Destigmatizing perceptions about Black adolescent depression: randomized controlled trial of brief social contact–based video interventions

Abstract

Objective:

To test the utility of brief social contact–based video interventions of a Black adolescent girl to reduce stigmatized attitudes and increase help-seeking intentions around adolescent depression.

Methods:

We conducted a randomized controlled trial (RCT) with 14- to18-year-old healthy volunteers drawn from the general US population. We enrolled participants through a crowdsourcing platform (n = 1,093) and randomly assigned participants to one of three video conditions (117 s each): depressed (DEP); depressed, adjusted to aspects unique to being a Black adolescent girl (including experienced or internalized racism; ADJ); and control (CONT). The primary outcome was the Depression Stigma Scale (DSS); secondary outcomes were the General Health-Seeking Questionnaire (GHSQ), and thermometers for Black and white race perception “warmth”.

Results:

Following the intervention, the DSS changed from baseline across the three conditions (p < .001). ADJ outperformed both DEP (p = .031) and CONT (p < .001). A race-by-intervention interaction (p < .001) revealed different response profiles between Black (ADJ = DEP = CONT; p = .726) and non-Black participants (ADJ > DEP > CONT; p < .001). DEP and ADJ both resulted in higher treatment-seeking intentions for both the emotional problems and the suicidal thought subscales of the GHSQ. We found a race-by-intervention interaction (p = .01) for the Black thermometer, which revealed a significant 2° increase in warmth among white (p < .001), but not Black, viewers (p = .06).

Conclusions:

On a short-term basis, brief social contact–based videos proved effective among adolescents in reducing depression-related stigma, increasing help-seeking intentions, and providing an “empathic foothold” in the lives of racially stigmatized groups. Even as the enduring effects of these interventions remain to be determined, the deployment on social media of short videos opens new opportunities to reach a large number of at-risk youth.”

Introduction

Black youth are twice as likely to die by suicide compared to their white peers (Calhoun & Martin, 2021; Coleman and Congressional Black Caucus, 2021). The COVID-19 pandemic has further exacerbated these sobering trends: a study based on decedent data from the office of Maryland’s chief medical examiner showed a doubling in the rate of suicide mortality among Blacks as closures advanced; the comparable rates halved among whites (Bray et al., 2021). These findings highlight the pressing need to develop interventions to reduce depression-related stigma and enhance help-seeking intentions, and that are tailored for the unique experiences that contribute to depression and high suicide risk among Black youth: a “one size fits all” model is simply not sufficient.

Contact-based interventions, in which individuals with mental illnesses interact socially with members of their communities, have proven to be the most potent interventions to combat stigma (Thornicroft et al., 2016). Antistigma interventions using short videos as a means for indirect contact with individuals with severe and chronic mental illnesses have proven to be effective (Morgan, Reavley, Ross, Too, & Jorm, 2018). There is a dearth of video-based antistigma interventions for adolescents, and none specifically designed to target the unique needs of Black youth that have been empirically tested. Such interventions could potentially have significant public health impact if deployed through social media platforms, an opportunity that could complement the rapid embrace of online-facilitated mental health interventions during the COVID-19 pandemic (Barney, Buckelew, Mesheriakova, & Raymond-Flesch, 2020).

We previously conducted a study in which brief social contact–based stimulus videos (<100 s long) had a significant impact on adolescents’ negative perceptions about depression and increased their intent to seek treatment if ever in need (Amsalem & Martin, 2021). In that study, we used as video stimuli two white actors and found differential responses based on viewer race: the response by Black adolescent participants was 58% lower than that by their white counterparts. Given that social contact–based video interventions are predicated on personal identification and emotional engagement, we assumed that Black viewers had more difficulty relating to a white protagonist (Moreland & Topolinski, 2010).

Based on those findings of race-moderated response, we conducted a randomized controlled trial (RCT) in which over one thousand healthy volunteers drawn from the general US population were assigned to one of three parallel conditions. Our specific goals were to examine if a short video adjusted to aspects unique to the living experiences of Black adolescents with depression could be more effective than an unadjusted one and if response to the intervention would vary depending on the viewer’s gender or race.

