ABSTRACT
Students born in the days before the school-entry cutoff date are one year younger at school entry than their counterparts born in the days following the cutoff, which causes an achievement gap between the youngest and oldest students within each cohort. I assess this so-called relative age effect in the Spanish context, where the school-entry rule is strictly enforced. By means of a sharp Regression Discontinuity Design, I report a remarkable effect in fourth grade that is still visible in eighth grade. The heterogeneity of the effect by social origin and gender is explored. Following the compensatory advantage model, I reason that the relative age effect should be smaller among high-SES students when assessing academic outcomes that might endanger the probability of attaining higher education (such as repetition), but not for low-stakes external examinations. I confirm this expectation. As for gender, I argue that the relative age effect might be lower among girls either because younger girls are not ever affected by their birth month or because they are better able to catch up with their older peers than boys. I provide evidence supporting this second hypothesis.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 This form of academic differentiation in an otherwise highly comprehensive educational system has been reported to significantly affect the formation of educational expectations (Valdés, Citation2021).
2 According to the Spanish Ministry of Education, the percentage of 10-year-old students that had ever repeated a grade was 11.4% in the academic year 2010-2011, and 31.1% for of 15-year-old students. Although parents participate in the decision to hold back a student during primary education, the final word rests with the teacher staff.
3 According to the Spanish Ministry of Education, around 7% of 8-year-old students have already repeated a grade.
4 Both datasets are freely available to the public on the website of the Spanish Institute for Educational Assessment (INEE, https://www.educacionyfp.gob.es/inee/portada.html).
5 I might employ the direct question about repetition included in the survey, but I consider these responses less reliable than the birthyear. In any case, the differences between both operationalizations are minimal. In fourth grade, 398 students responded to the direct question contradictorily with their year of birth, which represents 1.69% of the sample. In eighth grade, those numbers drop to 238 students and 0.98%. Results hold completely for this alternative operationalization.
6 Cultural wealth is measured using information on the number of books at home, while economic wealth is measured using information such as the possession of cars or a washing machine, or the number of bathrooms in the household.
7 Note the term local in the context of a sharp RDD model means that the estimate is valid a priori only for those individuals in the vicinity of the threshold, which diverges from what is traditionally referred to as a Local Average Treatment Effect (LATE) in the context of IV and fuzzy RDD designs that pertains to the population of compliers with the treatment.
8 Boys outperform girls in mathematics scores by 0.127σ, while girls outperform boys in linguistic scores by 0.122σ.