The benefits of single-sex schooling

Many people claim that single-sex (SS) education is better for students than co-educational (CE) e.g. Jackson (2016). There have been criticisms of this idea e.g. Halpern et a. (2011) but generally it is believed to be beneficial. But what does the evidence suggest? A large-scale meta-analysis by Pahlke et al. (2014), involving 184 studies and 1,663,662 students, compared them on a variety of variables (mathematics performance; mathematics attitudes; science performance; science attitudes; attitudes about school; gender stereotyping; self-concept; interpersonal relations; aggression; victimisation; and body-image) to see if attending a SS school benefited males, females, or both.

They looked at studies involving kindergarten to college level students (4-19 year old students). This wide age range (and the use of moderator analyses) allows the authors to see if there’s a difference in effect for different ages. They also used moderator analyses to tease out effects of socioeconomic status (SES) and dosage of SS instruction (is the whole school SS or is it just one class in a CE school?). But others weren’t e.g. whether the school was public or private. So it’s good they are looking to examine some of the reasons for the possible differences but it’s not great that they ignored factors that may have an effect on student outcomes (though the evidence is contradictory: Dronkers & Robert, 2003; Braun, Jenkins, & Grigg, 2006; Cobbold, 2015).

They overcame the “file-drawer” effect; the tendency for non-significant results to remain unpublished (Rosenthal, 1979), which is a good thing. Some studies were classified as controlled (they controlled for ONE of the possible selection effects) e.g. random selection of participants into CE or SS schooling; SES of parents, controlled for initial performance for target domain; and checked for initial differences between CE or SS. Weighted and unweighted effect sizes were calculated (so they could see the impact that studies with enormous sample sizes had on the results).

I’ve included some of the results below:

For controlled studies on mathematics performance, the effect sizes were very low for both boys and girls in favour of SS (weighted and unweighted). The same for uncontrolled, unweighted studies was also seen (though slightly stronger). However, for uncontrolled and weighted studies there was a moderate positive effect for SS. This result is due to a few samples with very large samples reporting large effect sizes e.g. Jackson (2012). There were some differences in outcomes across age: there was a weak effect size in favour of SS for females in middle school and trivial effect sizes for elementary and high school. There was also a weak effect size in favour of SS for females from middle/upper SES. For boys, there was a small effect size for SS in elementary school, a small effect size for CE in middle school.

For science performance, controlled and weighted studies, controlled and unweighted and uncontrolled and unweighted showed almost no effect sizes for males and females. Only for weighted uncontrolled studies showed a positive effect size for SS.

Across 8 controlled studies, they found that girls in CE schools were more significantly more likely to endorse gender stereotypes (e.g. women are not as good at maths as men). However, there was a large variation seen between weighted and unweighted studies so the authors urge caution when interpreting these results.

For most of the variables, there were too few studies to be able to see what moderating effect SES had on outcomes. It would have been interesting to know what moderating effect (if any) SES had on outcomes between SS and CE schools. Another criticism of the study is that they didn’t examine confidence intervals (though this is almost certainly due to most of the studies it is based on not examining confidence intervals). This also isn’t as big as a problem as it might be because the results indicate there is no real positive effect for SS schooling.

This meta-analysis indicates that the better quality the study, the lower the advantage conferred by attending a SS school. Studies that used at least one control generally found very little to no benefit of SS schooling. A few studies suggested a slight benefit of SS schooling but overall this meta-analysis suggests there is no significant benefit of attending a SS school.

This was further supported by Sohn (2016) who found the benefits of SS schools was very small when you control for teacher- and parental-sorting (but the effect is larger for average-performing students). These findings though were contradicted by Jackson (2016) who made use of 20 low-performing pilot secondary schools in Trinidad and Tobago being converted in SS schools. After controlling for student selection, he found a small increase in grades for academic subjects on national exams. He also found the male cohorts were less likely to have arrests. This suggests there might be some nuance to the benefits of SS education but the gains seem to be quite small even in studies that find a positive impact.

Braun, H.; Jenkins, F.; & Grigg, W. (2006). Comparing Public and Private Schools Using Hierarchical Linear Modeling. National Assessment of Educational Progress, 1-54.
Cobbold, T. (2015). A Review of Academic Studies of Public and Private School Outcomes in Australia. Education Research Brief.

Dronkers, J. & Robert, P. (2003). The Effectiveness of Public and Private Schools from a Comparative Perspective. EUI Research Repository, 1-63.

Halpern, D.F.; Eliot, L.; Bigler, R.S.; Fabes, R.A.; Hanish, L.D.; Hyde, J.; Liben, L.S.; & Martin, C.L. (2011). The Pseudoscience of Single-Sex Schooling. Science, 333 (6050), 1706-1707.

Jackson, C.K. (2012).  Single-sex Schools, School Achievement, and Course Selection: Evidence from Rule-Based Student Assignments in Trinidad & Tobago. Journal of Public Economics, 96 (1-2), 173-187.

Jackson, C.K. (2016). The Effect of Single-Sex Education on Academic Outcomes and Crime: Fresh Evidence from Low-Performing Schools in Trinidad and Tobago. National Bureau of Economic Research, 2222, (DOI): 10.3386/w22222

Pahlke, E.; Shibley Hyde, J. & Allison, C.M. (2014). The Effects of Single-Sex Compared With Coeducational Schooling On Students’ Performance and Attitudes: A Meta-Analysis. Psychological Bulletin, 140 (4), 1042-1072.

Rosenthanl, R. (1979). The file drawer problem and tolerance for null results. Psychological Bulletin, 86, 638-641.

Sohn, H. (2016). Mean and distributional impact of single-sex high schools on students’ cognitive achievement, major choice, and test-taking behavior: Evidence from a random assignment policy in Seoul, Korea. Economics of Education Review, 52, 155-175.