Social capital and core network ties: A validation study of individual-level social capital measures...

9
Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health Spencer Moore a,b,n , Ulf Bockenholt c , Mark Daniel b,d,e,f , Katherine Frohlich d , Yan Kestens b,d , Lucie Richard g,h a Queen’s University, School of Kinesiology and Health Studies, 69 Union St. PEC Rm. 215, Queen’s University, Kingston, ON, Canada K7L 3N6 b Centre de recherche du Centre Hospitalier de l’Universite´ de Montre´al, Canada c Desautels Faculty of Management, McGill University, Canada d De´partement de me ´decine sociale et pre´ventive, Universite´ de Montre´al, Canada e School of Health Sciences, University of South Australia, Australia f Department of Medicine, St Vincent’s Hospital, The University of Melbourne, Australia g Faculte´des sciences infirmi eres, Universite´ de Montre´al, Canada h Institut de recherche en sante´ publique, Universite´ de Montre´al, Canada article info Article history: Received 10 June 2010 Received in revised form 8 December 2010 Accepted 9 December 2010 Available online 16 December 2010 Keywords: Social capital Health Neighborhood environments Social networks Trust abstract Research on social capital and health has assumed that measures of trust, participation, and perceived cohesion capture aspects of people’s neighborhood social connections. This study uses data on the personal networks of 2707 Montreal adults in 300 different neighborhoods to examine the association of socio-demographic and social capital variables with the likelihood of having core ties, core neighborhood ties, and high self-rated health (SRH). Persons with higher household income were more likely to have core ties, but less likely to have core neighborhood ties. Persons with greater diversity in extra- neighborhood network capital were more likely to have core ties, and persons with greater diversity in intra-neighborhood network capital were more likely to have core neighborhood ties. Generalized trust, perceived neighborhood cohesion, and extra-neighborhood network diversity were shown associated with high SRH. Conventional measures of social capital may not capture network mechanisms. Findings suggest a critical appraisal of the mechanisms linking social capital and health, and the further delineation of network and psychosocial mechanisms in understanding these links. & 2010 Elsevier Ltd. All rights reserved. 1. Introduction Over the past decade, research on social capital and health has become an important sub-field within social epidemiological research. Yet, precise identification of the mechanisms by which social capital affects health has been limited. This limitation reflects in part a lack of knowledge of the validity of current measures of social capital. In health and place research, social capital is often defined as ‘‘network-accessed resources’’ but seen to represent a place-based feature (Moore et al., 2005). Individual proxy indicators of social capital, such as trust and participation, are often aggregated to the census-tract level to represent neighborhood social capital. Associations found between high neigh- borhood social capital and better health might be attributed in this regard to variations in the degree to which neighborhoods possess cohesive and resourceful social networks. Despite the reliance of health and place research on proxy indicators of social capital, no research as far as we are aware has examined whether these proxy indicators reflect aspects of people’s social connections within neighborhood settings using formal network data. Greater knowledge of the validity of current socio-relational indicators would assist in the identification of the mechanisms linking neighborhood social environments to health. Using population-based data on the personal social networks of 2707 randomly-selected adults residing in 300 different Montreal neighborhoods, the following study examines the association of (1) individual socio-economic and -demographic factors, such as socio-economic status (SES) and age, with characteristics of partici- pants’ social networks, (2) individual-level social capital variables with characteristics of participants’ networks, and (3) social capital variables with a person’s subjective health status. For this study, two social network characteristics are being treated as main outcomes. The first is whether individuals have people in their social networks with whom they can discuss important matters; the second is whether those who Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/healthplace Health & Place 1353-8292/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.healthplace.2010.12.010 n Corresponding author at: Queen’s University, School of Kinesiology and Health Studies, 28 Division St., Queen’s University, Kingston, ON, Canada K7L 3N6. Tel.: + 1 613 533 6000x78667; fax: + 1 613 533 2009. E-mail address: [email protected] (S. Moore). Health & Place 17 (2011) 536–544

Transcript of Social capital and core network ties: A validation study of individual-level social capital measures...

Page 1: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

Health & Place 17 (2011) 536–544

Contents lists available at ScienceDirect

Health & Place

1353-82

doi:10.1

n Corr

Studies,

Tel.: +1

E-m

journal homepage: www.elsevier.com/locate/healthplace

Social capital and core network ties: A validation study of individual-levelsocial capital measures and their association with extra- andintra-neighborhood ties, and self-rated health

Spencer Moore a,b,n, Ulf Bockenholt c, Mark Daniel b,d,e,f, Katherine Frohlich d,Yan Kestens b,d, Lucie Richard g,h

a Queen’s University, School of Kinesiology and Health Studies, 69 Union St. PEC Rm. 215, Queen’s University, Kingston, ON, Canada K7L 3N6b Centre de recherche du Centre Hospitalier de l’Universite de Montreal, Canadac Desautels Faculty of Management, McGill University, Canadad Departement de medecine sociale et preventive, Universite de Montreal, Canadae School of Health Sciences, University of South Australia, Australiaf Department of Medicine, St Vincent’s Hospital, The University of Melbourne, Australiag Faculte des sciences infirmi�eres, Universite de Montreal, Canadah Institut de recherche en sante publique, Universite de Montreal, Canada

a r t i c l e i n f o

Article history:

Received 10 June 2010

Received in revised form

8 December 2010

Accepted 9 December 2010Available online 16 December 2010

Keywords:

Social capital

Health

Neighborhood environments

Social networks

Trust

92/$ - see front matter & 2010 Elsevier Ltd. A

016/j.healthplace.2010.12.010

esponding author at: Queen’s University, Sch

28 Division St., Queen’s University, Kings

613 533 6000x78667; fax: +1 613 533 2009.

ail address: [email protected] (S. Moore).

a b s t r a c t

Research on social capital and health has assumed that measures of trust, participation, and perceived

cohesion capture aspects of people’s neighborhood social connections. This study uses data on the

personal networks of 2707 Montreal adults in 300 different neighborhoods to examine the association of

socio-demographic and social capital variables with the likelihood of having core ties, core neighborhood

ties, and high self-rated health (SRH). Persons with higher household income were more likely to have

core ties, but less likely to have core neighborhood ties. Persons with greater diversity in extra-

neighborhood network capital were more likely to have core ties, and persons with greater diversity in

intra-neighborhood network capital were more likely to have core neighborhood ties. Generalized trust,

perceived neighborhood cohesion, and extra-neighborhood network diversity were shown associated

with high SRH. Conventional measures of social capital may not capture network mechanisms. Findings

suggest a critical appraisal of the mechanisms linking social capital and health, and the further delineation

of network and psychosocial mechanisms in understanding these links.

