Comparing Sensitivity Rules for Protecting Tabular Data

• Activity Number: 423
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Jacob Bournazian*
• Companies: Energy Information Administration
• Keywords: Statistical disclosure control ; sensitivity rules ; tabular data ; p-percent rule ; dominance rule ; interval rule

Abstract

The p percent rule and dominance rules are common sensitivity rules used by statistical organizations to determine whether table cells are at risk for disclosing identifiable information. These rules have limitations that affect the results for identifying sensitive cells. A new approach, named the Interval Rule, attempts to overcome the limitations of these other sensitivity measures and is based on information that a data intruder knows and not necessarily based on the information that the statistical organization knows. This paper applies the Interval rule to volumetric energy data and evaluates its effectiveness for identifying sensitive cells by comparing the rule's performance with results from applying the p percent and dominance rule.