Video Abstract

Video Abstract

Close modal
BACKGROUND AND OBJECTIVES

Academic tracking is a widespread practice, separating students by prior academic performance. Clustering lower performing students together may unintentionally reinforce risky peer social networks, school disengagement, and risky behaviors. If so, mixing lower performing with high performing youth (“untracking”) may be protective, leading to better adolescent health.

METHODS

Advancement via Individual Determination (AVID), a nationally-disseminated college preparatory program, supports placing middle-performing students in rigorous college-preparatory classes alongside high-performing peers. We conducted the first randomized, controlled trial of AVID in the United States, randomizing 270 students within 5 large public high schools to receive AVID (AVID group) versus usual school programming (control group). Participants completed surveys at the transition to high school (end of eighth grade/ beginning of ninth grade) and the end of ninth grade. Intent-to-treat analyses tested whether AVID resulted in healthier social networks (primary outcome), health behaviors, and psychosocial wellbeing.

RESULTS

At follow-up, AVID students had lower odds of using any substance (odds ratio [OR] 0.66, 95% confidence interval [CI] 0.48–0.89) and associating with a substance-using peer (OR 0.74, 95% CI 0.45–0.98), and higher odds of associating with a peer engaged in school (OR 1.73, 95% CI 1.11–2.70). Male AVID students had lower stress and higher self-efficacy, grit, and school engagement than control students (P < .05 for all). No adverse health effects among high-performing peers were observed.

CONCLUSIONS

AVID positively impacts social networks, health behaviors, and psychosocial outcomes suggesting academic untracking may have substantial beneficial spillover effects on adolescent health.

What’s Known on This Subject:

Academic tracking is a widespread practice, separating students by prior academic performance. Clustering lower performing students together may unintentionally reinforce risky peer social networks and health behaviors. However, there are no studies investigating the health effects of academic tracking.

What This Study Adds:

Advancement via Individual Determination is an “untracking” intervention mixing lower performing and high-performing youth in college preparatory courses. In this randomized trial, we found positive impacts of Advancement via Individual Determination on social networks, health behaviors, and psychosocial outcomes suggesting academic untracking may impact adolescent health.

Through near-daily exposure, schools have potential to shape adolescents’ relationships, social norms, and social-emotional skills–1  factors strongly associated with health behaviors.2,3  Educational interventions that alter these factors may have important spillover effects on health. However, this has been rarely studied.

Academic tracking is a widely used strategy that groups students in classrooms according to prior academic performance. Critics argue tracking perpetuates structural disadvantage and racism by limiting access to educational opportunities for students of color and from low-income families.4  Although the educational merits of tracking are vigorously debated,59  no known studies examine its health implications. Academic tracking might directly influence social networks by grouping students together with peers of similar academic performance and engagement.1012  Although potentially beneficial to high-performing students, tracking may reinforce school disengagement and risky health behaviors like substance use, violence, and delinquency among lower-performing students.13,14 

Studies suggest teens tend to form friendships with peers based on similar levels of school engagement and risk behaviors.15  Within schools, this process can be reinforced by placing similar students in the same classrooms.16  This theory is strongly supported by evidence that adolescent health behaviors, including substance use, violence, and delinquency, are closely tied to behaviors and attitudes of individuals in their social network.14,1721  Academic tracking may determine to which peers a student is exposed, further impacting sources of support, transmission of social norms, and access to and opportunities for risky health behaviors, and psychosocial wellbeing.19,20,22,23 

Advancement via Individual Determination (AVID) is a college preparatory program operating in nearly 5000 US high schools across 46 states (about 20% of all public high schools).24  AVID targets students in the academic middle (earning B or C grades, on average) who would not typically be placed in high-performing, college preparatory academic tracks.25  Through AVID, participating students are encouraged to enroll in college preparatory courses and as a result, AVID has been described as an academic untracking intervention.6  Although prior studies examined AVID’s educational outcomes,2631  there are no randomized trials of AVID in the United States and no studies examining its impact on health.

To fill this gap, we conducted the first randomized trial of AVID in the United States to test whether AVID improved adolescent health. We hypothesized that students randomized to AVID would be exposed to more academically successful peers, resulting in more prosocial networks, and would have improved psychosocial wellbeing and reduced risky health behaviors. Because prior studies suggest the influence of schools on social networks and health behaviors may be stronger for boys versus girls, we aimed to test whether intervention effects vary by sex.3234  Finally, we examined whether social connections with AVID students were associated with risky peer networks and behaviors among high-performing students. Our initial study aimed to follow students through 11th grade, however the coronavirus disease 2019 pandemic-related school closures disrupted typical AVID delivery. Hence, we present our findings before COVID-related school closures, following student during their transition to high school, through the end of ninth grade.

