Toxic dilatation and infective diarrhoea.

different scales are used.In such a case the intraclass correlation coefficient computed for each method will indicate which is the more reliable, provided each method of measurement uses the same population, as in a dataset we recently examined.'In summary, we agree with Chinn when she says that "the intraclass correlation coefficient should be used only for comparative purposes.The within-subject standard deviation or 95 per cent range for change should be calculated as the absolute measure of repeatability."2