Community attachment and satisfaction: the role ... - Wiley Online Library

A R T I C L E

COMMUNITY ATTACHMENT AND SATISFACTION: THE ROLE OF A COMMUNITY’S SOCIAL NETWORK STRUCTURE Jessica Crowe St. Mary’s College of Maryland

This paper links the micro and macro levels of analysis by examining how different aspects of community sentiment are affected by one’s personal ties to the community compared with the organizational network structure of the community. Using data collected from residents of six communities in Washington State, network analysis combined with negative binomial regression is used to determine the effect of personal networks and community networks on community attachment and satisfaction. Findings suggest that while individual-level variables, such as length of residence and individual ties, affect one’s attachment to community, a community’s network structure does not significantly affect community attachment. However, a community’s network structure significantly affects one’s evaluation of community. Regardless of one’s ties to the community, residents of cohesive communities are more likely to evaluate the community’s social and physical environments more positively. C 2010 Wiley Periodicals, Inc.

The drivers of local social attachment (e.g., community sentiment, sense of community, social participation) have an important research tradition in sociology that date back to the early writings of Toennies (1887/1957) and Wirth (1938). Researchers have long been concerned with the effects of urbanization and industrialization on the social fabric of urban communities (Fisher, 1972; Reissman, 1964; Short, 1971). In this view, urbanization and industrialization transform relationships in society from primary contacts to secondary contacts and local community thus declines. This can lead to a host of social problems as people are less inclined to look out for the best interests of their neighbors. However, more contemporary researchers (Kasarda & Janowitz,

Correspondence to: Jessica Crowe, 18952 E. Fisher Road, St. Mary’s City, Maryland 20686. E-mail: [email protected] JOURNAL OF COMMUNITY PSYCHOLOGY, Vol. 38, No. 5, 622–644 (2010) Published online in Wiley InterScience (www.interscience.wiley.com). & 2010 Wiley Periodicals, Inc. DOI: 10.1002/jcop.20387

The Role of a Community’s Social Network Structure

623

1974; Sampson, 1988) argue that although community-level factors (e.g., urbanization, residential stability) may play a role in creating or destroying social bonds, individual factors, such as length of residence and age, play a more important role in creating community sentiment. Goudy (1990) in extending the work of Kasarda and Janowitz (1974) to rural communities, argues that because social bonds and sentiments are the products of individual choice, as they are in urban areas, then individual characteristics are better predictors of community sentiment than community characteristics, such as size and density. Although individual characteristics, such as length of residence and age, are important characteristics that influence community sentiment, they do not sufficiently account for a community’s atmosphere that may influence individual choices. One key factor that researchers have not thoroughly examined with respect to its influence on community sentiment is a community’s social network structure. Most studies of community sentiment that analyze community-level factors, focus on population and density—factors that Toennies (1887/1957) and Wirth (1938) wrote about as weakening social bonds and others (Goudy, 1990; Kasarda & Janowitz, 1974; Sampson, 1988) have since deemphasized. However, because local community members are embedded within their local surroundings and research shows that a community’s social network structure has an effect on community awareness, capacity for action, (Sharp, 2001) and economic development (Crowe, 2007), it is important to analyze the ways in which a community’s social network structure affects an individual’s attachment to the local community. In this article, I attempt to link the micro and macro levels of analysis by examining how different aspects of community sentiment are affected by one’s personal ties to the community compared with the organizational network structure of the community. Questions remain as to whether a community’s network structure has a significant effect on individual community sentiment and whether or not it affects all types of community sentiment. Drawing on the community sentiment and social network literatures and analyzing data from a survey of 97 community members in six rural communities, I address these questions. COMMUNITY SENTIMENT Research on community sentiment can generally be separated into three broad categories: work focusing on community attachment, on community satisfaction, and on identity and community life (Hummon, 1992). Although there is some overlap among these three categories, distinctions can be made with respect to their conceptualizations of community sentiment, disciplinary roots, and methodological strategies. Therefore, it is important to distinguish the causal mechanisms that enhance the different types of community sentiment. For the purposes of this study, I examine the first and second categories of community sentiment while breaking up community satisfaction into two components: evaluations of social environment (characteristics of people in the community) and evaluations of physical environment (conditions of community spaces). Community Attachment Researchers have studied residents’ relationships to their communities in several different settings and from many perspectives. However, a common strategy has been to analyze residential attachment to a particular community (Goudy, 1990; Hummon, Journal of Community Psychology DOI: 10.1002/jcop

624

Journal of Community Psychology, July 2010

1992; Kasarda & Janowitz, 1974; Liu, Ryan, Aurbach, & Besser 1998; O’Brien & Hassinger, 1992; Sampson, 1988). Community attachment refers to an individual’s commitment to his or her place of residence (Liu et al., 1998). These expressions of commitment can either be subjective (affective) or behavioral (Gerson, Steuve, & Fischer, 1977). Affective community attachment is demonstrated in four different ways: (a) a sense of belonging to the community, (b) a belief that one can have an impact on the community, (c) a feeling that the community can meet personal needs of its members and is satisfying those needs, and (d) expressions of emotional connections with the community and its members (McMillan & Chavis, 1986). Behavioral expressions of community attachment are less clearly defined than affective indicators. Confusion exists over the relationship between affective and behavioral elements. Although some (Beggs, Hurblert, & Haines, 1996; Gerson et al., 1977) claim that both types are distinct indicators of attachment, other researchers (Hummon, 1992; Kasarda & Janowitz, 1974; Sampson, 1988, 1991) limit the definition to include only affective indicators. Behavioral attributes (e.g., organizational participation) are treated as predictors of affective indicators. Interest in community sentiment as attachment has deep roots in community and urban sociology. Concerns about local attachment are linked to a central question that has triggered debate among classical social theorists. Toennies, Marx, Weber, and Durkheim were concerned with how the emergence of modern society would affect social and sentimental bonds. For these social theorists, the transformation to an urban, capitalist society would ultimately lead to a decline in the quality of local community life (Hunter, 1978). Wirth (1938) directly outlines the causes for this concern by arguing that increasing population, density, and heterogeneity associated with urban life lead to fewer ties among individuals, which ultimately leads to weaker emotional attachments to one’s locality. Although the ‘‘community lost’’ thesis still has supporters (e.g., Nisbet, 1976; Webber, 1963), this once dominant perspective has received criticism for relying on deductive, philosophical, or moralistic evidence rather than on empirical research (Lyon, 1999). For instance, ethnographic studies of urban neighborhoods have provided evidence that attachment to the local community is strong (Gans, 1962; Rivlin, 1982). Community attachment studies have shown that broad settlement patterns characterized by community size, density, or type are not strong predictors of one’s attachment to community (Goudy, 1982; Kasarda & Janowitz, 1974; Sampson, 1988). These findings suggest that when analyzing community attachment, one needs to consider the influence of processes other than large demographic patterns. Community Satisfaction Although community attachment is one way to conceptualize community sentiment, an additional way is by studying how community members evaluate their place of residence. A substantial majority of Americans evaluate their communities favorably when directly asked if they are satisfied with their place of residence (Hummon, 1992). For instance, the U.S. Department of Housing and Urban Development (1978) found that three out of four people reported that they were satisfied with their places of residence. Thus, it is important to look at specific predictors of satisfaction as these predictors show more variability. Goudy (1977) asserts that social dimensions are of significant importance in determining community satisfaction. Goudy (1977, p. 380) concludes that ‘‘y Journal of Community Psychology DOI: 10.1002/jcop


