Professional Documents
Culture Documents
SENSE OF COMMUNITY
IN NEIGHBORHOODS
AS A MULTI-LEVEL CONSTRUCT
Sharon Kingston
University of Rhode Island
Roger Mitchell
Brown University
Paul Florin and John Stevenson
University of Rhode Island
■
Correspondence to: Sharon Kingston, 3407 Fort Independence St., Bronx, NY 10463. E-mail:
PreEvalSK@aol.com
social problems (Coie & Jacobs, 1993; Jason & Kobayashi, 1995; Pretty, Andrews, & Col-
lett, 1994), there has also been debate as to whether late-20th century urban neighbor-
hoods still provide a sense of community. Several social scientists have characterized ur-
ban dwellers as having “portable personal communities” made up of social networks
which are detached from any specific locality (Crump, 1977 as cited in Glynn, 1986; Fis-
cher, 1976, 1982; Nisbit, 1962 as cited in Glynn, 1986; Wellman, 1988). Such factors as
increased urbanization, the growth of centralized bureaucratic and governmental struc-
tures and increasing imbalance between local and centralized structures have been im-
plicated in the “decline” of neighborhood community (Glynn, 1986). This debate has im-
portant intervention implications, however, because there are compelling arguments
that the design of one’s intervention should depend upon the level of neighborhood
identity and social interaction present in the neighborhood or community (Israel, Check-
oway, Schulz, & Zimmerman, 1994). For example, Warren (1978) found that resident use
of formal and informal resources was clearly linked to neighborhood type, suggesting
that capacity-building work in neighborhoods with low-levels of neighborhood identity,
and social interaction required different kinds of intervention strategies.
However, when an aggregation of residents’ responses is used to infer a neighbor-
hood-level variable, one must be careful that one is measuring a meaningful neighbor-
hood-level construct. One approach is to use cross-level analysis techniques that look at
the degree to which there is consensus among residents about neighborhood conditions
such as the degree of neighboring, and feeling of membership. In a study of block-group
residents, for example, Coultin, Korbin, and Su (1996) found a significant degree of sim-
ilarity in ratings of neighborhood quality, but not on ratings of social interaction among
block-group residents. Although this measure of neighborhood interaction showed high
reliability for individual-level responses (Cronbach’s alpha 5 .85), the disagreement
among residents of the same block group, relative to the total variation in the sample,
was remarkably high, suggesting the inappropriateness of aggregating individual per-
ceptions in this case to derive neighborhood-level measures. The authors conclude “It
may be more appropriate to think of neighborhoods as presenting opportunities for in-
teraction as a function of their spatial qualities, their institutions, and make these the fo-
cus of measurement in the future” (p. 25). Thus, further insight into whether elements
of sense of community actually represent a neighborhood-level construct can be exam-
ined through analyses such as intraclass correlations, which indicate the degree of asso-
ciation between scores of individuals within a group (in this case, residents of a particu-
lar neighborhood) on variables of interest. In addition, some programs can look at how
relationships are changed by statistically removing the effects of individual level covari-
ates (Kenny, 1985).
CORRELATES OF NEIGHBORHOOD-LEVEL
SENSE OF COMMUNITY
Research on the predictors of neighborhood sense of community has focused to a large
extent on individual-level predictors of the individual-level sense of community (e.g.,
Davidson, Cotter, & Stovall, 1991) with a few exceptions (e.g., Sagy, Stern, & Krakover,
1996; Sampson, Raudenbush, & Earls, 1997). Buckner (1988), however, suggests that fu-
ture research focus on determining the extent to which sense of community is influ-
enced by environmental as well as individual variables to determine the degree to which
sense of community can be influenced by system-level variables (Buckner, 1988). In this
684 • Journal of Community Psychology, November 1999
study, we will be examining the effects of individual-level variables (i.e., income and ed-
ucation) as well as environmental aspects of the neighborhood (e.g., physical-space at-
tributes, presence of neighborhood-focused community organizations) on neighbor-
hood-level sense of community. Knowledge of the factors that are correlated with sense
of community is an initial step toward determining what factors lead to the development
of a sense of community and to the development of interventions which might increase
a sense of community.
unteer efforts have an effect. The presence of a voluntary group dedicated to fostering
positive conditions specifically in the neighborhood could conceivably lead to a greater
sense of community among residents.
