Abstract
Understanding how civic associations shape individuals' understanding of and participation in local political affairs requires a focus on the structural and affective networks engendered by associations. Using micro-level network data, we analyse the networks developed by two different forms of voluntary associations: neighbourhood councils and faith-based organizations. These associations were chosen to examine common assumptions about the bridging and bonding characteristics of differing associational types and the contention that faith-based associations provide their members with compensating network resources. We find that associations' internal (bonding) networks varied more within type than across types. Moreover, the affective content of relationships play an important contextualizing effect, calling into question the usefulness of common associational typologies. We also find that neighbourhood councils have richer bridging networks, which qualifies the degree to which faith-based associations can compensate for inequitable distribution of political resources.
Acknowledgements
The authors were supported by grants from the National Science Foundation (Technology Research Grant #0112899), the John Randolph Haynes and Dora Haynes Foundation, the James Irvine Foundation, and the Urban Initiative of the University of Southern California. They also wish to thank the tremendous research efforts of all the students involved with the Neighborhood Participation Project and the Civic Engagement Initiative.
Notes
The survey was made available on a password-protected Web site, and board members were contacted by both mail and e-mail requesting that they take the survey. Follow-up requests for participation were made by phone. The survey was also conducted by phone (in both Spanish and English) to capture non-Internet users. The average response rate for these five councils was 59% with a high of 79% and a low of 47%.
The comparisons of internal network characteristics between associational types must be interpreted with caution due to the differences in the survey method. Researchers have found that alternative question structures can have significant impacts on network measures such as density, though centralization measures appear to be more robust (Borgatti & Everett, Citation1999; Brewer, Citation2000; Brewer & Webster, Citation1999; Costenbader & Valente, Citation2003). The differences in the survey structure do not affect external network measures or comparisons within associational types. Because NC respondents could nominate more discussion partners (up to the entire size of their board minus themselves) based on a roster presented to them compared to the FBO respondents who were limited to six free-form nominations, we improved comparability by analysing a subset of NC networks. We selected the six nominations that were rated as the most important contacts because respondents are most likely to identify strong ties in free-form questions. If several contacts had the same rating, we randomly selected among this set. A goal of future work is to develop more robust data gathering methodologies, but for the illustrative purposes of this paper these data suffice.
All of the survey and focus group questionnaires are available from the main author upon request.
The core is identified by an interative process that seeks to maximize the similarity between an observed network and an ‘ideal’ core-periphery in which all members of the core are connected and none of the members of the periphery are connected. Borgatti and Everett Citation(1999) describe the algorithm in detail.
Empirical work has rarely encountered observed densities greater than 0.6.
Knox et al. Citation(2006) provide an insightful analysis and review of the theoretical and empirical issues that arise in such multi-method studies.