What is neighbourhood?

An externality-space approach
by George C. Galster
1 Context
Ironically, while all those concerned with cities seem to understand the term
neighbourhood in its common usage, no one seems to be able to agree on exactly
what it means or how it should be spatially specified. Weseem to be conceptually
impaled on the horns of a dilemma. On the one hand, views of neighbourhood
grounded in individual cognition and collective sentiment and symbolism, though
perceptually meaningful, have had little operational content since they have not
been employed in the specification of precise, geographic neighbourhood boun-
daries. On the other hand, views of neighbourhood as defined by clear administra-
tive boundaries have had no necessary correspondence with the perceptual reality
of individuals in the given area.
Undoubtedly, there is a consensus that the neighbourhood is a social/spatial
unit of social organization . . . larger than a household and smaller than a city
(Hunter, 1979, 270). But here is where consensus ends. Academicians (especially
sociologists) have focused on questions of the social meaning of space: how far do
social interaction networks extend over space? What social support is drawn from
surrounding residents? What social symbolic significance does an area assume?
Numerous definitions can be found in the social science literature, varying in their emphases
and degree of ambiguity. A representative sampling follows:
Ecological: A place with physical and symbolic boundaries (Keller, 1968, 89). Place
and people, with the common sense limit as the area one can easily walk over (Morris
and Hess, 1975, 6). A physical or geographical entity with specific (subjective)
boundaries (Golab, 1982, 72). A distinct housing submarket (Ahlbrandt and Brophy,
1975.6).
Interactive: A social organization of a population residing in a geographically proximate
locale (Warren, 1981, 62).
Ecological interactive: A limited territory within a larger urban area, where people inhabit
dwellings and interact socially (Hallman, 1984, 13). Geographic units within which
certain social relationships exist (Downs, 1981, 15). Common named boundaries, more
than one institution identified with area, and more than one tie of shared public space or
social network (Schoenberg, 1979,69).
Perceptual: In the last analysis, each neighborhood is what the inhabitants think it is
(National Commission on Neighborhoods, 1979, 7).
244 What is neighbourhood? An externality-space approach
Unfortunately for the most part, equally important questions have to be begged:
do neighbourhoods as specified by unambiguous, consensual boundaries exist?3 If
so, what role does social meaning play in their establishment? Can they exist in the
absence of local social interaction? Lack of consideration of these issues is under-
standable, given that the emphasis on social meaning leads one exorably to the con-
clusion that in the last analysis, each neighborhood is what the inhabitants think it
is (National Commission on Neighborhoods, 1979,7).
An exception has been Suttles (1972, Chapter 3), who argued that the social
aspect and spatial aspect of neighbourhood are intrinsically interrelated, and that
particular social functions are associated with different spatial levels of neighbor-
hood4 As valuable as this insight is, shortcomings remain. First, it is unclear how
one moves from individual perceptions to collective representations of neighbour-
hood; i.e. at whatever spatial level, how do neighbourhood boundaries spawned by
one individuals perceptions coincide with those of other, proximate residents?
Does the aggregation of individual perceptions result in the establishment of
unanimously agreed upon boundaries which delinate a mutually exclusive and
exhaustive set of neighbourhoods over urban space? If not (as is likely): what
leads to variations in this degree of perceptual concidence? At what point does a
lack of perceptual coincidence render the existence of neighbourhood prob-
lematic? Second, there is no compelling a prion reason that a meaningful ecological
neighbourhood can only be defined in terms of social interaction, sentiment and
symbolism. Such dimensions indeed may prove to be sufficient for the unam-
biguous specification of neighbourhood boundaries, but not necessarily so. Third, a
spatial neighbourhood delineated solely by these social dimensions may be the
inappropriate scale when one is attempting to analyse investment or mobility
behaviours which lead to changes in the physical condition or demographic com-
position of a given area.6
Most planners and public officials (and residents?) are interested in exactly these
Exceptions are Firey (1945), Lynch (1960) and Hunter (1974).
3Wellman (1972) raises a similar challenge when he chastizes those who disregard the caveats of
researchers who studied unique urban villages and uncritically accepted the existence of
neighbourhood.
4Suttles built upon the seminal insights of J anowitz (1952) and Greer (1962). In his view, the
most elemental spatial unit was viewed as the block face, the area defined by where children
are allowed to play without supervision. The second level was labelled the defended neigh-
bourhood: the smallest area possessing a corporate identity as defined by mutual opposition to
another area. The third level, the community of limited liability, typically consisted of an
administrative district in which individuals social participation was selective and voluntary. The
highest geographic level of neighbourhood, the expanded community of limited liability, was
viewed as an entire sector of a city. Surveys conducted by Birch et al. (1979, Chapter 3) have
revealed that residents do, indeed, conceive of four distinct spatial levels of neighbourhood
which correspond closely to Suttless theoretical hierarchy.
See Keller (1968) and Hunter (1974) for evidence on the interresident variability of neigh-
bourhood boundary specification.
6For empirical analyses of the appropriate geographic scale of neighbourhood vis-d-vis such
behaviour, see Galster (forthcoming).