We conducted a multisite randomized trial testing whether AVID impacts adolescent health and wellbeing during the transition to high school among students attending schools in predominantly low-income, minority communities (Clinical Trials registration number NCT03059433).

AVID’s secondary school program targets ninth through 12th grade students from demographic groups underrepresented in higher education who are performing in the “academic middle” (earning B or C average grades) and are less likely to be placed in and succeed in college preparatory coursework without additional academic and social support.25  Students enter in ninth grade and are encouraged to remain in AVID through 12th grade. AVID students enroll in rigorous college-preparatory courses, placing them in an academic track typically targeting higher-achieving students, and attend an AVID elective class during which teachers provide academic skills coaching, explain the college application process, and facilitate social-emotional skill development, including persistence in the face of challenges, problem solving, and coping skills. Finally, the program emphasizes the student and teacher relationship and cultivates a family-like atmosphere.27 

We partnered with a large urban school district in Southern California to recruit high schools (serving only grades 9–12) into the study. We invited schools that served low-income minority families, had been certified by the national AVID office as achieving adequate program fidelity, and had more students who meet AVID eligibility criteria than the AVID program had the capacity to serve. We sought to enroll 5 schools to achieve adequate power. Of the first 6 schools approached, 5 agreed to participate. School and participant recruitment took place over 2 consecutive school years (2017–2018). Like the participating school district, participating schools served largely low-income Latinx students (Supplemental Table 2).

At participating schools, AVID recruitment followed the school’s typical practice, including presentations at feeder middle schools and student and parent meetings. Interested students completed an application and interview with the school’s AVID coordinator. Eligibility for AVID participation included: eighth grade, grade point average (GPA) of 2.0 to 3.5, enrolling in ninth grade at a study school, student commitment to taking rigorous college-preparatory courses, and parent permission for AVID participation. AVID-eligible students entered into a random admission lottery, regardless of study participation. Approximately twice as many students as each school had the capacity to serve entered the lottery (330 applicants for 138 AVID slots). Although study schools oversaw identification of students entering the lottery, the investigators conducted the lottery via a random number generator. A separate lottery was conducted for each school. Given the nature of the intervention, blinding was not feasible. Although students typically remain in AVID throughout high school, students were permitted to drop out of AVID over the course of the school year and open slots were filled, as per usual practice, on a first-come-first served basis, regardless of initial lottery result. These practices were agreed upon with each participating school before study initiation according to the tenants of community-based participatory research.35 

All students entering the AVID lottery were eligible for study participation and received a study enrollment packet, including consent forms. Students returning a signed parental consent form and student assent form were enrolled in the study. Study participation had no bearing on AVID lottery results. Of the 2625 students matriculating into a study school, 330 entered the AVID lottery and 270 consented to participate in the study (participation rate = 81.8%). Of those enrolled in the study, 124 “won” the lottery and were offered an AVID program spot (AVID group), whereas the remaining 146 participants were not offered an AVID spot (Control group) (Fig 1).

FIGURE 1

Recruitment and retention consort diagram.

FIGURE 1

Recruitment and retention consort diagram.

Close modal

Finally, we recruited a comparison group of high-performing incoming ninth grade students from the same schools at the same time as the AVID and Control groups. We identified these students (High performing group) by their grade point average of >3.5 during eighth grade. Of the 214 eligible students, 161 consented to participate in the study (participation rate = 75.2%).

Students completed a baseline computer-assisted survey in school at the end of eighth grade or beginning of ninth grade (April–October), and follow-up interview at the end of ninth grade (May–June). At the time of survey administration, students were reminded that the study goal was to learn about schools, social networks, and substance use and all answers would remain confidential. There was no difference in survey completion or retention by study arm. Overall, 418 of the 431 initially enrolled students (117 AVID; 141 control; 160 high performing) completed the follow up survey (retention rate = 97.0%). Of the 13 students lost to follow up, 12 switched schools, and 1 refused participation.

Social Network: we assessed students’ personal social networks using a standard procedure.36  At baseline, students named 20 people (alters) in their network and answered questions describing each person. Alters could include friends or family. At follow up, students named 10 alters outside their family. At each wave, alters identified as “about my age” were considered peers. For each peer, students reported whether that peer is in AVID, has ever been drunk, and ever used marijuana. Participants also reported whether the peer “tries hard in school,” “thinks it’s important to do well in school,” “thinks they should attend every class,” “does not disrupt class,” and “does not cause trouble.” Peers having all of these characteristics were considered highly engaged in school.

