Performance testing at central test stations and on farm testing are primarily used to select superior breeding stock. Since there may be environmental differences between batches, contemporary comparisons are often used in which each pig's performance is compared with its group's mean.When batch sizes are small, the accuracy of selection can be improved by using an exponentially smoothed average for the contemporary mean of the form where xt is the exponentially smoothed average, the mean of the present batch and po the smoothing parameter.

When batch sizes are variable, too much emphasis is placed on the current mean when the batch is smaller than average, and conversely for batches larger than average, there is too little emphasis on the current mean. Two alternative smoothing parameters are proposed which take into account variation in batch sizes. The accuracy with which pigs of superior genetic merit could be identified using the conventional and the alternative smoothing parameters was investigated by simulating batches of pigs and calculating the proportion of pigs correctly identified as being of superior genetic merit.