Effects of the Pathways for African American Success Program: Reductions in Adolescent Depression Through the Promotion of Racial Equity-Informed Parenting and Racial Pride

Abstract

African American adolescents face persistent disparities in depression and related behavioral health outcomes, which have been attributed to experiences with discrimination. The integrative model for the study of stress in Black American families provides a comprehensive perspective on how historical and current discrimination has direct and indirect effects on child behavioral health. Culturally based protective mechanisms (e.g., racial socialization) have been demonstrated to buffer adolescents from the negative effects of discrimination. The Pathways for African American Success (PAAS) program was developed as a scalable eHealth preventive intervention to facilitate parents’ use of strategies to protect their children from exposure to racism and the disproportionate consequences of risk behaviors, which we label as racial equity-informed parenting. Using data from a randomized effectiveness trial of PAAS and following stress and resilience approaches, we tested multiple hypotheses about the nature of discrimination as a source of risk, as well as program-driven improvements in adolescent racial pride as a source of protection. Our risk model indicated adolescent depression was associated with their own (direct) and their parents’ (indirect) experience of discrimination. Mediation analyses showed that program-driven improvements in adolescent racial pride served as a risk reducer/compensatory factor. Findings of moderation analyses, however, showed the negative effects of discrimination, even when racial pride was high. Implications for theory and cultural tailoring of evidence-based preventive interventions are presented.

Theory and extant research suggest that racial discrimination may help to explain the persistent disparities in depression and related behavioral health outcomes faced by African Americans (Centers for Disease Control and Prevention, 2019; Ghandour et al., 2019; National Center for Health Statistics, 2020; Price & Khubchandani, 2019; Substance Abuse and Mental Health Services Administration, 2020). Evidence for the link between racial discrimination and depression has been found across age, gender, and socioeconomic status (SES; Clark et al., 1999; Coker et al., 2009; Hudson et al., 2016; Klonoff et al., 1999; Simons et al., 2002; Yoon et al., 2019). Children and adolescents report discriminatory experiences in their microsystemic contexts, including schools, neighborhoods, and with the police (Berkel et al., 2009; Fisher et al., 2000; Saleem & Byrd, 2021; Seaton & Douglass, 2014; Simons et al., 2002; Spears Brown & Bigler, 2005; Wong et al., 2003). The integrative model for the study of stress in Black American families provides a comprehensive perspective on how historical and current discrimination have direct and indirect effects on child behavioral health (Murry, Butler-Barnes, et al., 2018). Extensive evidence shows direct pathways from discrimination during adolescence to immediate and long-lasting effects on mental health (Brody et al., 2006; Del Toro et al., 2019; Hughes et al., 2016; Simons et al., 2002; Smith-Bynum et al., 2014). In addition, vicarious experiences of discrimination (like the effect of exposure to secondhand smoke on chronic disease) in which children observe discriminatory acts against family members or other community members can also diminish mental health (Ford et al., 2013; Simons et al., 2002). Some evidence suggests that vicarious discrimination experiences may be even more insidious than personal experiences for children (Simons et al., 2002), perhaps by increasing perceptions that discrimination is ubiquitous and unavoidable. Given how powerful children perceive their parents to be, the fact that these experiences could happen to their parents may shake children’s perceptions about the security of the world to their very foundations. Other evidence suggests that the effects of discrimination on adolescent depression may filter through parents’ depressive symptoms (Odom & Vernon-Feagans, 2010), with sustaining negative effects from middle childhood to late adolescence (Murry et al., 2021).

Protecting African American Adolescents through Racial Socialization

Racial socialization refers to the strategies African American parents and other individuals in children’s microsystemic contexts use to prepare children for development in a context marked by racism (Saleem & Byrd, 2021). The form of racial socialization messages may be implicit or explicit, deliberate, or unintended, and proactive or reactive (Hughes & Chen, 1999); parents often use a combination of these strategies. Racial socialization messages have been characterized in several ways, including preparing children for race-related bias, promoting mistrust of other racial groups, coping with discrimination, cultural/ethnic socialization, celebrating culture and racial heritage, and appreciating diversity (Anderson et al., 2020; Bentley-Edwards & Stevenson, 2015; Brown et al., 2010; Granberg et al., 2012; Hughes & Johnson, 2001; Paasch-Anderson & Lamborn, 2013; Stevenson et al., 2002). Cultural/ethnic socialization messages are the most common; preparation for bias and promotion of mistrust, while especially rare during early childhood, often increase over the transition to adolescence (Hughes & Chen, 1997; McHale et al., 2006). Research has found that the use of cultural socialization messages leads to better adjustment for adolescents, whereas messages more focused on promotion of mistrust are associated with worse developmental outcomes (Berkel et al., 2009; Fisher et al., 2000; Grindal & Nieri, 2016; Murry et al., 2009; Neblett et al., 2008; Saleem & Lambert, 2016; Smalls, 2009; Stevenson, 1994; Varner et al., 2018). Furthermore, there is evidence to suggest that parents’ own experiences with discrimination are associated with their racial socialization practices (Brown et al., 2007; Hughes, 2003; Kurtz-Costes et al., 2019; Saleem et al., 2016).

