Abstract
Confidentiality protection of data published in tables is a major problem for statistical offices. To obtain full cooperation of the respondents, it is required that information of individual respondents with a confidential character is kept from being disclosed. One method to avoid disclosure is the method of cell suppression, in which the values of a number of statistical cells are not published but are suppressed from publication. We discuss the method of cell suppression in general two-dimensional tables, in which row totals and column totals are always published. The values of the sensitive cells are replaced by a cross (X). Usually, additional suppressions are necessary to prevent the values of the sensitive cells from being calculated from the row or column totals. Because of these additional suppressions, useful information gets lost. We want to minimize the loss of information by making the best choice for the additional suppressions. Therefore, we introduce and compare the performance of some heuristics for solving this problem.