Designing a social capital dashboard: study

2. Issues and approaches in measuring social capital

Definitions and dimensions

Social capital as a concept has been of interest to policy makers and researchers for several decades. Initially popularised in the US (particularly by Robert Putnam), social capital theory refers to the notion that the social networks people have, and the values they share are ‘assets’ that can contribute to a range of positive outcomes, from lower crime to better physical and mental health. However, in spite of its relatively widespread currency and use, agreement on exactly what constitutes social capital remains somewhat elusive. Van Deth (2003) argues that “the bewildering number of different aspects, characteristics, indicators or dimensions of social capital makes a common understanding rather unlikely” (2003:81), while Scrivens and Smith (2013) suggest that the term “has been used by different audiences to refer to very different underlying concepts”. Key areas of contention in defining social capital (all of which are relevant to how you then attempt to measure, display and interpret it) include:

• Debate about whether it is a community or an individual level phenomenon. Do individuals have social capital, or does social capital reside within the networks between individuals? If it is a community-level phenomenon, then are survey measures that measure, for example, the number of hours each person spends volunteering, actually capturing it?
• Debates about the relative nature and value of so-called ‘bonding’ and ‘bridging’ capital. Broadly speaking, ‘bonding’ capital refers to links between family and friends –people who are strongly tied and may provide social and emotional support for each other. In contrast, ‘bridging’ capital occurs between more distant contacts (e.g. colleagues), who may have fewer similarities with each other. Putnam has argued that while ‘bridging’ capital is inclusive, linking people from different backgrounds, ‘bonding’ capital can be exclusive, as mutual support may not necessarily extend beyond the particular ‘in’ group.
• What specific dimensions of social capital exist (and whether or not these are actually dimensions of the same underlying concept). Different authors have proposed numerous different dimensions of ‘social capital’. For example, Paxton (1999) identified social trust, trust in institutions and satisfaction with relationships, while Park (2006) added ‘social tolerance’ as a fourth dimension. The OECD’s work on measuring social capital (which formed the basis for ONS’s recent UK-level measurement framework) proposed four broad aspects of social capital, including:
  • personal relationships – referring to the “structure and nature of people’s personal relationships” (OECD, 2013), and concerned with who people know and what they do to establish and maintain personal relationships
  • social network support – referring to “the level of resources of support that a person can draw from their personal relationships” (OECD, 2013), but also including what people do for others on a personal basis
  • civic engagement – “actions and behaviours that can be seen as contributing positively to the collective life of a community or society” (OECD, 2013), including volunteering, political participation and other community actions, and
  • trust and cooperative norms – trust, norms and shared values that shape the way people behave towards each other and as members of society.

A recent internal Scottish Government paper (OCSPA, 2017) also set out four domains. However, while the content of these overlap with the ONS/OECD domains to a considerable degree, there are variations (for example, ‘personal relationships’ and ‘social networks’ have been merged into a single domain, ‘social networks’):

• social networks – “focusses on social networks and considers the role of family, friends and relationships. It is concerned with who people know and what they do to establish and maintain their relationships”
• social participation – “covers volunteering and other types of participation in social and community activities”
• community cohesion – “covers the way people relate to each other within geographical communities. It includes trust and co-operative norms and the shared values that shape the way people behave towards each other and as members of society”
• community empowerment – “covers civic engagement, political participation, and the extent to which people in communities have control and can influence decisions and actions to improve their local area.”

While all these domains or areas are presented as related to ‘social capital’, many authors would argue that they are not necessarily aspects of a single unified concept. Rather the different domains reflect different views of what social capital is, and different underlying research agendas (Scrivens and Smith, 2013). This is particularly relevant when considering whether to try and combine different measures of social capital into a composite measure – if social capital is not a single, unified concept, then attempting to create a single indicator may be unenlightening at best, and misleading at worst.

• Separating ‘form’ from ‘consequence’ – or separating what constitutes social capital, from what follows on as a consequence of it. For example, are volunteer rates actually a measurement of social capital itself, or are they an advantage of having social capital (defined perhaps as social trust or other shared values)?

The fact that there is no agreed definition of social capital has implications for any dashboard that attempts to track it – these fundamental issues around meaning and content will be carried over into decisions of what data is included and views on its appropriateness.

Measurement and interpretation

Three broad approaches to measuring social capital are apparent across the literature, with preferences for particular approaches reflecting people’s position on some of the debates above.

In his early work in the US, Putnam created a single ‘social capital index’ for each US state, comprising numerous variables from the General Social Survey. Each state was given a combined index score, which he then correlated with different outcomes like crime, education, etc. This approach has been criticised both for conflating ‘form’ and ‘consequence’ (by including variables like voter turnout, volunteering etc.), which others view as consequences of social capital, rather than components, and for treating social capital as a single construct when others argue there are multiple dimensions which cannot simply be added together to calculate ‘total social capital’.

In response to the perceived weaknesses of Putnam’s approach, others have developed indices or scales to measure specific dimensions or aspects of social capital. Each scale typically consists of a number of survey items that, when combined, are purported to measure social tolerance, bridging capital, and so on. However, constructing a set of questions all of which are clearly and unambiguously related to the particular dimension of social capital of interest may again be a difficult task. The interpretation of individual questions in terms of what they say about social capital is often far from straightforward – for example, Siegler (2014) notes that while interest in politics is included in ONS’s set of social capital measures, a lack of interest in politics is not necessarily correlated with political disengagement, since young adults in particular can be “disillusioned by traditional politics but concerned by issues covering a broad political agenda” (2013: 12). Combining multiple questions into a smaller number of indices might risk hiding these complexities. Moreover, government researchers do not always have the resources to add several different social capital indices to surveys of a sufficient size and quality to allow robust analysis of social capital at a national level or over time.

In this context, ONS has recently adopted an approach that involves presenting a relatively large number of separate social capital indicators (25), grouped under the four broad domains discussed above, but with no attempt to further reduce them to a single composite measure. Both ONS and OECD note the challenges that nonetheless remain in identifying robust and agreed measures of the various dimensions of social capital, and in interpreting these. For example, ONS include the “proportion of people who have used the internet for social networking in the last three months” as a headline measure of the Personal Relationships dimension. However, they note that the implications of online social networks for social capital are contentious – while they could help build social capital by “allowing people to maintain contact despite being geographically separated”, research has also indicated reduced wellbeing, particularly among young users. In the context of considering a social capital dashboard, these kinds of issues have implications for what guidance is provided alongside particular indicators – what further information will users require to help them make sense of specific measures, and to understand what they reveal about social capital?

Finally, if social capital is viewed as something that exists within relationships and social networks, then quantitative findings alone may not enable a full understanding of how social capital is generated and sustained. For example, Dudwick et al (2006) have argued that wherever possible, quantitative data on social capital should be supplemented by qualitative and participatory methods “to understand the causes and nuances of relationships and the contexts within which they exist”. Presenting qualitative data alongside quantitative measures in an online dashboard raises a number of practical and methodological challenges – how would qualitative data be kept up to date and relevant? Would the geographic coverage of specific qualitative data limit its relevance to wider users? However, there nonetheless might be scope for linking to qualitative resources as a way of providing wider context and a more nuanced understanding of the nature, transmission and significance of social capital in Scotland.

Practical considerations

Leaving aside the philosophical and definitional considerations discussed above, decisions about how to measure social capital are also subject to a number of more mundane practical constraints, including the availability of data covering the dimensions of interest which is both robust and can be analysed at the desired level (e.g. by local authority). ONS set out a number of specific considerations they took into account in determining their set of 25 headline indicators, including that they be:

• robust, and meet the standard statistical requirements of accuracy, reliability and validity
• relevant, easy to interpret and monitor by policy makers and members of the public
• considered acceptable by specialists in the area and draw on well-recognised research work
• based on accessible and consistent information available over time (past and future), and
• available for the UK, whenever possible.

The Scottish Government has already identified a number of potential measures of social capital in Scotland (OCSPA, 2017), which are primarily drawn from the Scottish Household Survey (SHS), Understanding Society, Scottish Crime and Justice Survey (SCJS), and the Scottish Social Attitudes survey (SSA). A number of new questions about social capital have recently been added to the SHS and data will be available from 2019.^[1] In determining which of the measures might be included as indicators in a future dashboard, the Scottish Government may wish to consider:

• Whether there are concrete plans to repeat them at regular intervals (several of the SSA questions, for example, have only been included in single years as part of modules on other topics).
• What level of disaggregation is required, and whether this should be standard across indicators – while SHS data can be presented at local authority level, the sample size for SSA is too small to facilitate this.
• What exactly the question is measuring, and how this is thought to relate to different aspects of social capital – for example, would area differences in feelings of safety when home alone at night actually show different levels of community cohesion, or would this simply reflect differential access to resources that might support both community cohesion and feelings of safety?

The next chapter sets out additional, broader considerations for the Scottish Government as it embarks on developing a social capital dashboard, identified in the review of existing dashboards.