George C Galster 245
mobility and investment behaviours, of course. They care less about social meaning
than operational precision. Given their need to specify some precise definition of
neighbourhood for which objective data can be gathered and analysed, they have
resorted to establishing ad hoc geographic boundaries. The result has been a
precise spatial definition of neighbourhood - planning district, census tract, etc. -
but one which may have little or no perceptual or behavioural significance for
anyone besides the boundary maker^.^Yet, if ones goal is to explain why certain
changes occur in a particular area and to predict how that area is likely to respond
to, e.g. a policy which alters its physical, demographic or social dimensions, one
cannot blithely rely on neighbourhood defined by fiat.
2 Purpose
The purpose of this paper is to present a new conceptual definition of neighbour-
hoods that is realist,m i.e. is grounded in the perceptions of people who are living
and investing in them. Algorithms are developed whereby individual perceptions
can be aggregated and the differences in these perceptions can be quantified. It is
suggested how operational, geographic specifications of neighbourhood might be
made based on empirical investigations that follow from these definitions and
algorithms. The operational specifications, in turn, open up a host of new possi-
bilities for testing hypotheses of interest to theoreticians and policy makers alike.
In other words, an attempt will be made to link explicitly perceptual and spatial
dimensions in order to derive a concept of neighbourhood that is both quantifiable
and holds the promise of improving our understanding of behavioural response to
environmental changes.
The fundamental way in which this approach to defining neighbourhoods differs
from previous attempts is that it employs an inductive versus deductive approach.
That is, prior attempts to specify neighbourhood in a way that was perceptually
meaningful have foundered on their inability to deduce precise geographic neigh-
bourhood boundaries from the theoretical constructs. I believe such a deductive
approach is inherently fruitless. Rather, a meaningful theoretical conceptualization
of neighbourhood is herein viewed as laying the groundwork for empirical tests in
which the geographic boundaries of neighbourhoods themselves are viewed as
For a more detailed discussion of this dichotomy between the social versus geographic view of
neighbourhoods, see Keller (1968,126,133).
Taub el al . (1977) has also noted that the need to identify a particular neighbourhood associa-
tion as a contact agent for the administration of social programmes can lead to the artificial
creation of neighbourhoods.
90ne possible exception is the Census Bureaus Statistical Neighbourhood Program which,
according to personal correspondence with a planning professor at the University of Cincinnati,
has been successful in specifying precise yet meaningful boundaries for Cincinnati neighbour-
hoods. For a description of a similar effort in Pittsburgh, see Ahlbrandt, Charney and Cunning-
ham (1977).
DFor more on the distinction between realist and nominalist approaches to boundary
definition, see Laumann, Marsden and Prensky (1983,20-22).
246 What is neighbourhood? An externality-space approach
dependent variables. In this way, the question of the existence of such boundaries
is not begged, but rather becomes the subject of hypotheses which may be tested
inductively. And here is where previous conceptual views about neighbourhoods
come into play; for it is precisely such characteristics as local social interaction,
culture, sentiment and symbolism which potentially can affect individuals percep-
tions in an area so as to produce varying degrees of coincidence among their
ecological views of neighbourhood.
It must be noted at the outset that this paper is conceptual and not empirical.
Its purpose is to explicate new concepts of defining and measuring neigh-
bourhood and to demonstrate thier potential usefulness in investigating a wide
variety of issues. Specific suggestions are provided, however, for operational-
izing the theory. It is beyond the scope of the present paper to conduct such
investigations.
I Neighbourhood as externality space
1 Overview
The fundamental claim of this paper is that neighbourhood can be meaning-
fully and usefully specified in terms of extemoliq space. Externality space is de-
fined as the area over which environmental changes initiated by others are per-
ceived as altering the wellbeing (psychological or financial) a given individual
derives from the given location. Three characteristics of these externality spaces are:
a) congruence - the degree to which an individuals externality space corresponds
to predefined geographic boundaries;
b) generality - the degree to which an individuals externality spaces for different
types of externalities correspond;
c) accordance - the degree to which externality spaces for different individuals
in the same area correspond.
Neighbourhood is not viewed in this paper as a unidimensional, dichotomous
entity, but rather as a variable whose magnitude can be measured along interval
scales in each of these three dimensions. In other words, for any specific dimen-
sion of neighbourhood one does not consider here whether it exists or does not
exist, but rather the degree to which it exists over a given space.u
How the concept of neighbourhood is built upon the notion of an indi-
viduals externality space and how the three dimensions of neighbourhood may
be specified from this foundation are explained in detail in the next three sub-
sections.
This view of neighbourhood owes an intellectual debt to the seminal suggestions of two scholars.
Segal (1979, 6) claimed that neighbourhoods might be specified in terms of externalities, and
Warren (1972, Chapter 1) noted that a vital dimension of a community is the extent to which
service areas of local units coincide.
George C. Galster 247
2 The individual perspective
From the perspective of a particular individual, say a resident or owner of property
at a given urban location, the externality space is defined as the area over which
changes in the environment initiated by others are perceived as altering the degree
of wellbeing (psychological and/or financial) the individual derives from the given
location. Note first that this definition considers the space over which particular
changes are both perceivedand are adjudged non-trivial. In other words, it is the space
over which stimuli indicative of changes in the quality of the individuals residential
milieu provide the necessary (though not sufficient) conditions for behavioural
response: outmigration, alteration in home maintenance, dwelling sale, revisions in
amount, type or locus of social interaction, etc. Changes are herein specified as
resulting from the actions of others which are external to the given individual, i.e.
are not the result of direct market or social transactions between the individual and
the other person(s).n The focus is on changes which are exogenous to the indivi-
dual because the ultimate proposed use of this definition is as a tool for analyzing
and predicting individual behavioral responses to such.u All sorts of externalities
could be considered in principle. Examples would be: the construction of a public
playground on a vacant lot on the block, a person of a different race moves into the
house next door, several close friends die or move out of the neighbourhood.
