca
Sw
d eb te
s
d ers M
eff oo Be st
Co d
oo
Fe
Edward
Dirigo t m enn cas

Gra
W
s
Sh nw
s
an
on
Ha
In
na
France
d
So
y
ilm
Douglas
t K an dia
Whitne
Ke odm Ha
n
n
man
nt
k
Fros
Portland Harbor
ot
Ha
L
co
Bolton
roo Pe
De
St
Rud m
Westb rith o
ck
rd a psh
yw
4
e
n W rl es
Jam
olls xfo C
rin
De
Ha
husetts
Pe
Ex
ire
Pa
e
M
oo
Kn O h m
g
dg a
yrt
ete
es
an
no
in Scott
ys
rtri l Th
es
d
Tw tn ra ry
le
rn
on
ver
Pa de
r
e bu ut
bu dle
Bood
Brat
er
nit Ex Fe
Massac
ash ew Mid
Pa
ch
ew
ra
Peaks Island
Elm
Clif W an N
tle M
G
rris
d
Br
6
f n ge
Waldo
rla
l
Ho
Lassel
Te
Pre
Riv ba W
be Cu Main
ain
Davis
est
m
ers rt
Mitton
ble
m City
n
ec
ple
Park
M
la
Cu
e
Captains
sto
sland
e
ur
Ed nd
ark
Sta
Ca
han
W
Bro
m
hb
ge
Ca
rib ha
ll
et Lo an ery Wh
te
H
Po
Sewa
Un
Ce
ou
w
sc
rf
Was
ic
Fe O ou
Grant f
Pow
an
n
Pie
ak
rt
o
nw
n
3
nd
har
ion
Sherm se
la
te
ay
r
on W
la
nd
r
Marston
d
Garris ton W
ng dle
De
Ch idg ion
oo
Mellen
Hu tch
ha
Pie
yn g
Lowell
e
rin
W
Osg
Bo
Cro
Old Mast rf
W
Fre
r
ha
Sp
ss
Deering
rf
Tid
h
ins
Un
Un
ll
rs
ort
n
Gilm
Sewa
u e Hil
Avon
Milla
m nt
Yello
W arf
sw
M
Douglas
l
Pa bb Fo
W arf
ver ill
ric asa
ha
Ell
rk
Blueberry
ld
re
Co
an
Ando Ple
York
de
Mac
rf
ie
wb
Riv
ll
A
tf
Walker
ha
Fre Pa Po
es
er Hig
ird
Dow
rk rt
W
h la
Bra
e
nd
Western Promenad
In te W
Perim
Sta Fis
County
rnat C in Ho
te te h
Carleton
Congress iona West r bso Pie

e
y
Ny
ns
Jo
eter
l Ho W r
y
ort lyok
Chadwick
Bra
n
ha
Gra
so
District
Skyway tp Lo ck De eW rf
Pine
hn
Je op D ett ak
Emery
Ty es har
Jo
Cla Ta ng W f
City Line
Thomas
rk te
5
ha
St John
lin
Vaughan
Carroll rf
m
Neal
May
Ha
rth
Valley
Westbrook
fo
er
Bri ool
n
Da
m
Sch
gg
Bowdoin rd
m
s
fo
Su
Sto
Perimeter Clif
Cas
em
Fle
re
Sal
co
r
rd ercial
tc
ha
Bay
he
Orc Comm
r
Veter
ans
nt
City of South Portland
dy Poi
Cassi
City of South Portland 0 0.4 0.8 1.6 2.4 3.2
River Miles
Fore
POLICE BEATS
Portland Police Department
Crime Analysis 1 3 5 7 9 11
December 2016
2 4 6 8 10

APPENDICES
Portland
Appendix C: Portland PD Arrests Chi-Square Results
Tables
Table 21
Portland PD Adult Population Estimate Compared to the Portland PD Arrest
Records Sample, All Years, Maine Residents Only
CHI-SQUARE RESULTS N X2 DF P-VALUE

