Van Ewijk, Reyn, & Sleegers, Peter
School Effectiveness and Achievement: A Meta-Analysis into the Compositional Effect
University of Amsterdam
Meta-analysis on the effects of ethnic minority share in school on achievement test scores.
Journal Name or Institutional Affiliation:
School Effectiveness and School Improvement
Vol. 21, Number 3, pp. 237-265
- Studies find average effect size of -0.18, indicates that increasing the share of students from an ethnic minority by 10 percentage points is associated with a decrease in individual student's test scores by 0.018 standard deviation.
- Increasing the Afro American or indigenous students share by 10 percentage points is associated with a decrease in test scores of students from those same minorities by about 0.027 standard deviation.
- Correcting for average SES affects the size of the ethnicity compositional effect to a much lesser extent
- Effects from Afro American and indigenous students seem equal to each other and appear larger than effects of immigrants.
- Compositional effects on test scores in the existing literature are generally not very large but that there is some important variation.
- Effects related to the share of African Americans (and to share of indigenous students) seems considerably stronger that effects related to share of immigrants.
- Increasing the ethnic minority share seems to have a stronger effect on students from that same ethnic minorty group than on students belonging to the ethnic majority or on other minority groups.
Journal Article Empirical Research
Academic Achievement, Ethnicity, Racial Composition, Segregation
Method of Analysis:
Studies about effect of ethnic minority share
Unit of Analysis:
- Inclusion criteria:
- (1) Report gives estimates of the effect of an increase in the proportion of ethnic minority students
- (2) the ethnic minority in question scores lower on achievement tests than the country's ethnic majority
- (3) the dependent variable is individual students' scores on tests of math, language, science, or general academic achievement
- (4) the estimation (regression) model used in the study includes individual ethnicity as a covariate
- (5) The students in the sample are in primary or secondary school
- (6) The study is published or presented no earlier than January 1996 and before January 2006
- (7) The study is written in English
- (8) The study uses students' present ("level") test scores, not gain scores, as the dependent variable in its model.
- Final database included 177 estimates from 13 studies
- Used a fixed-effects meta regression analysis.
- DV: Coefficient of the effect of ethnicity on math achievement
- IV: Effect is of immigrant share, effect is of indigenous share, effect is on entire student population, no individual prior achievement covariate included , model included covariate for average SES, etc.