Published online by Cambridge University Press: 02 September 2010
A study has been made of the relative accuracy of twelve linear body measurements of dairy cattle. The results of several repeatability trials are presented together with estimates of observer error derived from the routine measuring of identical twins. Various aspects of precision of measurement have been analysed. The accuracy found from any one trial can rarely be used as a reliable estimate of the accuracy of measuring in any subsequent experiment.
The importance of observer error in identical twin studies has been assessed. It was found that from 10% to 70% of the total variation within pairs of uniformly treated identical twins could be due to measuring error. It was therefore concluded that in such studies, where observer error is a fairly large part of the total variation, a single routine repetition of every measurement would be advisable in long term experiments and several in the case of short term experiments with a small number of animals.
Measurements have been ranked for relative accuracy and the ranking was found to be broadly similar on several different criteria: (a) relative to the variation within pairs of uniformly treated identical twins, (b) relative to the total variation in a population of unrelated animals and (c) relative to the mean size of the body part. The choice of measurements on the basis of relative accuracy is discussed.