Stirling vs Philip Morris: A car-crash for social science?

A busy summer means the blog can be overtaken by events. I had intended to write on 11 July about the threat represented by the Information Commissioner’s decision to force University of East Anglia climate researchers to release their raw dataset – but I wrote about journalists and ethics instead. So this is not quite the ‘I told you so’ text that I could have written. However, the car crash that I would have predicted has now happened anyway with the FOI request to the University of Stirling from Philip Morris, a leading cigarette manufacturer, for the raw data from a major survey of teenagers on their responses to tobacco marketing. Although the case has not yet reached a conclusion, the Scottish Information Commissioner has pressed the university not to treat this as a vexatious request. Disclosure has also been supported by figures as varied as Richard Smith, former editor of the BMJ and now a director of United Health Group, and Heather Brooke, the journalist who broke the MP’s expenses story.
Why is this a car crash? Essentially, if Philip Morris succeed, no UK social scientist can promise to maintain the confidentiality of personal data from their informants. This negates the most fundamental principle of social science research, for which US researchers have occasionally risked imprisonment, and jeopardizes our ability both and to win the trust of informants in sensitive environments of all kinds – from bankers to muggers – and to participate in international collaborations, where other countries’ ethical regulators will regard confidentiality as critical.
The Scottish Information Commissioner may need reminding of the duty to protect informants’ privacy, which many of Philip Morris’s supporters seem to regard as trivial. Modern data mining techniques ensure that this cannot be guaranteed simply by removing obviously personal data from the set. Richard Smith is clearly out of touch with this threat, which is well described by Paul Ohm in a recent paper in the UCLA Law Review. Ohm discusses the privacy implications of the rapid advance of de-anonymization methods by computer scientists and suggests that these require a major change in the attitude of regulators. Anything short of fully aggregated results will fail to protect the teenagers who took part in this study, although it is not clear that these would offer any benefit to Philip Morris’s analysts beyond the research team’s published accounts. Assurances from the company that they would not seek to de-anonymize the data are really insufficient: the company is not party to the moral contract between the research team and the teenagers that forms the basis of the data collection. Social researchers do not promise informants to protect data unless and until someone files an FOI request – we promise confidentiality full stop.
Stirling’s resistance was also denounced by Heather Brooke in The Guardian. This is a nice example of the hypocrisy that overcomes journalists when they are dealing with their rivals in social commentary. Ironically, this blog’s delay means that I am writing just as the journalism profession is rallying round the same newspaper’s resistance to police attempts to extract the names of informants who have disclosed information about the phone-hacking scandal. This is seen as a major encroachment on press freedom and on the ability of journalists to conduct inquiries into sensitive social and political issues. I would not disagree – but sauce for the goose is also sauce for the gander, Ms Brooke. The personal data assembled by social scientists, even with the support of public funds, cannot be a public asset in the same sense as, let us say, the results of astronomical observations at one of the UK’s observatories. Raw questionnaires, interview transcripts and field notes need to be placed clearly beyond the scope of FOI legislation, as much as the notebooks of Ms Brooke and her fellow journalists. Whether the research is publicly funded or not, the duty of confidentiality to informants is absolute and cannot be compromised. There are enough regulatory threats to free and robust social research without adding this one.
The social science community should be backing Stirling’s resistance, much as journalists around the world are backing The Guardian. We are rarely as energetic in our own cause, but this is a brave move by Stirling, where many universities would have rolled over in the face of a legal threat, and deserves public endorsement by our representative bodies.