View article

Software development is a complex process requiring aspects of social, cognitive, and technical skills. Software engineers face high levels of uncertainty and risk during functional and security decision making. This preregistered study investigates behavioral measures of cognitive reflection, risk aversion, and optimism bias among professional freelance software developers and computer science students, to expose relationships between uncertainty-associated language and risk sensitivity. We employ content analysis with a mixed-effect model to understand how psychological dimensions influence risk sensitivity in secure software development. We show an interaction between cognitive reflection and optimism bias in the proportion of uncertainty-related language used. Overly optimistic outlooks combined with higher cognitive reflection drives up expressions of uncertainty, while pessimistic or realistic individuals reduce uncertainty as cognitive reflection increases. Software engineers who hold average or pessimistic views on the security of their code are more likely to speak more intuitively about security and risk. We discuss the potential of our findings in relation to understanding how to leverage language used by engineers as markers of risk aversion. Encouraging increased discourse could be used as a catalyst for increased cognitive reflection and grounding optimistic behaviors, leading to more careful decisions.