Methods of geographical perturbation for disclosure control

Disclosure control methods are used to protect the confidentiality of individuals and households in aggregate census data. With growth in computational power, the disclosure control problem has been rapidly transformed. Increased analytical power has stimulated user demand for more detailed information for smaller geographic areas and to customized geographical boundaries. However, the possibility of allowing census users to create their own aggregates from census microdata, and for small areas, can lead to problems of disclosure by differencing. Traditionally, methods of statistical disclosure control have been aspatial in nature. This thesis describes a new framework of geographical perturbation methods designed to deal with the spatial nature of disclosure risk. The research offers several new contributions, specifically; . (1) Aframework of new geographical perturbation methods is defined, based on creating uncertainty around geographical location. Zone-independent methods are designed for protection in a flexibletabulation scenario'and to account for the spatial dimension of risk. • (2) Techniques for implementation of t~ese methods are tested on a synthetic census dataset which· show comparable risk-utility outcomes to RRS (an existing method used for the US and UK Censuses). The advantages and disadvantages of the proposed methods are discussed with regard to ease of implementation and flexibility of parameter values. (3) One of these new methods; LOS, is then explored in more detail showing a significant improvement over RRS in terms of the risk-utility outcome. Risk reduction is illustrated in a geographical differencing scenario and distortion to utility explored in a spatial context of typical census users' analyses.EThOS - Electronic Theses Online ServiceGBUnited Kingdo