& 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Over the past decade, research on social capital and health hasbecome an important sub-field within social epidemiological research.Yet, precise identification of the mechanisms by which social capitalaffects health has been limited. This limitation reflects in part a lack ofknowledge of the validity of current measures of social capital. In healthand place research, social capital is often defined as ‘‘network-accessedresources’’ but seen to represent a place-based feature (Moore et al.,2005). Individual proxy indicators of social capital, such as trust andparticipation, are often aggregated to the census-tract level to representneighborhood social capital. Associations found between high neigh-borhood social capital and better health might be attributed in thisregard to variations in the degree to which neighborhoods possess

ll rights reserved.

ool of Kinesiology and Health

ton, ON, Canada K7L 3N6.

cohesive and resourceful social networks. Despite the reliance of healthand place research on proxy indicators of social capital, no research asfar as we are aware has examined whether these proxy indicatorsreflect aspects of people’s social connections within neighborhoodsettings using formal network data. Greater knowledge of the validityof current socio-relational indicators would assist in the identificationof the mechanisms linking neighborhood social environments tohealth.

Using population-based data on the personal social networks of2707 randomly-selected adults residing in 300 different Montrealneighborhoods, the following study examines the association of(1) individual socio-economic and -demographic factors, such associo-economic status (SES) and age, with characteristics of partici-pants’ social networks, (2) individual-level social capital variables withcharacteristics of participants’ networks, and (3) social capital variableswith a person’s subjective health status. For this study, two socialnetwork characteristics are being treated as main outcomes. The first iswhether individuals have people in their social networks with whomthey can discuss important matters; the second is whether those who

Page 2: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

S. Moore et al. / Health & Place 17 (2011) 536–544 537

do have those persons in their neighborhoods. Discussant networkshave been shown to overlap with people’s core social relationships(McPherson et al., 2006). Core relationships tend to represent strongerties, and be small and comparatively dense and homogenous comparedto other relationships (Marsden, 1987). Core networks have impor-tance for health since they often exert normative pressures on behavior,and individuals can often use such networks for advice and emotionaland instrumental support (McPherson et al., 2006). Having or nothaving core network members should not however be conflated withhaving or not having social support. Having core networks representsthe possession of a key social structural feature from which support orother resources might emerge.

1.1. Individual social capital and self-rated health (SRH)

Social capital is often defined as the resources that individuals andpotentially groups have access to through their networks (Bourdieu,1986). The definitional simplicity belies the theoretical and measure-ment debates surrounding social capital’s use in health research.Theoretically, debate has often centered on the relative merits of morecommunitarian versus network and political economy approaches(Moore et al., 2005). Communitarian approaches tend to frame socialcapital as the property of places, e.g., neighborhoods, and thus a publicgood equally available to all; network approaches tend to frame socialcapital as an emergent, inter-personal dimension of social relation-ships. This theoretical debate has often converged with measurementissues. Within the communitarian approach, studies have tended toaggregate proxy indicators of social capital, such as trust, to theneighborhood level to assess the contextual effect of social capitalon health (Subramanian et al., 2003). More recently, individual-levelstudies of social capital have used indicators such as trust, participation,and perceived cohesion to examine the association of individualsocial capital with a range of health status or health behavior outcomes(Kim et al., 2008; Poortinga, 2006a).

In this study, the broad concept of social capital, which encompassessocio-relational indicators such as trust, participation, perceived cohe-sion, and network social capital, will be eschewed in favor ofterminology that reflects the actual measures used. Trust is differ-entiated into two types: (1) generalized trust and (2) particularized orlocalized trust (Abbott and Freeth, 2008). Generalized trust involvesindividual perceptions of how trustworthy overall the social environ-ment may be, whereas particularized trust, in contrast, asks about trustin specific others (e.g., neighbors) (Abbott and Freeth, 2008). Extensiveresearch has shown associations at the individual level betweengeneralized trust and SRH (Helliwell and Putnam, 2004; Kim andKawachi, 2006; Kim et al., 2008; Pollack and Knesebeck, 2004;Poortinga, 2006a, 2006b; Subramanian et al., 2002; Veenstra et al.,2000, 2005). Less research has been devoted to particularized trust,such as ‘‘trust in neighbors,’’ although this construct has been shownassociated with depression (Fujiwara and Kawachi, 2008).

Individual social participation has also been shown associatedwith SRH (Helliwell and Putnam, 2004; Kim et al., 2006; Poortinga,2006a; Lindstrom et al., 2004; Hyppa and Maki, 2003). Recentresearch has suggested that the form of participation might varyaccording to neighborhood characteristics (Swaroop and Morenoff,2006). Yet, little research has examined whether it matters forindividual health if participation takes place inside or outside aperson’s neighborhood. If participation facilitates the developmentof social networks, individuals who participate closer to home maylikely have more neighborhood ties. Indices of perceived socialcohesion and informal social control, i.e., collective efficacy, havealso been used to measure the neighborhood social environment.Although these indicators tend to be modeled as a neighborhood-level construct, research has shown individual perceptions of theneighborhood environment and informal social control associated

with SRH (Weden et al., 2008; Moore et al., 2010). Given theamount of research on trust, participation and SRH, SRH provides asa well-documented outcome for assessing the validity of currentsocio-relational and -capital indicators.

Network approaches using formal network measures of socialcapital have only recently surfaced within the health literatureand have focused primarily on individual-level relationships(Moore et al., 2010). These studies have demonstrated an associa-tion between network-accessed resources and obesity and over-weight status and mastery (Moore et al., 2009a, 2009b). Althoughproxy indicators of social capital are assumed to reflect in somesense these individual network connections, little is known in thisregard. Even less known is the degree to which the differentdimensions of network social capital, or whether extra- or intra-neighborhood network social capital is more strongly associatedwith SRH.

1.2. On the theoretical importance of neighborhood attachment

Recent research on neighborhood social environments and healthhas identified neighborhood attachment as central to understandingthe relationship between neighborhood social environments andhealth (Carpiano, 2006, 2007). Neighborhood attachment refers toan individual’s degree of integration into neighborhood networks;those better connected are more likely to access local resources, andexperience either the positive or negative consequences of theneighborhood social environment (Carpiano, 2006). For example, interms of the potential negative consequences, Caughy et al. (2003)showed in impoverished Baltimore neighborhoods that children whoseparents had few neighborhood social connections had lower levels ofbehavioral problems than the children of parents who had more socialconnections.

Previous studies using the concept of neighborhood attachmenthave applied global measures. Global measures ask respondents foroverall assessments of their sets of friends, neighbors, or otherrelationships, e.g., how close one feels to the friendliest neighborone knows. Global measures are relatively efficient, require lessinterview time, and have often shown strong associations with health(Marsden, 2006); yet, such questions place a high cognitive demand onrespondents in defining terms such as ‘‘close’’ and ‘‘friends’’ (Marsden,2006). In addition, it is more difficult to distinguish the differentmechanisms linking social networks with health using global com-pared to network measures (Marsden, 2006). Another advantage offormal network methods in this case is that it allows identification of aperson’s core network members first and only after is their place ofresidence discovered. For this study, having core ties in the neighbor-hood may be seen to reflect a person’s degree of integration intoneighborhood networks.

