ARTICLE IN PRESS

1353-8292/$ - se

doi:10.1016/j.he

�CorrespondHealth, The Li

30 Bond Street,

fax: +1416 864

E-mail addr

Health & Place 15 (2009) 56–68

www.elsevier.com/locate/healthplace

Neighbourhoods and mental well-being: What are the pathways?

Patricia O’Campoa,b,c,�, Christina Salmona, Jessica Burkec,d

aThe Centre for Research on Inner City Health, The Li Ka Shing Knowledge Institute of St. Michael’s Hospital, 30 Bond Street,

Toronto, Ont., Canada M5B1 W8bPublic Health Sciences, University of Toronto, Ont., Canada

cDepartment of Population and Family Health Sciences, Johns Hopkins Bloomberg School of Public Health, USAdGraduate School of Public Health, University of Pittsburgh, Pittsburg, PA, USA

Received 30 August 2007; received in revised form 3 January 2008; accepted 4 February 2008

Abstract

The last decade has seen an acceleration in research examining the impact of neighbourhood characteristics on a health outcomes. Yet,

identifying specific pathways by which neighbourhoods affect mental well-being has received less attention. The article describes findings

from a qualitative study conducted in Toronto, Ont., Canada, examining the pathways by which neighbourhood characteristics are

related to mental well-being. Methods of concept mapping were used to answer our research questions. Participants identified over 100

neighbourhood characteristics grouped into six cluster that were important for either good or poor mental well-being. Clusters were

rated in terms of their importance to mental well-being. Finally, participants drew diagrams describing how the neighbourhood factors

and domains were contributing to poor and good mental well-being. Concept mapping provides unique contributions and challenges and

may provide a useful means of examining specific pathways from neighbourhood to health outcomes.

r 2008 Elsevier Ltd. All rights reserved.

Keywords: Neighbourhoods; Mental health; Qualitative methods; Urban

Introduction

In developed countries, the percentage of people who suffermental disorder during the course of a year is approximately20–30% (Bijl et al., 2003). This 1-year mental disorderprevalence is tied to considerable health risks and economicstrain on society, and upon the sufferer. Studies suggest thatmental illness may be as disabling as leading chronicconditions including, heart disease and diabetes (StatisticsCanada, 2002). The World Health Organization states that 5of the 10 leading causes of disability are related to mentaldisorders (Statistics Canada, 2002).

The majority of research on mental well-being focuses onthe personal or family determinants of mental healthoutcomes (Dohrenwend and Dohrenwend, 1971; Kessler,

e front matter r 2008 Elsevier Ltd. All rights reserved.

althplace.2008.02.004

ing author at: The Centre for Research on Inner City

Ka Shing Knowledge Institute of St. Michael’s Hospital,

Toronto, Ont., Canada M5B1 W8. Tel.: +1 416 864 6060;

5485.

ess: ocampop@smh.toronto.on.ca (P. O’Campo).

1979; Mirowsky and Ross, 1990; Wu et al., 2003).However, the past decade has uncovered growing evidencefor the impact of area level determinants on a range ofphysical and mental health outcomes (Galea et al., 2005;Muntaner et al., 2004; O’Campo et al., 1995; Rajaratnamet al., 2006; Veenstra, 2004). Most of these studiesdemonstrate that an association exists between a fewcommunity factors, such as neighbourhood socioeconomicposition or impoverishment, and physical or mental health(Goldsmith et al., 1998; Ostir et al., 2003; Ross, 2000a, b;Ross et al., 2000; Silver et al., 2002). Yet, a growingnumber are focusing beyond those few characteristics toinclude multiple neighbourhood domains (Kirby andKaneda, 2005; Kubzansky et al., 2005). These latter studiessuggest that the pathways from residential neighbourhoodcharacteristics to mental well-being differs from the path-ways by which they impact physical health (Veenstra, 2004;Ziersch et al., 2005; Wainwright and Surtees, 2003; Foneet al., 2007; Guite et al., 2006).Few studies have tried to uncover the pathways by which

neighbourhood factors lead to good or poor mental

ARTICLE IN PRESSP. O’Campo et al. / Health & Place 15 (2009) 56–68 57

well-being. Identification of the pathways by whichneighbourhoods affect well-being is critical to moving thefield forward toward a more complete understanding of thecomplex relationships between residential environmentsand well-being. Such information can inform specifictheories and conceptual frameworks that are more appro-priate for the study of health than are the current theories‘borrowed’ from the social sciences on outcomes such ascrime and delinquency (Rajaratnam et al., 2006; O’Campo,2003). Moreover, data on pathways will inform programsand interventions to improve well-being.

A few studies have recently begun to identify pathwaysby which community characteristics contribute to mentalwell-being. In an attempt to explain the pathways by whichimpoverishment affects mental health problems, someauthors suggest that physical and social disorder (e.g.,deteriorating or poorly maintained building or streets,public disorderly conduct of residents) signals a breakdown in community social control leading to increasinglythreatening and distressing environments which negativelyimpact mental health (Galea et al., 2005; Ross, 2000a; Rosset al., 2000; Steptoe and Feldman, 2001). Others havefocused on the role of features such as the levels andquality of green areas within neighbourhoods (Dilani,2001; Kaplan and Kaplan, 1989; Nielsen and Hansen,2007). While neighbourhood social and health serviceshave been theorized to have a positive impact on health ormight even provide a buffer to the negative impact ofpoverty on health, empirical findings fail to providesupport for this idea (Kubzansky et al., 2005).

Another line of inquiry involves social capital. Area levelsocial capital, which is a consequence of many individuals’involvement in local social and political activities, mayimprove health by increasing network members’ socialcontact, support and cohesion. Alternatively, the conse-quences of high community social capital, such asneighbourhood improvements or neighbourhoods withabundant social and health resources, can also positivelyimpact physical and mental health (Veenstra, 2004; Foneet al., 2007; Lindstrom, 2004). Social capital or socialcohesion has been hypothesized to play a complex role inthe production of mental well-being by acting as amediator between neighbourhood disorder and health(Kruger and Reischl, 2007) or a mediator of the relation-ship between income deprivation and mental health (Foneet al., 2007). Still, other explanations have focused ondemographic homogeneity in neighbourhoods among non-majority populations such as racial minorities or the elderly(Ostir et al., 2003; Kubzansky et al., 2005).

