Ethnicity and social capital in Canada, A Aizlewood, R Pendakur

Tags: social capital, interpersonal trust, Canada, community, neighbours, visible minority, visible minorities, scores, individuals, civic attitudes, friends, behaviours, dummy variable, respondents, dummy variables, variables, models, social networks, formal participation, community size, ethnic groups, ethnic diversity, visible minority groups, cultural diversity, Canadian Ethnic Studies, community level, community characteristics, Social Development Canada, community levels, Department of Canadian Heritage, Community-level, immigrant status, social interactions, variable, Princeton University Press, Yale University Press, University of Chicago Press, organisations, Canadian census, Canadian Journal of Political Science, Harvard University, Equality Security Community Survey, Marital Status, social contact, ethnically diverse, traditional ties, Montreal neighbourhoods, social interaction, diverse communities, regression coefficients, Table entries, Statistics Canada
Content: ETHNICITY AND SOCIAL CAPITAL IN CANADA
Amanda Aizlewood Department of Canadian Heritage
by Ravi Pendakur Social Development Canada 2005
Canadian Ethnic Studies, volume 37 (2)
** The opinions expressed in this report are those of the authors and do not necessarily reflect the views of the Department of Canadian Heritage or Social Development Canada.
Abstract A review of recent research suggests that ethnic heterogeneity negatively influences attitudes and behaviours necessary for community cohesion. Using data drawn from the 2000 Equality Security Community survey, we test the applicability of these findings to the Canadian context. Our results suggest that individual ethnocultural diversity characteristics do not appreciably affect scores on standard social capital measures but that broad, designations of ethnicity such as visible minority and immigrant status do so, however weakly. Larger community size is a consistent predictor of lower interpersonal trust, lower propensity to join organizations, and less time spent with friends. We speculate that in Canada, where community size, diversity, wealth, and education are so closely and positively correlated, an urban lifestyle, or `city effect' may be a more accurate predictor of civic attitudes and behaviours. 1. Introduction. Social capital is, in theory at least, about working together to pursue common objectives. It manifests itself as specific forms of social conduct ­ including networks, norms, and trust ­ that allow people to interact in ways that encourage the production of social goods. The past decade has spawned a good deal of work that has sought to define social capital and to identify the appropriate personal and community characteristics that lead to higher levels of community cooperation, trust and participation (see in particular Putnam, 1993; Putnam, 1995; Putnam et al, 1993; Fukuyama, 1995; Brehm and Rahn, 1997). To date, social capital has been shown to be positively associated with high levels of education, confidence in public institutions, and participation in social, cultural and political activities. These positive associations are appealing to policy makers given the hypothesis that trust, reciprocity, and cooperation will correlate with higher outcomes on quality of life indicators such as public safety, health, and life satisfaction and create value not only for individuals, but for communities and societies in general. Not all studies have yielded encouraging results. Though a great deal of attention has been focused on the identification of `good' conditions for the creation of social capital, a small body of work carried out in the United States suggests that other things ­ ethnic diversity in particular ­ may work to reduce it. These findings are not limited to qualitative conjecture; a modest body of survey-based empirical research consistently shows that negative relationships exist between measures of social capital and indicators of ethnic diversity at the individual and community levels (see Alesina and Ferrara, 1999, 2000; Saguaro Seminar, 2000; Marschall and Stolle, 2002; Glaeser, et al, 2000, Johnston and Soroka, 2001). Some hypothesise that it is the differing levels of `civic-ness' exhibited by ethnic groups themselves that account for lower overall stocks of social capital (see Putnam, 1993; Inglehart, 1988, 1990; Fukuyama, 1995; Banfield, 1958; Verba, Schlozman and Brady, 1995; Rice and Feldman, 1997; and Black, 1982, 1987). Others maintain that the answers lie in the social-psychological aspects of majority-minority societal interaction and that the presence of diverse elements in communities inhibits trust and reduces civic and cooperative behaviour among dominant groups (see Forbes, 1997, Uslaner, 2002; Miller, 1995, Coleman, 1990). 2
From a Canadian perspective, these arguments suggest that there is reason to worry. Canada is a country with high immigration associated with settlement in large cities that serve as homes to ethnically diverse populations. A domestic line of inquiry seems prudent: do indicators of social capital vary based on ethnic origin and degree of ethnic diversity in Canada as they do in the United States? Using data from the 2000 Equality Security Community survey, we test these hypotheses in the Canadian context. We look at the broad relationships between ethnic diversity and social capital in order to further our understanding of the impact of ethnicity and ethnic diversity on civic attitudes and behaviours in Canada. This study moves beyond the scope of previous efforts by examining not only visible ethnic differences, but also the specific attitudes and behaviours of twenty-two distinct ethnocultural groups. We compare the relative and absolute strengths of individual and community level characteristics in predicting outcomes on social capital measures. Individual-level characteristics refer to personal markers, such as ethnocultural ancestry, immigrant status, and visible minority status. Community-level characteristics refer to the features of the environment in which an individual lives, such as the local proportion of visible minorities, the area's cultural diversity, and the size of the community. We seek to further our understanding of a) the impact of community-level characteristics on civic attitudes and behaviours, b) the impact of individual-level characteristics on civic attitudes and behaviours, and c) which, if any, holds greater explanatory power in the Canadian context. Our results suggest that in Canada, individual ethnocultural characteristics do not appreciably affect scores on social capital measures but that broad designations of ethnicity such as visible minority and immigrant status do so, however weakly. Community-level factors play a more ambiguous role: we find that larger community size in particular is a consistent predictor of lower interpersonal trust, lower propensity to join organizations, and less time spent with friends. We also find that scores on social capital indicators tend to increase as community-level diversity increases but that such differences do not withstand the addition of more individual-level ethnocultural attributes. Individual sociodemographic characteristics such as age, education and income hold greater explanatory power that remains consistent despite the presence of other statistical controls. We speculate that in Canada, where community size, diversity, wealth, and education are so closely and positively correlated, an urban lifestyle, or `city effect' may be a more accurate predictor of civic attitudes and behaviours. 2. The Literature. Why would diversity interfere with or reduce trust and cooperative behaviour among citizens? Social capital writ large represents the potential willingness of individuals to work together and to engage in cooperative civic endeavours collectively. The concept of social capital has thus been operationalised to mean the willingness and ability of individuals to create and sustain relationships through voluntary associations (Portes and Landolt, 1996). This role is an important one in societies that require senses of community, solidarity and mutual obligation in order to support collective undertakings such as a welfare state (see Soroka, Johnston and Banting, forthcoming). There is a fundamental weakness with such characterizations, however, since social capital does not necessarily have to be inclusive in nature. High levels of trust and association within a group do not necessarily translate to high levels of trust and networking when dealing with people from outside the group. It is well documented, for example, that ethnic 3
enclave economies draw from in-group members, and rarely from those outside the group (see for example, Portes and Stepick, 1993; Li, 1998). In ethnically diverse settings such as urban areas, broader community cohesion may be threatened if individuals from both majority and minority populations bond on ethnic terms to the exclusion of others, thereby creating socially isolated communities that do not trust or cooperate with each other. Despite the fact that Social Networks based on ethnic criteria present outward signs of high social capital and are useful to members in social and economic terms, they are based on exclusive rather than inclusive criteria and thus do not necessarily result in high levels of cohesion across groups (see Varshney, 2002; Portes and Landolt, 1996; Etzioni, 2001). This dilemma is noted in a broad cross-section of social capital literature, from which two hypotheses can be identified. The first hypothesis suggests that broad social capital is easier to build in ethnically homogeneous communities and is more difficult to sustain in ethnically diverse settings. Individuals are more disposed to trust others like themselves, and less likely to trust those they perceive to be different (Uslaner, 2002). In this view, ethnocultural homogeneity produces a more smoothly functioning society into which the insertion of diverse elements creates friction and stress. Homogeneity facilitates trust and in turn, cooperation because `like' individuals share cultural values and expectations, or what Granovetter called `strong ties' (1973; see also Coleman, 1990). Conversely, diversity creates real or perceived barriers in communication and understanding among people and leads to apprehension among so-called `unlike' individuals (see in particular Miller, 1995, Eisenberg, 1999 and Eisenberg, forthcoming). Majority groups in particular are believed to be negatively affected by diversity because the presence of `difference' threatens individual or collective identities (Forbes, 1997; Wilton, 1998). The results are reduced senses of trust and lower levels of cooperative activity. If group identity is threatened, it may lead to the withdrawal of formerly dominant groups from participation in the community and, in turn, lower social capital. In this context-based line of reasoning, it is the diversity that surrounds an individual or group on a micro- or macro-scale that influences attitudes and in turn, behaviours. A second hypothesis proposes that it is an individual's group-based ethnocultural characteristics that drive his or her attitudes and behaviours, or what Johnston and Soroka (2001) term `compositional' arguments. In this view, social capital is a cultural attribute that is inherited or learned, and not easily shed. As a result, some ethnocultural groups have more or less `civic' qualities than others, and these qualities present themselves in personal attitudes and behaviours that remain constant across generations, and even oceans (see Putnam, 1993; Inglehart, 1988, 1990; Fukuyama, 1995; Banfield, 1958; Verba, Schlozman and Brady, 1995; Rice and Feldman, 1997; and Black, 1982, 1987). The social capital of a community is the sum of an individual's, and in turn, group's contribution. In the aggregate, if a community has high social capital, it is because it is composed of civic-minded groups. The overall stock of social capital can be reduced if groups with low social capital enter the community, a situation that seems particularly salient for societies with large-scale, ongoing immigration. In this composition-based line of reasoning, it is the personal ethnocultural identification of an individual or group that determines outcomes on social capital measures, and in turn the social capital of a community. Both hypotheses can and have been supported empirically, and as such, most surveybased studies incorporate both individual and contextual indicators in an attempt to identify the key predictors of social capital measures. The vast majority of research in this domain is 4
