Safety in numbers: Minimum thresholding, Maximum bounds and Little White Lies : The case of the mean and standard deviation

Derrick, Ben; Green, Elizabeth; Kember, Kristian; Ritchie, Felix; White, Paul

Abstract

Reporting the sample mean, sample standard deviation and sample size could in some cases lead to the unique identification of the underpinning sample. The likelihood of this reveal via direct enumeration of the possible search space decreases with increasing sample size and increasing domain size and the degree of obfuscation increases with sample size. An R routine, uwedragon, is presented to assist analysts in evaluating the risk of disclosure and to help publish useful information whilst minimising the degree of risk. The identification of unusually large observations (sample maximum) would also be on interest. The use of “reverse marching observations” is used to place bounds on estimated maximum values.

Citation

Derrick, B., Green, E., Kember, K., Ritchie, F., & White, P. (2022, April). Safety in numbers: Minimum thresholding, Maximum bounds and Little White Lies : The case of the mean and standard deviation. Paper presented at Scottish Economic Society, Glasgow

• Presentation Conference Type: Conference Paper (unpublished)
• Conference Name: Scottish Economic Society
• Conference Location: Glasgow
• Start Date: Apr 25, 2022
• End Date: Apr 27, 2022
• Deposit Date: Jun 16, 2022
• Public URL: https://uwe-repository.worktribe.com/output/9644887
• Publisher URL: https://congress-files.s3.amazonaws.com/2022-04/Disclosure%20Control%20Version%201.0.pdf