625

residents find most satisfying those communities in which they think they have strong primary group relationships, where local people participate and take pride in civic affairs, where decision making is shared, where residents are heterogeneous, and where people are committed to the community and its upkeep.’’ Similarly, Herting and Guest (1985) in their examination of 44 qualities of the local environment find that evaluations of the social environment are one of the most important in accounting for community satisfaction. From these two classic studies and more recent ones (e.g., Filkins, Allen, & Cordes, 2000), evaluation of the social environment is an important indicator of community satisfaction. In addition to evaluations of the social environment, Herting and Guest (1985) find that of the 44 local environment variables, evaluation of the physical environment is the other most important indicator of community satisfaction. Personal Networks and Community Sentiment With their systemic model of community attachment, Kasarda and Janowitz (1974) show that community attachment is influenced by an individual’s local ties. More specifically, length of residence through informal local social ties positively influenced one’s level of community attachment. These informal local ties include acquaintances and friends. However, one’s formal local ties (organizational memberships) did not appear to have an effect on community attachment. By partitioning informal ties into weak and strong ties, Ryan, Agnitsch, Zhao, and Mullick (2005) find that strong ties (close personal adult friends) have a larger positive direct effect on community attachment than do weak ties (acquaintances). However, in contrast to Kasarda and Janowitz (1974), Ryan et al. (2005) show formal ties through organizational membership to have a significantly positive indirect effect on community attachment. Communty Social Networks and Community Sentiment Recent discussions of social capital often distinguish between ‘‘bonding’’ and ‘‘bridging’’ social capital (Putnam, 2000; Woolcock & Narayan, 2000). Bonding social capital is typically characterized as having dense relationships and networks within communities (Taylor, 2004). Bonding social capital acts as the social glue that binds groups together. This is often typified by the existence of tightly woven networks in which members are directly tied to many other members in the network. On the other hand, bridging social capital is often described as the weaker relationships and networks across social groups and communities. It consists of the weak ties described by Granovetter (1986). Bridging social capital is characterized by weak network connections that link groups together in a loose manner. The concepts of bonding and bridging social capital can be useful in viewing a community’s network structure as falling on a continuum ranging from cohesive ties, to loose ties, to factional ties (the lack of connections linking different groups of people). Researchers have shown a community’s network structure to affect a variety of community actions (Crowe, 2007; Sharp, 2001). For instance, Sharp (2001) provides evidence that communities characterized by a pyramidal or hub-like structure are able to generate community-field like capacity. That is, communities with a densely interlocked core clique (pyramidal structure) and communities with one central organization to which others are connected (hub-like structure) are able to mobilize and apply resources, increasing community awareness, and organize local actions. In Journal of Community Psychology DOI: 10.1002/jcop

626


my analysis of community economic development (Crowe, 2007), I argue that communities with a coalitional organizational network structure (typified by dense networks of organizations connected to each other in a nonredundant fashion) are the most likely to implement local economic development. On the other hand, communities with a factional organizational network structure (typified by two or more connected groups that are not connected to one another) are the least likely to implement local economic development. Both studies (Crowe, 2007; Sharp, 2001) acknowledge the influence of a community’s network structure on community actions. With respect to community sentiment, Allen and Dillman (1994) argue that it is important to address the difference in community organization as some communities may retain structures and processes that lead to different levels. Community Sentiment in Rural Communities Rural communities have recently witnessed drastic changes to their economy and population. Although some rural communities, such as those rich in natural amenities, have witnessed population and economic growth, other rural communities have witnessed population and economic decline (Flora & Flora, 2008). Still others have witnessed a switch in economic sectors and increasing diversity of residents (Diaz & Miraftab, 2009). Allen and Dillman (1994) argue that the importance of local community in rural people’s lives can be understood in the relative importance of three distinct eras of social and economic organization—community-control, masssociety, and information eras. The community-control era represents a time in which internal ties within a community are enforced by mutually supportive technological, social, and economic forces. The mass-society era is characterized by extra-local ties that lead to a decline in local community sentiment. The information era links people to the global economy, irrespective of national or community ties. Allen and Dillman claim that for some rural communities, forces consistent with the community-control era may predominate while the other two eras may predominate in other communities. In other words, community structure differs among rural communities as some communities resist forces of one or more eras or actively embrace them. While Allen and Dillman (1994) theoretically conceptualized the need for addressing community structure in explaining the different levels of community sentiment found among rural communities, they focused their study on one community. By sampling individuals in multiple communities, a primary goal of the present study is to examine the effect of a community’s network structure on community attachment and satisfaction while controlling for individual ties. I accomplish this by combining methods of network analysis with negative binomial regression. Hypotheses Consistent with previous research on the effect of individual social ties on community attachment, I expect to find that individuals with many informal and formal ties in their community will have higher levels of community attachment than individuals with fewer informal and formal ties in their community. However, I also expect a community’s social network structure to influence community sentiment. I anticipate that when controlling for an individual’s ties to the community, individuals living in communities characterized by dense network Journal of Community Psychology DOI: 10.1002/jcop


627

Table 1. Name and Broad Description of the Sampled Communities Community Creston Davis Grove Gwenville Heights Mayfield Rowans View Soundberry

Population sizea

White (%)

4,000–5,000 1,000–2,000 8,000–9,000 2,000–3,000 2,000–3,000 2,000–3,000

60–70 80–90 80–90 80–90 80–90 70–80

Primary economic baseb Farming Nonspecialized Nonspecialized Service industry Farming Nonspecialized

Rural typology Nonfederal lands Federal lands Metro Island Nonfederal lands Federal lands

a

Population size and White % provided by the U.S. Census Bureau. Primary economic base and rural typology provided by the Economic Research Service of the United States Department of Agriculture.

b

structures will have greater attachment and higher evaluations of their community than individuals living in communities characterized by loose network structures.