conducted in the United States, and results for the Canadian context are few. Results from the Social Capital Community Benchmark project in the United States show that scores on trust and participation are substantially lower in ethnically diverse communities (Saguaro Seminar, 2001). Individuals in such communities are less likely to trust others, are more likely to be personally isolated, have fewer friends, spend less time socialising with friends and family, and express a lower sense of community. These results, suggest the authors, hint at a pervasive, generalized mistrust in diverse settings: despite the expectation that individuals may be less trusting of members of other ethnocultural groups, researchers have found that in areas of high ethnic diversity, lack of trust is indiscriminate and extends to members of the respondents' own ethnic group. Johnston and Soroka's study of the Canadian component of the World Values Survey examines these issues directly, but poor available indicators of ethnicity and community-level diversity hamper the usefulness of its findings (2001). They do find that broadly speaking, both individual ethnic and community level diversity affect scores on social capital measures, but their results fail to definitively uphold the hypotheses of either the contextual or compositional arguments. They find that ethnic differences in social capital do exist, and while somewhat blunted, remain important even in the face of statistical controls such as age and education. In Canada, provinces with high ethnic diversity do not demonstrate lower scores on civic measures, but rather the opposite: diverse regions present high scores on social capital measures. In the United States, results do not appear to vary based on which element of social capital ­ be it trust, participation or interaction ­ is being studied. Alesina and Ferrara (2000) find that ethnic origin and Religious affiliation alone are not significant predictors of trust, but being unsuccessful in terms of income and education and living in an ethnically diverse area all negatively influence the degree to which trust is extended to others. In a separate study, the same researchers find that after controlling for individual characteristics, scores on participation measures are significantly lower in more ethnically fragmented communities. They conclude that on the whole, individuals "prefer to interact with others who are similar to themselves in terms of income, race or ethnicity [and] prefer to join groups composed of individuals with preferences similar to their own" (Alesina & Ferrara, 1999: 2). Overall, the authors' "aversion to heterogeneity" hypothesis is supported: for participation and the development of trust, individuals require similarity rather than difference. This seemingly broad negative relationship also extends to more intimate settings. The acquisition of close, personal networks reflects position in the social structure which in turn intersects with "dimensions of opportunity and constraint", argues Fischer in his study of urban networks (1982: 11). Educational resources, employment status, and ethnic membership all combine to expose individuals to specific groups of contacts from whom opportunities and advantages are gained. These findings are reinforced by the results of Glaeser, et al, who observe that in personal encounters, differences in race and nationality reduce the level of perceived trustworthiness (2000). The notion that likeness and familiarity breed trust is also reinforced by social-psychological literature, in which individuals are shown to be most strongly influenced by members of their own `primary' groups, namely those with whom they engage in regular, daily interaction (Frank and Yasumoto, 1998). A 1994 study of self-help support groups found that interaction in small, diverse settings leads to lower trust among participants (Wuthnow, 1994). Using a survey of 1000 American support group members, Wuthnow (1994) finds that in general, individuals in heterogeneous groups are less likely to report high levels of trust than those in homogeneous groups. In a related argument, Wilton suggests that negative reactions occur in 5
small, yet diverse settings because individuals encounter one another in places they think of as their own, "with people who don't appear to belong" (1998: 178). Other studies find that an accounting of both context and contact is needed to explain the relationship between diversity and civic attitudes and behaviours. With respect to context, beneficial social interactions can be obtained not simply through participation in formal activities, but also in informal activities such as social encounters (Foley, Edwards, and Diani, 2001). The frequency of such contact is important, but encounters are made especially meaningful if interaction take place in ethnically diverse settings. Stein et al (2000) do suggest that for majority groups at least, living in diverse settings result in lower trust scores. But they offer an interesting corollary: that an increase in the frequency of contact between groups significantly offsets the negative effects of a diverse context. Simply living near minority groups but not interacting with them increases feelings of threat among the majority group, a finding that reinforces the results of Jackman and Crane's earlier study of black and white communities in the United States in which similar effects were observed (1986). Marschall and Stolle (2001) also find that context plays an important positive role in shaping civic attitudes and behaviours. Despite their initial observation that individuals in heterogeneous areas are much less likely to trust others than individuals in more homogeneous areas, they find that interaction among individuals in ethnically diverse settings plays an important role in the development of generalized trust. When individuals living in heterogeneous neighbourhoods have positive, direct interactions with members of other ethnic groups, they are also more likely to extend a sense of trust to other strangers. Positive social interactions among `unlike' individuals contribute more to the development of generalized trust and civic orientations than similar experiences among individuals who shared common characteristics, attitudes and behaviours. The relationship between an ethnically diverse context and civic attitudes and behaviours seems particularly relevant to Canada. Canada is one of the most urbanized countries in the world, with close to 80 percent of citizens living in cities or metropolitan areas (Statistics Canada, 2002). Urban living occupies a central role in studies of social contact because of the implication of city life in the decline of traditional forms of community-based interaction (see Guest and Wierzbicki, 1999). Traditionally, `community' has referred to something resembling a neighbourhood, an entity that implies physical ties based on geographical proximity or spatial location. People interact and are familiar because they encounter one another regularly in the street. The term has since been reconceptualized in terms of social networks, especially nongeographic, non-local ones (see for example, Wellman, 1999). These are non-local because they arise from something other than geographic proximity and focus on interests, social relations and social networks as a basis for relationships. These networks may threaten the production or maintenance of social capital if urbanization is seen as the producer of "a social order in which the traditional ties of community-shared space, close kinship links, shared religious and moral values were being replaced by anonymity, individualism and competition." (Forrest and Kearns, 2001). There is some evidence to support this fear. Guest and Wierzbicki find in their 1999 study in the United States that there is "a relatively continuous, albeit slow, decline in the importance of social ties on the basis of neighbourhood and, in turn, an even smaller upward growth in the importance of non-neighbourhood social ties...neighbourhood and non-neighbourhood ties are becoming disassociated over time" (109). 6
Coupled with the effects associated with urbanization is the reality that immigration and ethnic diversity in Canada are essentially urban phenomena. In Canada, diverse communities are analogous to large communities. In 2001, 48 percent of immigrants and refugees settled in Toronto, 15 percent in Vancouver, 12 percent in Montreal. Compared to the United States, however, there is less ethnocultural segregation in Canadian urban spaces, thus altering the impact of ethnocultural diversity in comparison to the way cities are characterized in American social capital studies. Germain's 2000 study that finds that in Montreal neighbourhoods, recent immigrants tend to share residential space with others of different origins, and with people of French or British origins, "creating a much more cosmopolitan landscape where diversity rather than ethnic or racial concentration is the rule" (13). Poverty is neither as spatially nor as ethnically determined in Canada, and there is greater diversity of ethnocultural groups in poorer neighbourhoods than in comparable neighbourhoods of American cities. 3. Data, Measures and Methods. 3.1 The Data. What is the impact of ethnicity and ethnic diversity on social capital in Canada? In order to study these questions directly, we use the Equality Security Community Survey (ESC), a large, national, and stratified random sample of the Canadian population who speak an official language (English or French) and are over eighteen years of age. In order to include the contextual factors of local diversity and city size as controls in the analysis, we augment the ESC by linking it to municipal data from the 1996 Census. The ESC survey is particularly suited for the investigation of issues of ethnicity and social capital for a number of reasons. First, the survey was administered to 5,152 respondents, 1,051 of whom live in census tracts in Montreal, Toronto and Vancouver with four times the average number of visible minorities for that Census Metropolitan Area (CMA). This urban over-sample means that the total number of visible minorities available for meaningful analysis is substantial for a survey of this size. Second, the ESC features proportional weight, cluster and strata information, all of which we use to allow results to be generalized to the population.1 Third, the survey has a large and varied set of questions on ethnic identity, ancestry and affiliation that far exceed the depth publicly available in other surveys, including the Canadian census. Fourth, it poses detailed questions related to interpersonal and political trust, forms of participation, and formal and informal kinds of social interaction, all of which are widely used to operationalise the concept of social capital. 1 The general consensus is that on repeated draws from the population, weighting will get you to the true estimate quicker than not weighting. However, there is a debate as to whether analysis of non-random samples should be weighted. It is our feeling that given the ESC is a complex, multi stage sample (because of the oversamples), analysis should recognize the sample design. If one chooses not to weight, one option is to include all the variables which make up the weighting structure in the regression. This is the approach favoured by Soroka et al. However, one-third of respondents did not provide postal code information. We chose to weight the data because our interest was in looking at detailed ethnicity characteristics and we could not afford the loss of sample size. Our approach uses the full weighting information available on the dataset. These include 13 strata and 153 clusters. 7