In summary then, the purpose of this research is to use existing data to begin to ex-
plore the presence of elements of a sense of community as neighborhood-level con-
structs, and if present, to explore whether these constructs are associated with individ-
ual- and neighborhood-level variables. This represents an attempt to extend previous
work in sense of community into cross-level analysis. An underlying goal of cross-level re-
search is to develop theories and research designs that allow the researcher to consider
issues of level without incorrectly specifying the level at which phenomena occur
(Rousseau, 1985). Shinn (1988) has proposed that this type of multi- or cross-level analy-
sis should comprise the majority of community psychology research. Investigating the re-
lationships between individual characteristics and extra-individual environments not only
allows community psychologists to better understand the phenomena they are studying
but allows psychologists to determine how these relationships can be used to impact on
well-being (Shinn, 1988).
METHOD
Design
Secondary analyses were performed on data collected as part of an evaluation of a 5-year
community-based demonstration project designed to reduce alcohol and other drug
abuse in a Northeastern city. A random sample of residents in each of 21 neighborhood
areas were contacted through a random-digit dialing procedure, and asked to participate
in a structured interview regarding perceptions of neighborhood conditions. Appropri-
ate respondents were identified as individuals age 18 or older who had lived in their com-
munity for at least a year.
Residents of a Northeastern city (2,409; at least 100 residents from each of 21 dif-
ferent neighborhoods in the city) were interviewed by telephone during the months of
October, November, and December, 1993. The neighborhood designations used in the
investigation generally matched the official neighborhood boundaries used for city plan-
ning purposes (n 5 25 neighborhoods), except in four cases where neighborhood areas
were combined for project intervention purposes.
Participants
Households (4,757) were randomly chosen and contacted; of those, 2,409 households
completed the interview and 2,348 refused to participate, could not participate due to
hearing difficulties, or terminated the conversation in the midst of the interview. This
constitutes a participation rate of 50.6%. It is unknown if there is any systematic bias
among individuals who chose not to participate in the survey.
Almost two-thirds (63.2%) of the participants were women and 36.4% were men.
Over one-third (34.6%) reported that at least one person under the age of 18 was living
in their household. The mean age of participants was 46 years and the median age of
participants was 43 years old. One fifth of the participants (20.4%) had not graduated
from high school, 31.4% were high school graduates, 20.4% had completed “some col-
lege” or trade school, 16.9% had completed college and 10.8% had at least some post-
686 • Journal of Community Psychology, November 1999
college education. As for employment rate, 42.7% of participants were employed full-
time, 10.9% were employed part-time, 18.6% were unemployed, 23.9% were retired, and
3.6% were disabled. When compared to the 1990 census, unemployed, disabled, and re-
tired individuals are over-represented. This is likely due to the fact that people who are
not working were more likely to be home when the survey interviewers called.
The median annual income range of participants was $15,000 to $24,999. The tele-
phone survey underrepresented individuals in the higher income ranges. This under-
representation of higher income households may be due to the large number of retired
and unemployed individuals interviewed for the survey.
The sample was comprised of 11.3% African Americans, 3% Asians, 5.2% Latinos,
1.4% Native Americans, 72.8% White, and 4.7% of individuals who classified themselves
as “Other.” A small minority (1.6%) of participants refused to respond to this question.
This telephone survey under-represents minorities. According to the 1990 census,
African Americans represented approximately 14% of the population in this city, Asians
represented approximately 6% and Latinos represented approximately 16% (Bureau of
the Census, 1990). The under-representation of minorities may be explained by several
factors: a) members of minority groups live in larger households than Whites, b) mi-
norities tend to be over-represented among low-income groups and low-income groups
tend to participate in telephone surveys at a lower rate than middle-income groups, and
c) members of certain groups such as Latinos and Asians may have been unable to par-
ticipate due to limited English skills.
Measure s
Sense of community has been operationalized in a variety of ways (Bishop, Chertok, & Ja-
son, 1997; Buckner, 1988; Davidson & Cotter, 1991). Buckner’s (1988) neighborhood co-
hesion instrument tries to synthesize elements of psychological sense of community, at-
traction to neighborhood, and social interaction. McMillan and Chavis’ (1986) definition
of a sense of community involves four critical aspects: a) Membership: a feeling of be-
longing or of sharing a sense of potential relatedness, b) Influence: a sense of mattering
and making a difference to a group, c) Integration and Fulfillment of Needs: feeling that
members needs will be met by the resources received through membership in the group,
and d) Shared Emotional Connection: the commitment and belief that members have
shared and will share history, common places, time together, and similar experiences.
