Social Indicators Research

, Volume 108, Issue 2, pp 215–238

Measuring Social Capital in Hamilton, Ontario

Authors

    • McMaster Institute of Environment and HealthMcMaster University
  • Allison Williams
    • School of Geography and Earth SciencesMcMaster University
  • Dylan Simone
    • School of Geography and Earth SciencesMcMaster University
Article

DOI: 10.1007/s11205-012-0063-3

Cite this article as:
Kitchen, P., Williams, A. & Simone, D. Soc Indic Res (2012) 108: 215. doi:10.1007/s11205-012-0063-3

Abstract

Social capital has been studied by academics for more than 20 years and within the past decade there has been an explosion of growth in research linking social capital to health. This paper investigates social capital in Hamilton, Ontario by way of a telephone survey of 1,002 households in three neighbourhood groups representing high, mixed and low socio-economic status (SES). A Social Capital Measurement Tool is proposed as a straightforward way to account for differences in social capital perceptions and actions among residents. Consistent with the literature, the paper found that there was a strong association between social capital perceptions and health, particularly mental health and life stress. Social capital was greater in the high and mixed SES neighbourhoods and much weaker in the low SES neighbourhoods. With respect to social capital actions, both volunteering and voting was associated with positive overall health and mental health. Finally, the paper identified the social capital ‘elite’—respondents with high or above average perceptions and who participated in the two social capital actions—voting and volunteering. Prominent among the social capital elite in Hamilton is the ‘health wealthy’ senior, a positive development for the continued social well-being of the city.

Keywords

Social capitalHealthSocio-economic statusPlace of residence

1 Introduction

The idea of social capital has been studied by social scientists for several decades. Debate has emerged among academics on how best to define and measure the concept and within the past 10 years there has been an explosive growth in research linking social capital to health. A review of the extensive literature on this topic reveals a complex and sometimes contentious discourse that is geared largely towards an academic audience. Social capital has important policy implications that are of interest to both academic and non-academic communities. This paper follows two approaches to the study of social capital. The first proposes a concise and relatively straightforward method for measuring social capital in a community, one that we hope is accessible to a broader audience; an audience that includes members of the public, planners and policy makers. A Social Capital Measurement Tool (SCMT) is developed to examine social capital perceptions and social capital actions in Hamilton, Ontario. The second approach is more academic in nature and employs quantitative methods to assess the association between social capital and factors related to health, socio-demography and place of residence.

Social capital is closely related to the concept of sense of belonging, which is also explored in this special issue. Rose (2000) observes that social capital can refer to the properties of a community be they physical, social or cultural. For instance, communities with high levels of social capital ‘‘might be those that offer opportunities for interaction and that have well-developed resources such as parks, libraries and recreational facilities’’ (Rose 2000, p. 35). Rose (2000) succinctly defines social capital as follows:

At its most basic level, the term refers to the notion that relationships with others have important implications for well-being. Individuals can possess social capital by having a large network of friends and acquaintances, but social capital can also be thought of a type of social savvy (Rose 2000, p. 34).

There have been several research initiatives that have successfully engaged communities in measuring social capital and making links to public policy. For instance, the University of Minnesota conducted an assessment of social capital in the Waseca community of that state (Chazdon et al. 2008). It included the development of a Community Social Capital Model, which focused on residents’ social and economic relationships as determined by connections, engagement and trust. The process included strong community input and resulted in the development of action items aimed at improving social capital in Waseca. In an earlier initiative, Bullen and Onyx (1998) measured social capital in five communities in New South Wales Australia. The project involved significant community participation with more than 1,200 questionnaires being completed and resulted in the development of a social capital scale.

This paper analyzes data from a telephone survey of 1,002 households in Hamilton, Ontario. The data was collected among residents in three neighbourhood groups representing high, mixed and low SES. The objective is to examine variations in social capital according to SES, health and place of residence. For the purpose of this study, social capital is defined as a series of perceptions and actions on behalf of the individual, which leads to enhanced community engagement. The micro-scale of our study justifies the implementation of an individual-level conceptualization of social capital, which utilizes a six-point scale for the measurement of the concept. Four variables compose the ‘perceptions’ of social capital: (1) safety after dark, (2) inter-personal trust, (3) accessibility to help, and (4) views on multiculturalism. These four perceptions are used in conjunction with two measures of social capital ‘actions’: (1) volunteering and (2) municipal voting—which as a whole represent a validated operational framework for the empirical measurement of an individual’s level of social capital.

The data is analyzed through descriptive statistics, contingency tables and regression modeling. The paper is divided into four sections. The first provides a review of the recent literature on social capital highlighting the link to health. The second section describes the data and methods employed in the study while the third presents the results of the analysis. The fourth section consists of a discussion of social capital in Hamilton and points to areas of continued research.

2 Social Capital: A Review of the Literature

2.1 Theoretical Framework

The concept of social capital has received considerable attention since its introduction in the sociological literature more than 20 years ago (Bourdieu 1986; Coleman 1988). Commonly defined as the resources which individuals and groups have access to by virtue of their membership in networks (Bourdieu 1986), social capital describes the value of social relationships, mutual trust and norms of reciprocity to both the individual and society (Coleman 1988). However, the precise definition and inherent conceptualization is quite contentious, reflected in the lack of consensus among scholars (Poortinga 2006; Giordano et al. 2011; Moore et al. 2011). In an attempt to understand social capital, one has to be conscientious of the scale at which the concept is measured; while it is generally accepted as a collective resource, it can also be defined in terms of the individual (Veenstra 2005; Schultz et al. 2008; Giordano and Lindstrom 2010). To illustrate, Portes (1998) defines social capital as the ability of individuals to command scarce resources through their membership in networks or broader social structures. This definition emphasizes the importance of individual-level effects, which contrasts political scientist Putnam’s (1993) ‘contextual’ definition of social capital. Putnam (1993) defines social capital as being features of organizations, such as trust, norms, and networks that can improve the efficiency of society through facilitating coordinated actions. Putnam claims the ‘collective level’ is most accurate in measuring social capital. One solution to the definitional problem is to accept both conceptualizations—that is, to carry out studies that include both individual -and collective-level social capital measures (Kawachi et al. 2008). However, this too is not without contention, as these ‘levels’ may be defined by data availability or aggregation and, as such, may not represent an individual’s actual social networks and interactions (Giordano et al. 2011).

