Automated cell suppression to preserve confidentiality of business statistics

Article type: Research Article

Authors: Sande, G.¹

Note: [1] Gordon Sande works with Statistics Canada. Earlier employment was with the World Bank, the University of Waterloo and the University of Chicago. An undergraduate of the University of Alberta, graduate student of Princeton University. Current work is on economic modelling from a systems approach. Recent work related to confidentiality procedures and to edit and imputation methods. Earlier work included performance evaluation, numerical methods, applied statistics and time series methods. Member of the International Statistical Institute, American Statistical Association (Committees on Energy Statistics and Privacy and Confidentiality), Statistical Society of Canada and others.

Abstract: A statistical agency must balance the competing requirements of preserving the confidentiality of the data obtained from respondents and of publishing statistical summaries of the data obtained from respondents. This note will describe the components of an experimental suite of software which seeks to resolve this conflict in the case of economic censuses. The components described will be those which identify sensitive statistics, determine complementary suppressions and audit the suppressed publications. These components are supported by an infrastructure of utility programmes. Experience with the software on various economic censuses and opportunities for further work will be discussed.

DOI: 10.3233/SJU-1984-2104

Journal: Statistical Journal of the United Nations Economic Commission for Europe, vol. 2, no. 1, pp. 33-41, 1984