ArticlePDF Available

Measuring Social Cohesion: An Experiment using the Canadian National Survey of Giving, Volunteering, and Participating


Abstract and Figures

Social cohesion is a concept difficult to define and to measure. As there can be many definitions, so there can be many measurements. The main problem, either in defining or measuring the concept, is its multilevel and multidimensional nature. At one extreme, country is the most commonly used level to view social cohesion but measurement at this level is of little use for any interventions. At the other extreme, community is the most useful level but it is a social construct for which data are difficult to get, given the administrative boundaries used in social surveys. As an initial attempt to measure social cohesion at a subcountry level, this study focuses on census metropolitan areas for which data on several dimensions of social cohesion are available. We use the information gathered by the National Survey on Giving, Volunteering and Participating (NSGVP) on three dimensions of social cohesion: political (voting and volunteering), economic (occupation, income, labour force participation) and social (social interactions, informal volunteering). Using statistical techniques including factor analysis and standardization, we create an overall index of social cohesion for each CMA. We point out use of this measure for further analysis of social dynamics.
Content may be subject to copyright.
(Accepted 26 February 2006)
ABSTRACT. Social cohesion is a concept difficult to define and to measure. As there can be
many definitions, so there can be many measurements. The main problem, either in defining or
measuring the concept, is its multilevel and multidimensional nature. At one extreme, country is
the most commonly used level to view social cohesion but measurement at this level is of little
use for any interventions. At the other extreme, community is the most useful level but it is a
social construct for which data are difficult to get, given the administrative boundaries used in
social surveys. As an initial attempt to measure social cohesion at a subcountry level, this study
focuses on census metropolitan areas for which data on several dimensions of social cohesion are
available. We use the information gathered by the National Survey on Giving, Volunteering
and Participating (NSGVP) on three dimensions of social cohesion: political (voting and
volunteering), economic (occupation, income, labour force participation) and social (social
interactions, informal volunteering). Using statistical techniques including factor analysis and
standardization, we create an overall index of social cohesion for each CMA. We point out use
of this measure for further analysis of social dynamics.
KEY WORDS: latent scores, national survey of giving, social cohesion, structural equation
modeling, volunteering and participating
... social solidarity is completely a moral phenomenon which,
taken by itself, does not lend itself to exact observation and
indeed to measurement. To proceed to this classification and
this comparison, we must substitute for this internal fact which
escapes us an external index which symbolizes it and study the
former in the light of the latter.
(Durkheim, 1893 [1965], p. 64)
More than a century ago, Durkheim (1893) stated there was neither a clear
definition of the concept of social cohesion nor was there a possibility of its
direct measurement. A century of advances in empirical observation and
Social Indicators Research (2007) 80: 461–492 Springer 2006
DOI 10.1007/s11205-006-0011-1
analytical techniques have not overcome the problem. There is still no
universally recognized definition of social cohesion, and conceptualizations
found in the literature are at times contradictory and difficult to opera-
tionalize. For example, a definition by Rosell (1995, p. 78), also adopted by
Maxwell (1996), states that social cohesion involves ‘‘building shared values
and communities of interpretation, reducing disparities in wealth and in-
come, and generally enabling people to have a sense that they are engaged in
a common enterprise, facing shared challenges and that they are members of
the same community.’’ Stanley (2003, p. 9) criticizes the ambivalence of the
expression ‘‘shared values’’ and notes that social cohesion does not mean
‘‘social sameness, homogeneity of values or opinions’’. He offers his own
definition of social cohesion as ‘‘the sum over a population of individuals’
willingness to cooperate with each other without coercion in the complex set
of social relations needed by individuals to complete their life courses’’.
While this definition avoids assuming commonality of values, it remains
limited by not recognizing social cohesion as a group property that is greater
than the sum of individual parts. As Mudrack (1989, p. 38) pointed out,
such a ‘‘legacy of confusion’’ arises because cohesiveness as a property of the
group is often not measurable, and so researchers largely start with data on
Although we are not able to define exactly what social cohesion is, we
often understand it as ‘‘something that glues us together’’. It is also clear
that social cohesion is a multidimensional and multilevel concept. Any at-
tempt at measurement needs to take both these aspects into consideration.
This paper tries to measure social cohesion in its multidimensional aspect at
the level of census metropolitan area. Taking advantage of past research
we discuss the data and methods used in modeling the index of social
cohesion, relegating the advanced and complex technical details to the
Endnotes and Tables in the Appendix. We then present the results of our
analysis, discuss what we have learned through this study, and suggest
improvements in measuring the concept.
The concept of social cohesion has two basic components (Moody and
White, 2003). One refers to the psychological identification of members
within a collectivity, called ideational component. The other refers to the
observed relationships among members, called relational component.
Durkheim identified the theoretical link between these two components by
connecting changes from ‘‘mechanical’’ to ‘‘organic’’ societies. Present-day
research unfortunately separates these two components, depending on the
focus of study, leading to a wide variety of definitions and measurements.
Studies that restrict the concept to the ideational component inquire about
individuals’ feelings, such as sense of belonging. In contrast, those that focus
on the relational component examine the relationships between members of
different groups. To cite a few examples, cohesion has been examined in
terms of individual psychological feelings (Bollen and Hoyle, 1990), global
structural relationships (Freeman, 1992), and relationships in various pos-
sible intermediate groups (McPherson and Smith-Lovin, 1986). All these
perspectives touch on different levels at which cohesion can be measured.
Apart from Durkheim’s two basic components, we can also find in the
literature some specific group properties classified under different dimen-
sions. These also help clarify the concept of social cohesion (see for example,
Berger-Schmidt, 2000). The dimensions that seem most amenable to oper-
ationalization and measurement are the five discussed by Jenson (1998),
subsequently expanded to six by Bernard (1999). In this study, we use the
term domain to indicate the three major aspects of social cohesion, namely
Social Domain, Political Domain and Economic Domain. And, we use the
term dimension to point to measurable components of each domain, namely
Recognition, Belonging, Legitimacy, Participation, Inclusion and Equality,
as shown in Figure 1.
The inclusion/exclusion dimension under the Economic domain points to
the market forces. It addresses the questions of who has opportunities to
participate or who is marginalized or excluded from participation in the
economy. The dimension of equality/inequality was suggested by Bernard
(1999) who argued that equality is an essential dimension of social cohesion
that cannot be simply expressed in attenuated forms such as ‘‘equality of
opportunity’’. As a specific dimension of social cohesion, it rather calls for
reducing inequality of conditions. The legitimacy/illegitimacy dimension
under the Political domain refers to how adequately the institutions (such as
the government, political parties, and unions) represent the people. Partic-
ipation/passivity under the same domain relates to people’s involvement in
governance or in politics. The recognition/rejection dimension under the
Social domain recognizes the virtue of pluralism, while the dimension
belonging/isolation relates to shared values or sense of being part of a
community (Jenson, 1998). While these six dimensions are theoretically
interesting and meaningful, the measurement illustrated in this paper
focuses on the three broad domains (Economic, Political and Social) for two
reasons. First, the survey data that we use do not have all the information
needed as indicators of these six dimensions. Second, these six dimensions
are themselves interrelated, which can lead to problems in statistical
discriminant analysis.
Having mentioned the relevant multidimensional aspects of social cohe-
sion, let us now turn to its multilevel aspect. Measures of cohesion for the
nation as a whole may be interesting and useful especially for cross-national
comparisons; in fact, many studies do that either in studying social cohesion
or related ideas like social capital. Putnam’s (1995) ‘‘social capital’’, for
example, is measured for the whole country [but see Portes’ (1998) criticism
of the measures used for social capital]. Similarly, the indicators suggested
by Thomas (1999) are calculated for the country as a whole, using different
time points. However, it would seem suitable to measure cohesion at the
‘‘community’’ level, as communities are where people live, share, and engage
in day-to-day activities. But ‘‘community’’ or ‘‘neighbourhood’’ is another
social construct that is difficult to identify based on geographic maps. Thus,
people in the same geographic area may have different ‘‘communities’’ or
‘‘neighbourhoods’’ that are meaningful to them; or neighbourhoods and
communities can span over, and slice across, two or three geographic areas.
The literature also debates whether space and geographic closeness are
essential features of ‘‘communities’’.
Inclusion Equality Legitimacy Participation Recognition
Exploratory Factor Analysis and identify major indicators and loadings
Confirmatory Factor Analysis and identify major indicators for each domain
Economic Political Social
Domain Domain Domain
Index Index Index
(Ranked) (Ranked) (Ranked)
Standardize all distributions with weights for each domain
Suggested weight: Economic - 40% Political and Socio-cultural - 30% each
Overall Index of Social Cohesion (ISC) for each CMA
Political Socio-cultural
Fig. 1. Methodology used for construction and analysis of indicators of social cohesion.
One way of capturing ‘‘communities’’ in a meaningful way is to consider
the smallest possible geographic areas. Surveys do not usually measure such
small geographic areas. Using the census enumeration areas (EA) is one
possibility, although it has no intrinsic meaning besides that of units that are
convenient for enumerators. There are about 44,000 enumeration areas in
Canada. Our initial work at this level soon ran into the problem of small
numbers in many EAs, not only with the survey data but also with the
census data. The next higher geographic level is census tracts (CTs). There
are about 4400 CTs in Canada, and missing data could be handled through
imputation methods. However, the problem with the survey data still exists.
[See Myles et al. (2000) for an example of using CTs as ‘‘neighbourhoods’’.]
The Census Metropolitan Areas (CMA) is a feasible unit of analysis as each
has administrative and other features of a ‘‘community’’ distinct from all
others. The CMAs however may not represent the ‘‘true community’’ of the
residents, especially since the CMA sizes can vary greatly. Yet, we show in
this paper a measurement of social cohesion for this level mainly because the
data needed as indicators of its domains are available from a survey. We
hope future surveys will be able to provide the needed data at a more
meaningful level.
