6.6.1 Informed consent

Researchers should, can, and do follow the rule: some form of consent for most research.

Informed consent is a foundational idea—some might say a near obsession (Emanuel, Wendler, and Grady 2000; Manson and O’Neill 2007)—in research ethics. The simplest version of research ethics says: “informed consent for everything.” This simple rule, however, is not consistent with existing ethical principles, ethical regulation, or research practice. Instead, researchers should, can, and do follow a more complex rule: “some form of consent for most research.”

First, in order to move beyond overly simplistic ideas about informed consent, I want to tell you more about field experiments to study discrimination (these were covered a bit in Chapter 4 as well). In these studies, fake applicants who have different characteristics—such as a man or women—apply for different jobs. If one type of applicant gets hired at a higher rate, then researchers can conclude that there may be discrimination in the hiring process. For the purposes of this chapter, the most important thing about these experiments is that the participants in these experiments—the employers—never provide consent. In fact, these participants are actively deceived by the fake applications. Yet, field experiments to study discrimination have been performed in at least 117 studies in 17 countries (Riach and Rich 2002; Rich 2014).

Researchers who use field experiments to study discrimination have identified four features of these studies that, collectively, make them ethically permissible: 1) the limited harm to the employers, 2) the great social benefit of having reliable measure of discrimination, 3) the weakness of other methods of measuring discrimination, and 4) the fact that deception does not strongly violate the norms of that setting (Riach and Rich 2004). Each of these conditions is critical, and had any of them not been satisfied, the ethical case would be more challenging. Three of these features can be derived from the ethical principles in the Belmont Report: limited harm (Respect for Persons and Beneficence) and great benefit and weakness of other methods (Beneficence and Justice). The final feature, non-violation of contextual norms, can be derived from the Menlo Report’s Respect for Law and Public Interest. In other words, employment applications are a setting where there is already some expectation of possible deception. Thus, these experiments do not pollute an already pristine ethical landscape.

In addition to this principles-based argument, dozens of IRBs have also concluded that the lack of consent in these studies is consistent with existing rules, in particular Common Rule §46.116, part (d). Finally, US courts have also supported the lack of consent and deception in field experiments to measure discrimination (No. 81-3029. United States Court of Appeals, Seventh Circuit). Thus, the use of field experiments without consent is consistent with existing ethical principles and existing rules (at least the rules in the US). This reasoning has been supported by the broad social research community, dozens of IRBs, and by the US Court of Appeals. Thus, we must reject the simple rule “informed consent for everything.” This is not a rule that researchers follow, nor is it a rule that researchers should follow.

Moving beyond “informed consent for everything” leaves researchers with a difficult question: what forms of consent are needed for what kinds of research? Naturally, there has been substantial prior debate around this question, although most of it is in the context of medical research in the analog age. Summarizing that debate, Eyal (2012) writes:

“The more risky the intervention, the more it is a high-impact or a definitive ‘critical life choice’, the more it is value-laden and controversial, the more private the area of the body that the intervention directly affects, the more conflicted and unsupervised the practitioner, the higher the need for robust informed consent. On other occasions, the need for very robust informed consent, and indeed, for consent of any form, is lesser. On those occasions, high costs may easily override that need.” [internal citations excluded]

An important insight from this prior debate is that informed consent is not all or nothing; there are stronger and weaker forms of consent. In some situations, robust informed consent seems necessary, but, in other situations, weaker forms of consent may be appropriate. Next, I’ll describe three situations where researchers will struggle to obtain informed consent from all affected parties and a few options in those cases.

First, sometimes asking participants to provide informed consent may increase the risks that they face. For example, in Encore, asking people living under repressive governments to provide consent to have their computer used for measurement of Internet censorship might place those who agree at increased risk. When consent leads to increased risk, researchers can ensure that information about what they are doing is public and that it is possible for participants to opt-out. Also, they could seek consent from groups that represent the participants (e.g., NGOs).

Second, sometimes having fully informed consent before the study begins could compromise the scientific value of the study. For example, in Emotional Contagion, if participants had known that researchers were doing an experiment about emotions, this might have changed their behavior. Withholding information from participants, and even deceiving them, is not uncommon in social research, especially in lab experiments in psychology. If informed consent is not possible before a study begins, researchers could (and usually do) debrief participants after the study is over. Debriefing general includes explaining what actually happened, remediating any harms, and obtaining consent after the fact. There is some debate, however, about whether debriefing in field experiments is appropriate, if the debriefing itself can cause harm to participants (Finn and Jakobsson 2007).

Third, sometimes it is logistically impractical to receive informed consent from everyone impacted by your study. For example, imagine a researcher who wishes to study the Bitcoin blockchain (Bitcoin is a crypto-currency and the blockchain is a record of all Bitcoin transactions (Narayanan et al. 2016)). Some people that use Bitcoin expect and desire anonymity, and some members of the Bitcoin community might object to certain forms of research on their community. Unfortunately, it is impossible to obtain consent from everyone who uses Bitcoin because many of these people are anonymous. In this case, the researcher could try to contact a sample of Bitcoin users and ask for their informed consent.

These three reasons why researchers might not be able to receive informed consent—increasing risk, compromising research goals, and logistical limitations—are not the only reasons that researchers struggle to obtain informed consent. And, the solutions that I’ve suggested—informing the public about your research, enabling an opt-out, seeking consent from third-parties, debriefing, and seeking consent from a sample of participants—may not be possible in all cases. Further, even if these alternatives are possible, they may not be sufficient for your study. What these examples do show, however, is that informed consent is not all or nothing, and that creative solutions can improve the ethical balance of studies that cannot receive full informed consent from all impacted parties.

To conclude, rather than “informed consent for everything” researchers should, can, and do follow a more complex rule: “some form of consent for most things.” Expressed in terms of principles, informed consent is neither necessary nor sufficient for the principles of Respect for Persons [Humphreys (2015); p. 102]. Further, Respect for Persons is just one of the principles that needs to be balanced when considering research ethics; it should not automatically overwhelm Beneficence, Justice, and Respect for Law and Public Interest, a point made repeatedly by ethicists over the past 40 years [Gillon (2015); p. 112-113]. Expressed in terms of ethical frameworks, informed consent for everything is an overly deontological position that falls victim to situations such as Timebomb (see Section 6.5).

Finally, as a practical matter, if you are considering doing research without any kind of consent, then you should know that you are in a gray area. Be careful. Look back at the ethical argument that researchers have made in order to conduct experimental studies of discrimination without consent. Is your justification as strong? Because informed consent is central to many lay ethical theories, you should know that you will likely be called on to defend your decisions.