DATA AND METHOD Data for this study are drawn from structured interviews and surveys conducted in six rural communities in Washington, in the summer and fall of 2003. The six communities for this study had previously been chosen to receive a new high school centered around project-based learning. To be considered for the new high school, a community had to be small and rural. As a result, all six communities share a number of characteristics: they are of relatively equal size (between 1,000 and 9,000 residents), have similar levels of racial/ethnic composition, and are rural. The data used for this study are part of a larger project that assessed each community’s social infrastructure prior to the implementation of the new school. Table 1 broadly describes each community on a number of characteristics. To evaluate each community’s organizational network structure, I analyze data from 15 to 34 interviews with local leaders and citizens from each community, with a total of 150 participants among the six communities. Thirty-five percent of the sample consisted of people whom several others in the community identified as being a leader.1 These leaders tended to hold government positions, were very civically active, or both. The remaining 65% were nonleaders who represented one of 20 categories of people that characterized the community. Sixteen of the categories of people were consistent for each community, whereas four wild card slots were made available to fill with people from categories that were unique to each particular community. A list of the categories can be found in the Appendix. A local community coordinator from each community helped identify and recruit the participants. 1 Although the percentage of respondents who are leaders in the community may overrepresent the true percentage of leaders in the community, similar percentages are represented in each community. Thirty-four percent of participants in communities characterized by complete or coalitional network structures are considered leaders compared to 36% of participants in communities characterized by bridging or factional network structures. Regardless, being a leader is not strongly associated with community sentiment. The Pearson correlation coefficients associated with being a leader and each of the three measures of community sentiment are very low (see Table A1). Likewise, leaders are not significantly different than nonleaders in levels of community attachment (t 5 1.22) or how they evaluate the community with respect to social indicators (t 5 .95) or physical indicators (t 5 1.30).

Journal of Community Psychology DOI: 10.1002/jcop

628


Respondents were asked a series of open-ended questions about general community action. The purpose of the interviews was to capture the social network structure of each community. The researcher asked participants to recall all local organizations and government institutions that they belonged to, the number of years they had been a member, as well as all leadership positions that they had held in the previous 5 years. After the interview, participants were given a survey on community perceptions and social infrastructure to complete on their own time and were asked to return the completed survey in the provided stamped envelope. Past questionnaires and surveys used by Flora, Sharp, Flora, and Newlon (1997) and Sharp (2001) in their analyses of entrepreneurial social infrastructure served as the basis for the interview guide and survey. Respondents who were interviewed and responded to the survey were included in this analysis. The final sample consisted of 97 participants (65% response rate). Measurement of Community Attachment The first dependent variable, community attachment, is measured by responses to five questions. Respondents were asked to indicate whether they agreed or disagreed with the following statements: (a) ‘‘Being a resident of (name of community) is like living with a group of close friends’’; (b) ‘‘If you do not look out for yourself, no one else in (name of community) will’’; (c) Most everyone in (name of community) is allowed to contribute to local governmental affairs if they want to’’; (d) When something needs to get done in (name of community), the whole community usually gets behind it’’; and (e) ‘‘Community clubs and organizations are interested in what is best for all residents.’’ The response scales for these items ranged from 1 (strongly disagree) to 5 (strongly agree). Confirmatory factor analysis loaded all items onto one factor with most loadings .6 or higher (considered as high loadings) and only one item with a loading of .4. Because of the unidimentionality of the five word pairs coupled with the internal consistency of the items (Cronbach’s a 5 .67), an additive index was created by coding all of the questions in the same direction and summing the responses to the five statements. Measurement of Community Evaluation Social environment. The first dependent variable under community evaluation, evaluation of the social environment of the community, is measured by responses to five questions. Respondents were asked to imagine a scale that ranged from 1 to 7 for five word pairs. For example, in the first pair, 1 on the scale indicated unfriendly and 7 indicated friendly. For each pair of words, respondents circled the one number that best described the community. The five word pairs included unfriendly/friendly, indifferent/supportive, not trusting/trusting, prejudiced/tolerant, rejecting of new ideas/open to new ideas. Because of the unidimentionality of the five word pairs coupled with the internal consistency of the items (Cronbach’s a 5 .69), an additive index was created by coding all of the questions in the same direction and summing the responses to the five word pairs. Physical environment. The second dependent variable under community evaluation, evaluation of the physical environment of the community, is measured by responses to four questions. Respondents were asked to rate the community spaces in the community: senior citizen center, meeting space in city offices, formal meeting space in Journal of Community Psychology DOI: 10.1002/jcop


