There is a long way to go from normative ethics schools to researchers' practices on ethics. However, these practices are not uniform. They are different across countries, institutions and disciplines. In some places there are institutional safeguard mechanisms where the planned study is examined before a researcher is allowed to engage with a human subject. More familiarly, these are known as institutional review boards (IRB) in the United States or research ethics committees (REC) in the United Kingdom or Finland. However, the rules related to their operations differ a lot.
According to Finnish rules, an ethical review by the research ethics committee is not required in most cases for social scientists. Research requires an ethical review only if any of the points presented in Table 9.1 are ticked. Thus, most of my research is automatically exempt from this kind of review. At the University of California, Berkeley, any research involving human subjects would most likely require an ethical review, no matter the methods of conducting the study or if there are minimal harm and informed consent practices. (However, if the study will clearly have minimal harm, it is processed differently with a light process.) These differences lead us to approach research ethics quite differently.
These cultural differences in ethics play out in real-life research. When we studied differences in social networks, we installed a mobile phone app that tracked to whom participants called and sent text messages (Karikoski and Nelimarkka, 2010). Some might even say that we installed spyware on participantsâ devices. As the focus was on consenting adults and research took place in Finland, no ethical review was required. (No data were collected outside the group of participants, i.e. calling a friend who was not a participant in our study did not leave a data point for us to study.) When presenting our results at a conference taking place in the United States, one of the first results was how the study was approved in the IRB because of privacy concerns. We did explain that an institutional review was not required as the case was exempt according to our guidelines. Rather, we evaluated the study through thinking a lot about the privacy consequences the study had and balanced it against the gains of the study. Based on these considerations, we agreed it followed key principles like informed consent and participants' anonymity. The U.S.-based scholars were astonished by our response, and they might have thought our ethical norms were perverted. However, I would still give a green light for this type of study Participants knew quite well what we planned to do with the data. We discussed with them about the study settings and asked them for explicit permission to install the software. Participation was voluntary (though compensated). There are no obvious red flags that I consider harmful to participants - especially when we anonymised the data for analysis. On the opposite side, I think they were concerned about such an invasive data collection strategy on extremely personal devices. Or maybe they were worried about opportunities for a data leak or someone recognising our study participants, thus breaking anonymity. A third concern could relate to the sensitivity of mobile phone calls and text messages. They could be seen as highly private forms of communication. We did not engage in this debate as we were not able to move beyond the ethics process and protocols. However, I believe examining the concerns in detail would have benefited our discussion and helped everyone to understand why a particular choice was made.
Research ethics are not only contextualised by country, but there are indicators that different disciplines may put their emphasis into different questions. When reflecting on multidisciplinary research ethics in internet research, Buchanan (2017) suggest that these challenges stem from differences in the ways disciplines conduct and approach science. Research on institutional review systems show a divergence of perspectives and recommendations about research ethics in regards to using online data (McCann, 2016; Vitak et al., 2017). Furthermore, as computational social scientists work both in industry and in academia, even the notation of ethics differs. As computational social science can involve novel data, novel methods or novel disciplines participating in the research, challenging discussions about research ethics are foreseeable. For example, a fundamental question guiding how humans are protected against the harms caused by research (i.e. research ethics) is: What constitutes a human subject? While an easy question, it appears that different answers and justifications emerge depending on researchersâ context and disciplinary background.
It is clichÃ© to state that research ethics should not be a checklist but instead focus on discussing these hard questions. (However, based on my experience, discussions on practical research ethics discuss mostly the process and kinds of approvals required, driving the ethical questions towards a checklist. Having worked on many projects, I also have hoped there would be a clear process to externalise ethical considerations without needing to address these difficult questions.) As research ethics is contextual, disciplinary and temporally bounded, it is clear that this chapter cannot provide a list of doâs and dont's. Instead, there seems to be (at least) two ways to help researchers navigate the ethical questions of research.
First is the principles of ethics approach, where researchers draw from guidelines and (international) regulations. According to Menlo (Dittrich et al., 2012), for information and communication research (in the United States) these principles are:
The second approach is the ethics questions approach, which similarly does not provide one-size-fits-all answers to ethical issues. Instead of explicating principles to consider, researchers could use commonly asked questions about ethical practice to examine the research topic. Markham and Buchanan (2012) propose questions such as:
In terms of research practices and ethics, as the section shows, there are many concerns emerging from the need to protect humans from harms related to research. Questions may relate to topics like consent, privacy or benefits from the research activity. However, in computational social sciences, challenges for ethical discussion emerge from its inherent heterogeneity. As we have in previous chapters shown, there are versatile families of methods. When interacting directly with human subjects through experiments, ethical questions may be different than when analysing digital traces left by anonymous humans. Beyond these, as noted above, computational social sciences take place in different disciplinary and societal contexts, which naturally impact how research ethics are perceived.
Therefore, research ethics and protection of human subjects in computational social science require discussion and dialogue. We have already witnessed cases where the discussion has taken place ex post after the study with open ethical questions has been published. This is one way of developing our understanding of research ethics. Alternatively, we could develop more reflective and discussion-oriented practices among our colleagues to navigate these challenges ex ante. Scholars may benefit from a principles-based approach or questions-oriented approach when considering the ethical dimensions of the work and discussing how they may be present in human subject research.