Promoting Resilience Through Family-Based Prevention: The Strong African American Families (SAAF) Program

Given the previous research on ways in which African American caregiving practices buffer children from the potential negative effects of racial discrimination, the challenge for prevention scientists is to harness African American families’ protective capacities that originate in the family environment and test their efficacy in preventive intervention trials (Anderson et al., 2019; Coard et al., 2004). Pathways for African American Success (PAAS) is an eHealth adaptation of the SAAF program designed to dissuade youth from internalizing the negative consequences of discrimination and engaging in risky behavior. SAAF is a small group format preventive intervention, developed 20 years ago by Murry and Brody in partnership with African American communities in rural Georgia (Brody et al., 2004; Murry & Brody, 2004). At the time, there were no evidence-based programs developed for rural African American families. Community stakeholders identified their concerns for their children, including mental health, substance abuse, and sexual risk behavior; these concerns became the distal outcomes of the program. Input from stakeholders was combined with a decade of longitudinal research from those same communities (Murry & Brody, 1999, 2004), as well as general and culturally informed theories of adolescent development (Bandura, 1997; Gibbons & Gerrard, 1997; McAdoo, 1997) to develop a theoretical model of risk and protective mechanisms that guided the program content and delivery. In addition to common prevention targets included in other preventive interventions (e.g., warm and supportive parenting, communication, positive discipline, and youth attitudes toward risk), stakeholders emphasized the need to include attention to discrimination, the Black Church, and racial socialization. Prior to SAAF, these program targets had not previously been included in evidence-based preventive interventions for African American families (Jones & Neblett, 2016). Results of a randomized trial demonstrated that a cluster of intervention-targeted parenting strategies, which included racial socialization plus universal parenting skills related to parent–child relationships and positive discipline, were associated with improvements in adolescent racial pride and multiple behavioral health outcomes (Berkel et al., 2024; Brody et al., 2005; Murry et al., 2005, 2007, 2009, 2011; Wills et al., 2007). Since then, there has been a growing recognition of the importance of racial socialization as a core component in prevention programs for African Americans and members of other minoritized populations (Anderson et al., 2019; Anderson & Stevenson, 2019; Caldwell et al., 2010; Coard et al., 2004; Yasui & Dishion, 2007).

The PAAS Program

The PAAS program uses similar content and structure as the SAAF program but is adapted in light of implementation barriers identified in the SAAF trial (Berkel et al., 2013). Specifically, to address challenges to program fidelity and attendance, the delivery model was adapted from in-person small groups to an eHealth format delivered to individual families (for additional details, see Murry, Butler-Barnes, et al., 2018, Murry et al., 2019). Similar to SAAF, PAAS includes concurrent parent and youth sessions, followed by a conjoint family session. Importantly, a focus on discrimination, racial socialization, and racial pride is interwoven throughout the program. Parent sessions were designed to promote racial socialization and universally adaptive parenting practices, which included the establishment of clear expectations about alcohol/substance use and sexual risk (Murry et al., 2007, 2011). Youth sessions also include both universal (e.g., risk resistance skills) and culturally specific (e.g., dealing with racism) content. After concurrent parent and youth sessions, family members come together to reinforce what each learned in their respective sessions. A “highway to success” framework organizes the session topics with off ramps and side streets to illustrate associations between choices and consequences. In an attempt to incorporate some of the benefits of the original small group format of SAAF, the program includes characters that look and sound like members of the local community and participants can customize avatars to represent themselves. Family avatars interact with the characters in the program via a menu of preprogramed responses.

The PAAS Effectiveness Trial

To test the effectiveness of the PAAS program, a 3-arm (technology format, traditional small group format, and literature control) randomized trial was conducted with 418 sixth graders and their primary caregivers. Families were randomized to one of the three conditions: technology (n = 141), group (n = 141), and literature control (n = 136). Families assigned to the technology condition participated in the eHealth version of PAAS. The group condition was delivered via traditional small group format and was similar to SAAF, but updated to be more similar to the eHealth version (e.g., using the “highway to success” framework) for the sake of comparison. Families in the literature control received mailings on topically related brochures and pamphlets similar to those covered in the two active arms. Additional details about program content and delivery are provided in the methods section.

The overall goal of the trial was to determine whether the eHealth version of PAAS would be more effective than the control and comparable to the traditional small group version. We also sought to address concerns that African American families in low-income, rural communities would be willing to engage in a computer-based preventive intervention. Results show that, irrespective of parent age, education, or SES, families in the technology condition had significantly higher rates of program initiation, lower rates of attrition, and attended more sessions overall than the traditional small group condition (Murry, Berkel, & Liu, 2018). This demonstrates that translating the program to an eHealth format achieved the intended goal of increasing access to families.

In terms of effectiveness, previous analyses demonstrated evidence for two clusters of parenting practices, which differed somewhat from the SAAF trial and demonstrated distinct patterns of improvement by condition (Murry et al., 2019). The group condition reported higher increases in general warm and supportive parenting, compared to the eHealth and control conditions. The eHealth condition reported higher improvements relative to the other conditions in parenting related to challenging topics faced by African American parents, including racial socialization and communicating expectations for sexual behavior and substance use. The fact that this set of parenting practices clustered together was particularly enlightening and may be explained by parents’ understanding of the fact that, due to race-related bias, their offspring may face disproportionate consequences of engagement in risk behaviors compared to White peers (Pflieger et al., 2013; Zapolski et al., 2014). Consequently, we conceptualize this parenting construct as “racial equity-informed parenting.” Moreover, these three topics comprise a set of issues that parents may find particularly challenging to discuss in a group format, making them particularly well-suited for programs delivered via technology (Murry et al., 2019). In terms of adolescent outcomes, the eHealth condition reported reductions in intent to engage in risk behaviors at posttest, and long-term reductions in sexual risk behavior and substance use compared to both of the other conditions. Based on our primary interest in testing the eHealth version of the program, and our previous finding that the group format did not differ from the control, here we collapsed the group and control conditions to conduct a two-arm comparison of the eHealth version of the program relative to the group and control conditions combined.