Yet, despite the emergence of studies going beyond anexamination of neighbourhood socioeconomic conditionsand health, the task of uncovering the pathways fromneighbourhoods to mental well-being is largely undone.With regard to quantitative studies, the challenge, in part,has been a lack of comprehensive data at multiple levels(e.g., neighbourhoods and individual) to explore a full setof neighbourhood characteristics as well as examining the

presence of cross-level pathways. One promising approachat this stage is to undertake qualitative research on thistopic, some of which has begun to emerge (Nielsen andHansen, 2007; Burke et al., 2005, 2006; O’Campo et al.,2005; Warr et al., in press). Given the range of qualitativemethods available, consideration should be given to themost appropriate methods to employ for the question ofexploring pathways and mechanisms. Concept mapping,which is a structured mixed-method process that yields aconceptual framework for how a group views a particulartopic is an appropriate method for generating informationabout neighbourhoods and well-being (Burke et al., 2005,2006; O’Campo et al., 2005). This method has successfullybeen used in the past to examine whether and howresidential neighbourhoods impact the risk of intimatepartner violence and the cessation of partner abuse forthose who are in violent relationships (Burke et al., 2006;O’Campo et al., 2005).In this paper, we contribute to the literature on pathways

between neighbourhood characteristics and mental well-being. We examine the perspectives of those with livedexperiences with how neighbourhoods affect mental well-being. To achieve our goals we use methods of Conceptmapping, a semi-qualitative method that provides aconceptual framework to learn how specific groupsconceptualize a particular subject.

Methods

Concept mapping was used as the methodology in thisstudy. Trochim (1989) describes concepts mapping as ‘‘astructured process, focused on a topic or construct ofinterest, entailing input from one or more participants, thatproduces an interpretable pictorial view (concept map) oftheir ideas and concepts and how these are interrelated’’Trochim (1989). Specific data gathering activities arecompleted by each individual participant (brainstorming,sorting rating, diagramming) in order to fully representeach individual’s viewpoint while also incorporating groupconsensus. Qualitative and quantitative methods are usedin concept mapping by creating a visual display of how theparticipants and the group as a whole conceptualize aparticular topic. This method has recently emerged as anew tool to aid in the gathering of information from theperspective of consumers or community within publichealth (Burke et al., 2005, 2006; O’Campo et al., 2005;Trochim et al., 2003).

Recruitment

Participants were recruited from Toronto, a large urbanarea comprised of a dense downtown core containing thecentral business district surrounded by inner suburbanareas that, while still urban, are less crowded. Accessiblesocial and health services are concentrated in the down-town area (i.e., availability and ease of access via publictransit) while with the inner suburbs, such services are

ARTICLE IN PRESS

Table 1

Working definitions, focal question, and rating questions used in the study

Neighbourhood A physically bounded area characterized

by some degree of homogeneity and

sometimes social cohesion.

Mental well-being Mental health refers to the psychological

well-being of a person. This may include

positive mental states such as being

satisfied with life, happiness, or being

stress-free. Mental health also includes

poor psychological states such as being

highly stressed, feeling anxious, being

fearful, bored or unhappy. Finally,

mental health can include mental illness

like major depression or even substance

abuse like alcoholism.

Focal question for

brainstorming

What are some characteristics of

neighbourhoods that could relate in any

way, good or bad, to a persons’ mental

well-being?

Rating question about good/

positive mental well-being

Please rate on a scale of 1–5 how each of

these neighbourhood characteristics is

related to a person’s good or positive

mental well-being.

Rating question about poor/

negative mental well-being

Please rate on a scale of 1 to 5 how each

of these neighbourhood characteristics is

related to a person’s poor mental well-

being and/or mental illness.

P. O’Campo et al. / Health & Place 15 (2009) 56–6858

fewer, far between and less accessible. For the purposes ofcomposing our groups, we used two stratification factors:geographic area (downtown and inner suburbs) andsocioeconomic position (as measured by a proxy forincome, low and non-low). Area of residence wasdetermined through asking the address of the individualas well as the name of the area they lived in to identifywhether they resided downtown or in the outskirts of town.During the screening process, income levels were approxi-mated by asking if they routinely visited certain communityorganizations (e.g., homeless shelters, free health clinics)which serve low income individuals, if they were employedand if they owned their own home.

We posted flyers in doctor’s offices, community centres,listservs, and day labour centres containing very generalinformation about the study and a number to contact ifpotential participants were interested. At some locations,flyers were also given to potential participants by staff.Those interested were asked to phone in and were screenedby one of the authors (C.S.). A number of different criteriahad to be met in order for them to be eligible toparticipants in the study. They had to have lived in theirneighbourhoods for 6 months or more, be older then 18,have good written and oral English skills and had to havetime to attend up to three discussion sessions. Therequirement of knowledge of English was necessary asconcept mapping activities require manipulation of con-cepts written on cards and short surveys.

Data collection

The data collection process entails three primaryactivities; brainstorming activities; sorting and ratinggroups; and groups to discuss pathways from neighbour-hood factors to mental well-being. All data collectionactivities were approved by St. Michael’s Hospital Re-search Ethics Board. While detailed descriptions of theconcept mapping method and the specific data collectionprocedures are described in detail elsewhere (Burke et al.,2005), we describe each step briefly here. The groups wereheld at two different locations that were easily accessible bypublic transportation. During the first group we adminis-tered a short survey to all participants which collectedinformation on income, length of time they resided in theirneighbourhoods and a short screen for depression (CES-D)(Radloff, 1977).