In computing the indicators of social cohesion, we set the following criteria
for the data: (a) timeliness; (b) measures should be statistically robust; and
(c) indicators should directly or indirectly measure a major aspect of
cohesion. While we have sought data from various sources, the data col-
lected through the National Survey of Giving, Volunteering, and Partici-
pating (NSGVP) seem to be a good starting point to measure social
The NSGVP, conducted in 2000, collected information from 14,724
respondents residing in Canada excepting those from the Territories and
residents of institutions. This study focuses on the CMAs since these are
identifiable units with sufficient number of cases. The survey covered 64
CMAs with a total of 8374 respondents. However, to meet the criterion that
the data should allow the computation of statistically robust estimates, we
limit our analysis to CMAs with 30 or more respondents. This reduces our
sample to 8093 respondents from 49 CMAs that make up our units of
analysis (see Table I for details).
As for the third criterion, the main aim of the NSGVP was to gather
information on giving, volunteering, and civic participation, which are all
Number of respondents by census metropolitan areas, by province – 2000 national survey of giving, volunteering, and participating
Newfoundland Quebec Manitoba British Columbia
St. John’s 153 Chicoutimi-Jonq 141 Winnipeg 573 Vancouver 371
Corner Brk-Deer Lk 46 Que
´bec 172 Brandon 48 Victoria 153
Total CMA 199 Montre
´al 413 Total CMA 621 Kelowna 50
Non-CMA 394 Hull 126 Non-CMA 554 Kamloops 21
Total Province 593 Trois-Rivie
`res 128 Total Province 1175 Matsqui 67
Sept-Iles 35 Chilliwack-Hope 30
Baie-Comeau 46 Nanaimo 16
Prince Edward Island Rimouski 11 Prince George 32
Charlottetown 116 Sherbrooke 161 Saskatchewan Dawson Creek 6
Summerside 69 Ryn Nrnda/ValDOr 45 Regina 265 Total CMA 746
Total CMA 185 Total CMA 1278 Saskatoon 278 Non-CMA 394
Non-CMA 252 Non-CMA 1090 Moose Jaw 55 Total Province 1140
Total Province 437 Total Province 2368 Prince Albert 73
Total CMA 671 Total respondents in:
Ontario Non-CMA 680 Canada 14724
Nova Scotia Ottawa 267 Total Province 1351 Non-CMA 6350
Halifax 257 Sudbury 261 CMA 8374
Sydney-SdnyMines 88 Toronto 687 CMAs with < 30 resp. 281
New Glasgow 25 Hamilton 219 CMAs with 30+ resp. 8093
Truro 16 St. Cath-Niagara 220 Alberta
Total CMA 386 London 253 Calgary 306
Non-CMA 670 Windsor 165 Edmonton 287
Total Province 1056 Kitchnr-Waterloo 251 Lethbridge 35
Thunder Bay 221 Medicine Hat 29
Oshawa 249 Red Deer 35
Cornwall 18 Grande Prairie 15
New Brunswick Kingston 45 Fort McMurray 21
St. John 151 Peterborough 11 Total CMA 728
Bathurst 29 Guelph 89 Non-CMA 461
Chatham-Newcast 23 Brantford 56 Total Province 1189
Moncton 138 Sarnia-Clrwater 42
Fredericton 57 Sault Ste. Marie 25
Edmunston 15 North Bay 68
Total CMA 413 Total CMA 3147
Non-CMA 482 Non-CMA 1373
Total Province 895 Total Province 4520
indicators of a specific dimension of social cohesion, namely participation
under the Political domain. Nonetheless, there were also questions related to
the other dimensions such as on voting behaviour (legitimacy)
, labour force
participation and income (inclusion and equality), and socialization and
ethnicity (belonging and recognition). Information from these questions
provides the following variables measured either as proportions or measures
of heterogeneity
, estimated from weighted data:
We also examined many other measures (such as proportions employed,
union membership, giving donations, length of stay in community, and age
heterogeneity). But initial exploratory factor analyses helped us to narrow
the list to the above measures that had high loadings on the factors (see
details below).
As for method, we follow the schema presented in Figure 1. This method
actually extends the methods used for computing the Indices of Deprivation
2000 in England (Department of the Environment, Transport and the
Regions, 2000). Our idea is to create an overall index of social cohesion for
each CMA. This overall index, however, needs to be calculated from the
three domain indices, which in turn are to be calculated from the relevant
dimension indices. The procedure therefore seeks to calculate dimension
indices from the set of theoretically relevant indicators available from the
Variables Description Domain-Dimension
Voted – Fed Proportion of people voting in the last federal
Voted – Pro Proportion of people voting in the last pro-
vincial election
Voted – Mun Proportion of people voting in the last
municipal election
Volunteer Proportion volunteering Political- Participation
Civic Part Proportion participating in organizations
Full-time Proportion in full-time job Economic – Inclusion
Tenured Proportion with job tenure
Pincgt20T Proportion with personal income greater than
Economic – Equality
Wkly–Fam Proportion socializing weekly with family and
Social – Belonging
Wkly–Fri Proportion socializing weekly with friends
Wkly–Spt Proportion joining weekly in sports and
recreation with friends
Ethnic Het Heterogeneity measure of major ethnic
Social – Recognition
survey data. It is here the Factor Analysis, both exploratory and confir-
matory, becomes useful, and a brief description of this technique is given
Factor Analysis is a statistical technique used to identify a rather small
number of unobserved ‘‘factors’’ or ‘‘latent variables’’ that represent
relationships among many interrelated metric variables or indicators. In
our case, for example, we have a few indicators such as those listed above
that collectively represent an underlying latent (unobserved or unobserv-
able) characteristic or concept such as legitimacy or equality or belonging.
Similarly, the concept of social cohesion itself is a latent concept that
collectively represents the three domains of interest. Factor analysis finds
the weights or loadings that show the relationships between the indicators
and the latent construct, with larger loadings implying closer relationships.
One can use the technique therefore either to explore the data for any
latent constructs underlying the observed indicators (called exploratory
factor analysis) or to confirm the existence of theoretically established
latent constructs through the available indicators (called confirmatory
factor analysis, elaborated further into what is known as structural
equation model or SEM). We are using both the exploratory and confir-
matory approaches here since we need first to select useful indicators of
the six underlying dimensions as well as to confirm the theoretical rela-
tionships shown in Figure 1.
From the set of indicators available from the survey, the exploratory
factor analysis helps us to select more useful indicators of the six dimen-
sions, and thus to eliminate the redundant ones. As mentioned above,
indicators such as union membership, donations, length of stay in com-
munity have all been proposed in the literature as good indicators of social
cohesion. But these were not identified as good indicators in the exploratory
factor analysis and consequently were dropped from further consideration.
The confirmatory factor analysis or the structural equation model has an
added advantage besides confirming the relationships. It also finds the error
(co)variances between the selected indicators and the underlying constructs
– a feature provided by statistical techniques but rarely clarified in theo-
retical reflections. This specific feature can help us reformulate and refine
our theoretical relationships between the constructs.
The relationships and error (co)variances identified by the structural
equation model can then be used to estimate the latent scores for each
domain. Unlike the factor scores produced by factor analysis, these latent
scores (produced by structural equation modeling) are not orthogonal since
the model suggests some relationships between the domains. These latent
scores are already in a standardized form and therefore will have a mean of
zero and standard deviation of 1. Some scores will be positive and others
negative. Negative scores stand for the least cohesive and positive scores the
most cohesive. In this study, the latent scores for the three domains Social,
Political and Economic range from )2.58 to 1.91, from )3.82 to 1.69, and
from )2.44 to 1.92 respectively (see Appendix Table B).
The latent scores however can have different measures of skewness and
kurtosis for each domain. The skewness measures for the Social, Political
and Economic domains are )0.893, )1.058 and )0.329, and their kurtosis
measures are 0.482, 2.956, and )0.137 respectively. It is necessary, therefore,
to convert them all into one and the same metric, having the same statistical
measures such as mean, standard deviation, skewness and kurtosis (in other
words, they all have a common distribution). The use of a common distri-
bution for all the three domains safeguards against many pitfalls. One such
pitfall, for example, is: while combining the domain scores, a high score in
one domain can be fully cancelled out by a low score in another domain
simply because of the differences in their distributions.
One can transform either the latent scores themselves or their ranks into a
common distribution. We have used the latter procedure and an exponential
transformation as follows:
(a) Domain ranks (R) range from 1 to 49, 1 standing for the most cohesive
(corresponding to the highest positive latent score) and 49 the least
cohesive under that domain. [Note that ‘‘least cohesive’’ does not mean
absence of cohesion.] These ranks can be rescaled to the range of (0, 1)
by computing NR=R/49.
(b) To transform these values into a common (exponential) distribution, we
used the following procedure. For example for the Economic Domain:
Ecotr ¼20 ln½ð1NRÞð1expð100=20Þ
The value 20 stands for the mean of the exponential distribution. Trial
and error will suggest the best value that gives a good exponential shape.
These transformed values – call them exponentials of ranks – range from
0 (strictly 0.41=1/49) to 100, zero standing for most cohesive and 100
least cohesive. This transformation results in a proper distribution that
is common to all domains, with a mean of 20.43, a standard deviation of
20.38, a skewness of 1.853 and a kurtosis of 0.34. The skewness and
kurtosis measures are such that they reduce any ‘‘cancellation effect’’
that will occur when high scores in one domain are combined with low
scores in another.
Results of factor analysis: final model
Panel A: KMO and Bartlett’s Test
Kaiser–Meyer–Olkin Measure of Sampling Adequacy 0.57
Bartlett’s Test of Sphericity Approx. Chi-Square 287.188
Df 66
Sig. 0
Panel B: Factor extraction
Total variance explained
Component Initial eigenvalues Rotation sums of squared loadings
Total % of Variance Cumulative % Total % of Variance Cumulative %
1 3.821 31.838 31.838 2.853 23.773 23.773
2 1.983 16.527 48.365 2.665 22.21 45.983
3 1.65 13.752 62.117 1.847 15.394 61.377
4 1.337 11.142 73.259 1.426 11.882 73.259
5 0.944 7.867 81.126
6 0.562 4.681 85.808
Extraction Method: Principal Component Analysis.