The current study was conducted to expand on previous findings by testing a core conceptual mechanism of the eHealth PAAS program; that is, does the program protect adolescents from the negative consequences of discrimination? Following the stress process model (Roosa et al., 1997) and resilience frameworks (Fergus & Zimmerman, 2005; Zimmerman, 2013), we test multiple pathways by which the PAAS program may confer resilience in the face of discrimination. These approaches offer guidance on how protective processes may work against the negative effects of stressors and analytic approaches to test these conceptual models. Risk reducers (as they are labeled in the stress process model)/compensatory factors (as labeled in the resilience frameworks) have direct or indirect effects on an outcome of interest that works in opposition to the stressor. Protective factors (as labeled in both approaches) modify the association between the stressor and the outcome and are tested via moderation. Berkel and Colleagues (2024) compared similar pathways using data from the SAAF efficacy trial and found evidence of program-induced improvement in Black Pride as a risk reducer/compensatory factor predicting decreases in a latent construct of psychological functioning, which included depression, hope, and perceived life chances. The evidence did not, however, support Black Pride as a protective factor in the association between adolescent discrimination and psychological functioning. By contrast, moderation analyses of adaptations of SAAF for older adolescents (AIM and SAAF-T) have demonstrated better program effects on conduct problems or internalizing behaviors when discriminatory experiences were high (Brody et al., 2021). Furthermore, in an adaptation of SAAF for co-parenting families (ProSAAF), results showed that program-induced improvements in parenting were associated with a weaker relation between discrimination and adolescent depression (Lei et al., 2021).

In the current study, we first conducted analyses to establish the risk processes by which discriminatory experiences were associated with adolescent depression. We hypothesized adolescent depression would be directly associated with personal experiences of discrimination and vicarious exposure to parents’ experience of discrimination. In addition, we tested an indirect pathway, by which parent experiences of discrimination would be associated with adolescent depression through parent depression. These risk processes are represented by the dotted lines in the conceptual model in Figure 1. Next, we tested two protective processes using moderation and mediation. We examined adolescent racial pride as a moderator of each of the risk processes described above, hypothesizing that it would reduce the associations between the predictor and adolescent depression. We also examined racial pride as a mediator of program effects working in opposition to discrimination. We hypothesized that PAAS would contribute to a cascade of enhancement of racial equity-informed parenting and adolescent racial pride, which would reduce adolescent depression, above and beyond the negative effect of personal or vicarious exposure to discrimination.

Figure 1.

Conceptual Model

Methods

Recruitment and Randomization

Recruitment and pretest data collection took place in five rural counties in the western region of Tennessee between 2009 and 2012. Criteria for county selection were based on rurality, proportion of African American residents, evidence of high rates of teen pregnancy, and negative overall health indicators. Specifically, counties were selected based on the following characteristics:

• rurality index scores greater than 11 (scale of 0 = least rural to 16 or greater = most rural)

• over 30% African American residents

• over 600 African American adolescents in the targeted age range

• higher than state average teen pregnancy rates

• poor health determinant outcomes, including healthcare access, health behaviors, socioeconomic factors related to health, and physical environment (U.S. Census Bureau, 2012).

Potential participants were excluded from the study if the primary caregiver or adolescent did not speak English.

Recruitment procedures began with middle schools in each of the selected counties providing a list of 6th-grade African American students. Children were assigned a recruitment ID, and their order was permuted randomly to contact families for eligibility screening and recruitment. A letter was mailed to all parents/guardians informing them about the study. A community liaison (well-known local community leader) either called or visited families’ homes to provide information about the study. Active consent was obtained from primary caregivers and assent from adolescents. Randomization was carried out with computer-generated random numbers kept in opaque envelopes. After the pretest, participants were randomized to one of three conditions based on the assignment found in the numbered envelope in the corresponding enrollment list. Of the 550 families contacted, 78% consented to participate.

Program Delivery

EHealth Condition.

The ultimate goal is for PAAS to be available for families to use at any time on their own devices. Because an important aim of the trial was to establish effectiveness, the technology condition was conducted in a community setting with two rural African American community members serving as onsite technology intervention assistants (TIAs). TIAs received 6 hr of training on program content, procedures for setting up and managing weekly computer interactive sessions, and troubleshooting on-site computer-related issues. In preparation for the sessions, laptops from the research center were set up by TIA at designated community centers, such as churches, youth centers, and libraries. A 3-hr time block was allocated for each session. A member of the research staff sent a schedule to families informing them of the dates and times over the course of 6 weeks when the PAAS program would be available in their community. Each family selected a time that fits their schedule and received a follow-up call from the research team to confirm attendance. When families arrived on site, the TIAs set up each parent and adolescent on available laptops and remained present to provide any needed technical assistance. To ensure privacy and reduce distractions, each laptop was placed inside a trifold partition. After each family member completed their respective individual sessions, a TIA escorted the adolescent to the parent’s laptop to complete the conjoint family session. Individual and family sessions lasted an average of 45 min, for a combined 1.5 hr per week and 9 hr of total dosage.