Psychosocial wellbeing: at baseline and follow up, students completed the Mental Health Inventory to assess general mental health (range 1–25, α = .80)37 ; the Perceived Stress Scale (range 0–16, α = .61)38 ; General Self-Efficacy Scale (range 8–40, α = .95)39 ; Duckworth Grit Scale (range 13–40, α = .67)40 ; and a 29-item school engagement scale from the High School Survey of Student Engagement (range 29–116, α = .96).41  For these outcomes, higher scores indicate better mental health, more stress, and higher levels of self-efficacy, grit, and school engagement, respectively.

Health risk behaviors: at baseline and follow up, using measures from the Youth Risk Behavior Survey,42  students reported their frequency of alcohol, marijuana, vaping device use, tobacco, or other drug use in the previous 12 months. We created dichotomous measures of any alcohol, any marijuana, any vaping, and any substance use in the prior 30 days and prior 12 months. Using measures from the National Longitudinal Study of Adolescent to Adult Health,43  students reported their engagement in 8 delinquent behaviors in the previous 12 months (painting graffiti, damaging property, lying to a parent or guardian about where they had been or who they were with, stealing, running away from home, driving a car without permission, entering a house or building to steal something, using or threatening to use a weapon to get something from someone, or selling marijuana or other drugs) and if they had been in a physical fight in the last 12 months.

Socio-demographic characteristics and intervention exposure: at baseline, students reported demographic (birthplace, home language, family structure, race and ethnicity) and parental characteristics (educational attainment, employment, birthplace), and whether they participated in AVID during middle school. At follow-up students reported whether they participated in AVID during the fall and spring semesters of ninth grade. Sex and grade point average came from eighth grade academic transcripts.

T-test and χ2 analyses compared demographic characteristics and baseline health behaviors across groups. Intent-to-treat analyses tested whether intervention students had improved outcomes relative to control students. We used multilevel mixed effects models to account for clustering within schools and control for baseline values of the outcome of interest. Psychosocial outcome measures were standardized on follow-up sample values with a mean of 0 and SD of 1. As prespecified, interaction terms tested whether effects were moderated by sex and, when significant, we estimated sex-stratified models. Our prespecified primary outcomes on which the study was powered was the difference in proportion of prosocial peers in the network and difference in 30-day marijuana use between AVID and Control groups at the end of 11th grade. Although we present interim findings from the end of ninth grade, we maintained these primary outcomes. We also examined the odds of naming a prosocial peer in the network and past-12 month substance use. Finally, similar to other studies seeking to measure negative peer influence,13,14,18  we tested whether naming an AVID student in the social network was associated with more risky networks and higher odds of substance use, violence, or delinquency for students in the high-performing comparison group. All outcome data were complete.

The sample is similar to low-income communities in Southern California with 82.8% identifying as Latinx (Table 1) and 78.9% reporting at least 1 parent born outside the United States. Just over half the sample (53.6%) had at least 1 parent who graduated high school and 22.0% participated in AVID during middle school. There were no significant differences in demographics, baseline health behaviors, social network, or psychosocial outcomes between AVID and control arms. However, compared with the AVID group, the high-performing group had significantly fewer males, more Asian students, more students from immigrant families, and lower rates of fighting (12% vs 27%, P = .001) and marijuana use (2% vs 6%, P = .05).

TABLE 1

Baseline Characteristics of Study Participants by Study Arm

AVID Group, N = 124, %Control Group, N = 146, %High Performing Group, N = 161, %AVID Versus Control Difference, PAVID Versus High-Performing, P
Male 48 38 34 .10 .01 
Latinx 88 88 74 .91 .003 
Asian 25 .76 <.001 
Born in US 95 92 86 .36 .01 
English first language 70 60 43 .09 <.001 
≥1 parent born outside US 71 72 91 .86 <.001 
2-parent household 77 75 86 .71 .05 
≥1parent graduated high school 56 60 46 .33 .02 
≥1 full-time working parent 65 73 76 .16 .04 
MS AVID participant 27 22 19 .37 .11 
Mean eighth grade GPA 2.69 (0.57) 2.67 (0.59) 3.74 (0.24) .80 <.001 
Any fighting 27 30 12 .60 .001 
Any delinquency 38 45 31 .23 .23 
Any alcohol use 11 .42 .85 
Any marijuana use .73 .05 
Any vaping device usea 10 .53 .11 
Any substance useb 15 19 12 .41 .48 
AVID Group, N = 124, %Control Group, N = 146, %High Performing Group, N = 161, %AVID Versus Control Difference, PAVID Versus High-Performing, P
Male 48 38 34 .10 .01 
Latinx 88 88 74 .91 .003 
Asian 25 .76 <.001 
Born in US 95 92 86 .36 .01 
English first language 70 60 43 .09 <.001 
≥1 parent born outside US 71 72 91 .86 <.001 
2-parent household 77 75 86 .71 .05 
≥1parent graduated high school 56 60 46 .33 .02 
≥1 full-time working parent 65 73 76 .16 .04 
MS AVID participant 27 22 19 .37 .11 
Mean eighth grade GPA 2.69 (0.57) 2.67 (0.59) 3.74 (0.24) .80 <.001 
Any fighting 27 30 12 .60 .001 
Any delinquency 38 45 31 .23 .23 
Any alcohol use 11 .42 .85 
Any marijuana use .73 .05 
Any vaping device usea 10 .53 .11 
Any substance useb 15 19 12 .41 .48 