Panel C: Factor loadings
Rotated Component Matrix
12 3 4
Voted in last federal election )0.153 0.910 )0.207 )0.096
Voted in last provincial election )0.199 0.898 )0.159 0.118
Voted in last municipal election )0.035 0.845 0.086 0.156
Civic participation 0.706 )0.099 0.098 )0.219
Volunteer 0.766 0.138 0.173 )0.257
Personal income >20,000 0.014 0.089 0.838 )0.342
Full time 0.003 )0.090 0.686 0.482
Tenured job )0.062 0.202 )0.035 0.763
Ethnc heterogeneity 0.198 )0.333 0.714 0.097
Weekly socializing with family and relatives 0.733 )0.117 0.147 0.489
Weekly socializing with friends 0.747 )0.233 0.07 0.291
Weekly sports and recreation with friends 0.751 )0.228 )0.142 0.008
(c) Finally, the exponentials of ranks for each domain are combined to give
an overall index of social cohesion. There is a practical problem at this
stage in terms of weights to use to combine the domain scores. If one were
to use weights of 40% for the Economic, and 30% each for the Social and
Political Domains, the resultant scores are as given in Table III and
Appendix Table B. The Economic domain is assigned a greater weight
as discussions on social cohesion or inclusion/exclusion predominantly
focus on the economic aspect, with less attention to the social aspect.
Obviously assigning different weights would produce different results.
For the moment we leave the weights as above, although the LISREL
model used in this study suggests otherwise. Given the set of indicators
for the three domains used in this study, the social domain comes out as
more important than the other two, as revealed by the standardized
estimates of the LISREL model, which assigns weights of 46% for the
Social, 19% for the Economic and 35% for the Political Domains. Such
weights, however, may reflect the importance and relevance of the
indicators used in the structural equation model. We need to do more
research with more powerful indicators, but for now we have decided to
stay with the weights of 40+30+30 as mentioned above. Given the
arbitrary nature of these weights, we will focus on the domain scores
that differentiate the CMAs rather than the overall score.
All the above procedures assure that the overall index of social cohesion for
each CMA would be a weighted, exponentially distributed and ranked score,
independent of population size. A specific usefulness of these measures of
cohesion is that they can be used either as dependent or independent
variables in other studies.
4.1. Exploratory Factor Analysis
As seen in Table II, four factors were drawn from the selected indicators.
These factors explain 73% of the relationships between these indicators. The
four factors classify the indicators into the theoretical domains (Social,
Political, and Economic) but the classification does not neatly follow the
theoretically driven six dimensions discussed earlier. Factor 1, for example,
includes civic participation, volunteering and socializing variables, which
combines the political dimension of participation and the social dimension of
belonging. This might suggest that while we distinguish formal involvement
in organizations from informal socializing with family and friends, there
CMA overall rank and rank by major domains
Rank by Domains Overall Rank
Social Political Economic
1 Lethbridge 1 Ryn Nrnda/ValDOr 1 Toronto 1 Hamilton
2 Kelowna 2 Trois-Rivie
`res 2 Kitchnr-Waterloo 2 St. Cath-Niagara
3 Red Deer 3 Summerside 3 Windsor 3 Red Deer
4 St. Cath-Niagara 4 Que
´bec 4 Vancouver 4 Sudbury
5 Summerside 5 Sydney-SdnyMines 5 Matsqui 5 Charlottetown
6 Guelph 6 Sherbrooke 6 Edmonton 6 Fredericton
7 Prince George 7 Kelowna 7 Ottawa 7 St. John
8 Charlottetown 8 Prince Albert 8 London 8 Moose Jaw
9 Sydney-Sdny
9 St. John’s 9 Hamilton 9 Edmonton
10 St. John 10 Baie-Comeau 10 Chilliwack-Hope 10 Victoria
11 Brandon 11 Montre
´al 11 Oshawa 11 Winnipeg
12 North Bay 12 Moncton 12 Fredericton 12 Sept-Iles
13 Saskatoon 13 Charlottetown 13 Red Deer 13 Thunder Bay
14 Calgary 14 Brandon 14 St. Cath-Niagara 14 Kelowna
15 Edmonton 15 St. John 15 Kingston 15 Brantford
16 Hamilton 16 Brantford 16 Montre
´al 16 London
17 Kingston 17 Victoria 17 Sudbury 17 Kitchnr-Waterloo
18 Moose Jaw 18 Chicoutimi-Jonq 18 Calgary 18 Guelph
19 Sudbury 19 Thunder Bay 19 Hull 19 Kingston
20 Sept-Iles 20 Hull 20 Winnipeg 20 Ottawa
21 Kitchnr-Waterloo 21 Fredericton 21 Sarnia-Clrwater 21 Calgary
22 Victoria 22 Sudbury 22 Moose Jaw 22 Chilliwack-Hope
23 Thunder Bay 23 Winnipeg 23 Brantford 23 Regina
24 Regina 24 North Bay 24 Sept-Iles 24 Summerside
25 Moncton 25 Sept-Iles 25 Guelph 25 Windsor
26 Winnipeg 26 Moose Jaw 26 Victoria 26 Prince Albert
27 Fredericton 27 Chilliwack-Hope 27 Thunder Bay 27 Brandon
28 Vancouver 28 Ottawa 28 Sherbrooke 28 Halifax
29 Halifax 29 Hamilton 29 Charlottetown 29 Montre
30 Prince Albert 30 Halifax 30 Regina 30 Oshawa
31 Matsqui 31 Regina 31 Halifax 31 St. John’s
32 London 32 St. Cath-Niagara 32 St. John 32 Moncton
33 Brantford 33 London 33 Corner Brk-Deer
33 Matsqui
34 CornerBrk-Deer
34 Red Deer 34 Lethbridge 34 Toronto
35 Oshawa 35 Sarnia-Clrwater 35 Saskatoon 35 Sherbrooke
36 Toronto 36 Windsor 36 St. John’s 36 Sarnia-Clrwater
37 St. John’s 37 Guelph 37 Prince Albert 37 Vancouver
38 Ottawa 38 Edmonton 38 Chicoutimi-Jonq 38 Lethbridge
39 Windsor 39 Kingston 39 Kelowna 39 Chicoutimi-Jonq
40 Chicoutimi-Jonq 40 Oshawa 40 Baie-Comeau 40 Hull
41 Chilliwack-Hope 41 Kitchnr-Waterloo 41 Moncton 41 Saskatoon
42 Sarnia-Clrwater 42 Calgary 42 Brandon 42 North Bay
could be an underlying (unobserved) phenomenon common to both
dimensions, which is captured by this factor. For practical purposes, we can
call this factor Social Domain.
Factor 2 mainly consists of voting variables, which represent the political
dimension of legitimacy, according to which the democratic exercise of the
right to vote leads to institutions representative of the people. We take this
factor to represent the Political Domain in the subsequent steps of the analysis.
We initially considered the variable Full-time Job as an indicator of the
economic dimension of inclusion, Proportion Tenured and Personal Income
greater than $20,000 as indicators of equality, and Ethnic Heterogeneity as
an indicator of the social dimension of recognition (see section 3 on Data
and Methods). But as shown in Table II, Factor 3 has high loadings on Full-
time Job, Personal Income and Ethnic Heterogeneity while Proportion
Tenured falls separately under Factor 4. Both factors 3 and 4 clearly capture
an economic domain although which factor represents the dimension of
equality and which one represents inclusion is difficult to tell. In addition, the
fit between Ethnic Heterogeneity and the other economic indicators (which
is confirmed in further analysis below) is unexpected. This tells us an
important point for theoretical development. While we think of recognition
(or the related concept of tolerance for pluralism) as a social domain, its
outcome is mainly to be seen in the economic domain.
4.2. Structural Equation Model
Confirmatory factor analysis (using structural equation modeling) assured
the usefulness of the indicators selected through the exploratory procedure.
Rank by Domains Overall Rank
Social Political Economic
43 Ryn Nrnda/ValDOr 43 Toronto 43 Summerside 43 Ryn Nrnda/ValDOr
44 Sherbrooke 44 Matsqui 44 Ryn Nrnda/
44 Sydney-SdnyMines
45 Montre
´al 45 Lethbridge 45 Que
´bec 45 Baie-Comeau
46 Trois-Rivie
`res 46 Saskatoon 46 Prince George 46 CornerBrk-DeerLk
47 Hull 47 Vancouver 47 North Bay 47 Que
48 Baie-Comeau 48 CornerBrk-Deer
48 Sydney-SdnyMines 48 Prince George
49 Que
´bec 49 Prince George 49 Trois-Rivie
`res 49 Trois-Rivie
The results from the LISREL path diagram are summarized in Figure 2.
The LISREL model on which the diagram is based has a good fit, judging
from the goodness of fit parameters
for the model.
Figure 2 shows that the three indicators of voting behaviour capture very
well the Political domain, with voting in provincial elections standing out
very clearly with reliability (R
) value of 0.95. The Social domain is captured
moderately well by the socializing variables (with R
values around 65–70%)
but there is much to be wished with the other indicators like civic partici-
pation and volunteering which have large error variances. The indicators of
the Economic domain are also somewhat weak (with R
values hovering
around 25–30%). In general, however, the LISREL model shows that all
these indicators are good, and most of them have significant effects on their
respective domains. However, they are not sufficient in the sense that most
of their error variances are also significant, thus calling for more powerful
indicators than what we have here.
LISREL ML Estimates - Measurement
VOTED_FED = 0.081*Politica, Error var.=
VOTED_MUN = 0.074*Politica, Err orvar.
FULLTIME = 0.043*Economic, Error var.=
CIVIC PART = 0.038*Social, Errorv ar.= 0.
WKLY_FAM = 0.099*Social, Err orvar.= 0.