Small Group Condition.

For the traditional small group-based condition, teams of three rural African American community members served as facilitators (one for parents and two for youth sessions). Prior to implementing the program, facilitators received a total of 36 hr of training over the course of 6 days. A total of 12 groups of roughly 12 families met weekly. Facilitators presented the PAAS curriculum, organized role-playing activities, guided discussions among group members, and addressed participants’ questions. Each parent/youth concurrent session and family session lasted 1 hr on average, resulting in 2 hr per session and 12 hr of total dosage.

In determining the effectiveness of interventions, it is necessary to evaluate whether the program was delivered as intended (Gottfredson et al., 2015). In programs delivered via traditional, in-person program formats, implementation can vary widely (Berkel et al., 2011, 2013). In the group condition, sessions were videotaped to assess fidelity using a checklist with one item for each instruction in the manual. For each group, two-parent, youth, and family sessions were randomly selected for fidelity coding (33% of sessions). Reliability checks were conducted on 23% of the fidelity assessments, and interrater reliability exceeded 80% for parent, youth, and family sessions. Fidelity to the curriculum exceeded 80%.

Literature Control.

Families received a weekly mailing of brochures and pamphlets containing the same topical content as the eHealth and traditional small group conditions.

Incentives.

Upon completion of each session, for eHealth and group conditions, each family received a $25 financial incentive. Murry’s prior work (Murry & Brody, 2004) shows monetary incentives yield high adherence in assessment completion, retention, and program attendance, which was necessary for establishing the effects of the program with a relatively small sample in this study. The importance of monetary incentives also has been noted by other researchers (Guyll et al., 2003) and is used in community agencies that deliver programs to low-income, rural families. Because both conditions received the same incentives, it is unlikely that these incentives had differential impact across conditions. However, as the literature control condition was not assigned to attend sessions, they did not receive incentives for session attendance.

Interview Procedures

Participating families were interviewed at the pretest, posttest (M = 14.5 [SD = 4.4] months after the pretest), and long-term follow-up (M = 22.6 [SD = 3.7] months after the posttest), approximately 3 years after the pretest. Retention from the pretest to the posttest exceeded 80%. Reasons for attrition in the eHealth condition were as follows: the death of a parent (n = 5), moving (n = 8), and time constraints (n = 1). In the group condition, reasons included loss of custody and other family problems (n = 17), work schedules (n = 9), moving (n = 11), and time constraints (n = 2). Finally, in the literature control, all dropout was due to moving (n = 28). Due to the timing of the grant, only families who enrolled during the first 18 months of the trial were included in the long-term follow-up (n = 167; eHealth = 76; group = 31; control = 60; total comparison = 91).

Several steps were taken to enhance rapport and cultural understanding during the data collection procedures; these methods followed the approach developed as part of the SAAF efficacy trial (Brody et al., 2004; Murry & Brody, 2004). The constructs selected for inclusion were driven by existing family-centered, evidence-based preventive interventions (e.g., Spoth et al., 1998) and by input from community partners. In particular, the program’s conceptual model, study measures, and procedures were reviewed and refined with feedback from 40 African American community members (Murry & Brody, 2004). Trained African American community members served as interviewer staff. Self-report questionnaires were administered to parents and adolescents via laptop computers. To maintain confidentiality and address potential literacy concerns, questions were read via computer using the Audio Computer-Assisted Self-Interview program and participants entered responses with a remote keypad. Each interview lasted approximately two hours. To reimburse families for each data collection point, parents received $100, and youth received $50.

Several steps were undertaken to ensure that research activities did not cause harm. First, community research staff administered all questionnaires to each participant individually in a private area where no other household members could hear. Respondents keyed their own responses into a remote keypad so that interviewers could not see the responses. In addition, community research staff were instructed to report any adverse incidents to the research project coordinator. After data collection in families’ homes, community research staff completed comment cards to report any unusual or adverse events that occurred during an interview. These cards and the data obtained from participants were monitored carefully and were reviewed daily by the project coordinator, who was instructed to inform the principal investigator if any serious adverse event occurred. No adverse events occurred during the course of the study. All study procedures were approved by the Institutional Review Board at Vanderbilt University (#090217).

Measures

Parent Discrimination.

At the pretest, parents reported on their experiences with discrimination using the 9-item Racial Hassles Questionnaire (Harrell, 1997), rated on a 4-point scale from 1 (never) to 4 (frequently). Example items were “blamed of something or treated suspiciously (as if you have done something or will do something wrong) because of your race” and “called a name or harassed because of your race.” Cronbach’s α was .91.

Parent Depression.

At the pretest, parents reported on their depression symptoms using the 20-item Center for Epidemiological Studies—Depression scale (Radloff, 1977), rated on a 4-point scale from 0 (once in a while) to 2 (all the time). Example items were “I could not get ‘going’” and “I felt depressed.” Cronbach’s α was .87.

Racial Equity-Informed Parenting.

