Relative Risk Analysis of Educational Data

This paper demonstrates the use of relative risk, a statistic which is widely used in other areas but currently under-utilised in education. Relative risk analysis provides a language for comparing educational outcomes as well as statistical tests of significance. We illustrate this statistic with data on students ’ understanding of decimal notation. In particular, we determine differences in how misconceptions operate at different ages, by analysing the relative risk of primary and secondary students persisting with particular misconceptions, and becoming experts. This paper illustrates an approach to reporting, comparing and analysing educational outcomes, which is widely used in other fields but which has not been often used in mathematics education. The concept of relative risk is widely used in reporting results of medical, environmental and epidemiological research, in both scientific papers and in the popular press. Educators could capitalise on this popular familiarity in reporting their results. As well as being useful for describing results, measures of relative risk are amenable to statistical analysis. There are simple tests of statistical significance, confidence intervals and effect sizes, which are easily calculated; either manually, with