What is the impact of the data resorting in the HBAT example?
The HBAT dataset was re-sorted during some preliminary examination of the data before the hierarchical and nonhierarchical analyses were performed. In doing so, the re-ordering of the dataset has an impact on the labeling of output in the hierarchical analysis and small changes in the final cluster solution in the nonhierarchical approach versus what would be obtained with the dataset in its original order. The necessary sorting of the data to exactly replicate the results in the text is illustrated in the syntax file and a documentation file. The intent was never to cause any confusion, but this illustrates the somewhat “exploratory” nature of cluster analysis in that many different elements (e.g., data standardization or not, use of hierarchical or nonhierarchical procedures, use of see points versus random starting points in nonhierarchical analyses, etc.) have an impact on the resulting cluster solution. Users should always avoid the temptation to select one “correct” cluster solut
