Given the rising interest in the potential benefits of single-sex education in the United States, particularly in the STEM fields of science, technology, engineering, and math, Penn researchers Hyunjoon Park, Jere Behrman, and former Ph.D. student Jaesung Choi turned to a similar setting on the other side of the world: Seoul, South Korea, where two-thirds of high school students attend gender separate schools.

“At the high-school level in Seoul, about a third of students attend all-male schools, a third all-female schools, and a third co-ed schools,” says Behrman, the William R. Kenan, Jr. Professor of Economics. “Students are randomly assigned to those schools, which is in sharp contrast to the U.S., where almost all who go to single-sex schools choose to do so.”

In addition, most students in South Korea who want to attend college take a national college entrance exam akin to, but with higher stakes than, the SAT in the U.S., as well as apply to a particular major, says Park, the Korea Foundation Professor of Sociology. Those interested in fields like STEM take a science-specific math exam; all others take one general to math. And the government collects statistics on all of it.

Taking advantage of the unique dataset, the Penn researchers analyzed scores from 248,885 female and 290,627 male students from 11 school districts between the years 2004 and 2011 (minus 2007). The findings intrigued them.

“Boys in all-boys’ schools do better on the general math test than boys in co-ed schools. They’re also more likely to take the science-focused math test,” Park says. But, test scores for girls do not improve in all-girls’ schools. Rather, the outcomes are similar to those seen in co-ed schools.

Moreover, the researchers analyzed student activities after two years in college using data from a nationally representative survey called the Korean Education and Employment Panel. The results held: Attending an all-female high school doesn’t improve the chances of females entering the STEM fields, unlike males, who have higher odds of doing so coming from such a setting.

“Historically, single-sex schools were thought to provide an environment in which females could express themselves without being overridden by aggressive males or worrying about social interaction,” Behrman says. “Although there is this history, there is almost no previous evidence of the effects, just associations that may be due, in part or completely, to who chooses to attend single-sex schools.”

For the effect of these schools on STEM outcomes, this work provides somewhat of a missing link.

The Penn findings are not the first conclusions culled from this data. A 2013 Demography paper from the same three authors showed that for both genders, single-sex schools improved English and Korean test scores and produced more graduates who went on to four-year colleges. Another, from 2015, focused on weight: “Students attending single-sex schools are more likely to be overweight,” they wrote in Social Science & Medicine. “The effects are more pronounced for girls.” 

The researchers surmise that instructor gender may play a role in the gender disparity, particularly related to STEM.

“Boys attending schools with a larger share of male math teachers show better performance in this subject than the boys attending schools with a lower share,” Park says, adding that this is something that happens more in all-male schools that those that are co-ed. “We aren’t sure what actually makes a difference, but it seems like gender matching between students and teachers could be a factor.”

Park, Behrman, and Choi plan to continue research on this topic, focusing next on the long-term implications of single-sex school settings on outcomes like job placement, marriage, and gender identity.