Brainstorming

Brainstorming activities were comprised of groupsessions involving free listing of ideas as well as interviewand written responses to questions. Most of the brain-storming data were collected through groups of about 1.5 hin length, two in downtown and two inner suburbanneighbourhoods. Participants were provided descriptionsof the terms ‘‘neighbourhood’’ and ‘‘mental well-being’’(see Table 1) to ensure that the participants had similar

notions of these terms within and across groups. Duringthe group, participants were asked to generate a list ofitems that described the ‘‘characteristics of your neigh-bourhood that can affect mental well-being in a good orbad way.’’ Three groups were held, two low income groupsfrom the inner suburban area and downtown Toronto, anda non-low income group from downtown Toronto. Weobtained input from the non-low income inner suburbanparticipants via interviews and written responses toquestions about neighbourhood characteristics that affectmental well-being.The 381 non-unique items generated across the brain-

storming activities were reviewed. We eliminated duplicatesand combined similar items to reduce the list to one masterlist of 120 items (see Table 2). This master list was used inthe subsequent sorting and rating groups.

Sorting and rating

Sorting and rating groups of about 4 h in duration wereheld, two in downtown Toronto and two in the innersuburbs. For the sorting task, participants were asked toplace the 120 items from the master list into piles that‘‘made sense to them’’ and label the piles accordingly.Specifically, participants grouped similar items togetheraccording to their own perspectives. So, one participantmay group ‘‘walkable areas’’ with ‘‘pedestrians’’ whileanother participant might put ‘‘walkable areas’’ with bikepaths and lanes’’. Labels were given to the piles by eachparticipant reflecting the theme of the items in the pile.

ARTICLE IN PRESS

Table 2

Items and their ratings generated by participants in the concept mapping session. Items are presented by cluster

Item name (item no.) Good mental

well-being

Poor mental

well-being

Item name (item no.) Good mental

well-being

Poor mental

well-being

Cluster 1: Necessary human and social services Cluster 2: Neighborhood support for each other

Good public and social services (100) High Moderate Safe environment (1) High Moderate

Access to good public transportation

(29)

High Moderate Cleanliness (38) High Moderate

Access to medical or health facilities (8) High Moderate Friendliness of neighbours (48) High Moderate

Afterschool programs (90) High Moderate Communication between residents (91) High Moderate

Community activities and festivals (92) High Low Knowing neighbours (120) High Low

Recreation centers (22) High Low Police presence (63) High Moderate

Community centers (40) High Low Sense of community (69) High Moderate

Crisis intervention programs (93) High Moderate Privacy (20) High Low

Affordable programs for adults (89) High Moderate Residents being involved in neighbourhood

change (23)

High Low

Accessibility for the disabled (86) High Moderate Interaction between neighbours (54) High Moderate

Accessible and affordable day care (87) High Low Neighbourhood pride (17) High Low

Libraries (55) High Low Diverse neighbourhoods (44) High Moderate

Employment placement & training

services (96)

High Moderate Social cohesion (74) High Low

Seniors homes (24) Moderate Low Neighbourhood governance (58) High Low

Mental health clinics (116) Moderate Moderate Multicultural neighbourhood (4) High Low

Churches and other places of worship

(3)

Moderate Low Neighbourhood watch (18) High Low

Homeless shelters (102) Moderate Moderate Visibility of residents (82) High Low

Eviction prevention services (97) Moderate Moderate Neighbourhood change (57) Moderate Low

Fire departments (14) Moderate Low Surveillance cameras (110) Moderate Moderate

Accessible rehab programs (107) Moderate Moderate Discrimination (43) Moderate High

Women’s shelters (28) Moderate Low Neighbourhood stress (59) Moderate Moderate

Duplicative public and social services

(45)

Moderate Moderate Indifference about neighbours’ well-being (53) Moderate Moderate

Fear of helping neighbours (47) Low Moderate

Media portrayal of neighbourhood (e.g.,

slandering, 105)

Low Moderate

Construction (41) Low Moderate

Cluster 3: Green area and natural environment Cluster 4: Social makeup of people in the neighbourhood

Parks (106) High Low Good paying jobs (99) High Low

Green areas (51) High Low Affordable housing (34) High Moderate

Trees (111) High Low Well maintained houses (118) High Low

Air quality (2) High Moderate Efficient/effective property management (94) High Low

Accessible and accountable politicians

(85)

High Low Rent prices (68) High Moderate

Street lighting (75) High Moderate Neighbourhood economic development (32) High Low

Response time of emergency services

(95)

High Moderate Subsidized housing (25) High Moderate

Walkable areas (83) High Moderate Children (37) High Low

Good lighting (15) High Low Youth (84) Moderate Low

Gardens (49) High Low Average income (36) Moderate Moderate

Waterfront (112) High Low Proportion renters (67) Moderate Low

Sidewalks (70) High Low Average educational levels (30) Moderate Low

Bike paths and lanes (31) Moderate Low Gentrification (50) Moderate Low

Signage (109) Moderate Low Population density (64) Moderate Moderate

Traffic (7) Moderate Moderate Age diversity (35) Moderate Low

Garbage on sidewalks and street (113) Moderate Moderate Pets (19) Moderate Low

Poorly maintained parks/open space

(119)

Low Moderate Single family homes (71) Moderate Low

Poor road/sidewalk maintenance (117) Low Moderate Town homes (79) Moderate Low

Cobblestone roads (39) Low Low Proportion home owners (66) Moderate Low

Mixed residential and commercial neighbourhood

(56)

Moderate Low

Student density (76) Moderate Low

Unemployment (80) Moderate High

High poverty (114) Moderate High

Property taxes (65) Moderate Moderate

Highrises (52) Moderate Moderate

Irresponsible landlords (103) Moderate High

Run down buildings (108) Low Moderate

Factories (46) Low Low

P. O’Campo et al. / Health & Place 15 (2009) 56–68 59

ARTICLE IN PRESS

Table 2 (continued )

