Affiliations: [a] National Institute for Applied Statistics Research Australia, University of Wollongong, New South Wales, Australia | [b] Australian Bureau of Statistics, Australia | [c] Australian National University, Canberra, Australia
Correspondence:
[*]
Corresponding author: James Chipperfield, National Institute for Applied Statistics Research Australia, University of Wollongong, New South Wales, Australia. Tel.: +61 2 6252 7301; E-mail:[email protected]
Abstract: Important questions can be answered using frequency tables. However, when
publishing or releasing frequency tables, many data custodians are legally
required to ensure that the risk of disclosing information about a person or
organisation is acceptably low. A relatively recent way of releasing
frequency tables is via a remote server. A remote server automatically
responds to queries on the micro-data that are submitted by analysts, but it
does not allow analysts to view the underlying micro-data. The challenge is
to develop an automatic response to queries that has an acceptable
disclosure risk \textitand releases tables that are useful to analysts. This paper
describes the methodology underlying the Australian Bureau of Statistic'
remote server, called TableBuilder, for releasing frequency tables. We
introduce a framework for measuring the disclosure risk of allowing analysts
access to micro-data via TableBuilder. We apply this framework to meaure the
disclosure risk of micro-data created by linking the Australian Population
Census to administrative migration data, supplied by the Australian
Department of Immigration and Customs. We conclude that a remote server
provides a very effective balance between disclosure risk and utility.