Methods

Intervention

We used three brief stimuli videos (each lasting 117 s) edited down from filmed interviews with a young professional actor, a 16-year-old Black girl (“Jasmin”). In a direct manner, she describes difficulties coping with depressive symptoms, thoughts that life is not worth living, false assumptions about treatment, and how and when she decided to seek help. In a first video (“DEP”), the actor follows a script identical to the one used by a white actor in our previous study. A second video (“ADJ”) is adjusted to aspects unique to being a Black adolescent girl (e.g., experienced or internalized racism). To that end, we conducted a focus group of four Black women (median age 39; range, 17–53) to inform the adjusted script. For the second video, based on the adjusted script, the actor spoke “in the way you talk with friends or family or when white people are not around”. Later on, in the same videos, the actor discusses how life has changed since getting treatment. The videos seek to humanize mental health symptoms through social contact with an individual perceived to be at an equal power status. In a third video, the same actor describes hobbies (soccer and social media) and what she likes to do with her friends. That control video (“CONT”) lacked the description of depressive symptoms or any other mental health–related content.

We hired the professional actor through the Youth Simulated Patient Program of the Child Study Center, Yale School of Medicine. We trained, debriefed, and compensated the actor following best practices for standardized patients (Cleland, Abe, & Rethans, 2009), including those specifically pertaining to underage actors (Budd, Andersen, Harrison, & Prowse, 2020). The three stimulus video clips are available for viewing through the url links in Appendix S1.

Participants, recruitment procedure, and ethics approval

We recruited participants using CloudResearch (Chandler, Rosenzweig, Moss, Robinson, & Litman, 2019), a crowdsourcing platform widely used in social science with ample experience enrolling groups underrepresented in research, including minors. We included only English-speaking youth, 14–18 years of age, and living in the United States. We chose this age range as it overlaps with the median age of the onset for major depression and as it includes the potential peer group of adolescents with depressive symptoms and/or suicidal ideation. The participants were randomly assigned to one of the three video stimuli conditions on a 2:2:1 ratio (DEP, ADJ, and CONT) and stratified by gender (female vs. male) and race (Black vs. non-Black). For the purposes of randomization to the video conditions, individuals who self-identified as belonging to a race other than Black or White were randomly assigned in equal parts to the Black / non-Black groups.

We used several accepted methods to exclude invalid participants to ensure quality of the collected data. First, we used an open-ended question format requiring the participant’s age as a screen question. We allowed only a two-digit number as a valid answer. Second, we added a CAPTCHA question (Completely Automated Public Turing test to tell Computers and Humans Apart) to prevent bots from continuing on to the survey itself. Third, we added a timer to the “next” button to ensure participants read the survey instructions (7-s minimum) and again for watching the video (100-s minimum). Fourth, we scanned the participants to exclude people attempting to answer the survey more than once, completing the assessment in less than the minimum expected time, for locations (GPS coordinates) outside of the United States, and for unusual or suspicious IP addresses. Finally, we excluded participants who failed any of the two valid questions, each one phrased in a consistent way that required a single, forced answer (e.g., “mark the fourth option below”).

Volunteers accessed the study through their preferred WiFi-enabled personal device, whether mobile or desktop, and were compensated $3.50 for their participation. Before initiating the study, respondents reviewed an informed assent document; parental consent was waived as part of the Institutional Review Board (IRB) approval. Adolescents who agreed to participate were directed to complete the study using a secure, online data-collection platform (Qualtrics; Provo, UT). This study was approved by the Yale Human Investigations Committee (Protocol #2000028980; MOD00040747). The study was registered in ClinicalTrials.org before data collection began (ID: NCT04890990).

Instruments

We assessed stigma toward depression using the Depression Stigma Scale (DSS; Griffiths, Christensen, Jorm, Evans, & Groves, 2004) and treatment-seeking intentions using the General Help-Seeking Questionnaire (GHSQ; Wilson, Deane, Marshall, & Dalley, 2008). We only used the personal subscale of the DSS (DSS-Personal), with its total score as the primary outcome. We analyzed the scale’s individual items as secondary endpoints. The DSS is rated on a 5-point Likert scale ranging from strongly disagree (1) to strongly agree (5). The total score comprises the sum of its item scores (range, 9–45), with a higher score indicating more stigma. The DSS-personal subscale has shown adequate psychometric properties: 0.71 test–retest reliability, 0.76 internal consistency (Griffiths et al., 2004), and Cronbach’s α = .83 in our previous study.

