Protecting confidentiality in statistical analysis outputs from a virtual data centre
Christine M. O’Keefe, Mark Westcott, Adrien Ickowicz, Maree O’Sullivan, Tim Churches · 2013
In this paper we are concerned with protecting confidentiality in statistical analysis outputs from a virtual data centre. This is an increasingly popular approach in which data are held in a secure environment and are made available to a researcher over a secure internet connection. The researcher has unrestricted access to the data, which is regulated by applicable legislation and researcher agreements. Current systems generally rely on expert manual checking of analysis outputs and/or confidentiality requirements in the applicable legislation and researcher agreements. We believe that a desirable, though potentially interim, solution is to train researchers to conduct the output confidentialisation themselves, while recognising that they will probably not be experts in confidentiality protection methods. Eventually automated systems for output confidentialisation may become available. In this paper we describe a proposal for a two-stage process involving: • Dataset preparation by the data custodian before loading the data into the virtual data centre • Confidentialisation of the analysis outputs by the researcher on removal from the secure environment The second stage makes use of a checklist developed to assist researchers. However, it would be essential to provide researchers with training in disclosure control as part of the virtual data centre researcher and project approval process.