Sociologist Bill Reimer has conducted important research on social capital in rural Canada highlighting its relation to community and individual characteristics (see for example: Teipot and Reimer 2004; Reimer 2006; Reimer et al. 2008). He proposes that social capital is based on four fundamental types of normative relations: market, bureaucratic, associative and communal. These “reflect four ways in which norms guide the type of acceptable behaviour within social relations, justify the distribution of resources to the actors, and assign sanctions when those norms are violated” (Reimer 2006, p. 164). While much of Reimer’s work has focused on rural areas, his conceptualization of social capital is certainly applicable to many types of communities including urban. Referring to the ways in which communities can develop, Reimer observes:

Available social capital can be measured by the institutions and organizations within which the social relations are organized. A school, a baseball league, a food bank, or a card club all represent social capital that may be used by people or groups. For those outside the organization, however, the social capital they represent may remain only potentially available….Much of community development practice is directed toward recognizing the potential social capital that may be unused or unrecognized by community members when mobilizing this social capital in new ways (Reimer 2006, p. 164).

2.2 Social Capital and Health

Research on social capital and health has increased exponentially over the past 10 years (Kawachi et al. 2008). As explained by Veenstra et al. (2005) social capital is a key component in a contentious, broadly defined health discourse. Veenstra et al. discuss the compositional and contextual health effects of social capital. The compositional refers to an individual’s health being affected as result of direct and beneficial effects on his or hers activities and attributes. The contextual speaks to an indirect effect on health, through social capital’s influence on the larger factors which act as determinants of population health (e.g., social, economic, and environmental factors).

There is consensus in the literature over the two components constituting social capital: bonding and bridging (Kawachi et al. 2008; Elgar et al. 2011). Bonding capital refers to resources accessed within social groups by homogeneous members (e.g., who share similar socio-demographic characteristics such as class, race, religion, and so on). In contrast, bridging capital denotes resources that are accessed by individuals or groups who are dissimilar as a result of race, income or other socio-economic traits (Kawachi et al. 2008). Putnam (2000) suggests that bonding social capital can be used to help people meet their basic needs in a mutually beneficial manner by bringing people together who are similar in some respect. On the other hand, bridging capital may help people of differing backgrounds (be they demographic, economic or social) access external assets and share information in ways that allow them to get ahead and reach higher goals (Putnam 2000).

Kawachi et al. (1997, 1999) were among the first to test relationships between social capital and health at a contextual scale. Using Putnam’s conceptualization of social capital, Kawachi et al. (1999) showed that people reported poorer health while living in US states with low levels of social trust, compared to people living in states with medium or high levels of trust. Putnam (2000) confirmed these findings, showing that states scoring high on a social capital index also scored high on a public health index. More recently, international research has been conducted, attempting to further validate the effects of social capital on health. Notably, in a study of 69 countries using the World Values Survey (n = 160,436) Jen et al. (2010) found increased trust was significantly associated with better health, after controlling for individual socio-demographic and income variables. This study focused on a variant of Wilkinson’s hypothesis: that the greater the lack of social cohesion in a country, the less healthy individuals will be. Through multilevel modeling, it was confirmed that self-rated health was positively linked to both country- and individual-levels of social capital.

Similar findings are evident in Elgar et al’s research (2011), which employed the World Values Survey of 50 countries (n = 69,725 adults). The study showed a correlation between country-level social capital and health. Specifically, cross-level interactions demonstrated that people who are more trusting and affiliated with others reaped greater benefits from social capital. In another study, Poortinga (2006) used data from the European Social Survey—covering 22 countries (n = 42,358)—to examine whether social capital was an individual or collective resource for health. The author found a complex cross-level interaction for social capital (similar to Kim et al. 2006; Subramanian et al. 2002) where individuals who were trusting and socially active reported good or very good levels of health more often in countries with high levels of social capital, compared to individuals with low levels of trust and social activeness. However, individuals reporting good health were less likely to be trusting and active in countries with low levels of social capital. By contrast, a number of studies at the global scale have found no link between social capital and health, at either the compositional or contextual level (Lynch et al. 2001; Lindstrom and Lindstrom 2006; Kennelly et al. 2003).

A large body of research has investigated individual-level associations between measures of trust and self-rated health (Kim and Kawachi 2006; Kim et al. 2008; Subramanian et al. 2002), as well as social participation and self-rated health (Helliwell and Putnam 2004; Hyyppa and Maki 2003). A multitude of studies have found positive associations between individual-level social capital and health (Hyyppa and Maki 2001; Rose 2000; Schultz et al. 2008; Moore et al. 2011), while others have found no such compositional effects (Veenstra 2000). For instance, in a study of Duluth, Minnesota, and Superior, Wisconsin, Schultz et al. (2008) found that after controlling for individual and economic characteristics, compositional social capital measures (such as trust, volunteering, and participation in organized groups) were a significant predictor of self-rated health. Veenstra (2005) conducted a study on the compositional and contextual health effects of social capital in British Columbia, Canada, using two original data sets. The first consisted of information relating to the physical features of communities in the province while the second comprised the characteristics of the individuals living within these communities. Utilizing multilevel modeling, only one significant health effect, depressive symptoms, was found at the contextual level. However, at the individual-level, income, trust in politicians and government, and trust in other community members were significant predictors of health.

An important but understudied theme in the social capital and health literature is change in health effects over time due to social capital. Using data from the British Household Panel Survey (BHPS) Giordano and Lindstrom (2010) addressed this issue by investigating how changes in social capital and material conditions are associated with self-rated health over a six-year period (1999 and 2005). The study found a significant association between individuals unable to trust and deteriorating self-rated health. Likewise, improved health over time was significantly associated with increased levels of social participation. A second study utilizing longitudinal data from the BHPS found no evidence of a link between civic participation at the contextual scale and self-rated health (Snelgrove et al. 2009).

This literature review has highlighted the debate on social capital as a collective or individual concept. Research shows that social capital can benefit health at both the individual and group level and across geographic scales ranging from the local community to the international. Further, social capital utilizes measures that capture both aggregate and individual effects. In addition, while contextual studies are limited through their potential interpretation at both individual and collective levels, compositional research may distort valid contextual effects. However, studies which employ multilevel modeling techniques, using both individual and aggregate measures, have shown that rather than being a true contextual effect, social capital’s benefits on health are best observed through implementation of individual-level measures (Subramanian et al. 2002; Poortinga 2006).

