It’s well-known that there are variations between communities in the distribution of voluntary organisations. John Stuart Mill identified these and in recent years the Centre for Social Justice has picked up this theme, albeit in a fairly simplistic way. But what impact does the distribution of organisations have on the pattern of engagement? Are people more likely to volunteer when there’s a stronger organisational base in their community? And does a strong third sector presence have an effect on other elements of community well-being?
To test this you need to be able to link together survey data on what individuals do with administrative data on the distribution of organisations. In the UK this is possible using the ESRC-funded Secure Data Service. Confidentiality constraints normally prevent survey data being released for small areas but the SDS carries out linkage between data in a secure environment and the researcher can then analyse it and release the results, as long as they are non-disclosive (i.e. it is impossible to identify individuals or communities).
In TSRC work as part of a broader European project we have linked together data from the Citizenship Survey on volunteering to information for local authorities in England and Wales on the distribution of charitable organisations. We constructed a number of measures of the presence of charities for local authorities depending on the scale of operation of these organisations (within-local authority up to international) and depending on other characteristics (e.g. were they definitively “active” as revealed by non-zero financial returns to the Charity Commission). This was a fairly extensive and laborious task as there are over 160 000 charities. The literature on social capital and health has many such exercises which use a range of indicators but we think this is the first to use Charity Commission data in this way.
There are large variations between local authorities in the distribution of charities – by up to a factor of 25 depending on how we measure it – so do these variations affect the likelihood of volunteering? In practice most of the variation is due to individual characteristics. Our models first include a long list of individual predictors such as socioeconomic status, education, religious affiliation, etc. We then add in the area-level measure of the distribution of charities. This is the result:
|Charities per 1000 population||No other contextual controls||With the addition of urban/rural indicator, and ethnic heterogeneity||With the addition of urban / rural indicator, ethnic heterogeneity and social deprivation|
|Local registered charities||+***||+*||Not significant|
|Local general charities||+***||+**||Not significant|
|Local active general charities||+***||+**||Not significant|
|Note: * p<0.05; ** p<0.01; *** p<0.001;
So when you have a contextual measure of the distribution of charities the only ones which are significant are those for local organisations – those operating within a single local authority. But these do not survive the addition of other socioeconomic characteristics of the communities – such as deprivation, or ethnic heterogeneity.
What are we to make of this? The finding of a negative effect of urban residence, and of ethnic heterogeneity, on volunteering are both consistent with earlier literature. When social deprivation is added as a contextual measure, however, the effects of the distribution of third sector organisations are no longer statistically significant. This effect holds after controlling for interactions between deprivation and the distribution of charities. So these findings present a challenge to the belief that investment in the third sector infrastructure on its own will have much effect on volunteering rates. There are clearly other individual- and community-level influences that we need to address.