The GHSQ was developed to measure help-seeking intentions from different sources (friends, parents, mental health professionals, and others) and is divided into personal–emotional problems and suicidal thoughts (Ibrahim et al., 2019). The instrument consists of 10 items repeated twice for each part, measured on a 7-point Likert scale ranging from 1 (extremely unlikely) to 7 (extremely likely). One item is reverse scored (phrased as “I would not seek help from anyone”). We used the individual items as secondary endpoints. Higher scores on the scale and its items indicate more help-seeking intent. The GHSQ has shown good psychometric properties: Cronbach’s α = .70 and test–retest of .86 for personal–emotional problems and Cronbach’s alpha α = .83 and test–retest of .88 for suicidal thoughts. Cronbach’s α was .86 for both components in our previous study.

As an additional exploratory measure, we used a “race thermometer”, modeled after the study by Norton and Herek (2013), in order to gauge racial attitudes. The thermometers provide the following prompt: “Using a scale from 0 to 100, please tell us your personal feelings toward each of the following groups of friends, teachers, or colleagues. As you do this task, think of an imaginary thermometer. The warmer or more favorable you feel toward the group, the higher the number you should give it. The colder or less favorable you feel, the lower the number. If you feel neither warm nor cold toward the group, rate it 50.” To familiarize respondents with the response format, they were first presented with thermometers for their own sex and race (concordant: male or female and Black or White). They next rated the opposite (discordant) sex and race. Higher ratings (maximum 100) indicate warmer, closer, and more favorable feelings toward the target group, whereas lower ratings (minimum 0) indicate colder, more distant, or negative feelings. Given that the race thermometers are an adaptation of a previous scale designed for attitudes regarding sexual orientation, there are no available data on its psychometric properties as applied to race.

Data analysis

We used Pearson’s chi-square and one-way ANOVA to compare demographic variables between groups. For the primary outcome measure (DSS total score), we used general linear models (GLMs) to compare the adjusted mean of the difference before and after the intervention (PRE minus POST) across the three intervention video groups (DEP, ADJ, or CONT). We next repeated the GLM analysis twice: the first stratified by gender (female / male) and the second by race (Black/non-Black). For models with significant interaction effects (gender-by-intervention or race-by-intervention), we used one-way ANO-VAs to determine significant differences within separate strata and post hoc Tukey HSD tests to determine pairwise differences between interventions. For instances of baseline differences in DSS scores, we used PRE as a covariate. We used the same analytic strategy to compare change in Black and White thermometer ratings. For item-level analyses of the DSS and GHSQ, we use paired t-tests to compare change from baseline to endpoint in each of the three intervention groups and independent sample t-tests to compare group means between gender and race groups. For all t tests, we used Bonferroni’s correction, considering as significant only those results with p values under .001 (i.e., adjusted for at least 50 comparisons). We conducted all statistical analyses using IBM SPSS software, version 26.0 (Armonk, NY).

Results

We screened 1,425 individuals and excluded 134 (9.4%) who did not meet the inclusion criteria. We recruited 1,291 participants who completed baseline and endpoint assessments after watching the video they were randomly assigned to. Participants took a median of 8 min to complete the tasks (interquartile range, 7–10 min; p > .05 across groups). We excluded from analysis 198 individuals (15.3%) for failure on the embedded validity questions, yielding a total working sample of 1,093, evenly divided on a 2:2:1 ratio across the two active and one control conditions, respectively. Figure 1 shows the study’s overall flow, and Table 1 summarizes the demographic characteristics of the analyzed sample.

Figure 1.

Study flow chart

Table 1.

Demographic characteristics of randomized participants (N = 1,093)

	Intervention								Statistic
	Depressed n = 448		Depressed adjusted n = 456		Control n = 189		Total n = 1,093
	n	%	n	%	n	%	n	%	χ2/F	df	p
Gender
Female	200	45	213	47	82	43	495	45	5.753	6	.450
Male	201	45	202	44	84	44	487	45
Non-binary	44	10	35	8	18	10	97	9
Prefer not to say	3	1	6	1	5	3	14	1
Race
White	261	58	277	61	107	57	645	59	8.833	6	.183
Black	74	17	80	18	42	22	196	18
Other	101	23	79	17	32	17	212	19
Prefer not to say	12	3	20	4	8	4	40	4
Ethnicity
Latino/a/x/Hispanic	108	24	96	21	42	22	246	23	2.735	4	.603
Prefer not to say	14	3	10	2	7	4	31	3
Age
Mean ± SD	16.8	1.1	16.8	1.1	16.8	1.2	16.8	1.1	0.037	2	.964

