10 is the safest number that there’s ever been

Ritchie, Felix

Abstract

When checking frequency and magnitude tables for disclosure risk, the cell threshold (the minimum number of observations in each cell) is the crucial statistic. In rules-based environments, this is a hard limit on what can or can’t be published. In principles-based environments, this is less important but has an impact on the operational effectiveness of statistical disclosure control (SDC) processes.
Determining the appropriate threshold is an unsolved problem. Ten is a popular number for both national statistics institute (NSI) outputs and research outputs, five and twenty less so. Some organisations use multiple thresholds for different data sources.
Unfortunately, these are all entirely subjective. Three is the only threshold which has a solid statistical foundation, but many argue that this leaves little margin for error. There is no equivalent statistical case for any larger number: ten is popular because it is popular
This paper tries to provide some empirical analysis by modelling alternative threshold assumptions on both synthetic data and real datasets. The paper demonstrates that there is no ‘best’ option; moreover, there is no linear relation between a threshold and risk, as higher thresholds can increase disclosure risk in some cases. It also notes that there are disclosure checking practices which can reduce risk irrespective of the threshold.

• Presentation Conference Type: Conference Paper (unpublished)
• Conference Name: Workshop on statistical data confidentiality 2019
• Start Date: Oct 29, 2019
• End Date: Oct 31, 2019
• Deposit Date: Jun 10, 2021
• Publicly Available Date: Jun 11, 2021
• Keywords: confidentiality, privacy, statistical disclosure control
• Public URL: https://uwe-repository.worktribe.com/output/7457485
• Publisher URL: https://unece.org/statistics/events/SDC2019