School entrance recommendation: a question of age or development?

Abstract

Fixed cutoff dates regulating school entry create disadvantages for children who are young relative to their classmates. Early and late school enrollment, though, might mitigate these disadvantages. In this paper, we analyze in a first step which factors determine school entry, if entrance screenings allow for early and late enrollment. Second, we study whether children benefit from a delayed school entry. Using data on a compulsory school entrance screening of a German federal state, we show that children with impairments in cognitive, socio-emotional, and motor development as well as health but also young children are less likely to be recommended to start school. Delaying school entry allows the delayed children to improve, although their developmental status remains below average. School entrance screenings, thus, induce more flexible school entry rules that attenuate performance differences within a class and, as a result, mitigate disadvantages for children being young compared with their classmates.

Keywords:

• child development
• school entrance
• school recommendation
• age
• screening

JEL classification::

• J13
• I21
• I38

Acknowledgements

The authors are grateful to Colin Green and the two anonymous referees for comments and suggestions during the revision of the paper as well as to Maresa Sprietsma, Katja Coneus, Andrea Mühlenweg, and Bernd Fitzenberger for helpful comments and discussions at the initial stage. This study uses administrative data from the school entrance examination in the German federal state of Brandenburg. Thanks are due to the health ministry of Brandenburg for the provision of the data, in particular Andreas Böhm and Gabriele Ellsäßer from the federal health office for their generous support and cooperation. All remaining errors are the authors' own.

Notes

1. For the details of the legislation, consult the school law of Brandenburg (Landtag Brandenburg 2006).

2. As shown in Section 3, the vast majority of the children (93–94%) who delay school entry have been recommended to do so by a pediatrician. Thus, only very few parents and school principals deviate from the pediatrician's recommendation by retaining children who received a school recommendation.

3. Our data hint at this as we do observe more children repeating the screening in 2007 than were not recommended in 2006 (Section 4.2).

4. The examinations are based on a core set of indicators of the International Classification of Functioning, Disability and Health developed by the World Health Organization (Simeonsson et al. 2003; Üstün et al. 2003). See Horstschräer and Muehler (2010) for details on the sub-tests and the parental questionnaire.

5. The questions used for diagnosis are displayed in Horstschräer and Muehler (2010).

6. The development of motor functions is important for the overall development of perception and cognition. Although there seems to be no systematic relationship to later school performance (Blomeyer et al. 2009), motor skill problems and poor performance in physical education might put these children at a higher risk of being bullied (Bejerot and Humble 2007; Bejerot, Edgar, and Humble 2011).

7. Researchers may request access to the data by applying at the Landesgesundheitsamt Brandenburg, Dr Gabriele Ellsaesser, Wuensdorfer Platz 3, D-15806 Zossen, Germany. The authors are willing to advise others about access to the data and the application process.

8. As responsibility for education is on the federal level in Germany, no nationwide standardized testing procedures for school entrance screening exist. Moreover, the screenings are administered at the county level, so comparable data and comprehensive reports on federal level are scarce. Data for research are only available for the federal state of Brandenburg, which we used.

9. The pediatricians are given detailed and comprehensive information on all parts of the examination by a handbook. This is constantly updated, and regular quality meetings are held to reinforce the standards set by the ministry of health in Brandenburg. Although most pediatricians commit to these standards, the data show that not all of them deduce their diagnosis from the sub-tests. We therefore consider both, subtests and the pediatrician's judgment, to classify developmental impairments. See Section 2 and Horstschräer and Muehler (2010) for details on the sub-tests.

10. We use the SES classification provided and published by the federal health Office of Brandenburg (Böhm, Ellsäßer, and Lüdecke 2007). Parents are classified into high socio-economic status if both have a high-school diploma (more than 10 years of compulsory school education) and at least one of them is full-time employed. In case of missing values or single parenthood, the characteristics of the remaining parent are also assigned to the second parent. We include dummy variables for single parenthood and missing values in our regression. Results do not change when using education and labor market participation of the mother instead of the predefined socio-economic status.

11. The examination is predominantly taken from January to May; however, there are very few children tested in the months June to December. The control variable for month January thus includes also the months October, November, and December. The variable for month May includes also the months June, July, August, and September. The reference category is March as this is the month where most examinations are taken.

12. The addition of 346 children is caused by parents who hold back their children although they have been recommended.

13. For the full sample of 2007, the recommendation rate is 91.1% (19,535 out of 21,448). For the sample of 2007, the analysis of group means yields similar results as for 2006. Some of the variables have a small amount of missing values; see notes of Table 1. The regressions include dummies for missing values.

14. Estimation results are not shown here but are available from the authors upon request.

15. The marginal effects of the interaction terms are computed using the method introduced by Ai and Norton (2003). Effect sizes and significance levels are similar to estimating all interaction effects together in a linear probability model with robust standard errors.

16. In order to calculate this effect, we need to sum the age effect for children without impairments (1.6 percentage points) and the age effect for children with an impairment as opposed to children without any developmental disorders (2.2 percentage points).

17. As a robustness check, we estimated the equations on a sample of 18,551 children, excluding early and late attendees. These estimations yield similar results for all covariates, e.g. the age effect is between 1.7 and 1.9 percentage points depending on the specification.

18. The relationship also pertains for the other developmental indicators (socio-emotional development, motor skills, and health).

19. Most sub-tests also indicate a higher share of impairments for young children. For example, young children in particular show ADHD behavior (Figure A1 in the appendix). However, this should not be misinterpreted as ‘misdiagnosing’ young children with ADHD as the questions inducing an ADHD impairment are quite rudimentary and, thus, only constitute an initial suspicion (see ADHD questions displayed in Horstschräer and Muehler (2010)).

20. For cognitive impairments, the 2006 gap of 39 percentage points is calculated as the difference between nonrecommended children, of whom 50% show cognitive impairments, and recommended children, of whom 11% show cognitive impairments. For 2007, it is the gap between repeaters (who are most likely the nonrecommended children of 2006 examined 1 year later), of whom now 33% show cognitive impairments, and recommended children in 2007, of whom 13% show cognitive impairments (Table 3).