that Denmark, Sweden and to a somewhat lesser extent the Netherlands always stick
together. In the first wave they are accompanied by Britain and Canada but these
countries move to the continental European cluster in the second and third wave
respectively. We further see that Italy moves from a separate position to the
continental cluster in the second wave.
In short, the data provide only partial support for the „stable regimes of social
cohesion’ hypothesis. Whether a predicted regime can be identified substantively or in
terms of the countries it is said to include depends on the indicators and countries
examined. Two clusters seem to be quite stable - a continental European one with
Belgium, Germany, France, and Spain as core countries and a Scandinavian one
grouping Denmark and Sweden - and these could thus possibly be conceived of as
regimes. However, the unexpected companions in these clusters (Ireland and the
Netherlands respectively) and the transfers of Britain, Canada and Italy undermine
their external distinctiveness and put their longevity into question.
Conclusion
The analyses of this study have shown that none of conceptions of social cohesion
proposed by a number of macro-level approaches refers to a coherent empirical
phenomenon visible in a large number of states that could be labeled social cohesion.
As these conceptions thus constitute merely multidimensional theoretical constructs,
their use for empirical and policy-oriented approaches can legitimately be questioned.
Working with such constructs is problematic in policy terms because interventions
aimed at improving social cohesion as a whole could well have differential effects
(positive and negative) for the constituent components of social cohesion.
A factor analysis on all the components of social cohesion suggested by the
macro-level approaches produced four factors, two of which captured most of the
variance. The first of these I labeled „solidarity’ because of the high loadings of social
trust, equality, consensus on basic values and social order on this factor. The second
was labeled „participation’ as it comprised civic participation as its main component
and was further found to be related to ethnic tolerance, national pride, and (a lack of)
institutional trust. Both factors showed a strong positive correlation with GDP per
capita. I considered this finding to be supportive of the universalist/modernist notion
that social cohesion should be related to stages of socio-economic development.
More difficult to reconcile with the universalist perspective was the finding
that the post-communist and Latin-American countries differed conspicuously on both
factors while showing approximately equal levels of socio-economic development. It
suggested that historical processes unique to each region also play a role in shaping
social cohesion to the point that it is appropriate to speak of regionally distinct and
relatively enduring „regimes’ of social cohesion. This finding is more in line with an
ideographic/particularist understanding of social cohesion. Putting this perspective to
the test also produced mixed evidence, however. Verifying the substance, the country
membership and the stability of the four regimes of social cohesion postulated by
Green et al (2009), I found evidence for a reasonably distinctive and stable
Scandinavian model characterized by high trust, (declining) inequality and low crime
rates. I also identified a continental European cluster but this group saw several
English-speaking countries joining it in the 1990s and exhibited unexpectedly low