BIPOC 6355 373.28 1 <.001
Black or African American 6355 2078.787 1 0
Latinx or Hispanic 6369 1.682 1 0.195

Asian/Pacific Islander 6355 87.657 1 <.001
Native American/Indigenous 6355 10.649 1 0
Men 6368 1849.16 1 0.001
18–24 6366 144.937 1 <.001
25–29 6366 39.793 1 <.001
30–39 6366 699.092 1 <.001
UNDER 40 6366 1088.304 1 <.001
40–49 6366 128.768 1 <.001
50–59 6366 9.499 1 0.002

60 or older 6366 1838.981 1 0
Table 22
Portland PD Juvenile Arrest Records Compared to the Juvenile Population, All
Years, Maine Residents Only
CHI-SQUARE RESULTS N X2 DF P-VALUE

Boys 165 22.813 1 <.001
BIPOC 164 23.614 1 <.001
Black or African American 164 42.528 1 <.001
Asian/Pacific Islander 164 3.055 1 0.08 Table Notes: A p-value of .05 or less indicates a
statistically significant result. A statistically signifi-
Latinx/Hispanic 165 21.159 1 <.001 cant result indicates that the observed differences
10–14 years 165 1.146 1 0.284 between the population estimate and the arrest
dataset are greater than we might expect by
15–17 years 165 351.99 1 <.001 chance alone.

APPENDICES
Portland
Appendix D: Portland Residents Adult Arrest Rates
Average annual arrest rates represent the average number of individuals in each group arrested in the city of Portland
each year out of 100 people in that group. Rates were calculated by dividing the total number of arrests by 3, then
dividing by the total population for each group and multiplying by 100 ((total arrests subgroup/3)/total population
subgroup*100).
Table 23
Portland Annual Average Adult Arrest Rates by Gender, Race, and Age (18+ Population)
Portland Population PPD Adult Arrests Sample Average Annual Rate

N % N % (per 100)
Total 18+ Population (2020) 56,442 4,514 2.7
GENDER
Male 27,383 49% 3,450 76% 4.2
Female 29,059 51% 1,063 24% 1.2
RACE/ETHNICITY
BIPOC 8,623 15% 1,047 23% 4
Black or African American 3,561 6% 870 19% 8.1
Asian/Pacific Islander 2,265 4% 48 1% 0.7
Native American/Indigenous 199 0% 7 0% 1.2
Latinx/Hispanic 1,489 3% 130 3% 2.9
White, Not Hispanic 47,819 85% 3,459 77% 2.4
AGE
18–24 6,647 12% 662 15% 3.3
25–29 8,413 15% 661 15% 2.6
30–39 11,498 20% 1,381 31% 4
40–49 7,932 14% 942 21% 4
50–59 8,190 15% 719 16% 2.9
60 or older 13,762 24% 148 3% 0.4

APPENDICES
Portland
Table 24
Portland Annual Average Adult Arrest Rates Black or African American Individuals Compared to White (18+ Population)
Black or African American Individuals White Individuals
Portland Population Arrests Annual Rate Per 100 Portland Population Arrests Annual Rate Per 100
MEN 2,004 735 12.2 22,845 2,578 3.8

18–24 401 173 14.4 2,516 283 3.7
25–29 242 120 16.5 3,714 352 3.2
30–34 223 125 18.7 2,656 386 4.8
35–44 474 209 14.7 3,635 687 6.3
45–64 540 106 6.5 6,350 823 4.3
65+ 124 2 0.5 3,974 47 0.4
WOMEN 1,557 135 2.9 24,974 880 1.2

18–24 162 39 8 2,814 123 1.5
25–29 144 35 8.1 3,523 129 1.2
30–34 250 16 2.1 2,407 148 2
35–44 462 26 1.9 3,616 236 2.2
45–64 282 18 2.1 7,221 231 1.1
65+ 257 1 0.1 5,393 13 0.1