A latent construct established by Murry and Colleagues (2019) was used to examine change in parenting practices targeted by PAAS, including racial socialization and communication about expectations with respect to substance use and sexual behavior. At pre- and posttest, parents reported on their positive racial socialization practices using the 5-item celebration of racial heritage subscale of the Racial Socialization Scale (Hughes & Johnson, 2001), rated on a 3-point scale from 1 (never) to 3 (3–5 times). Example items were “done or said things to encourage your child to do other things to learn about the history or traditions of your racial group” and “celebrated cultural holidays of your racial group.” Items were coded such that higher scores indicate higher levels of positive racial socialization. Cronbach’s αs were .85 at pretest and .88 at posttest. At the pretest and posttest, parents reported on their establishment of clear rules about substance use using an expanded 9-item version of the Substance Use Rules Communication Scale from Strengthening Families (Spoth et al., 1998), rated on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). Example items included “I have explained my rules concerning drug use to my child” and “I have explained the consequences of not following my rules concerning drug use to my child.” Items were coded such that higher scores indicate more endorsement of clear rules. Cronbach’s αs were .91 at pretest and .95 at posttest. At pretest and posttest, parents reported on their communication with children about sex using the eight-item Frequency of Sexual Communication scale (Miller et al., 1998), rated on a four-point scale from 1 (never) to 4 (often). Example items were “how often have you and your child talked about pressures in his/her life to have sex?” and “how often have you and your child talked about the importance of young people using condoms when they are sexually active?” Items were coded such that higher scores indicate more frequent sexual communication. Cronbach’s αs were .96 at pretest and posttest. Confirmatory factor analysis of the latent parenting construct indicated adequate fit (X²[96] = 246.36, p ≤ .05; RMSEA = 0.06 [90% CI = 0.05, 0.07]; CFI = 0.91; SRMR = 0.08) and all loadings were over 0.30, p ≤ .001 (Murry et al., 2019).

Adolescent Discrimination.

At the pretest and long-term follow-up, adolescents reported on their experiences with discrimination using the 9-item Racial Hassles Questionnaire (Harrell, 1997), rated on a 4-point scale from 1 (never) to 4 (frequently). Example items were “blamed of something or treated suspiciously (as if you have done something or will do something wrong) because of your race” and “called a name or harassed because of your race.” Cronbach’s αs were .86 at pretest and .90 at the long-term follow-up.

Adolescent Racial Pride.

At the posttest and the long-term follow-up, adolescents reported on their racial pride using the 7-item Black Pride subscale of the Multidimensional Model of Racial Identity (Sellers et al., 1997), rated on a 4-point scale from 1 (strongly disagree) to 5 (strongly agree). Example items were “I am happy that I am Black” and “I believe that because I am Black, I have many strengths.” Cronbach’s αs were .81 at posttest and .90 at the long-term follow-up.

Adolescent Depression.

At the posttest and the long-term follow-up, adolescents reported their symptoms of depression using the 27-item Child Depression Inventory (Helsel & Matson, 1984), rated on a 3-point scale from 0 (once in a while) to 2 (all the time). Example items were “I am sure that a terrible thing will happen to me” and “I feel alone all the time.” Cronbach’s αs were .81 at the posttest and .90 at the long-term follow-up.

Intervention Condition.

To model the effects of the eHealth program relative to the control or traditional group format, we created a dummy coded PAAS eHealth variable (Hayes & Preacher, 2014), for which the eHealth program was coded as one, and the two comparison conditions (group and control) were coded as 0.

Participants

Primary caregivers were predominantly female (84%) and on average 40 years old. The majority (87%) had completed high school. Half were single parents, 37% were married, and the remaining were in grandparent-headed households. Most primary caregivers (63%) were employed and worked approximately 40 hr per week; 50% owned their own home; 56% reported that their income was adequate income to meet their needs; and 14% received public assistance. On average, there were 2.7 children in the home. Just over half of the adolescents were female (54%) and were on average 11 years old at the pretest and 14 years old at the long-term follow-up.

Analytic Strategy

General Analytic Approach.

Study hypotheses were tested via structural equation modeling in Mplus 8.1 (Muthén & Muthén, 2018) in two separate models (see details below). To address missing data, we conducted Little’s (1988) Missing Completely at Random (MCAR) test, which demonstrated that data were likely missing completely at random (X²[30] = 26.16, p = .67). Full Information Maximum Likelihood was used to address missing data (Enders & Bandalos, 2001). Nesting of individuals within clusters can result in violations of independence. Based on the fact that only one of the three conditions was conducted in a group format, we determined that county was the most appropriate cluster variable. Intraclass Correlation Coefficients (ICCs) for each variable were all under .05, indicating independence of the data by county (Kreft & de Leeuw, 1998).

We evaluated model fit based on a nonsignificant X² or a combination of SRMR close to .08, RMSEA close to .06, and CFI close to .95, based on simulation studies that revealed using this combination rule resulted in low type I and type II error rates (Hu & Bentler, 1999). We also examined modification indices to determine whether additional paths were indicated by the data. The significance of standardized βs represents tests of study hypotheses. Bias-corrected bootstrap confidence intervals (MacKinnon et al., 2002) were used to assess the significance of the standardized indirect effects for hypothesized mediational pathways (i.e., considered significant if the 95% CI did not cross zero).

Risk Processes.

