Evaluating a model program for improving law enforcement officers’ perceptions of and interactions with youth in a diverse urban setting

Abstract

Negative and, at times, violent encounters between police and youth have received increasing attention in recent years, leading to calls for more targeted training of law enforcement. In Pennsylvania, legal stakeholders collaborated to create a manualized curriculum training designed to educate law enforcement on developmentally appropriate ways to interact with youth and generate conversation between police and young people (particularly youth of color) with the goals of reducing stereotyping and improving relationships between these two groups and, ultimately, reducing racial and ethnic disparities in youth arrest rates. The current study evaluated change in law enforcement perceptions from 24 trainings conducted in Philadelphia – a city populated predominantly with residents of color – which included both formal and informal interactions with young, predominantly Black, volunteers. Data from the Police-Youth Engagement Survey, completed both before and after trainings by 1,344 police recruits, indicated curriculum acceptability and recruits’ willingness to use information from the training in their future interactions with youth. Recruits’ attitudes towards youth and their perceptions of the relationship between police and young people also significantly improved. Results indicate that this training may be a promising initial intervention for improving interactions between law enforcement and youth, including youth of color.

Over the past decade, shocking examples of police misconduct in the United States have exacerbated concerns about interactions between police and the communities they serve. Research investigating factors associated with how police officers interact with community members (e.g. seriousness of any alleged wrongdoing, citizen demographics including race, gender, and age) has become particularly salient given the recent surge in media coverage depicting violent – and sometimes deadly – encounters between police officers and citizens. Aggressive policing strategies, in particular, appear to disproportionately affect youth and communities of color (e.g. Brunson & Pegram, 2018). Often described as intended to prevent crime before it occurs, aggressive or proactive policing tactics include increasing surveillance in areas with perceived higher crime rates and stopping pedestrians to perform searches for weapons or other illicit materials (National Academies of Sciences, Engineering, and Medicine, 2018). These strategies are frequently applied within marginalized communities and can lead to increased reports of stereotyping, harassment, and verbal and physical mistreatment among residents of color (e.g. Brunson & Miller, 2006). Further, youth are disproportionately exposed to surveillance and involuntary contact with police, especially in schools and public spaces where they commonly spend time (Jackson, 2021; Nordberg et al., 2016). As such, young Black men and boys are more likely to experience aggressive proactive policing strategies and to be killed by police than individuals from other age and racial/ethnic groups (Edwards et al., 2019).

To enhance skills and improve interactions between police and the communities they serve, officers sometimes receive supplemental trainings on specialized topics and populations (e.g. child abuse, mental health); however, outcomes regarding the effectiveness of existing trainings are limited, and best practices for the content, attendees, and structure of these trainings have yet to be established (Chappell, 2008). Additionally, few studies have explored the benefits of including, in trainings, members of the unique populations for which officers are receiving instruction. This paper will briefly describe how prior research about interactions between youth and law enforcement demonstrates a need for training curricula specifically designed to enhance these relationships and will describe a specific training program designed to accomplish that goal by bringing law enforcement and youth together: the Pennsylvania (PA) Disproportionate Minority Contact (DMC) Youth-Law Enforcement Curriculum (hereinafter ‘PA DMC Curriculum’). We will then discuss outcomes associated with the curriculum’s implementation in Philadelphia, which involved law enforcement recruits and youth volunteers, including an overview of how recruits perceived the program’s acceptability and a summary of important changes in police officers’ reported perceptions of youth after participating in the training.

Youth and police contact

Research has demonstrated that youth and young adults (i.e. individuals between the ages of 10 and 24; Centers of Disease Control, 2014) are significantly more likely to interact with police than older adults (Brown et al., 2009; Eith & Durose, 2011). Moreover, although young people between the age of 16 and 19 make up only 8% of the U.S. population, they constitute 30% of all use-of-force interactions by police (Strategies for Youth, 2013). As a result of several developmental characteristics specific to youth (Steinberg, 2009; Sweeten et al., 2013; Weisburd & Majmundar, 2018), including differences in brain and psychosocial development between youth and adults (e.g. peer influence, reward sensitivity, impulsivity), young people may be more likely to engage in behaviors that would bring them into contact with police (e.g. fighting, drinking, speeding) and more likely to act in ways that could heighten their risk of arrest (e.g. behaving in ways perceived as disrespectful, non-compliant, or resistant) when they do come into contact with police (Morrow et al., 2018; Steinberg, 2009). These observed differences between youth and adults underlie well-known criminological theories, such as developmental theory, which highlights the fact that most young people engage in misbehavior during adolescence but ultimately desist during young adulthood (Moffitt, 1993), and differential association theory, which suggests that youth learn from their peers and, therefore, having close friends who engage in offending behavior increases the likelihood that a youth will engage in such behavior as well (Church et al., 2009; Sutherland, 1947).

Importantly, the risk of police contact and arrest for youth engaging in normative adolescent risky behavior is particularly elevated for youth of color in economically disadvantaged neighborhoods, where police tend to be present more often (Boehme et al., 2022; Gaston & Brunson, 2020). As such, Black youth face heightened surveillance by police and, despite engaging in similar rates of misbehavior as youth from other racial/ethnic groups (Bishop & Leiber, 2011; Leiber & Fox, 2005), experience disproportionately higher rates of arrest (e.g. 2.4 times that of White peers; OJJDP Statistical Briefing Book, 2020). Adding to concerns, youth of color are especially likely to be the victims of police mistreatment and excessive use of force (Edwards et al., 2019; National Juvenile Justice Network, 2017). For example, Black individuals below the age of 25 are 4.5 times more likely to be killed by law enforcement than individuals of other races and ages (Males, 2014). Such disparities further underscore the necessity of understanding and improving officer attitudes towards young people in communities of color.

Although research on law enforcement attitudes toward youth is sparse (Development Services Group, Inc., 2018), studies suggest that police officers may view the same behavior – including being out late at night and behaving in a reportedly disrespectful manner (Allen, 2005) – as more serious and indicative of illicit activity when conducted by a young person than by an adult, especially for youth from economically distressed communities (Brown et al., 2009). This difference might reflect police officers’ perceptions of youth as predisposed to engage in misbehavior and, thus, requiring additional surveillance and intervention (e.g. Richards et al., 2019). Furthermore, prior research has demonstrated distorted and racially biased perceptions of youth of color among law enforcement. For instance, police officers overestimated the ages of Black suspects compared to White suspects by an average of 4.6 years, contributing to the perceptions that Black youth are more responsible for their actions and, therefore, are more appropriate targets for violence and deserving of punishment by the police (Goff et al., 2014; Graham & Lowery, 2004). Thus, considerations of stereotyping and bias are integral to police trainings focused on improving interactions between youth and police.

