The analysis of repeated measurements in sensory and consumer studies

Methods of analysing repeated measurements from the perspective of a medical statistician are described. These are illustrated using two previously published data sets and one new data set. The key tool in the analysis is the mixed effects linear model. Analyses appropriate for normally distributed and ordered categorical assessments are described. Two of the examples use a cross-over design and the third a between-groups design. An important aspect is that we show how the correlations between the repeated measurements can be allowed for in the model. For ordered categorical data we illustrate the use of the generalized estimating equations (GEE) methodology. We also briefly consider the problem of multiple testing.