MS, middle school.

a

Vaping includes any use of a vaping device, regardless of whether it was used to vape nicotine or cannabis.

b

Any substance use includes any positive response to items assessing any alcohol use, marijuana use, tobacco use, vaping device use, or other drug use in the previous 12 mo.

Of those who won the AVID lottery (AVID group), 81% participated in the program for at least 1 semester and 66% participated for both ninth grade semesters; 5% of the Control group participated in AVID for 1 semester and 5% for both semesters. Lottery “winners” who participated for both semesters were more likely to have a full-time working parent, less likely to report any delinquent behaviors at baseline, and had a higher eighth grade GPA compared with those who participated for just 1 semester (Supplemental Table 3).

Intent-to-treat analyses (Fig 2, Supplemental Table 4) revealed a greater proportion of peers who do not disrupt class (odds ratio [OR] 1.44, 95% confidence interval [CI] 1.11 to 1.86) but no difference in the proportion of substance using peers between AVID and control students (OR 0.95, 95% CI 0.73 to 1.28). However, the AVID Group had lower odds of naming a peer in their social network who has been drunk or used marijuana (OR 0.74, 95% CI 0.56 to 0.98) and higher odds of naming a peer who does not disrupt class (OR 1.23, 95% CI 1.07 to 1.41), was highly engaged in school (OR 1.73, 95% CI 1.11 to 2.70) and who was in AVID (OR 2.19, 95% CI 1.01 to 4.73), compared with the Control group. In addition (Fig 3), although there were no differences in 30-day substance use between groups, the AVID group had lower odds of any substance use (OR 0.66, 95% CI 0.48 to 0.89) and any delinquent behaviors (OR 0.65, 95% CI 0.42 to 0.99) compared with the Control group. These effects did not vary by sex (interaction terms all >.05). Ad hoc analyses tested whether AVID was associated with lower odds of substance use, controlling for having a substance using peer, a highly engaged peer, or a peer in AVID (Supplemental Table 5), to explore whether intervention effects on substance use are explained by social network changes. Results suggest having a peer in AVID may account for some AVID effects on substance use.

FIGURE 2

Intervention effects on the odds of naming a peer in the social network with the following characteristics. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1.

FIGURE 2

Intervention effects on the odds of naming a peer in the social network with the following characteristics. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1.

Close modal
FIGURE 3

Intervention effects on risky health behaviors. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1.

FIGURE 3

Intervention effects on risky health behaviors. All models used intent-to-treat mixed-effects regression models with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1.

Close modal

For psychosocial wellbeing, intervention effects varied by sex (interaction terms <.05) for all outcomes except general mental health, hence we conducted sex-stratified analyses. Although there were no differences between AVID and Control girls, for boys, the AVID group had lower levels of stress (β −.21, 95% CI −0.40 to −0.02) and higher self-efficacy (β .32, 95% CI 0.12 to 0.51), grit (β .28, 95% CI 0.04 to 0.52), and school engagement (β .23, 95% CI 0.05 to 0.41) (Fig 4 and Supplemental Table 6) compared with Control boys. There were no effects on general mental health.

FIGURE 4

Intervention effect on psychosocial wellbeing, stratified by sex. All models used intent-to-treat mixed effects models stratified by sex with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1.

FIGURE 4

Intervention effect on psychosocial wellbeing, stratified by sex. All models used intent-to-treat mixed effects models stratified by sex with a random intercept for school and cluster-robust standard errors to account for clustering within schools, after adjusting for baseline values of the outcome of interest to test whether outcomes for students randomized to AVID differed from those randomized to the control group. Statistical significance is represented by a 95% confidence interval bar that does not cross 1.

