Epistemic risk: empirical science and the fear of being wrong

Abstract

Throughout its history, empirical scientists have taken a conservative approach towards epistemic risk, minimizing inferences beyond observable data. However, at the same time, scientific research is conducted with the aim of advancing scientific knowledge and furthering its practical application in areas such as public health and medicine. Twentieth-century scientists and philosophers have developed methods for managing epistemic risk and quantifying degrees of uncertainty and evidential support. In the biomedical sciences, these efforts have focused primarily on the use of the p value, along with a threshold for statistical significance (p < 0.05) and avoidance of Type I error. However, excessive reliance on p values comes at a high cost. The p value threshold avoids Type I error at the expense of Type II error. Moreover, this imbalance is counter to the aims of scientific research to increase knowledge and reduce uncertainty. Ultimately, there is no single rule for determining what type or amount of epistemic risk is acceptable. Biomedical scientists should be aware of how assumptions and values may influence, often surreptitiously, epistemic risk taking.