WKLY_SPT = 0.044*Social, Er rorvar.= 0.
Voted - Fed
Voted - Pro .092*
Voted - Mun
Full-tim e .043*
Tenured 0.094*
Ethnic Het
Civic Part
Volunteer .038*
Wkly - Fam .099*
Wkly - Fr i .044*
Wkly - Spt
0.0015 ,R2 = 0.81 VOTED_PR0 = 0.092*Politica, Er rorvar.= 0.00048 ,R2 = 0.95
= 0.0047 ,R2 = 0.54 PINC>20 = 0.040*Economic, Errorvar.= 0.0033 ,R2= 0.32
0.0053 ,R2 = 0.26 TENURED = - 0.026*Economic, Errorvar.= 0.0053 , R2 = 0.11
0047 ,R2= 0.24 VOLUNTEER = 0.038*Social, Errorvar.= 0.0045 ,R2 = 0.25
0055 , R2 = 0.64 WKLY_FRI = 0.073*Social, Er rorvar.= 0.0022 ,R2 = 0.71
0023 ,R2 = 0.45 ETHNIC HET = 0.023*Social + 0.094*Economic, Errorvar .= 0.0039 ,R2
-.002 .003*
.004* .002*
Fig. 2. LISREL model of indicators of social cohesion.
Other findings through the exploratory factor analysis are confirmed by
the LISREL model (Figure 2). Thus, Ethnic heterogeneity is shown to be
related to both Social and Economic domains with an R
value of 71%,
although the path coefficient connecting it to the Social domain is not
significant at 5% level. The model also reveals what cannot be seen in
traditional factor analyses; that is, there are significant error covariances
between different indicators, for example, between Civic Participation and
Volunteering, and between Full-time Job and Tenured Job. These covari-
ances make a lot of sense. The LISREL model points to possible improve-
ment in the model by adding covariances between certain indicators (through
the so-called modification indices). For example, it points out a significant
covariance between Full-time Job and Voting in Federal elections, though
why this is so is not clear a priori. Similarly, LISREL points to a possible
(although nonsignificant) covariance between Ethnic heterogeneity and Full-
time Job. These points, subtle as they are, need further reflection in theorizing
the relationships between the various dimensions of social cohesion.
4.3. The Domain Scores and Ranks
The LISREL model allows estimate of latent scores for the three domains of
social cohesion. As pointed out in the previous section, these scores are
turned into ranks, ranks into exponentials of ranks, and finally into an
overall index of cohesion
, which are all provided in Appendix Table B.
Combining the three domain scores (indices) into an overall index of
cohesion for each CMA was done by averaging the domain scores with
weights of 30% for the Social and Political domains and 40% for the
Economic domain. It needs stressing here that the domains’ exponentiated
scores are proper distributions and therefore they matter more than the
ranks and should be used for further analysis. The ranks of these scores may
be helpful for interpretation and comparison of CMAs, but they should not
be used for further analysis of the determinants and consequences of social
Table III presents the ranks of CMAs under each domain and the overall
ranks. The first rank in the Social domain (meaning the most ‘‘cohesive’’ in
that domain) is held by Lethbridge, followed by Kelowna and Red Deer, all
fairly small CMAs. Of the top ten CMAs in the Social domain, four are
from the Atlantic region (Summerside, Charlottetown, Sydney-Sydney
Mines, and St John) with none from Quebec. In contrast, the first and
second rank in the Political domain are both in Quebec (Rouyn-Noranda/
ValDOr and Trois Rivie
`res) with three more in the top ten (Quebec,
Sherbrooke, and Baie-Comeau) but none from Ontario. However, Ontario
CMAs dominate the Economic domain with Toronto in the first rank and
with five others in the top ten (Kitchener-Waterloo, Windsor, Ottawa,
London, and Hamilton). British Columbia is well represented as well with
three CMAs in the top ten (Vancouver, Matsqui, and Chillliwack-Hope).
The lowest ranked CMAs within each domain include a predominance of
cities from Quebec in the social domain, while those lowest on the political
domain tend to be from the West plus Toronto. While the top ranked cities
on the economic domain tend to be the larger cities that are west of Quebec,
those ranked lowest tend to be smaller cities, or from Quebec, but not
Montreal. Thus, the domain ranks are strikingly clustered in the provinces,
which brings to the fore the significant differences that exist in the provinces,
economically, socially and politically.
As for the overall rank, in general, we normally expect those CMAs that
rank high in all three dimensions to have higher overall ranking. That,
however, is a rare phenomenon in our data set. And, the method followed in
this study has specifically taken account of possible cancellation effects that
will arise in the case of a CMA that has a high rank in one domain but a low
rank in another. Consider, for example, Toronto which holds the first rank
in the Economic domain but only the 43rd rank in the Political domain and
the 36th rank in the Social domain. The technique of using a common
distribution for all domains (with the same skewness and kurtosis measures)
reduces this anomalous effect as much as possible. It also indirectly produces
the result that the CMAs which have ‘‘average’’ ranks in all domains move
up in the overall index, depending also on the weights associated with each
domain. As shown in the last column of Table III, the CMAs that rank high
overall do not necessarily rank high in all three domains and the largest
cities tend not to be among the top or bottom ranked cities. For example,
Hamilton, the highest ranked CMA overall, holds the 16th in the Social, the
29th in the Political, and the 9th in the Economic domain. Most CMAs
holding the top ten ranks overall have moderately high ranks in at least two
domains. This can be taken to mean that to be at the high-end in overall
ranking of cohesion, CMAs must have better than average ranking on at
least 2 dimensions.
The above ranks are all based on relevant data available from the NSGVP
and derived from models that we have used. Needless to say, the ranks will
be different with different sets of data and with different statistical models of
cohesion. We think, however, there is no better statistical model than one
that adopts the latent construct approach. There is also a need to collect
more meaningful and refined indicators of those latent constructs.
As the title of this paper implies, our main aim was to try out a measure of
social cohesion that considers the multidimensional aspect of the concept.
This was done using only one source of data, the Canadian National Survey
of Giving, Volunteering, and Participating conducted in 2000. We think the
experiment succeeds to an extent, so we can make a tentative substantive
interpretation of the results.
A look at the separate results of each of the three domains – economic,
social, and political – brings out some of the commonly known impressions
about the different regions of the country. Many cities in the province of
Ontario, particularly Toronto, have a strong economy, as do the other two
big metropolitan areas of Vancouver and Montreal in the provinces of
British Columbia and Quebec respectively. They all have high ranking in the
economic domain. The CMAs in the Atlantic Provinces, on the other hand,
are generally seen as places where communities are closely knit. This is
revealed in their prominence in the social domain, but their economy is not
as good as in Ontario, Alberta, or British Columbia. And, the long-standing
issue of separation of Quebec from Canada may have politicized its resi-
dents more than in other parts of Canada. It is no surprise, then, that many
cities in Quebec rank high in the political domain.
The multidimensionality of the concept of social cohesion calls for a
measure that combines all three domains together. The regional differences
in Canada are certainly reflected in the domain scores of the CMAs from
each region. And, the end-result of the overall ranking is such that CMAs
with the highest ranks are generally small CMAs that hold a moderate to
high rank in at least two domains. What this means is that high level of
cohesion requires a balancing of the three domains. This could be seen as
similar to a point made by Bernard (1999). That is, in a democracy, a
distortion of social order occurs when one or two elements of social cohe-
sion, namely liberty, equality, and solidarity, are neglected.
The overall measure also shows the highest ranking CMAs are well
scattered in different parts of Canada – Hamilton, St. Catherines-Niagara,
and Sudbury in Ontario; Red Deer, Moose Jaw, Edmonton, and Victoria in
the West; and Charlottetown, Fredericton, St. John in the Atlantic Prov-
inces. This makes it clear that no one region can claim to be more conducive
to social cohesion than any other.
An examination of domain scores reveals another interesting point: No
CMAs rank high in all three domains. This suggests that CMAs differ in
their base on which social cohesion is (and can be) built. When weak in one
domain, they compensate by being strong in another. Thus, in the CMAs
that rank high in the social dimension like those in the Atlantic region,
people may need to band together to make up for their economic disad-
vantage. Conversely, in economically strong CMAs such as big metropoli-
tan areas of Toronto or Vancouver, people may not have a compelling need
for strong social ties or political involvement. This ‘‘compensation effect’’
results in CMAs not greatly polarized, which would have been the case had
some CMAs ranked very high in all three domains and others very low in all
three domains. The resulting balance among the CMAs possibly contributes
to the cohesion of the country as a whole.
While these substantive discussions of the results are reasonable, the
social cohesion measure used here has a number of limitations, the most
conspicuous one being lack of information on the ideational aspect, an
important component of social cohesion as discussed in an earlier section. A
few of these limitations are discussed in the concluding section.
Making use of data from a major Canadian social survey and a social
cohesion paradigm developed by Canadian researchers, this study shows the
validity of some of the important aspects of the paradigm as well as the utility
of the procedures and models. Both the exploratory and confirmatory factor
analyses highlighted the multidimensionality of social cohesion that
encompasses the economic, social, and political domains. However, clear
distinctions between the associated six dimensions proved to be difficult to
validate. In the political domain, for example, volunteering and association
membership indicative of political participation did not statistically fit in with
voting behaviour. Rather, they fitted in better with the socializing variables
taken to represent the socio-cultural dimension of belonging. Similarly, while
ethnic heterogeneity is generally assumed to be related to the social domain,
it is positively and more strongly related to the economic domain.
