Empirical analysis of sensitivity rules: Cells with frequency exceeding 10 that should be suppressed based on descriptive statistics

Abstract

In official statistics, it is a serious problem to be able to estimate the original data from the numerical values of result tables. To limit such problems, cell suppression is frequently applied when creating result tables, and when researchers create summary tables that include levels of descriptive statistics for remote access, statistical disclosure control must be applied. This research therefore focuses on higher-order moments in descriptive statistics to perform empirical analysis of the safety of statistical levels. The results from standard deviation (variance), skewness, and kurtosis confirm that cells with frequency of 10 or higher are unsafe.