Equality of school choice: a study applied to the Spanish region of Aragón

Abstract

The aim of this paper is to test whether the distribution of students by social, cultural and racial characteristics is homogeneous between Spanish public schools (PS) and publicly subsidised private schools (PSPS) or whether segregation exists between the profile of pupils attending each type of school. The theoretical framework is based on the contributions of researchers into school choice policies, while the empirical application uses a 2005 questionnaire answered by the final year secondary school students of the Spanish region of Aragón. We quantify the degree of internal segregation within each sector (PS and PSPS) and estimate a probit model in order to discover which factors determine the choice of a PSPS. We conclude that the distribution of pupils between PS and PSPS follows a clear socioeconomic pattern which favours privately owned schools. Our study offers an additional result, namely, that cream-skimming processes are more recurrent within the publicly subsidised sector, which is shown to be far more selective than the public sector in its distribution of pupils. Finally, it is found that the higher the socioeconomic status, the higher the probability of choosing PSPS, suggesting that the segregation found in this paper may be caused partly by the choice patterns of Spanish families.

Keywords:

• school choice
• publicly subsidised private schools
• segregation
• cream skimming

Acknowledgements

The authors are grateful for the financial support received for project SEC 2000-0581 (Spanish Ministry of Science and Technology), for project P148/2001 (Regional Government of Aragón), for scholarship AP2003-4486 (Spanish Ministry of Education and Science) and for the research group ADETRE (Regional Government of Aragón).

Notes

1. According to data provided by the Spanish Ministry of Education, the distribution of students enrolled in secondary education among different school types in Spain in 2005 was as follows: PS 67%, PSPS 26% and private-independent schools 7%. For a detailed description of the Spanish non-university educational system, and of its historical evolution, see Bernal (2005).

2. Many of the Spanish PSPS require pupils to wear a uniform. In addition, extra-scholar activities are more expensive in PSPS than in PS and PSPS usually ‘invite’ parents to give donations to a private institution (such as a foundation or an association linked to the school). Although these donations are ‘voluntary’, they may discourage less well-off families from enrolling their children in this type of school. Villarroya (2003) estimates that the additional tuition that parents have to pay for their children to attend a PSPS, compared to a PS, was 75,235 pesetas per year (€452.17).

3. Aragón, like all Spanish regions, has had decision capacity for some education issues since 2001. However, the basic education legislation concerning school choice is the same for the whole country. Besides, the distribution of students among different types of schools is quite similar for all the Spanish regions, according to data provided by the Spanish Ministry of Education. For these reasons, the results obtained in this paper about Aragón can be generalised to Spain as a whole.

4. The results of the empirical studies in this field are varied. While Bast and Walberg (2004) summarise various studies which demonstrate beneficial effects of school choice policies upon academic results, others, such as that of Bettinger (2005), do not reveal any significant effect upon academic success.

5. As Levin (1991) explains, both the access to information and its efficient use require a certain experience in the making of choices, in addition to a certain educational level which permits rational decisions to be made.

6. Carroll and Walford (1997) report the results of a qualitative interview study and show that school choice pattern is very complex and multifaceted. They found that there are strong relationships between both SES and educational level and the degree to which families participate in school choice processes.

7. The empirical evidence is not, however, conclusive with regard to the selection procedures of schools. Some studies, such as that of Bryk, Lee, and Holland (1993), demonstrate that private catholic schools are not selective about admissions, while others, like that by Vandenberghe (1996), show that the opposite is true.

8. This questionnaire was given to all students enrolled in the final year of secondary school, that is, it is a universe sample, and the response rates were 80.2, 76.5 and 89.5 for the total sample, PS and PSPS, respectively. Students were helped to answer their questionnaires by an interviewer.

9. Appendix 1 summarises the variables defined on the basis of the items in the questionnaire.

10. There exist other statistics that measure the ‘degree of association’ between two nominal variables such as Phi, Cramer’s V and the contingency coefficient. These measures confirm that there is a significant association between the variables studied. Results are available upon request.

11. The self-confidence measure is actually a mixture of self-confidence and a self-assessment of personal preparedness. For simplicity we refer to it only as self-confidence.

12. Standardisation is carried out because of the sensitivity of the contact index c_ij to the total proportion of students from group j in the sector.

13. We have statistically compared the differences between PS and PSPS variances of p_ki by means of an F-test. In spite of the slight differences between sectors, we have detected statistical differences for all items except for father’s education.

14. In Bernal (2005), three types of electors are identified in the educational market of the city of Zaragoza (Spain): non-electors, moderate electors and demanding electors. Most of the first group belong to the working class and do not choose schools, but send their children to the nearest public school. Most of the moderate electors belong to the middle class and enrol their children to subsidised private schools. Demanding electors send their children to both private and subsidised schools and belong to the upper and middle classes.

15. The estimations we performed using logit models produced very similar results, which are available to any reader requiring them.

16. The selection of the variables to be incorporated into the model is based on several earlier studies of the subject (Kingdon 1996; Bedi and Garg 2000; Figlio and Stone 2001; Lankford and Wyckoff 2001; Epple, Figlio, and Romano 2004).

17. Measurements of discrimination, which permit us to determine to what extent each variable is explained by the indicator, are included in Appendix 2.

18. The variables measuring academic achievement reflect the experiences of students in public or private schools as well as factors that may have contributed to the initial choices of school type by their parents. Thus, we do not consider previous academic achievement as an explanatory variable in the probit model because mixed effects are difficult to separate.

19. Escardíbul and Villarroya (2009) found no statistically significant effect of mother’s education (measured as years of schooling) on choosing a PSPS, whereas the effect of father’s education is positive and highly significant.

20. The results of the probit model including the interaction terms SES*Prestige, SES*Tradition and SES*Proximity are not presented in the paper, but are available upon request from the authors.