In the Netherlands there is an increasing demand for statistical information related to social capital. Current information at Statistics Netherlands describes aspects of social capital separately instead of integrating information into an overall statistic. Because social capital is a multidimensional concept covering different aspects traditional methods, such as factor analysis, could not be used. The social capital index developed in this paper is comprised of two main dimensions based on a theoretical framework; (1) participation and (2) trust. Each consists of three subdimensions related to (a) social relations, (b) organisations, and (c) politics. The index has two main objectives: (1) monitoring social capital and (2) comparing subpopulations.
The index is based on the Permanent Survey on Living Conditions 2009, administered by Statistics Netherlands, and includes 7560 respondents of 12 years or older. This is combined with registration data of the population. A set of core indicators related to the six subdimensions was determined.
A Structural Equation Model (SEM) with Partial least squares (PLS) has been used as the estimation method to generate the index. This method enables formative measurement models and does not rely on strict data assumptions. In PLS the score of each latent variable or dimension is estimated by optimizing the exact linear combination of the indicators. This maximizes the explained variance of the construct. Various additional analyses, such as bootstrapping, have been conducted to ensure the model is robust and reliable.
Results show that the elderly (65 years and older) and the lower educated lag behind in social capital. Groups with higher shares of social capital are: young people (12-25 years), highly educated people, and Protestants. In addition, the social capital index is significantly and positively related to perceived health and well-being demonstrating nomological validity of the index.