Statistical aspects of measuring change in oral health status of older adults

There are many statistical aspects involved in the measurement of change and some of these are discussed in this paper. First, a general overview of the three types of index, discriminative, predictive and evaluative, is presented. Each type of index should be developed in a different manner and this will depend on its use. For example, the reliability of an evaluative index is different from that of the other types of index and would depend on the observations on each individual remaining stable over time, and having a large change score when the dental status improves or deteriorates. Second, the issue of when we should be measuring change as opposed to just using the final measurement, or to using an analysis of covariance, is discussed. The final section of the paper looks at some of the fallacies which occur in the measurement of change and discusses whether or not we should try to correct for these.