We examine the relationship between personal trust, political trust, participation and social interaction as dependent variables on one hand and an array of personal and communitylevel characteristics as independent variables on the other. Ours is a general study; we intend only to reveal the strength of relationships between dependent and independent variables in each of five separate models, and do not attempt to uncover causal relationships across the dependent variables themselves.2 3.2 Dependent Variables: Measures of Social Capital. We operationalise social capital according to four thematic indicators: interpersonal trust, political trust, formal participation, and informal social interaction. The indicator of interpersonal trust is constructed using the aggregate of four related questions concerning the fate of a lost wallet. This instrument of inquiry in the area of trust is unique to the ESC survey. The respondent is asked whether a wallet containing $200 would be returned intact if it were found by a neighbour, a police officer, a grocery clerk at a local shop, or a complete stranger. The resulting variable is measured on a nine-point scale, on which a low figure represents low trust and a high figure represents high trust. Our measure for political trust is the sum of two questions that ask whether the respondent trusts the federal or provincial government "to do what is right". The new variable is measured on a seven-point scale, ascending numerically from low trust to high trust. We enlist two variables to measure informal social interaction: frequency of seeing friends and frequency of talking to neighbours. Both are measured on seven-point scales, ascending numerically from low to high frequency. Formal participation is measured using a binary variable that taps respondents' propensity to join organisations. It is drawn from seven questions that probe membership in service, community, and recreational clubs, political groups, youthoriented service organizations, volunteer organisations, or any other type of formalised organisation. As long as the respondent is a member of any club or organisation, he or she is considered to have joined. 3.3 Independent Variables: Contextual Level. Our independent variables come in two varieties: individual-level variables and contextual/community-level variables. Community-level variables are so called because they refer to the general characteristics of the area in which a respondent resides, rather than to the personal attributes of each respondent. Each variable is linked on a case basis to census subdivision (CSD) population and ethnicity data from the 1996 census. In this way, three variables are derived for the analysis. The first is the natural log of the total population of the CSD in which a respondent lives (size of the community). Community size is important for two reasons: first, larger communities in Canada are more likely to be home to a diverse population and a larger proportion of immigrants, and second, larger communities are expected to exhibit different attitudes and behaviours from smaller communities. The two other contextual variables measure the diversity of a respondent's community. The `index of diversity' variable measures the chance that two randomly selected cases within the CSD will differ. As the index value 2 For discussions of causal relationships between indicators of social capital, see Brehm and Rahn, 1997; Muller and Seligson, 1994. 8
increases, contextual diversity increases. The final contextual variable is the percentage of visible minorities in the respondent's CSD.3 3.4 Independent Variables: Individual Level. Individual level variables are divided into two types. The first type includes the basic demographic markers that exist independently of ethnocultural ancestry but are implicated in discussions of social capital.4 This group includes age, sex, educational attainment, marital status, employment status, subjective income, and whether the respondent has children living in the household. The second group includes variables that are ethnocultural in nature. This group includes ethnic ancestry, household language, religious affiliation, visible minority affiliation, immigration status, and for immigrants only, the length of time a respondent has lived in Canada.5 The particular derivation of the ethnic ancestry variable merits a more detailed explanation. Categorizing ethnic identity is a complex and amorphous task because there are myriad ways in which people may define membership in a particular group. Identity in ethnocultural terms weaves together a broad spectrum of concepts, including ancestry, religion, language and socialization to name just a few. In general terms, however, ethnicity refers to a group's distinctiveness and is therefore also a measure of culture. This ethnocultural grouping comprises a bundle of membership criteria, including: self categorization and identification as means to identify group members and non-members; a shared descent that suggests commonality and includes a notion of history; specific cultural traits such as customs or language; and a social organization for interaction both within the group and with people outside the group. These are, by and large, ascribed criteria; individuals have these characteristics through cultural reproduction and knowledge acquisition. It is not necessary for members to fulfill all criteria, but a few of the criteria must be fulfilled for the essence of the ethnicity to be maintained. The ESC is a rich source for information on ethnic affiliation thanks to its unique collection instrument. A set of four questions asks specifically about ethnic affiliation. The first asks "To what ethnic or cultural group(s) do you belong?" and allows the respondent up to four mentions, prompted by the question "Are there any others?" With these data, it is possible to construct a clear picture of ethnocultural origins, including in particular individuals with multiple origins. If the respondent answers "Canadian" as a first response, a probe question seeks additional information: "In addition to being a Canadian, to what ethnic or cultural group did you or your ancestors belong upon first coming to this continent?" For our purposes, we use responses from the first two questions (first and second mention) and the Canadian probe question to categorise respondents into a set of ethnic groups that includes both people who report only one ethnic origin and those who report more than one origin. Sixteen categories of single responses can be drawn from the first mention. If the first response is Canadian, the answer(s) to the probe question is used instead, if any are provided. If a second response is provided, it is then combined into a series of multiple response categories. 3 Further detail as to the construction of these variables can be found in the Appendix. 4 For a study on the impact of age on social capital, see Robinson and Jackson (2001); for education, see Helliwell and Putnam (1999) and Nie, Junn, and Stehlik-Barry (1996). 5 The method of preparation of these variables for analysis is described in the Appendix. 9
Our review of other studies in this area prompts us to consider that ethnocultural origin alone is not sufficient to examine the impact of diversity on social capital. Ethnic identity can also ­ rightly or wrongly ­ be ascribed by others. Indeed, "the man in the street can see that men differ in physical appearance and he is certain that the differences are more than skin deep" (Simpson and Yinger, 1965). To account for the relative visibility of some ethnocultural groups, we aggregate separately Europeans, visible minority groups and Aboriginals in ways analogous to the employment equity categories. In order to maintain the cultural element of the compositional social capital debate, we further divide the European group to differentiate between majority whites (i.e. British, French, and Canadian) and non-majority whites (i.e. German, Italian, Ukrainian, etc). 3.5 Analysis. The data are explored in two stages. In the first stage, we present simple means comparisons of the five social capital measures alone, and according to selected indicators. From these results, we can identify basic patterns in the data and prepare the analysis for the second, more advanced, analytical stage. In the advanced stage, our hypotheses are tested according to the criteria discussed in the previous section. To test our hypotheses, we use a three-phase multiple regression model.6 In the base-line phase, standard demographic control variables and community-level population and diversity characteristics are examined. Added at the second phase are broad ethnocultural indicators: employment equity (e.g. visible minority) categories, immigrant status, recentness of arrival in Canada, household language and religion. In the third and final phase, we augment the model with the inclusion of specific ethnocultural ancestry markers. At each phase, we explore the impact of each independent variable on the respective dependent variables. We also identify changes to the relative explanatory power of each independent predictor due to the inclusion of increasingly precise control variables. If contextual-level arguments apply to Canadian communities, we should expect indicators of community diversity to bear significant negative weight on our dependent social capital variables. If individual-level arguments are valid, we expect personal ethnocultural attributes to assume positions of explanatory strength in our models. Finally, if neither contextual nor compositional indicators hold exclusive weight, we should expect standard controls such as age, education and income to play important roles in explaining variance in the five dependent variables. Although we do take the liberty of drawing conclusions as to the usefulness of independent variables both within and among the models, each regression is a separate analysis and thus must be considered independently. 4. Findings. 4.1 Comparisons of means. 6 Data are analysed using STATA's survey regression procedures. Using this program allows us to specify the characteristics of the sampling design (e.g. strata, clusters, and weights) and obtain unbiased point estimates and efficient standard errors. See StataCorp, 1999. 10
Table 1 presents the basic characteristics of the five social capital variables in the analysis, including the number of observations, proportions, weighted means, and Standard Deviations. Formal participation is measured dichotomously (i.e. a respondent is a member of an organization or s/he is not). In this sample, 73 percent have joined at least one organization. The variables for informal interaction are measured on seven-point scales. Overall, respondents see friends fairly often (5.23) and talk to their neighbours slightly less often (4.95). Respondents have mid-level trust in others (5.95 on a scale of 9), and low overall trust in government (3.28 on a scale of 7).
