This document was added 08/11/2017.
NEW LINKING METHOD AND REVISED 2016 DATA
The ABS has reviewed the record linking method used in the National Early Childhood Education and Care Collection following the release of Preschool Education, Australia, 2016. The review has resulted in the development of a new, improved, record linking method which the ABS will use for future Collections.
The ABS has used the new method to compile revised data for 2016 to help users of preschool education statistics understand the effect of the change and enable statistics for 2016 and 2017 to be compared on a similar basis. A number of tables compiled using the revised 2016 data are available in an additional datacube on the ‘Downloads’ tab. Revised data are also available in TableBuilder and the ABS DataLab – refer to Microdata: Preschool Education, Australia, 2016 (cat. no. 4240.0.55.003).
The new linking method enhances the accuracy of preschool education statistics for use by government and the community, and builds on quality improvements made in prior years through the combined efforts of the ABS and data custodians.
Preschool education statistics are compiled by the ABS from unit record level data supplied mainly by the Australian, state and territory education departments. When the component datasets are combined there are multiple records for individual children who are enrolled in more than one preschool program and/or have their information captured by more than one data provider. Records are linked to connect all of a child’s enrolments together and remove duplicates.
The new method links records based on:
- letters from each child’s first name and surname, their sex and their date of birth;
- an additional ‘Match Key’ which data providers use to show multiple records within their dataset that belong to a single child; and
- the state of the child’s place of residence.
In the previous method the comparison across records of child’s place of residence was based on a very small geographic area (the Statistical Area level 1) instead of the state, and all of a child’s records needed to have the same area before any would be linked. These criteria were problematic because the address information in the component datasets can be imprecise or from different points in time. This led to different geographic areas on a child’s records simply because of misspelt/missing addresses or a change in their place of residence.
Using the new linking method it is estimated that 330,676 children aged 4-5 years were enrolled in a preschool program during 2016. This is 14,000 fewer than in the estimate published in February 2017.
The change is lower (around 13,000) for the number of children enrolled in state-specific Year Before Full-time Schooling cohorts.
Most of the changes between the original and revised estimates are in the data for Queensland and Victoria. The education departments of these states supply data to the ABS for both preschools and long day care centres. Long day care centres are also covered by the Australian Government’s Child Care Management System (CCMS).