interest had data missing or outside the possible range. By these selection criteria, 820 of the
original 4875 respondents were eliminated. Also, three entire national groups (394 respondents)
were eliminated: a) the Finns because respondent's sex had been miscoded such that no males
were present even though the sample site report sheets in the SIV manual (Iversen, 1969) noted
only a slight over-representation of females; b) the British because they had not completed the
SIV; and c) the West Germans because only 20 of the original 65 respondents remained after
eliminating females and foreign students. The South African samples were separated into two
national groupings, South African Bantu and South African Whites. Finally, three sub-samples
(173 respondents) were eliminated because of doubtful reliability of the SIV scales. For the first
Indian sample, Cronbach alpha coefficients ranged from .37 to .58, with a mean of .49. For the
two Ceylonese samples that used questionnaires that were translated into Tamil and Sinhalese,
alpha coefficients ranged from .11 to .74, with a mean of .57. After eliminating these three
sub-samples, mean national alpha coefficients, with one exception, were all above .67. Only the
South African Bantu had an alpha coefficient below .50 (a = .44 on the Support scale). A
summary of the characteristics of the samples appears in Table 2.
Before analyses, the questionnaire data were transformed to make them more
interpretable and comparable when represented graphically. First, PQ1 was reversed by
subtracting it from 8, so that both property question scores represented positive attitudes
towards private property. Then both items had one scale point subtracted to set their range
from 0 to 6. They were then divided by six and had .50 subtracted. Thus, the transformed
property questions had a range of -.50 to +.50, with zero indicating indifference towards private
property, a negative value indicating a negative attitude, and a positive value indicating a
positive attitude.
On the SIV, the standard scoring procedure tabulates the number of times each scale was
preferred over the other scales. In the earlier example triad, if a respondent chose the
Leadership item as most important, it would be scored 2, indicating that it was preferred twice
(once over Recognition and once over Conformity). If Conformity were chosen as least
important, it-would get a score of 0, indicating that it was not preferred over any items.
Recognition, the neutral preference, would get a score of 1, indicating that it was preferred
once, over Conformity. Three of the SIV scales have 15 items, two have 16 items and one has
13 items. Following the practice of Gordon (1967), each scale was divided by the maximum