3 Data and Methods

One of the goals of this paper is to develop a Social Capital Measurement Tool (SCMT) that can be readily applied to a number of scales including the neighbourhood, municipality, and region. It is employed here to measure individual and community level social capital in Hamilton, Ontario. The data for the analysis was derived from the Hamilton Household Quality of Life Survey (n = 1,002) administered in three neighbourhood groups between November 2010 and March 2011. The telephone survey included several questions relating to social capital, which are employed as dependent variables in the analysis. These are listed in Table 1 and are divided into two categories: social capital perceptions and social capital actions. Each of the six questions has been used in previous research. As shown in the table, social capital perceptions relate to safety, trust, help from friends and multiculturalism while social capital actions refer to volunteering and voting.
Table 1

Dependent variables measuring social capital 2010/2011 Hamilton household telephone survey

Variable

Survey question

Coded responses

Section A: social capital perceptions

 Safety

Do you feel safe walking down your street after dark?

(0) Not at all safe, (1) not very safe, (2) somewhat safe, (3) very safe

 Trust

Do you agree that most people can be trusted?

(0) Strongly disagree/somewhat disagree, (1) neither agree or disagree, (2) somewhat agree, (3) strongly agree

 Help from friends

Can you get help from friends when you need it?

(0) Never, (1) sometimes, (2) often, (3) always

 Multiculturalism

Do you think that multiculturalism makes life in your area better?

(0) Somewhat worse/much worse, (1) area not multicultural, (2) somewhat better, (3) much better

Section B: social capital actions

 Volunteer

In the past 12 months, did you do unpaid volunteer work for any organization?

(1) Yes, (2) no

 Voting

Did you vote in the last municipal election (held in Hamilton on October 25, 2010)?

(1) Yes, (2) no

Another objective of this paper is to examine the association between social capital and factors related to health, socio-demography and place of residence. Table 2 lists the 13 independent variables employed in the analysis, all derived from questions asked in the Hamilton Household Quality of Life Survey. Three health related indicators are included among the independent variables: self-perceived health, self-perceived mental health and perceived life stress. In addition, nine socio-demographic indicators are assessed: immigrant status, sex, age, marital status, household income, education, dwelling type, housing tenure and years lived in neighbourhood. Finally, the variable denoting the three neighbourhood groups (Southwest Mountain, Central and Lower City) is used to measure place-based variations in social capital in Hamilton. As described in the Introduction to this special issue, the neighbourhood groups were selected to represent high SES (Southwest Mountain), mixed SES (Central) and low SES (Lower City).
Table 2

Independent variables 2010/2011 Hamilton household telephone survey

Variable

Survey question

Coded responses

Self-perceived health

In general, would you say your health is?

Excellent/very good–good–fair/poor

Self-perceived mental health

In general, would you sat your mental health is?

Excellent/very good–good–fair/poor

Perceived life stress

Thinking about the amount of stress in your life, would you say that most days are?

Not at all/not very stressful–a bit stressful–quite a bit/extremely stressful

Immigrant status

Were you born in Canada?

Yes–no

Sex

Is respondent male or female?

Male–female

Age

What is your age category?

Age 18 to 29–age 30 to 44–age 45 to 64–age 65 and over

Marital status

What is your marital status?

Married/common law–widowed/separated/divorced–single/never married

Household income

What is your total annual income before taxes?

Less than $20,000–$20,000 to $49,999–$50,000 to $79,999–$80,000 to $99,999–$100,000 or more

Education

What is your current level of education?

Less than high school–high school–some college or university–community college–bachelors degree–post graduate/professional degree

Dwelling type

What type of dwelling do you live in?

Single detached–semi detached/row/duplex–low-rise–high-rise

Housing tenure

Do you or a member of your household own or rent the dwelling you live in?

Own–rent

Years in neighbourhood

How long have you lived in this or a nearby neighbourhood?

Less than 5 years–5 to 10 years–11 to 19 years–20 or more years

Neighbourhood group

Determined through sampling

Southwest mountain–central–lower city

3.1 A Social Capital Measurement Tool: Examining Perceptions and Actions

A scale was created to measure social capital perceptions based on the four questions listed in Section A of Table 1. Each question has four coded responses ranging from code 0 (denoting the least positive response) to code 3 (denoting the most positive response). The objective was to calculate a single score based on these four questions and to thereby establish an overall measure of social capital perceptions for each of the 1,002 survey respondents. This was achieved by simply adding the coded responses across the four questions creating a scale that ranges from a low of 0 to a high of 12. For instance, if respondent 001 reported that her perceptions on each of the four questions (safety, trust, help from friends and multiculturalism) were the least positive on the four-point scale (coded response 0) then her overall score is 0. On the other hand, if respondent 002 reported that his perceptions on the four questions were the most positive (coded response 3) then his overall score is 12. The result of this approach is a raw score ranging from 0 to 12 for each of the 1,002 respondents of the telephone survey.

The next step was to calculate a standard score for each respondent based on his or her raw score (Z = x−mean x/SD). A standard score has a mean of 0 and a standard deviation of 1 permitting the relative position (with respect to the mean score) of each respondent to be assessed with scores above 0 indicating higher social capital and scores below 0 denoting lower social capital. Based on the Z-scores, the following classification was devised to categorize levels of social capital:

 

Z-score

Social capital

Z ≥ 1

High

< 1 Z > 0

Above average

< 0 Z > −1

Below average

Z ≤ −1

Low

The two indicators relating to social capital actions (volunteering and voting) are binary with a simple ‘Yes’ or ‘No’ answer to each question. The data analysis involved descriptive statistics, by way of bar charts and contingency tables, to convey the basic health and socio-demographic characteristics of social capital in Hamilton and the three neighbourhood groups (Southwest Mountain, Central and Lower City).

3.2 Regression Modelling: Identifying Predictors of Social Capital

The next step in the analysis involved regression modelling. Ordered logistic regression was employed to examine the predictors of social capital perceptions. Three models were devised where the dependent variable represents the four-point social capital classification described above (high, above average, below average, low). In Model 1, the independent variables consist of the three health measures only (self-perceived health, self-perceived mental health, and life stress). In Model 2, they consist of the health variables plus nine socio-demographic measures (immigrant status, sex, age, marital status, household income, education, dwelling type, housing tenure and years lived in neighbourhood). In Model 3, the variable denoting the three neighbourhood groups is added to the health and socio-demographic measures as a single place of residence indicator.

Regression coefficients are employed to estimate odds ratios for each of the independent variables in the three models. The objective is to identify factors associated with changes in social capital perceptions in Hamilton. Bootstrap techniques were utilized in the regression to ensure appropriate inference by correcting for downward bias standard errors and adjusting for intra-cluster correlation. For a more detailed description of ordered logistic regression and the rational for selecting it rather than OLS or binary logistic see Kitchen et al. (2012).