Table 1: Social Capital Measures
Formal Participation Interpersonal Trust Political Trust Seeing Friends Talking to Neighbours
N 4999 N 4738 4908 5117 5096
Proportion who join 73% Mean 5.95 3.28 5.23 4.95
Standard Deviation 1.85 1.47 1.33 1.69
Scale 1 to 9 1 to 7 1 to 7 1 to 7
What patterns emerge with respect to markers of diversity? Table 2 presents a comparison of means for each measure of social capital according to employment equity categories and immigrant status. Group-based differences can be observed for each component of social capital, with the exception of `seeing friends', for which differences are marginal. In general, majority whites and non-majority whites exhibit higher scores, while visible minorities and Aboriginal peoples exhibit lower scores. Visible minorities exhibit the lowest scores for `talking to neighbours' and have a lower propensity to join organizations. There are some clear exceptions. Visible minorities and immigrant visible minorities are the most trusting of government, and Canadian-born visible minorities socialize with friends most often.
Immigrant status plays some role, though a clear pattern is not easy to identify. Immigrant status reduces most scores as compared to Canadian-born respondents. The story changes, however, when the category is broken down into white and visible minority groups. The most striking group-based differences appear within the interpersonal trust variable. Here, visible minorities exhibit the lowest scores regardless of immigrant status. Interestingly, the immigrant factor also plays a role for whites, though in the opposite direction: this group exhibits the highest score for interpersonal trust. For visible minority groups, it is likely that recentness of arrival may explain part of the story.
Table 2: Comparisons of Means
Informal Social
Interaction
Formal Interpersonal Political
Seeing
Talking to
Participation
Trust
Trust
Friends Neighbours
All
0.73
5.95
3.28
5.23
4.95
11
Majority white
0.73
Non-majority white
0.75
Visible minority
0.67
Aboriginal
0.73
Immigrant
0.69
Immigrant white
0.68
Immigrant visible minority
0.69
Canadian-born
0.73
Canadian-born white
0.74
Canadian-born visible minority
0.62
6.02
3.25
5.25
5.03
6.01
3.20
5.26
4.97
5.48
3.66
5.21
4.42
5.72
3.11
5.34
4.85
5.88
3.54
5.11
4.49
6.16
3.43
5.13
5.00
5.50
3.76
5.13
4.55
5.97
3.22
5.26
4.98
6.01
3.23
5.27
5.03
5.46
3.38
5.46
3.97
These initial results suggest that differences do exist among individual-level indicators of ethnicity and with community-level diversity. We now turn to multiple regression modelling to uncover the impact of individual and community-level factors on these differences. If distinctions remain after repeated plausible controls are introduced, we can conclude that these variations are indeed the product of diversity-related factors.
4.2 basic model.
The basic model includes a group of individual-level demographic predictors and the respondents' community-level population and diversity characteristics. Results for five independent regressions are presented in Table 3. The leftmost column lists the independent variables included in the analysis. Two types of regression are used: for the formal participation model, a weighted survey logistic regression model is used to measure the propensity to join organizations. The other four models use weighted survey linear regression analyses. Each cell of the table shows the coefficient of the independent variable and its significance.
The five models vary in strength. Political trust is the weakest model, with only 2 percent of the variance of the dependent variable explained. Interpersonal trust is the strongest model, with 13 percent of the variance explained.
Contextual-level variables fare well in these models, of which and community size and index of diversity display the most consistent results. With the exception of political trust, as community size increases, scores on social capital measures decrease. This effect is most powerful for the interpersonal trust model and weakest for `talking to neighbours'. The effect appears to be counteracted by the index of diversity variable. In all cases, the negative impact of community size is balanced by the positive impact of the diversity of the municipality. The percentage of visible minorities in a respondent's CSD displays a consistently negative relationship among its coefficients, though this result is significant for formal participation rates, interpersonal trust, and seeing friends. Overall the picture is mixed; community size and the presence of visible minorities reduce social capital scores, and ethnocultural heterogeneity increases scores.