1.3. Research questions

To decipher more clearly the association of trust, participation,and the different dimensions of network social capital with having(1) core network ties, (2) core neighborhood network ties, and(3) high SRH, this analysis will distinguish the socio-relational and -capital indicators according to whether they occur inside or outsidethe neighborhood setting. The study anticipates that trust inneighbors, neighborhood social participation, and neighborhoodsocial capital will be more strongly associated with having coreneighborhood social ties than if those socio-relational factors occurfarther away from one’s neighborhood. Three sets of questionsguide this study:

Q1.

To what degree are socio-economic and -demographic, andsocial relational and capital factors (i.e., trust, participation,
Page 3: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

S. Moore et al. / Health & Place 17 (2011) 536–544538

perceived cohesion, and the different dimensions of networkcapital) associated with the likelihood of having core ties?

Q2.

To what degree are socio-economic and -demographic, andsocial relational and capital factors associated with the like-lihood of having core neighborhood ties?

Q3.

To what degree are social relational and capital factors, particularlythe different dimensions of network capital, associated with thelikelihood of reporting a higher level of SRH?

As a test of the validity of current social capital indicators inmeasuring social connections, knowledge of the individualcorrelates of having core neighborhood ties is important for severalreasons. First, if having core neighborhood ties is associated withsocio-economic or -demographic factors in meaningful ways, itmay suggest the potential unequal distribution of an effectmodifier, i.e., neighborhood attachment, across neighborhoods.For example, if persons with low SES tend to have core neighbor-hood ties and low SES persons tend to cluster within particularneighborhoods, neighborhood attachment may play a greater rolefor low SES persons in certain neighborhoods. Second, if having coreneighborhood ties is associated with proxy indicators of socialcapital, such as trust and participation, it would help validatecurrent measures of social capital as measuring neighborhoodsocial connectivity. Finally, examining the degree to whichtrust, participation, and the different dimensions of social capitalis associated with self-reported health will assist in the identi-fication of the specific mechanisms by which social capital or socialrelations influence health. The parsing of social capital intoinside and outside neighborhood ties enables greater attentionto the relative importance of neighborhood social connections onself-rated health.

2. Methods

2.1. Sample

Data came from the 2008 Montreal Neighborhood Networksand Healthy Aging Study (MoNNETs-HA). The MoNNETs-HA studyused a two-stage stratified cluster sampling design. In stage one,Montreal Metropolitan Area (MMA) census tracts (N¼862) werestratified using 2001 Canada Census data into tertiles of high,medium, and low household income. One hundred census tractswere selected from each tertile (nj¼300). In stage two, potentialrespondents within each tract were stratified into three age groups:25–44 years old, 45–64, and 65 or older. Three respondents wererandomly selected within each age stratum and CT for a total of9 respondents per tract, except for seven tracts in which fourparticipants were selected (ni¼2707). To be selected, individualshad to (1) be non-institutionalized, (2) have resided at their currentaddress for at least one year, and (3) able to complete thequestionnaire in French or English. Random digit dialing of listedtelephone numbers was used to select households and a computerassisted telephone interviewing system guided questionnaireadministration. Participants completed the telephone interviewbetween mid-June and early August 2008.

2.2. Measures

2.2.1. Outcomes

2.2.1.1. Social ties. To measure core ties, we used the ‘‘discussimportant matters’’ name generator. The name generator questionasked participants to name up to three individuals, i.e., alters, withwhom they may have discussed important matters in the last sixmonths. This question elicits an individual’s close set of confidants(McPherson et al., 2006). To reduce potential measurement error

due to variation among respondents in their reporting of personalnetworks (Brewer, 2000; Feld and Carter, 2002), participants wererestricted to identifying no more than three alters. If participantsreported not having discussed important matters with anyone inthe last six months, interviewers repeated the question to parti-cipants, and asked whether they had not spoken with anyone orpreferred not to answer the question. Participants who said thatthey preferred not to answer this question were dropped fromthese analyses (n¼72). Among those who answered the namegenerator question and had no missing responses to covariates(n¼2556), the first set of analyses examined the likelihood of aparticipant having ‘‘one or more ties’’ versus ‘‘no ties.’’

2.2.1.2. Neighborhood social ties. Participants who nominated oneor more alters in the preceding name generator were administeredthe name interpreter. The name interpreter consists of a series ofquestions about the nominated alters, e.g., their age. One questionasked participants if their alters resided in their (i) household,(ii) neighborhood, (iii) in the MMA, or (iv) outside the MMA. Nospecific definition of neighborhood was provided to respondents,thereby allowing participants to conceptualize their own neigh-borhood boundaries. Among those participants who had at leastone core tie and no missing responses from model covariates, asecond set of analyses examined the likelihood of ‘‘having one ormore neighborhood ties’’ versus ‘‘no neighborhood ties’’ (n¼2268).Household ties were distinguished from neighborhood ties.

2.2.1.3. Self-reported health. To assess self-reported health status,participants were asked if they, generally speaking, would say thattheir current health was excellent, very good, good, fair, or poor. Tocompare current findings with other research on social capital andself-reported health, we dichotomized participant responses intohigh (excellent and very good) and low (good, fair, and poor)categories, and modeled the likelihood of high self-reported health.To facilitate comparison across the analyses, we used the samesample in the SRH analysis as that used in the ‘‘core neighborhoodties’’ analysis.

2.2.2. Main exposure variables

The following socio-relational variables were used: (1) outsideneighborhood social capital, (2) neighborhood social capital,(3) generalized trust, (4) particularized trust, i.e., trust in neighbors,(5) inside neighborhood social participation, (6) outside neighborhoodsocial participation, and (7) perceived neighborhood cohesion.

Outside- and inside-neighborhood network capital were assessedusing a position generator. Position generators have been recentlyapplied in public health research (Moore et al., 2009a, 2009b).Position generators assess social capital by asking participants toindicate whether they know someone (on a first-name basis) whoholds certain occupations in society, e.g., teacher or taxi driver. Bylinking these occupations to a context-relevant prestige score, thediversity and potential value of a person’s social connections can beassessed (Lin, 2001). Ten occupations were selected from a listing of90 occupations that had been ranked according to gender-neutraljob prestige scores within Canada (Goyder et al., 2003). The list wasdivided into octiles ranging from high to low prestige jobs. Fromeach octile, one occupation was randomly selected; two additionaloccupations (i.e., physician and musician/artist) were selected.These ten occupations were randomly listed in the positiongenerator. If a participant indicated that they knew someone inany of the ten occupations, they were asked if that person resided intheir household, neighborhood, outside their neighborhood but inthe MMA, or outside the MMA (relationship location). If a respon-dent reported that they knew more than one person in an

Page 4: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

S. Moore et al. / Health & Place 17 (2011) 536–544 539

occupation, they were asked to consider the person who wasclosest emotionally to them.