The geographic boundaries of this individuals externality space cannot be more
precisely specified until one knows more about the externality(ies) under considera-
tion, the individuals beliefs and attitudes, the local social networks between the
individual and proximate others, and the spatial pattern of the individuals habitual
travels. In other words, the externality space is a function of how individuals
evaluate the particular externality, how their peers evaluate it, and whether they are
aware of its existence, either first-hand through direct observation or second-hand
through interpersonal communication.
Consider, initially, the externality of litter as an illustrative example. An indi-
vidual may not readily perceive or be concerned about the presence of litter on an
adjacent block, but more likely would if it were on the individuals own block. In
such a case the individuals externality space for litter would consist of the block-
face. Analogously, the externality space over which the individual is sensitive to the
immigration of households of lower socioeconomic status is probably more
expansive. Still larger may be the externality space for such externalities as new
nNote this usage is somewhat more general than the way in which the term externality is
conventionally employed in economics; see Schreiber and Clemmer (1982).
? h e approach of speafying neighbourhood in the context of individual behavioral responses
to external stimuli is consistent with the theory propounded by Franz (1982). He asserts that it
is only in these reactions to these disturbances that the boundaries of neighbourhoods become
empirically visible.
It has been suggested that individuals ability to produce clear mental maps of urban space
is enhanced by their frequency of travel through the area; see Lynch (1960), Jacobs (1961),
Mlligram el al . (1972).
248 What is neighbourhood? An externality-space approach
shopping centres or public facilities.5
Of course, exactly when some externality comes close enough to be perceived
and to be assessed as influencing ones residential wellbeing depends on the char-
acteristics of the observer. Elderly households and households with children may
be, for example, relatively more sensitive to the approach of hgher crime.
Similarly, white bigots would have a larger externality space vish-vis new black
residents than would more tolerant whites.
In the case of households and owner occupants, the externality space may also
be influenced by the social character of the surrounding area. Although the social
dimension of space may, itself, be the source of an externality, it also can influence
the likelihood of individuals in that space perceiving a given physical, demographic,
or social interactive externality and the way in which they evaluate its conse-
quences. For instance, if a person is embedded within a spatially-dense social
network (e.g. a traditional ethnic enclave), that person will be exposed to much
second-hand information about and interpretation of spatial events which otherwise
might not have been perceived through first-hand observation. The individuals
evaluation of perceived information may also depend on the acculturation which
has occurred in the local social milieu. As illustration, one who has resided for a
long time amid professional-status residents might grow more favourably-disposed
toward valuing everything in its place, i.e. rigid separation of land uses. This
individual might therefore specify a more spatially-extensive perceptual boundary
for non-residential land uses compared to a person inculcated with the values of
proximate working-class residents. These latter considerations may be included
under the rubric symbolic communities (Hunter, 1974).
Given an individual ( n) with particular characteristics who is embedded in a
particular spatial-social environment, two dimensions of his/her externality space(s)
can be specified. The first, individual congruence (C,,), is the degree to which
individual ns perceptual boundaries of the externality space (Yen) for a particular
externality ( e) correspond to some predetermined geographic boundaries defined
by streets, topographical features, or administrative fiat that delineate a space (X)
containing the property which the nth individual invests or resides in.& Formally,
the congruence between this individuals particular externality space and the given
area is specified as:
'Sit should not be inferred from the above that externality spaces necessarily have regular
shapes. The algorithms presented below make no assumptions in this regard. None of the
measures are unique, however; an infinite number of topological configurations could produce
the same value for C, G or A.
mNote that the definition of congruence used here is very different than that employed by
Michelson (1976, 26), which refers to states of variables in one system coexisting better with
states of variables in another system than with other alternative states.
Georpe C Galster 249
where n and U signify intersection and union, respectively, in set terminology.
Heuristically, X n Ye, represents the area of the region where person ns perceptual
map for the impact of externality e overlaps the map of the specified area; X U Y ,
represents the sum of the areas of X and Y. Visually, this measure is portrayed in
Figure 1 c. C,, ranges from a minimum of zero to a maximum of one.
The second dimension, generality, is the degree to which individual ns exter-
nality spaces (Y) correspond across E number of different externalities. Formally,
individual ns generality is:
Heuristically, GE, represents the summed ratios of overlapping to total areas of
externality spaces specified for a variety of potential externalities, for all possible
permutations of externality combinations excluding identities (e=fl . C, , ranges
from a minimum Of Zero to a maXimUm Of 2 e . Visually. This measure is par-
E- 1
z
e- 1
trayed in Figure 1 b.
3 The aggregate perspective
It is at the point where individual perceptions of neighbourhood are aggregated
over a given spatial grouping of individuals that the operational geographic meaning
of neighbourhood has been typically rendered ambiguous in previous studies. I
would suggest that this can be avoided by focusing on the question: to what degree
does neighbourhood exist in a given place? The following discusses how this focus
provides the basis for simultaneously defining and measuring neighbourhood at
the group level. The analysis revolves around aggregate measures of the two dimen-
sions introduced above - congruence and generality - and adds a third -
accordance.