Police legitimacy and law enforcement training

The many examples of tense and sometimes fatal interactions between police and communities over time have likely contributed to citizens’ poorer perceptions of police and their legitimacy as authority figures – particularly in marginalized neighborhoods and those with high crime rates, where policing tends to prioritize aggressive order and maintenance (Fagan & Davies, 2000). For example, in Philadelphia, a city comprised predominantly of residents of color,¹ which also consistently ranks above the national crime average, young people have expressed acute distrust of law enforcement, based primarily on their lived experiences of negative interactions (e.g. Carr et al., 2007). In one study, for example, among 147 young people interviewed, 62% expressed a negative disposition toward police and less than 10% of interviewed youth indicated they would call the police in any circumstance (Carr et al., 2007).

To improve relationships between police and citizens and otherwise enhance skills, law enforcement officers sometimes receive supplemental training on specialized topics and populations (e.g. substance use, mental health crises, LGBTQ community members). Such trainings have demonstrated success in increasing officers’ knowledge about training topics (e.g. Ellis, 2014; Israel et al., 2014; Wagner et al., 2016); however, changes in behavioral intentions (e.g. attitudes toward specific behaviors), are also critical precursors to behavior change (Kim & Hunter, 1993). Per the theory of planned behavior, behavioral intentions serve as the best predictor of future conduct (Ajzen, 1991). Furthermore, attitudes and intentions – including self-reported predictions of behavior – have been significantly linked to future performance (Armitage & Conner, 2001; Steinmetz et al., 2016). Therefore, effective training programs for police should highlight the specific actions expected from officers when interacting with community members (Demir et al., 2009) by creating curricula that actively engage police and encourage participation from all attendees to promote changing intentions (Birzer & Tannehill, 2001; Knaak et al., 2019). Additionally, intergroup contact theory posits that positive encounters with diverse individuals can help reduce racial biases and stereotypes (Fridell, 2016; Spencer et al., 2016). As a result, incorporating discussions between police officers and other groups (e.g. young people, youth of color) into police training can offer an opportunity for meaningful conversations that may reduce biases.

Notably, law enforcement officers have expressed a need for additional training focused on how best to interact with youth (Schulenberg & Warren, 2009), as most police academies spend only 1% of recruits’ training hours on juvenile and youth justice issues (Strategies for Youth, 2013). Existing training programs focused on improving officer and youth interactions have aimed to contextualize youth behavior as related to their neurodevelopmental capabilities and unique environmental stressors, teach officers how to appropriately de-escalate emotionally heated interactions with youth, and increase awareness of the role of bias and disproportionate legal system contact for youth of color (e.g. Aalsma et al., 2018; LaMotte et al., 2010). Evaluations of these programs have observed increased knowledge of youth behavior and developmentally informed practices (e.g. de-escalation), improved attitudes toward youth, and reduced agreement with punishment as the primary goal of the juvenile justice system (Aalsma et al., 2018; LaMotte et al., 2010). However, few trainings have included youth participation as part of their programming, and most prior evaluations of such trainings were conducted with small sample sizes (e.g. Broaddus et al., 2013).

Given existing gaps in research related to police training with youth participation, there is a need for larger-scale evaluation to determine whether empirical support exists for these programs. The present paper aims to address this need through its initial evaluation of the Pennsylvania DMC Youth-Law Enforcement Curriculum in Philadelphia – a manualized training program designed to improve attitudes, perceptions, and understanding between law enforcement officers and young people, with the ultimate goal of behavior change facilitating more amicable interactions between these two groups and reduced racial and ethnic disparities in interaction quality and arrest rates.

Method

Participants

Data were collected from a total of 1,419 law enforcement recruits who attended one of 24 PA DMC Curriculum trainings that took place in Philadelphia between June 2015 and January 2020. Recruits were encouraged to complete surveys both before and after the training. Pre-training, 1,344 recruits (94.7% of the total sample) completed the Police-Youth Engagement Survey, which focuses on beliefs about police-youth interactions and relationships. Additionally, post-training, 1,231 recruits (86.8% of the total sample) completed the Police-Youth Engagement Survey once more and responded to additional items about training acceptability.

As a result, analyses regarding training acceptability included 1,231 recruits, of whom 77.2% were male (21.2% female, 0.2% another gender, 1.5% did not report), and mean age was 27.2 years (SD = 4.8, range = 19–63). Regarding race/ethnicity, 55.9% identified as White, 18.0% as Black, 14.4% as Hispanic/Latine, 8.9% as another race/ethnicity or more than one race/ethnicity, and 2.8% did not report.

Analyses regarding changes in officer responses to items on the Police-Youth Engagement Survey from pre- to post-training involved imputation of missing data in the context of exploratory factor analysis (described below); therefore, all 1,344 recruits who completed pre-training surveys were included in these analyses. This subsample was 78.1% male (20.9% female, 0.2% another gender, 0.9% did not report); 55.7% identified as White, 18.6% as Black, 15.0% as Hispanic/Latine, 8.3% as another race/ethnicity or more than one race/ethnicity, and 2.3% did not report their race. Mean age was 27.1 years (SD = 4.8, range = 19–63).

Procedure

Informed by research highlighting the need to both increase knowledge and change attitudes to influence behavior change, multiple Philadelphia agencies (e.g. police, district attorney’s office, public defenders’ office), experienced training developers, and individuals with lived justice system experience collaborated to create the PA DMC Curriculum, as well as the Police-Youth Engagement Survey to assess the training’s impact on recruit beliefs and attitudes toward police interactions with youth. Module I of the PA DMC Curriculum training involves a facilitated panel discussion in which officers and young people, as both panelists and participating audience members, speak openly about their perceptions of the other group and propose ideas for improving relations. Audience members both answer questions and participate in role-play activities centered around common community interactions between law enforcement and youth. Module II involves facilitated small-group discussions between officers and young people, allowing for continued dialogue and the opportunity to jointly brainstorm suggestions for enhancing community interactions. By encouraging sharing of personal experiences and invoking productive dialogue in a non-threatening environment, the first two modules of the PA DMC Curriculum attempt to create space for more positive experiences between law enforcement and young people.

