Numerous procedures are available for performance of pairwise multiple comparisons among the observed treatment means from designed experiments.
Since the magnitudes of the critical values vary among procedures, results obtained from the application of one procedures to a given set of data will often differ from those obtained if another procedure is utilized.
In this study, the simulation was performed using hypothetical data for each of 16 sets of treatment means.
The results were summarized as follows :
1. In the case of equal variance of each treatment levels, equal or not equal sample size, and the magnitudes of variance regardlessly, Tukey and Scheffe´ procedures are more conservative than LSD, Duncan and SNK procedures in specifying the real difference of treatment means.
2. In the case of unequal variance of each treatment levels, LSD, Duncan procedures are more sensitive than SNK, Tukey, Scheffe procedures, especially Scheffe procedure is more conservative than other procedures under the treatments variance in large.
3. By the magnitudes of variance, of all multiple comparison procedures, the group if large variance is more sensitive than small variance group.
|