THE EFFECTS OF WITHIN-GROUP COVARIANCE STRUCTURE ON RECOVERY IN CLUSTER-ANALYSIS .1. THE BIVARIATE CASE

In a preliminary investigation, two Monte Carlo studies investigated t he effects of within-group covariance structure on subgroup recovery b y 10 hierarchical clustering methods. Data sets were 100 bivariate obs ervations from two subgroups. Study 1 manipulated subgroup size, withi n-group correlation, within-group variance, and distance between centr oids. Negative within-group correlation yielded much poorer recovery f or all clustering methods. In addition, clustering method interacted w ith within-group variance. Study 2 manipulated subgroup size, within-g roup correlation, direction of the vector separating subgroup centroid s, and distance between subgroup centroids. Superior recovery was asso ciated with within-group correlation that matched the direction of sub group separation. Results are interpreted according to the weakness of Euclidean distance as a measure of (dis)similarity.