socio-economic development, which would endorse the particularist school of
thought? These are the key additional questions this study seeks to answer. From a
policy perspective these questions are most relevant. If social cohesion is manifested
in ways that are in agreement with the particularist perspective it makes no sense for
countries to emulate social cohesion models from other societies because these
models are culturally specific and path-dependent phenomena. If, on the other hand,
social cohesion follows the logic of the universalist perspective, policies can be
devised that promote socio-economic development and thereby engender more
desirable forms of social cohesion.
As the aforementioned pioneering works of Green et al have partially
addressed these questions as well, it is important to highlight that this study aims to
complement Green’s studies in the following ways: (a) it not only seeks to explore the
validity of the particularist perspective but also that of the universalist view, which
was not an explicit objective of Green’s studies; (b) it makes use of a much larger
sample of countries (70 states worldwide, while Green’s studies were restricted to
OECD states); (c) it seeks to explore hypothesized regimes of social cohesion
diachronically over a period of two decennia.
Data and indicators
The data presented in this section serve in first instance to tap the eight components of
social cohesion as distilled from the four definitions discussed previously (i.e. the
components displayed in Table 1). They also make it possible to scrutinize whether
syndromes or regimes of social cohesion can be identified which are in agreement
with the universalist or particularist perspective.
I found one or more indicators for each of the eight components (see Table 3).
These indicators all have their limitations, but compared to alternatives they seemed
to be the best option. I acknowledge, for instance, that (the inverse of) the homicide
rate is a very crude and one-sided measure of social order, but given the deplorable
comparability of other cross-national crime statistics I preferred it over other
indicators. Similarly, discussing politics with friends may be a good indicator for
political engagement but it need not say anything about actual political participation.
Using indicators of actual participation (such as voter turnout), however, have the
drawback of not only tapping political engagement but also the possibility to
participate (i.e. in authoritarian states people may feel very engaged but they are
likely to be prohibited from expressing this in terms of actual participation).
As Table 3 further shows, the data have been taken from a variety of sources.
The data for all the subjective indicators have been drawn from the 1999-2004 Wave
of the World Values Survey, the only international opinion survey with a global
scope. The data of five of these indicators are national means, those of three indicators
represent percentages, while those of the four indicators for the „shared values’
component represent the inverse of national-level standard deviations (SDs). I made
sure that the indicators based on administrative data (the gini coefficient and the
homicide rate) matched the years of the 1999-2004 Wave as closely as possible. Thus
all data pertain to the end of the 1990s and early 2000s. I compiled data on as many as
70 countries worldwide. The analyses were sometimes based on less than 50
countries, however, due to missing data.