| Popis: |
There have been intensive research efforts to improve Bayesian reasoning over the last 25 years. Much of this research focuses solely on improving performance on Bayesian tasks. In addition to performance, however, it is also important to establish an understanding of the effect on the positive predictive value when parameters of Bayesian formula are changed. We call this ability “covariation” in Bayesian tasks. To this end, training courses were developed to support understanding of covariation, based on strategies that have been proven helpful by previous studies concerning performance, using: (a) natural frequencies and (b) visualisations, i.e., double trees and unit squares. Results of a comparative study in a pre-, post-, and follow-up test design show that the developed training courses can improve understanding of covariation. |
