social hierarchy (see Appendix 2 for the full wording of the item). Since the
postulated regimes did not make specific claims about the level or the strength of a
shared sense of belonging I omitted the two indicators on identity. I further had to
omit all four indicators on value diversity because of missing values in the first wave.
Instead, I selected the materialism-postmaterialism values scale and used the SDs of
this scale as a measure of value diversity. This scale consists of four items and is a
subset of the aforementioned survival-selfexpression values scale (see Appendix 2).
Lastly, data from the UN and World Bank on homicides (social order) and income
gini (equality) were collected for the years 1981 and 1990 to match the survey data of
the first and second wave of the WVS. I thus proceed with ten indicators for which
data were found for the three points in time noted above. Table 5 lists these indicators
and the components of the proposed regimes they are meant to tap.
Table 5 about here
Testing the regimes proposed by Green et al (2009) involves asking three questions:
(1) to what extent do the patterns in the data match the expected substance of the
regimes?; (2) to what extent can we find the hypothesized country clusters?; (3) how
stable are the substance and country clusters found in the data? I used group means
and hierarchical cluster analysis to explore questions 1 and 3, and 2 and 3,
respectively. To begin with group means, I assigned sixteen OECD countries to the
regimes they are expected to exemplify and calculated the regime mean score on each
of the indicators for each of the three waves. Table 5 presents the results of these
computations. It also includes the overall mean so that we can assess whether a
particular group has a relatively high or low score on some indicator. Low scores
more than one standard deviation (SD) from the overall mean are given in italics; low
scores less than one SD from the overall mean are in normal style; high scores less
than one SD from the mean are given in bold; high scores more than one SD are in
bold and italics.
It turns out that the data of Table 5 are broadly in line with the proposed
regimes substantively. Indeed we see that the liberal group of countries has relatively
high levels of inequality, crime, tolerance and civic participation (both active and
passive). Likewise, the social democratic Nordics are relatively high on social trust
and low on crime and inequality. The continental group conforms to the expected
regime by showing relatively low levels of crime and passive and active participation,
and medium levels of income inequality. The predicted strong social hierarchies and
exclusionary ethno-cultural identities of the East-Asian regime are confirmed by East
Asia’s high score on respect for parents and very low score on ethnic tolerance. Yet,
we also find scores not in line with the predicted regimes. Thus, for a „conservative
regime demanding moral consensus’ we find surprisingly high value diversity in the
European continental group of countries. By contrast, value pluralism is surprisingly
low in the liberal societies. Likewise, the assumed individualism of these societies
does not prevent them from showing relatively high levels of respect for parents.
Lastly, the high ethnic tolerance levels of continental Europe are difficult to match
with their supposedly exclusionary ethnic identities.
Table 5 also allows us to assess the stability of the proposed regimes. The
overall picture is ambiguous. On some components the mean scores are fairly stable
over the three points in time for all four groups, which is consistent with the idea of