Standardised Test Scores and Educational Achievement in Australia

Abstract

This thesis is comprised of four self-contained papers utilising standardised test score data, specifically data from the Australian National Assessment Program - Literacy and Numeracy (NAPLAN). The analysis presented by this thesis is particularly pertinent given the recent publication of the findings of the Review to Achieve Educational Excellence in Australian Schools, chaired by David Gonski AC. The first paper provides a survey on the uses of standardised test score data for economic analysis, and a discussion of the uses and limitations of the NAPLAN data set. This paper establishes two main areas for analysis. Firstly, the analysis of school funding policy, given the provision of schools’ financial data enabled by the My School website, which will be explored in the second paper of this thesis. Next, the analysis of how student background characteristics may impact achievement, which is analysed further in the third and fourth papers of this thesis. As outlined above, the second paper focuses on the topic of school funding. This paper explores the causal impact of school funding on student achievement in NAPLAN. Using school-average test score data paired with funding information for each school, we determine how the three different types of funding received by Australian schools impact test scores differently depending on sector and state. We find, in general, that funding from the federal government has the least beneficial impact, with state government funding and parent fees more likely to provide the greatest benefit to schools. These results have a significant policy impact, indicating that funding is most beneficial when provided at as local a level as possible. The third paper of this thesis turns to the socio-educational determinants of educational achievement in Australian schoolchildren. We find that students with an Indigenous or language other than English background are at risk of poor performance, as well as students with a parent who did not complete year 12, does not have a university degree or is not employed. Secondly, we find that private schooling makes a student more likely to meet and surpass national benchmarks for achievement, on average. However, the probability of a private school student performing in the higher NAPLAN bands changes based on their other socio-educational features. This thesis concludes with a short fourth paper that provides another perspective to predicting the event of ‘low achievement’ by implementing machine learning strategies. Together, these papers constitute an overview of the possibilities for econometric analysis of the NAPLAN data.