Binary logistic regression was employed to assess the predictors of social capital actions. Two analyses were carried out; the first with volunteering (code 1 = Yes) as the dependent variable and the second with voting (code 1 = Yes) as the dependent variable. Binary logistic regression tests the statistical association between the presence and absence of a condition (such as volunteering or not volunteering) and a set of independent characteristics. Following the same approach as ordered regression described above, the binary logistic analyses involved the phasing in of groups of variables: (1) health, (2) socio-demographic and (3) place of residence (neighbourhood). Similarly, regression coefficients are employed to estimate odds ratios for each of the independent variables in the three models with the objective being to identify factors associated with volunteering and voting. Bootstrap techniques are again employed in the analysis.

3.3 A Social Capital Measurement Tool: Identifying the Social Capital ‘Elite’

The final phase of the analysis employed descriptive statistics to identify the ‘highly-engaged’ citizens of Hamilton, the so-called ‘social-capital elite’. These are people who rank highly on both social capital perceptions and social capital actions. To qualify for this group, a person must satisfy three conditions. First, he or she must have ‘above average’ or ‘high’ social capital perceptions. Second, he or she must have volunteered during the past year. Third, he or she must have voted in the 2010 Hamilton municipal election. The objective is to examine how variations in health, socio-economic status and place of residence influence social capital among residents.

3.4 Weighting

A ‘probability of selection’ weight was employed in the statistical analysis. The weight was calculated as the inverse of the proportion of household survey respondents to the actual number of households in each census tract as determined by the 2006 Census. The equation is as follows:
$$ {\text{W}} = 1/\left( {{\text{n}}/{\text{N}}} \right) $$
The weight adjusts for the fact that the probability of a household being selected for the survey differed among the 11 census tracts comprising the three neighbourhood groups. All analysis in this paper was conducted using the statistical software Stata 11 (http://www.stata.com).

4 Social Capital in Hamilton: Perceptions, Actions and the Social Capital Elite

4.1 Descriptive Statistics: Social Capital Perceptions

Figure 1 shows the distribution of survey respondents (n = 1,002) according to the four categories of social capital perceptions. A majority (53 %) had either ‘high’ (16 %) or ‘above average’ (37 %) perceptions of social capital. Just under half (47 %) of respondents had either ‘below average’ (28 %) or ‘low’ (19 %) perceptions. Figure 2 reveals the distribution of these perceptions among the three neighbourhood groups. Positive perceptions were highest in the Southwest Mountain and Central neighbourhoods—a total of 59 and 56 % respectively in the ‘high’ and ‘above average’ categories. By comparison, substantially fewer respondents in the Lower City had positive perceptions of social capital. Just 7 % had ‘high’ perceptions while 30 % had ‘above average’ perceptions. Nearly two-thirds (64 %) of Lower City respondents had either ‘below average’ or ‘low’ perceptions of social capital.
https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig1_HTML.gif
Fig. 1

Social capital among survey respondents (aged 18+ n = 1,002). Based on Section A questions: social capital perceptions

https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig2_HTML.gif
Fig. 2

Social capital among survey respondents by neighbourhood group (aged 18+ n = 1,002). Based on Section A questions: social capital perceptions

Figures 3, 4 indicate that there is a clear association between social capital and health in Hamilton. As depicted in Fig. 3, about 60 % of survey respondents who reported ‘excellent or very good’ self-perceived health had ‘high’ or ‘above average’ perceptions of social capital. On the other side of the health spectrum about three-quarters (73 %) of respondents who reported ‘fair or poor’ health had ‘below average’ or ‘low’ perceptions of social capital. This gap is even more pronounced with respect to mental health. Again, about 60 % of respondents who said they had ‘excellent or very good’ mental health had ‘high’ or ‘above average’ perceptions. Tellingly, more than three-quarters (77 %) of respondents reporting ‘fair or poor’ mental health had either ‘below average’ (33 %) or ‘low’ (44 %) perceptions of social capital. Figure 5 shows the impact of stress with respondents reporting ‘not at all/not very stressful’ lives having greater levels of ‘high’ (17.5 %) or ‘above average’ (42 %) perceptions of social capital.
https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig3_HTML.gif
Fig. 3

Social capital among survey respondents by self-perceived health (aged 18+ n = 1,002). Based on Section A questions: social capital perceptions

https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig4_HTML.gif
Fig. 4

Social capital among survey respondents by self-perceived mental health (aged 18+ n = 1,002). Based on Section A questions: social capital perceptions

https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig5_HTML.gif
Fig. 5

Social capital among survey respondents by self-perceived life stress (aged 18+ n = 1,002). Based on Section A questions: social capital perceptions

4.2 Descriptive Statistics: Social Capital Actions

Figure 6 reveals that 40 % of all respondents said that they performed unpaid volunteer work during the past year. This figure was higher in the Southwest Mountain (47 %) and Central (48 %) neighbourhoods and substantially less in the Lower City (32 %). Figure 7 indicates that education is an important factor with the most educated respondents having the highest rates of volunteering. Particularly noteworthy is the fact that more than two-thirds (67 %) of respondents with a post-graduate or professional degree indicated that they had volunteered during the past year. With respect to the second social capital action, Fig. 8 shows that 62 % of respondents voted in the 2010 Hamilton municipal election. Voting was highest in the Southwest Mountain (67 %), somewhat lower in the Central neighbourhood (64 %) and markedly less in the Lower City (55  %).1 Similar to volunteering, voting was highest among the most educated respondents—70 % among those with a Bachelors degree and 67.5 % of people with a post-graduate or professional degree (Fig. 9). Finally, Fig. 10 shows a clear link between voting and neighbourhood longevity. An impressive 79 % of respondents who have lived in their neighbourhood 20 or more years said that they voted in the 2010 election.
https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig6_HTML.gif
Fig. 6

Volunteered during the past year by neighbourhood group (Hamilton: respondents aged 18+ n = 1,002). Based on Section B questions: social capital actions

https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig7_HTML.gif
Fig. 7

Volunteered during the past year by highest level of education (Hamilton: respondents aged 18+ n = 1,002). Based on Section B questions: social capital actions

https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig8_HTML.gif
Fig. 8

Voted in the 2010 municipal election by neighbourhood group (Hamilton: respondents aged 18+ n = 1,002). Based on Section B questions: social capital actions

https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig9_HTML.gif
Fig. 9

Voted in the 2010 municipal election by highest level of education (Hamilton: respondents aged 18+ n = 1,002). Based on Section B questions: social capital actions

https://static-content.springer.com/image/art%3A10.1007%2Fs11205-012-0063-3/MediaObjects/11205_2012_63_Fig10_HTML.gif
Fig. 10