As above, begin by taking as given some predetermined geographic area of a city
with clearly-specified cartographic boundaries: area X. This spatial set will demar-
cate a particular number ( I ) of residents who live in X, ( J ) owners of property in
X , and ( K ) others with financial interests in the property in X , such as real estate
brokers or financial institution officers. From this group all or some subset may be
selected as the basis for analysis. Let this group consist of N members.
Now the aggregate degree of congruence for the space X for given externality
e over N members of the given group is the summation of individual congruence as
defined in (1):
250 What is neighbourhood? An externality-space approach
0 ........
--1
I mm.0
I ---
I -
I-
I
I
0
0 ........
Key to hypothetical situation
Boundary for Person A, externality 1
Boundary for Person A, externality 2
Boundary for Person 8, externality 1
'Official' boundary
1 a 'Accordance'
For persons A, B; externality 1
Aggregate
Accordance =ratio of %area to total area
covered by combined boundaries
0
0
0
.........
George C Galster 251
0 0 0 0 0 0 0 0 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 0 0 0 0 0 0 0 0
1 b 'Generality'
For person A; externalities 1 and 2:
Individual
Generality =ratio of //,area to total area
covered by combined boundaries
Aggregate
Generality =sum of individual generalities
for persons A and B, externalities 1 and 2
0 0 0 0 0 0 0 0 0
0
0
0
0
0
0
l c 'Congruence' 0
For person A vs 'official' area; externality 1 :
Individual
Congruence =ratio of ;/area to total area
covered by combined boundaries
Aggregate 0
Congruence =sum of individual
congruences for persons A and B,
externality 1 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 0 0 0 0 0 0 * 0
Figure 1
fied example
Graphic illustrations of three dimensions of neighbourhood for simpli-
252 What is neighbourhood? An externality-space approach
where N =Z, J, K or combinations or subsets thereof. Thus, aggregate congruence
for a particular externality is specified in terms of a group sum of the ratios of
overlapping to total areas of X versus the members externality spaces. Maximum
aggregate congruence is obtained (Cm =N)when each group members perceptual
space for the given externality corresponds to the specified area; minimum con-
gruence is obtained (CeN =0) whea there is no correspondence whatever.
As will be indicated in a later section, it sometimes might be desirable to define the
aggregate congruence of an area only in terms of a subset of individuals and a single
e~ternal i ty. ~ If, however, one wishes to define a more aggregated level of con-
gruence, (3) can be easily expanded to include summations across all individuals in
area X and all E externality types:
where 0 C,, < E. Nand N =I + J + K
Next we can consider the second dimension of the aggregate neighbourhood -
generality - the aggregate degree to which the individuals externality spaces
coincide for E different externalities, as given by ( 2 ) summed over N individuals:
N E E
N =I, if residents; J , if owners; K , if others; or combinations of Z, J, K . Aggregate
generality is thus the total of the ratio of overlapping to total areas of all possible
externality space comparisons, summed over all group members in area X. The
generality of area X over a variety of E different externalities would be maximized
(GEN =2N ye) for a given group in X if each individual in the group perceived
the same boundaries for all externalities (although these boundaries need not
coincide fiom individual fo individua2). Generality would be minimized (GEN =0)
if each and every individuals externality space for any given externality did not
overlap with that for any other externality.
The final aggregate dimension is accordance: the degree to which all N indivi-
duals externality spaces for a given externality e overlap. It may be specified as:
c = 1
N may consist of a subset of residents, owners or brokers based on race, income, age, etc.
George C Galster 253
N - 1
where N = I, J, K or combinations thereof; 0 Q A , < 2 7 n. Accordance for a
L
n-1
given externality is thus the total of the ratios of overlapping to total areas of all
possible interpersonal comparisons of e externality space for N groups members in
area X . This is visually portrayed in Figure l a.
Accordance over all externalities is found by summingAeN over E:
N-1
Such accordance for area X would be maximized (AEN =2E n) if each member
L
n-1
in the group perceived the identical externality space for a given externality, and
this was true for every externality (although the spaces across externaZities need not
be identical). Accordance would be minimized (AEN =0) if no two individuals
externality spaces overlapped for a given externality, and this was true for all
externalities.
It should be noted at this point that CEN, C,, A , as specified are interval
measures of three distinct dimensions of the aggregate relationships between exter-
nality spaces existing for some predefined area. As such they are not comparable
in a cardinal sense. Certain logical connections do exist between them, of course.
For instance, AEN and G,, maximization is a necessary (but not sufficient) condi-
tion for C,, to be maximized. In other words, for everyone in an area to agree
that its predetermined boundaries accurately reflect their externality space for all
externalities, they must also agree that their externality space is the same across all
people and all externalities. The converse is not true, however; there may be
complete accordance and generality, but the common space thus specified may
have little congruence with the predetermined area. Through analogous logic it may
be demonstrated that A and G minimization is neither a necessary nor a sufficient
condition for CE, minimization. Finally, A , and C , serve as constraints on
the maximum CEN level attainable for a given spatial set X . Although no precise
mathematical relationship can be given at this level of generality, it is intuitively
clear that if, e.g. A,, and G, are low, CE, will tend to be also. That is, if there
is little overlap of externality spaces across various individuals and across various
externalities, there cant be much consistent overlap between these spaces and the
given area X .