There were no differences in demographic characteristics across the three intervention groups (p > .05). Males and females were evenly distributed, with 111 (10%) of participants identifying as non-binary or preferring not to say their gender. Black participants (n = 196) comprised 18% of the overall sample, a 1.3-fold higher proportion than in the 2019 US census (14%; Tamir, 2021). Individuals of a race other than Black or White (n = 212, 19%) primarily self-identified as “mixed or other” (n = 122, 58%), Asian (n = 74, 35%), or Indigenous / Native American (n = 16, 8%). Nearly one quarter of the sample (n = 246, 23%) identified their ethnicity as Latino/a/x or Hispanic.

We found a robust difference in our primary outcome (DSS change from baseline) across the three conditions (omnibus GLM: F = 15.21, df = 2, p < .001; Table 2). ADJ outperformed both DEP (1.95 ±.41 vs. 1.34 ±.92, Tukey HSD p = .031) and CONT (0.72 ±.41, Tukey HSD p .001), but DEP did not differentiate from CONT (p = .120).

Table 2.

Comparison between social-based video interventions on Depresson Stigma Scale (DSS) composite scores, stratified by gender and race (N = 1,093)

		Intervention
		Depressed n = 448						Depressed, adjusted n = 456						Control n = 189						Omnibus statistic
		Baseline		Post				Baseline		Post				Baseline		Post				GLM
	n	M	SD	M	SD	Delta	p	M	SD	M	SD	Delta	p	M	SD	M	SD	Delta	p	Fdf = 2	p	Tukey HSD
Gender
Female	530	17.07	5.25	16.06	5.63	1.01	<.001	17.48	5.08	15.51	5.07	1.97	<.001	16.85	4.63	15.85	4.75	1.00	.001	3.984	.019	ADJ = DEP > CONT
Male	563	18.52	5.56	16.92	5.51	1.60	<.001	19.00	5.65	17.08	6.02	1.93	<.001	18.37	5.59	17.87	6.52	0.50	.084	4.501	.011	ADJ > DEP = CONT
Race
Black	196	20.14	5.95	19.01	6.23	1.13	.01	19.75	6.20	18.23	6.46	1.52	<.001	20.19	5.77	19.21	5.97	0.98	.038	0.321	.726	ADJ = DEP = CONT
Non-Black	897	17.43	5.26	16.05	5.30	1.38	<.001	17.98	5.22	15.94	5.38	2.04	<.001	17.01	4.87	16.36	5.73	0.65	.011	8.592	<.001	ADJ > DEP > CONT

Values are reported as mean ( ± SD); DSS composites range from 9 to 45, with higher scores indicating higher stigma. Between-group analyses conducted with general linear models (GLM), covaried for differences in baseline scores. Bold font indicates p < .001, Bonferroni-corrected for within-group differences (paired t, Cohen’s d effect size range, 0.27–0.44). Video interventions: ADJ, depressed, adjusted; DEP, depressed; CONT, control.

Baseline DSS means were similar for the full sample (F = 1.046, df = 2, p = .352) but differed across gender (F = 33.02, df = 2, p < .001) and race (F = 20.72, df = 2, p < .001), leading to our use of DSS PRE as a covariate in stratified GLM analyses. Black participants had higher stigma scores at baseline than their non-Black counterparts (20.00 ± 5.99 vs. 17.59 ± 5.19, t = 5.747, df = 1,100, p < .001). There was a significant interaction between gender and intervention (F = 16.91, df = 3, p < .001), but females and males showed a similar overall pattern, with either of the depressed conditions outperforming control (ADJ = DEP > CONT; F ≥ 3.984, df = 2, p ≤ .019). By contrast, the interaction between race and intervention, which was also significant (F = 17.248, df = 3, p < .001), revealed different response profiles for Black (ADJ = DEP = CONT; F = 0.321, df = 2, p = .726) and non-Black participants (ADJ > DEP > CONT; F = 8.571, df = 2, p < .001; Figure 2). At the DSS item level, we found a significant decrease in six out of nine items in the ADJ group, compared to only three out of nine in the DEP group (Table S1).

Figure 2.