Voted in 2010 municipal elections by years lived in neighbourhood (Hamilton: respondents aged 18+ n = 1,002). Based on section B questions: social capital actions

4.3 Regression Modelling: Social Capital Perceptions

Table 3 shows the results of the ordinal regression analyses of social capital perceptions in Hamilton. As indicated in Model 1 and consistent with the results of the descriptive statistics, there is a strong association between changes in social capital and health with each of the three variables being statistically significant. First, survey respondents who reported ‘excellent or very good’ health were 2.8 times (OR = 2.76) more likely to have a positive increase in perceptions of social capital than those reporting ‘fair or poor’ health—the reference group. Second, respondents who said they had ‘excellent or very good’ mental health were 2.3 times (OR = 2.30) more likely to have a positive increase in social capital than those with ‘fair or poor’ mental health. Third, life stress was also found to have an influence. Respondents with ‘not at all or not very’ stressful lives were 1.7 times (OR = 1.67) more likely to have an increase in perceptions of social capital than people with ‘quite a bit or extremely’ stressful lives—the reference group.
Table 3

Results of ordinal logit regression analyses of social capital perceptions in Hamilton, Ontario

Independent variables

Model 1

Model 2

Model 3

Odds ratios

95 % CI

Odds ratios

95 % CI

Odds ratios

95 % CI

Self-perceived health

 Excellent/very good

2.769***

2.143–3.578

2.321***

1.971–2.732

2.011***

1.287–3.140

 Good

2.010***

1.725–2.341

1.786***

1.552–2.056

1.683**

1.093–2.592

 Fair/poor

Reference

 

Reference

 

Reference

 

Self-perceived mental health

 Excellent/very good

2.300***

1.426–3.711

2.163***

1.246–3.755

2.350***

1.451–3.805

 Good

1.161

0.911–1.48

1.218

0.943–1.574

1.303

0.806–2.106

 Fair/poor

Reference

 

Reference

 

Reference

 

Perceived life stress

 Not at all/not very

1.677***

1.617–1.738

1.656***

1.144–2.397

1.690**

1.113–2.567

 A bit

1.270***

1.094–1.474

1.260*

0.971–1.634

1.228

0.805–1.874

 Quite a bit/extremely

Reference

 

Reference

 

Reference

 

Immigrant status

 Immigrant

  

0.860

0.670–1.104

0.848

0.631–1.139

 Canadian born

  

Reference

 

Reference

 

Sex

 Male

  

Reference

 

Reference

 

 Female

  

0.905

0.786–1.041

0.911

0.708–1.174

Age (years)

 18–24 

  

Reference

 

Reference

 

 25–44

  

0.738

0.368–1.479

0.789

0.505–1.231

 45–64

  

0.872

0.518–1.468

0.945

0.602–1.483

 65 and over

  

1.035

0.293–3.657

1.020

0.593–1.754

Marital status

 Married/common law

  

0.977

0.644–1.484

0.927

0.68–1.263

 Widowed/separated/divorced

  

1.351

0.724–2.522

1.286

0.861–1.92

 Single/never married

  

Reference

 

Reference

 

Household income

 Less than $20,000

  

Reference

 

Reference

 

 $20,000 to $49,999

  

1.101

0.826–1.467

0.990

0.65–1.507

 $50,000 to $79,999

  

1.198

0.856–1.677

1.121

0.689–1.823

 $80,000 to $99,999

  

1.850***

1.559–2.195

1.699*

0.927–3.113

 $100,000 or more

  

1.855***

1.461–2.355

1.537*

0.934–2.528

Education

 Less than high school

  

Reference

 

Reference

 

 High school

  

1.506***

1.107–2.048

1.317

0.783–2.213

 Some college or university

  

1.115

0.788–1.578

0.949

0.546–1.649

 Community college

  

1.676***

1.184–2.373

1.435

0.854–2.412

 Bachelors degree

  

1.601***

1.219–2.103

1.243

0.717–2.155

 Post-graduate/professional degree

  

2.013***

1.273–3.185

1.466

0.810–2.653

Dwelling type

 Single detached

  

1.288

0.550–3.018

1.588*

0.922–2.736

 Semi-detached/row/duplex

  

1.705

0.770–3.776

1.972**

1.163–3.342

 Low-rise

  

1.639

0.602–4.464

1.796**

1.039–3.105

 High-rise

  

Reference

 

Reference

 

Housing tenure

 Owner

  

1.444**

1.019–2.046

1.529**

1.026–2.278

 Rent

  

Reference

 

Reference

 

Year lived in neighbourhood

 Less than 5

  

Reference

 

Reference

 

 5–10

  

0.996

0.955–1.039

1.036

0.741–1.45

 11–19

  

1.400**

1.037–1.889

1.386*

0.961–1.998

 20 or more

  

0.987

0.832–1.172

0.945

0.647–1.38

Neighbourhood group

 Southwest Mountain

    

2.020***

1.443–2.826

 Central

    

2.582***

1.824–3.655

 Lower city

    

Reference

 

Cut 1

0.093

−0.122–0.308

0.989

0.01–1.968

1.442

0.61–2.275

Cut 2

1.576

1.425–1.727

2.539

1.449–3.63

3.033

2.191–3.875

Cut 3

3.457

3.248–3.666

4.489

3.446–5.533

5.033

4.141–5.925

Observations

1,002

 

1,002

 

1,002

 

Pseudo R2

0.0515

 

0.0744

 

0.0882

 

Wald χ2

1.34

(df = 6)

1.72

(df = 9)

211.5

(df = 31)

The dependent variable is social capital using a 4 point scale: 1 high, 2 above average, 3 below average, 4 low. The model used for estimation is Ordered Logit. Bootstrap confidence intervals provided

* Significant at 10 %; ** significant at 5 %; *** significant at 1 %. Reference categories are included in the table

In Model 2, socio-demographic indictors were added as control variables. Again, the three health-related measures were significantly associated with perceptions of social capital. It is interesting to note that immigrant status, sex, age and marital status did not have an influence on changes in social capital. Respondents living in households with higher incomes were more likely to have a positive increase in perceptions of social capital ($80,000–$99,999: OR = 1.85 and $100,000 or more: OR = 1.85) than people in households with the lowest income category (less than $20,000)—the reference group. In addition, Model 2 indicates that respondents with the highest level of education (post-graduate or professional degree: OR = 2.013) those who own their homes (OR = 1.44) and people who have lived in their neighbourhood from 11 to 19 years (OR = 1.40) were all more likely to have a positive increase in perceptions of social capital.