Item name (item no.) Good mental

well-being

Poor mental

well-being

Item name (item no.) Good mental

well-being

Poor mental

well-being

Cluster 5: Neighbourhood affordability Cluster 6: Negative community factors

Affordable grocery stores (11) High Low Violence (81) Moderate High

Smells (73) Moderate Moderate

Availability of good schools (12) High Low Crimes (42) Moderate High

Accessible and affordable parking (88) Moderate Low Noise (6) Moderate Moderate

Accessible and affordable shopping (10) Moderate Low Pests (e.g., rats, wildlife, insect infestation) (61) Moderate Moderate

Tourists (78) Moderate Low Police harassment (62) Moderate High

Accessible and affordable restaurants

(9)

Moderate Low Substance abuse (77) Moderate High

Proximity to university (33) Moderate Low Drug dealing (13) Low High

Pedestrians (60) Moderate Low Prostitution (21) Low High

Vibrant nightlife (27) Low Low Gunshots (101) Low High

Nightclubs and bars (5) Low Moderate Vandalism (26) Low High

Impatient bus drivers (115) Low Moderate

Graffiti (16) Low Moderate

Kids congregating and smoking (104) Low Moderate

Gang symbols (98) Low Moderate

P. O’Campo et al. / Health & Place 15 (2009) 56–6860

Participants were told to not have too few piles (e.g., 2) ortoo many piles (e.g., 100).

After completing the sorting, each participant was askedto complete four rating sheets. The 120 items were placedon a sheet and were rated according to two areas of mentalwell-being (see Table 2).

During the sorting and rating session, the data wereimmediately entered into the concept systems softwarewhere multidimensional scaling and cluster analysis wereperformed to obtain information about the distance ofeach item to all other items and to create ‘‘clusters’’ oritems representing conceptual domains. The conceptsystems software performs these tasks quickly whichenabled us to share the results with the participants duringthe latter half of the sorting and rating group session. Thismade it possible for the participants to immediately view,understand and participate in the analyses of the conceptmaps that yielded a final group map which representedwhat they believe as a group was the correct representationof their viewpoints. To analyze the data with participantswithin the groups, an emphasis was placed on viewing,confirming and correcting the concept maps. Participantsgave input into the number of total clusters that wouldrepresent their ideas. The Concept Systems softwareallowed for easy display of maps with different numbersof clusters. The groups ended up with maps that rangedfrom five to six clusters in size.

After holding the four groups, we had only groupspecific representations of the concept map. Moreover,each group had only processed data from their own groupand did not have the benefit of seeing the maps generatedin the other groups. The researchers, therefore, processedthe data further to arrive upon the final map concept map.Stress values, a multidimensional scaling diagnostic statis-tic that reflects the goodness of fit of the map to the originaldissimilarity matrix that served as input Kruksal and Wish

(1978), were calculated for group specific and overall mapsolutions and were found to be very similar. A lower stressvalue implies a better fit. The stress value for the mapgenerated with all the data was 0.33, within the range ofmost concept maps generated (Kane and Trochim, 2006).The final map was taken back to each group, incorporatingfurther feedback from all groups, and confirmed. See Fig. 1for the final cluster map.

Pathway identification groups

The third type of group held was the pathway diagram-ming groups were about 1.5 h in length. Here theparticipants extended the information of the maps andunderwent group discussions about how neighbourhoodsaffect mental well-being. Small groups of three and fourworked together to create the diagrams that representedtheir ideas about the pathways from the items and clustersto mental well-being. Participants had access to the clusterdata but did use the rating information (although ratinginformation had been shared with each group in the priorsorting and rating sessions). We focused upon both thewhole set of clusters as well as those clusters that tended toreceive higher ratings in their relation to good mental well-being to discuss pathways. Figs. 5–7 illustrate three suchfigures generated by the groups.

Results

We had 36 persons participate in our study; 17 in thedowntown groups and 19 in the inner suburb groups.Among the 21 persons we had in our low-socioeconomicposition (SEP) group, 90% had incomes under $20,000 peryear. Of the 15 non-low SEP group, 90% earned over$35,000 with some earning over $100,000 per year. Therewere 11 men and 25 women in our groups. We spent

ARTICLE IN PRESS

1

2

3

4

5

6

7

8

9 10

11 12

13

14

15

16

17 18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37 38

39

40

41

42 43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86 87

88

89 90

91

92

93

94

95

96

97

98

99

100

101

102

103

104 105

106

107

108

109

110

111

112

113

114 115

116

117

118

119

120

Negative community factors

Social make-up of peopleIn the neighborhood

Neighborhoodaffordability

Necessary human and socialservices

Neighborhood support for eachother

Green area and naturalenvironment

Fig. 1. Cluster map for the relationship between neighborhoods and mental well-being, six cluster solution.

P. O’Campo et al. / Health & Place 15 (2009) 56–68 61

approximately 6 h with each individual talking about andundertaking activities regarding neighbourhoods andmental well-being across all the groups and interviews.

Brainstorming: Table 2 presents our results of thebrainstorming process. We present the 120 items (sortedinto their final cluster location) in that table.

Sorting data: The final six cluster concept map ispresented in Fig. 1. The items comprising the clusters arenoted by numbers in the figure and the numberscorrespond to the item numbers contained in Table 2.Items that are closer to one another are more closelyrelated to one another as determined by the multidimen-sional scaling analyses. For example, item 47 (fear ofhelping neighbours) at the top of Neighbourhood Supportfor Each Other is closely related to item 53 (indifferenceabout neighbour’s well-being) but is not as closely relatedto item 1 (safe environment) which is further down in thatsame cluster (Fig. 1). The clusters are heterogeneous intheir content and can be considered larger domainsimpacting positive and negative mental well-being.

