Innovation in statistical methodology for predicting growth in academic achievement

Innovation in statistical methodology for predicting growth in academic achievement of school students: Making the most of Australia’s longitudinal administrative education data. 

Chief Investigators

Professor Michele Haynes
Dr Melanie Spallek
Hon. Professor Joy Cumming

Overview

The main aim of this project is to advance statistical methodology for modelling growth in student academic achievement to inform the education priority area of maximising student learning progress by utilising large-scale sources of Australian data. This will be achieved by advancing methodology to:

1. Derive estimators of measurement error and reliability of measurement for longitudinal assessment scores in which the responses to items that contribute to the score are dependent; this will also contribute an extension to item response modelling methodology.

2. Profile achievement trajectories through an extension to the method of multi-channel sequence analysis that allows a range of longitudinal measurement types.

3. Develop statistical methods for modelling longitudinal academic achievement using administrative data for both standardised (NAPLAN) and non-standardised (A-E) assessment scores, incorporating adjustments for missing observations and measurement error in multilevel growth models; this method will investigate the stabilisation of the trajectories of individual A-E scores relative to less frequent NAPLAN scores.

4. Develop a dynamic multilevel growth model that predicts individual academic achievement trajectories with consideration of socioeconomic status, ethnicity, early capabilities and prior achievement of the student, school factors (e.g. geolocation, resources), and changes in circumstances.

5. Establish baseline evidence of individual growth in academic achievement, utilising the methodologies developed and applied to existing data, that will provide the foundation for development of a new education model focussed on maximising individual learning progression.

This project will utilise data from:

• The longitudinal survey of Australian children (LSAC);

• The longitudinal survey of Indigenous children (LSIC);

• The longitudinal studies of Australian youth (LSAY);

• Equivalent international longitudinal surveys;

• Australian census longitudinal data (ACLD).

Duration 1 July 2019 – 30 November 2022

Publications

Goldstein, H., Haynes, M., Leckie, G., Tran P. (2020). Estimating reliability statistics and measurement error variances using instrumental variables with longitudinal data. Longitudinal and Life Course Studies. Published online 20 April 2020. doi.org/10.1332/175795920X15844303873216

Spallek M, Haynes M, Baxter J, Kapelle N. (2020) The value of administrative data for longitudinal social research: a case study investigating income support receipt and relationship separation in Australia. International Journal of Social Research Methodology, 1-15. doi: 10.1080/13645579.2019.1707984

Research Status

Active

Research Partners

Previously Professor Harvey Goldstein, School of Education, University of Bristol

Professor George Leckie, School of Education, University of Bristol

ILSTE Research team members

Professor Michele Haynes
Dr Melanie Spallek
Dr Andrew Smith
Ms Karen Kusuma