Module III teaches officers about normative adolescent development and youth culture, including the environmental, physical, and biological underpinnings specific to adolescent behavior and the impacts of trauma on decision making. This module provides officers with potential explanations for adolescent misbehavior to promote creative problem solving in lieu of arrest when intervening with youth, with encouragement to de-escalate situations and engage in more positive interactions. At the same time, youth and young adult attendees at the training learn strategies for staying safe during police officer stops and how to report police misconduct; they also are given the opportunity to ask any questions they have about police procedures. At the conclusion of this half-day training,² the officers and young people enjoy lunch together.

Across the 24 PA DMC Curriculum trainings implemented in Philadelphia during the study time period, 436 youth volunteers attended (between 5 and 44 young people at each training session). These volunteers were predominantly male and were 17 years old, on average. Youth predominantly identified as Black or African American. Most youth reported having had prior interactions with police in some capacity. Youth and young adult training volunteers were identified and contacted via community partners (e.g. schools, community-based service providers) that commonly interact with legally involved youth. Young-person volunteers agreed to attend the trainings after learning about the program and, as appropriate, after having an opportunity to consult with their guardian about participating. Recognizing that these young people may have lived or vicarious experiences of trauma involving law enforcement that could be triggered by additional interactions, training facilitators informed participating youth that they could leave the session temporarily or permanently at any point and, during the training, they could sit with a support person or simply attend without actively participating. Youth volunteers were not paid but were provided with a free lunch at the conclusion of the training.

At the beginning of a PA DMC Curriculum training session, each participating police recruit received a survey packet, which included a pre-training survey (i.e. the Police-Youth Engagement Survey) and a post-training survey (i.e. the Police-Youth Engagement Survey and additional questions about training acceptability) marked with a unique identifying number. Survey participation was anonymous, as no identifying information was collected, and voluntary (though encouraged). Recruits completed the pre-training survey immediately before the session began and completed the post-training survey immediately following the half-day training session.

Measures

The Police-Youth Engagement Survey – created specifically for this project – aimed to measure changes in participants’ attitudes, beliefs, and behavioral intentions related to perceptions of and interactions with youth. Initial versions of the Police-Youth Engagement Survey were based on the manualized curriculum and were revised through discussions with and written feedback from the multi-agency PA DMC Curriculum leadership committee. The final version of the survey consisted of 18 items which elicited information about respondents’ agreement with statements regarding attitudes and beliefs about youth-police interactions, such as ‘Youth are often afraid when they interact with police’, and ‘Police should show respect when approaching youth’. Response options for these survey items ranged from 1 (‘Strongly disagree’) to 5 (‘Strongly agree’). The survey also asked recruits about their behavioral intentions when interacting with youth, such as whether they would ‘Explain to the youth why you stopped him or her’, and ‘Approach the youth differently than you would an adult’; response options on these items ranged from 1 (‘Definitely not’) to 5 (‘Definitely’).

Post-training surveys included the 18-item Police-Youth Engagement Survey plus 14 additional items asking respondents to indicate their level of agreement with statements about the utility of the training program itself, such as ‘I will use the information I learned today in my work’, and ‘Today’s program should be shared with other police departments across the country’ on a scale of 1 (‘Strongly disagree’) to 5 (‘Strongly agree’). Finally, the post-training survey included two open-ended questions about the utility of the training and suggestions for future trainings, for a total of 34 questions in the post-training survey.

Method of analysis

Analyses were conducted in R (versions 4.0.1 and 4.2.2; R Core Team, 2020, 2022). Training acceptability was assessed using descriptive statistics for the 14 relevant items from the post-training survey and qualitative analysis of two open-ended responses.

We examined training-related changes in recruits’ beliefs about police-youth interactions using the 18-item Police-Youth Engagement Survey included at both pre- and post-training time points. To identify sources of common variance in these 18 items and reduce dimensionality for subsequent analyses, we conducted an exploratory factor analysis (EFA) with parallel analysis using the ‘fa.parallel’ function of the psych package (version 2.3.3; Revelle, 2023).³ After determining the optimal number of factors using parallel analysis, we estimated loadings for each item on each of the factors using the ‘fa’ function in psych. We also applied an oblimin rotation, allowing the resulting factors to correlate with one another. In the interest of parsimony, we removed factors that had fewer than three items with loadings ≥ .30, as they explained insufficient variability in survey items; we then reran the analysis with the remaining factors. We examined the pattern matrix of item loadings for the final factor model and assigned descriptive factor labels based on the content of the highest-loading items. We extracted pre- and post-training factor scores for each recruit and addressed missing items using mean imputation.

We then conducted separate multi-level models (MLMs) to analyze changes in each factor of the Police-Youth Engagement Survey from pre- to post-training, using the ‘lmer’ function of the lme4 package (version 1.1–32; Bates et al., 2015). Given the training program’s focus on improving interactions between police and youth, especially youth of color, we also ran the same models for a single item of particular interest (i.e. ‘Race/ethnicity affects interactions between police and young people locally’), using z-scores to facilitate interpretability. Importantly, cohorts of police recruits attended the trainings together; thus, individuals at one training date were likely to respond more similarly to others attending the same training than those who attended a different training.

Estimates of the effect of training (pre/post) were therefore adjusted to account for within-recruit and within-cohort clustering. Specifically, pre- and post-training scores (Level 1) were nested within recruits (Level 2), who were, themselves, nested within training date (Level 3).

In each model, factor score (or the score on the single race/ethnicity item) served as the outcome variable, and the main predictor of interest was time (fixed effect; 0 = pre-training; 1 = post-training). We added random intercepts for recruit identification number and training date to account for mean differences in responses that were unrelated to training effects and included a random slope to allow time effects to vary by training date (i.e. accounting for variability in efficacy by the specific training attended).

We interpreted fixed-effects coefficients for time as the change in score (in standard-deviation units) from pre- to post-training. As the MLM corollary of Cohen’s d, we interpreted the effect sizes for these estimates as follows: .2, small; .5, medium, .8, large (Cohen, 1988). Initial models for each factor (i.e. Model A) included time as the only fixed effect. For Model B, we controlled for recruit gender (male, female/another gender), race (White, Black, Hispanic/Latine, another race/ethnicity or more than one race/ethnicity), and age (z-scored) to determine whether any observed time effects would be attenuated and whether these demographic characteristics demonstrated main effects of their own on responses. For Model C, we incorporated interactions between time and each demographic variable to assess whether the training demonstrated differential effectiveness based on recruit characteristics. Across all analyses, p-values for fixed effects were adjusted using the false discovery rate correction to reduce the likelihood of Type I error while maintaining adequate power.

