Premium
Illustration of a Cluster Analysis Method for Mean Separation 1
Author(s) -
Gates C. E.,
Bilbro J. D.
Publication year - 1978
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj1978.00021962007000030024x
Subject(s) - desk , separation (statistics) , cluster (spacecraft) , statistics , variance (accounting) , range (aeronautics) , computer science , contrast (vision) , separation method , mathematics , data mining , artificial intelligence , chemistry , chromatography , engineering , programming language , accounting , business , aerospace engineering , operating system
New statistical methods of separating means into different groups should be brought to the attention of researchers so they can decide if the new methods can be used advantageously in their research programs. Our objective was to illustrate the use of a cluster analysis method (which we have called the Scott‐Knott method after the developers) and compare it to the commonly used Duncan's multiple range test. We applied the two methods to four sets of data. The results showed that the smaller the variance of the treatment means, the more similar were the separation groupings produced by the respective methods. However, in contrast to Duncart's procedure, the Scott‐Knott method never produced overlapping mean separation groups and distinct groups of non‐overlapping means are often desired by the researcher. The Scott‐Knott method has more complex calculations than Duncan's method, but it is readily programmable for computers and many desk top electronic calculators.