The results point to the need for more refined conceptualization of the
complex relationships among the various dimensions of social cohesion. The
confirmatory factor analysis provides a good start as it presents, for
example, interrelationships between domains and dimensions (see curved
arrows in Figure 2). The analysis also shows the need for better indicators of
the dimensions. Particularly missing in our analysis are ideational (as
opposed to relational) indicators. In Stanley’s (2003) definition (quoted in
the introductory section), the ‘‘willingness to cooperate’’, for example, calls
for an ideational indicator. Certainly, indicators of economic dimensions
require ‘‘hard’’ data such as the indicators that we have used (income,
employment), but economic inclusion connotes certain attitudes as well (for
example, attitudes towards immigrants as co-workers). And, the socio-cul-
tural dimension of recognition (or related concept of tolerance) is more
attitudinal than behavioural. Thus, it is possible that a strong sense of
belonging, measured here by frequencies of socializing with family and
friends, may be accompanied by low tolerance for diversity, which however
can be measured only by attitudinal variables.
Even if surveys like NSGVP collect ideational variables in future surveys,
we would still need to combine data from different sources in order to
provide a holistic picture of social cohesion. However, data linkage assumes
that we have found the level that best corresponds to our concept of
‘communities’ or ‘neighbourhoods’. This study has considered the level of
CMAs, which is not ideal but justifiable given the limitations of social
surveys, not to speak of problems of anonymity and confidentiality. After
all, CMAs are entities, each characterized with distinct economic, political,
and social features. But concentrating on CMAs leaves out the rest of the
country – the nonCMAs and, in this study, very small CMAs. Also, CMAs
vary greatly in size, and size is correlated with dimensions of social cohesion.
It is imperative that we define a level of aggregation that is not too disparate
in size, more inclusive, and yet would not pose an excessive problem either
in data collection or in preserving the confidentiality requirement.
The usefulness of a study such as this lies not so much on the ranking but
on the latent scores generated by the model. The latent scores can be used to
examine the impact of social cohesion on other outcomes such as population
health or the well-being of children and youth. It would also be possible to
examine the effect of many other processes like market penetration, aging
and family change on social cohesion.
Observed measures of variables by CMAs
Volunteer Per. Inc.
Full-Time Tenured Ethnic
Fam Soc
Fr Soc
St. John’s 0.845 0.828 0.728 0.459 0.306 0.621 0.831 0.424 0.656 0.587 0.389 0.321
Corner Brk-Deer Lk 0.555 0.609 0.497 0.601 0.261 0.539 0.894 0.461 0.677 0.677 0.333 0.428
Prince Edward Island
Charlottetown 0.816 0.817 0.692 0.548 0.338 0.572 0.780 0.386 0.715 0.692 0.400 0.332
Summerside 0.806 0.898 0.709 0.483 0.271 0.420 0.782 0.373 0.668 0.750 0.401 0.261
Nova Scotia
Halifax 0.750 0.710 0.602 0.611 0.324 0.626 0.841 0.320 0.673 0.556 0.457 0.327
Sydney-Sdny Mines 0.911 0.889 0.835 0.505 0.358 0.511 0.886 0.580 0.710 0.762 0.514 0.388
New Brunswick
St. John 0.790 0.793 0.786 0.444 0.354 0.593 0.806 0.365 0.697 0.695 0.459 0.311
Moncton 0.831 0.824 0.597 0.491 0.327 0.528 0.780 0.481 0.677 0.633 0.382 0.298
Fredericton 0.762 0.789 0.579 0.459 0.375 0.702 0.866 0.416 0.817 0.656 0.453 0.268
Chicoutimi-Jonq 0.798 0.727 0.698 0.375 0.218 0.459 0.666 0.426 0.540 0.402 0.191 0.252
´bec 0.855 0.846 0.664 0.473 0.212 0.539 0.770 0.339 0.383 0.238 0.234 0.202
´al 0.778 0.819 0.592 0.353 0.154 0.560 0.800 0.370 0.709 0.364 0.236 0.195
Hull 0.736 0.763 0.659 0.484 0.232 0.591 0.823 0.356 0.616 0.327 0.199 0.184
`res 0.849 0.890 0.637 0.391 0.209 0.481 0.668 0.432 0.432 0.295 0.324 0.270
Volunteer Per. Inc.
Full-Time Tenured Ethnic
Fam Soc
Fr Soc
Sept-Iles 0.731 0.752 0.473 0.444 0.301 0.411 0.860 0.231 0.823 0.513 0.420 0.313
Baie-Comeau 0.776 0.854 0.703 0.375 0.393 0.592 0.873 0.406 0.634 0.333 0.286 0.321
Sherbrooke 0.843 0.827 0.634 0.353 0.221 0.526 0.753 0.388 0.632 0.351 0.193 0.262
Ryn Nrnda/Val DOr 0.917 0.867 0.552 0.486 0.278 0.512 0.668 0.407 0.451 0.341 0.232 0.230
Winnipeg 0.712 0.759 0.640 0.548 0.343 0.567 0.798 0.331 0.755 0.585 0.365 0.320
Brandon 0.819 0.784 0.613 0.548 0.337 0.445 0.684 0.359 0.543 0.646 0.260 0.320
Regina 0.729 0.717 0.512 0.655 0.399 0.577 0.832 0.305 0.678 0.572 0.413 0.313
Saskatoon 0.651 0.629 0.456 0.631 0.413 0.525 0.873 0.359 0.721 0.625 0.429 0.429
Moose Jaw 0.747 0.739 0.623 0.655 0.557 0.614 0.837 0.416 0.763 0.640 0.380 0.287
Prince Albert 0.834 0.840 0.722 0.631 0.345 0.499 0.848 0.485 0.701 0.665 0.286 0.310
Ottawa 0.743 0.700 0.590 0.486 0.313 0.633 0.794 0.355 0.808 0.472 0.360 0.274
Sudbury 0.754 0.756 0.673 0.521 0.275 0.622 0.775 0.402 0.766 0.620 0.384 0.306
Toronto 0.657 0.644 0.603 0.433 0.200 0.624 0.845 0.330 0.875 0.538 0.343 0.253
Hamilton 0.738 0.707 0.541 0.520 0.276 0.599 0.827 0.292 0.739 0.605 0.341 0.325
St. Cath-Niagara 0.696 0.718 0.641 0.433 0.276 0.559 0.804 0.374 0.800 0.721 0.413 0.319
London 0.709 0.699 0.619 0.545 0.329 0.607 0.795 0.370 0.791 0.570 0.312 0.302
Windsor 0.698 0.698 0.619 0.434 0.243 0.604 0.895 0.394 0.838 0.598 0.290 0.202
Kitchnr-Waterloo 0.675 0.665 0.562 0.470 0.271 0.645 0.853 0.293 0.833 0.624 0.356 0.270
Thunder Bay 0.783 0.780 0.722 0.544 0.305 0.576 0.804 0.361 0.738 0.587 0.427 0.290
Volunteer Per. Inc.
Full-Time Tenured Ethnic
Fam Soc
Fr Soc
Oshawa 0.689 0.677 0.640 0.475 0.281 0.642 0.883 0.367 0.796 0.585 0.398 0.348
Kingston 0.701 0.620 0.602 0.433 0.278 0.615 0.655 0.302 0.666 0.499 0.337 0.352
Guelph 0.632 0.728 0.609 0.433 0.300 0.551 0.806 0.362 0.821 0.667 0.478 0.463
Brantford 0.779 0.782 0.720 0.545 0.194 0.507 0.853 0.215 0.734 0.511 0.352 0.387
Sarnia-Clrwater 0.732 0.735 0.735 0.434 0.224 0.606 1.000 0.549 0.770 0.569 0.292 0.337
North Bay 0.722 0.777 0.719 0.544 0.238 0.533 0.794 0.457 0.556 0.724 0.511 0.342
Calgary 0.704 0.649 0.531 0.622 0.460 0.646 0.835 0.285 0.751 0.560 0.471 0.343
Edmonton 0.691 0.665 0.565 0.558 0.345 0.634 0.794 0.266 0.805 0.553 0.418 0.313
Lethbridge 0.707 0.550 0.536 0.622 0.500 0.584 0.622 0.280 0.638 0.601 0.427 0.506
Red Deer 0.657 0.715 0.548 0.491 0.329 0.544 0.719 0.306 0.855 0.622 0.438 0.378
British Columbia
Vancouver 0.636 0.550 0.408 0.491 0.221 0.572 0.748 0.373 0.837 0.493 0.351 0.281
Victoria 0.821 0.770 0.552 0.602 0.286 0.578 0.734 0.461 0.788 0.544 0.383 0.326
Kelowna 0.827 0.795 0.594 0.484 0.277 0.402 0.564 0.392 0.717 0.645 0.399 0.324
Matsqui 0.629 0.648 0.499 0.601 0.231 0.598 0.754 0.365 0.805 0.550 0.290 0.333
Chilliwack-Hope 0.783 0.738 0.496 0.548 0.326 0.616 0.934 0.566 0.855 0.542 0.296 0.197
Prince George 0.417 0.421 0.261 0.505 0.279 0.380 0.894 0.523 0.638 0.721 0.581 0.334
Latent scores, ranks, and transformed ranks by CMAs
Latent Scores Rank of Latent Scores Exponentials of Ranks Overall
Social Political Economic Soc Pol Eco Social Political Economic Overall
St. John’s )0.2155 0.8952 ).5363 37 9 36 27.73 4.03 26.17 19.99 31
Brk-Deer Lk
0.0071 )1.8309 )0.3183 34 48 33 23.37 72.23 22.11 37.52 46
Prince Edward Island
Charlottetown 0.8711 0.7908 )0.2128 8 13 29 3.54 6.12 17.73 9.99 5
Summerside 0.9917 1.3491 )0.9473 5 3 43 2.14 1.25 41.06 17.44 24
Nova Scotia
Halifax 0.1501 )0.2757 )0.2472 29 30 31 17.73 18.74 19.80 18.86 28
Sydney-Sdny Mines 0.8444 1.2931 )2.0050 9 5 48 4.03 2.14 72.23 30.74 44
New Brunswick
St. John 0.8176 0.5426 )0.2678 10 15 32 4.53 7.25 20.92 11.90 7
Moncton 0.2613 0.7919 )0.8426 25 12 41 14.14 5.57 35.57 20.14 32
Fredericton 0.1948 0.2023 0.7818 27 21 12 15.85 11.09 5.57 10.31 6
Chicoutimi-Jonq )0.9908 0.4525 )0.6838 40 18 38 33.30 9.08 29.42 24.48 39
´bec )2.5829 1.3049 )1.6602 49 4 45 100.00 1.69 48.65 49.97 47
