Source: Authors’ calculations; see Table A1 for specific
references.
Notes: Symbols plot reported effect sizes for potentially stereotyped
groups (African American and/or Hispanic students) for the first year of the
self-affirmation intervention, and lines represent 95% confidence
intervals (+/− 1.96 standard errors). Shapes represent distinct school
or district contexts. For instance, Sherman et
al. (2013) studies 1 and 2 were conducted in different schools in
different states. Dee (2015) reports
subgroup results from the same sample of Philadelphia-area schools. The dashed
line represents the overall mean effect size (0.07), calculated by weighting
individual estimates according to the inverse of their squared standard error.
The impact estimates are lower in the large-scale replication studies (Dee 2015, Borman et al. 2016, and Current Study), but these differences could
reflect heterogeneous effects across local context, research team, and
implementation. This paper investigates two effects observed within the trial
conducted in a single school district (represented by circles), for which
context and procedures were consistent.