4
At this point we can now discuss what neighbourhood means. Neighbourhood is
both defined and measured by its congruence, generality and accordance. Each
Congruence, generality, accordance and neighbourhood
254 What is neighbourhood? An externality-space approach
dimension may only be specified in terms of particular perceptual group(s) and
externality(ies) being perceived. Once having specified such a context, the neigh-
borhood associated with a given place or group of individuals can be considered in
terms of the degree to which it manifests congruence, generality, and accordance.
For a predetermined spatial set of individuals, should there be an area over which
accordance and generality (for particular exteralities) were high, it would imply
not only that a neighbourhood exists but its boundaries would be defined. If
accordance and generality were low we would conclude that no meaningful spatial
neighbourhood existed for this group. For a predetermined area (and the indivi-
duals implicitly contained therein), the degree to which neighborhood is mani-
fested there would be primarily measured by congruence, although as noted above
the degree of congruence possible can be influenced by accordance and generality.
If congruence were low we would conclude that the given boundaries did not pro-
vide good proxies for the neighbourhood (potentially) perceived by these indi-
viduals.
I1
The preceding theoretical analysis implies that the way to convert the concept of
neighbourhood into an operational measure useful for empirical research and
planning purposes is to specify boundaries in terms of how, in aggregate, people
map their externality spaces. If one is attempting to designate meaningful neigh-
bourhoods a priori, the goal would be to specify an area X such that a predeter-
mined form of congruence is maximized. Although it is beyond the scope of this
paper to undertake such an empirical investigation, an outline of a suggested pro-
cedure will be presented.
Depending on the ultimate analytical reason for the investigation, the research
would first need to choose one particular externality or subset of them, as well as
one or more particular groups of people to focus on; i.e. the particular dimensions
over which neighbourhood will be assessed must be specified. For example, one
might only want to explore behavioural responses of a certain type of household to
the proximate inmigration of persons of lower socioeconomic status. Or, at the
other extreme, for planning purposes one might be interested in the parameters of
a more general neighbourhood for a broad set of externalities that might be regu-
lated by zoning, as perceived by residents, owners, real estate brokers, and officers
of financial institutions alike.
Second, the research would need to ascertain how each individual in the chosen
sample perceived his or her externality space(s). At least two methods for doing this
come to mind. The first would involve the administration of carefully designed
opinion polls to reveal the distance at which respondents begin to perceive that
certain generators of externalities affect them. In this vein one could either query
about alternative hypothetical situations and likely perceptions or ascertain actual
perceptions of particular externalities and, after noting the specific locations of
Making the neighbourhood concept operational
George C. Galster 255
respondent and externality(ies), infer a perception-distance relationship. A
technique analogous to the former has already been extensively used to investigate
how whites would evaluate alternative proximities of black neighbours (Pettigrew,
1973; Farley et al., 1978). The potential weakness of this approach is that evalua-
tions of hypothetical externalities may not correspond to evaluations of real ones if
such are not cognitively perceived in a particular instance. The latter approach
above is superior in this regard since perceptions and evaluations of existent extern-
alities are being recorded. Some investigations along this line have been conducted
by Scotland and Galster (1977), which deal with externalities associated with deter-
iorated housing.s Of course, the shortcoming of this latter approach is that the
range of analysis is restricted to the set of externalities currently perceived (in
varying degrees) by the given sample.
The second method would be based on indirectly eliciting information about
individuals externality spaces which is implicit in their perceptions of property
values. More specifically, data could be gathered about the characteristics of indivi-
duals homes, their surrounding environment, and what they currently perceive
their property to be worth. A multiple regression of this self-assessed property value
on all these characteristics would yield estimates (in the form of regression coef-
ficients) of the relative contributions made by each characteristic. If variables were
carefully specified as proxies for the existence of a variety of externalities at various
distances from the given observation, the magnitude and statistical significance of
their coefficients would provide the desired information about the point at which
externalities begin to impinge. Several studies have employed this hedonic index
methodology to assess the spatial extent of externalities generated by, e.g. non-
residential and multifamily land uses (Grether and Mieszkowski, 1980), racial com-
position (Galster, 1982), and human service facilities (Gabriel and Wolch, 1984).
The conceptual difficulty with this approach are the assumptions that: a) all
relevant social, physical and demographic externalities are capitalized into property
values and b) self-assessed housing value is more a reflection of the respondents
own perceptions and preferences, and not their view on what the rest of the market
believes. The operational difficulty is the requirement of an extremely rich and
large household data base for local areas.
Whichever of the above techniques is used in the second step, the end product
must be a mapping of each particular externality space for each sampled individual.
The third step would be to specify geographic neighbourhood boundaries over the
given space which maximize the congruence between themselves and the extern-
ality spaces obtained from the sampled individuals in the second step. Undoubt-
edly, the only feasible way to accomplish this for large samples would be by
computer iteration/simulation. Such a computer program might well begin with the
criteria for selecting a spatial grouping out of a large number of alternatives as
proposed by Cliff et al. (1975). Such a simulation could employ, if desired, various
A multivariate linear probability equation indicated that people noted that the deteriorated
housing became bothersome only when it was located on their blockface.