629

local restaurants or other businesses, and restaurants and coffee shops. The response scales for these items ranged from 1 (not available) to 5 (very good). Confirmatory factor analysis loaded all items onto one factor with all loadings .6 or higher (considered as high loadings). Because of the unidimentionality of the four items coupled with the internal consistency of the items (Cronbach’s a 5 .71), an additive index was created by summing the responses to the four statements. Independent Measures Demographics. Age is measured in years. Respondent’s sex is coded 1 for female and 0 for male. Respondents were asked how long they had lived in the community. Length of residence is measured in years. Weak informal ties. On a scale of 1 to 5, respondents were asked the proportion of adults living in their community that they knew by name: 1 5 none or very few; 2 5 less than half; 3 5 about half; 4 5 most of them; 5 5 all of them. Strong informal ties. On a scale of 1 to 5, respondents were asked the proportion of all their close personal friends lived in the community: 1 5 have no friends or none live here; 2 5 less than half; 3 5 about half; 4 5 most of them; 5 5 all of them. Formal ties. Respondents were asked to list all the local organizations they belonged to. These included service organizations, recreational and environmental groups, political and civic groups, job-related organizations, and church-related groups. Formal ties is measured as the total number of local organizations a respondent participates. Community Measures Population. Because population size of communities varies from 1000 to 9000, I control for population size (U.S. Census, 2000). Community network structure. To represent community network structure, I use UCINET to analyze interlocking membership among local community organizations and institutions, based on the open-ended surveys. One can either focus on the linkages among organizations created by members or the linkages among members created by organizations. Here I focus on links among organizations created by members. Figure 1 depicts the reduced network structure for each community. Using k-core and cut-point analysis, each community is placed into one of four categories representing the bonding structure of a community. It is useful to examine kcores (Seidman, 1983) to help interpret the level of bonding capital in each network structure. A k-core is a maximal subgraph in which each point is directly connected to at least k other points.2 Thus, an isolate is a ‘‘0-core’’ because the single point is not connected to any other points in the network. Because the current study is interested in bonding and bridging network structures in how they relate to community attachment and evaluation, the analysis of k-cores is an improvement over a measure of density for 2

Because the highest value of k for each of the six community network structures ranges from 3 to 8, I will compare the proportion of organizations that belong to a three-core or higher (i.e., the proportion of organizations that are directly connected to at least three other organizations).


630


Core Mayfield Interorganizational Network Structure

Creston Interorganizational Network Structure

Figure 1. Interorganizational network structures of six communities.

measuring bonding structures.3 It is also important to look at the number and proportion of cut-points in a network to measure the level and type of bridging capital in each network. Cut-points determine the extent of nonredundant contacts: contacts that are either not directly connected or have contacts that are different from one another. A cut-point is a node in which its ‘‘removal would increase the number of components by dividing the sub-graph into two or more separate subsets between which there are no 3 Although the density of each community network can be measured, a fundamental problem exists with this measure. The density of a network depends on the size of the graph. This prevents density measures from being compared across networks of different sizes (Friedkin, 1981; Scott, 2004). Although measuring the mean degree of each network overcomes this limitation, it does not measure the bonding type of structures that are theoretically important for the current study. This is because one member may have direct ties with many other members thus raising the mean number of ties for all other members of a network who may not have many direct ties with other members in the network. Because the analysis of k-cores overcomes both of these limitations, the current study uses k-cores to measure bonding network structures.



631

Rowans View TVAC-Centered Subnetwork

Gwenville Heights Interorganizational Network Structure

Figure 1. Continued.

connections’’ (Scott, 2004, p. 107).4 Each subgraph that either stands alone or is connected to a larger graph by a cut-point is referred to as a block. Thus, cut-points are essential in measuring the extent and type of bridging capital in a given network. 4

This is what Burt (1992) refers to as ‘‘structural holes.’’


632


Soundberry Interorganizational Network Structure

Davis Grove Interorganizational Network Structure

Figure 1. Continued.

Communities characterized by a high proportion of organizations that belong to a three-core or higher and a small proportion of cut-points are classified as having a complete network structure. Communities characterized by a high proportion of organizations that belong to a three-core or higher and a mid-range proportion of cut-points are classified as having a coalitional network structure. Communities characterized by a low proportion of cut-points with several blocks of organizations not linked to other blocks of organizations are classified as having a factional network structure. Communities characterized by a low to mid-range proportion of organizations that belong to a three-core or higher and a low proportion of cut-points are classified as having a bridging network structure. Journal of Community Psychology DOI: 10.1002/jcop


633

Communities with a complete network structure are coded as one; all other communities are coded as zero. Communities with a coalitional network structure are coded as one; all other communities are coded as zero. Communities with a factional network structure are coded as one; all other communities are coded as zero. Communities with a bridging network structure are coded as one; all other communities are coded as zero. Analytic Strategy The first stage of the analyses focuses on the descriptive statistics of each dependent and independent variable. I examine individual ties and community network structures for community attachment and community evaluation. For the second stage of the analyses, I test my hypotheses by performing negative binomial regression to assess how a community’s social network structure in conjunction with individual ties to the community affect different types of community sentiment.5 All dependent variables are overdispersed count variables (a6¼0). Although Poisson regression is typically used for count variables, the preferred model for overdispersed count variables is negative binomial regression because the negative binomial distribution assumes that the variance is larger than the mean (Barron, 1992; Hoffmann, 2004).

RESULTS Descriptive Statistics for Each Community Table 2 provides measures of central tendency and variability for the dependent and independent variables as well as measures for each community. For community attachment, the mean score for all community members was 18.98 with a range from 11 (low community attachment) to 25 (high community attachment). Community members in Rowans View on average had the highest levels of community attachment (21.36), whereas residents of Gwenville Heights on average had the lowest levels of community attachment (17.26). For community evaluations of the social environment, the mean score for all community members was 24.05 with a range from 14 (unfavorable community evaluation) to 35 (favorable community evaluation). Residents of Mayfield had the most favorable evaluations of their community’s social environment (26.44), whereas once again residents of Gwenville Heights had the least favorable evaluations of the community (21.79). For community evaluations of the physical environment, the mean score for all community members was 12.55 with a range from 5 (unfavorable community evaluation) to 20 (favorable community evaluation). Residents of Mayfield and Soundberry had the most favorable evaluations of their community’s physical environment (13.61), whereas residents of Davis Grove had the least favorable perceptions of the community’s physical environment (9.88). 5 A limitation of cross-sectional data is the uncertainty of causation. Although negative binomial regression analyses can reveal correlations between the dependent and independent variables, caution must be taken in asserting causality from data collected from one point in time. Nevertheless, the systemic model of community attachment asserts that local ties influence community attachment and not vice versa. Furthermore, measures of community sentiment asked about current perceptions and evaluations, while most participants were members of organizations and institutions for several years before the survey was conducted (6.6 years on average). Because the study’s hypotheses were made prior to data collection, I infer causal relationships, albeit with caution.



