ABSTRACT
In our analysis of experience in the preceding chapter we proceeded on the assumption that the observations and measurements constituting the data were worth using. In this chapter we take a more critical look at the problems of observing and measuring. Clinical medicine abounds with difficult problems of measurement such as the conformity of measurements made by different clinicians under similar working conditions, the identification of sources of variability, the transferability of data from one clinic to another, the comparison of different methods of measurement, the development of measurement strategies to eliminate or at least to counter any sources of irreproducibility and the consensus problems between different observers for both continuous and categorical data. Since all of these problems involve variability and uncertainty the statistician has often an important role to play in assessing the merits of a method of observing or measuring, in advising on how the method may be improved and in quantifying the effects of any remaining imprecision.