Do the demographic controls shed any light on these results? The answer is yes, but only within models and not across them. Our basic model for explaining political trust provides few answers. Growing older appears to reduce trust in government, and having a higher perceived income appears to increase it. Educational attainment has a strong positive impact on formal participation and interpersonal trust. These effects are particularly strong among those who have
12
earned post-secondary credentials. This may in part explain why students also appear as strong factors in the participation model, as well as those who believe that they earn higher incomes relative to others. Demographic variables such as age, sex, marital status and the presence of children in the home have no significant impact on an individual's propensity to join. Men are as likely to join as women, and people who are married are as likely to join as those who are divorced, separated, single or widowed. Interpersonal trust is seen to increase with age, but diminish among men. Our two models for informal social interaction show dissimilar trends. Age has a positive effect on talking to neighbours, but the frequency of seeing friends decreases with age. Being single has the opposite effect: single individuals talk with neighbours less, but socialize with friends more. Men are more likely to see friends than women, but are no more likely to talk with neighbours on a regular basis. Education, employment status and income do not play major roles, suggesting that sociability is not a product of socioeconomic circumstance. 4.3 Broad model: adding general ethnocultural attributes. The basic models suggest that social capital measures are driven by a combination of community-level diversity characteristics and basic demographic attributes. What impact do broad compositional attributes have on the model? The second phase explores the impact of broad ethnocultural characteristics on measures of social capital. The results can be found in Table 4. Perhaps the most striking change is the reduction in predictive power of the contextual level variables, save community size. What could account for this effect? In phased regression modelling, if the addition of control variables reduces the predictive power of existing variables, we can conclude that one of the new variables now explains that variance more effectively. If a coefficient does not shrink, it can be said that the added control variable has failed to rule out the original coefficient as an explanatory factor. The strength of each model is increased, with the most significant gain made for the model for interpersonal trust. Political trust remains the weakest model. In almost all cases, we observe a reduction in the size of the contextual-level coefficients and their loss of statistical significance. The key exception is the continued power of community size in predicting interpersonal trust, formal participation and seeing friends. The magnitude of the coefficients is reduced, but not sufficiently to affect their significant role in the model. In this case, the addition of ethnocultural markers does not impinge on the statistical impact of city living on interpersonal trust. The same cannot be said for markers of contextual diversity, all of which are statistically insignificant. The broad compositional controls must explain the change, since age, education, marital status and income all remain statistically significant, and their relative impacts do not change. With the exception of speaking French at home, no clear pattern among the ethnocultural coefficients can be noted across the models. Religious identification does not appear to play a strong role in interpersonal trust, but French speakers, Aboriginals, visible minorities and longtime visible minority immigrants all exhibit lower scores for this variable. The broad political trust model reveals the continued positive influence of ethnocultural attributes. In this model, community-level indicators are weakened by the positive effect of immigrant controls, in particular recent immigrants. Religion does play a role in propensity to join and talking to 13
neighbours. Being a Protestant exerts a positive influence on participation. Among the broad ethnocultural variables, being a visible minority, being a recent white immigrant, or speaking French as one's household language all exert strong negative influences. Interestingly, however, long time visible minority immigrants exhibit a strong positive propensity to join, and nonmajority whites are not as likely to join. The models for informal social interaction share some similarities. Sociodemographic predictors significant in the basic model continue to be so in the new model. Religion plays an important role in neighbourly interaction. Aside from Protestants, other respondents reporting a religious affiliation also report higher contact with neighbours. No similar effect is seen for the `seeing friends' model. As seen with interpersonal trust, speaking French as one's household language is a negative influence on both measures of informal social interaction. 4.4 Full model: adding specific ethnocultural attributes. Results from the broad models allow us to conclude that contextual measures of diversity do little to explain variance in the five social capital variables in the face of broad ethnocultural controls. These results favour compositional hypotheses over contextual ones by suggesting that broad group characteristics assume greater predictive strength in our five social capital models. Community size remains an important determinant in three models. What of individual-level ethnocultural attributes? The final phase of regression analysis allows us to examine the compositional argument with greater precision. In Table 5, the broad ethnocultural groups have been broken out into twenty-two distinct categories ­ sixteen single ethnicities, and six groups in which respondents claim more than one ethnic origin. Also added are immigrant status variables by recentness of arrival.7 If compositional hypotheses are valid, we should see significant differences between ethnic groups. It is apparent from the results that the addition of specific ethnocultural characteristics has little impact on the variables in the previous model. No patterns appear with respect to the detailed ethnic groups across the models. At the same time, contextual-level indicators of diversity remain insignificant, with the exception of the impact of `community size' on interpersonal trust, formal participation and seeing friends, which remain strong. No ethnic groupings have significant impacts within the `seeing friends' model. In general, groups do not appear to differ significantly from the British reference category. There are some exceptions: Chinese talk to their neighbours less and exhibit lower interpersonal trust. Latin Americans talk to their neighbours less, but trust government more. Blacks and South Asians also trust government more. eastern Europeans and Filipinos are less likely than the British to join 7 The former ethnic comparison group has been broken out into 3 single origins ­ British, Canadian and French and one multiple origin group made up of individuals claiming combinations of British, French or Canadian. In this table, the comparison ethnic group is people claiming only British origin. The European group has been split into 6 single origin groups (German, Ukrainian, Northern European, Eastern European, Southern European and Other European) and 1 group for people claiming combinations of European origins (European multiple). The visible minority category has also been split into 6 single origin groups (Black/Caribbean, Latin American, Chinese, Filipino, South Asian, and Other Asian) and one group for those claiming a visible minority origin in combination with another origin. The Aboriginal group has been split into those claiming only Aboriginal origins and those claiming Aboriginal origin(s) in combination with another origin. As well, there are two categories for people claiming a combination of European origins with majority origins (Majority with Northern European, and Majority with other European). 14
organizations. Ukrainians and "other Asians", a group comprised of, among others, those of Vietnamese, Korean, and Japanese ancestry, also extend less trust to others. Standard individual-level demographic attributes appear to withstand the addition of the new controls with ease. The contribution of the demographic variables in Table 5 closely mirrors that seen in Table 4: higher education and income positively affect participation and interpersonal trust, younger individuals trust government more and socialize more with friends but less with neighbours, and older individuals trust others more. Men and single individuals are more likely than women or married people to see their friends and talk to their neighbours, but are also less trusting of others. Broad, structural identifiers of diversity such as immigrant status, religion and language, remain important contributors to several models. These effects remain strong despite the presence of more precise markers of ethnicity. There is a clear immigration effect within the political trust model, with immigrants exhibiting a clear positive pattern. Not surprisingly, immigrants who have arrived in Canada within the last five years are significantly less likely to join an organization. French speakers continue to exert strong downward pressure across four of the models, while speaking a non-official language makes no impact whatsoever on any model. 