Does market-oriented education systems improve performance or increase inequality: A configurational comparative method for understanding (un)intended educational outcomes

摘要

Since the introduction of PISA in 2000, school choice has been featured as a mechanism by which students gain equal educational opportunities and schools are pressured to improve their performance. In opposition, critics have argued that geographic and socioeconomic disparities may cause choice to unintentionally contribute to further educational inequalities, while performance gains are only marginal. To further explore the market rationale behind these claims, this study develops a conceptual framework that theoretically identifies complementary school-level characteristics of market-oriented education systems. These include choice, autonomy and accountability. Both educational outcomes - educational performance and inequality - are then analyzed with regard to these characteristics of market-oriented education systems using fuzzy-set qualitative comparative analysis (fsQCA). This configurational comparative approach further allows for the analysis of complex causality in order to understand how features of market orientation (i.e. choice, autonomy and accountability) also interact with a country's socio-structural context. Therefore, market orientation is modeled along with two well known contributing factors of the outcomes explored, namely social stratification and the institutional stratification of educational systems. Data is derived from the 2009 PISA study and aggregated for 21 OECD countries. The results demonstrate that a strong trade-off between performance and inequality is demonstrated in most countries sampled here. Although many of these countries indeed have market-oriented education systems, few cases contradict this association - the primary difference being the socio-structural context. Theoretically and empirically relevant configurations are identified to explain cross-national variation of educational outcomes; however, these solutions are highly sensitive and possibly subject to model ambiguities.