## ABSTRACT

From a validation study, data have been acquired on inter-analyst reliability (how consistently different people use the method), intra-analyst reliability (how consistently the same person uses the method on different occasions), and validity (how well the method predicts what it is supposed to). The coefficients of each of these variables lie between 0 and 1, so a simple multiplication will provide information on the overall accuracy of the method (which will also be between 0 and 1). That is, given a random analyst at any point in time, how well can the analyst be expected to perform with a given method? Accuracy can therefore be summarised thus:

accuracy = inter-analyst reliability × intra-analyst reliability × validity

The study was based on an in-vehicle device, so the retooling figures we refer to here are also based on this type of system. Of course, analysts can substitute these figures for their own if they are interested in a different product. Retooling costs for a car radio can be between £3000 (for minor changes to the product line) and £150 000 (for complete retooling) using figures from 1999 (which are likely to have doubled by 2014).