The effects of system type and characteristics on skills inequalities during upper secondary education: a quasi-cohort analysis of OECD data

ABSTRACT

This article examines the effects of education system types and characteristics on changes in the distributions of literacy and numeracy skills during the upper secondary phase of education and training. Whereas there is a substantial literature on system effects on skills during the primary and lower secondary phases of education, much less has been written about these effects in relation to the upper secondary phase. This article reports on research using quasi-cohort data for 15-year-olds in PISA (Programme for International Student Assessment) and 18- to 20-year-olds in the Survey of Adult Skills to test the effects of a range of system indicators on changes in skills distributions during the upper secondary phase. Consistent with some dominant theories, our difference-in-difference analysis identifies a range of system characteristics associated with skills inequality reduction in upper secondary education and training, which relate to ‘system standardisation’ and ‘parity of esteem’, and which can explain why some ‘system types’ are more effective than others in reducing skills inequality.

KEYWORDS:

• Skills inequality
• literacy
• numeracy
• education systems

Disclosure statement

No potential conflict of interest was reported by the author(s).

Declarations

Funding and support for this project was provided by the Institute for Adult Learning, Singapore University of Social Science. The authors are grateful to the OECD for providing customised data for 18- to 20-year-olds in SAS and to the Leibniz Institute for the Social Sciences for granting access to the German PIAAC survey data via the GESIS Data Archive: ZA5845 – PIAAC, Germany (Reduced Version) – Data file Version 2.2.0, https://doi.org/10.4232/1.12660. The authors have no competing interests to declare that are relevant to the content of this article. The authors would like to thank the anonymous reviewers for their very helpful comments.

Notes

1. Some countries, such as Denmark and the Netherlands, also have substantial element of work-based training as part of their vocational school provision and represent a growing trend towards hybridisation of upper system types (Verdier 2013).

2. Spours, Hodgson and Rogers (2018) note short duration for England and Spain particularly.

3. See Eurostat data at: http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/School_enrolment_and_early_leavers_from_education_and_training

4. Note that this includes the sub-national systems of Flanders in Belgium, and England and Northern Ireland in the UK. Scores for SAS are available separately for English- and French-speaking Canada, whilst corresponding PISA scores are available for Canada as a single national unit. Where possible, sub-national indicators and controls are included for these regions of Canada, whilst where data are not available, national-level measures are included as proxies.

5. The formula for the Gini coefficient is given as: $\mathit{GINI}=\frac{1}{\mathit{\mu N}\left(\mathit{N}-1\right)}\sum _{\mathit{i}>\mathit{j}}\mathit{\hspace{0.17em}}\sum \left|y_i-y_j\right|$, where μ is the average skill level, N is the total number of observations, and y_i and y_j represent individuals’ levels of skill within the distribution.

6. In the US all high school completers are categorised as having achieved general upper secondary education.

7. i.e. classroom contact hours for teachers in upper secondary programmes.

8. The Gini coefficient can be used to assess the relative mean absolute difference – that is, for two randomly selected students the expected difference in skills would be equivalent to twice the value of the Gini coefficient (Atkinson and Bourguignon 2014). A 1 standard deviation increase in the Gini coefficient in our models is equivalent to a 20% (literacy) or 26% (numeracy) increase in the mean expected difference between two randomly selected students.

9. SAS data unavailable to calculate inter-decile differences for Australia.

10. PISA data for Reading for the US are unavailable for 2006 (see main text); here we include 2009 PISA data for comparability.

Additional information

Funding

This work was supported by the Institute for Adult Learning Singapore University of Social Science [MSAPG017].