Biased and inattentive responding contribute to apparent metacognitive biases in mental health.

Abstract

Large-scale online studies with healthy adults have documented consistent associations between transdiagnostic psychiatric traits and metacognitive biases. Here, analysis of existing and new large-scale datasets reveals that such correlations may be largely due to surface-level dimensions of questionnaire-filling behavior: systematic rating biases and inattentive responding. Specifically, a bias to report positive or negative values in self-report scales may generalize to confidence ratings, producing spurious correlations between the two. Additionally, systematic overconfidence among inattentive responders produces spurious positive correlations between confidence and the endorsement of rare symptoms. We show that previously identified transdiagnostic dimensions of "anxiety-depression" and "compulsivity and intrusive thought," both shown to correlate with decision confidence, map neatly onto these two biases of questionnaire-filling behavior. In a preregistered experiment, we further show that decision confidence and self-reported obsessive-compulsive tendencies are correlated with independent measures of inattentive and biased responding. Taken together, we find substantial influence of inattentive and biased responding over both self-report psychiatric measures and confidence ratings. When not accounted for, these factors can produce a mirage of apparent metacognitive alterations in mental health. We discuss concrete precautionary measures that are needed to control for these biases.