Lin (2001) argues that the position generator captures threeseparate dimensions of social capital: (1) upper reachability, (2)diversity, and (3) range. Upper reachability, which is the highestprestige occupation a person knows, represents the uppermostresource a person can reach through their social ties; diversity, whichrefers to the number of different occupations accessed, reflects, in thiscase, a person’s network size; and, range, which is the differencebetween the highest and lowest prestige job accessed, reflects thedifferent types of resources a person might access (Lin, 2001). To aid inthe identification of how these different dimensions are associatedwith the main outcomes, we elected to assess their associationsseparately rather than bundling them into a single measure. Giventhat the dimensions were at different scales, the three were standar-dized to facilitate their comparison among each other. The odds ratio ofeach standardized network social capital dimension is the amount theodds of the outcome occurring increases (decreases) when thatdimension increases (decreases) by one standard deviation. Correla-tions among the dimensions were not found to pose multicollinearityproblems. Ancillary analyses showed that the pattern of results formodels with a variable representing a single socio-relational or net-work-capital dimension, or different subsets of variables were similarto the complete model.

Generalized trust in others was assessed using the U.S. General SocialSurvey question ‘‘Generally speaking, would you say that most peoplecan be trusted or that you can’t be too careful in dealing with people?’’Participants chose from the following responses: (1) most people canbe trusted, (2) can’t be too careful, (3) depends, (4) most people cannotbe trusted, and (5) don’t know. ‘‘Don’t know’’ responses were treated asmissing (n¼20). Responses were reverse coded so that higher numbersindicated greater trust. To maintain the full content of the question, thisvariable was modeled in ordinal form.

Trust in neighbors was assessed with the single item – ‘‘people inyour neighborhood can be trusted’’ – with five-point Likert responsescale ranging from strongly agree to strongly disagree. Responses werereverse coded so that higher numbers indicated greater trust. Thevariable was modeled in ordinal form with ‘‘don’t know’’ responsesforming the neutral category (n¼124). Sensitivity analyses wereconducted to ensure that this decision did not impact findings.

Inside neighborhood participation was assessed by asking parti-cipants to indicate whether or not they had been active in the lastfive years as a volunteer or officer in a neighborhood group orassociation. Outside neighborhood participation was assessed byasking participants if they had been active in the last five years as avolunteer or officer in a group or association outside the neighbor-hood. Both were modeled in dichotomous form.

Perceived neighborhood social cohesion was assessed with fouritems: (1) you have trouble with your neighbors, (2) people in yourneighborhood are willing to help each other, (3) most people inyour neighborhood know you, and (4) your neighborhood is clean.Responses were on a five-point Likert scale from strongly agree tostrongly disagree; ‘‘don’t know’’ responses formed the neutralcategory. Responses were reverse coded with the exception of itemone, and centered on the neutral category so that higher numbersindicated a greater perceived cohesion. The cohesion scale had aCronbach’s alpha of 0.41. Sensitivity analyses, in which theperceived environment items were entered separately or a per-ceived cohesion factor score was used, held similar results howeveras those reported using the scale.

2.2.3. Study covariates

Covariates included gender, income category, educational attain-ment, employment status, age, marital status, primary householdlanguage, birthplace, and length of residence at current address.

Participants self-identified gender. Participants selected their incomefrom five categories: (a) less than $28,000, (b) $28,000–$49,000,(c) $50,000–$74,000, (d) $75,000–$100,000, and (e) more than$100,000. Missing responses to the income question were imputedusing ordinal regression for 20% of the respondents using (a)questionnaire data on socio-demographic variables, including edu-cation, age, and employment status, and (b) Canada census data onmedian household income for the census tract in which respondentsresided. Study participants were asked to select their highest level ofeducational attainment from seven categories: (a) no high schooldegree or certificate, (b) high school diploma or equivalent, (c) tradecertificate or diploma, (d) college certificate or diploma belowBachelor’s degree, (e) Bachelor’s degree, (f) Master’s degree, or (g)earned doctorate, medical, or professional degree. For employmentstatus, participants indicated if they were currently employed.

Participants’ ages were grouped into six categories: (1) 25–34,(2) 35–44, (3) 45–54, (4) 55–64, 65–74, and (6) 75 years or more,with the oldest group used as the reference. Participants wereasked to indicate their marital status as: (1) married or in acommon-law relationship, (2) single, (3) separated, (4) divorced,or (5) widowed. Those who were married or in a common-lawrelationship were contrasted with those who were in one of theother categories. Participants were asked to indicate the primary

language spoken in their homes. French-speaking households werecontrasted with English or other language households. Foreign-born status was based on whether participants reported being borninside or outside of Canada. Residential duration, which was theamount of time in years and months that a person had resided attheir current residence, had a skewed distribution and wasmodeled in continuous terms after logarithmic transformation.

2.3. Statistical analysis procedures

The MoNNET-HA response rate was calculated according toAmerican Association for Public Opinion Research standard defini-tions (AAPOR, 2008). The response rate was calculated as thenumber of completed interviews divided by the number of inter-views, non-interviews, and an estimated proportion of cases ofunknown eligibility. The MoNNET-HA study had a response rate of38.7%. To assess the representativeness of the MoNNET-HA sample,chi-square analyses were used to compare on a CT-by-CT basis theobserved sample counts to the expected counts based on the mostrecent 2006 Canada census. These comparisons were made for arange of census variables, including percentage adults 65 years andolder and average length of residence. Minimum cell counts werevalidated by comparing the results with those from exact binomialtests. Results from these analyses were summarized across the 300CTs and compared to the null hypothesis of no difference betweenthe sample and the 2006 census. Results of these analyses showedthat the MoNNET-HA sample over-represented (1) older adults (bysampling design), (2) individuals with an income less than 50,000per year, (3) persons who lived in their places of residence for morethan five years, (4) females, and (5) those with more than a highschool degree.

Multilevel logistic regression analysis was used to account forthe clustered sampling design. CTs were specified as a randomeffect. No area-level measures were included in this study.Analyses proceeded in several stages. First, the variance betweenneighborhoods (i.e., census tracts) in having at least one core tie,having at least one neighborhood tie, and having high SRH wascalculated, and reported as the intraclass correlation coefficient(ICC) along with a 95% plausible value interval (PVI). The ICC was (i)0.00 in having core ties, (ii) 0.02 (95% PVI: 0.02–0.03) in having atleast one core neighborhood tie, and (iii) 0.04 (95% PVI: 0.00–0.08)in high SRH. Three statistical analyses were conducted: (i) having at

Page 5: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

Table 1Characteristics of Montreal Neighborhood Networks and Healthy Aging Study

(MoNNETs-HA), social ties sample, 2008.

Variables Core ties analysis(n¼2556)

Core neighborhood tiesanalysis (n¼2268)

Social capital measures

S. Moore et al. / Health & Place 17 (2011) 536–544540

least one core network tie versus none, (ii) having at least one coreneighborhood network ties versus none, and (iii) reporting highversus low SRH. For each analysis, the first model examined theassociation between the socio-demographic and -economic studycovariates and the outcome. The second model introduced thesocio-relational variables and the separate social capital dimen-sions into model one.

