Suppression criteria for inaccurate estimates

Article type: Research Article

Authors: Marker, David A.

Affiliations: 7131 Willow Brook Way, Columbia, MD 21046, USA | Tel.: +1 301 807 2957; E-mail: [email protected]

Correspondence: [*] Corresponding author: 7131 Willow Brook Way, Columbia, MD 21046, USA. Tel.: +1 301 807 2957; E-mail: [email protected].

Abstract: Statistical offices regularly decide at what level of aggregation to publish results of their data collection. These decisions are typically driven by two separate concerns: first, they do not want to publish estimates with large amounts of uncertainty; second, they do not want to provide potentially-identifying information that could disclose an individual person or company. This article focuses on the first concern, when are data so uncertain that an agency should not publish the results? We focus on policies adopted by 16 statistical offices around the world.

Keywords: Relative standard error (RSE), coefficient of variation (CV), minimum sample size

DOI: 10.3233/SJI-210794

Journal: Statistical Journal of the IAOS, vol. 37, no. 4, pp. 1337-1346, 2021