First, we sought to establish the direct and indirect influence of discrimination on adolescent depression as described in Figure 1. Baseline data were used for this model to enable an examination of the hypothesized relations without the influence of the intervention. We included direct paths from adolescent discrimination, parent discrimination, and parent depression to adolescent depression. We also modeled the indirect pathway from parent discrimination to parent depression to adolescent depression.

Protective Processes.

Next, we conducted moderation and mediational analyses to examine the protective processes of racial pride as depicted in Figure 1. Because we wanted to examine longitudinal effects on adolescent depression, as noted above, models assessing protective processes were conducted only with families who completed the long-term follow-up. We tested the interaction of racial pride with parent discrimination, parent depression, and adolescent discrimination in three separate models. Specifically, we examined whether racial pride (long-term follow-up) moderated the negative effects of parent discrimination (pretest) and depression (pretest) and adolescent discrimination (long-term follow-up) on adolescent depression (long-term follow-up). In each of these models, we centered all predictors prior to creating interaction terms, then entered the centered predictor, moderator, and control (posttest adolescent depression) variables plus the interaction term.

We also examined the indirect effects of the PAAS eHealth program on adolescent depression, through improvements in parenting and racial pride. For this model, we used the parenting construct identified in our previous work, which we label as racial equity-informed parenting (Murry et al., 2019). As noted above, we collapsed the group and control conditions to conduct a two-arm comparison of the eHealth version of the program relative to the group and control conditions combined. To assess change in parenting and adolescent outcomes, we included autoregressive pathways controlling for the previous wave of each variable. That is, we included baseline parenting to determine the effect of the eHealth program on improvements in parenting at posttest, and we included posttest adolescent depression to model the association between racial pride and improvements in depression at the long-term follow-up. Baseline variables were allowed to covary.

Results

Study Descriptives

Means, standard deviations, and correlations for all study variables are presented in Table 1. Parents’ discrimination was positively correlated with their depressive symptoms and racial socialization practices. Adolescents’ discrimination was positively correlated with their depressive symptoms, but not their racial pride. Racial pride was negatively correlated with depression. Parenting indicators were intercorrelated within each wave and across waves. Pretest parent depression was correlated with pretest parenting (but not with posttest parenting) and with adolescent depression at posttest and follow-up.

Table 1.

Descriptives and Correlations for All Study Variables

Study variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14
Wave 1
1. Parent discrimination	—
2. Parent depression	0.24***	—
3. Subs use rules	−0.02	−0.17***	—
4. Sex risk communication	−0.02	−0.11*	0.14**	—
5. Racial socialization	0.21***	0.12*	0.16***	0.10*	—
6. PAAS eHealth	−0.01	−0.01	−0.03	0.03	0.04	—
Wave 2
7. Subs use rules	−0.03	−0.12*	0.41***	0.13*	0.14**	0.11+	—
8. Sex risk communication	−0.12*	−0.06	0.16**	0.28***	0.01	0.05	0.19***	—
9. Racial socialization	0.10+	0.04	0.21***	0.10+	0.42***	0.09	0.13*	0.02	—
10. Adolescent racial pride	0.12*	0.00	0.02	−0.08	0.08	−0.08	−0.01	0.09+	0.04	—
11. Adolescent depression	0.05	0.20***	−0.16**	−0.08	−0.02	−0.01	−0.07	0.03	−0.01	−0.08	—
Wave 3
12. Adolescent discrimination	0.11	0.15+	0.02	0.02	0.19**	0.01	0.11	0.05	0.10	−0.07	0.17*
13. Adolescent racial pride	−0.03	−0.15*	0.09	−0.04	0.01	0.07	0.23**	0.05	−0.02	0.36***	−0.20**	−0.13
14. Adolescent depression	0.06	0.27***	−0.05	0.06	0.07	−0.02	−0.09	0.03	−0.09	−0.04	0.47***	0.25***	−0.32***
M	14.4	13.5	4.4	2.6	2.0	—	4.5	2.7	1.9	12.8	6.2	13.3	6.1	13.1
SD	5.6	9.2	0.7	0.9	0.5	—	0.8	0.8	0.6	2.3	5.3	2.2	6.9	4.7

PAAS: Pathways for African American Success.

^*p ≤ .05.

^**p ≤ .01.

^***p ≤ .001.

Risk Processes

Results of the risk process analyses are presented in Figure 2. The model demonstrated a good fit to the data (X²[1] = 0.80, p = .37), and no changes were suggested by modification indices. Standardized βs indicated that discrimination was significantly and positively associated with depression for both adolescents (β = .28, p ≤ .001] and parents (β = .24, p ≤ .001). Parent discrimination was not directly associated with adolescent depression (β = −.02, p > .05); however, the positive indirect effect through parent depression was significant (β = .047 [95% CI: 0.022, 0.084]).

Figure 2.

Concurrent Relations Between Discrimination and Adolescent Depression at Pretest

Notes: X²(1) = 0.80; p = .37; *p ≤ .05. **p ≤ .01. ***p ≤ .001.

Protective Processes

The moderation effects of racial pride on the association between risk variables and adolescent depression are presented in Table 2. None of the interaction terms were significantly associated with adolescent depression at the long-term follow-up, indicating that racial pride did not act as a moderator in these analyses. Hence, the interaction terms were not included in the model testing mediational effects.

