Abstract
This paper integrates a set of independent social capital indicators into one index by using structural equation modeling
(SEM) based on partial least squares estimation (PLS). The social capital index consists of two dimensions: participation
and trust. In each of the two dimensions, three levels are distinguished: social (micro), organizational (meso), and political
(macro). The main objectives of the index are to: (1) provide a coherent overview of social capital in the Netherlands; (2)
monitor social capital over time; and (3) compare subpopulations. A broad spectrum of indicators is included, however, these
are only weakly correlated and consequently treated as distinct. Therefore, traditional index development methods such as
factor analysis and reflective modeling cannot be applied. Consequently, formative modeling in which the indicators are specified
as causes rather than as effects is used. We employ the Permanent Survey on Living Conditions 2009 (POLS), administered face-to-face
to 7,560 respondents in the Netherlands. We find that the index predicts well-being and health, which demonstrates nomological
validity. Subsequently, bootstrapping is conducted to test the robustness of the index. And the 2009 index is replicated based
on the 2010 POLS data. The replication shows that the 2009 model is relatively stable and results are robust. Finally, the
model is extended by including volunteering and informal help, which changes the model but the results remain largely the
same. This index, which is shown to be valid and robust, contributes to a further understanding of the concept of social capital.
(SEM) based on partial least squares estimation (PLS). The social capital index consists of two dimensions: participation
and trust. In each of the two dimensions, three levels are distinguished: social (micro), organizational (meso), and political
(macro). The main objectives of the index are to: (1) provide a coherent overview of social capital in the Netherlands; (2)
monitor social capital over time; and (3) compare subpopulations. A broad spectrum of indicators is included, however, these
are only weakly correlated and consequently treated as distinct. Therefore, traditional index development methods such as
factor analysis and reflective modeling cannot be applied. Consequently, formative modeling in which the indicators are specified
as causes rather than as effects is used. We employ the Permanent Survey on Living Conditions 2009 (POLS), administered face-to-face
to 7,560 respondents in the Netherlands. We find that the index predicts well-being and health, which demonstrates nomological
validity. Subsequently, bootstrapping is conducted to test the robustness of the index. And the 2009 index is replicated based
on the 2010 POLS data. The replication shows that the 2009 model is relatively stable and results are robust. Finally, the
model is extended by including volunteering and informal help, which changes the model but the results remain largely the
same. This index, which is shown to be valid and robust, contributes to a further understanding of the concept of social capital.
- Content Type Journal Article
- Pages 1-28
- DOI 10.1007/s11205-012-0129-2
- Authors
- Jacqueline van Beuningen, Statistics Netherlands, CBS-weg 11, 6412 EX Heerlen, The Netherlands
- Hans Schmeets, Statistics Netherlands, CBS-weg 11, 6412 EX Heerlen, The Netherlands
- Journal Social Indicators Research
- Online ISSN 1573-0921
- Print ISSN 0303-8300