Finally, to supplement quantitative analyses, we examined responses to two open-ended post-training survey questions asking for brief feedback about the training, including what was helpful about the training and suggestions for training improvement. To identify repeated themes, the first author reviewed all responses to open-ended questions and grouped them based on commonalities. The second author then independently reviewed the grouped responses and generated a list of themes. Finally, the two authors conferred to generate consensus about the final list of themes and how best to apply them to each response. Themes observed in at least 7% of open-ended responses were presented as part of study results.

Results

Training acceptability

Post-training survey responses indicated that more than three-quarters of recruits (79%) reported they would use the information they learned during the training in their future work and 75% of respondents said they would share the information they learned during the training with their fellow officers. Table 1 displays all descriptive results related to training acceptability. Additionally, 69% of recruits agreed or strongly agreed that the training helped them understand how adolescent development affects young people’s behavior, 66% indicated they learned new ways to interact with youth during the training, and 66% reported feeling more prepared to interact with youth after the training session. Moreover, most recruits agreed or strongly agreed that the program should be shared with all young people in the community (79%) and with other police departments across the country (78%). More than 64% of respondents reported they were able to share, during the training, their thoughts and feelings about how young people and police interact, and 75% indicated they would like to participate in more positive activities with youth in the community. Most (62%) attendees who completed post-training surveys rated the overall training session as ‘very’ or ‘extremely’ helpful. Regarding the specific components of the training, recruits most often rated the small group discussions as ‘very’ or ‘extremely’ helpful (63%), and each training component received endorsements from more than 50% of respondents, reporting it was ‘very’ or ‘extremely’ helpful.

Table 1.

Training acceptability results from post-training survey (n = 1,231).

	Strongly disagree (%)	Disagree (%)	Neutral (%)	Agree (%)	Strongly agree (%)
I will use the information I learned today in my work	0.3	0.7	7.4	33.8	45.1
Today’s workshop helped me understand how adolescent development affects youth behavior	0.8	2.8	14.9	33.0	35.5
I feel more prepared to interact with youth	0.9	2.3	17.2	34.7	31.7
Today, I learned new ways to interact with youth	0.9	3.5	16.6	31.3	34.9
I will share information I learned today with my fellow officers	0.6	0.9	10.8	34.9	39.9
Today’s program should be shared with other police departments across the country	0.5	0.7	7.5	26.6	51.8
Today’s program should be shared with all youth in the community	0.4	0.2	7.4	25.4	53.5
I would like to participate in more positive activities with youth in the community	0.6	0.6	11.0	32.3	42.5
I was able to share my thoughts and feelings about how young people and police interact	1.0	0.9	13.4	33.5	30.9
	Not at all	Slightly	Moderately	Very	Extremely
How helpful was each of the following in today’s trainings:
Adolescent development module	1.9	5.6	20.9	31.7	19.1
Panel discussion	1.1	3.2	14.8	32.6	27.9
Role plays	2.1	4.4	15.8	30.0	26.9
Small group discussions	1.3	3.6	11.5	26.1	37.1
Overall workshop	0.8	2.9	14.1	31.8	29.8

Of the recruits (n = 550) who responded to an open-ended question asking what they liked most or found most helpful about the training, responses most commonly included hearing the youths’ perspective (22%). More specifically, some recruits mentioned the small group discussions (11%) and/or role plays (9%) as most helpful. Other recruits noted that interacting with youth (11%), having discussions with youth (10%), and hearing others’ viewpoints (5%) represented the curriculum components they liked most or found most helpful. When asked for suggestions about improving the training, the most common response from recruits (n = 384) was to make it longer (13%), with some recruits (7%) specifically suggesting more time for small group discussions. Additionally, some recruits (7%) recommended including more young people at the training.

Changes in beliefs, attitudes, and intended behaviors

As shown in Table 2, average responses to nearly all pre-training Police-Youth Engagement Survey items (including the race/ethnicity item) were greater than neutral (i.e. M > 3.0). Therefore, respondents appeared to generally agree with statements about the importance of having positive interactions with youth, utilizing behaviorally appropriate strategies for interacting with youth, and attending to social dynamics between police and youth.

Table 2.

Descriptive statistics of pre- and post-training police-youth engagement survey items.

	Pre-training survey					Post-training survey
Survey item	n	Mean	SD	Median	Mode	n	Mean	SD	Median	Mode
Relationships between police and young people can be improved	1344	4.21	0.69	4	4	1219	4.47	0.65	5	5
Having positive interactions with young people in the community is an important part of my job	1342	4.66	0.54	5	5	1212	4.64	0.59	5	5
Race/ethnicity affects interactions between police and young people locally	1337	3.31	1.08	3	4	1211	3.50	1.02	4	4
Police have a role to play in keeping young people out of the justice system	1337	3.75	0.92	4	4	1204	3.99	0.86	4	4
Young people often overreact when dealing with police	1340	3.73	0.81	4	4	1216	3.60	0.82	4	4
Police often overreact when dealing with young people	1333	2.65	0.84	3	3	1208	3.07	0.89	3	3
Young people are more likely to take risks with a group of friends than when alone	1221	4.55	0.63	5	5	1103	4.48	0.68	5	5
Young people are often afraid when they interact with police	1339	3.55	0.83	4	4	1212	3.84	0.80	4	4
I have some control over whether young people show me respect	1325	3.72	0.87	4	4	1213	3.94	0.85	4	4
Police should show respect when interacting with young people	1222	4.48	0.66	5	5	1095	4.56	0.62	5	5
There are techniques I can use to make stopping young people safer	1329	3.79	0.72	4	4	1204	4.15	0.71	4	4
I trust many of the young people that live in my district	1339	3.03	0.84	3	3	1204	3.33	0.87	3	3
Explain to the young person why you stopped him or her	1342	4.72	0.55	5	5	1212	4.73	0.55	5	5
Approach the young person differently than you would an adult	1340	3.75	1.09	4	4	1204	3.95	1.03	4	5
Try to talk to one young person at a time away from the group	1342	3.96	0.85	4	4	1208	4.33	0.77	4	5
Work to remain calm when young people get upseta	1336	4.57	0.61	5	5	1203	4.64	0.58	5	5
Try to keep another officer calm if you see him or her getting worked upa	1342	4.76	0.50	5	5	1211	4.74	0.54	5	5
Say hello and talk to young people in your district even when not responding to a call	1223	4.61	0.61	5	5	1099	4.68	0.55	5	5

Note: Scales ranged from 1–5 (i.e. ‘strongly disagree’ to ‘strongly agree’ or ‘definitely not’ to ‘definitely’).

