Cell suppression has been a commonly used method at the Census Bureau and at other agencies for protecting sensitive cells in statistical tables whose cells contain magnitude data. In this method, the sensitivity of each cell depends on the distribution of the respondent values which are summed to form the cell value. Those cells determined to be sensitive are suppressed and then a cell suppression program is run to determine which additional cells (called secondary suppressions) need to be suppressed in order to protect the sensitive ones. In this study, we compare two ways of protecting sensitive cells and their effects on the suppression patterns, i.e., the set of secondary ones. These ways are (1) fixed interval protection and (2) sliding protection. In studies done over a decade ago by researchers at the University of Maryland, it was shown that sliding protection often leads to fewer secondary suppressions than fixed interval protection. Here we show that this result does not hold when the cell suppression program incorporates the following assumption: if v is any respondent value, then any table user knows, from publicly available information, that the value lies in the interval [0, 2*v]. In other words, respondent values are known by interested parties to within 100% of their actual value.
