Closing Achievement Gaps at Scale Through Self-Affirmation: First-Year Findings From Two Cohorts

• In Event: Scaling Up a Social-Psychological Intervention to Combat Stereotype Threat and Narrow Achievement Gaps

Abstract

Objectives

Small-scale studies suggest that self-affirmation writing exercises can help reduce the gap in grade point average (GPA) between African-American and Caucasian students by as much as 40% (Cohen et al., 2006). This is the first district-wide effort to implement self-affirmation writing, thus, a key objective is to investigate how the impacts of the exercises may scale-up.

Perspectives

Using an experimental design, we document the extent to which the delivery of the self-affirmation intervention caused the hypothesized equity effects on students’ academic outcomes. This proposal summarizes the intervention’s estimated impacts for the 2011-2012 7th grade cohort on GPA and two achievement tests administered in the fall and spring; outcome data for the 2012-2013 cohort will be available in time to present at AERA in 2014.

Methods

The experiment was conducted during the 2011-2012 and 2012-2013 academic years in the MMSD. During both years, we invited all 7th grade students to participate in the trial, and 61% (N=1,041) consented to participate in 2011-2012 and 67% (N=1,243) consented in 2012-2013. Students were randomly assigned within schools to receive either the self-affirmation exercise or a similar non-affirming exercise in which students considered values that may matter to others but not to them.

We estimated the impact of being assigned to complete the self-affirmation exercise on students’ academic performance (the “intent-to-treat”) using the following multilevel regression model:

Yij= α + β(AFFIRMEDij) + γ(AFFIRMEDij* THREATENEDij)+ ∑ δ Xij + uj + εij

The model includes assignment to self-affirmation (β), an affirmation by threatened-group-status interaction (γ), a set of individual covariates (δ), a school-level random intercept (uj), and a student error term (ϵij).

Data Sources

We relied on district-provided data, which revealed Cohort 1 students were 50% female, 14% limited English proficient, 12% special education, and 37% free/reduced lunch. Thirty-five percent of the analytical sample belonged to non-Asian and non-Caucasian racial/ethnic groups subject to stereotype threat. The affirmation and comparison samples were equivalent at baseline on all background characteristics.

Results

Assignment to the affirmation condition did, indeed, impact academic outcomes for students vulnerable to stereotype threat. The strongest evidence of impacts is in mathematics. Most effect sizes for statistically significant impacts are on the order of d = 0.10. These results, though smaller than those found in prior research (Cohen et al. 2006, Sherman et al. 2013), are consistent with our predictions. Importantly, assignment to the affirmation condition reduced by 50-55% the portion of the residual achievement gap not explained by student demographics and prior achievement.

Scientific Significance

The evidence suggests self-affirmation can narrow racial/ethnic gaps on a large scale. Impacts are small relative to the overall gaps, but large relative to the residual performance gaps unexplained by student characteristics and prior achievement. This unexplained portion of the gap is theorized to be partially attributable to stereotype threat (Cohen & Sherman, 2005, Steele & Aronson, 1995). Given how easily that expressive writing can be integrated into everyday classroom practice, this self-affirmation intervention offers a viable strategy for helping school districts achieve more equitable outcomes.

Authors

• Jeffrey A. Grigg, Johns Hopkins University

• Paul Hanselman, University of Wisconsin - Madison

• Geoffrey D. Borman, University of Wisconsin - Madison