´al )1.7591 0.8128 0.5950 45 11 16 48.65 5.05 7.84 19.24 29
Hull )2.2178 0.2880 0.3456 47 20 19 61.03 10.40 9.73 25.32 40
`res )1.8153 1.6058 )2.4365 46 2 49 53.90 .83 100.00 56.42 49
Sept-Iles 0.3449 0.0200 0.1215 20 25 24 10.40 14.14 13.33 12.69 12
Baie-Comeau )2.2389 0.8635 )0.7734 48 10 40 72.23 4.53 33.30 36.35 45
Latent Scores Rank of Latent
Exponentials of Ranks Overall
Social Political Economic Soc Pol Eco Social Political Economic Overall
Sherbrooke )1.5969 1.1591 )0.1108 44 6 28 44.50 2.59 16.77 20.83 35
Ryn Nrnda/Val Dor )1.4212 1.6892 )1.5579 43 1 44 41.06 .41 44.50 30.24 43
Winnipeg 0.1992 0.0794 0.3306 26 23 20 14.97 12.56 10.40 12.42 11
Brandon 0.7676 0.7657 )0.9222 11 14 42 5.05 6.68 38.13 18.77 27
Regina 0.2689 )0.3015 )0.2141 24 31 30 13.33 19.80 18.74 17.43 23
Saskatoon 0.6581 )10.3232 )0.4771 13 46 35 6.12 53.90 24.72 27.89 41
Moose Jaw 0.3701 )0.1124 0.1627 18 26 22 9.08 14.97 11.81 11.94 8
Prince Albert 0.0605 0.9140 )0.6237 30 8 37 18.74 3.54 27.73 17.77 26
Ottawa )0.3295 )0.2081 1.0171 38 28 7 29.42 16.77 3.06 15.08 20
Sudbury 0.3476 0.1937 0.4910 19 22 17 9.73 11.81 8.45 9.84 4
Toronto )0.1877 )0.9012 1.9256 36 43 1 26.17 41.06 .41 20.33 34
Hamilton 0.3779 )0.2473 0.8600 16 29 9 7.84 17.73 4.03 9.28 1
St. Cath-
1.0298 )0.3183 0.6485 4 32 14 1.69 20.92 6.68 9.45 2
London 0.0440 )0.3389 1.0090 32 33 8 20.92 22.11 3.54 14.32 16
Windsor )0.4186 )0.4908 1.6757 39 36 3 31.27 26.17 1.25 17.73 25
0.3442 )0.7546 1.8270 21 41 2 11.09 35.57 .83 14.33 17
Thunder Bay 0.2819 0.4518 )0.0720 23 19 27 12.56 9.73 15.85 13.03 13
Latent Scores Rank of Latent Scores Exponentials of Ranks Overall
Social Political Economic Soc Pol Eco Social Political Economic Overall
Oshawa )0.0412 )0.7275 0.8056 35 40 11 24.72 33.30 5.05 19.43 30
Kingston 0.3763 )0.6643 0.6188 17 39 15 8.45 31.27 7.25 14.81 19
Guelph 0.9137 )0.5402 )0.0222 6 37 25 2.59 27.73 14.14 14.75 18
Brantford 0.0322 0.5236 0.1255 33 16 23 22.11 7.84 12.56 14.01 15
Sarnia-Clrwater )1.3412 )0.3832 0.1773 42 35 21 38.13 24.72 11.09 23.29 36
North Bay 0.6737 0.0571 )1.8505 12 24 47 5.57 13.33 61.03 30.08 42
Calgary 0.5408 )0.8538 0.3531 14 42 18 6.68 38.13 9.08 17.07 21
Edmonton 0.5029 )0.6333 1.1100 15 38 6 7.25 29.42 2.59 12.04 9
Lethbridge 1.9088 )1.2242 )0.3574 1 45 34 .41 48.65 23.37 24.07 38
Red Deer 1.1581 )0.3437 0.6923 3 34 13 1.25 23.37 6.12 9.84 3
British Columbia
Vancouver 0.1766 )1.6071 1.3939 28 47 4 16.77 61.03 1.69 24.02 37
Victoria 0.2840 0.4741 )0.0308 22 17 26 11.81 8.45 14.97 12.07 10
Kelowna 1.7540 0.9937 )0.7461 2 7 39 .83 3.06 31.27 13.67 14
Matsqui 0.0483 )0.9375 1.2601 31 44 5 19.80 44.50 2.14 20.14 33
Chilliwack-Hope )1.0411 )0.1974 0.8487 41 27 10 35.57 15.85 4.53 17.24 22
Prince George 0.8922 )3.8246 )1.7429 7 49 46 3.06 100.00 53.90 52.48 48
Descriptive Statistics
Minimum Maximum Mean Std.
Skewness Kurtosis
Latent social )2.583 1.909 0.006 1.009 )0.893 0.482
Latent Political )3.825 1.689 )0.011 1.007 )1.058 2.956
Latent Scores Rank of Latent Scores Exponentials of Ranks Overall
Social Political Economic Soc Pol Eco Social Political Economic Overall
Latent Economic )2.437 1.926 )0.010 1.008 )0.329 )0.137
1.000 49.000 25.000 14.289 0.000 )1.200
1.000 49.000 25.000 14.289 0.000 )1.200
1.000 49.000 25.000 14.289 0.000 )1.200
Expon ranks
social with 20
0.410 100.000 20.430 20.378 1.853 4.199
Expon ranks
political with 20
0.410 100.000 20.430 20.378 1.853 4.199
Expon ranks
economic with 20
0.410 100.000 20.430 20.378 1.853 4.199
Overall Index
9.282 56.418 20.430 10.827 1.801 3.323
One of the most recent noteworthy contributions towards understanding the concept of interest
is The Problem of Solidarity: Theories and Models, edited by Doreian and Fararo (1998). The
central idea of this book is that we need a synergy between theorizing and advanced mathematical
modeling in understanding what cohesion means and how it is related to other social realities.
Legitimacy, which refers to whether or not organizations (usually, political) duly represent
their constituents, is inherently a group attribute. An individual level counterpart of legitimacy
is a basic political right of citizenship – to vote or to select one’s representative in the
government in federal, provincial, or local elections. People in a cohesive society participate
more in the political processes, one of which is the exercise of their right to elect their
representatives in the government. As Stanley (2003, p. 12) says, ‘‘...increased social cohesion
means increased political support for action to produce collective goods...’’, and voting is one
way of expressing such political support.
The heterogeneity measure is computed in this study only for those variables that have three or
more categories (e.g. job types, ethnic groups, etc.); simple proportions are used for
dichotomous variables. In general, the heterogeneity measures, called also qualitative variation,
can be computed as follows:
QV ¼Pijfifj
where f(i)=(weighted) frequency of the i-th category, n=number of categories, and F=total
(weighted) frequency. The measure takes values from 0 to 1, indicating the degree of
Percentile distribution of exponentiated ranks for all domains and of the overall score
Percentile Exp. ranks Overall
5 1.0413 9.6444
10 2.1373 9.9878
15 3.2996 11.9211
20 4.5306 12.068
25 5.8459 12.8586
30 7.2498 14.0071
35 8.7644 14.5397
40 10.3978 15.0798
45 12.1829 17.3355
50 14.1354 17.732
55 16.309 18.8124
60 18.736 19.4252
65 21.5143 20.1417
70 24.7212 20.8336
75 28.5727 24.0416
80 33.3018 25.3203
85 39.5918 30.1616
90 48.6492 36.3495
95 66.6328 51.2222
heterogeneity. QV is highest when the proportions for all categories are equal – for example, in
the case of a trichotomous variable, when the three categories have almost equal frequencies.
This is confirmed by these statistics for the model: Model v
=52.55 with p=0.24, Root Mean
Square Error of Approximation (RMSEA)=0.054, Comparative Fit Index (CFI)=0.93.
As was described in the text, these transformed ranks have the same distribution across all the
three domains with a mean of 20.43, a standard deviation of 20.38, a skewness of 1.853 and a
kurtosis of 0.34. The interpretation of these transformed domain scores is straightforward. For
example, let us consider Toronto. It has scores of 26.16, 41.06, and 0.41 for the Social, Political
and Economic dimensions (recall that the smaller the score, the greater the ‘‘cohesiveness’’).
Thus, Toronto falls 6 points above the mean for the Social domain, but one standard deviation
above the mean for the Political domain, and 20 points below the mean for the Economic
domain (thus holding the first rank). It may be easier to interpret a CMA’s position on a
domain scale by using the percentile distribution (see Appendix Table C). This percentile
distribution holds for all the three domains. Toronto’s Social score, for example, falls near the
75th percentile, its Political score around 80th percentile.
For example, let us consider Hamilton which gets the first rank in overall score. It holds the
9th and the 16th place in Economic and Social dimensions, which are much above the average
rank of 25; its 29th place in the Political dimension is near the average. Thus, a greater weight
attached to the Economic domain pushes it to the top place in overall ranking. In contrast, let
us consider those CMAs which manifest average scores/ranks in all the 3 domains – for
example, Thunder Bay, Victoria, Sept-ıˆ les and Winnipeg. These CMAs have their overall ranks
in the tens, and looking at the percentile distribution of exponentiated ranks, they fall around
the 25th percentile. As a third contrast, consider Halifax. It has greater than average scores in
all three domains with an overall score of 18.86 that places it in the 55th percentile. Quebec
CMA shows the lowest overall ranking because it has scores for Social and Economic domains
falling above the 95th percentile, although it has a score on the Political domain falling below
the 10th percentile.
