Educational Outcomes of Young Indigenous Australians
Type of Publication: Research report
Lead Organisation: Flinders University
Year Published: 2015
Lead Researcher: Stephane Mahuteau
Written by Stephane Mahuteau, Tom Karmel, Kostas Mavromaras and Rong Zhu, National Institute of Labour Studies at Flinders University
Improved educational outcomes are seen as a key lever for addressing the disadvantage faced by Indigenous Australians. Poor educational outcomes have been observed at all levels of education, from early childhood through to tertiary education. While the increase in school retention rates of Indigenous Australians in recent years is encouraging, the more critical issues are whether there have been improvements in educational performance at earlier years for Indigenous students and the extent to which educational performance at say, year 10, is flowing through to education outcomes such as year 12 completion.
By tracking two cohorts from the Longitudinal Survey of Australian Youth (LSAY) – the first aged 15 in 2006 and the second in 2009 – we can look at a number of key issues:
- The size of the gap between the Indigenous and non-Indigenous in education performance at the end of compulsory education, as captured by academic performance at age 15 from the Programme for International Student Assessment (PISA). As well as looking at the size of the gap we can also assess the extent to which it is explained by differences in socio-economic status and other background characteristics.
- Whether there has been any improvement in academic performance at age 15 across the two cohorts among Indigenous students.
- The extent to which educational outcomes for Indigenous students are affected by the final years of schooling, given academic performance at age 15. This is important from a policy perspective by allowing us to disentangle the influence of earlier education to that of the latter years of secondary schooling.
The data used in this project come from the Longitudinal Survey of Australian Youth (LSAY), including both the 2006 and 2009 student cohorts. The first wave of each LSAY is the PISA survey. One of the advantages of PISA scores is that they allow overtime direct comparisons and between-cohort comparisons. The subsequent waves of LSAY allow us to follow the students throughout their compulsory education and beyond. The latest wave of LSAY was released in 2013 for the 2012 wave of both cohorts. The 2009 cohort has (mostly) left school in the last LSAY observation window.
This makes it the first year when a full comparison between the 2006 and 2009 cohorts can happen. The data allow us to conduct a full set of comparisons between Indigenous students and non-Indigenous students in the two cohorts up to their choice of tertiary education.
Our methodology tackles the sequential nature of students’ education pathways by first modelling PISA scores, and then modelling a series of subsequent educational outcomes conditional on PISA, namely:
- School dropout and year 12 completion
- Intention to attend University
- ATAR request
- University participation
- VET participation
The approach we take in modelling PISA is a multi-level one capturing individual background characteristics and school level characteristics, including an estimate of (unobserved) ‘school quality’ (identified through a random coefficient in the model). This approach allows us to decompose the difference in the average PISA score between Indigenous and non-Indigenous students into a component attributed to differences in personal characteristics, a component due to differences in school characteristics, and a component due differences in ‘returns’, that is differences in the coefficients of the characteristic variables. It is the differences in these ‘returns’ that capture the specific disadvantage associated with being Indigenous, over and above socio-economic and other background characteristics. A policy aim would be to reduce the differences in returns to zero, such that the PISA scores for Indigenous students are the same as non-Indigenous students, after controlling for background characteristics.
We take a similar approach to modelling the subsequent educational outcomes, but with the difference that we also condition on academic achievement at 15 (i.e. PISA). An issue here is that PISA itself is an outcome variable (endogenous) and therefore its inclusion can lead to bias in the coefficients. Our approach is to control for this endogeneity by using the expected PISA score rather the observed score.
Educational Outcomes of Young Indigenous Australians – Report