256 What is neighbourhood? An externality-space approach
constraints on, e.g. a maximum number of neighbourhoods defined over the space
or a minimum area per neighbourhood. In addition, the simulation program could
calculate generality and accordance values for each of the neighbourhoods resulting
from the congruence-maximizing algorithm.
The preceding three-step methodology would reveal to the researcher what the
neighbourhoods for the given externality(ies) and group(s) geographically looked
like in the particular area. In addition, tests conducted with the congruence-
maximizing algorithm could reveal how sensitive congruence was to slight alter-
ations of boundaries or in the number of neighbourhoods specified. For instance,
one might be particularly interested in how much congruence is lost when one
adopts some previously established administrative or traditional boundaries instead
of the best ones.The accordance and generality values calculated for the neigh-
bourhoods generated in the simulation would also tell the investigator the degree
to which these neighbourhoods exist in similar ways across individuals in the
sample and across externalities. In other words, the boundary of maximum con-
gruence need not imply that congruence is high absolutely; accordance and gener-
ality may still be relatively low. Thus, whether neighbourhood is meaningful for
all or some of the locales in a given analysis can be assessed from the method-
ology.
As a final point, consider the contrasts between the suggested methodology and
that conventionally used: asking residents to draw or name the boundaries of
their neighbourhood; e.g. see Hunter (1974). Certainly, the latter technique is
operationally simpler. Once obtained, such maps could be subjected to accordance
and congruence calculations according to equations 6 and 3, respectively. And
this, indeed, would be an interesting exercise. It would not, however, produce very
precise measurements because there would remain residual ambiguity concerning
what type or level or neighbourhood (re: externality space) respondents were
implicitly considering when making their maps.
I11 Discussion
1
How does the foregoing analysis correspond to the conventional sociological
notion of neighbourhood as arena for social interaction and symbolic com-
munity? (Hunter, 1974; 1979). Quite simply, the social dimension is herein viewed
as a prime factor delineating an individuals externality space. Borders specified
by tradition and sentiment can serve as diodes for the perception of externality:
within the border all externalities are important; without they are not. The degree
to which group members become aware of the existence of externalities within
the border depends on the areal information nexus: how quickly and compre-
hensively news is transmitted over space by interpersonal communication. Finally,
whether recognized externalities within the border are seen as threats, windfalls,
The social dimension of neighbourhood
George C Galster 257
or inconsequential is influenced by the socialization which has transpired in the
area.
The algorithms outlined in this paper provide for the first time a framework for
testing quantitatively these various roles of the social neighbourhood. Numerous
potential research questions come to mind. What is the relationship between the
spatial density of social networks and the areas accordance and generality? If, as
Hunter (1 974) has suggested, neighbourhood perceptions vary between categories
of individuals, which specific groups (age, sex, race, SES, etc.) manifest the least
degree of accordance? Does a given intergroup dearth of accordance persist over
various types of externality spaces?
2
Lynch (1960), J acobs (1961) and Milgram et al. (1972) have shown how people
mentally map their urban surroundings and how the visual character of the
physical environment influences the clarity of these maps. Hunter (1974, Chapter
2) has found that 80% of Chicago residents descriptions of neighbourhood boun-
daries involved streets. Of these streets, 63% were within one block of vacant land,
55% were within one block of a railroad. In the context of the present model, these
findings can be interpreted in two ways. First, distinctive physical features can serve
as a rough-and-ready referent for judging the proximity of externalities, much as
social symbolism functions. Second, certain types of externalities may, in fact, be
physically impeded by certain physical barriers.
Perhaps of more interest is the potential for testing empirically these proposi-
tions. For example, do areas of high Lynchian imageability possess larger degrees
of accordance and generality? Can the creation or modification of edges, paths,
landmarks and nodes (see Lynch, 1960) dramatically augment these measures
and, indeed, create neighbourhood?
The visual-physical dimension of neighbourhood
3
The notion of generality relates to Suttless (1972) and Birch etal.s (1979) afore-
mentioned observations that people are cognitive of four distinct spatial levels of
neighborhood. I would posit that these results stem from individuals categoriza-
tion of various potential physical, demographic and social-interactive changes into a
hierarchy of externality spaces. At each level the appropriate subset of external-
ities would indicate high generality. The veracity of this interpretation awaits
further empirical tests, of course. What are the groups of externalities which cluster
with high degrees of generality at different spatial levels?
Four spatial levels of neighbourhood
sSee Wellman (1979) and Wellman and Leighton (1979).
258 What is neighbourhood? An externality-space approach
4 An array of neighbourhoods
The foregoing analysis suggests that neighbourhood is a concept which varies in
three distinct dimensions. Actual urban spaces thus may be arrayed within a matrix
according to their scores on these dimensions, analogous to a social area analysis.
Such an array provides the means for integrating disparate concepts of neighbour-
hood appearing in the literature. Cases where we get high accordance, generality,
and congruence in an area (e.g. clustered people with similar ethnic and socioecon-
omic backgrounds, little intergenerational mobility, dense spatial social networks,
high imageability of the physical environment) have been called the urban
villages (Cans, 1962). At the other extreme, areas characterized by low scores on
these three measures (e.g. people of diverse backgrounds, beliefs and preferences,
who move frequently, have spatially-dispersed networks, and inhabit a visually-
indifferentiated space) have provided testimony for the death of neighbourhoods
(Hunter, 1975; Wellman, 1972; 1979; Wellman and Leighton, 1979). Places having
values of accordance, generality and congruence between these two poles might
be referred to as generic neighourhoods.
