Statistical disclosure control for Caribbean census tables: A proposal to expand the availability of disaggregated census data

Abstract

Among Caribbean statistical offices, there is widespread awareness of the issue of statistical confidentiality which has often led to quite strong restrictions on the availability of detailed and disaggregated census tables. In North America, Europe, Australia and New Zealand, statistical offices use disclosure control methods, particularly methods involving some form of data perturbation, to safely publish detailed and disaggregated census tables. These methods are not used in the Caribbean (or Latin America) but could safely facilitate the release of much more detailed census tables than has hitherto been possible in many Caribbean countries. This study reviews the problem of statistical disclosure control for census tables and international best practice, focusing particularly on the use of perturbative methods; it carries out comparative analysis and testing of the cell perturbation method and random rounding methods; and recommends that these methods should be made available to statistical offices through the REDATAM software.