5. Discussion. Our goal was to explore the relationship between diversity and social capital across a broad range of indicators. The research was motivated by findings that suggested that diversity exerts a negative influence on civic attitudes and behaviours. Our findings differ markedly from previous research, particularly those studies carried out in the United States. We find that in Canada, while differences based on ethnocultural identifiers do exist, they are the exception rather than the rule. If diversity were detrimental to social capital, we would expect to see negative effects across either immigrant or visible minority groups. By and large, we do not. There is no consistency across groups in any of our measures. An examination of bivariate means suggested slight differences across broad ethnic categories in such attributes as political and interpersonal trust and informal social interaction. However, controlling for individual and contextual characteristics using survey regression methods caused most of these differences to disappear. Indeed, in the case of trust in government, visible minorities and immigrant groups exhibit higher scores than the comparison groups. We also find that scores on social capital indicators tend to increase as community-level diversity increases but that such differences do not withstand the addition of individual-level ethnocultural attributes. We suggest that in Canada, where community size, diversity, wealth, and education are so closely and positively correlated, an urban lifestyle may be a more useful explanation for variance in civic attitudes and behaviours. Cities are places where myriad world-views, languages and cultures meet. Perhaps the dominant finding is the effect of community size on measures of social capital. In three of the five models ­ participation, interpersonal trust and seeing friends ­ the larger the city of residence, the less likely people are to participate, trust, and socialize. This may be accounted for by the effect of living in a city ­ constant presence of strangers, social networks based on individual interest more than local attachments (non-local networks). 15
Generalized trust in cities is reduced because familiarity is a more selective, network-based phenomenon. Mitigating the effect of city size are the individual characteristics of education and income. Higher levels of schooling and having a higher perceived income result in increased scores on measures of participation and interpersonal trust. From a political perspective, these findings are encouraging. Indicators of social capital are politically important only if we know what levers can be used to make a situation better, How do we increase levels of trust and membership? Based on our research, controlling diversity is neither justifiable nor realistic, but more importantly, does not appear to be the answer. Education and income appear to be far more effective levers for affecting social capital. 6. Appendix. This appendix describes the variables used in the analyses and their basic descriptives (see also Table A1). Variables are obtained from the 2000 Equality Security Community survey or from the 1996 Canadian Census linked through forward sortation postal codes and Census subdivisions. Most variables have been recoded, combined or reconstructed for the purposes of the analyses. 6.1 Social capital variables. Frequency of seeing friends: a seven-point scale variable derived from the question: "How often do you see close friends - not your husband or wife or partner or family member, but people you feel fairly close to?", ranges from 1 ­ `never' to 7 ­ `every day'. Frequency of talking to neighbours: a seven-point scale variable derived from the question: "And how often do you talk with your neighbours?", ranges from 1 ­ `never' to 7 ­ `every day'. Formal participation: dummy variable, membership=1, derived from the following questions: "How many service clubs, such as Lions or Meals on Wheels, do you belong to?", "How many recreational groups, such as sports leagues or clubs, music or hobby clubs, or exercise classes are you involved in?", "How many organisations active on political issues, such as the environment or taxpayers' rights, do you belong to?", "Sometimes people give time to various types of organisations. For instance, how many youthoriented groups, such as Girl Guides or Minor Hockey, have you given time to in the last 12 months?", "How about organisations providing cultural services to the public, such as a museum or music festival. How many of these have you given time to in the last 12 months?", "How about organisations that help people, such as the Cancer Society or a food bank? How many of these have you volunteered time to in the last 12 months?", and "Do you belong to or volunteer for any other groups or organisations that we have not asked about?". The reference category is 0 - `no reported memberships'. Interpersonal trust: a nine-point scale constructed from four questions: "If you lost a wallet or a purse that contained two hundred dollars, how likely is it to be returned with the money in it if it was found by (someone who lives close by/a clerk at the nearest grocery store/a police officer/a complete stranger); would you say very likely, somewhat likely or not at all likely?", ranges from 1 ­ `lowest trust' to 9 ­ `highest trust'. 16
Political trust: a seven-point scale constructed from two questions: "How much do you trust the government in Ottawa to do what is right? Do you trust it almost always, most of the time, only some of the time, or almost never?", and "How much do you trust the government in [province] to do what is right? Do you trust it almost always, most of the time, only some of the time, or almost never?", ranges from 1 ­ `lowest trust' to 7 ­ `highest trust'. 6.2 Community-level variables. Community size: the natural log of the population of the respondent's municipality of residence, as derived from the 1996 Canadian census. Index of diversity: the index of diversity = 1-sum of squares of the proportions of all the groups in the CSD (1996 Canadian census). The minimum value (for a CSD in which everyone shares the same ethnicity) is zero ­ ID = (1-12)=0. The maximum value varies depending on the number of groups represented. All majority ethnic groups were combined (British, French and Canadian) in order to avoid problems of interpretation in cases of CSDs reporting only British and Canadian ethnicity. Counts for twenty-five minority ethnic groups were included in order to calculate the index in each CMA. It should be noted that not all CMAs have people from each ethnic group. Percentage of visible minorities: the percentage of visible minorities in a respondent's municipality of residence, as derived from the 1996 Canadian census. 6.3 Individual-level variables. Education: dummy variables for `high school', `some technical school', `technical certificate', `some university', `bachelor's degree' and `MA/PhD'. Reference category is `less than high school'. Religion: dummy variables for `Catholic', `Protestant', `Evangelical Protestant', and `other religion'. Reference category is `no religion'. Household language: dummy variables for `household language ­ French', `household language ­ other'. Reference category is `household language ­ English'. Subjective Income: five-point scale derived from the following questions "If you had to guess, would you say your household income is above average, average, or below average?", "Would you say it is a little bit above average or a lot above average?" and "Would you say it is a little bit below average or a lot below average?" The resulting categories are 1 'a lot below', 2 - 'a little below', 3 - 'average', 4 - 'a little above', and 5 - 'a lot above' . This variable was used because over ј of respondents did not answer the reported income variable. Sex: dummy variable, =1 if the respondent is male. The reference category is female. Immigrant status and recentness of arrival in Canada: for immigrant respondents only, dummy variables for `immigrant who has arrived in the last 5 years', `immigrant who arrived between 5 and 9 years ago', `immigrant who arrived between 10 and 20 years ago', `immigrant who arrived more than 20 years ago'. The reference category is `Canadianborn'. Employment equity markers: dummy variables for `visible minority', `Aboriginal', and `nonmajority white'. Reference category is `majority white'. 17
Visible minority/immigrant status/recentness of arrival: dummy variables for `white immigrant who has arrived in the last 5 years', `white immigrant who arrived between 5 and 9 years ago', `white immigrant who arrived between 10 and 20 years ago', `white immigrant who arrived more than 20 years ago', `visible minority immigrant who arrived less than 5 years ago', `visible minority immigrant who arrived between 5 and 9 years ago', `visible minority immigrant who arrived between 10 and 20 years ago', and `visible minority immigrant who arrived more than 20 years ago'. Marital Status: dummy variables for `single', `divorced or separated', and `widowed'. The reference category is `married or living common law'. Employment status: dummy variables for `unemployed', `student', `homemaker', and `retired'. The reference category is `employed'. Children living at home: dummy variable =1 if respondent has at least one child living at home. The reference category is `no children living at home'. 7. References. Alesina, A., and E. La Ferrara. 2000. The Determinants of Trust. NBER Working Paper no. 7621. ______. 1999. "Participation in Heterogeneous Communities", The Quarterly Journal of Economics 115(3): 847-904. Almond, G.A. and S. Verba. 1963. The Civic Culture: Political Attitudes and Democracy in Five Nations. Princeton: Princeton University Press. Banfield, E.C. 1958. The Moral Basis of a Backward Society. Chicago: The Free Press. Black, J.H. 1982. "Immigrant Political Adaptation in Canada: Some Tentative Findings," Canadian Journal of Political Science XV:1 (March): 3-27. Black, J.H. 1987. "The Practice of Politics in Two Settings: Political Transferability Among Recent Immigrants to Canada," Canadian Journal of Political Science 22: 731-53. Brehm, J. and W. Rahn. 1997. "Individual-level Evidence for the Causes and Consequences of Social Capital," American Journal of Political Science 41(3): 999-1024. Coleman, J. S. 1990. Foundations of Social Theory. Cambridge, MA: The Belknap Press of Harvard University. Eisenberg, A. 2000. "Trust, Multicultural Politics and Political Dissent," Vancouver: UBC Centre for Research on Economic and Social Policy. Available online at http://www2.arts.ubc.ca/cresp/avigpap.pdf Edwards, B., Foley, M.W. and M. Diani. 2001. "Social Capital Reconsidered", Beyond Tocqueville: Civil Society and the Social Capital Debate in Comparative Perspective Hanover: University Press of New England, pp. 221-24. 18