Note that this represents a typology which is distinct from those which have
been used conventionally; for a review see Hunter (1979). The quantitative estima-
tion of the frequency of neighbourhoods at different points in the array and the
intertemporal and cross-sectional variations in these distributions provides a fertile
area of future research. Whats more, the meaning and significance of different
degrees of these indices deserves investigation. What kind of social, economic and
political reality is hidden behind a certain constellation of congruence, generality
and accordance? Such a question leads inexorably to the next area of discussion.
5 Bediction and policy evaluation
The model presented in this paper provides implications for. researchers and
planners who want to be able to predict and, if necessary, alter spatial change.
Alterations in the physical, demographic and social-interactive character of a given
area result from changes in the flows of households and resources into that area.
These flows are produced, in turn, by the individual mobility and investment deci-
sions of the residents, property owners, real estate agents, and financial institutions
who have interests in that space (Downs, 1981, Chapter 1). Changes in mobility and
investment behaviour are triggered by proximate events which produce external-
ities and thereby alter the wellbeing derived from the given residential environment.
Such behavioural responses can, of course, only ensue after ones externality space
has been penetrated.
Hence, understanding the degree to which neighbourhood exists over a given
space is a prerequisite for comprehending and predicting changes in that space. For
ZOFor contrasting schema for providing neighbourhood typologies see, e.g. Schoenberg and
Rosenbaum (1980) and Warren (1981).
George C. Galster 259
example, if the racial composition of residents on a certain block changes, how
many residents on adjacent blocks will perceive this as a change in their neighbour-
hood? If there is a wide disparity among residents on these adjacent blocks as to
their cognizance of the change, their sense that it is in their neighbourhood, and
whether it affects their wellbeing, there would be less likelihood of the spasmodic
tipping of the area. Thus, the accordance of an area becomes important for pre-
dicting responses to a particular externality generator. Similarity in the pattern of
such responses across different externality types will depend on the generality of
the area.
The measurement of neighbourhood in an area is also crucial for assessing the
impact of various public policies designed to alter the physical or demographic
dimensions of the local urban environment. Suppose, for instance, planners were
considering renovating a certain blockface or putting in a new park. Relevant ques-
tion would then be: who will view such a change as an alteration in their neighbour-
hood? Over what area are these people spread? How large a portion of all residents
in this area do they represent? In terms of the present model, such policy decisions
should be guided initially by an examination of the neighbourhood mapping pro-
duced by the maximizing-congruence algorithm (described above) as applied to the
particular externality generator being contemplated in the policy (renovation, park,
etc.) It is noteworthy that these boundaries may differ considerably from those
typically employed by planners, given the implications of several case srudies.n
Having established such, the planners could then measure accordance within these
areas in order to proxy for one measure of the efficacy of the policy.
A final way in which the above concepts may be relevant to contemporary pro-
grammatic concerns relates to the issue of local community political power. It has
been claimed that a central problem in microcommunity control of the environ-
ment is the absence of any authoritative way in which residents can appeal to a
single set of boundaries (Social Science Panel, 1974, 77). If, for this political
reason, one wanted to build neighbourhood(s), the existing levels of accordance
and generality within congruence-maximizing boundaries would provide an
indicator or where such community organizing efforts might be most propitious.
Furthermore, the success of such efforts to mobilize collective sentiment and
common symbolism could be measured by estimating increases in accordance and
generality over time.
6
As a closing comment, it should be noted that the approach proposed in ths paper
may have broader applicability than merely for the analysis of the urban neighbour-
Broader applicability of the approach
a
For a variety of studies indicating that residents perceptual boundaries differ from those
defined administratively, see Ahlbrandt, Charney and Cunningham (1977, 338), Schoenberg
and Rosenbaum (1980, Chapter 7). Warren (1981, 88), Hallman (1984, 57-59), Vaskowics
and Franz (1984. 152).
260 What is neighbourhood? An externality-space approach
hood. Indeed, the constructs of accordance, generality and congruence could be
applied to any number of networks having a geographic content that is potentially
bounded. For instance, the constructs could be applied to a given set of individuals
so as quantitatively to contrast their social ties of differing levels of frequency of
intimacy, or their various types of resources obtained from alternative institutions;
see Berkowitz (1982). Whether such applications can indeed provide an important
contribution to structural analysis must await further research.
Acknowledgements
I wish to gratefully acknowledge the provocative comments and questions tendered
by Peter Franz, J ohn Macionis, David Varady, Patricia Wittberg and anonymous
referees on an earlier draft of this paper.
IV References
Ahlbrandt, R. and Brophy, P. 1975 : Neighborhood revitalization. Lexington:
D.C. Heath.
Ahlbrandt, R., Chamey, M. and Cunningham, J. 1977: Citizen perceptions of their
neighborhoods. Journal of Housing 7,338-41.
Ahlbrandt, R. and Cunningham, J . 1979: A new public policy for neighborhood
preservation. New York: Praeger.
Berkowitz, S.D. 1982: An introduction t o structural analysis. Toronto: Butter-
worth.
Birch, D. Brown, E., Coleman, R., Da Lomba, D., Parsons, W., Sharpe, L. and
Weber, S . 1979: The behavioral foundations of neighborhood change. Wash-
ington: USGPO/HUD.