In Model 3, the neighbourhood variable was added to the analysis to test the effect of place on social capital. The health indicators (first category) were again found to be associated with social capital. The inclusion of place of residence resulted in some changes in the socio-demographic factors. While high household income and home ownership were again associated with social capital, education was not. Rather, dwelling type was found to have an influence with respondents living in single detached homes (OR = 1.58), semi-detached/row/duplex (OR = 1.97) and low-rise structures (OR = 1.79) all more likely to have a positive increase in perceptions of social capital than people residing in high-rise buildings. Consistent with the findings of the descriptive statistics, Model 3 reveals that residents of the Southwest Mountain (OR = 2.02) and Central (OR = 2.58) neighbourhoods were more likely to have a positive increase in perceptions of social capital than people residing in the Lower City.

4.4 Regression Modelling: Social Capital Actions

Table 4 shows the results of the binary logistic regression analysis with volunteering as the dependent variable. Perhaps not surprisingly, health is associated with volunteering. As Model 1 indicates, respondents with ‘excellent or very good’ health were 2.4 times (OR = 2.45) more likely to volunteer than those with ‘fair or poor’ health, the reference group. Mental health and stress were not associated with volunteering. In Model 2, the nine socio-demographic variables were added to the analysis and health was again associated with volunteering. Interestingly, stress was found to be factor. Respondents reporting that their lives were ‘a bit’ stressful were 1.5 times (OR = 1.53) more likely to volunteer than people with ‘quite a bit or extremely’ stressful lives. Among the socio-demographic indicators, immigrants were less likely (OR = 0.83) to volunteer than Canadian born respondents. In addition, those who are married (OR = 1.60), people living in a household with an income of $50,000–$80,000 (OR = 1.37) and respondents with a Bachelors degree (OR = 3.68) and post-graduate or professional degree (OR = 5.35) were all more likely to volunteer. In Model 3, the neighbourhood variable was added. Again, health, stress, marital status and education were associated with volunteering. The model reveals that residents of Southwest Mountain were 1.4 times (OR = 1.48) and residents of Central also 1.4 times (OR = 1.45) more likely to volunteer than people living in the Lower City, the reference group.
Table 4

Results of logistic regression analyses unpaid volunteer work in the past 12 months, Hamilton, Ontario

Independent variables

Model 1

Model 2

Model 3

Odds ratios

95 % CI

Odds ratios

95 % CI

Odds ratios

95 % CI

Self-perceived health

 Excellent/very good

2.453***

1.833–3.282

1.906***

1.219–2.979

1.785**

1.092–2.919

 Good

1.257***

1.156–1.367

1.143

0.896–1.457

1.106

0.694–1.762

 Fair/poor

Reference

 

Reference

 

Reference

 

Self-perceived mental health

 Excellent/very good

1.251

0.779–2.009

1.071

0.679–1.69

1.082

0.566–2.068

 Good

0.962

0.599–1.545

0.923

0.58–1.467

0.934

0.493–1.769

 Fair/poor

Reference

 

Reference

 

Reference

 

Perceived life stress

 Not at all/not very

1.058

0.923–1.212

1.156

0.853–1.567

1.162

0.748–1.807

 A bit

1.407*

0.975–2.031

1.537***

0.966–2.444

1.523*

0.979–2.369

 Quite a bit/extremely

Reference

 

Reference

 

Reference

 

Immigrant status

 Immigrant

  

0.832***

0.772–0.896

0.820

0.574–1.172

 Canadian born

  

Reference

 

Reference

 

Sex

 Male

  

Reference

 

Reference

 

 Female

  

1.154

0.801–1.663

1.155

0.856–1.559

Age (years)

 18–24

  

Reference

 

Reference

 

 25–44

  

0.917

0.649–1.294

0.949

0.576–1.564

 45–64

  

0.830

0.464–1.484

0.866

0.513–1.462

 65 and over

  

1.071

0.676–1.698

1.061

0.568–1.981

Marital status

 Married/common law

  

1.605***

1.302–1.979

1.555**

1.051–2.299

 Widowed/separated/divorced

  

1.283

0.743–2.214

1.245

0.775–1.999

 Single/never married

  

Reference

 

Reference

 

Household income

 Less than $20,000

  

Reference

 

Reference

 

 $20,000–$49,999

  

1.137

0.963–1.342

1.103

0.697–1.744

 $50,000–$79,999

  

1.378***

1.198–1.585

1.355

0.805–2.28

 $80,000–$99,999

  

1.026

0.542–1.94

0.981

0.49–1.963

 $100,000 or more

  

0.951

0.753–1.201

0.872

0.511–1.49

Education

 Less than high school

  

Reference

 

Reference

 

 High school

  

1.002

0.836–1.200

0.939

0.556–1.585

 Some college or university

  

2.604***

1.794–3.779

2.445***

1.338–4.465

 Community college

  

1.500***

1.134–1.983

1.405

0.809–2.440

 Bachelors degree

  

3.680***

2.301–5.888

3.379***

1.888–6.049

 Post-graduate/professional degree

  

5.354***

2.290–12.52

4.795***

2.561–8.980

Dwelling type

 Single detached

  

1.395

0.746–2.606

1.484

0.779–2.827

 Semi-detached/row/duplex

  

1.221

0.728–2.049

1.245

0.674–2.3

 Low-rise

  

1.357

0.619–2.977

1.393

0.729–2.662

 High-rise

  

Reference

 

Reference

 

Housing tenure

 Owner

  

1.095

0.527–2.275

1.099

0.674–1.792

 Rent

  

Reference

 

Reference

 

Year lived in neighbourhood

 Less than 5

  

Reference

 

Reference

 

 5–10

  

1.226

0.754–1.993

1.236

0.824–1.854

 11–19

  

1.192

0.850–1.671

1.170

0.763–1.794

 20 or more

  

0.954

0.687–1.326

0.953

0.578–1.569

Neighbourhood group

 Southwest Mountain

    

1.483*

0.975–2.256

 Central

    

1.459*

0.967–2.203

 Lower city

    

