The Use of the Method of Partial Regression in the Analysis of Comparative Feeding Trial Data. Part II

Abstract

A previous statistical study of the relationship between the initial weight and feed consumption of swine of four different age groups and their performance in experimental feeding trials has been extended, and further examples of the application of these relationships to the reduction of the experimental error in such trials have been given. In all age groups a highly significant relationship between final weight and feed eaten is demonstrable, the increase in final weight per unit of additional feed consumed decreasing with advancing age of pig. The final weight attained is also influenced by the initial weight of the pig, except possibly in the case of young pigs fed through to market weight. It is shown that identical results are obtained whether the gains or final weights attained are corrected for variations in feed consumption by means of the appropriate regression formula. The usual calculation of gains thus leads to no increase in accuracy, and unless of particular interest may be omitted. The application of the method of partial regression to the analysis of comparative feeding trial data has been illustrated, using the results of an experiment recently completed at Macdonald College in which five lots of ten pigs each were fed for a period of 105 days on rations differing in protein level.