[1] The Executive Summary was prepared by the Saguaro Seminar: Civic Engagement in America, a project of the John F. Kennedy School of Government at Harvard University.  The Executive Summary summarizes the macro trends and findings of the Social Capital Benchmark Survey.  Individual community sponsors with help from local academic partners may have other local interpretations of the data.

[2] We refer to the sponsors as community foundations for shorthand; in actuality, less than a handful of them were other sponsors (generally private foundations).

[3] Social capital refers to value of the social networks embodied in various communities (both geographically and communities of interest), and the trust and reciprocity that flows from those networks.

[4] Throughout this Executive Summary, "white" is used as shorthand for non-hispanic white.

[5] There were a few countertrends of blacks having more "non-family" members treated as family than whites or hispanics, or blacks going to marches and rallies more often than whites, or participating more in religious services, but these trends of lower civic participation and social capital were remarkably persistent.

[6] Unadjusted, along some dimensions there are huge discrepancies in the results, for example: social trust ("do you think most others can be trusted or you can't be too careful") ranges from 36% believing that most others can be trusted in one community to 75% in another; or the percent saying they signed a petition in the last 12 months ranged from 17% in one community to 60% in another.

[7] The members of the Scientific Advisory Committee were Lawrence Bobo (Harvard University Department of Sociology), Xavier de Souza Briggs (Harvard University Kennedy School of Government), Michael delli Carpini (Columbia University Department of Political Science), Michael Dawson (University of Chicago Chairman of the Department of Political Science), Tom Guterbock (U. Virginia), Robert D. Putnam (Harvard University Department of Government and Kennedy School of Government), Wendy Rahn (University of Minnesota Department of Political Science), Robert Sampson (University of Chicago Department of Sociology), and J. Phillip Thompson (Columbia University Department of Political Science).

[8] In most cases, the survey area was one county or a cluster of contiguous counties; some of the community samples are municipalities and others are entire states.  Most of the community surveys called for proportionate sampling, that is, no over- or under-sampling of sub-areas or population groups.

[9] To gauge the extent to which the results of a survey are representative, pollsters commonly use two measures: response rate and cooperation rate.  Both are expressed as percentages.  Response rate refers to the number of completed interviews relative to the estimated number of eligible individuals (or households) whose phone numbers were dialed at least once. [The calculation ignores ineligible phone numbers such as numbers that are no longer working, fax machines, business numbers, etc.] For individuals who were not contacted (due to devices like answering machines and caller ID or due to non-answered phone numbers), the response rate estimates the number of eligible respondents.

The cooperation rate refers to the percentage of completed interviews out of the number of eligible individuals who were contacted.  The cooperation rate is equal to 100% less the percent of eligible individuals contacted who refused to complete an interview (refusal rate).  Because the cooperation rate does not include those who could not be contacted, it is almost always higher than the response rate.  Both are important measures of the quality of the data and results, but some researchers worry more about low cooperation rates because they fear that persons who consciously refuse to participate are more likely to hold different survey-relevant views than those who do. In theory, of course, some of those who "cannot be contacted" may also be consciously avoiding being surveyed (through caller ID, etc.).  In addition, in theory those who are hard to contact may also hold different views from those easier to contact.

The AAPOR (the American Association of Public Opinion Researchers) RR2 formula for response rates is: RR2 = I / ((I + R + NC + O + e(UH)), where: I = the number of completed interviews; R = the number of refusals and terminations; NC = the number of households where the designated respondent was not reached (and there was no explicit refusal); O = other (health or language barriers); UH = unknown eligibility / unknown if household – mostly repeated busy signal or Caller ID block. The proportion of unknowns estimated to be eligible (e) was .25.  In most samples, there was no geographic or race/ethnicity screening, so all adults qualified (incidence = 100%).

In the community surveys where screening occurred (as in the national survey), incidence was less than 100% – requiring an adjustment to make the screened and unscreened sample response rates comparable.  The adjustment consisted of multiplying the sum of the non-response categories in the denominator of the formula [R, NC, O, e(UH)] by the estimated incidence and recalculating RR2. The incidence proportion was calculated as the sum of (the completed interviews plus partial interviews plus terminates) divided by the sum of (the completed interviews plus partial interviews plus terminates plus the number of households screened and determined to be ineligible).

The Adjusted Cooperation Rate uses the same logic as the RR2 Response Rate – only it deletes the NC, O, and e(UH) terms from the denominator.