APPENDICES
Portland
Appendix E: Portland PD Traffic Data Chi-Square Results Tables
Table 25 Table 26
Portland PD Traffic Citations Compared to the Portland Portland PD Traffic Stops (Warnings and Citations)
PD Population Estimate of Drivers, Maine Residents Compared to the Portland PD Population Estimate of
Only, All Years Drivers, All Residences, All Years
CHI-SQUARE RESULTS N X2 DF P-VALUE CHI-SQUARE RESULTS N X2 DF P-VALUE

Men 4671 216.484 1 <.001 Men/Boys 13738 953.758 1 <.001
BIPOC 4573 23.302 1 <.001 BIPOC 13537 450.401 1 <.001
Black or African American 4573 588.033 1 <.001 Black or African American 13537 3804.114 1 0
Asian/Pacific Islander 4573 15.586 1 <.001 Asian/Pacific Islander 13537 11.481 1 <.001
Native American/Indigenous 4573 19.136 1 <.001 Native American/Indigenous 13537 47.711 1 <.001
Latinx/Hispanic 4573 53.577 1 <.001 Latinx/Hispanic 13537 249.945 1 <.001
15 to 17 4664 30.082 1 <.001 15 to 17 13589 297.932 1 <.001
18 to 24 4664 66.833 1 <.001 18 to 24 13589 701.628 1 <.001
25 to 29 4664 117.247 1 <.001 25 to 29 13589 452.106 1 <.001
30 to 39 4664 181.546 1 <.001 30 to 39 13589 488.716 1 <.001
40 to 49 4664 35.468 1 <.001 40 to 49 13589 67.42 1 <.001
50 to 59 4664 3.251 1 0.071 50 to 59 13589 68.173 1 <.001

60 or older 4664 653.573 1 <.001 60 or older 13589 1983.247 1 0
Table Notes: A p-value of .05 or less indicates a statistically significant result. A statistically significant result indicates that the observed differences observed differences
between the population estimate and the traffic stops dataset are greater than we might expect by chance alone.

APPENDICES
Portland
Appendix F: Portland PD Logistic Models for Multiple
Regression Analysis
Table 27
Logistic Regression Final Model Results for the Arrested Individual Being
Black/Latinx
ARRESTED INDIVIDUAL
BEING BLACK/LATINX B S.E. WALD DF SIG. EXP(B)
Officer-Initiated Incident 0.161 0.064 6.387 1 0.011 1.174
Male 0.687 0.082 69.739 1 0 1.989
Age Under 40 0.812 0.072 127.492 1 0 2.251

Warrant -0.297 0.082 12.984 1 0 0.743
Unhoused -0.161 0.069 5.427 1 0.02 0.851
Patrol Area 2 0.411 0.127 10.509 1 0.001 1.508
Patrol Area 5 -0.512 0.118 18.703 1 0 0.599
Patrol Area 6 0.247 0.113 4.806 1 0.028 1.281
Time 9PM to 12AM 0.187 0.086 4.688 1 0.03 1.206
Time 12AM to 3AM 0.192 0.084 5.172 1 0.023 1.211
Constant -2.551 0.109 551.939 1 0 0.078
Table Notes:
This logistic regression model was the final model testing Black/Latinx as the dependent variable. Many
variables were tested (such as gender, race, age, location, time of day, call-source) but only the variables which
were found to be significant are included in the table above.
The model showed that the above variables were all significantly associated with the arrested individual being
Black or Latinx (X2(10)=296.673, p<.001). These variables explained 6.8% (R2) of the variance in the race/ethnic-
ity and correctly predicted 80.4% of the cases.
The odds of the arrestee being Black or Latinx increased when the incident was officer-initiated, when the
individual was male and under the age of 40, when the incident occurred in Patrol Areas 2 or 6, and during
the hours of 9:00 pm to 3:00 am. The odds of an individual being Black or Latinx decreased for the remaining
variables.