Table 2.

Moderational Effects of Racial Pride on Adolescent Depression

Model results	B	Std. Error	t	p
Model 1: parent discrimination
Regression constant	12.74	2.80	4.54	≤.001
Autoregressive path (covariate)	0.62	0.09	6.65	≤.001
Parent discrimination (X)	−0.17	0.38	−0.45	.66
Racial pride (M)	−0.46	0.20	−2.14	.03
Interaction term (X*M)	0.01	0.03	0.46	.64
Model 2: parent depression
Regression constant	11.78	2.90	4.06	≤.001
Autoregressive path (covariate)	0.57	0.10	5.97	≤.001
Parent discrimination (X)	0.31	0.32	0.96	.34
Racial pride (M)	−0.46	0.21	−2.14	.03
Interaction term (X*M)	−0.02	0.03	−0.74	.46
Model 3: adolescent discrimination
Regression constant	5.85	0.45	12.88	≤.001
Autoregressive path (Covariate)	0.55	0.10	5.55	≤.001
Adolescent discrimination (X)	0.25	0.10	2.39	.02
Racial pride (M)	0.36	0.17	2.13	.03
Interaction term (X*M)	−0.02	0.03	−0.58	.56

In the mediation model (see Figure 3), the p value for the X² test of model fit was close to significant (X²[63] = 77.10, p = .11); to be conservative, we also examined the RMSEA, CFI, and SRMR, which also indicated adequate fit of the model (RMSEA = 0.04 [90% CI = 0.00, 0.06]; CFI = 0.93; SRMR = 0.07). No changes were indicated by the modification indices. Standardized βs demonstrated adolescent experiences of discrimination were associated with increases in adolescent depression from the posttest to the long-term follow-up (see Figure 3). Assignment to the eHealth version of PAAS led to increases in racial equity-informed parenting from pretest to posttest, which, in turn, predicted increases in adolescent racial pride at the long-term follow-up. Racial pride was associated with decreases in adolescent depression from posttest to follow-up. Mediational analyses demonstrated a significant indirect program effect on adolescent racial pride through racial equity-informed parenting (β = .097 [95%CI = 0.006, 0.223]), and on adolescent depression through racial equity-informed parenting and racial pride (β = −.021 [95%CI = −0.083; −0.002]).

Figure 3.

Examination of the Protective Effects of the Pathways for African American Success eHealth Program on Adolescent Depression

Notes: Autoregressive pathways to model change across time for parenting, racial pride, and depression were included in the model, but are excluded from the figure for visual clarity.

Model fit: X²(63) = 77.10, p = .11; RMSEA = 0.04 (90% CI = 0.00, 0.06); CFI = 0.93; SRMR = 0.07.

*p ≤ .05. **p ≤ .01. ***p ≤ .001. +p ≤ .10.

Indirect intervention effects on racial pride through parenting: β = .097 (95%CI = 0.006, 0.223).

Indirect intervention effects on adolescent depression through parenting and racial pride: β = −.021 (95%CI = −0.083; −0.002).

Discussion

The goals of this study were twofold: (a) to expand on the evidence for the direct and indirect pathways by which discrimination influences depression among rural African American adolescents and (b) to examine two potential protective processes within a culturally tailored intervention administered via an eHealth delivery format. Results supported prior research showing a direct association between adolescent experiences of discrimination and depression, as well as an indirect association between parent discrimination and adolescent depression through parent depression. Furthermore, we found the eHealth version of PAAS reduced adolescent depression by improving racial equity-informed parenting and adolescents’ racial pride. However, racial pride did not attenuate the effect of discrimination on adolescent depression.

Persistent disparities in adolescent behavioral health outcomes have been attributed to discriminatory experiences (Berkel et al., 2009; Brody et al., 2006, 2008; Clark et al., 1999; Coker et al., 2009; Ford et al., 2013; Simons et al., 2002), through direct and indirect mechanisms (Murry, Butler-Barnes, et al., 2018). Results of this study confirm the concurrent association between personal experiences of discrimination and depression, when children were just 11 years old. In contrast to previous research highlighting the negative consequences of second-hand exposure to discrimination against family or community members (Ford et al., 2013; Simons et al., 2002), this study found no direct associations between parents’ experiences with discrimination and their children’s depression, either concurrently or across time. A possible reason for the lack of a significant association may be due to measurement, as our measure of discrimination did not assess whether children were aware of their parent’s experiences. Alternatively, previous research suggests that the association between parents’ exposure to discrimination and their children’s mental health may be indirect (Murry et al., 2021). Our results provide support for this explanation, with indirect effects through parents’ own depressive symptoms.