Etzioni, Amitai. 2001. "Is Bowling Together Sociologically Lite?" Contemporary Sociology 30(3): 223-224. Fischer, C.S. 1982. To Dwell Among Friends: Personal Networks in Town and City. Chicago: University of Chicago Press. Foley, M., B. Edwards and M. Diani. 2001. "Social Capital Reconsidered," Beyond Tocqueville, ed. B. Edwards, M. W. Foley and M. Diani. University Press of New England, pp. 266280. Forbes, H.D. 1997. Ethnic Conflict: Commerce, Culture, and the Contact Hypothesis. New Haven: Yale University Press. Forrest, R. and Kearns, A. 2001. "Social Cohesion, Social Capital and the Neighbourhood," Urban Studies 38 (12): pp.2125-43. Frank, K.A. & Yasumoto, J. 1998. "Linking Action to Social Structure within a System: Social Capital Within and Between Subgroups," American Journal of Sociology 104(3): 642686. Fukuyama, F. 1995. Trust: The Social Virtues and the Creation of Prosperity. New York: Free Press. Germain, Annick. 2000. Immigrants and Cities: Does Neighbourhood Matter? Montreal: INRSUrbanisation. Guest A. and S.K. Wierzbicki. 1999. "Ties at the Neighbourhood Level: Two Decades of GSS evidence", Urban Affairs Review 35(1): 92-111. Glaeser, E.L., D. Laibson, J. A. Scheinkman and C.L. Soutter. 2000. "Measuring Trust," Quarterly Journal of Economics, 65 (August): 811-46. Granovetter, M. 1973. "The Strength of Weak Ties," American Journal of Sociology 78: 136080. Helliwell, J. and Robert Putnam. 1999. Education and Social Capital. Working Paper No. 7121. National Bureau of Economic Research, Cambridge, MA. Helliwell, J.F. and R. D. Putnam. 1995. economic growth and Social Capital in Italy, Eastern Economic Journal 21(3): 295-307. Helliwell, J.F. 2003. "Do Borders Matter for Social Capital? Economic Growth and Civic Culture in the U.S. States and Canadian Provinces", The Economic Implications of Social Cohesion. Toronto: University of Toronto Press, pp. 19-42. Inglehart, R. 1988. "The Renaissance of Political Culture," American Political Science Review 82: 1203-30. 19
Inglehart, R. 1990. Cultural Shift in Advanced Industrial Society. Princeton: Princeton University Press. Jackman, M. and M. Crane. 1986. "Some of My Best Friends Are Black...Interracial Friendship and White's Racial Attitudes," Public Opinion Quarterly 50: 459-486. Johnston, R. and S. Soroka. 2001. "Social Capital in a Multicultural Society: The Case of Canada," in Social Capital and Participation in Everyday Life, edited by P. Dekker and E. M. Uslaner. London: Routledge, pp. 30-44. Li, P. 1998. The Chinese in Canada. Toronto: Oxford University Press. Marschall, M and D. Stolle, 2001. "Race in the City: Neighbourhood Context and the Development of Generalized Trust". Paper presented at the APSA 2001. Article in progress and under review. Miller, D. 1995. On Nationality. Oxford: Clarendon. Muller, E. N. and Mitchell A. Seligson. 1994. "Civic Culture and Democracy: The Question of Causal Relationships," American Political Science Review 88: 635-53. Nie, N.H., J. Junn, and K. Stehlik-Barry. 1996. Education and Democratic Citizenship in America. Chicago: University of Chicago Press. Papillon, M. 2002. Immigration, Diversity and Social Inclusion in Canada's Cities. Canadian Policy Research Networks. Portes, A. and A. Stepick. 1993. City on the Edge: The Transformation of Miami. Berkeley: University of California Press. Portes, A. and P. Landolt. 1996. "Unsolved Mysteries: The Toqueville Files II ­ The Downside of Social Capital," The American Prospect 7(26). http://www.prospect.org/web/page.ww?name=Archives§ion=root.]. Putnam, R. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton: Princeton University Press. Putnam, R. 1995. "Bowling Alone: America's Declining Social Capital," Journal of Democracy 6:1 (January): 65-78. Rice, T. and J. Feldman. 1997. "Civic Culture and Democracy from Europe to America," Journal of Politics 59(4): 1143-1172. Robinson, R.V. and E.F. Jackson. 2001. "Is Trust in Others Declining in America? An AgePeriod-Cohort Analysis," Social Science Research 30: 117-45. 20
Saguaro Seminar. 2000. Social Capital Community Benchmark Survey: Executive Summary. Harvard: John F. Kennedy School of Government. http://www.kgs.harvard.edu/saguaro/communitysurvey/results.html Simpson G. and M. Yinger. 1965. "Types of Majority-Minority Situations" in Racial and Cultural Minorities: An Analysis of Prejudice and Discrimination. 4th Edition. New York: Harper Row. StataCorp. 1999. Stata Statistical Software: Release 6.0. College Station, Texas: Stata Corporation. Stein, R., S. S. Post, A. Rinden. 2000. "Reconciling Context and Contact Effects on Racial Attitudes," Political Research Quarterly 52 (June): 285-303. Uslaner, E. 2002. The Moral Foundations of Trust. Cambridge: Cambridge University Press. Varshney. A. 2002. Civic Life and Ethnic Conflict: Hindus and Muslims in India. New Haven: Yale University Press. Verba, S., K. Schlozman, and H. Brady. 1995. Voice and Equality: Civic Voluntarism in American Politics. Cambridge: MA: Harvard University Press. Wellman, B., (ed.). 1999. Networks in the Global Village: Life in Contemporary Communities. Boulder: Westview. Wilton, R.D. 1998. "The Constitution of Difference: Space and Psyche in Landscapes of Exclusion," Geoforum 29 (2) 173-85. Wuthnow, R. 1994. Sharing the Journey: Support Groups and America's New Quest for Community. Free Press. 21
Table 3: Basic Model
Formal Participation
Political Trust
Interpersonal Trust
N
4837
F
15.05 ***
r2
--
4651 11.94 *** 2%
4562 70.35 *** 13%
Contextual Variables
Community Size (ln) Index of Diversity % Visible Minorities
-0.09 *** 1.12 *** -0.73 *
0.02 -0.27 0.02
-0.25 *** 2.46 *** -0.81 **
Age
0.00
-0.01 **
0.02 ***
Sex (Female) Male
-0.08
-0.12
-0.27 ***
Subjective Income
0.17 ***
0.08 *
0.11 **
Children at home
-0.06
0.07
0.12
High school
0.60 ***
0.06
0.23
Individuallevel Variables
Education (less than High School)
Some technical Technical certificate Some university Bachelor's MA/PhD
0.44 * 0.59 *** 0.71 *** 0.98 *** 1.27 ***
-0.14 0.22 -0.02 0.21 0.22
0.27 * 0.29 * 0.31 * 0.52 *** 0.55 **
Employment Status (employed)
Unemployed Student Retired Homemaker
-0.11 0.56 * -0.19 -0.06
-0.06 0.17 0.09 0.24 *
-0.08 0.23 -0.08 0.00
Marital Status (married)
Single Divorced/Separated Widowed
-0.04 -0.05 -0.04
0.10 -0.01 0.28
-0.18 -0.29 * -0.43 **
Table entries are regression coefficients.
Equality Security Community Survey, first wave with metropolitan oversample, weighted.
Significance t and F tests: *p<0.05, **p<0.01, ***p<0.001.
Informal Social Interaction
Seeing Friends
Talking to Neighbours
4804 52.41 *** 8%
4780 17.03 *** 6%
-0.07 ***
-0.04 *
0.65 ***
0.27
-0.42 *
-0.26
-0.01 **
0.02 ***
0.17 **
0.08
0.04
-0.04
0.03
0.16
-0.04
0.12
0.05
-0.11
-0.09
0.18
0.02
0.27 *
-0.13
0.00
-0.09
0.03
-0.05
-0.06
0.16
0.08
0.00
0.12
0.20
0.28 *
0.59 ***
-0.34 ***
0.15
-0.23
0.17
-0.04
22
Table 4: Broad Model
Formal Participation
Political Trust
Interpersonal Trust
N
4837
4651
4562
F
11.42 ***
r2
--
21.41 *** 4%
65.58 *** 19%
Contextual Variables
Community Size (ln) Index of Diversity % Visible Minorities
-0.04 * 0.39 -0.33
0.00 -0.01 -0.17
-0.15 *** 0.20 -0.34
Age
0.00
-0.01 **
0.02 ***
Sex (Female) Male
-0.05
-0.12
-0.26 ***
Subjective Income
0.15 **
0.09 **
0.10 *
Children at home
-0.08
0.06
0.11
High School
0.64 ***
0.04
0.22 *
Education (less than high school)
Some technical Technical certificate Some university Bachelor's
0.49 * 0.61 *** 0.74 *** 1.10 ***
-0.14 0.17 -0.05 0.16
0.24 * 0.29 * 0.30 * 0.57 ***
MA/PhD
1.45 ***
0.16
0.65 ***
Employment Status (employed)
Unemployed Student Retired Homemaker
-0.10 0.72 ** -0.18 -0.07
-0.10 0.09 0.06 0.24 *
-0.15 0.24 -0.11 -0.07
Individuallevel Variables
Marital Status (married) Religion
Single Divorced/Separated Widowed Protestant Catholic
-0.04 -0.10 -0.05 0.27 * 0.10
0.12 0.01 0.28 * 0.16 0.25 **
-0.17 -0.31 ** -0.47 ** 0.20 * 0.06
(no religion) Evangelical Protestant
0.41 *
0.11
0.19
Other religion
0.49
0.02
0.11
Household Language
Language ­ French Language - Other
-0.45 * 0.07
0.10 -0.19
-1.46 *** -0.27
EE category (majority white)
Visible Minority Non-majority white Aboriginal
0.06 -1.01 ** 0.03
-0.08 0.26 -0.07
-0.19 ** -0.27 -0.39 *
VM Immigrant (<5)
-0.31
0.76 *
0.07
VM Immigrant (5-9)
0.44
0.47
0.08
Immigrants (Canadian- born)
VM Immigrant (10-20) VM Immigrant (20+) White Immigrant (<5) White Immigrant (5-9)
0.25 0.66 * -1.29 * -1.22 *
0.38 0.20 1.10 * 0.60
-0.26 -0.75 * 0.10 0.09
White Immigrant (10-20)
-0.25
0.74 **
0.05
White Immigrant (20+)
-0.28
0.18
-0.15
Table entries are regression coefficients.
Equality Security Community Survey, first wave with metropolitan oversample, weighted.
Significance t and F tests: *p<0.05, **p<0.01, ***p<0.001.
Reference category for household language is English.