Mean change in the Depression Stigma Scale (DSS) following social contact–based video interventions. Panels depict DSS changes stratified by participants’ race (A) or gender (B). Error bars indicate standard error of the mean (SEM). *** indicates differences within groups (paired t) or between groups (Tukey HSD) significant at p < .001

Black and White race ‘thermometers’ did not change between times across the three conditions (F ≤ 0.492, df = 2, p ≤ .782). For the Black thermometer only, there was a significant interaction between race and intervention (F = 15.188, df = 5, p = .01), which revealed a significant increase in warmth among White, but not among Black viewers (2.05 ± 1.01, paired t = 3.86, df = 375, p < .001; vs. 2.28 ± 0.13, paired t = 1.89, df = 79, p = .06; Figure 3).

Figure 3.

Mean change in racial perceptions following social contact–based video interventions. Panels depict Black (A) or White (B) “temperature” changes between Black and non-Black participants. Error bars indicate standard error of the mean (SEM). *** indicates differences within groups (paired t, p < .001)

Table S2 summarizes results of the GHSQ, stratified by its two subscales. Both DEP and ADJ resulted in increased treatment-seeking intentions for emotional problems and suicidal thoughts. There were no comparable changes in the control group. Post hoc analyses did not reveal difference across gender or race in a preferred source for seeking help from (e.g., family or friends, vs. doctors or religious leaders).

Discussion

In this RCT, we replicated that our primary earlier finding: short social contact–based videos are an effective short-term intervention to reduce depression-related stigma and to increase help-seeking intentions among adolescents.

We found that the video of the same Black girl sharing a history of past depression, treatment, and recovery made a greater impact on viewers when its narrative was true to the realities of depression among Black girls – aspects unique to the Black experience when compared to those of White and other non-Black adolescents. Specifically, the overt mention of experiencing racism was a major difference between the two depressed conditions. For example, in the adjusted video, “Jasmin” describes feeling out of place in a school “in the suburbs”, where peers often ask her if her hair “is real” or comment on her “ghetto” tastes and preferences. She describes the internalized racism that compounded her depressive episode, when she felt that “popular girls looked nothing like me” or that she could not be beautiful “unless I was white”. Her depiction is consistent with unique challenges faced by many Black girls, such as gendered racism or the intersection of racism and sexism, and specifically misogynoir, the anti-Black racist misogyny that Black girls and women often experience (Bailey & Trudy, 2018). “Jasmin” goes on to mention how her family dismissed her concerns at the time as there was “nothing physically wrong”.

In the rehearsal and ultimate filming of the videos, we were attentive to the input of the focus group that informed the scripts. It was critical that the videos reflected some of the unvarnished realities of the Black experience in a way that was legitimate and rang true, but that was not an exaggeration that could further entrench racialized stereotypes. For example, the actor was instructed to speak as she normally would at home or with close friends, outside of predominantly “White spaces” (Anderson, 2015), and not necessarily in a manner consistent with African American Vernacular English (AAVE; Labov, 2010).

Our data revealed a gender-by-intervention effect: among female viewers, only the adjusted video separated from the control condition, whereas among males, both conditions separated, more prominently for the adjusted condition. This finding is inconsistent with two of our previous studies: In the first (Amsalem et al., 2020), we showed that women reported greater stigma reduction when viewing a female, supporting the hypothesis that identification with a familiar protagonist reduces stigma (Reis, Maniaci, Caprariello, Eastwick, & Finkel, 2011; Verosky & Todorov, 2010). In the second, the protagonist’s gender did not affect the viewer response pattern (Amsalem & Martin, 2021). We speculate several possible explanations for our current findings of a muted response in female when compared to male viewers: (a) Female participants, having lower stigma scores at baseline, in keeping with other studies (Cheung, Mak, Tsang, & Lau, 2018), may have approached a floor effect, with more limited “room to improve” than their male counterparts, (b) Female viewers may be more prone to consider depression as not that different from normative ebbs and flows in mood states, or (c) Females may preferentially respond to video stimuli that reach a clinical severity threshold – in this instance, less about depressive symptomatology alone than about the compounding effects noted above, particularly of racism.

