The hobgoblin of consistency: algorithmic judgment strategies underlie inflated self-assessments of performance

Abstract

People often hold inflated views of their performance on intellectual tasks, with poor performers exhibiting the most inflation. What leads to such excessive confidence? We suggest that the more people approach such tasks in a "rational" (i.e., consistent, algorithmic) manner, relative to those who use more variable or ad hoc approaches, the more confident they become, irrespective of whether they are reaching correct judgments. In 6 studies, participants completed tests involving logical reasoning, intuitive physics, or financial investment. Those more consistent in their approach to the task rated their performances more positively, including those consistently pursuing the wrong rule. Indeed, completely consistent but wrong participants thought almost as highly of their performance as did completely consistent and correct participants. Participants were largely aware of the rules they followed and became more confident in their performance when induced to be more systematic in their approach, no matter how misguided that approach was. In part, the link between decision consistency and (over)confidence was mediated by a neglect of alternative solutions as participants followed a more uniform approach to a task.