Informal Social Interaction
Seeing
Talking to
Friends
Neighbours
4804 42.29 *** 10%
4780 30.74 *** 8%
-0.04 **
-0.01
-0.26
-0.32
-0.24
-0.01
-0.01 **
0.02 ***
0.18 **
0.12
0.03
-0.05
0.02
0.13
-0.03
0.13
0.06
-0.08
-0.09
0.19
0.02
0.28 *
-0.10
0.06
-0.05
0.13
-0.07
-0.05
0.17
0.10
0.01
0.12
0.19
0.26 *
0.61 ***
-0.33 ***
0.14
-0.24 *
0.17
-0.05
0.09
0.16
-0.04
0.34 ***
0.08
0.24 *
-0.01
0.35 *
-0.52 ***
-0.59 ***
0.00
0.16
-0.04
-0.10
-0.12
-0.71 **
0.03
-0.23
-0.19
0.07
0.25
0.11
-0.15
0.01
0.14
0.34
0.43 **
-0.27
-0.26
-0.18
-0.12
-0.34
-0.01
-0.14
23
Table 5: Full Model
Formal Participation
Political Trust
Interpersonal Trust
N
4837
F
14.95 ***
r2
--
4561 17.45 *** 5%
4652 63.37 *** 20%
Contextual Variables
Community Size (ln) Index of Diversity % Visible Minorities
-0.05 * 0.40 -0.40
0.00 -0.03 -0.15
-0.14 *** 0.16 -0.28
Age
0.00
-0.01 *
0.02 ***
Sex (Female) Male
-0.04
-0.12
-0.25 ***
Subjective Income
0.15 **
0.09 *
0.09 *
Children at home
-0.07
0.07
0.10
High school
0.64 ***
0.04
0.24 *
Education (less than High school)
Some technical Technical certificate Some university Bachelor's
0.49 * 0.63 *** 0.74 *** 1.10 ***
-0.13 0.18 -0.06 0.16
0.26 * 0.30 ** 0.31 * 0.59 ***
MA/PhD
1.43 ***
0.15
0.69 ***
Employment Status (employed)
Unemployed Student Retired Homemaker
-0.12 0.66 ** -0.21 -0.12
-0.11 0.11 0.06 0.23 *
-0.14 0.27 -0.10 -0.06
Marital Status (married)
Single Divorced/Separated Widowed
-0.05 -0.05 -0.03
0.12 0.00 0.28 *
-0.15 -0.32 ** -0.47 **
Protestant
0.28 *
0.16
0.20 *
Religion (no religion)
Catholic Evangelical Protestant
0.13 0.37 *
0.23 *
0.08
0.11
0.14
Other religion
0.63
-0.07
0.04
Individuallevel Variables
Household Language Immigrant Status (Canadianborn)
Language ­ French Language - Other < 5 years 5-9 years 10-20 years 20+ years
-0.46 * 0.10 -0.24 -0.75 * -0.21 -0.15
0.12 -0.27 0.62 ** 0.97 *** 0.48 * 0.14
-1.39 *** -0.22 0.25 0.27 -0.01 -0.28 *
Majority (multiple)
-0.02
-0.07
-0.09
French
0.01
-0.03
-0.15
Canadian
0.00
-0.11
-0.19
Northern European
-0.20
-0.10
-0.07
Maj. with Northern European
0.03
-0.13
-0.27
German
0.56
-0.10
0.25
Ukrainian
0.21
-0.09
-0.76 **
Southern European
0.19
0.19
-0.35
Eastern European
-0.85 **
0.07
-0.21
Group Status (British)
Other European Maj. with Other European European ­ Multiple
-0.10 0.32 0.58
0.04 -0.17 0.17
-0.25 -0.23 0.01
Black/Caribbean
-0.31
0.58 *
-0.32
Latin American
-0.15
0.91 **
-0.28
Chinese
-0.16
0.23
-0.80 **
South Asian
-0.57
0.54 *
-0.28
Filipino
-1.06 *
-0.02
-0.40
Other Asian
-0.94 *
-0.13
-0.95 **
Visible Minority and Other
-0.96 **
0.07
-0.03
Aboriginal
-0.29
0.20
-0.57
Aboriginal with Other
0.36
-0.30
-0.25
Table entries are regression coefficients.
Equality Security Community Survey, first wave with metropolitan oversample, weighted.
Significance t and F tests: *p<0.05, **p<0.01, ***p<0.001.
Reference category for household language is English.
Informal Social Interaction
Seeing Friends
Talking to Neighbours
4804 50.18 *** 10%
4780 30.93 *** 9%
-0.03 *
-0.01
-0.25 -0.20
-0.32 0.01
-0.01 **
0.02 ***
0.18 **
0.12
0.03
-0.04
0.02
0.14
-0.03
0.13
0.04
-0.09
-0.08
0.20 *
0.02
0.29 *
-0.11
0.06
-0.06
0.13
-0.08
-0.06
0.16
0.09
0.00
0.10
0.19
0.27 *
0.60 *** -0.34 ***
0.15
-0.21
0.15
-0.05
0.10
0.17
-0.03
0.34 ***
0.10
0.26 *
-0.04
0.16
-0.53 *** -0.57 ***
-0.03
0.28
0.10
-0.21
0.01
-0.20
-0.20
-0.34
-0.02
-0.10
-0.02
-0.12
0.02
-0.01
0.17
0.06
0.06
0.05
-0.02
0.04
-0.04
-0.36
-0.19
0.06
-0.18
-0.23
0.06
-0.36
-0.17
-0.07
0.25
0.30
0.10
0.11
-0.15
-0.42
0.16
-0.74 *
-0.09
-0.72 ***
0.01
-0.07
-0.16
-0.22
0.29
-0.40
-0.33
-0.65 *
0.23
-0.44
-0.10
0.01
24
Variable Community size (ln) Index of diversity Subjective income High school Some technical school Technical certificate Some university Bachelor's degree MA/PhD Age Male R has children at home Divorced/separated Single Widowed Household language ­ other Household language ­ French Retired Unemployed Student Homemaker Protestant Evangelical protestant Catholic Other religion Recent immigrant (<5 yrs) Recent immigrant (5-9 yrs) Immigrant (10-20 yrs) Immigrant (20+ yrs) Canadian French Northern European German Southern European Eastern European Ukrainian Other European Chinese Filipino Other Asian Black/Caribbean South Asian Latin Native British/French/Canadian British with Other British with other European Multiple European origin VM with other origin Aboriginal with other origin
Table A1: Descriptives N Mean SD 5142 10.90 2.31 5142 0.60 0.23 4933 2.96 0.94 5152 0.24 0.43 5152 0.07 0.25 5152 0.17 0.38 5152 0.08 0.27 5152 0.18 0.39 5152 0.07 0.25 5016 44.30 16.34 5152 0.46 0.50 5152 0.74 0.44 5152 0.12 0.32 5152 0.26 0.44 5152 0.06 0.24 5152 0.11 0.31 5152 0.18 0.38 5152 0.16 0.37 5152 0.08 0.27 5152 0.07 0.26 5152 0.05 0.22 5152 0.25 0.43 5152 0.08 0.27 5152 0.39 0.49 5152 0.06 0.23 5152 0.03 0.17 5152 0.03 0.17 5152 0.04 0.21 5152 0.10 0.30 5152 0.08 0.27 5152 0.15 0.35 5152 0.02 0.13 5152 0.04 0.19 5152 0.02 0.15 5152 0.02 0.14 5152 0.02 0.12 5152 0.02 0.13 5152 0.04 0.19 5152 0.01 0.11 5152 0.02 0.13 5152 0.02 0.13 5152 0.03 0.17 5152 0.01 0.09 5152 0.01 0.11 5152 0.04 0.20 5152 0.04 0.19 5152 0.01 0.12 5152 0.02 0.14 5152 0.01 0.12 5152 0.01 0.10
Min 4.17 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 18.00 0.00 0.00 0.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Max 13.81 0.89 5.00 1.00 1.00 1.00 1.00 1.00 1.00 95.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
25

A Aizlewood, R Pendakur

File: ethnicity-and-social-capital-in-canada.pdf
Title: The 1990s spawned a great deal of work defining social capital and identifying the personal and community characteristics that
Author: A Aizlewood, R Pendakur
Author: SM-AizlewoodA
Published: Thu Feb 8 15:13:06 2007
Pages: 25
File size: 0.25 Mb


ON THE LONGING FOR HOME, 15 pages, 0.06 Mb

JIM DENISON, 93 pages, 1.66 Mb

, pages, 0 Mb
Copyright © 2018 doc.uments.com