APPENDIX - A.1 Data construction - Teacher data, NAPLAN data, and SEIFA comparisons

Sep 5, 2026

Appendix section A.1 explains how teacher-quality indicators are constructed for analysis across SA3 areas in NSW. Teacher mean ATAR is estimated by linking Census 2016 place-of-work information for secondary teachers with HEIMS ATAR records for domestic university entrants from 2005 onwards, with a noted coverage limitation. It also details school-level NAPLAN standardisation and SA3 aggregation via weighted means, plus sorting-related comparisons using occupation selection criteria and SEIFA-linked patterns. Additional sections discuss prior academic performance, cognitive ability, and illustrative effect sizes on NAPLAN outcomes.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages2

Reading Time5 min

AccessFree to download

A.1 Data construction

A.1.1 Teacher data

A.1.2 NAPLAN data

A.1.3 Sorting and SEIFA data

A.2 Prior academic performance and teacher quality

A.3 Example: the effect of lifting teacher quality

A.4 Additional results

APPENDIX A.1. Data construction A.1.1 Teacher data We focus our analysis on teachers employed in 2016 who attended university between 2005 and 2016. We estimate the mean ATAR of teachers in a given SA3 area by combining Census 2016 data and Higher Education Information Management System (HEIMS) data. The Census 2016 data allows us to identify secondary school teachers using information on their occupation and their place of work (SA2 area). We then link this data to HEIMS data that includes the Australian Tertiary Admission Rank (ATAR) of domestic students who enrolled in an Australian university from 2005 onwards. With this information, we calculate the mean ATAR of secondary school teachers working in each SA3 area in NSW. One limitation of this approach is that we are only able to identify the ATAR of teachers who attended university between 2005 and 2016 because the HEIMS data begins in 2005 and we rely on Census 2016 place of work data. A.1.2 NAPLAN data We use school-level NAPLAN results from the Australian Curriculum, Assessment and Reporting Authority to calculate the mean standardised NAPLAN score for all secondary schools in a given SA3 area. To do so we first calculate standardised NAPLAN scores at the school-grade-domain level by subtracting the average NAPLAN score across NSW for a given gradedomain (e.g. Year 9 Numeracy) and dividing by the standard deviation of that grade-domain’s scores across NSW. This produces a standardised score for each school-grade-domain level score (e.g. 1 = 1 s.d. above the NSW mean for that student grade and test domain). We then aggregate these scores to the SA3 level by taking a weighted mean using the number of students in each grade-domain who sat the test. A.1.3 Sorting and SEIFA data To compare the sorting behaviour of teachers with other occupations we extend our analysis of ATAR scores to include other similar occupations. We select these occupations based on two criteria: • Does the job require a university qualification? • Are workers in this occupation spread fairly evenly across the population? (i.e. they are not clustered in the CBD like bankers or lawyers or in regional areas like mining engineers or geologists) Using these two criteria we select a group of largely health-related occupations for comparison. A.2. Prior academic performance and teacher quality Of all the commonly used measures of teacher quality, teacher cognitive ability and prior academic achievement have been the most consistently related to student outcomes (Hanushek et al. , 2019). However, the results have not been entirely consistent. A review of the literature by Hanushek et al. (2019) suggests that among the early estimates, 37 per cent of studies found positive and statistically significant effects and 27 per cent found positive, but statistically insignificant effects. In comparison, 10 percent found negative and statistically significant effects and 15 percent found negative and statistically insignificant effects. However, they note that most studies in this area have relied on small and idiosyncratic data sets and that more recent, higher-quality work has identified a cleaner link between teacher cognitive ability and student performance (Hanushek et al. , 2019). There is also a precedent for using prior academic performance as a measure of teacher aptitude in Australia. For instance, Leigh and Ryan (2008) use PISA data to examine trends in the aptitude of individuals studying teaching at university. More broadly, it is well understood that ATAR is the single most consistent predictor of university performance (Manny et al., 2019), and in most other occupations, higher ATAR workers earn more on average, suggesting that they are more productive (Dwyer et al., forthcoming). Of course, it is also important to recognise that prior academic performance is far from the only determinant of teacher quality and a teacher’s ability to improve students’ test performance is on

FAQs

How is teachers' mean ATAR estimated for each SA3 area in NSW?

The method links Census 2016 place-of-work data identifying secondary school teachers by SA2 area with HEIMS data containing domestic students’ ATAR at Australian universities from 2005 onwards, then calculates a mean ATAR per SA3 area in NSW.

What is the process for standardising NAPLAN results before aggregating to SA3?

Standardised NAPLAN scores are computed at the school-grade-domain level by subtracting the NSW average for that gradedomain and dividing by the standard deviation across NSW. These scores are then aggregated to SA3 using a weighted mean based on the number of students in each grade-domain who sat the test.

Why is the teacher ATAR estimate limited in time, according to the appendix?

Identification is constrained to teachers who attended university between 2005 and 2016 because HEIMS ATAR data begins in 2005, and the approach relies on Census 2016 place-of-work information.

Sep 5, 2026

Uploaded by Quinn Holloway