Close modal

As treated analyses examined outcomes among students who participated in 1 or 2 semesters of AVID relative to those who did not participate in AVID to check the robustness of our findings (Supplemental Table 7). Results show a similar pattern to intent-to-treat analyses and improved outcomes for those who participated for 2 versus only 1 semester.

Finally, among high-performing students, naming an AVID-involved peer in the social network was associated with increased odds of naming a substance using peer in the network (OR 1.99, 95% CI 1.11 to 3.55 for marijuana use, OR 2.21, 95% CI 1.16 to 4.22 for marijuana or alcohol use), but was not associated with peer school-related behaviors nor self-reported health behaviors (Supplemental Table 8).

We found significant health benefits to adolescents randomized to AVID during the transition to high school, including more prosocial peer networks, decreased substance use, and, for boys, improved psychosocial wellbeing. These findings are notable because not only is this the first experimental study of AVID in the United States, but it demonstrates that academic interventions can have substantial spillover benefits to health.

AVID expands access to rigorous courses for middle-performing students, thereby creating more connections between and among academically middle- and high-performing youth. In addition, AVID simultaneously improves health behaviors. We found that connections with AVID-related peers may partially explain the program’s impact on substance use. Of note, this study focuses on the transition to high school, which may be a sensitive period when social networks are in flux and high-risk behaviors often emerge. Together, findings suggest the social network and health implications of academic tracking may be substantial and are critical to consider, particularly in light of critiques that low-income, Black, and Latinx students may be less likely to gain access to more advanced academic tracks.44  Although not studied here, it is possible that low-performing and less engaged students may also benefit from increased access to rigorous college-preparatory courses. Applying AVID school-wide may be 1 strategy to accomplish this. Identifying the health effects of such an approach to more broadly reduce academic tracking can provide important insights into the public health implications of education tracking policies.

Our findings are consistent with prior research demonstrating educational practices in schools can influence adolescent health behaviors.45  Potential mechanisms include changing adult and peer social networks, school engagement, and noncognitive skills. Traditional school-based substance use prevention strategies rarely target a school’s social and academic environment, despite recognition that school environments are consistently related to substance use.32,4648  The appeal of targeting the larger school environment for substance use prevention is two-fold. First, this approach may be more successful and can simultaneously influence multiple education and health outcomes. Second, this approach does not require carving out instructional time to deliver specific substance use prevention curricula, which is often a barrier to implementation.49 

AVID improved psychosocial wellbeing particularly for males, including grit, self-efficacy, stress, and school engagement, which are associated with multiple positive life and health outcomes.13,5053  The coronavirus disease 2019 pandemic has highlighted the impact of schools on adolescent wellbeing and the potential need for interventions to support these areas during the recovery period.54,55  AVID might be one such tool. It is notable that these effects were largely observed in males. Although we cannot determine the reasons for this finding, it is consistent with studies suggesting that supportive school environments have a greater impact on boys of color, perhaps by buffering the negative impact of traditional school environments.16,32,56 

Although multisite, all study schools were from the same district serving mostly low-income and Latinx students and findings are from only 1 school year. Future analyses are needed to test whether effects are generalizable and sustained over time and identify specific mechanisms through which AVID impacts substance use. Though all study schools met national certification criteria, we did not directly observe or measure AVID implementation and cannot examine whether higher fidelity improves outcomes. We did not examine whether AVID achieves its primary objective of increasing college enrollment and persistence. Although we examine multiple outcomes, they are highly correlated. We rely on self-report substance use measures, though studies suggest high correlation with biologic testing, particularly when measured via computerized surveys employed here.57  Blinding participants to study arm was not possible and we cannot rule out the possibility that AVID participants were more susceptible to social desirability bias.

Despite these limitations, this study has important implications for the role schools play in influencing adolescent health trajectories. First, AVID might be considered an evidence-based program that simultaneously prevents adolescent health risk behaviors and promotes wellbeing. More importantly, many of the strategies AVID uses, including disrupting typical academic tracking practices and providing academic and social-emotional support, could be implemented outside the program. Ensuring schools have the resources and structures necessary to expand access to educational opportunities and facilitate healthy social connections, particularly in marginalized communities, may be key to achieving education and health equity more broadly.

