We are trying to perform hierarchical clustering on ranked data (participants were asked to rank their preferences on a series of items). Which procedure is best to do this? VARCLUS appears to work more like factor analysis, and other clustering procedures only seem to cluster observations instead of variables. Any help with this would be appreciated. Thanks!
If you can calculate a distance matrix (a dataset of type=distance) between your objects (observations or variables), then you can use proc cluster to do the hierarchical clustering. Check the data= option in the proc cluster statement doc.