Reference

 

Observations

1,002

 

1,002

 

1,002

 

Pseudo R2

0.0378

 

0.1106

   

Wald χ2

14,554

(df = 6)

 

(df = 8)

 

(df = 31)

The dependent variable is binary (volunteering) 1 yes, 0 no. The model used for estimation is Logistic. Bootstrap confidence intervals provided

* Significant at 10 %; ** significant at 5 %; *** significant at 1 %. Reference categories are included in the table

Table 5 shows the results of the binary logistic regression with voting in the 2010 municipal election as the dependent variable. Interestingly, Model 1 indicates that respondents with ‘good’ self-perceived health were less likely (OR = 0.67) to vote than people with ‘fair or poor’ health, the reference group. Stress was also a factor with respondents reporting that they had ‘not at all or not very’ stressful lives being 1.6 times (OR = 1.64) more likely to vote than those with ‘quite a bit or extremely’ stressful lives. However, when the socio-demographic variables were added to the analysis in Model 2, health was no longer associated with voting. The model indicates that immigrants (OR = 0.54) were less likely to vote than Canadian born respondents and females (OR = 0.64) were less likely to vote than males. There was also a clear association between age and voting with respondents 45–64 years being 3.3 times (OR = 3.33) and seniors (65+) being 8.7 times (OR = 8.66) more likely to vote than those aged 18–29. Model 2 also reveals that voting was associated with higher education (post-graduate/professional degree: OR = 2.22) and more lengthy neighbourhood longevity (20 or more years: OR = 2.21). In Model 3 when the neighbourhood variable was added, the statistical structure (in terms of significant odds ratios) of the results remained largely the same as Model 2. While respondents living in Southwest Mountain (OR = 1.44) and Central (1.46) were found to be more likely to vote than residents of the Lower City, the results were significant at the 10 % level only.
Table 5

Results of logistic regression analyses voted in last municipal election, Hamilton, Ontario

Independent variables

Model 1

Model 2

Model 3

Odds ratios

95 % CI

Odds ratios

95 % CI

Odds ratios

95 % CI

Self-perceived health

 Excellent/very good

0.993

0.667–1.479

1.332

0.850–2.087

1.231

0.701–2.163

 Good

0.678***

0.592–0.777

0.873

0.693–1.100

0.836

0.483–1.447

 Fair/poor

Reference

 

Reference

 

Reference

 

Self-perceived mental health

 Excellent/very good

1.644

0.919–2.943

1.376

0.659–2.871

1.406

0.776–2.549

 Good

1.154

0.715–1.862

0.997

0.516–1.926

1.016

0.561–1.841

 Fair/poor

Reference

 

Reference

 

Reference

 

Perceived life stress

 Not at all/not very

1.644***

1.32–2.047

1.175

0.698–1.976

1.190

0.762–1.86

 A bit

1.383***

1.131–1.692

1.303

0.904–1.878

1.295

0.847–1.98

 Quite a bit/extremely

Reference

 

Reference

 

Reference

 

Immigrant status

 Immigrant

  

0.546***

0.416–0.718

0.536***

0.369–0.778

 Canadian born

  

Reference

 

Reference

 

Sex

 Male

  

Reference

 

Reference

 

 Female

  

0.644***

0.481–0.863

0.648***

0.467–0.899

Age (years)

 

 18–24

  

Reference

 

Reference

 

 25–44

  

1.433

0.413–4.969

1.489

0.908–2.442

 45–64

  

3.332***

1.331–8.342

3.516***

2.087–5.924

 65 and over

  

8.667***

4.504–16.68

8.568***

4.398–16.692

Marital status

 Married/common law

  

1.307

0.842–2.028

1.273

0.849–1.910

 Widowed/separated/divorced

  

0.995

0.625–1.582

0.962

0.587–1.576

 Single/never married

  

Reference

 

Reference

 

Household income

 Less than $20,000

  

Reference

 

Reference

 

 $20,000–$49,999

  

1.151

0.616–2.151

1.104

0.683–1.784

 $50,000–$79,999

  

1.128

0.627–2.029

1.090

0.632–1.881

 $80,000–$99,999

  

1.116

0.549–2.269

1.056

0.520–2.145

 $100,000 or more

  

1.143

0.860–1.521

1.031

0.568–1.872

Education

 Less than high school

  

Reference

 

Reference

 

 High school

  

1.402***

1.118–1.756

1.310

0.799–2.148

 Some college or university

  

1.895**

0.993–3.614

1.755*

0.942–3.269

 Community college

  

1.611***

1.424–1.824

1.496

0.83–2.694

 Bachelors degree

  

2.250***

1.721–2.942

2.036**

1.112–3.726

 Post-graduate/professional degree

  

2.224**

1.185–4.175

1.968**

1.015–3.813

Dwelling type

 Single detached

  

1.469

0.582–3.704

1.561

0.838–2.909

 Semi-detached/row/duplex

  

1.177

0.675–2.053

1.207

0.64–2.273

 Low-rise

  

1.830

0.686–4.879

1.892*

0.961–3.724

 High-rise

  

Reference

 

Reference

 

Housing tenure

 Owner

  

1.471*

0.986–2.194

1.499*

0.982–2.286

 Rent

  

Reference

 

Reference

 

Year lived in neighbourhood

 Less than 5

  

Reference

 

Reference

 

 5–10

  

1.597***

1.350–1.889

1.619**

1.072–2.445

 11–19

  

1.597***

1.335–1.910

1.583**

1.009–2.484

 20 or more

  

2.213***

1.528–3.205

2.213***

1.345–3.639

Neighbourhood group

 Southwest Mountain

    

1.448*

0.960–2.184

 Central

    

1.465*

0.985–2.179

 Lower city

    

Reference

 

Observations

1,002

 

1,002

 

1,002

 

Pseudo R2

0.0251

 

0.1480

 

0.1516

 

The dependent variable is binary (voted in municipal election) 1 yes, 0 no. The model used for estimation is Logistic. Bootstrap confidence intervals provided

* Significant at 10 %; ** significant at 5 %; *** significant at 1 %. Reference categories are included in the table

4.5 Social Capital Elite: The Healthy Wealthy Senior

The final step in the analysis was to identify the social capital elite in Hamilton. These are people who have high or above average perceptions of social capital and who volunteered and voted. A total of 205 respondents, representing 20.5 % of the sample (n = 1,002), fell into this category. Table 6 shows the distribution of the social capital elite compared to other respondents according to selected indicators. In terms of place of residence, the table indicates that a higher proportion of the social capital elite live in the Central neighbourhood compared to other respondents, a difference of 16.5 % between the two groups. The Lower City is home to a small proportion of the social capital elite compared to other respondents, a difference of −23.1 %.
Table 6

