similarity ratings and Belk’s scale of Materialism did not have the redundancy or balance of the 
PRF, any missing data was grounds for exclusion from the analyses of those sections of the 
questionnaire data. [See Appendix D for a tabulation of the data used here.] 
Results 
The first task of the analysis was to determine the semantic structure of the verbs of 
ownership in a manner that would be comparable and legitimate for the two samples. Assuming 
these to be opportunistic samples of the same population of adult English Canadians, the data 
were initially pooled (n =245). Hierarchical cluster analysis of the verb correlation matrix using 
maximum distance criteria indicated six clusters of verbs. Save and hide were anomalous. This 
is shown in Figure 6, where, for example, save correlated most highly with keep (r=.25, p <.001), 
manage (r=.27, p<.001), plan (r =.28, p <.001) and share (r =.30, p <.001), while hide correlated 
most highly with make (r=.21, p<.001) and lend (r =.25, p<.001). Though statistically significant, 
these correlations were still much weaker than the within-cluster correlations of verbs that did 
cluster. 
These verb clusters appear on the left in Table 4, where they are tentatively labelled 
possession, dominion, acquisition, attachment, stewardship, and covetousness. To confirm 
these clusters, factor analysis was used on the pooled data, and Cronbach alpha coefficients 
were used as measures of cluster coherence in the sample data. Principal components factor 
analysis, based on eigenvalues greater than one and varimax rotation, essentially reproduced 
these clusters. Other factoring methods, rotations, and factor selection criteria did not improve 
the correspondence with the cluster analysis. The factor analysis, however, did indicate that 
be familiar with and share should each appear in two factors. Keep and cherish had relatively 
low factor loadings in their respective verb clusters, but were retained there because their 
presence did not appear to diminish sample Cronbach alpha coefficients for the verb clusters. 
Considering the verbs as equally weighted items in a scale, alpha coefficients were above 
65 for the summer students and above .71 for the ferry passengers on all but the possession 
cluster. For that two-item cluster, have and possess were correlated in the student sample 
(n= 170, r =.46, p <.001), but not in the ferry sample (n=75, r=.16, p >.05). The failure of have and 
possess to correlate in the ferry sample may stem from possess having less of a meaning of 
dominion for that group. For the students (n= 170), both have and possess correlated with