1352.0.55.060 - Research Paper: Assessment of risk for Unit Record File disclosure - Spontaneous recognition and population modelling (Methodology Advisory Committee), Nov 2003

The ABS makes unidentifiable microdata from its surveys available to users in the form of Confidentialised Unit Record Files (CURFs). To confidentialise these microdata, the ABS aims to protect the data against two specific scenarios: spontaneous recognition and matching to lists.

Spontaneous recognition of an individual occurs when a user, while looking at information on a microdata file, recognises a particular record as possibly corresponding to a particular person that they know of. In order to assess the risk of spontaneous recognition, data items may need to be collapsed or masked in some way, on the basis of how many individuals there are estimated to be in the population in a particular small-dimensional cross-classification.

A method being considered for assessing the risk of individuals from list matching is referred to as population modelling, or more fully, modelling population probabilities based on sample survey data. In particular, it is required that the method to be able to identify whether combinations of data items that are observed for individuals in the sample, are likely to be unique in the population. The paper concludes with a discussion of the applications of population modelling to confidentiality issues.

This page last updated 20 June 2006

	Mobile Survey Participant Information About Us Careers Contact Us
	Australian Bureau of Statistics

Page tools: Print Page Print All RSS Search this Product
About this Release	The ABS makes unidentifiable microdata from its surveys available to users in the form of Confidentialised Unit Record Files (CURFs). To confidentialise these microdata, the ABS aims to protect the data against two specific scenarios: spontaneous recognition and matching to lists. Spontaneous recognition of an individual occurs when a user, while looking at information on a microdata file, recognises a particular record as possibly corresponding to a particular person that they know of. In order to assess the risk of spontaneous recognition, data items may need to be collapsed or masked in some way, on the basis of how many individuals there are estimated to be in the population in a particular small-dimensional cross-classification. A method being considered for assessing the risk of individuals from list matching is referred to as population modelling, or more fully, modelling population probabilities based on sample survey data. In particular, it is required that the method to be able to identify whether combinations of data items that are observed for individuals in the sample, are likely to be unique in the population. The paper concludes with a discussion of the applications of population modelling to confidentiality issues. Document Selection These documents will be presented in a new window. This page last updated 20 June 2006

Australian Bureau of Statistics