Coalitional

2.00 (.87) 1–4 3.33 (1.22) 1–5 2.22 (1.86) 0–5 Bridging

2.80 (1.11) 1–4 2.85 (1.09) 1–4 2.55 (1.67) 0–6

52.32 (11.57) 22–74 45 22.74 (17.15) 0–66

(1.93) 7–14

(2.50) 8–15

50.89 (12.84) 25–69 44 26.00 (25.58) 0–67

(4.56) 14–31 9.88

(4.85) 14–31 12.00

Factional

2.05 (.71) 1–4 2.68 (1.00) 1–4 2.47 (1.74) 0–7

49.32 (10.96) 35–74 58 19.66 (20.82) 0–74

(2.31) 10–18

(4.33) 14–29 12.80

17.26 (2.71) 13–22 21.79

Gwenville Heights (N 5 19)

Complete

2.54 (.78) 1–4 3.33 (1.09) 2–5 4.17 (3.07) 0–10

48.76 (10.23) 24–71 52 13.06 (9.06) 1–35

(2.29) 10–19

(2.55) 21–31 13.61

19.75 (2.45) 15–24 26.44

Mayfield (N 5 25)

Coalitional

3.18 (.87) 2–4 3.27 (1.19) 1–5 4 (2.37) 0–7

51.7 (10.81) 29–65 70 27.2 (19.75) 1–65

(4.67) 8–20

(4.48) 21–35 13.44

21.36 (2.34) 16–25 25.45

Rowans View (N 5 11)

Note. For each variable, mean scores are located in the first row, standard deviation in the second, and range in the third.

Community measures Community network structure

Formal ties

Strong informal ties

Individual ties to community Weak informal ties

% Female Years resided

Independent variables Age

Community evaluation of physical env.

Community evaluation of social env.

18.83 (2.04) 15–22 22.47

19.00 (2.78) 13–22 24.33

Davis Grove (N 5 20)

Coalitional

2.38 (.96) 1–4 3.08 (1.26) 1–4 2.54 (1.71) 0–6

46.92 (17.44) 18–83 69 18.08 (17.24) 1–60

(3.93) 5–20

(4.15) 17–32 13.61

18.23 (3.14) 11–22 23.69

Soundberry (N 5 13)

–

2.50 (.94) 1–4 3.06 (1.12) 1–5 3.07 (2.3) 0–10

49.84 (11.95) 18–83 55 19.73 (17.75) 0–74

(3.13) 5–20

(4.32) 14–35 12.55

18.98 (2.78) 11–25 24.05

Total (N 5 97)

–

>0

1–5

1–5

– >0

>18

4–20

5–35

5–25

Scale

Dependent variables Community attachment

Creston (N 5 9)

Table 2. Mean, Standard Deviation, and Range for Dependent and Independent Variables by Community

634 Journal of Community Psychology, July 2010


635

The average age of respondents was 50 years, with Soundberry having the youngest with an average of 47 and Davis Grove having the oldest with respondents having an average age of 52 years. Fifty-five percent of respondents were female. With respect to personal ties to the community, respondents on average lived in the community for 20 years, knew less than half of community members by name, had half of their close personal friends living in the community, and belonged to an average of three local organizations. Respondents from Rowan’s View had resided in the community the longest with an average of 27 years, whereas respondents of Mayfield had resided in the community the shortest with an average of 13 years. On average, respondents of four communities claimed to know less than half of the community members by name, whereas respondents of two communities (Rowan’s View and Davis Grove) claimed to know about half of the community members by name. With respect to strong informal ties, very little difference at the community level exists, with respondents of all six communities, on average, having half of their close personal friends living in the community. At the community level, the number of local organizations’ respondents belonged to ranged from an average of two (Gwenville Heights and Creston) to four (Mayfield and Rowan’s View). With respect to communitylevel measures, Creston, Rowan’s View, and Soundberry have coalitional network structures. Mayfield has a complete network structure. Davis Grove has a bridging network structure, whereas Gwenville Heights has a factional network structure. Table 3 presents the mean scores on the items used to measure the three types of community sentiment by network structure. Towns with complete or coalitional network structures have significantly higher levels of attachment than communities with bridging or factional network structures for three of the five indicators of community attachment. Likewise, towns with complete or coalitional network structures have significantly higher averages for four of the five indicators of community evaluation of social environment and two of the four indicators of community evaluation of physical environment. When responses are summed to form a scale, towns with complete or coalitional network structures have significantly higher averages than communities with bridging or factional network structures for all three types of community sentiment. Although the above analyses indicate that communities with complete and coalitional network structures on average have higher levels of community sentiment than communities characterized by bridging or factional network structures, the above analyses do not control for the effect of individual-level ties to the community. To examine the effect of a community’s network structure in conjunction with individual-level ties toward the three types of community sentiment, I employ negative binomial regression. Community attachment. Table 4 presents negative binomial regression results for community members’ attachment to community while controlling for individualand community-level factors. Similar to other studies of community attachment, I find that the number of years a member has resided in the community is positively related to community attachment (b 5 .002, po.01). More specifically, the longer a person has resided in the community the more attached she or he will be to the community. Table 4 shows that community attachment is also affected by an individual’s formal ties, weak informal ties, and strong informal ties. As expected, individuals who have many formal ties through local organizations are more likely to be attached to the community than those community residents who are not active in formal local organizations (b 5 .010, po.05). Likewise, individuals who know many community members by name are more likely to be attached to the community than those community residents Journal of Community Psychology DOI: 10.1002/jcop

636


Table 3. Mean Scores on Items Measuring Type of Community Sentiment by Network Structure Type of community sentiment Community attachment Being a resident of the community is like living with a group of close friends. If you do not look out for yourself no one else in the community will. Most everyone in the community is allowed to contribute to local government affairs if they want to. When something needs to get done in the community, the whole community usually gets behind it Community clubs and organizations are interested in what is best for all residents Community evaluation of social environment Level of friendliness Level of support Level of trust Level of tolerance Level of openness to new ideas Community evaluation of physical environment Quality of senior citizen center Quality of meeting space in city offices Quality of formal meeting space in local restaurants and other businesses Quality of restaurants and coffee shops

Complete/ coalitional

Bridging/ factional

19.60 (2.79) 3.72

18.03 (2.50) 3.50

18.98 (2.77) 3.63

(.82) 4.03

(.86) 3.84

(.84) 3.96

(.92) 4.21

(.59) 3.86

(.81) 4.07

(.77) 3.60

(.79) 3.16

(.79) 3.43

(.99) 4.03

(.83) 3.66

(.95) 3.89

(.82) 25.31 (3.81) 5.84 (1.09) 5.24 (1.33) 4.88 (1.33) 4.66 (1.41) 4.69 (1.38) 13.31 (3.20) 4.11 (.92) 2.98 (1.17) 2.86