Generalized trust 3.3 3.3

Neighborhood trust 1.1 1.1

Outside neighborhood

social participation

0.23 0.25

Neighborhood social

participation

0.24 0.25

Perceived neighborhood

environment

0.77 0.77

Outside neighborhood

network social capital

0.01 0.10

Neighborhood network

social capital

0.00 0.05

IncomeLess than $28,000 20.4% 18.7%

$28,000–$49,000 28.2% 28.8%

$50,000–$74,000 26.8% 25.7%

$75,000–$100,000 12.9% 11.3%

More than $100,000 11.7% 15.5%

GenderFemale 64.7% 64.9%

Male 35.3% 35.1%

EducationLess than a high school

degree

11.6% 9.2%

High school degree or

trade certificate

29.4% 27.9%

College certificate 20.7% 21.6%

Bachelors degree and

higher

38.3% 41.3%

Age group25–34 years old 14.8% 16.1%

35–44 years old 17.9% 19.6%

45–54 years old 20.4% 21.0%

55–64 years old 16.2% 16.6%

65–74 years 20.7% 19.0%

or 75 years and more 9.9% 7.7%

Marital statusMarried/common-law

relationship

54.4% 56.1%

Single 20.2% 20.7%

Divorced/separated 14.8% 14.6%

Widowed 10.1% 8.6%

Household languageFrench (bilingual) 78.1% 78.2%

English 13.7% 13.9%

Other 8.2% 7.9%

Foreign born status 19.5% 19.5%

Residential duration 13.9 years 13.3 years

Social connection outcomesNo core ties 11.1% –

No core neighborhood ties – 51.5%

3. Results

Table 1 provides information on the socio-demographic character-istics of the MoNNET-HA sample. Table 2 provides information on thenetwork capital characteristics of MoNNET-HA participants.

3.1. Analysis one: having a core tie

Out of the MoNNET-HA sample of 2707 respondents, 2635participants answered the name generator question. From thissample, an additional 79 observations, or 3.0%, were dropped fromanalysis one for missing data on any of the socio-demographicvariables. Among the 2556 participants, 11.1% reported having nocore ties. Table 3 provides the results of analysis one: women (OR:1.52; 95%CI: 1.14–2.04), French-speaking households (OR: 1.53;95%CI: 1.06–2.22), persons with household income levels greaterthan $75,000 or educational attainment of a bachelor degree orhigher, and younger age groups were more likely to have a corenetwork tie. Respondents who had participated in associationsoutside the neighborhood in the last five years (OR: 1.93; 95%CI:1.21–3.07) were more likely to have core network ties. In terms ofthe different dimensions of network social capital, persons with amore diverse network of contacts outside the neighborhood (OR:1.77; 95%CI: 1.29–2.44), and those who had a greater reach of tiesinside the neighborhood (OR: 1.43; 95%CI: 1.08–1.90) were morelikely to have at least one core tie.

3.2. Analysis two: having a core neighborhood tie

From the sample of 2332 participants who had at least one coretie, an additional 64 observations, or 2.1%, were dropped in analysistwo for missing data on any of the socio-demographic or healthvariables. Among this sample of participants (n¼2268), 5979 alterswere named: 28.8% of the participants had at least one alter in theirhousehold, 48.5% had at least one alter in their neighborhood, 59.6%had at least one in the MMA but outside their neighborhood, and40.0% had at least one outside the MMA. Out of this sample, 10.6%had only neighborhood ties. Table 4 provides the results of analysistwo: higher income participants were less likely to have coreneighborhood ties compared to the lowest income group, whereasparticipants with higher levels of trust in their neighbors (OR: 1.09;95%CI: 1.00–1.18) and greater diversity in neighborhood socialcapital (OR: 1.98; 95%CI: 1.57–2.51) were more likely to have a coreneighborhood tie. A greater range of network social capital outsidethe neighborhood decreased the likelihood of having a coreneighborhood tie (OR: 0.79; 95%CI: 0.66–0.95).

3.3. Analysis three: having High self-reported health (SRH)

Socioeconomic variables (i.e., household income, educationalattainment, employment status) were associated with high SRH inexpected directions. Participants born outside of Canada were lesslikely to report high SRH (OR: 0.62; 95%CI: 0.48–0.79). Participants withhigh generalized trust (OR: 1.18; 95%CI: 1.04–1.33), high neighborhoodsocial participation (OR: 1.28; 95%CI: 1.02–1.60), and a more favorableperception of the neighborhood environment (OR: 1.24; 95%CI:1.08–1.42) were more likely to report high SRH. In terms of the

network social capital, greater diversity in outside network socialcapital was associated with high SRH (OR: 1.20; 95%CI: 1.01–1.42).

4. Discussion

To assess the validity of current socio-relational and -capitalmeasures used in social epidemiological research, this study addressedthree sets of questions related to the association of (i) individual-levelsocial capital with having social ties, (ii) having those ties inside theneighborhood, and (iii) reporting high SRH.

Page 6: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

Table 2Characteristics of position generator indicators of individual social capital and place of residence of named alter (n¼2707), sorted by order of appearance in the questionnaire.

Occupation Prestige value Do you know someone on a firstname basis?

Place of residence

No (%) Yes (%) Household(%)

Neighborhood(%)

Beyondneighborhood (%)

Don’tknow (%)

High school teacher 68 55.7 44.3 0.8 11.9 31.4 0.2

Carpenter 52 56.9 43.1 1.5 12.1 29.4 0.2

Musician/artist 60 49.3 50.7 2.8 11.5 36.1 0.2

Taxi driver 35 81.4 18.6 0.2 5.9 12.1 0.3

Physician 93 40.9 59.1 0.3 11.8 45.0 1.9

Janitor 34 67.7 32.3 0.5 14.3 17.0 0.5

Registered nurse 77 49.5 50.5 1.4 12.9 35.9 0.4

Welder 55 76.0 24.0 1.0 5.5 17.5 0.1

Accountant 67 34.4 65.6 1.6 13.4 50.1 0.5

Receptionist 40 57.5 42.5 0.9 8.6 32.2 0.9

S. Moore et al. / Health & Place 17 (2011) 536–544 541

4.1. Having core ties

Women, individuals residing in higher income households,higher educational attainment, and younger adults were morelikely to have confidants with whom they could discuss importantmatters. These findings support research in the U.S. that has shownmore educated individuals and women tend to have largerdiscussion networks (McPherson et al., 2006). Increasing age hasalso been found in other studies to be associated with having fewercore ties (McPherson et al., 2006; Oh, 2003; Stoller and Pugliesi,1988). In Montreal, individuals who resided in French languagehouseholds were more likely to have core ties than those who didnot reside in such households. Among the socio-relational and -capital measures, general social participation, general diversity inoutside social capital, and the range of social ties within theneighborhood were associated with having core ties. As antici-pated, greater diversity of social ties reported in the positiongenerator was associated with a greater likelihood of having coreties. For Lin (2001), the position generator tends to capture people’sweaker, socially- or emotionally-distant relationships. In thisregard, participants who reported a greater number of weakersocial connections outside the neighborhood were also more likelyto report core network ties outside the neighborhood. In terms ofthe different dimensions of neighborhood social capital, thosepersons who were able to access resources and persons with highersocial status within the neighborhood (i.e., greater reach) werealso more likely to have general core network ties. Longitudinalresearch is needed, but the finding highlights the relationshipamong within-neighborhood weak ties, social status, and coresocial connections outside the neighborhood.