Distribution of social capital elite and other respondents by selected characteristics (n = 1,002)

Variable

Social capital elite (%) (n = 205)

Other respondents (%) (n = 797)

Difference

Neighbourhood

 Southwest Mountain

33.9

27.3

6.6*

 Central

50.8

34.3

16.5***

 Lower city

15.3

38.4

−23.1***

 Total

100

100

Self-perceived health

 Excellent/very good

73.1

48.8

24.3***

 Good

19.3

31.6

−12.3***

 Fair/poor

7.6

19.6

−12.0***

 Total

100

100

Self-perceived mental health

 Excellent/very good

82.7

58.9

23.8***

 Good

14.2

29.9

−15.7***

 Fair/poor

3.1

11.2

−8.1***

 Total

100

100

Sex

 Male

40.3

41.2

−0.9

 Female

59.7

58.8

1.2

 Total

100

100

Age (years)

 18–29

10.5

12.3

−1.8

 30–44

25.5

27.0

−1.5

 45–64

35.8

40.9

−5.1

 65 and over

28.2

19.8

8.4**

 Total

100

100

Household income

 Less than $20,000

8.2

16.9

−8.7***

 $20,000–$49,999

27.8

35.0

−7.2**

 $50,000–$79,999

21.8

19.8

2.0

 $80,000–$99,999

10.1

7.7

2.4

 $100,000 or more

32.1

20.6

11.5***

 Total

100

100

Housing tenure

 Owner

79.5

62.3

17.2***

 Renter

20.5

37.7

−17.2***

 Total

100

100

Dwelling type

 Single-detached

71.5

56.1

15.4***

 Semi-detached/row/duplex

10.6

16.6

−6.0**

 Low-rise

9.2

10.5

−1.3

 High-rise

8.7

16.8

−8.1***

 Total

100

100

* Significant at 10 %; ** significant at 5 %; ***significant at 1 %

Health is clearly a defining feature of this select group of residents. Table 6 reveals that a significantly higher proportion of the social capita elite reported ‘excellent or very good’ health compared to other respondents (a difference of 24.3 %) and reported ‘excellent or very good’ mental health (a difference of 23.8 %). Interestingly, there was no significant difference in the distribution of males and females between the two groups. Similarly, there were no significant differences for the first three age groups—18–29, 30–44 and 45–64. However, a greater proportion of seniors (age 65+) were among the social capital elite compared to other respondents, a difference of 8.4 %. With respect to household income, the social capital elite were less represented in the two lowest income categories (under $20,000 and $20,000–$49,999) than other respondents (−8.7 and −7.2  % respectively) and significantly more represented in the highest income category ($100,000 or more), a difference of 11.5 %. Furthermore, nearly 80 % of the social capital elite are homeowners compared to 62 % of other respondents, a difference of 17.2 %. Finally, about 72 % live in single-detached homes, a proportion 15-percentage points higher than the other respondents in the study.

This brief analysis points to an association between the social capital elite and certain heath and socio-economic characteristics. Good health, higher income and home ownership are defining features. The data in Table 6 suggests that one prominent group among the social capital elite are senior citizens many of whom are maintaining good physical and mental health and are enjoying a relatively high quality of life. The ‘healthy, wealthy’ senior is certainly among the most highly engaged citizens in Hamilton, a positive development for the continued social well-being of the city.

5 Discussion

This paper followed two approaches in measuring social capital in Hamilton, Ontario. The first employed the Social capital Measurement Tool (SCMT) while the second used more complex modelling to test the association between social capital and factors relating to health, SES and place of residence. The SCMT is a simple additive measure based on the responses to four questions relating to social capital perceptions (safety, trust, seeking help, and multiculturalism). Each survey respondent was assigned a level of social capital: (1) high, (2) above average, (3) below average, (4) low. The SCMT also accounts for a person’s social capital actions relating to volunteering and voting. The tool can be used to identify respondent with different levels of perceptions and different types of action (such as those who volunteer and vote and those who don’t). We feel that the SCMT is a relatively straightforward way to account for differences in social capital among respondents and can be applied at a number of scales such as the neighbourhood, city or region. Furthermore, the SCMT can be readily applied by academics for research purposes and members of the non-academic community (e.g., planners and local organizations) for policy purposes.

In summary, the study found that there was a strong association between social capital and health in Hamilton, a finding consistent with the extensive literature on this topic (Kawachi et al. 1999; Putnam 2000; Kim and Kawachi 2006). Respondents with positive self-perceived health and mental health were far more likely to have high or above average social capital perceptions. Stress was also linked to social capital with people having less stressful lives being more likely to have positive social capital. With respect to place of residence, social capital was greater in the high and mixed SES neighbourhoods and much weaker in the low SES neighbourhoods. In terms of social capital actions, both volunteering and voting was associated with positive health, a result in accordance with research by Schultz et al. (2008) on civic participation. The study revealed that social capital perceptions and actions in Hamilton were associated with certain socio-demographic characteristics including higher levels of education, home ownership and residential longevity.

Finally, the paper found that seniors with good health and higher incomes were among the most highly engaged citizens in Hamilton. This finding provides an avenue for further research, which could focus on a more detailed analysis (possibly though regression modelling employing interaction terms) of the characteristics that influence the emergence of the so-called social capital ‘elite’ in a community. In addition, the relationship between social capital perceptions and actions requires further investigation with one likely having a significant influence on the other. For instance, a person with low social capital perceptions may be less inclined to volunteer or vote. Indeed, the paper found that the residents of the Lower City (low SES) had overall weaker social capital perceptions and were less likely to volunteer and vote compared to their counterparts in the Central (mixed SES) and Southwest Mountain (high SES) neighbourhoods. Another area of further research, geared specifically to Hamilton, would be an in-depth analysis of how the physical features of a community (e.g., layout and design, types of housing, proximity to services, accessibility to recreational and cultural amenities) affects a person’s social capital. The work conducted by Veenstra (2005) in British Columbia could be used as a guide in this respect.

Footnotes
1

The overall voter turnout for the 2010 municipal election was 41 %. An analysis of poll data, however, revealed that voter turnout was higher in the wards associated with the sampled neighbourhoods.

 

Copyright information

© Springer Science+Business Media B.V. 2012