We used data from the Survey of Police-Adolescent Contact Experiences (SPACE), 4,15 a cross-sectional, non-probability survey administered to a sample of 345 Black youth (~ 52% male) in Baltimore City, Maryland from August 2022 to July 2023. To be eligible for participation, youth were required to 1) live in Baltimore City, 2) self-identify as Black (including Black biracial and multiracial), and 3) be 12–21 years old. The project was co-developed with the Center for Adolescent Health (CAH) Youth Advisory Board (YAB) to ensure the inclusion of youth perspectives. Recruitment involved collaboration with 12 youth-serving organizations in Baltimore, some of which serve especially vulnerable youth populations (e.g., LGBQ youth, youth disconnected from school and/or work). Organizations helped disseminate recruitment information and assisted youth in accessing study materials, on-site or online. Surveys were conducted via Qualtrics, with youth participants receiving a $30 e-gift card. Approval was obtained from the University’s [BLINDED FOR REVIEW] Institutional Review Board (#18323). Written informed consent was obtained from participants ages 18 and older; written caregiver permission and written youth assent were obtained from participants under age 18.
Dependent Variable
Police Avoidance. Following prior research,5 we assessed young people's efforts or behavioral strategies to avoid attention from–or harassment by–police. Youth asserted how often they watched how they dressed; spoke loudly with others; felt pressure to remain calm; walked or drove; planned where to walk, drive or go to; and made sure there were in the company of only a small number of other Black youth all to avoid harassment by or attention from the police. Responses ranged from never (1) to very often (4). Items were averaged into an index (range = 1–4; alpha = 0.89).
Independent Variables
For more details on the measurement of all independent variables, see the appendix.
Youth and Family Demographics. The following five youth demographic measures were included in the present study: 1) youth age (in years), 2) youth sex (male = 1; female = 0), 3) youth LGBQ identity (i.e., lesbian/gay, bisexual, and queer, with heterosexual/straight as the reference category), 4) youth multiracial (Black/African American in conjunction with other race/ethnicity = 1), and 5) youth disconnection from school and work (youth neither working nor in school = 1). Additionally, the following 4 family demographic measures were included in the study: 1) household size, 2) biological parent relationship status (married, cohabiting, with other as reference category), 3) biological parent US citizenship status, and 4) maternal education.
Adverse Childhood Experiences (ACEs). The measurement of ACEs was adapted from the National Survey of Children’s Health16 and included 11 binary items on lifetime exposure to various stressors and adversities. Items were summed and participants were categorized into the following groups: None, One, Two or Three, and Four or More.
Environmental Safety Perceptions. We included three items asking youth how often they feel safe 1) at home, 2) at school, and 3) in their neighborhoods. Response options to all items included never (1), sometimes (2), most of the time (3), and always (4).
Negative Attitudes about Police and Delinquent Behaviors. To capture relevant youths’ attitudes and behaviors, we assessed 1) negative attitudes about police using an expanded, 14-item version adapted from prior research,17 and 2) delinquent behaviors using a 13-item index adapted from the Future of Families and Child Well-being Study.18
Police Exposures. Finally, we followed the lead of prior research15 and included five police exposures in the present study: 1) police violence news consumption (days per week), 2) known persons stopped by police, 3) in-person witnessed police stop, 4) in-person direct police stop, and 5) intrusive police stop.
Analytic Approach
We calculated descriptive statistics for the full sample (n = 345), then estimated five stepwise, multivariable ordinary least squares (OLS) regression models separately assessing associations between each grouping/domain of independent variables (i.e., youth and family demographics, ACEs, environmental safety perceptions, negative attitudes about police and delinquent behaviors, and police exposures) and police avoidance. After estimating these five models, we estimated the fully specified model with all independent variables under all domains included. Finally, we estimated associations between individual acts of officer intrusion (e.g., harsh language, racial slurs, threats of force, use of force) and police avoidance among the subsample of youth with in-person police stops (n = 184). Missing data were imputed in STATA 18.1 using mi commands, which offers several key advantages over listwise deletion.19 We employed multiple imputation with chained equations, imputing variables with less than 10% missing data, resulting in 20 multiply imputed data sets.20