The Five Safes
To demonstrate trustworthiness, statistics producers must do more than just ensure that robust safeguarding provisions are in place and are followed. There also must be transparency around these safeguarding procedures so they can be scrutinised. To meet this requirement, we expect statistics producers to provide consistent, coherent and accessible information about data safeguarding.
The Five Safes framework developed by ONS is an accessible tool that can be used to help make and communicate decisions about data safeguarding in a consistent and coherent way. It covers five areas.
The Five Safes
Safe data: what steps have been taken to remove items that could identify individuals?
Safe people: do the users have the necessary technical skills to use the data, do they understand the importance of data confidentiality and have they completed all necessary training?
Safe projects: is it an appropriate use of the data, ethical and clearly for the benefit of the public?
Safe settings: where will the data be used and what steps are in place to ensure the data are kept safe?
Safe outputs: what processes are in place to ensure that outputs produced from the data cannot be used to identify individuals?
Proportional to the situation
The steps that will need to be taken to address each of the five areas will vary depending on the type of data and how they are used. For example, ‘safe people’ will include the IT security training provided by organisations to all staff, and the minimum qualifications expected of government analysts, as well as the protocols to approve external researchers for accessing microdata in secure settings, if applicable. ‘Safe settings’ covers many aspects, including buildings and IT system security provided by an organisation for its staff, and the additional security procedures that apply in microdata labs for external users, either in person or via remote access. Organisations will need to use their own judgement to decide what steps need to be taken in each area. The framework provides a way of ensuring that all necessary aspects have been considered.
We want statements about keeping data safe to be as prominent and commonplace as statements about data quality. Using the Five Safes as the template for these statements will help provide some consistency and coherence across statistics producers.Back to top