4.2. Having core neighborhood ties and low income

The study showed a significant socio-demographic pattern in thelikelihood of having a core neighborhood tie with individuals residingin lower income households more likely to have core neighborhoodties. This finding has several implications. First, work in the U.S. hasshown that more educated individuals tend to have more spatiallydispersed ties (Fisher, 1982). Our study showed that low-incomepersons in Montreal may be more likely to have core neighborhoodnetwork ties than those of higher income. Second, Carpiano (2007)suggests that neighborhood attachment operates as an effect modifierof the relationship between the neighborhood social environment andhealth. Given that income is important in individual self-selection intoneighborhoods, study findings would suggest that neighborhoodattachment is unequally distributed across neighborhood groupings.Neighborhoods may not only differ on the distribution of an important

risk condition (i.e., income) for disease, but neighborhoods may alsodiffer in the prevalence of a key effect modifier in the relationshipbetween neighborhood social environments and health. Furtherresearch is necessary to understand the importance of neighborhoodattachment as an effect modifier in different neighborhood settings.

4.3. Social capital and core neighborhood ties

Previous research on social capital and health has suggested thatproxy indicators of social capital, such as trust, participation, andperceived cohesion, reflect aspects of people’s neighborhood socialconnections. Our study suggests that among current proxy indicators ofsocial capital, only trust in neighbors is associated with having coreneighborhood ties. Unlike questions on generalized trust, questionsregarding a person’s trust in neighbors capture assessments based onknowledge of particular individual personalities and past behavior(Abbott and Freeth, 2008). Among the different dimensions of networksocial capital within the neighborhood, greater diversity in weak,neighborhood social connections was found associated with havingcore neighborhood network ties. This finding paralleled that on overallnetwork diversity and core network ties, indicating the degree to whichthe position generator is able not only to capture a person’s weak tiesbut also their core ties. In addition, the study showed that persons witha greater range in the types of resources that they could generallyaccess outside the neighborhood were less likely to have coreneighborhood ties.

4.4. Social-relational and -capital measures and SRH

Our study supports previous research in showing an associationamong several social-relational indicators (generalized trust,neighborhood social participation, perceived neighborhood envir-onment) and SRH. In addition, however, our study found thatdiversity in extra-neighborhood social capital was also associatedwith high SRH. Among these variables associated with SRH, onlydiversity in extra-neighborhood social capital was associated withhaving general core network ties; none were associated withhaving core neighborhood network connections. These findingshave several implications when it comes to our work on socialcapital and SRH. First, our study questions whether generalizedtrust at the individual level is a valid, direct proxy of actualindividual social ties outside or inside the neighborhood. Althoughresearchers tend to develop neighborhood-level measures of socialcapital through the aggregation of individual responses to thegeneralized trust question, such measures may not be capturingthe degree to which people do or do not have social connections,particularly within the neighborhood. It may be that the

Page 7: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

Table 3Adjusted Odds ratios and 95% confidence intervals of ‘‘Having one or more core social ties versus no core social ties’’ (ni¼2556; nct¼300), and ‘‘Having one or more core

neighborhood ties versus no core neighborhood ties’’ (ni¼2268; nct¼300), Montreal Neighborhood Networks and Healthy Aging Study (MoNNET-HA), 2008.

Variable Analysis one: having core network ties (n¼2556) Analysis two: having core neighborhood ties (n¼2268)

Model 1 Model 2 Model 1 Model 2

Gender

Female 1.46 (1.10–1.93)** 1.52 (1.14–2.04)** 1.04 (0.87–1.25) 1.03 (0.85–1.24)

Male (referent) 1.00 1.00 1.00 1.00

Household income per year

Less than $28,000 (referent) 1.00 1.00 1.00 1.00

$28,000 – $49,000 1.58 (1.12–2.24)** 1.28 (0.89–1.85) 1.03 (0.78–1.35) 0.95 (0.71–1.27)

$50,000–$74,000 1.59 (1.04–2.43)* 1.15 (0.74–1.80) 0.78 (0.58–1.05) 0.70 (0.51–0.96)*

$75,000–$100,000 3.31 (1.64–6.70)** 2.21 (1.07–4.58)* 0.71 (0.49–1.03) 0.65 (0.44–0.96)*

More than $100,000 8.36 (2.86–24.47)*** 5.30 (1.78–15.82)** 0.67 (0.45–0.98)* 0.54 (0.36–0.82)**

Employment status

Unemployed (referent) 1.00 1.00 1.00 1.00

Employed 1.14 (0.77–1.68) 1.05 (0.70–1.57) 0.92 (0.73–1.16) 0.91 (0.72–1.17)

Education

Less than a high school degree (referent) 1.00 1.00 1.00 1.00

High school degree or trade certificate 1.44 (1.02–2.04)* 1.16 (0.80–1.67) 1.29 (0.93–1.80) 1.14 (0.81–1.62)

College certificate 2.25 (1.42–3.57)** 1.53 (0.94–2.49) 1.16 (0.81–1.66) 0.98 (0.67–1.43)

Bachelors degree and higher 3.51 (2.24–5.50)*** 1.91 (1.18–3.10)** 1.15 (0.81–1.62) 0.97 (0.67–1.40)

Age category

25–34 years old 7.74 (3.75–16.00)*** 8.32 (3.94–17.55)*** 1.34 (0.86–2.11) 1.42 (0.89–2.28)

35–44 years old 8.20(4.16–16.16)*** 8.99(4.44–18.23)*** 1.31 (0.85–2.01) 1.34 (0.85–2.11)

45–54 years old 3.00 (1.79–5.01)*** 2.89 (1.68–4.95)*** 1.23 (0.82–1.85) 1.36 (0.88–2.08)

55–64 years old 3.12 (1.96–4.97)*** 3.05 (1.88–4.97)*** 0.97 (0.66–1.44) 1.06 (0.70–1.60)

65–74 years old 1.90 (1.32–2.74)** 1.88 (1.28–2.75)*** 1.33 (0.92–1.92) 1.38 (0.94–2.03)

75 years or more 1.00 1.00 1.00 1.00

Marital status

Married/common law union 0.86 (0.63–1.17) 0.90 (0.65–1.24) 1.09 (0.89–1.32) 1.05 (0.85–1.29)

Household language

French language 1.36 (0.96–1.94) 1.53 (1.06–2.22)* 1.03 (0.82–1.28) 1.12 (0.88–1.42)

(English/other

language referent) 1.00 1.00 1.00 1.00

Birthplace

Foreign born 0.68 (0.47–0.98)* 0.72 (0.49–1.06) 0.91 (0.72–1.15) 0.95 (0.74–1.22)

Canada born (referent) 1.00 1.00 1.00 1.00

Residential length (log10) 1.15 (0.82–1.60) 1.03 (0.72–1.47) 1.16 (0.91–1.47) 0.94 (0.73–1.21)

High generalized trust – 1.12 (0.94–1.34) – 1.06 (0.93–1.20)

Low generalized trust – 1.00 – 1.00

Trust in neighbors – 1.00 (0.89–1.12) – 1.09 (1.00–1.18)*

Outside neighborhood social participation – 1.93 (1.21–3.07)** – 1.08 (0.86–1.35)

