High School Rank in Math and English and the Gender Gap in STEM

Judith Delaney, Paul J. Devereux

Research output: Contribution to journalArticlepeer-review

6 Citations (SciVal)
45 Downloads (Pure)


Using unique data on preference rankings for all high school students who apply for college in Ireland, we investigate whether, conditional on absolute achievement at the end of high school, within school-cohort rank in English and math affects choice of college major. We find that higher rank in math increases the likelihood of choosing STEM and decreases the likelihood of choosing Arts and Social Sciences. Similarly, a higher rank in English leads to an increase in the probability of choosing Arts and Social Sciences and decreases the probability of choosing STEM. The effects of subject ranks on STEM are larger for boys than girls while there is no evidence of a gender difference in the effect of subject ranks on Arts and Social Sciences. We also find that English and math rank can explain about 4% of the gender gap in the choice of STEM as a college major and 9% of the gender gap that is not explained by absolute achievement. Overall, the tendency for girls to be higher ranked in English and lower ranked in math within school-cohorts can explain about 10% of the difference in the STEM gender gap between mixed-sex schools and same-sex schools and about 25% of the difference that is unexplained by absolute achievement. Notably, these effects occur even though we control for an extensive set of measures of absolute achievement at the end of high school, and the institutional setup implies that within-school rank plays no role whatsoever in college admissions decisions.
Original languageEnglish
Article number101969
JournalLabour Economics
Early online date1 Feb 2021
Publication statusPublished - 30 Apr 2021


  • gender gap in STEM
  • high school rank
  • college field of study


Dive into the research topics of 'High School Rank in Math and English and the Gender Gap in STEM'. Together they form a unique fingerprint.

Cite this