^aFor every item other than the two with this superscript, recruit responses ranged from 1 to 5 on both the pre- and post-training survey. For the two items with this superscript, recruit responses ranged from 2 to 5 on the pre-training survey only, whereas recruit responses to these items on the post-training surveys ranged from 1 to 5.

Exploratory factor analysis (EFA)

To better investigate changes in Police-Youth Engagement Survey responses from pre- to post-training, we first sought to identify the underlying factor structure of the survey items. Parallel analysis suggested the presence of five factors; however, two of those factors showed appreciable loadings (≥.30) on only two items and, therefore, were rejected in the interest of parsimony. We then repeated the EFA, extracting three factors. Loadings for this final factor model are reported in Table 3. Based on the content of items showing the strongest loadings on each respective factor, the factors were labeled (1) Calm and Respectful Intentions (e.g. ‘Try to keep another officer calm if you see him or her getting worked up’), (2) Police Responsibility for Positive Youth Interactions (e.g. ‘Having positive interactions with young people in the community is an important part of my job’), and (3) Sensitivity to Youth Perspective (e.g. ‘Police often overreact when dealing with young people’; this factor included the item related to race/ethnicity that was also analyzed separately).

Table 3.

Final police-youth engagement survey exploratory factor analysis pattern matrix.

Survey Item	Factor 1: Calm and Respectful Intentions	Factor 2: Police Responsibility for Positive Youth Interactions	Factor 3: Sensitivity to Youth Perspective
Try to keep another officer calm if you see him or her getting worked up	.73	.05	−.08
Work to remain calm when young people get upset	.71	.02	.04
Say hello and talk to young people in your district even when not responding to a call	.62	.12	−.05
Explain to the young person why you stopped him or her	.56	.06	−.07
Try to talk to one young person at a time away from the group	.50	−.09	.23
Approach the young person differently than you would an adult	.40	−.17	.31
Young people are more likely to take risks with a group of friends than when alone	.25	.09	.06
Having positive interactions with young people in the community is an important part of my job	.14	.61	−.11
Police should show respect when interacting with young people	.11	.59	−.11
Police have a role to play in keeping young people out of the justice system	−.05	.53	.22
Relationships between police and young people can be improved	.04	.49	.13
There are techniques I can use to make stopping young people safer	.06	.49	.25
I have some control over whether young people show me respect	−.01	.38	.09
I trust many of the young people that live in my district	−.04	.26	.14
Police often overreact when dealing with young people	−.04	.08	.62
Young people are often afraid when they interact with police	.07	.18	.42
Race/ethnicity affects interactions between police and young people locally	.01	.07	.41
Young people often overreact when dealing with police	.09	−.10	.18

Note: Loadings ≥ .30 are bolded. Factor scores were extracted using the pattern matrix shown in this table, allowing for representation of any cross-loadings and without dropping low-loading items to compute the final scores.

Multi-level models (MLMs) of factor 1: calm and respectful intentions

Table 4 displays fixed-effects results for all models related to Factor 1. In Model A, recruit responses to items within the Calm and Respectful Intentions factor significantly increased by 0.13 standard deviations from pre- to post-training (p = .005) when adjusting for recruit and cohort, a small effect. This effect remained constant in Model B, which controlled for recruits’ age, gender, and race/ethnicity (b = 0.13, p = .009). We also observed significant main effects for gender, with recruits identifying as women/other gender demonstrating higher mean levels of Calm and Respectful Intentions items across time than those identifying as men (b = 0.33, p < .001; small-to-medium effect). Additionally, we observed a significant effect of recruit race/ethnicity, as Black and Hispanic/Latine recruits demonstrated higher mean levels of Calm and Respectful Intentions items over time relative to White recruits (bs = 0.21 and 0.17, ps = .004 and .028, respectively; both small effects). In Model C, time did not interact with any demographic factor (bs ≤ .03, ps ≥ .328), suggesting the pre/post effect on recruits’ reported Calm and Respectful Intentions did not differ by their demographic characteristics.

Table 4.

Fixed effects from multi-level models concerning factor 1: calm and respectful intentions.

	Model 1a		Model 1b		Model 1c
Variable	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)
Time	0.13 [0.06, 0.21]	.005	0.13 [0.05, 0.21]	.009	0.13 [0.04, 0.22]	.020
Age	–	–	0.02 [−0.02, 0.07]	.458	0.01 [−0.05, 0.06]	.956
Non-Male (vs. Male)	–	–	0.33 [0.22, 0.44]	<.001	0.33 [0.20, 0.45]	<.001
Black (vs. White)	–	–	0.21 [0.08, 0.33]	.004	0.20 [0.06, 0.34]	.014
Hispanic (vs. White)	–	–	0.17 [0.04, 0.30]	.028	0.18 [0.03, 0.33]	.038
Other Race (vs. White)	–	–	−0.07 [−0.24, 0.10]	.632	−0.05 [−0.25, 0.14]	.817
Time*Age	–	–	–	–	0.03 [−0.02, 0.09]	.328
Time*Non-Male (vs. Male)	–	–	–	–	0.01 [−0.11, 0.14]	.966
Time*Black (vs. White)	–	–	–	–	0.01 [−0.13, 0.14]	.968
Time*Hispanic (vs. White)	–	–	–	–	−0.03 [−0.18, 0.12]	.911
Time*Other Race (vs. White)	–	–	–	–	−0.03 [−0.22, 0.16]	.956

Note: CI, confidence interval; p(FDR), adjusted p-value using false discovery rate. Each model also contained random intercepts for recruit and cohort and random slopes for time within cohort.

MLMs of factor 2: police responsibility for positive youth interactions

Table 5 displays fixed-effects results for all models related to Factor 2. Results of Model A revealed a significant pre- to post-training increase of 0.31 standard deviations (p < .001) on this factor, when accounting for the random effects. The effect size was small-to-medium, and the 95% confidence interval did not overlap with that for Factor 1 (Calm and Respectful Intentions), suggesting a significantly larger effect for Police Responsibility for Positive Youth Interactions. Adding demographic covariates (Model B) did not diminish the training effect (b = 0.32, p < .001). Further, both Black and Hispanic/Latine recruits demonstrated higher scores on items within the Police Responsibility for Positive Youth Interactions factor than their White counterparts (bs = 0.37 and 0.21, ps = <.001 and .005, small-to-medium and small effects, respectively). In Model C, no demographic variable moderated the effect of time (bs ≤ 0.07, ps ≥ .480), suggesting the pre/post effect on recruits’ reported agreement with Police Responsibility for Positive Youth Interactions items did not vary with their demographic characteristics.

