Abstract
90 college student subjects participated in an experiment testing the effect of graphic representation on speed and accuracy of Bayesian reasoning. Five representation conditions were employed: a contingency table representation, three graphic representations, and a no-training control group. Tabular or graphic depictions were shown only as feedback and were not available as a student solved a problem. Control subjects remained inaccurate throughout the experiment, showing that practice alone is not sufficient to teach Bayesian reasoning. Subjects receiving graphic feedback were highly accurate and significantly faster than those in the contingency table condition. The reaction time findings are interpreted as evidence that a person using one of the present three graphic representations does not simply reduce the diagram to a contingency table representation as a substep in solving a Bayesian problem. Graphic repesentation seems to lead to a mental model that differs from the model prompted by tabular representation. Practically, the present findings suggest that a physician or nurse with no external aid can accurately diagram and solve a Bayesian diagnostic problem in about 30 seconds.
Full text
PDF
Selected References
These references are in PubMed. This may not be the complete list of references from this article.
- Casscells W., Schoenberger A., Graboys T. B. Interpretation by physicians of clinical laboratory results. N Engl J Med. 1978 Nov 2;299(18):999–1001. doi: 10.1056/NEJM197811022991808. [DOI] [PubMed] [Google Scholar]
- Christensen-Szalanski J. J., Beach L. R. Experience and the base-rate fallacy. Organ Behav Hum Perform. 1982 Apr;29(2):270–278. doi: 10.1016/0030-5073(82)90260-4. [DOI] [PubMed] [Google Scholar]
- Christensen-Szalanski J. J., Bushyhead J. B. Physician's use of probabilistic information in a real clinical setting. J Exp Psychol Hum Percept Perform. 1981 Aug;7(4):928–935. doi: 10.1037//0096-1523.7.4.928. [DOI] [PubMed] [Google Scholar]