Safeguarding privacy in public statistics

The UK's Office for National Statistics had a problem - how to usefully publish public information while protecting anonymity. A team at UWE Bristol provided the answer with an innovative piece of statistics software.

Research that tackles a real-world issue

Businesses, members of the public and governments all need access to statistics to help them make good decisions. But if information is made publicly available, it is important that it does not compromise anyone’s privacy. It can be a challenge to make sure no one can hack public statistics to extract the private data that was used to compile them.

This is a problem of vital importance to statistical agencies around the world, and it is one that research at UWE Bristol has tackled head on. Professor Jim Smith, the research leader, and Dr Martin Serpell, both of UWE Bristol's Department of Computer Science and Creative Technologies, collaborated with Dr Alistair Clark from the Department of Engineering Design and Mathematics.

In 2006, the UK's Office for National Statistics (ONS) approached them to investigate how best to maximise the information made public whilst ensuring appropriate confidentiality. The interdisciplinary team from UWE Bristol brought together mathematical and computing techniques to produce efficient methods that protect large tables of data from hacking.

This has led to a continuing research collaboration with ONS.

Benefits in the UK and beyond

As a result of the research, the ONS has greatly improved the ways that it ensures confidentiality. They can now do this with much larger and more complex data tables than before.

In 2011 and 2012 ONS tested the UWE Bristol findings extensively. So did the developers of software called tau-Argus, the most popular program used internationally to protect confidentiality in statistical disclosure. This confirmed that the UWE Bristol methods were substantially superior to the methods previously used for large, complex data tables.

As a result, ONS decided in 2012 to change their methodologies and take up UWE Bristol's algorithms and tools as part of their standard working practices.

In 2012, the tau-Argus software was modified to make use of UWE Bristol's algorithm. ONS recommends it to UK data providers such as the NHS. More impressively still, it is the standard software used by national statistics agencies around the world.

In the face of the increasing sophistication of hackers and of the greater computer power available to them, UWE Bristol research has helped ensure that respondents to data surveys, whether individuals or organisations, are now better protected against breaches in their confidentiality.

Data providers, such as local and central government agencies, the NHS and businesses, can now also validate their data tables before they make them public, helping them ensure they maintain their duty of trust.

National statistics agencies now have faster, automated tools that remove the need for the manual or ad hoc procedures they often had to rely on until now - especially for large or detailed datasets.

All of this should also mean that data consumers will now have access to a greater range of more detailed information, helping inform better decisions by companies, policy makers and the public - for the benefit of the whole of society.