Table 5.

Fixed effects from multi-level models concerning factor 2: police responsibility for positive youth interactions.

	Model 2a		Model 2b		Model 2c
Variable	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)
Time	0.31 [0.23, 0.40]	<.001	0.32 [0.23, 0.40]	<.001	0.33 [0.24, 0.43]	<.001
Age	–	–	0.00 [−0.05, 0.05]	.997	0.00 [−0.05, 0.05]	.997
Non-Male (vs. Male)	–	–	0.09 [−0.02, 0.20]	.222	0.08 [−0.04, 0.21]	.328
Black (vs. White)	–	–	0.37 [0.25, 0.49]	<.001	0.38 [0.24, 0.52]	<.001
Hispanic (vs. White)	–	–	0.21 [0.08, 0.35]	.005	0.25 [0.10, 0.40]	.004
Other Race (vs. White)	–	–	−0.02 [−0.19, 0.15]	.966	−0.01 [−0.20, 0.18]	.979
Time*Age	–	–	–	–	0.00 [−0.05, 0.05]	.997
Time*Non-Male (vs. Male)	–	–	–	–	0.01 [−0.11, 0.13]	.968
Time*Black (vs. White)	–	–	–	–	−0.02 [−0.15, 0.11]	.934
Time*Hispanic (vs. White)	–	–	–	–	−0.07 [−0.21, 0.07]	.480
Time*Other Race (vs. White)	–	–	–	–	−0.01 [−0.19, 0.17]	.968

Note: CI, confidence interval; p(FDR), adjusted p-value using false discovery rate. Each model also contained random intercepts for recruit and cohort and random slopes for time within cohort.

MLMs of factor 3: sensitivity to youth perspective

Table 6 displays fixed-effects results for all models related to Factor 3. Once adjusting for clustering by recruit and cohort, we observed a significant increase in recruits’ agreement with Sensitivity to Youth Perspective items from pre- to post-training in Model A (b = 0.60, p < .001). The effect size for this finding was medium-to-large, and based on the non-overlapping confidence intervals, responses to items within the Sensitivity to Youth Perspective factor showed a significantly greater pre- to post-intervention change than we observed for the items within either of the other factors. The significant effect remained once controlling for demographic factors in Model B (b = 0.60, p < .001). We also observed significant differences across time by racial/ethnic identity, with both Black and Hispanic/Latine recruits reporting greater agreement with Sensitivity to Youth Perspective items than White recruits (bs = 0.49 and 0.15, ps = <.001 and .047, medium and small effect sizes, respectively). In Model C, no demographic factors significantly moderated the effect of time (bs ≤ .20, ps ≥ .083), suggesting the pre/post effect on recruits’ reported agreement with Sensitivity to Youth Perspective items did not differ by their demographic characteristics.

Table 6.

Fixed effects from multi-level models concerning factor 3: sensitivity to youth perspective.

	Model 3a		Model 3b		Model 3c
Variable	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)
Time	0.60 [0.53, 0.67]	<.001	0.60 [0.53, 0.67]	<.001	0.63 [0.55, 0.72]	<.001
Age	–	–	0.03 [−0.02, 0.07]	.349	0.04 [−0.01, 0.09]	.194
Non-Male (vs. Male)	–	–	−0.11 [−0.21, 0.00]	.101	−0.12 [−0.25, 0.00]	.105
Black (vs. White)	–	–	0.49 [0.38, 0.61]	<.001	0.54 [0.40, 0.67]	<.001
Hispanic (vs. White)	–	–	0.15 [0.02, 0.27]	.047	0.15 [0.01, 0.30]	.080
Other Race (vs. White)	–	–	0.16 [0.00, 0.32]	.101	0.26 [0.07, 0.45]	.017
Time*Age	–	–	–	–	−0.03 [−0.08, 0.02]	.430
Time*Non-Male (vs. Male)	–	–	–	–	0.03 [−0.10, 0.16]	.866
Time*Black (vs. White)	–	–	–	–	−0.09 [−0.23, 0.05]	.349
Time*Hispanic (vs. White)	–	–	–	–	−0.02 [−0.17, 0.13]	.966
Time*Other Race (vs. White)	–	–	–	–	−0.20 [−0.40, −0.01]	.083

Note: CI, confidence interval; p(FDR), adjusted p-value using false discovery rate. Each model also contained random intercepts for recruit and cohort and random slopes for time within cohort.

MLMs of race/ethnicity item

Table 7 displays fixed-effects results for all models related to the survey item of particular interest (i.e. ‘Race/ethnicity affects interactions between police and young people locally’). Of note, Model A demonstrated singularity due to extremely small variances for the random intercept and slope for cohort (variances ≤ .003). Therefore, these random effects were deleted, resolving the singularity issue; random intercepts for recruits were retained. In Model A, we observed a significant change in responses from pre- to post-training after adjusting for clustering within recruit and cohort (b = 0.17, p < .001); this effect was considered small and of comparable magnitude with the observed effect for Factor 1. In Model B, the effect held even when controlling for age, gender, and race (b = 0.17, p < .001). Significant main effects also were observed for age and some racial groups, as age of recruit demonstrated a positive relationship with endorsement of the race/ethnicity item regardless of time point (b = .07, p = .020, extremely small effect size) and Black recruits and recruits identifying as another race/ethnicity or more than one race/ethnicity provided stronger endorsement of this item than White recruits (bs = .38 and .25, ps = <.001 and .016, small-to-medium and small effects, respectively). In Model C, time did not interact with any demographic factor (bs ≤ .08, ps ≥ .222), suggesting the pre/post effect on recruits’ reported agreement with the race/ethnicity item did not differ by their demographic characteristics.

Table 7.

Fixed effects from multi-level models concerning selected item: ‘Race/ethnicity affects interactions between police and young people locally’.

