Background: Prior research suggests that stereotype threat explains significant portions of school-based achievement gaps (Steele & Aronson, 1995). To combat stereotype threat, several recent school-based field trials have leveraged self-affirmation theory (Steele & Liu, 1983; Liu & Steele, 1986), wherein affirming one's important beliefs and values can buffer against threats. Cohen et al. (2006) reported that brief writing exercises, which prompted students to reflect on and write about their core beliefs and values, reduced the Black-white grade-point-average (GPA) gap among seventh-grade students by 40%. Sherman et al. (2013) reported similar impacts for Latinx middle-school students. Longitudinal studies, also targeting middle-school students, have revealed enduring impacts for students of color on high school GPA (Borman et al., 2018; Borman et al., 2020), on-time high school graduation rates (Borman et al., 2020), and college enrollment (Goyer et al., 2017). Purpose/Research Questions: This project provides a direct replication of these prior efforts at an unprecedented scale, spanning 11 districts and 33 schools, and answers two research questions (RQs): (1) Does assignment to the self-affirmation intervention impact seventh-grade students' GPAs, attendance, and suspensions? (2) Are these impacts moderated by students' race/ethnicity? Our study was also designed to assess students' standardized achievement and social-psychological outcomes, but due to COVID-19 related school closures we were unable to obtain these data for this cohort. Research Design: This is a multisite randomized controlled trial including three consecutive cohorts of 7th grade students (2019-20, 2020-21, 2021-22). Students were randomly assigned to experimental conditions within school and race/ethnicity blocks; both teachers and students were blind to treatment condition. Setting: Cohort 1 included 33 schools in 11 districts across 8 states. Participants: Cohort 1 included 5,536 students who were randomized to experimental conditions (2,783 to treatment and 2,753 to control). The sample is diverse with 57% classified as economically disadvantaged, 29% African American, 21% Latinx, 42% white, 2% Asian, and 6% other race/ethnicities. Intervention: The intervention consists of three or four 15-minute in-class writing exercises, which prompt grade 7 students to choose two or three personally important values from a list of 11 (e.g., family, music, sports) and write about why they find them important. Control students engage in the same amount of reading and writing and are prompted to focus on the same list of values, but are asked to choose their "least" important values and write about how they may be important to "others." The student-level response rate for exercise 1 was 83%, exercise 2, 70%, exercise 3, which was optional, 89%. No schools were able to implement Exercise 4. Data: Student-level administrative data include demographics (race/ethnicity; free and reduced-price lunch eligibility; special education participation, and ELL status), behavioral measures (suspension counts, attendance rates); and academic performance (overall GPA; GPA in core courses; number of Ds and Fs). Analytic Approach: To estimate the intention-to-treat (ITT) impacts (RQ1), we estimated the following two-level random intercept model (students nested within schools): Y[subscript ij] = [alpha]+[beta](Affirmed[subscript ij])+[sigma][phi]X[subscript ij]+u[subscript j]+[epsilon subscript ij]. In this model, Y[subscript ij] represents the outcomes (e.g., GPA) for students within schools; [alpha] represents the model intercept, [beta] is the coefficient representing the overall impact of self-affirmation for all student groups, and [sigma][phi]X[subscript ij] includes additional covariates (e.g., prior achievement), u[subscript j] is the school-specific error, and [epsilon subscript ij] is the student-specific error term. To evaluate whether students' membership in a potentially stereotype-threatened group (i.e., African American and Latinx students) moderated intervention effects (RQ2), we fit a similar multilevel model but include an interaction term. Y[subscript ij] = [alpha] + [beta](Affirmed[subscript ij]) + [gamma](Threatened[subscript ij]) + [delta](Affirmed[subscript ij]*Threatened[subscript ij]) + [sigma][phi]X[subscript ij]+u[subscript j]+[epsilon subscript ij]. In this model, [delta] represents the interaction between condition (1=affirmation; 0=control) and students' potentially threatened status (1=African American or Latinx; 0=white, Asian, other), which tells us whether potentially threatened students responded to the self-affirmation intervention differently than non-threatened students. We used a Poisson regression to estimate the treatment effects on attendance rate and a zero-inflated negative binomial regression model for the number of Ds and Fs and the number of suspensions to account for both the integer outcome and the prevalence of students in the sample with counts of zero Ds and Fs and suspensions. Findings: Overall, 4,000 students, 26 schools, and 7 districts have provided data thus far. We applied dummy variable adjustment for missing covariates. Missing outcomes were not imputed. Analysis of pre-randomization data demonstrates that the analytic treatment and control group samples are statistically equivalent with respect to demographic, GPA, and behavioral baseline measures (see Table 1). We hypothesized positive impacts of assignment to the self-affirmation intervention for potentially threatened Latinx and African American students and null effects for students from non-threatened white, Asian, and other racial/ethnic groups. As hypothesized, impact analyses indicate that being assigned to receive the self-affirmation intervention did not have statistically significant main effects on academic or behavioral outcomes (see Table 2). The results, unexpectedly, do not reveal moderating effects of treatment for potentially threatened students, and furthermore show unexpected increases in suspensions (d = 0.51, p < 0.05) (see Table 3). Conclusions: Although the results of this study for Cohort 1 did not replicate prior findings of short-term impacts (Borman et al., 2016), there are several explanations. First, the majority of students completed only 2 of the 3-4 exercises. Second, due to school closures and inconsistent policies within and across schools, fourth quarter grades, attendance, and suspension data were not viable outcomes. All outcome measures, instead, rely on only the first three quarters of the 2019-20 school year. Because self-affirmation impacts tend to accumulate with time through positive recursive processes (Cohen et al. 2009), this truncated analysis is likely to underestimate effects. Coding of the student writing exercises suggests that the materials did lead to students affirming core values. Among those who responded to the first exercise, 93.1% in the treatment condition self-affirmed in their responses as compared to 9.6% in the control condition and among those who responded to the second exercise, 95.3% of treatment students self-affirmed compared to 7.1% in the control condition. Future analyses of treatment on the treated effects will help discern whether the intervention shows clearer results for those that affirmed.