Rating data: Rating data on how strongly each item isrelated to both good and poor MWB are presented inTable 1. While the ratings were made on a five-point scale

ranging from 1 ¼ not related at all to 5 ¼ strongly related,we have categorized the average item ratings into high (3.8and higher), moderate (2.9–3.7) and low (2.8 or lower)(O’Campo et al., 2005). For example, if we examine ‘‘goodand public social services’’ the participants rated it as beinghighly related to good mental well-being, while it was onlymoderately rated to poor mental well-being; ‘‘recreationcentres’’ were rated high in relation to good mental well-being but low in relation to poor mental well-being.Participants gave far fewer high ratings for poor mental

well-being compared to good mental well-being. Severalitems within the negative community factors clusterwere given high ratings for poor mental well-being:violence, crimes, vandalism, and drug dealing amongothers. Outside of that cluster just a few items were ratedas being highly important for poor mental well-being: lowincome, high unemployment, discrimination and irrespon-sible landlords.Items that were related as highly important for good

mental well-being included social services and resourcessuch as accessible public transportation, affordablechild care, libraries, and access to health care services.Items related to safety, neighbourhood cohesion and

ARTICLE IN PRESSP. O’Campo et al. / Health & Place 15 (2009) 56–6862

neighbourhood green space were also considered to beimportant for good mental well-being.

Another way to examine the rating data is to use a graphcalled Pattern Matching (Fig. 2). Here, average ratingsfor each cluster are plotted. Ratings for both good andpoor mental well-being are contrasted in Pattern Matching(Fig. 2). This allows us an opportunity to graphicallyexamine the strength of relationship between the clusters,good mental well-being and poor mental well-being. Forexample, while the necessary human and social servicescluster was highly rated in relation to good mental well-being, it was ranked fourth in importance for poor mentalwell-being. More starkly, the negative community factorscluster was ranked most important for poor mental well-being, while it was ranked the lowest in relation to goodmental well-being. The correlation between the ratings ofgood mental well-being and poor mental well-being was�0.62.

Fig. 3 illustrates how we might combine cluster mapswith rating information to explore a single cluster ingreater detail. Fig. 3 is a cluster rating map for good mentalwell-being and displays the cluster as a stacked set ofplates, the higher the stack the higher the average rating forthat cluster in relation to good mental well-being. Thepoints corresponding to the items are also stacked, thehigher the tower the stronger the relationship of that itemto good mental well-being. By viewing the rating informa-tion within the context of a cluster map we can identifyitems that are both closely related to each other but alsostrongly related to good mental well-being. On the far right

Good mental well-being

4

2

Necessary human and social servicesNeighborhood support for each other

Green area and natural environment

Social makeup of people in the neighborhoodNeighborhood affordability

Negative community factors

3

r = -0.6

Fig. 2. Cluster map for the relationship between neighb

corner of the cluster, a set of items that are very closelyrelated with high point towers is identified; the specificitems are listed in the box. The theme of these related itemsis clear, positive neighbor interactions. There are threeother items that are close to this cluster—17 (neighbour-hood pride), 18 (neighbourhood watch) and 58 (neighbour-hood governance)—that were not listed along with thegrouped items. We used another source of information toinform this decision, bridging values. Bridging values,generated from multidimensional scaling, are an indicatorof how ‘‘anchored’’ an item is to that spot on the map orwhether it is ‘‘bridged’’ (or connected to many other itemson the map). The values go from zero to 1.0 with lowvalues indicating an anchored item and a higher valueindicating a bridged item. Our bridging values for all itemswent from a low of 0.00 to a high of 0.72. The items in thegroup in the box in Fig. 3 are all highly anchored withvalues around 0.13 to about 0.21. The adjacent items notincluded in the box had bridging values much higher, 0.34,0.42 and 0.47 suggesting that these items are moreconnected to other items all across the map than is thecase for the anchored items listed in the box. This exampleis just to illustrate the multiple sources of information thatcan be used to understand the data that are generatedthrough the concept mapping process.

Differences by geographic area: We found a fewdifferences in the average cluster rating between thosewho reside in the downtown versus the innersuburban areas. We illustrate one such examination ofour data, the Pattern Match (Fig. 4). Among those residing

Poor mental well-being

4

2

Necessary human and social services

Neighborhood support for each other

Green area and natural environment

Social makeup of people in the neighborhood

Neighborhood affordability

Negative community factors

3

2

orhoods and mental well-being, six cluster solution.

ARTICLE IN PRESS

Neighborhood Support for Each Other Cluster

Highly related group of items within the cluster:

23. Residents being involved in neighborhood change48. Friendliness of neighbors54. Interaction between neighbors69. Sense of community 74. Social cohesion91. Communication between neighbors120. Knowing neighbors

41

58

17

59

5347

43

105

63

11044

4

381

57

2018

Fig. 3. Within cluster examination of neighborhood support for each other, cluster rating map.

P. O’Campo et al. / Health & Place 15 (2009) 56–68 63

in the inner suburban areas, the average cluster rating washighest for necessary and human and social services, whilethis came in 3rd for those in the downtown area. This wasthe only difference between the downtown and innersuburb ratings and the correlation between the twowas 0.86.

Differences by depression: We administered the CES-Dto participants, a scale that measures depression. Weexamined the extent to which depression (a score of 16 orhigher on the CES-D) affected participants’ ratings ofneighbourhood items for both good mental well-being orpoor mental well-being. Pattern matching graphs indicatethat there were no differences in the average cluster ratingsfor those with and without CES-D scores higher than 16and the correlations between those were depressed com-pared to those who were not depressed for good MWBwere 0.99 and for poor mental well-being were 0.99 (graphsnot shown).