	Model 4a		Model 4b		Model 4c
Variable	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)	Estimate [95% CI]	p (FDR)
Time	0.17 [0.12, 0.23]	<.001	0.17 [0.12, 0.23]	<.001	0.18 [0.11, 0.26]	<.001
Age	–	–	0.07 [0.02, 0.11]	.020	0.08 [0.03, 0.14]	.007
Non-Male (vs. Male)	–	–	−0.02 [−0.14, 0.09]	.924	−0.03 [−0.16, 0.10]	.890
Black (vs. White)	–	–	0.38 [0.26, 0.51]	<.001	0.42 [0.28, 0.56]	<.001
Hispanic (vs. White)	–	–	−0.05 [−0.19, 0.08]	.632	−0.06 [−0.21, 0.09]	.632
Other Race (vs. White)	–	–	0.25 [0.07, 0.42]	.016	0.25 [0.05, 0.45]	.033
Time*Age	–	–	–	–	−0.04 [−0.10, 0.01]	.222
Time*Non-Male (vs. Male)	–	–	–	–	0.02 [−0.12, 0.15]	.966
Time*Black (vs. White)	–	–	–	–	−0.08 [−0.22, 0.07]	.449
Time*Hispanic (vs. White)	–	–	–	–	0.02 [−0.14, 0.17]	.966
Time*Other Race (vs. White)	–	–	–	–	0.00 [−0.20, 0.19]	.997

Note: CI, confidence interval; p(FDR), adjusted p-value using false discovery rate. Each model also contained random intercepts for recruit.

Discussion

The PA DMC Youth-Law Enforcement Curriculum was developed with the aim of preventing youth-law enforcement encounters from escalating into dangerous situations and unnecessary arrests by improving law enforcement officers’ and youths’ beliefs about one another, attitudes toward police-youth relationships, and behaviors during their interactions. The present study serves as an initial evaluation of the acceptability and effectiveness of the PA DMC Curriculum in educating police recruits about youth development and the extent to which recruits’ attitudes and beliefs about youth changed after participating in the training. The Police-Youth Engagement Survey, a measure developed in collaboration with community partners formed the basis of this evaluation, as it was designed to examine a wide range of beliefs relevant to officers and their work with youth.

Descriptive data collected from law enforcement recruits indicated that, overall, recruits received the PA DMC Curriculum well. Most recruits agreed that the information provided was helpful and taught them new ways to interact with youth, thereby enabling them to feel better prepared to interact effectively and appropriately with young people while working. Attendees most often indicated that the opportunities to interact with youth and young adults and hear their perspectives was a helpful or very helpful component of the training. Multilevel modeling results indicated that this training curriculum, which includes – among several components – small group discussions between young people and law enforcement, may contribute to changing reported beliefs and intended behaviors, echoing research demonstrating the importance of interpersonal dialogue and sharing of personal experiences when changing attitudes (e.g. Engberg, 2004). Although these findings do not support causal conclusions related to the effect of small group discussions, qualitative results indicated that some recruits wished to spend more time speaking with young people from the communities they serve. Recruits noted that trainings like the one they attended should be more widespread throughout the community, with suggestions that it should be shared with more police officers and more young people.

Curriculum goals and outcomes

A primary goal of the PA DMC Curriculum training is to improve interactions between law enforcement and young people through increased knowledge about youth development and understanding of youth perspectives. Results from this evaluation indicated that after the training, recruits’ reported increased intentions to stay calm and be respectful during their interactions with youth, improved understanding about officers’ responsibilities to positively interact with youth, and increased sensitivity to youths’ perspectives. Effects were particularly large for improving sensitivity to youth perspectives among recruits, whereas effects related to recruits’ own intentions and beliefs were more modest in size. Past studies of police trainings without youth participation have found similar increases in knowledge and attitudes but did not see significant increases on items related to understanding youth emotions and perspectives (e.g. no increase in belief that adolescence is a stressful time; Aalsma et al., 2018). This difference may suggest that hearing youth perspectives directly from young people during the training contributed to the observed relationship and its notable strength. Furthermore, the opportunity to share lunch at the conclusion of the training offered law enforcement officers and youth the opportunity to get to know one another on a personal level, in an informal setting, which can lead to improved relationships and reduced bias (Fridell, 2016; Spencer et al., 2016).

Our findings align with prior research suggesting that trainings with a social interaction component (e.g. sharing a meal with and/or hearing directly from youth, and especially youth of color) can positively impact attitude change more than training programs without an interactive component (e.g. Kalinoski et al., 2013). Items included in the first two factors, which address recruits’ calm and respectful intentions and understanding of their responsibility for positive interactions with youth, represented areas in which recruit responses increased to a small or small-to-medium degree, which may indicate that the training could incorporate more material aimed at improving these areas. However, these modest changes may be attributed to a ceiling effect, as officers already tended to provide strong endorsement of these items pre-training, limiting the potential to observe much positive change after training. Although new law enforcement recruits will likely need additional, in-depth training on both verbal and non-verbal de-escalation strategies for youth (Murphy, 2015; Stoughton, 2014), the PA DMC Curriculum may be a valuable first step in helping them understand how youth differ from adults and generating greater buy-in for using different policing techniques with each of these age groups.

In addition to goals related to improving officer beliefs and attitudes toward youth in general, this training, as implemented in Philadelphia – a diverse city with a uniquely large proportion of residents of color – also aimed to promote interactions with and better understanding of the young people with whom Philadelphia officers most often interact: youth of color. To do so, the training curriculum allowed officers to practice cross-cultural communication and hear the perspectives of several youth of color. Therefore, a particular point of interest was that after the training, recruits were more likely to agree that race/ethnicity impacts police interactions with youth. Although the increased awareness reported in the post-training iteration of the Police-Youth Engagement Survey represents a positive step forward, the size of the effect was small. It may be that this training, alone, is not sufficient to impact long-standing tensions between a predominantly White police force and the communities of color they serve (e.g. Braga et al., 2019; Nodjimbaden, 2020) and recruits and officers would benefit from receiving further training targeting this form of awareness. Importantly, extant research has yet to establish consensus about the longevity of outcomes related to racial bias training (Smith, 2015). Consequently, it is unclear whether officers’ heightened understanding of racial impacts on policing will translate into long-term behavioral change. If agency leadership seeks to enact long-lasting change with the PA DMC Curriculum, they may wish to modify the current iteration to include additional content that further emphasizes racial biases, stereotyping, racial disparities in legal outcomes, and/or historical origins of police mistrust in communities of color.

Interestingly, on average, Black and Hispanic/Latine recruits endorsed significantly higher levels of agreement with statements aligning with PA DMC Curriculum goals across time points – namely, even before receiving the training and afterwards. This effect could be a result of these recruits’ lived experiences as members of minoritized groups, making it easier to put themselves in the shoes of youth of color and support building positive relationships in the community (Novak et al., 2003; Sun, 2003). They may also already have positive relationships with the members of their communities, including young people and therefore be more likely to perceive them as individuals rather than as an ‘other’ group. The fact that recruits of color demonstrated heightened awareness and understanding of young people compared to their White counterparts also speaks to the need for diverse police departments that are representative of the neighborhoods they serve.

