Sotaro Shibayama (Lund University)

Risk in science: the socialisation of risk-taking in early-career training

Abstract. Science is a risky business, and risk-taking is a critical trait for scientists. This study uses a combination of survey and bibliometric data to measure the degree of individual scientists’ risk-taking and investigate its determinants. In particular, we focus on academic training as a critical arena for the socialization of risk-taking and examine whether and how risk-taking is transmitted from one generation to the next. Drawing on a sample of PhD students and their supervisors in life sciences in Sweden, we find that students’ risk-taking is significantly associated with that of their supervisors, and that this association is stronger when supervisors provide more frequent mentoring. Importantly, the association persists for 10 years even after students leave the PhD affiliation and switch research topics. Approximately 25% of the variation in students’ risk-taking can be attributed to the variation in supervisors’ risk-taking. We find other determinants of students’ risk-taking such as funding conditions and previous job experiences. As endogeneity due to selection is a concern, we employ the propensity score matching technique and obtained consistent results. We also exploit cases in which a supervisor was replaced by another supervisor, showing that the longer the initial supervisor is in place, the stronger the association between students’ risk-taking and the initial supervisor’s risk-taking and the weaker the association with that of their replacement. These findings suggest the critical role of early-career academic training in shaping scientists’ risk-taking and passing down the trait to the next generation.