Berger-Schmidt, R.: 2000, Social Cohesion as an Aspect of the Quality of Societies: Concept
and Measurement. European Union Reporting Working Paper No.14 (Centre for Survey
Research and Methodology (ZUMA), Mannheim).
Bernard, P.: 1999, Social Cohesion: A Critique (Canadian Policy Research Network, Dis-
cussion Paper No.F-09, Ottawa).
Blakeley, R.: 1997, Social Capital and Public Policy Development. New Zealand Minister. As
summarized in the Social Capital Database at the World Bank web site.
Bollen, K. A. and R. H. Hoyle: 1990, ÔPerceived cohesion: A conceptual and empirical exam-
inationÕ, Social Forces 69, pp. 479–504.
Buckner, J.C.: 1988, ÔThe development of an instrument to measure neighbourhood cohesionÕ,
American Journal of Community Psychology 16, pp. 771–91.
Burke, M. and J. Shields: 1999, The Job-poor Recovery: Social Cohesion and the Canadian
Labour Market. A Research Report of the Ryerson Social Reporting Network. Ryerson
Polytechnic University. Available on the internet at
Department of the Environment, Transport and the Regions: 2000, Indices of Deprivation
2000. Available on the internet at
Doreian, P. and Fararo, T. (eds.): 1998, The Problem of Solidarity: Theories and Models
(Gordon and Bleach Publishers, Canada).
Durkheim, E.: 64 1893 (1965), The Division of Labor in Society. Trans. by George Simpson
(The Free Press, New York), p. 64.
Freeman, L.C.: 1992, ÔThe social concept of ‘‘Group’’: An empirical test of two modelsÕ,
American Journal of Sociology 98, pp. 152–166.
Fukuyama, F.: 1995, Social capital: The problem of measurement. Available from the World
Bank Group at
Hirschfield, A. and K.J. Bowers: 1997, ÔThe effects of social cohesion on levels of recorded crime
in disadvantaged areasÕ, Urban Studies 34, pp. 1275–1295.
Jensen, J.: 1998, Mapping Social Cohesion: The State of Canadian Research (Canadian Policy
Research Network, Study No.F-03, Ottawa).
Kearns, A. and R. Forrest: 2000, ÔSocial cohesion and multilevel urban governanceÕ, Urban
Studies 37, pp. 995–1017.
Knack, S. and P. Keefer: 1997, ‘Does social capital have an economic payoff? A cross-country
investigation,’ Quarterly Journal of Economics 1997, 1251–1288.
Lavis, J. and G. Stoddart: 1999, Social Cohesion and Health. Working Paper No. 99–09 of
Centre for Health Economics and Policy Analysis (McMaster University, Hamilton, On-
tario, Canada).
Maxwell, J.: 1996, Social Dimensions of Economic Growth, Eric John Hanson Memorial
Lecture Series, Vol. VIII, University of Alberta.
McCracken, M.: 1998, Social Cohesion and Macroeconomic Performance (Paper presented at
the CSLS Conference on the State of Living Standards and the Quality of Life in Canada,
McPherson, J.M. and L. Smith-Lovin: 1986, ÔSex aggregation in voluntary associationsÕ,
American Sociological Review 51, pp. 61–79.
Moody, J. and D.R. White: 2003, ÔSocial cohesion and embeddedness: A hierarchical concep-
tion of social groupsÕ, American Journal of Sociology 68, pp. 103–127.
Myles, J., G. Picot and W. Pyper: 2000, Neighbourhood inequality in Canadian cities. Paper
presented at the Canadian Economics Association Meetings and CERF Conference, June
2000, Vancouver. Paper available from Statistics Canada at
Mudrack, P.E.: 1989, ÔDefining group cohesiveness: A legacy of confusionÕ, Small Group
Behaviour 20, pp. 37–49.
Putnam, R.: 1995, ÔBowling alone: America’s declining social capitalÕ, Journal of Democracy 6,
pp. 65–78.
Portes, A.: 1998, ÔSocial capital: Its origins and applications in modern sociologyÕ, Annual
Review of Social Sciences 24, pp. 1–24.
Rosell, S.A., : 1995, Changing Maps: Governing in a World of Rapid Change (Carleton
University Press, Ottawa).
Sennett, R.: 24 1998, The Corrosion of Character: The Personal Consequences of Work in the
New Capitalism (Norton & Co, New York), p. 24.
Stanley, D.: 1997, The Economic Consequences of Social Cohesion (SRFA-302, Heritage
Stanley, D.: 2003, What Do We Know About Social Cohesion: The Research Perspective of the
Federal Government’s Social Cohesion Research Network.
Thomas, D.: 1999, Indicators of Social Cohesion in Canada (Statistics Canada, Ottawa).
Woolley, F.: 1998, Social cohesion and voluntary activities: making connections. Conference on
the state of living standards and the quality of life in Canada, Ottawa, Centre for the Study of
Living Standards, October 30–31.
University of Western Ontario, Sociology,
Social Science Centre
N6A 5c2
London, Ontario, Canada
... Systematic monitoring and measurement of social cohesion require the use of adequate methods and reliable statistical data. Social cohesion can be assessed on the basis of primary data (cf., e.g., [2][3][4][5][6][7][8][9][10]) or secondary data (cf., e.g., [11][12][13][14][15][16]). Primary data about social cohesion come from sample surveys, while secondary data are provided mainly by official statistics. ...
... Modern understanding of social cohesion is the result of numerous theoretical and methodological approaches that can be found in the literature ( [1,4,8,41,42,[55][56][57][58][59][60][61]). They represent two main discourses: the academic discourse focuses on sociology and psychology and on the conceptual and analytic understanding of social cohesion [43], with emphasis on integration and social stability ( [6,9,10,57,62,63]). In contrast, in the policy discourse, social cohesion is viewed as a prerequisite of economic well-being ( [1,7,8,41,42,55,56]). ...
Full-text available
Composite indicators play an important role in the analysis of socio-economic phenomena. A number of different approaches to constructing composite indicators have been proposed in the literature. Depending on the degree of compensation, they can be divided into compensatory, partially compensatory, and non-compensatory. The following article focuses on the method of relative taxonomy and its dynamic modification. While this method is typically applied to metric data, the authors propose using the dynamic approach for interval-valued data, which describes objects of interest more precisely. Metric data are of an atomic nature; i.e., an observation of each variable is expressed as one real number. In contrast, each observation of an interval-valued variable is expressed as an interval. By making use of interval-valued data, it is possible to assess objects not only at the regional level but also at a lower level of territorial aggregation, taking into account spatial variation across districts that make up each region. The study described in the article was conducted by applying relative taxonomy in its dynamic approach to interval-valued data in order to measure the level of social cohesion in Poland’s NUTS2 regions during the period 2010–2019. The target dataset was obtained by aggregating numeric data about social cohesion in districts (LAU1) at the level of regions. The lower and upper limit of the interval for each region was based on district-level data and corresponded to the 2nd and 8th decile, respectively (60% of observations), which helped to mitigate the effect of outliers. By applying dynamic relative taxonomy to interval-valued data, it was possible to graphically represent changes in the level of social cohesion that took place across 17 Poland’s NUTS2 regions between 2010 and 2019. It was found that during the reference period, the level of social cohesion in the regions systematically improved. Despite the observed variation, the distance between the regions consistently decreased over time. The level of social cohesion was found to be higher in regions that had received more EU funding to support regional development.
... However, the meso-level indicators of social cohesion are almost absent in this scheme, and only partially present in blocks on voluntarism and donations; social participation and vibrancy of civil society.Chan's et al.'s position is to analyse social cohesion as a societal attribute that should be studied using the state as the primary unit of analysis. Other researchers, e.g.Rajulton (2007), have argued that social cohesion should ideally rather be measured at the local community level where people live their daily lives. In Ukraine, frequently regarded as a country of regions, the regional and local levels are arguably quite important for identity formation and political reference. ...
Focusing on the local community level, this study draws on representative survey data to examine how Ukrainians perceive social cohesion on the basis of objective and subjective criteria. Building on Chan et al.'s operationalisation of social cohesion, we use factor analysis to identify four dimensions of social cohesion: social engagement, connectedness, civic participation and intergroup concordance. Although we find no differences in assessments between the various regions of the country, social cohesion appears stronger in voluntarily formed amalgamated territorial communities compared to other territorial units. We discuss how and why Ukraine differs from other societies as to social cohesion features.
... Social cohesion thus serves as the major starting point for our concept of preferences for EU cohesion. It is both a multi-dimensional and a multi-level phenomenon (Dragolov et al., 2016;Rajulton et al., 2007). Moreover, it is closely related to other notions like social capital, social and political trust, and solidarityall of which are typically used to describe aspects of the foundations and workings of a society. ...
Although polarisation risks breaking up the European family, such conflicts paired with transnational challenges can also raise awareness of the need for more European Union (EU) cohesion. In order to approach the hitherto underspecified concept, this article connects the recent literature on social cohesion with research on various facets of public EU support – integration preferences, identity, voting behaviour, and solidarity. Building on survey data from 10 EU countries taken at the end of 2020, a latent class analysis identifies two distinct groups of preferences for EU cohesion: its friends and its foes. EU cohesion can be achieved by highlighting personal advantages of European integration and by stressing joint approaches to crises, civic education, and democratic processes.
Social cohesion is a multi-dimensional concept referring to social connectedness, or the ‘glue’ that connects members of a society through bonds of solidarity and trust, within and across communities and organizations, and within society at large. The concept of social cohesion continues to garner interest in public and policy circles, perhaps reflecting the intuitive appeal of the concept and the role that cohesion can play in societies’ abilities to respond to challenges, to function effectively, and to support rewarding lives. As a latent concept that is not directly observable or measurable, social cohesion is often measured through key dimensions. In this context, a dimension refers to a constituent part of social cohesion. Using factor analysis and data from Statistics Canada’s 2020 General Social Survey on Social Identity, this study identifies nine key dimensions of social cohesion. Latent class modelling is then used to sort respondents into three latent classes or groups (“Low”, high “Confidence-Belonging” and high “Trust-Participation” cohesion groups) of individuals that share common traits and prioritize certain dimensions of social cohesion. The probabilistic classification of individuals in accordance with latent classes provides valuable insights into social sorting mechanisms and how this extends to cohesiveness within Canadian society.