APPENDICES
Portland
Table 28
Multiple Logistic Regression Model Results for Multiple Charges Requested
MULTIPLE CHARGES
REQUESTED B S.E. WALD DF SIG. EXP(B)
911 Call 0.442 0.062 50.639 1 0 1.556
Unhoused 0.236 0.056 17.59 1 0 1.266
Time 3 to 6 AM 0.316 0.104 9.179 1 0.002 1.371

Age 25 to 29 0.172 0.076 5.083 1 0.024 1.188
Age 30 to 39 0.188 0.059 10.23 1 0.001 1.207
Patrol Area 6 0.257 0.099 6.751 1 0.009 1.293
Patrol Area 7 0.296 0.118 6.248 1 0.012 1.344
Patrol Area 8 0.268 0.1 7.124 1 0.008 1.307
Constant -1.146 0.05 535.088 1 0 0.318
Time 12AM to 3AM 0.192 0.084 5.172 1 0.023 1.211

Constant -2.551 0.109 551.939 1 0 0.078
Table Notes:
This logistic regression model was the final model testing Black/Latinx as the dependent variable. Many
The model showed that the above variables were all significantly associated with the arrested individual being
Black or Latinx (X2(10)=296.673, p<.001). These variables explained 6.8% (R2) of the variance in the race/ethnic-
ity and correctly predicted 80.4% of the cases.
The odds of the arrestee being Black or Latinx increased when the incident was officer-initiated, when the
individual was male and under the age of 40, when the incident occurred in Patrol Areas 2 or 6, and during
the hours of 9:00 pm to 3:00 am. The odds of an individual being Black or Latinx decreased for the remaining
variables.

APPENDICES
Portland
Table 29
Multiple Logistic Regression Model Results for Multiple Charges Requested
DRIVER IS BLACK OR
AFRICAN AMERICAN B S.E. WALD DF SIG. EXP(B)
Male 0.723 0.056 165.523 1 0 2.062
Citation Received -0.243 0.059 17.116 1 0 0.784
Officer Age -0.015 0.003 29.104 1 0 0.985

Driver Aged 18 to 24 0.467 0.059 61.891 1 0 1.595
Patrol Area 3 -0.592 0.168 12.422 1 0 0.553
Patrol Area 4 0.322 0.081 15.964 1 0 1.38
Time 9 PM to 12 AM 0.409 0.061 44.188 1 0 1.505
Time 12 AM to 3 AM 0.667 0.071 88.537 1 0 1.948
Constant -1.949 0.114 292.267 1 0 0.142
Time 12AM to 3AM 0.192 0.084 5.172 1 0.023 1.211

Constant -2.551 0.109 551.939 1 0 0.078
Table Notes:
This logistic regression model was the final model testing Multiple Charges as the dependent variable. Many
This logistic regression model showed that the above variables were all significantly associated with multiple
charges being requested (X2(8)=101.408, p<.001). These variables explained 2.1%(R2) of the variance in the
race/ethnicity and correctly predicted 68.9% of the cases.
The odds of the individual receiving multiple charges increased when the incident was a 911 call, when the in-
dividual was unhoused, when the individual was between the ages of 25 and 39, when the incident occurred in
Patrol Areas 6, 7, or 8, and when the incident occurred between 3:00 am and 6:00 am. Note race and gender
were not found to be significant in the models.

About The Project & Partners
This project was developed by the cities of Portland and South Portland and their Police
Departments in collaboration with the University of Southern Maine, Catherine Cutler Insti-
tute and the Northeastern University, Institute on Race and Justice. The project was funded
by the cities of Portland and South Portland and the Roux Institute at Northeastern Univer-
sity. The goal of this project was to examine and analyze available data to determine if there
was evidence of disproportionate police activities in either city.
The Catherine Cutler Institute is the research arm of the Muskie School of Public Service
at the University of Southern Maine. The Catherine Cutler Institute has a long history of
collaborating with local, state, and national partners to examine and find solutions to critical
societal issues. For more information about the Catherine Cutler Institute visit usm.maine.
edu/cutler.
The Institute on Race and Justice (IRJ) is based at Northeastern University’s School of Crim-
inology and Criminal Justice. The IRJ utilizes social science research methods to support
partners in the development of policy changes which advance the cause of social justice.