The scales used in this study capture several elements of a sense of community: a)
perceptions of neighborhood climate (i.e., a feeling of membership); b) perceptions of
residents’ control over neighborhood conditions (i.e., influence); and c) neighboring
behavior (i.e., social interaction). In addition, a measure of participation in community
organizations was included. Overall, the four scales used in this investigation provide
measures of neighborhood-related attitudes and neighborhood-related behaviors.
The association is known to at least one department of the city government that deals
with grassroots neighborhood groups.
The association was known to at least 10% of neighborhood residents who partici-
pated in the interview that was used in this study.
The association is a multipurpose organization (not just a crime watch) concerned
with local neighborhood conditions.
The association is responsible for a geographic area.
The association is composed primarily of volunteers (has fewer than five paid-staff
members).
The association is small scale (has fewer than 200 active members and an annual
budget of less than $100,000).
The criteria for a neighborhood association were adapted from the Block Booster Proj-
ect in New York City in 1987 (Chavis, Florin, Rich, & Wandersman, 1987).
The physical aspects of neighborhoods that were of interest in this study are taken
from the plan for urban neighborhoods developed by Clarence Perry in 1929 (Hallman,
1984; Sale, 1980). Perry attempted to design urban neighborhoods that would foster a
sense of community. Perry’s design included the following major points:
Analyses
The Levels Program developed by David Kenny was used to perform the data analysis
(Kenny, 1985). This program performs intraclass correlations (ICCs) for each variable of
interest. The ICC is evaluated for significant difference from zero with an F-test pro-
viding the test for nonindependence. A significant intraclass correlation indicates a sig-
nificant degree of association between scores of individuals within a group (in this case,
residents of a particular neighborhood) on variables of interest. The ICC may be inter-
preted as the variance accounted for by group membership. For example, for a variable
with an ICC of .25, group membership would be interpreted as accounting for 25% of
the variance in scores on that variable. The program is also capable of statistically re-
moving covariates such as income and education, and calculates correlation matrices of
the variables using both scores that have been corrected for variance attributable to the
covariates and uncorrected scores (Kerman, 1993).
To investigate whether individuals residing in the same neighborhood have similar
scores on neighborhood-related attitudes and neighborhood-related behaviors, intraclass
correlations were calculated for these variables using neighborhood residence as the in-
dependent variable.
The intraclass correlations were recalculated after controlling for income and edu-
cation to determine if shared variance attributable to neighborhood residence would re-
main once the effect of personal resources had been removed.
Further analyses were performed to investigate which neighborhood conditions cor-
related with neighborhood-related attitudes and behaviors. The relationship between
neighborhood-related attitudes and neighborhood-related behaviors with the presence
of a neighborhood association was investigated by performing point-biserial correlations
using the presence or absence of a neighborhood association in a neighborhood as the
independent variable, and scores on the two neighborhood-related attitude and two
neighborhood related behavior scales as dependent variables.
Intraclass F ratio
Variable Correlation (df 20,1660)
Intraclass F ratio
Variable Correlation (df 20,1660)
RESULTS
Intraclass correlations were used to test for group-level variance among residents of the
same neighborhood for each of the neighborhood-related variables: perceptions of
neighborhood climate, perceptions of residents’ control over neighborhood conditions,
self-reported neighboring behavior, and self-reported participation in community orga-
nizations. Intraclass correlations were also performed on the two measures of individual
resources used as covariates: income and education. Residents of the same neighborhood
were more similar to one another than to residents of other neighborhoods on all six
variables. Intraclass correlations ranged from .009 for participation in community orga-
nizations to .123 for education. The amount of variance accounted for in the neighbor-
hood-related variables by neighborhood residence was small. Table 1 provides the intra-
class correlation and F-ratio for each of the variables. All of the intraclass correlations
were significant at an alpha level of .05.
Given the existence of these constructs at the neighborhood level, to what extent are
they associated with individual-level resources (i.e., income and education) and neigh-
borhood-level variables (i.e., physical space, community-based organizations)? When the
variance attributable to the personal resources of income and education were removed,
intraclass correlations for neighborhood climate and perception of residents’ control
over neighborhood conditions were significant at an alpha level of .05. Self-reported
neighboring behavior and self-reported participation in community organizations were
no more similar among residents of the same neighborhood than to residents of a dif-
ferent neighborhood. Table 2 presents the intraclass correlation and an F-ratio for each
of the variables. Thus, neighborhood climate and control remained detectable as neigh-
borhood-level constructs even after controlling for the effects of individual-level re-
sources.