Diagramming pathways: In small groups, participantsdiscussed and their understanding of the pathways bywhich items, or clusters of items, were related to goodmental well-being. We select three here, out of more than adozen, to illustrate the kinds of diagramming produced inthese sessions. For purposes of illustration, we intention-ally selected diagrams that differed in the approach todepicting the pathways. Fig. 5 shows the relationship of theitems in the negative community factors cluster to poormental well-being. Here the group subdivided the itemsinto three smaller subgroups as illustrated in the linked‘clouds’. These subgroups represent similar items and eachsubgroup had separate links to poor mental well-being.Thus, Negative Community Factors could be furtherbroken down into smaller subgroups of similar itemsaccording to the authors of that figure. Fig. 6 presents thepathways drawn by one group for the items in the

necessary human and social services cluster and goodmental well-being. The group thought that ‘good Publicand Social Services’ were driving the pathway to goodmental well-being (see top of Fig. 5). The various serviceswere broken up into two groups (e.g., access to MentalHealth Clinics or Homeless Shelters) and in the othergroup, Recreational Centres or Libraries). These serviceshad direct relationships to good mental well-being. Therewere additional pathways identified. Access to the healthand social services would lead to Crisis Intervention andEviction Prevention programs which would, in turn,impact directly on good mental well-being. The otherresources (e.g., community centres) would lead to commu-nity action which would have a direct impact on goodmental well-being. Fig. 7 presents the pathways within thecluster Neighbourhood Support for Each Other asconceptualized by another group. Here the items withinthe cluster were arranged according to their negative orpositive influences on good mental well-being. Theinfluence from negative to positive was seen to lie on acontinuum as indicated in the diagram.

Discussion

Concept mapping was used in our study to gain anunderstanding of the ways in which neighbourhoods affectthe mental well-being of its residents. The concept mappingprocess is advantageous in that it yields an interpretableconceptual framework about various neighbourhooddomains important to consider for mental well-being,expresses this framework entirely in the language of theparticipants as the generation and interpretation of thedata are undertaken by the participants, and illustratesall major ideas and their interrelationships in the finalconcept mapping products. While concept mapping is

ARTICLE IN PRESS

Good MWB Downtown Good MWB Inner Suburbs

4

2

4

2

Necessary human and social servicesGreen area and natural environment

Neighborhood support for each other

Social makeup of people in the neighborhoodNeighborhood affordability

Necessary human and social services

Green area and natural environmentNeighborhood support for each other

Social makeup of people in the neighborhood

Neighborhood affordability

33Negative community factors

Negative community factors

r = 0.89

Fig. 4. Pattern match for good MBW: downtown versus inner suburban areas.

Violence

POOR MENTAL HEALTH & WELL BEING

SubstanceAbuse Drug

Dealing

Gun Shorts

Crimes

Prostitution

Vandalism

Graffiti

GangSymbols

Noise Pests

Smells

Fig. 5. Diagram of the pathways for the negative community factors cluster.

P. O’Campo et al. / Health & Place 15 (2009) 56–6864

semi-quantitative the sample is small and more appropriatefor qualitative studies and our sampling strategy was notintended by ensure generalizability to the city as a whole.Yet, the concept mapping process and software has uniquestrengths such as the built in tools to facilitate the analysesof data that ensure that the generated maps are sound (e.g.,

bridging values and stress tests) and the long time spentwith participants, 6 or more hours, helps to ensure that thedata and findings well represent the ideas of the partici-pants.Compared to the handful of community characteristics

examined in relation to mental well-being found in the

ARTICLE IN PRESS

Good Public & Social Services

Access to:

M.H. Clinics

Med: Health

Rehab

Woman shelter

Homeless

Public Trans

Crisis Inter.

Evict. Prev.

Recreational Centre

Community Centre

Libraries

Churches

Program Adults

Employment TrainingAfter School

Program

Good Mental Well Being

Community Action

Fig. 6. Diagram of the pathways in the necessary human and social services cluster.

Sur

veill

ance

C

amer

as

Sense of

Community

Commun

icatio

n

betw

een

Reside

ntsNeighborhoodWatch

Nei

ghbo

rhoo

dC

hang

e

Discrim

ination

Fear of Helpingneighbors Neighborhood

stress

PolicePresence

Residents

being involved

in neighbor.

change

Negative Positive

Midway

Positive

Negative Positive

Mid-way

Positive

Negative

Positive

Fig. 7. Diagram of the pathways in the neighborhood support for each other cluster.

P. O’Campo et al. / Health & Place 15 (2009) 56–68 65

literature, our participants generated over 100 neighbour-hood characteristics that they identified as being related tomental well-being in either a positive or negative way.Some of these items could have been predicted a priori asthey have been demonstrated in the literature as beingimportant for poor mental well-being (Galea et al., 2005;Muntaner et al., 2004; Goldsmith et al., 1998; Ross,2000a, b; Silver et al., 2002; Ziersch et al., 2005; Wain-wright and Surtees, 2003; Steptoe and Feldman, 2001;Subramanian et al., 2004; Takano et al., 2002). Such itemsincluded those related to poverty (e.g., unemployment and

high poverty) as well as neighbourhood problems (e.g.,violence, substance abuse, vandalism, crimes, and gun-shots). Past research has identified ‘‘neighbourhood pro-blems’’ and physical disorder as being related to adversemental health (Galea et al., 2005; Steptoe and Feldman,2001; Kruger and Reischl, 2007). Similarly, past researchhas investigated the role of social cohesion or social capitalon mental well-being (Fone et al., 2007; Kruger andReischl, 2007). Our participants reported that some of themost important items for good mental well-being includedsimilar items such as friendliness of neighbours, sense of

ARTICLE IN PRESSP. O’Campo et al. / Health & Place 15 (2009) 56–6866

community, interaction between neighbours, neighbour-hood governance, and residents being involved in neigh-bourhood change. Our participants also noted that greentrees, bike paths, parks, and walkable areas were stronglyrelated to good mental well-being. These findings areconsistent with existing studies illustrating that green areashave positive effects on both physical and mental health orthat absence of well-maintained green areas contribute tostress and negative mental well-being (Guite et al., 2006;Dilani, 2001; Kaplan and Kaplan, 1989; Nielsen andHansen, 2007).

Yet, some items that our participants reported as beingstrongly related to good or poor mental well-being werenot anticipated based upon prior literature on neighbour-hoods and mental health. Our participants reported thathealth and social services such as community centres, goodpublic transportation, recreation centres, day care foradults and children, affordable and well-maintained hous-ing, employment training or placement programs, grocerystores, schools, and crisis intervention programs were allhighly important for good mental well-being. While thishas been examined to a small extent in the existingliterature, it was found not to be strongly related todepression among the elderly (Kubzansky et al., 2005). Notonly have too few quantitative studies examined thesefactors to date to be able to make sound comparisons toour findings, but the services that our participantsidentified were far more comprehensive than those thathave been examined in the existing literature on mentalhealth problems and that may be one reason for thedifference in findings.