Dr Dudovitz conceived of and oversaw all aspects of the study, including study design, data collection, analysis, interpretation of results, and drafting of the manuscript; Drs Chung, Phillips, Tucker, Pentz, Tseng, and Wong contributed to the study design, analysis plan, data interpretation, and critical revision of the manuscript; Ms Dosanjh, Ms Galvez, and Ms Arellano contributed to data collection, interpretation of results, and critically reviewed and revised the manuscript; Mr Biely contributed to data analysis, interpretation of results, and critically reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Clinical Trial Identification: Leveraging School Environments to Shape Social Networks and Reduce Adolescent Substance Use—A Pilot Randomized Trial of a Social Networks Intervention. Registration number: ClinicalTrials.gov ID: NCT03059433, https://www.clinicaltrials.gov/ct2/show/NCT03059433?term=dudovitz&draw=2&rank=1. Registered February 16, 2017.

COMPANION PAPER: A companion to this article can be found at www.pediatrics.org/cgi/doi/10.1542/peds.2022-059051.

DATA SHARING STATEMENT: Data dictionaries, study protocols, the statistical analysis plan, and the informed consent form will be made available upon request. In accordance with our IRB approval from the participating school district, de-identified individual participant data will only be made available with approval from the school district. The data will be made available upon publication to researchers who provide a methodologically sound proposal for use in achieving the goals of the approved proposal and whose request is submitted and approved by the participating school district’s Committee for External Research Review. Proposals should be submitted to the corresponding author at rdudovitz@mednet.ucla.edu.

FUNDING: This study was funded by the Robert Wood Johnson Foundation (E4A 74086) and National Institutes of Health (1K23DA040733-01A1). The funders played no role in the study nor preparation of the manuscript.

CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no conflicts of interest to disclose.