(.75) 22.13 (4.40) 5.21 (1.12) 4.76 (1.22) 4.28 (1.00) 3.87 (1.22) 3.87 (1.34) 11.25 (2.55) 3.39 (1.24) 1.94 (1.03) 2.53

(.81) 24.05 (4.32) 5.59 (1.14) 5.05 (1.30) 4.64 (1.23) 4.34 (1.38) 4.36 (1.42) 12.54 (3.13) 3.82 (1.11) 2.57 (1.22) 2.74

(1.09) 3.51 (.93)

(.90) 3.49 (.84)

(1.03) 3.50 (.89)

Total

Note. Means are reported on a 5-point scale for community attachment (1 5 strongly disagree, 2 5 disagree, 3 5 neither agree nor disagree, 4 5 agree, 5 5 strongly agree) and for community evaluation of physical environment (1 5 not available, 2 5 poor, 3 5 fair, 4 5 good, 5 5 very good) and on a 7-point scale for community evaluation of social environment. Standard deviations are provided in parentheses. N 5 97. po.05 po.01 po.001 (two-tailed t test).

who know few people in the community by name (b 5 .028, po.10). As expected, individuals who have a high proportion of their close friends residing in the community are more likely to be attached to the community than those community residents who have few or no friends in the community (b 5 .021, po.05). On the other hand, all of the community network structure variables are not significant. That is, all else equal, Journal of Community Psychology DOI: 10.1002/jcop


637

Table 4. Individual and Community Models for Community Sentiment Community evaluation

Independent variables Intercept

Age

Female

Years resided

Weak informal ties

Strong informal ties

Formal ties

Social environment

Physical environment

b

b

2.76 (.088) .001 [.9] (.001) .000 [.0] (.000) .002 [3.9] (.001) .028 [2.4] (.018) .021 [2.4] (.012)

2.97 (.107) .000 [.6] (.001) .002 [.3] (.001) .001 [2.2] (.001) .009 [.8] (.022) .0301 [3.4] (.016)

1.707 (.140) .008 [11.5] (.002) .036 [3.7] (.040) .001 [1.7] (.001) .034 [3.2] (.031) .022 [2.4] (.021)

.010 [2.2] (.006)

.004 [.9] (.007)

.025 [5.3] (.009)

.065 [6.7] (.044)

.187 [20.5] (.068)

.458 [58.0] (.077)

.065 [6.7] (.073)

.126 [13.5] (.108)

.479 [61.5] (.129)

.050 [5.2] (.170) .000 [3.5] (.000)

.108 [11.4] (.273) .000 [4.1] (.000)

.753 [112.3] (.305) .000 [13.1] (.000)

Community attachment b

Complete

Coalitional

Factional

Population

Note. Standardized percentage change appears in brackets (unstandardized for dummy variables). Robust standard errors are in parentheses. For community attachment, test for intercept and years resided are two-tailed; all other tests are one-tailed. For community evaluation, all tests are two-tailed. po.10, po.05, po.01, po.001.

residents residing in communities with bridging network structures have relatively the same level of attachment to community as residents living in communities with complete, coalitional, or factional network structures. Community Evaluation Table 4 presents HLM results for community members’ evaluations of community while controlling for individual- and community-level factors. As expected, Journal of Community Psychology DOI: 10.1002/jcop

638


community network structure affects an individual’s evaluation of his or her community’s social environment. Specifically, individuals living in a community characterized by a complete (b 5 .187, po.01) network structure are more likely to evaluate their community in a favorable manner than communities characterized by bridging or factional network structures. When comparing residents of communities with a complete network structure to residents of communities with a bridging network structure, those living in a community with a complete network structure on average rate the social environment of the community 20% more favorably than those living in a community with a bridging network structure. In other words, a community member who resides in a community characterized by dense ties that connect different parts of the community together is more likely to have favorable perceptions of the community than a community member who lives in a community with very loose ties or an absence of ties that bind different parts of the community together, regardless of one’s personal ties to the community. As for individual ties, individuals who have a high proportion of their close friends residing in the community are more likely to be attached to the community than those community residents who have few or no friends in the community (b 5 .030, po.10). All other personal network variables are not significant. That is, members who have not resided in the community for many years, have few weak informal ties, and have few formal ties are as likely as those members who have resided in the community for many years, have many weak and strong informal ties, and have many formal ties to have positive (or negative) perceptions of the community. Community network structure also affects an individual’s evaluation of his or her community’s physical environment. As expected, individuals living in a community characterized by a complete (b 5 .458, po.001) or coalitional (b 5 .479, po.001) network structure are more likely to evaluate their community’s physical spaces in a favorable manner than communities characterized by a bridging network structure. When comparing residents of communities with a complete network structure to residents of communities with a bridging network structure, those living in a community with a complete network structure on average rate physical spaces in the community 58% more favorably than those living in a community with a bridging network structure. Residents of communities with a coalitional network structure, on average, rate physical spaces in the community 61.5% more favorably than those living in a community with a bridging network structure. Age and the number of formal ties to the community also affect one’s evaluation of the community’s physical environment. Older residents are more likely to evaluate the physical environment more favorably than younger residents. Residents with more formal ties in the community are also less likely to evaluate the physical spaces in a community favorably than those with few formal ties.

