77 
used in the various societies. Third, ecological Cronbach alpha coefficients (alpha coefficients 
based on societies as cases and societal item medians as item scores) were moderately high 
(M =.81, range =.64 to .95), indicating that the SIV scales were coherent across cuitures 
(Hofstede, 1980). Median scores were used as societal measures because medians are more 
conservative than means, being insensitive to extreme scores. 
By using societies as cases and examining the ecological correlations based on median 
societal scores, the attenuating effects of individual differences within cultures were eliminated 
and the relationships between the SIV scales and the property attitudes became salient. 
Examining the ecological correlations of the six SIV scales with the two property questions, four 
relationships were found to be statistically significant. As shown in Table 3, for PQ1, a favorable 
societal attitude towards private property was very positively related to a relative societal 
preference for Leadership (n=15, r=.80, p<.001). For PQ2, a favorable attitude towards private 
property was positively related to relative preferences for Support (n=15, r=.64, p =.005), for 
Independence (n= 15, r=.54, p =.02), for Leadership (n=15, r=.46, p =.02), and very negatively 
related to a relative preference for Conformity (n=15, r=-.80, p <.001). : 
Figures 1 and 2 plot the relationships of the two largest of these, median societal 
Leadership and Conformity scale scores, with the median societal property attitude scores. In 
Figure 1, when Yugoslav and Ceylonese samples are removed, the correlation decreases (n=13, 
r=.42, p=.08). However, because this correlation remains markedly positive and because there 
is no rationale for deleting the Yugoslav and Ceylonese data, the correlation of Leadership with 
PQ1 will be considered statistically significant until demonstrated otherwise by replication 
studies. Indeed, the positive correlation of preference for Leadership with positive attitudes 
towards private property is the only correlation that is statistically significant for both property 
questions, at both the societal or individual levels of analysis. 
To further examine these relationships, forward multiple regression of the SIV preference 
pair scores on PQ1 showed LB, LR and LC to account for 71% of the variance of PQ1. This 
means that preferring Leadership to Benevolence, Recognition and Conformity predicts a 
Savors attitude towards private property. Similarly, CL, CS and CR accounted for 86% of the 
variance of PQ2. This means that preferring Conformity over Leadership, Support and 
Recognition predicts that a society Is ill disposed to private property. This multivariate analysis