Effect of relative age in the first grade of primary school on long‐term scholastic results: international comparative evidence using PISA 2003

Abstract

In this paper, we estimate the effect of pupil’s relative age within the first grade of primary school on mathematics and reading test scores at age 15. The main objective is to evaluate the long‐term causal effect of relative age in the first grades of primary school on pupil’s test in 16 different countries. We use the national rule for admission to primary school to construct the predicted relative age of each pupil. We find that relative age at the start of primary school has a significant positive effect on test scores in about one‐half of the considered countries and regions. Moreover, we identify some of the channels through which the effect occurs.

Keywords:

• pupil performance
• relative age
• international comparison

Acknowledgements

The author is grateful for the financial support from the Belgian French Community’s programme ‘Action de Recherches Concertée’ (ARC 03/08‐302). Thanks are due to Hessel Oosterbeek for inspiring discussion and ideas. In addition, the author is particularly grateful to Andreas Ammermueller, Vincent Vandenberghe and the anonymous referee for helpful comments and suggestions.

Notes

1. See Bedard and Dhuey (2006) for the model of human capital accumulation that corresponds to the intuition developed here.

2. Tracking refers to pupils being selected into different tracks at high school. Often there will be, for instance, a general versus a vocational track.

3. The number of births decreased by 25% in 1966 due to a superstition about births in the 1966 ‘Fire‐horse’ year.

4. These data contain the birth month and maternal socio‐economic information for all US births.

5. These occur mainly for the variable ‘age at start of primary school’ that we use to compute EARLYSTART and LATESTART, and for the variable RETENTION.

6. ‘The Information Network on Education in Europe’ (European Commission 2006). We also check our dates are the same as those used by Bedard and Dhuey (2006) for countries present in both our studies.

7. Only two regions have been excluded, in Canada: Québec and Nova Scotia.

8. Current Serbia and Montenegro. We refer to the country by former Yugoslavia throughout the paper, as this was its name at the time of the survey.

9. We tested this hypothesis.

10. It is possible to include both grade dummies and RETENTION because RETENTION is a dummy variable equal to one if the pupil repeated one or more grades; it does not specify the number of grade repetitions.

11. Owing to the low number of observations of pupils in a higher grade in most countries or regions, we cannot perform a Probit estimation here.