Secondary Data

In broad terms, secondary data1 research involves an analysis of existing data that were collected during the conduct of a primary research study or that are publically available (e.g. newspaper archives, policy documents). Secondary data research should have the following characteristics:

• no manipulation of human subjects;
• no new data collection;
• no identification of research participants (unless participants have given consent to be identified e.g. if they have criticised the NHS and want credit for it or the data is in the public domain and widely available e.g. Gordon Brown’s speeches).

The key issues in determining the ethical implications of secondary data research concern:

• the potential risk of disclosure of sensitive information/privacy;
• the presumed consent of those who provided the data;
• the use of data for purposes which were not originally intended.

Categories of secondary data research where NO ethical approval is required include research

• involving data that are in the public domain, whether anonymised or not, such as newspaper articles, websites, blogs or television programmes;
• involving data sets held in data archives that are available to the public – for example the ESRC data archive, the Psychological Therapies Research Centre archive or census data;
• where the scope of the consent provided by participants in the primary study included provision for the future use of the data for other research projects (evidence of this must be available is requested);
• Further or re-analysis of a dataset by the original or other researchers.

Ethical scrutiny is required for ALL other secondary data analysis, including where the participant who provided the data in the primary project did not give consent for their data to be used in other research.

Example:
In cases where an agency or organisation can provide data in an anonymised form (for example, a police force supplying data about violent assaults) and where it is not possible or practical to obtain consent from those to whom the data relate (for example, because the data have already been stripped of identifying material or because the number of people to whom the data refer is very large or because contacting victims and offenders of violent assaults could re-stimulate distress), the researchers still need to apply for ethical approval and specify why they are unable to secure the consent of those who provided the original data or those to whom the original data relate.

1 Data is defined here as anything that can be subjected to analysis, including: words, images, numbers, sound, moving images.