Comparing treatment means correctly and appropriately

Publisher Website
Google Scholar
Abstract
Many researchers set up an experiment, make measurements, do an analysis of variance, calculate the mean response for each treatment, and then try to decide if the treatment means are significantly different and why. Duncan's multiple-range test is frequently used to test the difference among treatment means. It is, however, only one of a number of techniques that can be used to examine treatment means. Some researchers are unaware of the different techniques and that the interpretation of the results of an experiment can be strongly influenced by the technique used. In fact, using an inappropriate technique can lead to making incorrect recommendations and to completely missing major treatment effects. Selection of the appropriate technique to use for a particular experiment depends upon the nature of the treatments and the objectives of the research. This paper discusses four techniques (ranking treatment means, multiple comparison procedures, fitting response models, and using contrasts to make planned comparisons) that can be used to examine treatment means and presents examples of each one.

This publication has 2 references indexed in Scilit: