Daily Archives: August 7, 2008

Advances in social capital measurement (UPDATED 4/12/12)

Here is an update on our great progress on social capital measurement.

We should begin with a word about the concept of social capital measurement in general. Since social capital refers to the value of social networks, in principle if you were going to measure social capital, you’d ask everyone to detail all their friends, contacts, acquaintances and then ask them all sorts of questions about these folks (the demographics of each friend, how frequently they contact each person, for what purposes, what they could use these ties for, etc.). It is an interesting approach employed by social networks academics and practical for a business work group, or a university class but far too time-consuming for a city or a country. [One interesting area, on which I have blogged before in "Life In The Network" and "Life In the Network II" is the emerging field of digital traces, where digital footprints like one's e-mails, call logs, locations recorded through GPS/bluetooth devices in cellphones, etc. might collectively reveal our social networks on a grand scale without requiring such detailed surveying. It raises lots of privacy concerns, but it is certainly an area to watch. In principle, one could watch them dynamically change over time, and with demographic information about each person could figure out which links are social bridges across various dimensions or how social patterns differed by demographic groups. Some interesting work of David Lazer has at least found that one can use some of this information to quite accurately gauge who work and social friends are. But these data are not generally available.]

Thus, for now, we have gathered social capital data at the individual level by gathering proxies for social capital: volunteering, religious involvement, neighborliness, trust, participation and leadership in voluntary associations, philanthropy, political participation, etc. For more on the dimensions of social capital, click here. One can then aggregate random individual-level social capital data at a neighborhood, or town or city or state to understand social capital strengths or weaknesses of places and which places have overall greater connectedness. Of course, since there are differential benefits of being in the networks (job leads, lifetime earnings) from the spillover benefits of networks to isolated individuals (lower crime in areas, better performing governments, lower corruption rates, higher public health, etc.), not all the residents in a community with high social capital will necessarily get the same benefits if they are relying on others’ social capital rather than their own.

One of the things we’ve been pushing (given the strong connections of social capital with so many of these public goods), is government measurement of social capital.

The good news is that the US government has started annually measuring social capital on the Current Population Survey (the largest government survey other than the Census). While we’ve been urging this for a while among high level government contacts, the 2 key breakthroughs were a meeting of Robert Putnam with President George W. Bush where he personally committed to make this happen, and then the extremely diligent work of Robert Grimm and Nathan Dietz at the Corporation for National and Community Service, working with the folks at the Bureau of Labor Statistics with background help and reinforcement from the Saguaro Seminar. It is a terrific step forward for policy makers, civic leaders, academics and citizens.  [We're also grateful to the Ford Foundation and a consortium of about 3 dozen community foundations that partnered with us to measure social capital in 2000 and 2006 as the lessons learned from those surveys helped provide the answers to many of the questions that CNCS and the Bureau of Labor Statistics had.]

The US government began measuring volunteering annually (on the Current Population Survey September supplement in 2000), included questions on attending a public meeting and working with neighbors to fix/improve something (starting in 2006), but are now expanding the list by some 20 items starting with Fall 2008. This CPS is the gold standard as far as measurement and has a national sample size of about 57,000 households annually (although they obtain approximately 110,000 responses since they ask about other folks 15+ living in the household). The data is primarily used to construct monthly unemployment rates and has oversamples of larger cities. They plan to ask about volunteering and social capital every year (volunteering on the September CPS supplement and social capital mainly on the November CPS supplement).

As to the questions they are asking beyond volunteering, they ask about attending family dinners, working with neighbors to fix/improve something, attending a public meeting, talking to neighbors, talking to friends/family via the Internet, exchanging favors, and participation in various types of group (school, religious, service/civic, sports/recreation, other). The survey does NOT contain some key social capital items (like religious attendance, generalized social trust, inter-racial trust, subjective wellbeing, etc.). These may be asked in future years, but no guarantees.

Much of this past social capital data has been made available on the Corporation for National and Community Service-sponsored website Volunteering in America [See blog post on that terrific new resource here.] or on the Civic Life in America website.

If you click on Select a City/State and then choose *All*, you can see all the cities that they have enough volunteering data on which to develop reliable estimates. Over the next several years, they will have reliable social capital measurement for similar middle sized cities, states, or regions of states.  This will function, as Robert Putnam calls it, as a “social capital seismograph”, always going in the background, that will be very useful to researchers who want to produce natural experiments:  seeing how baseline levels of social capital affect the ability of two similar communities experiencing different events (a major plant closing or a hurricane or…).

For those of you dealing with smaller level geography (rural areas, cities with populations of under 100,000), I’m not positive that the CPS data even when 3-4 years are lumped together will produce reliable estimates for you. You may, if you want to measure your social capital, have to figure a way to band together with some other community foundations or other local groups to commission social capital surveys in your community and to commission a national survey to compare these data to. You can also always e-mail the Corporation for National and Community Service and make a request for a lower level of geography if they have it. They might be able to provide you with data they have already run but didn’t put on the website.

Three researchers at Penn State University (Anil Rupasingha, Stephen Goetz, and David Freshwater) developed county-level social capital measures that are reasonably good based on the density of civic and non-profit organizations, voting turnout, and census completion rates, among other factors. [You should note however that we found higher correlations, r=.37, between our social capital measures in 2000 Social Capital Community Benchmark Survey and RGF county measures than the Corporation for National and Community Service did in their analysis of their own social capital measures with RGF data at the MSA level.]
– These data are available for 1990, 1997 and 2005.  If you want the RGF data, you can download these county-level data here:

A note to the wise: I would urge that you NOT try to compare local social capital data that you gather to these CPS measures. CPS numbers are typically far LOWER and LESS civic than what you would get in a phone survey (both because the government survey is not about community or civic engagement and because they garner a far higher response rate, they hear more from people who are uninterested in civic engagement). The CPS numbers are probably more accurate but thus hard to compare with what you would get from the phone survey.

If you are interested in doing your own survey, you can, as always, find a copy of our Short Form Social Capital Survey on our website. We ask you e-mail us if you do use the Short-Form so we can keep track of who is using this.

The latest Social Capital Survey we administered was the 2006 social capital community survey. The national benchmark banners (what proportion of total, men, women, etc. gave various answers to the questions) is also available.  We also asked a lot of social capital questions (with lots of questions on religion) in our 2006 Faith Matters Survey, available here.

For more information on social capital measurement in general, visit here.

See related blog post “US expands social capital measures