Secondary data analysis

According to the ESRC Research Ethics Framework (as the BPS does not specifically address this issue), "Ethics review should always be proportionate to the potential risk, whether this involves primary or secondary data" (p. 4) and, indeed, not all types of secondary data require ethical review (1.3.2.6).

ESRC Research Ethics Framework

The ESRC distinguishes three categories of secondary data (1.17.5.2):

1. The first includes data which is not sensitive and where there is minimum risk of disclosure of the identity of individuals.
2. Second is data that is protected by legislation, such as census data and administrative data. Here, the data producer has a strong interest in how researchers will access the data, and may control access to it. This category of data may only be available via 'safe settings'.
3. A third category, such as the National Child Development Study (NCDS), includes data where the inclusion of information such as a birth date or postcode makes disclosure possible, perhaps via a link to other datasets. This means that such data is ethics sensitive.

The PREC requires that researchers intending to analyse data falling into the third category ('3') apply for ethical approval. If the data are not sensitive AND the individuals contributing that data are anonymous (and cannot be made unanonymous through data linkage) OR if access to and use of the data are already managed by an external organisation (that complies with all legal requirements concerning data protection, etc.), the researcher need not apply for ethical approval locally. Of course, such research still must "(adhere) to professional codes of practice, legal requirements and compliance with the Data Protection Act (DPA)" (1.17.3).

Note: Analysis of data garnered on the Internet (e.g., messageboards, chatrooms, weblogs, etc.) is a separate issue.