The PAAS program was designed to mitigate the negative effects of discrimination on adolescent behavioral health through a scalable eHealth delivery format. PAAS, and its predecessor the SAAF program (Murry et al., 2007, 2011), are relatively unique among family-centered interventions for adolescents from racial/ethnic minority backgrounds in that they have an explicit focus on discrimination, racial socialization, and racial pride (for noteworthy exceptions, see the work of Stevenson and Colleagues: Anderson et al., 2019; Coard et al., 2004; Stevenson, 2003). This approach is consistent with Barrera and Castro’s (2006) heuristic adaptation model, which promotes the consideration of unique adaptive elements to address stressors specific to a population of interest. For over a century, it has been noted that societal and interpersonal messages about an individual—and by extension their community—can mold self-image (Cooley, 1902). African American adolescents are regularly forced to confront systemic and personal acts of racism that portray their community as “less than.” Parents’ use of racial socialization to counter these negative messages with positive messages promotes racial pride and supports self-healing (Anderson & Stevenson, 2019; Hughes & Chen, 1999; Peters, 1985). This study builds on this understanding of racial socialization by supplementing these positive messages with clear and consistent rules and expectations about their children’s behavior, in recognition of the fact that the consequences of risky behavior may be disproportionately severe. Results demonstrated that the combination of racial socialization and rules about risk behavior led to improvements in adolescents’ pride in their community and themselves as members of that community.

This study sought to compare the mechanisms by which the PAAS program might serve a protective function against depression, either via a direct effect of program-induced improvements on racial pride (i.e., as a risk reducer or compensatory factor according to family stress and resilience models) or by moderating the effect of discrimination (i.e., as a protective factor) (Fergus & Zimmerman, 2005; Sandler et al., 1997; Zimmerman, 2013). Results confirmed racial pride as a risk reducer/compensatory factor but not as a protective factor. These results are similar to findings from the SAAF efficacy trial, which showed program-induced improvement in Black Pride as a risk reducer/compensatory factor predicting decreases in a latent construct of psychological functioning, which included depression, hope, and perceived life chances, but not as a moderator (Berkel et al., 2024). However, it is, in contrast to moderation effects, found in other adaptations of SAAF (Brody et al., 2021; Lei et al., 2021) and in the larger literature, showing greater program effectiveness for families at higher levels of baseline risk (Gardner, 2023; Pelham et al., 2017; Perrino et al., 2014; Tein et al., 2004). We interpret our findings to mean that discrimination is detrimental to adolescent mental health, regardless of their sense of racial pride, but on the flip side, racial pride is beneficial to adolescents, regardless of their experiences with discrimination. The implications of this finding are that programs to support racial pride should be offered universally to African American adolescents. Such programs are also likely beneficial for other groups experiencing discrimination; however, few evidence-based preventive interventions have incorporated discrimination as an explicit focus. A second implication is that effective and scalable interventions are needed to protect both parents and adolescents from toxic exposure to interpersonal and systemic discrimination (Berkel et al., 2024).

Limitations

Interpretation of study results should be made in light of a few limitations. In terms of measurement, analyses did not distinguish the types of discrimination (institutional, personally mediated, or internalized) experienced by parents or adolescents or whether adolescents were aware of the discrimination faced by their parents (Jones, 2000). Moreover, the measure does not capture how participants’ lives were impacted by structural forces that affected material conditions or access to resources or their appraisals of discriminatory experiences related to a sense of dehumanization or loss of self-worth (Jones, 2000), which likely contribute to the effects on adjustment and the use of racial socialization (Anderson & Stevenson, 2019). The use of parent self-report of racial socialization limits the ability to capture implicit messages, such as those transmitted through instinctive responses to racist situations that occur in front of adolescents. Moreover, SAAF and PAAS were developed in rural African American communities with a long history of exposure to slavery and Jim Crow era discrimination. It is unknown whether results would apply to the growing Black immigrant and refugee communities across the United States, which have unique experiences with discrimination.

In addition, the eHealth version of PAAS was developed to address challenges with the implementation of group-based interventions for African American families living in low-income, rural areas. However, due to the need to establish the effectiveness of a technology-based intervention in this population, artificial supports were in place to ensure access to the intervention. Specifically, computer laboratories with technology assistants were set up in community settings and families were able to attend sessions during established timeframes. This approach may have enhanced engagement by increasing access to computers; however, national data on technology access and our own work done in preparation for the trial indicated most families did have some form of access to computers, which has likely grown over time (Murry, Berkel, & Liu, 2018; Pew Research Center: Internet & Technology, 2019). Moreover, this design displaced some of the benefits of an eHealth intervention, specifically, the ability to participate at any place or at any time. Consequently, the effects presented in this study are likely to represent a conservative estimate of the program’s potential to reach rural African American families.

Conclusions

African American adolescents and parents are regularly exposed to discriminatory experiences that lead to disparities in behavioral health outcomes. Theory and formative research suggest that parents’ use of racial socialization strategies may reduce the negative impact of these experiences (Murry, Butler-Barnes, et al., 2018). In accordance with cultural adaptation frameworks (Barrera & Castro, 2006), the PAAS program focuses on both universal and culturally based protective mechanisms (i.e., racial equity-informed parenting and racial pride). Results of this study show the effectiveness of the eHealth version of the program against two levels of intervention controls (traditional small group and literature controls) in promoting racial equity-informed parenting, Black Pride, and reductions in depression. Moreover, it clarifies the specific protective processes at play. Specifically, Black Pride served as a risk reducer/compensatory factor but did not moderate the impact of discrimination on depression. It should be emphasized that exposure to discrimination is by its very nature an unjust element of our society, and investment in evidence-based strategies to prevent exposure to this seemingly intractable social ill is long overdue. In the meantime, results of this randomized trial suggest the implementation of culturally tailored, eHealth preventive interventions that promote parents’ use of racial equity-informed parenting and adolescent racial pride can contribute to more positive mental health outcomes, above and beyond the negative impact of discrimination.