Neighborhood social participation – 1.09 (0.74–1.60) – 1.12 (0.90–1.40)

Perceived neighborhood environment – 0.98 (0.79–1.23) – 1.09 (0.95–1.25)

Social capitalOutside neighborhood reach – 1.04 (0.89–1.20) – 1.12 (0.96–1.30)

Outside neighborhood range – 1.04 (0.81–1.33) – 0.79 (0.66–0.95)*

Outside neighborhood diversity – 1.77 (1.29–2.44)*** – 1.06 (0.89–1.26)

Neighborhood reach – 1.43 (1.08–1.90) * – 1.13 (0.97–1.32)

Neighborhood range – 0.78 (0.59–1.03) – 0.89 (0.76–1.04)

Neighborhood diversity – 0.93 (0.63–1.39) – 1.98 (1.57–2.51)***

Likelihood-ratio test (G2) (From null) 294.67*** 110.29*** (From null) 27.07 229.51***

* po0.05, ** po0.01, *** po0.001.

S. Moore et al. / Health & Place 17 (2011) 536–544542

association often found (as in this study as well) between general-ized trust and SRH operates through psychosocial (Abbott andFreeth, 2008), rather than network mechanisms. Second, althoughoverall participation was associated with having core ties ingeneral and has been shown associated with SRH in other studies,it was not shown to be associated with SRH in this study. Instead,neighborhood social participation was associated with SRH. Futureresearch might use a more sensitive typology of social participationthat distinguishes between different forms of participation andtheir location to assess more accurately the importance of parti-cipation for SRH. Third, although the study found the perceivedneighborhood environment associated with SRH, the study found

no association between the perceived neighborhood environmentscale or its separate items and the likelihood of having coreneighborhood ties. The scale had a low reliability and cautionshould be taken in interpreting this finding. Yet, as suggested aboutgeneralized trust, people’s perceptions of the neighborhood envir-onment may relate less to network mechanisms of resourceaccessibility and more to psychosocial mechanisms involvingpeople’s sense of social integration and control. Finally, the studysupports the validity of the position generator in measuring extra-and intra-neighborhood social ties, and shows that individualswith greater diversity in extra-neighborhood social capital aremore likely to report high SRH. More broadly, the study thus

Page 8: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

Table 4Adjusted Odds ratios of self-reporting ‘‘Excellent’’ or ‘‘Very Good’’ health versus self-

reporting ‘‘Good,’’ ‘‘Fair,’’ or ‘‘Poor’’ health (MoNNETs-HA) (ni¼2268; nct¼300)

2008.

Variable Model 1 Model 2

Gender

Female 1.13 (0.94–1.37) 1.12 (0.92–1.35)

Male (referent) 1.00 1.00

Household income per year

Less than $28,000 1.00 1.00

$28,000–$49,000 1.36 (1.03–1.79)* 1.28 (0.97–1.69)

$50,000–$ 74,000 1.86 (1.37–2.52)*** 1.69 (1.24–2.30)***

$75,000–$100,000 2.11 (1.45–3.08)*** 1.90 (1.30–2.78)***

More than $100,000 3.33 (2.21–5.02)*** 2.87 (1.89–4.36)***

Employment status

Unemployed 1.00 1.00

Employed 1.30 (1.02–1.64)* 1.30 (1.03–1.66)*

Education

Less than a high school degree 1.00 1.00

High school degree or trade

certificate 1.53 (1.09–2.15)* 1.53 (1.09–2.17)*

College certificate 2.21 (1.53–3.19)*** 2.11 (1.46–3.06)***

Bachelors degree and higher 2.27 (1.59–3.23)*** 2.11 (1.46–3.04)***

Age category

25–34 years old 1.02 (0.65–1.62) 1.07 (0.68–1.70)

35–44 years old 0.96 (0.62–1.48) 0.94 (0.60–1.46)

45–54 years old 0.94 (0.62–1.42) 0.98 (0.64–1.49)

55–64 years old 0.93 (0.63–1.39) 0.92 (0.61–1.37)

65–74 years 0.99 (0.68–1.43) 0.97 (0.66–1.40)

or 75 years and more 1.00 1.00

Marital status

Married/common law union 1.04 (0.85–1.27) 1.02 (0.83–1.25)

Household language

French language 1.04 (0.83–1.31) 1.07 (0.85–1.35)

(English/other language

referent)

1.00 1.00

Birthplace

Foreign born 0.59 (0.47–0.76)*** 0.62 (0.48–0.79)***

Canada born (referent) 1.00 1.00

Residential length (log10) 1.00 (0.78–1.28) 0.92 (0.71–1.18)

High generalized trust – 1.18 (1.04–1.33)**

Low generalized trust 1.00

Trust in neighbors – 1.05 (0.97–1.13)

Outside neighborhood social

participation

– 0.91 (0.73–1.14)

Neighborhood social participation – 1.28 (1.02–1.60)*

Perceived neighborhood

environment

– 1.24 (1.08–1.42)**

Social capitalOutside neighborhood reach – 1.02 (0.88–1.18)

Outside neighborhood range – 0.85 (0.71–1.01)

Outside neighborhood diversity – 1.20 (1.01–1.42)*

Neighborhood reach – 1.04 (0.90–1.20)

Neighborhood range – 1.14 (0.98–1.34)

Neighborhood diversity – 0.87 (0.71–1.06)

Likelihood-ratio test (G2) 182.67*** 41.87***

* po0.05, ** po0.01, *** po0.001.

S. Moore et al. / Health & Place 17 (2011) 536–544 543

suggests for the general adult population that high SRH is asso-ciated more with people’s connections and access to resourcesoutside the neighborhood than inside.

4.5. Limitations

Findings should be interpreted in light of the following con-siderations. First, this study has focused as an outcome on corenetwork ties in general and the neighborhood in particular. This isonly one aspect of a person’s social networks and other aspects,such as weak, acquaintance relationships, may be shown in thefuture to have different associations with socioeconomic and social

capital indicators. Second, the interview asked participants todefine for themselves where their neighborhood boundaries lied.This meant that within neighborhoods, participants may holddifferent conceptions of those boundaries. Theoretically, thisdecision is justifiable for this study since the key factor is whetherthe participants themselves perceived their social ties as spatially-proximate. Finally, the study is cross-sectional in design, meaningfor example that there may be recursive effects of trust in neighborson having core neighborhood ties. Trust is more upstream in thisstudy since research has posited trust as a proxy for social capital.Future longitudinal research would help delimit potential recur-sive relationships among trust, participation, social capital, andneighborhood connections.

5. Conclusion

The breadth of research that has shown significant associationsbetween generalized trust, participation, and health highlights theimportance of those constructs for social epidemiological research.Yet, the specific mechanisms by which social capital influenceshealth have remained unclear. This study suggests that conven-tional measures of social capital, such as generalized trust, parti-cipation, or the perceived neighborhood environment, may notsuccessfully capture network mechanisms related to having coreconnections within one’s neighborhood social environment. Forhealth and place research, more broadly, our findings call for a morecritical appraisal of the mechanisms linking social capital andhealth, and the further delineation of network and psychosocialmechanisms in understanding these links.