DISCUSSION This study increases our understanding of community sentiment by examining how the social network structure of communities can influence different types of community sentiment when controlling for an individual’s ties to the community. Overall, the findings provide support for my hypotheses, showing that an individual’s ties to the community are positively associated with community attachment. However, the biggest Journal of Community Psychology DOI: 10.1002/jcop


639

contribution was the finding in support of my second hypothesis: that a community’s network structure significantly influences an individual’s evaluation of the community. Drawing on research on community attachment and community network structures, I hypothesized community members residing in cohesive communities would be more attached to their communities than those residing in disjointed communities when controlling for individual-level factors. I do not find support for this hypothesis. When controlling for length of residence in the community, informal weak and strong ties, and formal ties to the community, residents of cohesive communities and more disjointed communities appear to have similar levels of attachment to their community. Although attachment to community does not appear to differ significantly by community network structure, I find strong support for the hypotheses relating to individual-level factors and community attachment. Specifically, I hypothesized length of residence, informal ties, and formal ties to positively influence community attachment. The findings indicate that the longer a member resides in the community, the more attached she or he will be to the community. Likewise the more acquaintance and close friends a person has in the community and the more formal organizations a person belongs to, the more attached he or she will be to the community. These findings support previous research relating length of residence and personal ties to the community to community attachment (Beggs et al., 1996; Goudy, 1990; Kasarda & Janowitz, 1974; Liu et al., 1998; Logan & Spitze, 1994; Sampson, 1988). The main contribution of this study is the evidence that a community’s social network structure influences an individual’s evaluation of his or her community. As expected, citizens of communities characterized by complete network structures (i.e., those with relatively closed, cohesive associational networks) would evaluate their community’s social environment more favorably than citizens of communities with bridging network structures (those with loose associational networks). One possible explanation for these findings is that communities with high bonding social capital may provide an atmosphere that minimizes controversy. Thus, negative statements about the community may be held to a minimum, thereby influencing a person’s perceptions of the community as being more open and friendly. As expected, citizens of communities characterized by complete or coalitional network structures (i.e., those with relatively closed, cohesive associational networks) evaluate their community’s physical environment more favorably than citizens of communities with bridging network structures (those with loose associational networks). Communities with more cohesive network structures are shown to have higher levels of economic development (Crowe, 2007). Thus, they may be more effective at developing the physical spaces of the community than are more disjointed communities. Community members with strong formal ties are less likely to have a favorable evaluation of the community’s physical spaces. This is opposite to the finding for community attachment. It appears that while joining formal local organizations increases one’s level of attachment to the community it also gives more of an awareness to the quality of a community’s physical spaces thus making a person more critical. Limitations Despite the various strengths of this study, some limitations must be acknowledged. One of the study’s limitations is the small sample size. Future studies can improve upon the sample size by sampling more communities and a larger total number of individuals. Furthermore, I did not oversample for racial minority residents. With Journal of Community Psychology DOI: 10.1002/jcop

640


respect to community evaluations, it may be the case that minority residents are not embedded in the same social networks of the community as the White residents. Thus, they may have different evaluations of the community. However, because all communities had between 60–90% White residents and the communities with the lower percentage of White residents (Creston and Soundberry) had more positive evaluations, I believe that community social network structures significantly influence community satisfaction independent of a resident’s race or ethnicity. Second, although the current study has a measure of community attachment and community satisfaction (in the form of evaluation of social and physical environments), it does not measure the outcomes each type of community sentiment has on the community. For instance, it would be instructive to know how one’s level of community attachment compared to one’s evaluation of the community affects the likelihood of staying in the community or returning to the community later in life. Finally, although the findings illustrate that community-level characteristics affect individual community evaluations, there is a need for more comparative studies to better evaluate the robustness of the findings. For instance, future studies may compare how individual ties and community-level variables influence community attachment and evaluation in cities to that in rural communities.

CONCLUSION Although individual-level ties have been used to account for differing levels of community attachment (Liu et al., 1998), until now little empirical work has systematically addressed the connection between a community’s social network structure and community attachment and evaluation. Accordingly, this study makes a unique contribution to our understanding of a community’s social network structure and different types of community sentiment. By using survey data from community members in six rural communities in Washington, this study shows that community sentiment, in general, is related to both individual-level and community-level variables. However, individual-level and community-level variables have varying effects on the different types of community sentiment. Findings suggest that although individual-level variables, such as length of residence and individual ties, affect one’s attachment to community, a community’s network structure does not significantly affect community attachment. This finding challenges the suggestions made by social scientists that strengthening social capital will lead to higher levels of community attachment (Putnam, 1993). Instead, it appears to be the strength of informal and formal ties that an individual holds that influences community attachment. Thus, communities may find it more beneficial to actively recruit residents into existing groups and to particularly recruit youth participants. However, a community’s network structure significantly affects one’s evaluation of community. Regardless of one’s ties to the community, residents of cohesive communities are more likely to evaluate the community’s social and physical environments more positively. The next steps include partitioning out the effects of community attachment and community evaluation on factors that build community such as civicness and youth retention. Based on the results of this study, I recommend researchers, policy makers, and others interested in understanding community sentiments pursue more in-depth investigation of the interaction between individual ties, community network structures, and community attachment and satisfaction. Journal of Community Psychology DOI: 10.1002/jcop


641

REFERENCES Allen, J., & Dillman, D. (1994). Against all odds: Rural community in the information age. Boulder, CO: Westview Press. Barron, D. (1992). The analysis of count data: Overdispersion and autocorrelation. Sociological Methodology, 22, 179–220. Beggs, J., Hurlbert, J., & Haines, V. (1996). Community attachment in a rural setting: A refinement and empirical test of the systemic model. Rural Sociology, 61, 407–426. Burt, R. (1992). Structural holes: The social structure of competition. Cambridge, MA: Harvard University Press. Crowe, J. (2007). In search of a happy medium: How the structure of interorganizational networks influence community economic development strategies. Social Networks, 29, 469–488. Diaz, E.M., & Miraftab, F. (2009). Sundown town to ‘little Mexico’: Old-timers and newcomers in an american small town. Rural Sociology, 74, 605–629. Economic Research Service. (2004). County Typology Codes. Washington, D.C.: United States Department of Agriculture. Filkins, R., Allen, J., & Cordes, S. (2000). Predicting community satisfaction among rural residents: An integrative model. Rural Sociology, 65, 72–86. Fischer, C. (1972). Urbanism as a way of life: A review and an agenda. Sociological Methods and Research, 2, 82–242. Flora, C., & Flora, J. (2008). Rural communities: Legacy and change. (3rd ed.). Boulder, CO: Westview Press. Flora, J., Sharp, J., Flora, C., & Newlon, B. (1997). Entrepreneurial social infrastructure and locally initiated economic development in the nonmetropolitan United States. The Sociological Quarterly, 38, 623–645. Friedkin, N. (1981). The development of structure in random networks. Social Networks, 3, 41–52. Gans, H. (1962). The urban villagers. New York: The Free Press. Gerson, K., Steuve, A., & Fischer, C. (1977). Attachment to place. In C. Fischer (Ed.), Networks and places: Social relations in the urban setting. (pp. 139–161). New York: The Free Press. Goudy, W. (1977). Evaluations of local attributes and community satisfaction in small towns. Rural Sociology, 42, 371–382. Goudy, W. (1982). Further consideration of indicators of community attachment. Social Indicators Research, 11,181–192. Goudy, W. (1990). Community attachment in a rural region. Rural Sociology, 55,178–198. Granovetter, M. (1986). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91, 481–510. Herting, J., & Guest, A. (1985). Components of satisfaction with local areas in the metropolis. The Sociological Quarterly, 26, 99–115. Hoffmann, J.P. (2004). Generalized linear models: An applied approach. Boston: Pearson Education. Hummon, D. (1992). Community attachment: Local sentiment and sense of place. In I. Altman & S. Low (Eds.), Place attachment (pp. 253–278). New York: Plenum Press. Hunter, A. (1978). Persistence of local sentiments in mass society. In D. Street (Ed.), Handbook of contemporary urban life (pp. 133–162). San Francisco: Jossey-Bass. Kasarda, J., & Janowitz, M. (1974). Community attachment in mass society. American Sociological Review, 39, 28–39. Journal of Community Psychology DOI: 10.1002/jcop