We found an even more prominent race-by-intervention effect: Black participants responded similarly to both video conditions, whereas the effect was stronger to the adjusted condition among non-Black viewers, even after adjusting for baseline differences in stigma levels between the two groups. We had predicted the opposite response, given that the difference between the two conditions depicted subtleties germane only to the Black living experience. We explain this unexpected finding assuming that for Black adolescent viewers, depressive symptoms are hard to contextualize without an inherent component of racism, even if one not made explicitly. By contrast, for non-Black viewers, who are not personally affected by anti-Black sentiment, hearing an adolescent peer articulate the devastating impact of racism may provide a window into an experience they may have only imagined before – if giving it any thought at all. This explanation is further supported by the 2-increment in warmth among non-Black viewers toward Black individuals, the only significant change among the eight-race “thermometer” comparisons we made.

Taken together, these findings have two main implications. The first and more clinically relevant implication regards the identification and treatment of depressive disorders. The findings from this study corroborate those from our earlier report: a short social contact–based video intervention decreases stigma related to depression and promotes health-seeking intentions. In addition, with interventions specifically developed for Black youth, this study yielded a comparable response and effect size among Black viewers as it did among non-Black viewers. By contrast, in our previous study, the response to two White actors following the same script (unadjusted version) had been significantly lower for Black than for White viewers. As predicted, video-based interventions tailored to the specific characteristics of their target viewers can enhance anticipated response.

The second implication is regarding the intervention as a complement to anti-racist initiatives. By individualizing, personifying, and giving a face and a voice to the experiences of racism regularly confronted by Black youth, other adolescents – especially those in different racial groups – can gain an “empathic foothold” into the lives of racially stigmatized groups, partially through the mere exposure phenomenon, in which exposure to a less familiar and short stimulus can be sufficient to improve attitudes toward that stimulus (Monahan, Murphy, & Zajonc, 2000; Moreland & Topolinski, 2010). Moreover, the video interventions can be an effective means to “unlearn” racial biases, including well-meaning yet damaging perspectives, such as colorblindness.

Short videos hold promise as interventions that can make a substantial public health impact. The scalability and replicability of the social contact–based video approach can be anticipated by virtue of low production costs and ease in adjusting to underlying scripts, target populations, and specific goals. For example, other stigmatized conditions to target could include a range that spans from the normative to the psychiatric (e.g., from transphobia toward gender-expansive youth, to diagnostic acceptance, and treatment adherence among youth with early-onset psychosis.) Brevity is an admittedly relative concept, and what passes today as a “short” 2 min may well have to be decreased commensurate to the putative communal “shrinking attention span” (Newman, 2010), facilitated by ubiquitous online platforms with 1-min time limits, such as Instagram or TikTok.

We recognize several limitations. First and most notably, we had only two time points, which took place within minutes of each other. Even as this is standard for crowdsourcing studies among adolescents, there is at least one study among young adults populations that included a follow-up assessment after 1 month (Amsalem et al., 2021). We are not aware of any such studies with underage participants. Second, we used a single adolescent girl for the stimulus video, such that we did not empirically test whether the response would have varied according to the protagonist’s gender or whether concordant or discordant pairings would have resulted in different result profiles. Third, we analyzed data from all non-Black participants as a single group as we had insufficient power to analyze smaller racial categories. Fourth, by randomly assigning non-binary individuals to one of two gender categories, we may have obscured response patterns unique to this sizable fraction of our sample (9%). Fifth, we only evaluated perceptions and intentions rather than actual behavioral change. Finally, we did not assess for psychopathology among participants, which could have led to different response patterns.

Notwithstanding these limitations, this study contributes to our ultimate goals of improving destigmatization and acceptance of mental health disorders among adolescents, to promote their seeking help and adhering to treatment, and to address racism in everyday life through the salve of anti-racist actions.

Supplementary Material

Table 1

Table S1. Comparison between social contact–based video interventions on individual items of the Depression Stigma Scale (DSS; N = 1,093).

Table 2

Table S2. Comparison of social contact–based video interventions on items of the General Help-Seeking Questionnaire (GHSQ; N = 1,093).

Appendix

Appendix S1. Links to three stimulus video conditions.

Key points.

• Short social contact–based videos can reduce depression-related stigma and increase treatment-seeking intentions among adolescents.

• Interventions tailored to the specific realities of Black adolescent lives – including experienced or internalized racism – are effective, including among race-discordant youth.

• Short videos hold promise as scalable interventions that can impact preconceived notions about depression, treatment-seeking, and race. They can be designed to address the needs of specific target populations and in the case of Black youth, complement sorely needed antiracist efforts.