The effects of the confirmation bias on diagnostic decision making

Abstract

Research has shown that the confirmation bias affects judgments in a wide array of contexts, including clinical settings. Studies have demonstrated that the confirmation bias not only affects one’s hypothesis testing strategy, but also how one interacts with others and how others in turn respond. In addition, several studies investigating the effects of diagnostic labels have found that such labels affect the way in which clinicians interpret information. Thus, once clinicians make a diagnosis, the confirmation bias has the potential to color their subsequent processing of new data, resulting in confirmation of initial diagnostic impressions even if these impressions are contradicted by subsequent data. In an attempt to mitigate the effects of the confirmation bias, several studies have investigated various debiasing techniques, including warning or educating subjects about biases. The results of these studies indicate that such techniques may be useful if subjects are adequately educated about a bias. The purpose of this study was to examine the possible effects of the confirmation bias in psychodiagnostic assessment, and to examine the effectiveness of educating clinicians specifically about the confirmation bias. Clinicians received two case vignettes, followed by additional information about each case one week later. The additional information was either consistent with the diagnosis indicated by the initial data, or inconsistent with it. Participants were asked to make a diagnosis after reading each vignette. Half of the participants received information about the confirmation bias at time two before receiving additional information about the cases. Results suggest that clinicians did evidence the confirmation bias during performance of this task. Specifically, clinicians tended to remain consistent with their original diagnosis when a change in diagnosis was required from time one to time two. In addition, the confirmation bias instructions did not have a significant effect on participants’ performance during this task. Participants did not overcorrect, or incorrectly alter their initial diagnoses, when they received confirmation bias instructions. Finally, age was a predictor of diagnostic accuracy; specifically, as age increased, performance decreased. There was also some evidence to suggest that those participants who identified their orientation as cognitive behavioral performed more accurately. The results hold important implications about commonly occurring biases in the psychodiagnostic assessment process.%%%%Ph.D., Clinical Psychology – Drexel University, 2006