Remote Access in Action – Design and Implementation of a Demonstration Remote Access System

Remote Access is a relatively new approach to balancing access and use of data with privacy and confidentiality protection. Its novelty lies in the fact that it is implemented as a confidentialisation process on analysis outputs, rather than on inputs as in traditional statistical disclosure control methods. This paper describes the design of a fully automated remote access system in which queries are restricted via a menu-driven web-based interface, analyses are run on unconfidentialised data, and outputs are modified or filtered to protect confidentiality. The paper discusses the implementation of the methods in the CSIRO demonstration software Privacy-Preserving Analytics (PPA). To illustrate the effect of the restrictions and modification filters on common statistical analyses, the paper provides examples enabling a side-by-side comparison of traditional analysis outputs with outputs of the PPA remote access system for exploratory data analysis and survival analysis. An example of PPA system output for linear regression is also provided, without corresponding traditional output due to space limitations. The paper gives an overview of the methodological challenges which arguably need to be addressed before data custodians can have confidence that remote access adequately addresses the balance between data use and confidentiality in real-world situations.