642


Liu, Q., Ryan, V., Aurbach, H., & Besser, T. (1998). The influence of local church participation on rural community attachment. Rural Sociology, 63, 432–450. Logan, J., & Spitze, G. (1994). Family neighbors. American Journal of Sociology, 100, 453–476. Lyon, L. (1999). The community in urban society. Prospect Heights, IL: Waveland Press. McMillan, D., & Chavis, D. 1986. Sense of community: a definition and theory. Journal of Community Psychology, 14, 6–23. Nisbet, R. (1976). The quest for community. New York: Oxford University Press. O’Brien, D., & Hassinger, E. (1992). Community attachment among leaders in five rural communities. Rural Sociology, 57, 521–534. Putnam, R. (1993). Making democracy work: Civic traditions in modern Italy. Princeton, NJ: Princeton University Press. Putnam, R. (2000). Bowling alone: The collapse and revival of American community. New York: Simon and Schuster. Reissman, L. (1964). The urban process. Glencoe: The Free Press. Rivlin, L. (1982). Group membership and place meanings in an urban neighborhood. Journal of Social Issues, 38, 75–93. Ryan, V., Agnitsch, K., Lijun Zhao, L., & Mullick, R. (2005). Making sense of voluntary participation: A theoretical synthesis. Rural Sociology, 70, 287–313. Sampson, R. (1988). Local friendship ties and community attachment in mass society: A multilevel systemic model. American Sociological Review, 53, 766–779. Sampson, R. (1991). Linking the micro- and macrolevel dimensions of community social organization. Social Forces, 70, 43–64. Scott, J. (2004). Social network analysis. London: Sage. Seidman, S. (1983). Network structure and minimum degree. Social Networks, 5, 269–287. Sharp, J. (2001). Locating the community field: A study of interorganizational network structure and capacity for community action. Rural Sociology, 66, 403–424. Short, J. (1971). The social fabric of the metropolis. Chicago: University of Chicago Press. Taylor, M. (2004). Community issues and social networks. In C. Phillipson, G. Allan, & D. Morgan (Eds.), Social networks and social exclusion. Burlington, VT: Ashgate. Toennies, F. (1957). Gemeinschaft und Gesellschaft [Community and society]. Leipizig: Fues’s Verlag. (Original work published 1887). U.S. Bureau of the Census. (2000). Summary File 1. Washington, DC: Author. U.S. Department of Housing and Urban Development. (1978). A Survey of Citizen Views and Concerns about Urban Life (Study no. P2795). Washington, DC: Author. Webber, M. (1963). Order in diversity: Community without propinquity. In L. Jr Wingo, (Ed.), Cities and space: The future use of urban land. Baltimore: John Hopkins University Press. Wirth, L. (1938). Urbanism as a way of life. American Journal of Sociology, 44, 1–24. Woolcock, M., & Narayan, D. (2000). Social capital: Implications for development theory, research, and policy. The World Bank Research Observer, 15, 225–249.



643

APPENDIX

Community Interview Categories

1.

Parent

2.

Nonprofit Youth-Focused Service Organization

3.

Nonprofit Adult-Focused Service Organization

4.

Major Employer

5.

Entrepreneurial Business

6.

Faith-Based Organization

7.

Elected Official

8.

Law Enforcement

9.

Social Services Agency

10.

Chamber/Economic Development Council

11.

School Employee (must live in community)

12.

School Board Member

13.

Hospital/Health Organization

14.

Senior Citizen

15.

Representatives of Ethnic Groups in Community

16.

Older youth

17–20.

Wild Cardy

y Wild Card category examples may include but are not limited to farming community, service club representative, arts community, representative from an environmental group, factory worker, and timber worker.


1. Community attachment 2. Social evaluation of community 3. Physical evaluation of community 4. Leader 5. Age 6. Sex 7. Years resided 8. Weak informal ties 9. Strong informal ties 10. Formal ties 11. Population 12. Complete network structure 13. Coalitional network structure 14. Factional network structure 15. Bridging network structure

Variable

1.0 .571 .138 .126 .051 .052 .274 .358 .364 .316 .284 .164 .135 .313 .026

1 1.0 .250 .097 .086 .096 .146 .126 .300 .118 .202 .330 .068 .261 .182

2

1.0 .141 .367 .050 .081 .018 .144 .001 .130 .206 .133 .037 .425

3

1.0 .246 .384 .110 .256 .095 .128 .082 .087 .065 .090 .106

4

1.0 .196 .461 .008 .120 .158 .042 .054 .019 .022 .104

5

1.0 .016 .053 .031 .156 .047 .038 .104 .027 .105

6

1.0 .244 .419 .119 .023 .227 .143 .002 .086

7

1.0 .379 .318 .298 .026 .035 .238 .165

8

1.0 .283 .110 .140 .097 .168 .098

9

1.0 .139 .275 .042 .130 .117

10

1.0 .321 .012 .907 .529

11

1.0 .109 .080 .083

12

1.0 .094 .096

13

1.0 .071

14

1.0

15

Table A1. Pearson Correlation Coefficients

644 Journal of Community Psychology, July 2010
