Social Capital in a Community Context: A Multilevel Analysis of Individual- and Community-Level Predictors of Social Trust

Abstract

The community structural pluralism model represents the earliest and most enduring research program to conceptualize media effects within a community context. The model, which has intellectual roots in social theory concerned with the impact of population size and density on human interaction, is relevant to the resurgence of interest in multilevel analysis in media research, particularly research regarding the media effects on indicators of social capital such as social trust. The present study uses multilevel modeling to test the linear development model (Berry & Kasarda, 1977) which predicts that community structural pluralism would be negatively associated with individual-level social trust, even while controlling for individual-level predictors including media use and length of residence. Using survey data from 21 communities included in the 2001 Social Capital Community Benchmark Study, the analysis includes a rigorous test of the community structural pluralism model by determining whether the relationship between media use and social trust varies across communities. In support of the linear development model, community structural pluralism was negatively associated with social trust even while controlling for individual-level predictors. Additional findings that challenge traditional assumptions about the role of social trust in political participation are discussed.

Notes

¹Descriptive statistics available upon request from the corresponding author.

Note. The factor analysis was conducted using robust weighted least squares and geomin rotation. The fit of the model is evaluated by the following fit indices: χ²(50) = 307.37, standardized root mean square residual = .027, root mean square error of approximation = .021, comparative fit index = .983. Bold text indicates significant factor loadings.

Note. Cell entries are hierarchical linear modeling coefficients; N = 10,000 for individual-level variables; N = 21 for the community-level variable.

*p < .05. **p < .01. ***p < .001.

Additional information

Notes on contributors

Douglas Blanks Hindman

Douglas Blanks Hindman (Ph.D., University of Minnesota, 1994) is an Associate Professor in the Edward R. Murrow College of Communication at Washington State University. His research interests include the influence of social contexts on communication.

Masahiro Yamamoto

Masahiro Yamamoto is a Ph.D. candidate in the Edward R. Murrow College of Communication at Washington State University, and a Visiting Assistant Professor in the Department of Humanities at Colby-Sawyer College. His research interests include mass media and community social organization, public health, and new media and politics.