Statistical Disclosure Control
It is critical that everyone complies with policies and procedures for protecting data and safeguarding the identity of individuals.
There is a very small risk of ‘statistical disclosure’, when specific information from a de-identified data collection can be indirectly associated with a particular individual, household or business.
It is possible that a researcher might already know something which can be linked to the data collection, either by:
- identification – associating records with a specific individual, household or business
- attribution – associating particular data values with a specific individual, household or business
For example, if a researcher was looking at a remote area of the UK and there was only one doctor in that area – or only one from a particular ethnic background – there could be a risk of identification.
The risk of this happening is very small, but the Network took it very seriously – and the de-identification process covered more than the removal of names, addresses and other personally identifying information.
Every Administrative Data Research Centre had a set of controls in place to assess and minimise the risk of disclosure when a researcher is using administrative data. Once the researchers finished their data analysis, Network staff cleared the findings as safe before releasing the results. This involved minimising the chances that anyone could be identified from the results.
Only then could the results of the research leave the secure environment. The dataset itself never left the secure environment.