Acknowledgements

S.M. holds a New Investigator Award from the CanadianInstitutes of Health Research (CIHR)—Institute of Aging. U.B. holdsa Canada Research Chair in e-Marketing. M.D. holds a ResearchChair for Social Epidemiology from the University of SouthAustralia, and held a Canada Research Chair for BiopsychosocialPathways in Population Health from the CIHR during project designand funding of the results reported here. K.F. holds a CIHR NewInvestigator Award and is a Humboldt Experienced Researcher. L.G.holds a CIHR/Centre de Recherche en Prevention de l’Obesite(CRPO) Applied Public Health Chair in Neighborhoods, Lifestyle,and Healthy Body Weight. L.R. is a FRSQ National Scholar (#16207).

Data collection and analysis was supported by a team grant fromthe Canadian Institutes of Health Research (Grant no. MOP-84584).

References

Abbott, S., Freeth, D., 2008. Social capital and health: starting to make sense of the role ofgeneralized trust and reciprocity. Journal of Health Psychology 13, 874–883.

Brewer, D.D., 2000. Forgetting in the recall-based elicitation of personal and socialnetworks. Social Networks 22, 29–43.

Bourdieu, P., 1986. In: Richardson, J.G.., Westport, CT (Eds.), ‘‘Forms of Capital’’ inHandbook of Theory for the Sociology of Education. Greenwood Press, pp. 241–258.

Carpiano, R.M., 2006. Toward a neighborhood resource-based theory of socialcapital for health: can Bourdieu and sociology help? Social Science & Medicine62, 165–175.

Carpiano, R.M., 2007. Neighborhood social capital and adult health: an empirical testof a Bourdieu-based model. Health & Place 13 (3), 639–655.

Caughy, M.O., O’Campo, P.J., Muntaner, C., 2003. When being alone might be better:neighborhood poverty, social capital, and child mental health. Social Science &Medicine 57 (2), 227–237.

Feld, S., Carter, W., 2002. Detecting measurement bias in respondent reports ofpersonal networks. Social Networks 24, 365–383.

Fisher, C., 1982. To Dwell Among Friends. University of Chicago Press, Chicago.Fujiwara, T., Kawachi, I., 2008. A prospective study of individual-level social capital

and major depression in the United States. Journal of Epidemiology andCommunity Health 62, 627–633.

Page 9: Social capital and core network ties: A validation study of individual-level social capital measures and their association with extra- and intra-neighborhood ties, and self-rated health

S. Moore et al. / Health & Place 17 (2011) 536–544544

Goyder, J., Guppy, N., Thompson, M., 2003. The allocation of male and femaleoccupational prestige in an Ontario urban area: a quarter-century replication.The Canadian Review of Sociology and Anthropology 40 (4), 417–440.

Helliwell, J.F., Putnam, R., 2004. The social context of well-being. Proceedings of theRoyal Society B: Biological Sciences 359, 1435–1446.

Hyppa, M.T., Maki, J., 2003. Social participation and health in a community rich instock of social capital. Health Education Research 18, 770–779.

Kim, D., Kawachi, I., 2006. A multilevel analysis of key forms of community- andindividual-level social capital as predictors of self-rated health in the UnitedStates. Journal of Urban Health 83, 813–826.

Kim, D., Subramanian, S.V., Kawachi, I., 2008. Social capital and physical health: asystematic review of the literature. In: Kawachi, I., Subramanian, S.V., Kim, D.(Eds.), Social Capital and Health. Springer Press, New York, pp. 139–190.

Lindstrom, M., Moghaddassi, M., Merlo, J., 2004. Individual self-reported health,social participation, and neighborhood: a multilevel analysis in Malmo, Sweden.Preventive Medicine 39, 135–141.

Lin, N., 2001. Social Capital: A Theory of Social Structure and Action. CambridgeUniversity Press, Cambridge.

Marsden, P., 1987. Core discussion networks of Americans. American SociologicalReview 52, 122–131.

Marsden, P., 2006. Network methods in social epidemiology. In: Oakes, M.,Kaufman, J. (Eds.), Methods in Social Epidemiology. Jossey-Bass, San Francisco,pp. 267–286.

McPherson, M., Smith-Lovin, L., Brashears, M., 2006. Social isolation in America:changes in core discussion networks over two decades. American SociologicalReview 71, 353–375.

Moore, S., Shiell, A., Hawe, P., Haines, V.A., 2005. The privileging of communitarianideas: citation practices and the translation of social capital into public healthresearch. American Journal of Public Health 95 (8), 1330–1337.

Moore, S., Daniel, M., Paquet, C., Dube, L., Gauvin, L., 2009a. Association of individualnetwork social capital with abdominal adiposity, overweight, and obesity.Journal of Public Health March 31, 175–183.

Moore, S., Daniel, M., Gauvin, L, Dube, L., 2009b. Not all social capital is good capital.Health & Place 15, 1071–1077.

Moore, S., Gauvin, L., Daniel, M., Kestens, Y., Dube, L., Bockenholt, U., Richard, L.,2010. Associations among socioeconomic status, perceived neighborhoodcontrol, perceived individual control, and self-reported health. Journal ofCommunity Psychology 38 (6), 729–741.

Oh, J-H., 2003. Assessing the social bonds of elderly neighbors: the roles of length ofresidence, crime victimization, and perceived disorder. Sociological Inquiry73 (4), 490–510.

Pollack, C.E., von dem Knesebeck, O., 2004. Social capital and health among theaged: comparisons between the United States and Germany. Health & Place 10,383–391.

Poortinga, W., 2006a. Social capital: an individual or collective resource for health?Social Science & Medicine 63, 292–302.

Poortinga, W., 2006b. Do health behaviors mediate the association between socialcapital and health? Preventive Medicine 43, 488–493.

Stoller, E., Pugliesi, K., 1988. Informal networks of community-based elderly.Research on Aging 10, 499–516.

Subramanian, S.V., Kim, D.J., Kawachi, I., 2002. Social trust and self-rated health in UScommunities: a multilevel analysis. Journal of Urban Health 79, S21–S34.

Subramanian, S.V., Lochner, K., Kawachi, I., 2003. Neighborhood differences insocial capital: a compositional artifact or a contextual construct? Health & Place9, 33–44.

Swaroop, S., Morenoff, J., 2006. Building community: the neighborhood context ofsocial organization. Social Forces 84 (3), 1665–1695.

Veenstra, G., 2000. Social capital, SES, and health: an individual-level analysis. SocialScience & Medicine 50, 619–629.

Veenstra, G., Luginaah, I., Wakefield, S., Birch, S., Eyles, J., Elliott, S., 2005. Who youknow, where you live: social capital, neighborhood and health. Social Science &Medicine 60, 2799–2818.

Weden, M., Carpiano, R, Robert, S.A., 2008. Subjective and objective neighborhoodcharacteristics and adult health. Social Science & Medicine 66 (6), 1256–1270.