Results also differed by recruits’ gender, as male recruits agreed less strongly with Calm and Respectful Intentions items than female recruits or recruits that identified with another gender. This finding may reflect socialization effects, such that female officers might be more likely to enter a situation empathizing with youth, while male officers might prioritize other facets of safety (Schuck, 2014; White et al., 2021). Among other recruit demographics, age was not associated with item endorsement across any of the three survey factors. Importantly, although race and gender demographics were associated with overall attitudes across time points, none of the demographic variables significantly moderated the effect of time. This consistent finding suggests that the training had similar effects on participants regardless of their age, gender, and race/ethnicity. Not only does this finding suggest good construct validity for the different survey factors, but it also supports the proposition that the PA DMC Curriculum can similarly change attitudes and understanding across trainees with diverse backgrounds.

Limitations

When considering the significant results in this study, we also consider several relevant limitations. First, the Police-Youth Engagement Survey provides entirely self-report data, and the recruits who completed them had limited field experience; therefore, recruits’ answers may not reflect how they will actually behave in the field and could be impacted by social desirability bias, given the charged nature of police accountability in the current political climate. We also did not have access to behavioral data (e.g. arrest logs, bodycam data) to assess whether the training led to actual behavioral changes in the way recruits interacted with youth on the street. Further, recruits were from a police academy in Philadelphia, a single, urban Pennsylvania county, so the generalizability of these findings to experienced officers and recruits from other counties throughout the state and across the country may be limited.

Strengths of the Police-Youth Engagement Survey were that community partner needs (i.e. police, juvenile justice agencies, community providers, advocacy groups representing young people with system involvement, individuals with lived experience) and real-world importance were maximized. However, as a result, the survey was new, and further psychometric refinement may be needed in the future to optimize its measurement properties, particularly given the fact that not all items loaded neatly onto the three-factor structure and ceiling effects were observed on some items. Additionally, we did not assess the effectiveness of specific training components, such as youth participation, by evaluating other trainings without this component for comparison purposes. As a result, no one component of the training can be directly credited with producing specific outcomes; nevertheless, qualitative feedback provided preliminary insight into which components participating recruits described as most helpful.

Post-training surveys were administered immediately after training completion and no additional follow-up surveys were administered. Consequently, it is unknown whether changes – and the amount of change – in recruits’ attitudes, beliefs, and behavioral intentions persisted over time. Research has demonstrated that changes in knowledge and attitudes do not always lead to action-oriented behavior change (Webb & Sheeran, 2006) and that increased knowledge and attitudinal shifts, especially around bias, that arise following police officer training may not significantly impact actual policing behavior (e.g. Seguino et al., 2020). However, the pre- to post-training changes observed across recruits indicate an important level of change (i.e. at the individual level) that, in tandem with reforms at the organizational and systemic levels, is necessary for overall change within the field of policing (e.g. Lumb & Breazeale, 2002).

Future directions

Despite the above-identified limitations, the PA DMC Curriculum can serve as a starting point for recruits as they learn best practices for interacting with youth in their assigned communities. Other strategies that have demonstrated success both locally and nationally – foot patrol (Ratcliffe et al., 2011), procedural justice (Mazerolle et al., 2013), problem-oriented policing, and community policing (Braga et al., 2019) – should also be included more broadly across policing organizations as part of ongoing efforts to improve relationships between police and community members and, ultimately, help reduce negative interactions while enhancing community safety. Future research should utilize official police data, whether through bodycam footage and/or records of youth stops and arrests, to examine changes in officers’ actual behaviors from before to after training implementation. Administrative police department data, in particular, would allow for an evaluation of whether the PA DMC Curriculum achieved its goals of improving police-youth interaction outcomes and, consequently, decreasing use of force with young people, reducing number of youth arrests, and remediating the racial and ethnic disparities in both of these outcomes.

Additionally, conducting a randomized controlled trial, comparing factors and changes in Police-Youth Engagement Survey responses pre- and post-training between recruits who receive the PA DMC Curriculum training and a control group of recruits who do not, would help isolate the impacts of the program on recruits’ beliefs and understanding, enhancing the internal validity of the training evaluation. Given that the PA DMC Curriculum has also been designed for more experienced law enforcement officers, future research also should compare outcomes from such officers with those of new recruits to examine similarities and differences in observed trends. Because recruits have limited field experience, they may be overly optimistic about their ability to translate their new knowledge into future behaviors. In contrast, training experienced officers may be more challenging if certain police department practices and ideologies have become ingrained (Huisman et al., 2005). The use of long-term follow-up surveys would also further the current research by determining whether recruits maintain changes in attitudes and beliefs over time. Lastly, analysis of data from multiple counties and states with different jurisdictional characteristics (e.g. suburban, rural) and demographic features of law enforcement (e.g. race, length of time on the force) would add to the findings presented here and provide information with which to assess the generalizability of results.

Conclusion

The PA DMC Curriculum is a foundational training for law enforcement to better understand and positively engage with young people, as well as to increase buy-in for developmentally appropriate policing techniques. This evaluation of the curriculum suggests that the training may be a promising intervention to improve relationships between law enforcement and youth, given recruits’ increased sensitivity toward youths’ perspectives and increased support for engaging in positive, calm, and respectful interactions with young people as part of their job responsibilities. Furthermore, these outcomes suggest the curriculum may have the potential to influence positive interactions between police and youth of color specifically, given its encouragement of cross-cultural communication with young volunteers of color – especially in Philadelphia, where the police force is predominantly White and the city’s population is not (Vargas, 2020). Although results indicate that the training might be adjusted to further strengthen recruits’ behavioral intentions, and further evaluation should determine whether attitude changes translate into behaviors that improve outcomes during future interactions with youth, the PA DMC Curriculum appears to be a positive first step in improving law enforcement officers’ perceptions of young people in their assigned communities. As localities seek to improve relationships between law enforcement and youth, trainings such as the PA DMC Curriculum might be adopted to give youth a platform to voice their perspective and give officers tools to better serve their communities.

Data availability statement

The data are not publicly available given that the data were originally collected as quality improvement data, not research data. Although permission to use deidentified data for research purposes was obtained, police recruits were told that the raw data would only be seen by the research team. Thus, we cannot make it available at this time without violating the understanding with which police recruits provided the data.