Secondary analysis
You need to consider what kind of consent – or whose permission – you need to get to conduct secondary research using data such as documentary records, case notes, archives and other data.
It is sometimes assumed that secondary analysis raises few (or no) ethical considerations, but that view is rarely taken by ethics committees, and so it is worth spending some time thinking about the specific ethical considerations of your planned research. Those considerations will depend on the nature of the data set – whether you are planning to use quantitative data sets (such as cohort study data), to conduct historical and archival research, or to draw on documentary analysis of information such as school reports, hospital records, court reports, and so on.
The following questions are a guide to prompt your thinking, not a comprehensive list:
- In secondary analysis the people whose details are recorded in the data are not directly approached for their consent, but there are still questions to consider. For example, would they object to how you are using and interpreting their data? Should they be, or can they be, contacted and asked for their consent?
- What are the implications of the Data Protection Act (1998) for your planned use of the data? Specifically, you should check what kind of consent was secured from participants when the data were collected. Under the terms of the Data Protection Act, they should have been asked if they were willing for their data to be archived and made available for further research.
- Are the original participants individually identifiable or recognisable in any way? Do you have a responsibility to protect their anonymity or confidentiality? If you won’t, can you justify that to an ethics committee?
- What are your responsibilities to the researchers who collected the original data? Do you need their permission to use the data set? Do you need to inform them of what you plan to do?
Even when data have been anonymised for secondary analysis, there is still a risk that participants could become identifiable. One obvious example would be when participants are distinctive in some way - see our section on Professional and Elite Interviews. But people can become identifiable even within large scale data sets - perhaps because they have distinctive characteristics (e.g. families with large numbers of children may stand out in cohort studies) or because a method of analysis combines variables in ways that identify small groups within a larger sample.
For example, imagine a group of researchers are analysing local authority school survey data to identify areas where there are high rates of young people who are sexually active under the age of 16. They use age, gender and postcode analysis, and analyses identify a street with three young men aged 15, one of whom reported being sexually active. Does this analysis respect the assurances of anonymity given to participants? What should the researchers do?
You may need to limit the analyses that you do to ensure that promises of anonymity are maintained. If you are applying for ethics approval for a study based on secondary analysis, the ethics committee may be concerned about the potential for anonymised data to become identifiable, and you should address this in your application.
