“6 
score of the scale, so that the scale score represents the proportion of occasions it was 
preferred. To highlight the null condition of no preference, the scores were then re-scaled by 
subtracting .50, so that transformed scale scores ranged from -.50 to +.50, with 0 indicating 
chance preference for the scale concept, negative values indicating rejection of the scale 
concept, and positive values indicating preference. 
Finally, a second set of SIV data was created for use in multivariate analyses. Because the 
SIV scores represent preference decisions, the data can be decomposed into preference pair 
scores which would be independent rather than ipsitive. For six scales, there are 15 preference 
pairs (e.g. Independence preferred to Support, Independence preferred to Benevolence, 
Independence preferred to Conformity, etc.). Labelling these preference scores with SIV scale 
initials, the 15 independent measures would be 1S, IB, IC, IL, IR, SB, SC, SL, SR, BC, BL, BR, CL, 
CR, and LR. In the earlier example in which Leadership was preferred to Recognition and 
Conformity, and Recognition was preferred to Conformity, LR would be incremented by 1, and 
CR and CL would not be incremented. The total preference counts for each of the 15 scale pairs 
was divided by the number of times the preference pair was offered on the SIV. This ranged 
from 4 for SR to 18 for IR. Thus, the range of the transformed preference pair scores is 0 to +1. 
The labelling of preference pairs can be reversed by using complementary scores (e.g. SI = 1 
- 1S). To make these graphically comparable to the other scores, 0.5 was subtracted from the 
preference pairs such that a positive value indicates a preference for the first scale of the pair 
over the second, a negative value indicates a A for the second scale of the pair over 
the first, and a zero indicates a balanced preference, which is chance preference or 
non-preference, of one scale over the other. Preference pair scores are independent of one 
another: a high or low value on one pair has no necessary influence on the value of another pair. 
All pair scores could simultaneously have maximum values. [See Appendix A for a tabulation 
of the societal data used here.] 
Results 
The questionnaire responses were comparable across cultures for three reasons. First, 
the sample groups within societies were similar in that they were composed of non-foreign, 
male. university students. Second, the societal Cronbach alpha coefficients were generally 
adequate (M =.75. range =.44 to .88), indicating that the SIV scales were internally coherent when