AVID

Advancement via Individual Determination

GPA

grade point average

OR

odds ratio

95% CI

95% confidence interval

1
Bonell
C
,
Parry
W
,
Wells
H
, et al
.
The effects of the school environment on student health: a systematic review of multi-level studies
.
Health Place
.
2013
;
21
:
180
191
2
Ellickson
PL
,
Hays
RD
.
On becoming involved with drugs: modeling adolescent drug use over time
.
Health Psychol
.
1992
;
11
(
6
):
377
385
3
Jessor
R
.
Problem-behavior theory, psychosocial development, and adolescent problem drinking
.
Br J Addict
.
1987
;
82
(
4
):
331
342
4
McCardle
T
.
A critical historical examination of tracking as a method for maintaining racial segregation
.
Educational Considerations
.
2020
;
45
(
2
):
1
15
5
Giersch
J
.
Academic tracking, high-stakes tests, and preparing students for college: how inequality persists within schools
.
Educ Policy
.
2018
;
32
(
7
):
907
935
6
Mehan
H
.
Tracking untracking: The consequences of placing low-track students in high-track classes
. In
Race, Ethnicity, and Multiculturalism
.
Oxfordshire, England
:
Routledge
;
2012
:
115
150
7
Yonezawa
S
,
Wells
AS
,
Serna
I
.
Choosing tracks: “freedom of choice” in detracking schools
.
Am Educ Res J
.
2002
;
39
(
1
):
37
67
8
Card
D
,
Giuliano
L
.
Can tracking raise the test scores of high-ability minority students?
Am Econ Rev
.
2016
;
106
(
10
):
2783
2816
9
Kolluri
S
.
Advanced placement: the dual challenge of equal access and effectiveness
.
Rev Educ Res
.
2018
;
88
(
5
):
671
711
10
Valente
TW
.
Network interventions
.
Science
.
2012
;
337
(
6090
):
49
53
11
Ronald
H
,
Price
CL
,
Thomas
SL
.
Tracks as emergent structures: a network analysis of student differentiation in a high school
.
Am J Educ
.
2004
;
110
(
4
):
321
353
12
Legette
K
.
A social-cognitive perspective of the consequences of curricular tracking on youth outcomes
.
Educ Psychol Rev
.
2020
;
32
(
3
):
885
900
13
Wong
MD
,
Strom
D
,
Guerrero
LR
, et al
.
The role of social-emotional and social network factors in the relationship between academic achievement and risky behaviors
.
Acad Pediatr
.
2017
;
17
(
6
):
633
641
14
Fujimoto
K
,
Valente
TW
.
Decomposing the components of friendship and friends’ influence on adolescent drinking and smoking
.
J Adolesc Health
.
2012
;
51
(
2
):
136
143
15
Brechwald
WA
,
Prinstein
MJ
.
Beyond homophily: a decade of advances in understanding peer influence processes
.
J Res Adolesc
.
2011
;
21
(
1
):
166
179
16
Dudovitz
RN
,
Perez-Aguilar
G
,
Kim
G
,
Wong
MD
,
Chung
PJ
.
How urban youth perceive relationships among school environments, social networks, self- concept, and substance use
.
Acad Pediatr
.
2017
;
17
(
2
):
161
167
17
Coronges
K
,
Stacy
AW
,
Valente
TW
.
Social network influences of alcohol and marijuana cognitive associations
.
Addict Behav
.
2011
;
36
(
12
):
1305
1308
18
Tucker
JS
,
de la Haye
K
,
Kennedy
DP
,
Green
HD
,
Pollard
MS
.
Peer influence on marijuana use in different types of friendships
.
Journal Adolesc Health
.
2014
;
54
(
1
):
67
73
19
Lundborg
P
.
Having the wrong friends? Peer effects in adolescent substance use
.
J Health Econ
.
2006
;
25
(
2
):
214
233
20
Zaharakis
N
,
Mason
MJ
,
Mennis
J
, et al
.
School, friends, and substance use: gender differences on the influence of attitudes toward school and close friend networks on cannabis involvement
.
Prev Sci
.
2018
;
19
(
2
):
138
146
21
Daw
J
,
Margolis
R
,
Verdery
AM
.
Siblings, friends, course-mates, club-mates: how adolescent health behavior homophily varies by race, class, gender, and health status
.
Soc Sci Med
.
2015
;
125
:
32
39
22
Osgood
DW
,
Feinberg
ME
,
Wallace
LN
,
Moody
J
.
Friendship group position and substance use
.
Addict Behav
.
2014
;
39
(
5
):
923
933
23
Berkman
LF
,
Glass
T
,
Brissette
I
, %
Seeman
TE
.
From social integration to health: Durkheim in the new millennium
.
Soc Sci Med
.
2000
;
51
(
6
):
843
857
24
Snyder
TD
,
De Brey
C
,
Dillow
SA
.
Digest of Education Statistics 2017, NCES 2018-070
.
Washington, D.C.
:
National Center for Education Statistics
;
2019
25
Swanson
MC
.
Advancement via individual determination: Project AVID
.
Educ Leadersh
.
1989
;
46
(
5
):
63
64
26
Day
SK
.
Comparison of College/Career Readiness Outcomes Between the Advancement Via Individual Determination (AVID) Program and the Traditional High School Academic Program
.
Omaha, NE
:
University of Nebraska at Omaha
;
2012
27
Bernhardt
PE
.
The Advancement Via Individual Determination (AVID) Program: providing cultural capital and college access to low-income students
.
Sch Community J
.
2013
;
23
(
1
):
203
222
28
Cronin
D
.
College Aspirations and Preparation: How AVID Students Beat the Odds
.
St. Paul, MN
:
University of St. Thomas
;
2013
29
Parker
MA
,
Eliot
J
,
Tart
M
.
An exploratory study of the influence of the Advancement Via Individual Determination (AVID) program on African American young men in southeastern North Carolina
