New and Recent Releases
New ABS Small Area Estimates
Users of ABS data often require small geographic regions, where survey sample sizes are often quite low. In response to the increasing user demand for small area estimates, together with practical difficulties in increasing sample sizes, the ABS has been evaluating statistical methods for producing small area estimates and determining their quality.
As part of this evaluation, an empirical study of small area estimates using the 2003 Survey of Disability, Ageing and Carers was undertaken. The culmination of this work is the Small Area Estimates Manual which has been released on the National Statistical Service website www.nss.gov.au
The manual will assist in standardising the methods used for small area data requests and improve the quality of the data. The manual draws on ABS experience with the implementation of small area methods, and over time will incorporate further small area exercises. It is also anticipated that a more technical manual will be produced later this year for those interested in the statistical and methodological issues surrounding small area estimation.
The manual provides a guide to the key issues to consider when first consideration is given to undertaking a small area data exercise. These issues include:
For further information contact Daniel Elazar Ph(02) 6252 6962 or email: firstname.lastname@example.org
A Census Inquiry Service catering for people speaking a language other than English was available from 28 July.
The Census Inquiry Service Language Helpline operated from: 8:30am to 8:00pm (local time), 7 days a week. Ph 1300 363 365
- user’s need for small area data;
- the feasibility of producing data of suitable quality;
- which technique to use to produce sufficiently reliable estimates (for user’s requirements) while being practical and cost effective to implement;
- validating the small area estimates produced and ensuring their quality; and
- communicating the quality of small area estimates to users, including the assumptions underpinning the statistical models used.