Computer simulation of a clinical trial as an aid to teaching the concept of statistical significance

The advent of cheap and powerful microcomputers provides opportunities for improving the teaching of statistics in medicine. This paper describes a practical exercise in which students study the concept of statistical significance using a simple computer program which simulates a clinical trial. The students are required to make decisions about patients with incomplete data and form contingency tables for analysis. This includes analysing contingency tables with several cells having small expected frequencies. The results from 25 groups of students are collected together and presented to the class. Two null hypotheses are tested, one of which is true. For the true null hypothesis the group is likely to produce, by chance, at least one significant result at the 0.05 level. For the false null hypothesis only a minority of results will be significant, because the sample size is small. This leads to a discussion of the concept of statistical significance. The simulation model is described and the advantages and potential of this type of teaching are discussed.