.
J Educ Students Placed Risk
.
2013
;
18
(
2
):
153
167
30
Llamas
JD
,
López
SA
,
Quirk
M
.
Listening to students: examining underlying mechanisms that contribute to the AVID program’s effectiveness
.
J Educ Students Placed Risk
.
2014
;
19
(
3-4
):
196
214
31
Kolbe
T
,
Kinsley
P
,
Feldman
RC
,
Goldrick-Rab
S
.
From the (Academic) middle to the top: an evaluation of the AVID/TOPS college access program
.
J Educ Students Placed Risk
.
2018
;
23
(
4
):
304
335
32
Dudovitz
RN
,
Chung
PJ
,
Reber
S
, et al
.
Assessment of exposure to high- performing schools and risk of adolescent substance use: a natural experiment
.
JAMA Pediatr
.
2018
;
172
(
12
):
1135
1144
33
Amialchuk
A
,
Ajilore
O
,
Egan
K
.
The influence of misperceptions about social norms on substance use among school-aged adolescents
.
Health Econ
.
2019
;
28
(
6
):
736
747
34
Poulin
F
,
Pedersen
S
.
Developmental changes in gender composition of friendship networks in adolescent girls and boys
.
Dev Psychol
.
2007
;
43
(
6
):
1484
1496
35
Wells
K
,
Jones
L
.
“Research” in community-partnered, participatory research
.
JAMA
.
2009
;
302
(
3
):
320
321
36
McCarty
C
,
Bernard
HR
,
Killworth
PD
,
Shelley
GA
,
Johnsen
EC
.
Eliciting representative samples of personal networks
.
Soc Networks
.
1997
;
19
(
4
):
303
323
37
Ostroff
JS
,
Woolverton
KS
,
Berry
C
,
Lesko
LM
.
Use of the Mental Health Inventory with adolescents: a secondary analysis of the Rand Health Insurance Study
.
Psychol Assess
.
1996
;
8
(
1
):
105
107
38
Lee
E-H
.
Review of the psychometric evidence of the perceived stress scale
.
Asian Nurs Res (Korean Soc Nurs Sci)
.
2012
;
6
(
4
):
121
127
39
Chen
G
,
Gully
SM
,
Eden
D
.
Validation of a new general self-efficacy scale
.
Organ Res Methods
.
2001
;
4
(
1
):
62
83
40
Duckworth
AL
,
Quinn
PD
.
Development and validation of the short grit scale (grit-s)
.
J Pers Assess
.
2009
;
91
(
2
):
166
174
41
Yazzie-Mintz
E
.
Voices of Students on Engagement: A Report on the 2006 High School Survey of Student Engagement
.
Bloomington, IN
:
Center for Evaluation and Education Policy, Indiana University
;
2007
42
Kann
L
,
McManus
T
,
Harris
WA
, et al
.
Youth risk behavior surveillance - United States, 2017
.
MMWR Surveill Summ
.
2018
;
67
(
8
):
1
114
43
Haynie
L
,
Delinquent Peers Revisited
D
.
Does network structure matter?
Am J Sociol
.
2001
;
106
(
4
):
1013
1057
44
Clotfelter
CT
,
Ladd
HF
,
Clifton
CR
, %
Turaeva
MR
.
School segregation at the classroom level in a southern ‘new destination’ state
.
Race Soc Probl
.
2021
;
13
:
131
160
45
Van Ryzin
MJ
,
Roseth
CJ
.
Enlisting peer cooperation in the service of alcohol use prevention in middle school
.
Child Dev
.
2018
;
89
(
6
):
e459
e467
46
Montgomery
SC
,
Donnelly
M
,
Bhatnagar
P
,
Carlin
A
,
Kee
F
,
Hunter
RF
.
Peer social network processes and adolescent health behaviors: a systematic review
.
Prev Med
.
2020
;
130
:
105900
47
Van Ryzin
MJ
,
Roseth
CJ
.
Peer influence processes as mediators of effects of a middle school substance use prevention program
.
Addict Behav
.
2018
;
85
:
180
185
48
Wong
MD
,
Quartz
KH
,
Saunders
M
, et al
.
Turning vicious cycles into virtuous ones: the potential for schools to improve the life course
.
Pediatrics
.
2022
;
149
(
Suppl 5
):
e2021053509M
49
Salas-Wright
CP
,
AbiNader
MA
, %
Vaughn
MG
, et al
.
Trends in substance use prevention program participation among adolescents in the U.S
.
J Adolesc Health
.
2019
;
65
(
3
):
426
429
50
Guerrero
LR
,
Dudovitz
R
,
Chung
PJ
, %
Dosanjh
KK
,
Wong
MD
.
Grit: a potential protective factor against substance use and other risk behaviors among Latino adolescents
.
Acad Pediatr
.
2016
;
16
(
3
):
275
281
51
Datu
JAD
.
Beyond passion and perseverance: review and future research initiatives on the science of grit
.
Front Psychol
.
2021
;
11
:
545526
52
Henry
KL
,
Knight
KE
,
Thornberry
TP
.
School disengagement as a predictor of dropout, delinquency, and problem substance use during adolescence and early adulthood
.
J Youth Adolesc
.
2012
;
41
(
2
):
156
166
53
Schwinn
TM
,
Schinke
SP
,
Hopkins
J
,
Thom
B
.
Risk and protective factors associated with adolescent girls’ substance use: data from a nationwide Facebook sample
.
Subst Abus
.
2016
;
37
(
4
):
564
570
54
Colao
A
,
Piscitelli
P
,
Pulimeno
M
, %
Colazzo
S
,
Miani
A
,
Giannini
S
.
Rethinking the role of the school after COVID-19
.
Lancet Public Health
.
2020
;
5
(
7
):
e370
55
Lee
J
.
Mental health effects of school closures during COVID-19
.
Lancet Child Adolesc Health
.
2020
;
4
(
6
):
421
56
Campbell
F
,
Conti
G
,
Heckman
JJ
, et al
.
Early childhood investments substantially boost adult health
.
Science
.
2014
;
343
(
6178
):
1478
1485
57
Boykan
R
,
Messina
CR
,
Chateau
G
,
Eliscu
A
,
Tolentino
J
,
Goniewicz
ML
.
Self-reported use of tobacco, e-cigarettes, and marijuana versus urinary biomarkers
.
Pediatrics
.
2019
;
143
(
5
):
e20183531

Supplementary data