The article is dedicated to the analysis of the Ukrainian society social cohesion challenges in the pandemic conditions of digitally mediated interactions. Based on the results of qualitative sociological research, as well as a secondary analysis of the results of other sociological studies and statistics data, the author explores the phenomenon of social cohesion in the unity of three social levels of its presence - macro, meso and micro. The study has shown the dialectical nature of social cohesion and the opposite tendencies of its reproduction at different social levels – inner circle, professional communities and society as a whole. In the macro-social perspective, the pandemic has become a «common challenge», intensifying social consolidation at the level of civil society institutions; on the other hand, it has created a new axis of social inequality, deepening local differences and awareness of the different degree of «protection» from crisis conditions. It is concluded that the digitally mediated everyday interactions, characterized by a lack of social actors' co-presence in space and time, contributes to the spread of systemic integration in those social systems for which it was previously uncharacteristic. This, in turn, leads to the «frozen» functioning of social systems. In contemporary digital age, social actors are increasingly beginning to experience «digital loneliness» due to the loss of coexistence in physical time and space and the weakening of binding social capital, making familiar identification patterns less stable and definite. The «shift» of the concentration of interactions from different social circles to the level of the inner circle – the family one sets a double scenario of its reproduction either as a space of cohesive values, the highest level of trust, participation and social security, or as an environment for everyday conflicts. The article emphasizes the urgency of further conceptualization of a new type of capital – digital, which in conditions of systemic integration of social systems «puts into action» the activation of other types of capital.
Full-text available
Promoting social cohesion is particularly critical in South Africa given the context of widening inequality, growing unemployment, and persistent racial inequalities. Without measurement, potential key contributing factors that may influence social cohesion (inequality, poverty, violence, gender conflicts, mistrust, etc.) will remain elusive, thereby complicating the formulation of effective policies and programmes in the local sphere of government towards improved social cohesion and inclusive development (Palmary, 2015; Rocha, Kunc, & Audretsch, 2020). Following a mixed method design, inclusive of a literature review, desktop survey, case study analyses and semi-structured interviews with three cohorts (n = 32), the purpose of this article was to outline the framework dimensions of an evidence-based measurement instrument (i.e., Municipal Social Cohesion Barometer) to gauge the extent to which municipalities succeed in promoting social cohesion. The findings revealed dimensions and indicators for such an instrument and concluded that without a measurement instrument, potential key determinants that influence social cohesion in local communities remain largely hidden, making it difficult to design initiatives aimed at enhancing social cohesion through a more inclusive approach to local development. The relevance of the findings is centred on the utilisation of an instrument for measuring social cohesion that enables policymakers to target key domains that currently limit national unity.
It is no longer possible to avoid the current digital transformation. This opinion seems to be shared by the majority of scholars and practitioners, when considering the recent attention paid to the challenges and opportunities related to this growing and diffused phenomenon. In addition, the industry 4.0 is depicted and considered as providing benefits for employees, for the environment, and for the processes as well as it creates new business venues (Müller et al. 2018).KeywordsDigital transformationTechnologiesWorkforce transformation
Full-text available
Two models of the structural form of small, informal groups are compared. One, derived by Winship, requires that patterns of social affiliation be strictly transitive. The other, based on Granovetter's ideas about weak and strong ties, requires only a special limited form of transitivity. When these alternative models are tested with data on human interaction, it turns out that the Winship model does not fit the data but that the model developed from Granovetter's work does.
We analyze the sex composition of 815 face-to-face voluntary organizations in 10 communities to determine the extent of sex integration produced by voluntary affiliation. The sex segregation in these groups is substantial; nearly one-half of the organizations are exclusively female, while one-fifth are all male. Instrumental organizations (business-related and political groups) are more likely to be sex heterogeneous, while expressive groups are likely to be exclusively male or female. From the point of view of the individual, the typical female membership generates face-to-face contact with about 29 other members, less than four of whom are men. Male memberships, on the other hand, produce contact with over 37 other members on the average, nearly eight of whom are female. Men's contacts are both more numerous and more heterogeneous. We conclude that there is little support for the sex integration hypothesis in these data, although the sex heterogeneity of instrumental groups (especially those which are job-related) indicates that this pattern may change as women move into the labor force in increasing numbers. The paper explores some consequences of segregation for the organizations and the social networks they generate.
Most existing measures of cohesion attempt to objectively measure cohesion while neglecting individual group members' perceptions of their cohesion to a particular group. We propose that group members' perceptions of cohesion are important for the behavior of the individual as well as the group as a whole. We offer a theoretical definition of perceived cohesion which says individuals' perceptions of their own cohesion to a group has two dimensions: sense of belonging and feelings of morale. We test this conceptualization and the adequacy of our Perceived Cohesion Scale in two random samples: students at a small college noted for its strong "school spirit" and residents of a midsized city. Our confirmatory factor analyses indicate a two-factor model, with a high degree of invariance across the two samples, and with the two dimensions correlated over .90 in both. However, as anticipated, perceived cohesion is significantly higher in the college than the city sample.
Journal of Democracy 6.1 (1995) 65-78 As featured on National Public Radio, The New York Times, and in other major media, we offer this sold-out, much-discussed Journal of Democracy article by Robert Putnam, "Bowling Alone." You can also find information at DemocracyNet about the Journal of Democracy and its sponsor, the National Endowment for Democracy. Many students of the new democracies that have emerged over the past decade and a half have emphasized the importance of a strong and active civil society to the consolidation of democracy. Especially with regard to the postcommunist countries, scholars and democratic activists alike have lamented the absence or obliteration of traditions of independent civic engagement and a widespread tendency toward passive reliance on the state. To those concerned with the weakness of civil societies in the developing or postcommunist world, the advanced Western democracies and above all the United States have typically been taken as models to be emulated. There is striking evidence, however, that the vibrancy of American civil society has notably declined over the past several decades. Ever since the publication of Alexis de Tocqueville's Democracy in America, the United States has played a central role in systematic studies of the links between democracy and civil society. Although this is in part because trends in American life are often regarded as harbingers of social modernization, it is also because America has traditionally been considered unusually "civic" (a reputation that, as we shall later see, has not been entirely unjustified). When Tocqueville visited the United States in the 1830s, it was the Americans' propensity for civic association that most impressed him as the key to their unprecedented ability to make democracy work. "Americans of all ages, all stations in life, and all types of disposition," he observed, "are forever forming associations. There are not only commercial and industrial associations in which all take part, but others of a thousand different types -- religious, moral, serious, futile, very general and very limited, immensely large and very minute. . . . Nothing, in my view, deserves more attention than the intellectual and moral associations in America." Recently, American social scientists of a neo-Tocquevillean bent have unearthed a wide range of empirical evidence that the quality of public life and the performance of social institutions (and not only in America) are indeed powerfully influenced by norms and networks of civic engagement. Researchers in such fields as education, urban poverty, unemployment, the control of crime and drug abuse, and even health have discovered that successful outcomes are more likely in civically engaged communities. Similarly, research on the varying economic attainments of different ethnic groups in the United States has demonstrated the importance of social bonds within each group. These results are consistent with research in a wide range of settings that demonstrates the vital importance of social networks for job placement and many other economic outcomes. Meanwhile, a seemingly unrelated body of research on the sociology of economic development has also focused attention on the role of social networks. Some of this work is situated in the developing countries, and some of it elucidates the peculiarly successful "network capitalism" of East Asia. Even in less exotic Western economies, however, researchers have discovered highly efficient, highly flexible "industrial districts" based on networks of collaboration among workers and small entrepreneurs. Far from being paleoindustrial anachronisms, these dense interpersonal and interorganizational networks undergird ultramodern industries, from the high tech of Silicon Valley to the high fashion of Benetton. The norms and networks of civic engagement also powerfully affect the performance of representative government. That, at least, was the central conclusion of my own 20-year, quasi-experimental study of subnational governments in different regions of Italy. Although all these regional governments seemed identical on paper, their levels of effectiveness varied dramatically. Systematic inquiry showed that the quality of governance was determined by longstanding traditions of civic engagement (or its absence). Voter turnout, newspaper readership, membership in choral societies and football clubs -- these were the hallmarks of a successful region. In fact, historical analysis suggested that these networks of organized reciprocity and civic solidarity...
This paper reports on recent research in the Federal Government on social cohesion. Social cohesion is defined as the willingness of members of a society to cooperate with each other in order to survive and prosper. Willingness to cooperate means they freely choose to form partnerships and have a reasonable chance of realizing goals, because others are willing to cooperate and share the fruits of their endeavours equitably. Social cohesion contributes to a wide variety of social outcomes such as health and economic prosperity. Part I of the paper relates how Federal Government researchers came to be interested in social cohesion. Part II discusses the concept of social cohesion. Part III presents a model of society to demonstrate how social cohesion works. Part IV concludes by discussing the policy significance of the model. /// Cette communication rende compte de la recherche récente du gouvernement fédéral au sujet de la cohésion sociale. Par la cohésion sociale, on comprend la volonté des membres de la societé à coopérer l'un avec l'autre dans le but de survivre et prospérer. La volonté de coopérer veut dire qu'ils choissisent librement de s'associer et qu'ils ont une bonne chance de réaliser leurs objectifs, puisque les autres sont également bien disposés à coopérer et à partager équitablement les résultats de leurs efforts. La cohésion sociale contribue à divers résultats sociaux, telles que la santé et la prospérité économique. La première partie de la communication raconte comment le gouvernement fédéral s'est interessé à la cohésion sociale. La deuxième partie explore le concept de la cohésion sociale. Partie trois présente un modèle de la société qui démontre comment fonctionne la cohésion sociale. Comme conclusion, partie quatre parle des implications du modèle pour la politique.