Table 3 about here
The indicators selected to measure the shared values component deserve further
explanation. Of course we need to ask ourselves „which values’ when we seek to
assess the degree of consensus on them. Unsurprisingly, there is no agreement about
the values that matter. While Bellah et al. (1985), for instance, deem a common
Christian morality to be the glue that holds society together (a morality that they think
has been eroded by individualism), Almond and Verba (1963) and Dahl (1967) argue
that values relating to the institutions of democracy are key. Social cohesion, in their
view, is assured when citizens agree on the political institutions and procedures and
on the ways to participate in them. Given the disagreement about which values should
be shared, we selected no less than four indicators to tap into a variety of values. Two
of these are composite dimensions comprising a range of socio-cultural attitudes and
thus partly addressing the morality Bellah et al consider crucial. These are the
dimensions traditional versus secular values and survival versus self-expression
values created by Inglehart and Welzel (2005) by means of factor analysis. Both
dimensions represent coherent syndromes of values and can be seen as continuums.
Low values on the first dimension represent religious and traditional orientations
while high values denote secular beliefs. Low values on the second dimension
represent materialist convictions while high values denote post-materialist
orientations. Appendix 1 shows the composition of both dimensions. The other two
indicators concern support for the democratic system and support for gender equality,
which thus tap more into political and civic values.
Results
Assessing the four theoretical approaches
I start by assessing whether the definitions of social cohesion offered by the four
approaches discussed previously represent coherent syndromes. For this to be the
case, the components included in these definitions have to co-vary and show a strong
relation to a latent factor. I conducted an exploratory factor analysis (using the default
option in SPSS) on the indicators of social cohesion shown in Table 3 to explore these
relationships. This analysis is based on a sample of 41 countries worldwide. The
default option is the un-rotated solution using listwise deletion of missing values. It is
programmed to produce a minimum number of latent unrelated factors explaining a
maximum amount of the variance in the indicators. In other words, it tries to
„squeeze’ the variation in as many indicators as possible in one factor. Consequently,
if the collection of indicators captured by the first factor is not in line with any of the
definitions, we can be fairly sure that none of these definitions reflect a coherent
syndrome of social cohesion.
The analysis produces four factors (see Table 4). This already tells us that
comprehensive theoretical understandings of social cohesion are not likely to reflect
coherent real-life phenomena. The first factor shows strong correlations (i.e. loadings
of more than .50 or less then -.50) with nine indicators and explains 32 per cent of the
variance. It comprises equality, social trust, social order, sexual tolerance, national
skepticism (as indicated by the negative loading of national pride), disparities on the
traditional/secular and survival/self-expression values indicators (as indicated by