The majority of the current literature focuses ondeterminants of adverse mental issues (e.g., depression,substance use) and does not focus on promoting ormaintaining good mental well-being. Our data suggest thatthere are differences in those neighbourhood factors thatpromote good mental well-being versus those that con-tribute to poor mental well-being. For poor mental well-being, our data suggest that the negative communityfactors cluster was strongly related to poor mental well-being but not to good mental well-being. Other clustersthat showed large differences in their relationship to goodversus poor mental well-being included green areas andnatural environment, necessary human and social servicesand neighbourhoood support for each other. Thesefindings might even be used to generate hypotheses thatmight be tested in future comprehensive quantitativestudies on mental well-being which might include measuresof both good and poor mental well-being to examineshared of divergent pathways between neighbourhoodfactors and health outcomes. In the area of research,future studies might contrast the relationships betweenneighbourhood factors important for good versus poormental well-being. In particular, our data suggest thatneighbourhood distress and neighbourhood problems willbe important for poor mental well-being but not for goodmental well-being. Similarly, factors such as parks and

green neighbourhood areas would be important for goodmental well-being but less so for poor mental well-being.One concern expressed in the literature is that those who

have high levels of depressive symptoms might view andcharacterize their environments more negatively comparedto those who do not have high levels of depressivesymptoms. In our study, we found no difference in thereporting of the characteristics that are important formental well-being for those with and without high levels ofdepressive symptoms as measured by the CES-D.Our findings might provide information on avenues to

pursue for those program designers interested in targetingneighbourhood level factors. Our finding suggest that forprograms concerned about reducing high levels of mentalhealth problems among residents, focusing on eliminatingneighbourhood problems and physical and social disorderwould be a priority. If the neighbourhood has fewproblems with mental health problems of its residents thenfocusing on facilities and resources to promote supportamongst neighbours or ensuring that communities haveample green space, bike paths and walkable areas would bea priority. Finally, our data suggest that such approachescan be used equally in dense downtown areas as well asneighbourhoods that are less crowded and located in theinner suburbs.A unique set of information that we gathered concerned

perceptions about the pathways by which the neighbour-hood level items and clusters are related to good and poormental well-being. While discussion about these pathwaysoccurred throughout the whole process of concept map-ping—from brainstorming to diagramming groups—wewere able to capture their ideas about these pathways in thediagramming sessions. These diagrams drew not only fromthe concept mapping results that were shared withparticipants but also from their own daily lived experi-ences. These pathways can be used in several ways. First,they can be used to confirm, or challenge, the concept mapfindings such as the content of the clusters or even therating information. The participants primarily relied oncontent cluster information for the diagramming sessionsand not the rating information. Yet, many of the diagramsseemed to mirror the rating data. For example, those itemsgiven higher ratings tended to appear in the diagrams as‘direct’ arrows to good or poor mental well-being whilethose with lower ratings tended to have less direct arrowsto the outcomes. The diagramming data can also be used togenerate hypotheses about associations and pathways thatmight be tested in larger, comprehensive quantitativestudies on mental well-being. Finally, these diagrams mightsuggest pathways of intervention that might be consideredin the design of neighbourhood level programs to promotegood or prevent poor mental well-being.Concept mapping is a useful tool to explore a group’s

ideas about a topic. We illustrated here how conceptmapping can be used to generate information abouthow neighbourhood factors are related to mental well-being. The advantages of concept mapping are numerous

ARTICLE IN PRESSP. O’Campo et al. / Health & Place 15 (2009) 56–68 67

including the structured process by which we gathered theinformation; the equal contribution of each persons’sorting and rating data to the final concept map; and thecontributions and input of participants to the analysis andinterpretation of the data to ensure that the findingsrepresent their own ideas and perspectives. The wholeprocess takes a few weeks which makes concept mapping auseful tool to increase access to the perspective of thosewith lived experiences into research studies.

Acknowledgements

This study was funded by Centre for Urban HealthInitiatives at the University of Toronto.

We would like to thank the Wellesley Institute and theScarborough Housing Help Centre for allowing us to holdgroups there. We would like to thank the SoutheasternToronto Project (SETo) for assistance with recruitment.

We would also like to thank the participants who weregracious enough to allow us their time.

We gratefully acknowledge the Ontario Ministry ofHealth and Long-Term Care in providing support to thelead author. The views expressed in this publication are theviews of the authors and do not necessarily reflect the viewsof the Ontario Ministry of Health and Long-Term Care.

References

Bijl, R., de Graaf, R., Hiripi, E., et al., 2003. The prevalence of treated and

untreated mental disorders in five countries. Health Affairs 22 (3),

122–133.

Burke, J., O’Campo, P., Peak, G., et al., 2005. An introduction to concept

mapping as a participatory public health research methodology.

Qualitative Health Research Methods 15 (10), 1392–1410.

Burke, J., O’Campo, P., Peak, G., 2006. Neighborhood influences and

intimate partner violence: does demographic setting matter. Journal of

Urban Health 83 (2), 182–194.

Dilani, A., 2001. Design and Health. The Therapeutic Benefits of Design.

Svensk Byggtjanst, Stockholm.

Dohrenwend, B., Dohrenwend, B., 1971. Social status and psychological

disorder: a causal inquiry. American Sociological Review 36 (2), 384–385.

Fone, D., Dunstan, F., Lloyd, K., et al., 2007. Does social cohesion

modify the association between area income deprivation and mental

health? A multilevel anlaysis. International Journal of Epidemiology

36, 338–345.

Galea, S., Ahmed, J., Rudenstine, S., et al., 2005. Urban built

environment and depression: a multilevel analysis. Journal of

Epidemiology and Community Health 59, 822–827.

