Participants are exposed to risk by taking part in research studies, and researchers should always consider these risks carefully. When considering a particular study, a natural first reaction is to ask whether the research exposes participants to any risks at all. While this reaction is well-meaning, it turns out to be not very useful. Quickly you will decide that all human subjects research could potentially cause some harm or discomfort. For example, even a simple survey collecting demographic information could possibly cause some person some discomfort. For this reason, a much more useful standard in research ethics is minimal risk. Roughly, minimal risk asks how the risks of the research study compare to the risks that people experience in their everyday life.
More formally, the Common Rule defines minimal risk as follows:
“Minimal risk means that the probability and magnitude of harm or discomfort anticipated in the researchers not greater in and of themselves than those ordinarily encountered in daily life of during the performance of routine physical or psychological examinations or tests.”
Under the Common Rule, research that is minimal risk has less stringent oversight than research that is not minimal risk. For example, minimal risk research is eligible for expedited review, and some elements of informed consent can be waived for research that is minimal risk. These procedures are consistent with the idea that the level of review required for a study is related to the level of risk.
One challenge when trying to apply the definition of minimal risk is that it is not clear whose daily life is to be used for benchmarking. For example, homeless people have higher levels of discomfort in their daily lives. Does that imply that that homeless people can be subject to research that is higher risk? Probably not. For this reason, there seems to be a growing consensus that minimal risk should be benchmarked against a “general population” standard, not a “specific population” standard. Returning again to a study of homeless people, when deciding if that study is minimal risk, researchers should compare the probability and magnitude of harm or discomfort not the daily life of a homeless person but to the daily life of someone in the general population.
While I generally agree with the idea of a general population standard rather than a specific population standard for assessing minimal risk, I think that for online field experiments, a specific population standard is reasonable. That is, when considering an experiment on an online platform (e.g., Facebook), I think that it is reasonable to benchmark against everyday risks on that platform. This kind of specific population standard is unlikely to conflict with the justice principle in the Belmont Report, which seeks to prevent the burdens of research failing unfairly on disadvantaged groups (e.g., prisoners and orphans).
Most social science research is minimal risk, and most online experiments run by companies are also minimal risk. However, if your research is not minimal risk, then you should really think carefully about whether the benefits are worth the risks and whether there are other lower risk ways of obtaining the same scientific knowledge.