December 22, 2019

Anchoring and adjusting heuristic

Effect of Teaching Bayesian Methods Using Learning by Concept vs Learning by Example on Medical Students’ Ability to Estimate Probability of a Diagnosis

Why do we need more bayesian reasoning?. Can we improve clinical decision making? The answer in this JAMA article.
Although bayesian reasoning has been widely promoted in the literature as a diagnostic strategy, there has been scant evidence that teaching bayesian reasoning actually improves diagnostic accuracy. Providing learners with relatively brief instruction on these abstract concepts appeared to significantly improve their diagnostic performance in comparison with simply providing a number of relevant examples or no relevant instruction.
 The previously reported discrepancy between human probability estimation and optimal bayesian probability estimation has become traditionally accepted in the psychology and medical literature as cognitive biases, including base-rate neglect, anchoring bias, confirmation bias, and representativeness, all of which suggest suboptimal revision and have been purported to be a primary cause of diagnostic error.13,14 However, on closer scrutiny, much of this evidence was derived from situations that were not representative of the typical diagnostic setting. One highly cited study9 used a screening situation in which the base rate was very low and any positive test result, even one with excellent operating characteristics, would have most likely been a false-positive result. Participants in that study also appeared to exhibit semantic confusion by confusing the posttest probability with the conditional probability that was presented in the problem-solving exercise.