Goldsmith, H., Holzer, C.r., Manderscheild, R., 1998. Neighborhood

characteristics and mental illness. Evaluation Program Plan 21,

211–225.

Guite, H., Clark, C., Ackrill, G., 2006. The impact of the physical and

urban environment on mental well-being. Public Health 120 (12),

1117–1126.

Kane, M., Trochim, W., 2006. Concept Mapping for Planning and

Evaluation. Sage Publications Inc., Thousand Oaks, CA.

Kaplan, R., Kaplan, S., 1989. The Experience of Nature. A Psychological

Perspective. Cambridge University Press, Cambridge, USA.

Kessler, R., 1979. A strategy for studying differential vulnerability to the

psychological consequences of stress. Journal of Health and Social

Behavior 20 (2), 100–108.

Kirby, J., Kaneda, T., 2005. Neighborhood socioeconomic disadvantage

and access to health care. Journal of Health and Social Behavior 46

(1), 15–31.

Kruger, D., Reischl, T., 2007. Neighborhood social conditions mediate the

association between physical deterioration and mental health. Amer-

ican Journal of Community Psychology 40, 261–271.

Kruksal, J., Wish, M., 1978. Multidimensional Scaling. Sage, Beverly

Hills, CA.

Kubzansky, L., Subramanian, S., Kawachi, I., et al., 2005. Neighborhood

contextual influences on depressive symptoms in the elderly. American

Journal of Epidemiology 162 (3), 253–260.

Lindstrom, M., 2004. Social capital: the miniturisation of community and

self-reported global and psychological health. Social Science and

Medicine 5, 595–607.

Mirowsky, J., Ross, C.E., 1990. Control or defense? Depression and the

sense of control over good and bad outcomes. Journal of Health and

Social Behavior 31 (1), 71–86.

Muntaner, C., Eaton, W.W., Miech, R., et al., 2004. Socioeconomic

position and major mental disorders. Epidemiologic Reviews 26,

53–62.

Nielsen, T., Hansen, K., 2007. Do green areas affect health? Results from

a Danish survey on the use of green areas and health indicators. Health

and Place 13, 839–850.

Ostir, G., Eschbach, K., Markides, K., 2003. Neighborhood composition

and depressive symptoms among older Mexican Americans. Journal of

Epidemiology and Community Health 57, 987–992.

O’Campo, P., 2003. Invited commentary: advancing theory and methods

for multilevel models of residential neighborhoods and health.

American Journal of Epidemiology 157 (1), 9–13.

O’Campo, P., Gielen, A., Faden, R., et al., 1995. Violence by male

partners against women during the childbearing year: a contextual

analysis. American Journal of Public Health 85 (8), 1092–1097.

O’Campo, P., Burke, J., Peak, G., et al., 2005. Uncovering the

neighborhood influences on intimate partner violence using concept

mapping. Journal of Epidemiology and Community Health 59 (7),

603–608.

Radloff, L.S., 1977. The CES-D scale: a self report depression scale for

research in the general population. Applied Psychological Measure-

ment 1, 385–401.

Rajaratnam, J.K., Burke, J.G., O’Campo, P., 2006. Maternal and child

health and neighborhood context: the selection and construction of

area-level variables. Health and Place 12 (4), 547–556.

Ross, C., 2000a. Neighborhood disadvantage and adult depression.

Journal of Health and Social Behavior 41 (2), 177–187.

Ross, C., 2000b. Neighorhood disadvantage and adult depression. Journal

of Health and Social Behavior 41, 177–187.

Ross, C., Reynolds, J., Geis, K., 2000. The contingent meaning of

neighborhood stability for residents’ psychological well-being. Amer-

ican Sociological Review 65, 581–597.

Silver, E., Mulvey, E., Swanson, J., 2002. Neighborhood structural

characteristics and mental disorder: Faris and Dunham revisited.

Social Science and Medicine 55, 1457–1470.

Statistics Canada, 2002. Canadian Community Health Survey: Mental

Health and Well-Being. Statistics Canada, Ottawa, Canada.

Steptoe, A., Feldman, P., 2001. Neighborhood problems as sources of

chronic stress: development of a measure of neighborhood problems,

and associations with socioeconomic status and health. Annals of

Behavioral Medicine 23 (3), 177–185.

Subramanian, S., Kim, D., Kawachi, I., 2004. Covariation in the socio-

economic determinants of self rated health and happiness: a multivariate

multilevel analysis of individuals and communities in the USA. Journal of

Epidemiology and Community Health 59, 664–669.

Takano, T., Nakamura, K., Watanabe, M., 2002. Urban residential

environments and senior citizens longevity in megacity areas: the

importance of walkable green spaces. Journal of Epidemiology and

Community Health 56, 913–918.

Trochim, W., 1989. An introduction to concept mapping for planning and

evaluation. Evaluation and Program Planning 12, 1–16.

ARTICLE IN PRESSP. O’Campo et al. / Health & Place 15 (2009) 56–6868

Trochim, W.M., Stillman, F.A., Clark, P.I., et al., 2003. Development of a

model of the tobacco industry’s interference with tobacco control

programmes. Tobacco Control 12 (2), 140–147.

Veenstra, G., 2004. Location, location, location: contextual and composi-

tional. Social Science and Medicine 60, 2059–2071.

Wainwright, N., Surtees, P., 2003. Places, people and their physical and

mental functional health. Journal of Epidemiology and Community

Health 58, 333–339.

Warr, D., Tacticos, T., Kelaher, M., Klein, H., 2007. Money, stress, jobs:

residents’ perceptions of health-impairing factors in ‘poor’ neighbour-

hoods. Health and Place 13, 743–756.

Wu, Z., Noh, S., Kaspar, V., et al., 2003. Race, ethnicity, and depression in

Canadian society. Journal of Health and Social Behavior 44, 426–441.

Ziersch, A., Baum, F., MacDougall, C., et al., 2005. Neighbourhood life

and social capital: the implications for health. Social Science and

Medicine 60, 71–86.