Explaining computation of predictive values: 2 × 2 table versus frequency tree. A randomized controlled trial [ISRCTN74278823]

Abstract

BACKGROUND: Involving patients in decision making on diagnostic procedures requires a basic level of statistical thinking. However, innumeracy is prevalent even among physicians. In medical teaching the 2 x 2 table is widely used as a visual help for computations whereas in psychology the frequency tree is favoured. We assumed that the 2 x 2 table is more suitable to support computations of predictive values. METHODS: 184 students without prior statistical training were randomised either to a step-by-step self-learning tutorial using the 2 x 2 table (n = 94) or the frequency tree (n = 90). During the training session students were instructed by two sample tasks and a total of five positive predictive values had to be computed. During a follow-up session 4 weeks later participants had to compute 5 different tasks of comparable degree of difficulty without having the tutorial instructions at their disposal. The primary outcome was the correct solution of the tasks. RESULTS: There were no statistically significant differences between the two groups. About 58% achieved correct solutions in 4-5 tasks following the training session and 26% in the follow-up examination. CONCLUSIONS: These findings do not support the hypothesis that the 2 x 2 table is more valuable to facilitate the calculation of positive predictive values than the frequency tree.