PPD Findings Final 121922

Uploaded by

Copyright:

Available Formats

PPD Findings Final 121922

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PPD Findings Final 121922

Uploaded by

Copyright:

Available Formats

Decembert 2022

Assessing Arrest & Traffic Stop

Jack McDevitt, PhD, Northeastern University

George Shaler, MPH, University of Southern Maine

Sarah Goan, MPP, University of Southern Maine

Stephen Abeyta, PhD Candidate, Northeastern University

Carlos Cuevas, PhD, Northeastern University

Layout and Design

Becky Wurwarg, University of Southern Maine

Assessing Arrest & Traffic Stop Patterns ii

About This Report

Summary of Key Findings

Samples & Population Estimates

Data & Analysis Limitations

Detailed Findings: Portland Police Data Analysis 9

Portland Police Department Traffic Stops, 2018-2020

Conclusion & Recommendations  29

Recommendations for Portland Police Department

Recommendations for the Portland Police Department, City of Portland, and

Appendix B: Portland Police Department Patrol Areas (Beat Map)

Appendix C: Portland PD Arrests Chi-Square Results Tables

Appendix D: Portland Residents Adult Arrest Rates

Appendix E: Portland PD Traffic Data Chi-Square Results Tables

Appendix F: Portland PD Logistic Models for Multiple Regression Analysis

Assessing Arrest & Traffic Stop Patterns iii

Assessing Arrest & Traffic Stop Patterns 1

Assessing Arrest & Traffic Stop Patterns 2

BLACK OR AFRICAN AMERICAN INDIVIDUALS EXPERIENCE DISPARITIES IN ARRESTS

THE HIGHEST NUMBER OF ARRESTS OCCURRED IN THE WEST BAYSIDE NEIGHBOR-

THE ODDS OF BLACK OR LATINX INDIVIDUALS BEING ARRESTED INCREASED WHEN

Assessing Arrest & Traffic Stop Patterns 3

PEOPLE EXPERIENCING HOMELESSNESS REPRESENTED OVER ONE-THIRD OF ALL

BLACK OR AFRICAN AMERICAN DRIVERS EXPERIENCE DISPARITIES IN TRAFFIC

Assessing Arrest & Traffic Stop Patterns 4

Data Extraction & Analysis

Assessing Arrest & Traffic Stop Patterns 5

Race Portland PD Call Source

Ethnicity Registration Violations

The Portland-South Portland Metro Area

The various data extracts resulted in multiple datasets (Table Table 1

Juveniles (Under 18) 250,967 10,264 4,248

Assessing Arrest & Traffic Stop Patterns 7

Assessing Arrest & Traffic Stop Patterns 8

Portland Police Department Arrests,

land-South Portland Metro Area.29 Only a small proportion of Missing 2 0.0%

Assessing Arrest & Traffic Stop Patterns 9

PORTLAND PD PATROL AREA NUMBER OF ARRESTS PERCENT

11 Oakdale/Back Cove/East Deering 564 7.5%

8 Riverton/Deering Center/Rosemont 553 7.3%

6 Parkside 533 7.1%

2 Upper East Bayside/Munjoy Hill 392 5.2%

10 North Deering/Back Cove 391 5.2%

7 Stroudwater/Libbytown/Nasons Corner 381 5.1%

1 Lower East Bayside/Munjoy Hill 239 3.2%

9 Riverton/North Deering 223 3.0%

Other (Peaks, Jetport, Out of Town) 42 0.6%

Assessing Arrest & Traffic Stop Patterns 10

Map of Total Number of Arrests in Portland by Census Tract, 2018-2020

Detailed Findings: Portland Police Data Analysis 9

Conclusion & Recommendations 29