Table 2 shows the means of the three dimensions comprising each cluster (upper panel) while their z-scores are graphically depicted in Fig. 1. The derived solution was tested for its stability by splitting the sample into two halves and performing k-means clustering in each half by using as initial centers the ones that were derived by the hierarchical clustering in the other half. The solution of each half was then cross-tabulated with the respective solution that was derived from the full sample. The results showed acceptable consistency (kappa's agreement > 0.95 for both halves).