Is the presence of a small-scale neighborhood organization dedicated to improving
neighborhood conditions associated with greater neighborhood-level sense of commu-
nity? Six of the twenty-one neighborhoods were rated as having a neighborhood associ-
ation meeting the criteria described in the results. Point-biserial correlations were per-
690 • Journal of Community Psychology, November 1999
DISCUSSION
Neighborhood-based interventions that seek to increase a sense of community as a cat-
alyst for citizen involvement in addressing social problems will be more likely to succeed
if they examine the robustness of the construct in their particular settings, as well as the
full range of individual and environmental factors that might contribute to its develop-
ment. In this study, we have illustrated how data on community residents can be exam-
ined to look for neighborhood-level effects. Indeed, the results of this study support the
idea that urban neighborhoods may act as a locus of at least some aspects of a sense of
community. Ratings of neighborhood climate, perceptions of control over neighborhood
conditions, neighboring behavior, and participation in community organizations were
more similar among people living in the same neighborhood than among people living
in different neighborhoods. Such neighborhoods did, indeed, exhibit distinct and co-
herent profiles regarding the level of neighboring, and the level of control over neigh-
borhood conditions that residents perceived.
What contributes to the emergence of distinct climates in different neighborhoods?
At the environmental level, the possibility of finding significant relationships may have
been hampered by the sensitivity of the measurement instruments used here, as well as
by the level of variance in this sample. For example, neighborhood-related attitudes and
behaviors were not significantly correlated with the presence of a neighborhood associ-
ation dedicated to improving neighborhood conditions. This finding may be due, in
part, to the fact that the definition of voluntary organizations used in this study was
rather narrowly construed. Only relatively small-scale organizations dedicated specifical-
ly to improving conditions within a neighborhood were included. It appeared plausible
that there would be a relationship between neighborhood-related attitudes and behav-
iors and organizations explicitly committed to fostering desirable neighborhood condi-
tions. Such organizations might increase the commitment of volunteers to be attentive
to and interested in neighborhood issues, strengthen the capacity of individuals to ad-
Neighborhood-Level Sense of Community • 691
dress problems in their informal neighborhood networks as a result of the skills they have
developed as volunteers, and increase the likelihood that outside resources might be
made available to address local issues. However, a broader array of voluntary organiza-
tions might be examined to explore the impact of community-based organizations on
neighborhood sense of community.
Neighborhood-related attitudes and behaviors were not significantly correlated with
one physical characteristic (arterial streets which bound the neighborhood) hypothe-
sized to foster a sense of community. However, when considering these results it is im-
portant to note that the neighborhoods in the city used in this investigation are very sim-
ilar in the physical characteristics included in this investigation. More complex ratings
of physical characteristics have been developed to assess block characteristics relevant to
fear of crime (Perkins & Taylor, 1996), and such work might be extended to examine
sense of community.
Despite the results of the present investigation, the relationship between physical
conditions and neighborhood-related attitudes and behaviors is a promising topic for
community psychology research. Presently, a number of architectural and urban plan-
ning projects are underway based on the assumption that certain physical characteristics
can be used to foster a sense of community. This approach is sometimes referred to as
the “new town” movement in urban planning literature (Plas & Lewis, 1996). In a qual-
itative study of Seaside, Florida, one of these “new towns,” a majority of residents spon-
taneously referred to one or more of the components of a sense of community (as defined
by McMillan and Chavis) in describing their middle-class resort community. Further,
many residents mentioned that the physical aspects of the town were at least partially re-
sponsible for engendering a sense of community. The presence of recreational spaces, a
town grocery store, and various town features designed to discourage automotive traffic
in the town were included in residents’ discussions of the community fostering physical
characteristics (Plas & Lewis, 1996).
How do individual-level variables contribute to the development of neighborhood
level constructs? The cross-level analyses used here allowed for statistically controlling for
the effects of individuals’ income and education. The intraclass correlations for neigh-
borhood attitudes (i.e., ratings of neighborhood climate and perception of control over
neighborhood problems), though small, remained significant after the effects of income
and education were removed. Such consensus among members of a neighborhood sug-
gests that there should be continued examination of environmental properties of the
neighborhood that might contribute to such climates.
However, regarding neighborhood-related behaviors (i.e., neighboring behavior,
participation in community organizations), the intraclass correlations were no longer sta-
tistically significant when income and education levels were statistically controlled. To
what extent do individual-level resources (i.e., income and education) influence either
opportunities or motivation for participation, or ratings of the quality of neighboring?
