MEDICAL TRAINEES ARE LEARNING BUT BIASED: TESTING KNOWLEDGE EFFECTS ON SUPPORT THEORY WITH A SIMPLE QUESTION

Purpose: To understand the role of knowledge on differing presentations of a simple medical question. It is implied in Support Theory that knowledge mitigates the effects of presentation manipulations on the assessment of an outcome.   

Method: All medical students and medical residents from Family Medicine, Surgery, and Internal Medicine at the University of Kansas School of Medicine - Wichita; and 80 undergraduates at Wichita State University; were approached to estimate the chances of causes of death on a questionnaire. The same causes of death were presented in varying levels of detail; consistent with Support Theory. Of those approached, 75 undergraduates, 47 medical students, and 39 medical residents agreed to participate. A total of 6 questions were shown to each participant, divided between contexts that included causes of death. For each context, question presentations had either minimal detail (I), much detail on the first cause (EF), much detail on the second cause (EA), or much detail on both causes (EFA). The questions and context order were randomly selected per participant, and no order effects were found.

Results:  Support Theory predicts EA < I, I < EFA, EFA < EF when participants are biased by presentation manipulations. This general trend was found with undergraduates (EA ³ I, I ≤ EFA, EFA < EF), but not as strongly with medical students (EA = I = EFA, EFA ≤ EF). Medical residents’ responses were consistent across the causes of death questions, except for EA (EA < I, I = EFA = EF).

Conclusions: We would expect that differing presentations of the same situation would not affect one’s judgment, and judgments would be consistent regardless of the detail manipulations. However, participants with less medical knowledge (undergraduates) were sensitive to presentation manipulations, and had a different response pattern than those with greater medical knowledge. Knowledge gained through training and education offers protection from being biased by presentation manipulations. However, even those with the most medical training demonstrated bias with at least one presentation. This may have implications for more complex diagnosis and treatment decisions that are based on a patient’s description or a colleague’s presentation of symptoms.