Guest post by Ilan Kelman, Professor of Disasters and Health, building on his captivating presentation in Session 2 of the UCL Open Science Conference 2024.

Open Science brings risks and opportunities regarding privacy. Making methods, data, analyses, disagreements, and conclusions entirely publicly available demonstrates the scientific process, including its messiness and uncertainties. Showing how much we do not know and how we aim to fill in gaps excites and encourages people about science and scientific careers. It also holds scientists accountable, since any mistakes can be identified and corrected, which is always an essential part of science.

Given these advantages, Open Science offers so much to researchers and to those outside research. It helps to make science accessible to anyone, notably for application, while supporting exchange with those inspired by the work.

People’s right to privacy, as an ethical and legal mandate, must still be maintained. If a situation might worsen by Open Science not respecting privacy, irrespective of it being legal, then care is required to respect those who would want or might deserve privacy. Anonymity and confidentiality are part of research ethics precisely to achieve a balance. Irrespective, Open Science might inadvertently reveal information sources or it could be feasible to identify research participants who would prefer not to be exposed. Being aware of possible pitfalls assists in preventing them.

Disaster decisions

Some research could be seen as violating privacy. Disaster researchers seek to understand who dies in disasters, how, and why, in order to improve safety for everyone and to save lives. The work can examine death certificates and pictures of dead bodies. Publicising all this material could violate the privacy and dignity of those who perished and could augment the grief of those left behind.

Sometimes, research hones in on problematic actions for improving without blaming, whereas society more widely might seek to judge. A handful of studies has examined the blood alcohol level of drivers who died while driving through floodwater, which should never be attempted even when sober (Figure 1). In many cases, the driver was above the legal limit for blood alcohol level. Rather than embarrassing the deceased by naming-and-shaming, it would help everyone to use the data as an impetus to tackle simultaneously the separate and unacceptable decisions to drive drunk, to drive drugged, and to drive through floodwater.

Yet storytelling can be a powerful communication technique to encourage positive behavioural change. If identifying details are used, then it must involve the individuals’ or their kin’s full and informed consent. Even with this consent, it might not be necessary to provide the full details, as a more generic narrative can remain emotional and effective. Opportunities for improving disaster decisions emerge in consensual sharing, so that it avoids violating privacy—while also being careful regarding the real need to publish the specifics of any particular story.

Figure 1: Researching the dangerous behaviour of people driving through floodwater, with the number plate blurred to protect privacy (photo by Ilan Kelman).

Small sample populations

Maintaining confidentiality and anonymity for interviewees can be a struggle where interviewees have comparatively unique experiences or positions and so are easily identifiable. Governments in jurisdictions with smaller populations might employ only a handful of people in the entire country who know about a certain topic. Stating that an interviewee is “A national government worker in Eswatini specialising in international environmental treaties” or “A megacity mayor” could narrow it down to a few people or to one person.

A similar situation arises with groups comprising a small number of people from whom to select interviewees, such as “vehicle business owners in Kiruna, Sweden”, “International NGO CEOs”, or specific elites. Even with thousands of possible interviewees, for instance “university chiefs” or “Olympic athletes”, quotations from the interview or locational details might make it easy to narrow down and single out a specific interviewee.

Interviewee identification can become even simpler when basic data on interviewees, such as sex and age range, are provided, as is standard in research papers. Providing interview data in a public repository is sometimes expected, with the possibility of full transcripts, so that others can examine and use those data. The way someone expresses themselves might make them straightforward to pinpoint within a small group of potential interviewees.

Again, risks and opportunities regarding privacy focus on consent and on necessity of listing details. Everyone including any public figure has some level of a right to privacy (Figure 2). Where consent is not given to waive confidentiality or anonymity, then the research process—including reviewing and publishing academic papers—needs to accept that not all interviewee details or data can or should be shared. With consent, care is still required to ensure that identifying individuals or permitting them to be discovered really adds to the positive impacts from the research.

Figure 2: Ralph Nader, an American politician and activist, still has a right to privacy when not speaking in public (photo by Ilan Kelman).

Caution, care, and balance

With caution and care, always seeking a balance with respect to privacy, any difficulties emerging from Open Science can be prevented. Of especial importance is not sacrificing many of the immense and much-needed gains from Open Science.