In examining several hundred Chicago neighborhoods, Sampson, Raudenbush, & Earls
(1997) found that individual-level SES, as well as a neighborhood-level measure of col-
lective disadvantage (i.e., an index of percentage below poverty level, degree of residen-
tial instability, etc.) was predictive of collective self-efficacy (i.e., a measure of social co-
hesion and informal social control). This suggests that there is the need to continue to
look at the degree to which sense of community is influenced at both the individual and
neighborhood level by the socioeconomic resources that are available. There is a need
to determine whether an absence of these resources (and perhaps, other measures of in-
692 • Journal of Community Psychology, November 1999
REFERENCES
Bannes, M., Bonauito, M., & Ercolani, A.P. (1991). Crowding and residential satisfaction in the ur-
ban environment: A contextual approach. Environment and Behavior, 23, 531–552.
Battistich, V., Solomon, D., Kim, D., Watson, M., & Schaps, S. (1995). American Educational Re-
search Journal, 32, 627– 628.
Bishop, P.B., Chertok F., & Jason, L.A. (1997). Measuring sense of community: Beyond local bound-
aries. American Journal of Community Psychology, 18, 193–212.
Neighborhood-Level Sense of Community • 693
Maton, K.I., & Salem, D.A. (1995). Organizational characteristics of empowering community set-
tings: A multiple case study approach. American Journal of Community Psychology, 23, 631–
676.
McMillan, D.W., & Chavis, D.M. (1986). Sense of community: A definition and theory. Journal of
Community Psychology, 14, 6–23.
Miller, F.D., Tsembergis, S., Malia, G.P., & Grega, D. (1980). Neighborhood satisfaction among ur-
ban dwellers. Journal of Social Issues, 36, 101–117.
Newman, O. (1972). Defensible space. New York: Macmillan.
Nisbit, R. (1962). Community and power. New York: Oxford University Press.
Plas, J., & Lewis, S. (1996). Environmental factors and sense of community in a planned town.
American Journal of Community Psychology, 24, 109–143.
Perkins, D., & Taylor, R. (1996). Ecological assessments of community disorder: Their relationship
to fear of crime and theoretical implications. American Journal of Community Psychology, 24,
63 –108.
Pretty, G.M.H., Andrews, L., & Collett, C. (1994). Exploring adolescents’ sense of community and
its relationship to loneliness. Journal of Community Psychology, 22, 346–358.
Rapoport, A. (1982). The meaning of the built environment: A nonverbal communication Ap-
proach. Newbury Park, CA: Sage.
Rousseau, D. (1985). Issues of level in organizational research: Multi-level and cross-level perspec-
tives. Research in Organizational Behavior, 7, 1– 37.
Royal, M.A., & Rossi, R.J. (1996). Individual-level correlates of sense of community: Findings from
workplace and school. Journal of Community Psychology, 24, 395–416.
Sale, K. (1980). Human scale. New York: Coward, McCann & Geoghegan.
Sagy, S., Stern, E., & Krakover, S. (1996). Macro- and microlevel factors related to sense of com-
munity: The case of temporary neighborhoods in Israel. American Journal of Community Psy-
chology, 24, 657– 676.
Sampson, R.J., Raudenbush, S.W., & Earls, F. (1997). Neighborhoods and violent crime: A multi-
level study of collective efficacy. Science, 277, 918–924.
Schwirian, K.P. & Schwirian, P.M. (1993). Neighboring, residential satisfaction, and psychological
well-being in urban elders. Journal of Community Psychology, 21, 285–299.
Shinn, M. (1988). Mixing and matching: Levels of conceptualization, measurement, and statistical
analysis in community research. In P. Tolan, C. Keys, F. Chertok, & L. Jason (Eds.), Research-
ing community psychology: Integrating theories and methods. Washington, DC: American Psy-
chological Association.
Unger, D.G., Wandersman A., & Hallman, W. (1992). Living near a hazardous waste facility: Cop-
ing with individual and family distress. American Journal of Community Psychology, 62, 55–70.
Warren, D. (1978). Explorations in neighborhood differentiation. Sociological Quarterly, 19, 310–
331.
Wellman, B. (1988). The community question re-evaluated. In M.P. Smith (Ed.), Power, commu-
nity and the city, comparative urban research. Vol. 1, p. 81–107. Transaction.
Weisenfeld, E. (1996). Individual-level correlates of sense of community: Findings from workplace
and school. Journal of Community Psychology, 24, 337–346.