Clay, P. 1979: Neighborhood renewal. Lexington: D.C. Heath.
Cliff, A., Hagget, P., Ord, J.K., Basset, K. and Daries, R. 1975: Elements of spatial
Downs, A. 198 1 : Neighborhoods and urban development. Washington: Brookings.
Farley, R. Schuman, H., Bianchi, S., Colastano, D. and Hatchett, S. 1978: Choco-
late City, Vanilla Suburbs: will the trend toward racially separate communities
continue? Social Science Research 7, 319-44.
Firey, W. 1945: Sentiment and symbolism as ecological variables. American Socio-
logical Review 10, 140-48.
Franz, P. 1982: Zur Analyse der Beziehung von Sozialokologischen Prozessen und
Socialen Problemen. In Vaskovics, L., editor, Zur Raumbezogenheit Socialer
Probleme, Opladen, Germany.
Gabriel, S. and Wolch, J . 1984: Spillover effects of human service facilities in a
racially segmented housing market. Journal of Urban Economics 16, 339-50.
Galster, G. 1982: Black and white preferences for neighborhood racial composition.
American Real Estate and Urban Economics Association Journal 10, 39-66.
forthcoming: Neighborhoods, homeowners, and housing upkeep. Wooster, Ohio,
monograph in press.
structure. Cambridge: Cambridge University Press.
George C Galster 261
Cans, H. 1962: The urban villagers. New York: Free Press.
Goetze, R. 1979 : Understanding neighborhood change. Cambridge: Ballinger.
Golab, C. 1982: The geography of neighborhood. In Bayor, R., editor, Neighbor-
Greer, S . 1962: The emerging city. New York: Free Press.
Grether, P. and Mieszkowski, P. 1980: The effects of nonresidential land use on the
prices of adjacent housing: some estimates of proximity effects. Journal of
Urban Economics 8, 1-15.
Hallman, H. 1984: Neighborhoods: their place in urban life. Beverly Hills: Sage.
Hunter, A. 1979: The urban neighborhood: its analytical and social contexts.
hoods in urban America. Port Washington, NY: Kennikat.
Urban Affairs Quarterly 15, 267-88.
1974: Symbolic communities. Chicago: University of Chicago Press.
1975: The loss of community. American Sociological Review 40, 537-52.
J acobs, J . 1961: The death and life of great American cities. New York: Random
J anowitz, M. 1952: The community press in an urban setting. New York: Free
Keller, S . 1968: The urban neighborhood. New York: Random House.
Laumann, E. Marsden, P. and Prensky, D. 1983: The boundary definition problem
in network analysis. In Burt, R. and Minor, M., editors, AppZied network
analysis, Beverly Hills: Sage.
House.
Press.
Lunch, K. 1960: The image of the city. Cambridge: MIT Press.
Michelson, W. 1976: Man and his urban environment. Reading: Addison-Wesley.
Milgram, S., Greenwald, J ., Kessler, S., McKenna, W. and Waters, J . 1972: A
Morris, D. and Hess, K. 1975: Neighborhood power. Boston: Beacon Press.
National Commission on Neighborhoods 1979: People, building neighborhoods.
Washington: USGPO.
Pettigrew, T. 1973: Attitudes on race and housing. In Hawley, A. and Rock, V.,
editors, Segregation in residential areas, Washington: National Academy of
Sciences.
Schoenberg, S. 1979: Criteria for the evaluation of neighborhood viability in
working class and low income areas in core cities. Social Problems 27,69- 85.
Schoenberg, S. and Rosenbaum, P. 1980: Neighborhoods that work. New Bruns-
wick, NJ : Rutgers University Press.
Schreiber, A. and Clemmer, R. 1982: Economics of urban problems, third edition.
Boston: Houghton-Mifflin.
Scotland, J . and Galster, G. 1977: Federal rehabilitation programs: are they worth
it? Urban Studies Working Paper, The College of Wooster, Wooster.
Sepal, D. 1979: The economics of neighborhood. New York: Academic Press.
Social Science Panel 1974: Toward an understanding of metropolitan America.
San Francisco: Canfield Press/National Academy of Sciences.
Suttles, G. 1972: The social construction of communities. Chicago: University of
Chicago Press.
Taub, R. Surgeon, G., Lindholm, S., Otti, P.B. and Bridges, A. 1977: Urban volun-
tary associations, locally based and externally induced. American Journal of
Sociology 83,425-42.
psychological map of New York City. American Scientist 60, 194-200.
262 What is neighbourhood? An externality-space approach
Vmkowics, L. and Franz, P. 1984: Residential areal bonds in the cities of West
Germany. In Peachey, P., Bodzenta, E. and Mirowski, W., edi tors, The residen-
tial areal bond, New Y ork: I rvington.
Warren, D. 1981: Helping networks. Notre Dame, I ndi ana: Notre Dame University
Press.
Warren, R. 1972: The community in America, second edition. Chicago: Rand-
M cNally .
Wellman, B. 1972: Who needs neighborhoods? In Powell, A., edi tor, The city:
attacking modern myths, Torono: McClelland and Stewart.
1979: The community question. American Journal of Sociology, 84, 1201-31.
Wellman, B. and Leighton, B. 1979: Networks, nei ghborhoods and communities.
Urban Affairs Quarterly 14, 363-90.
Cet article presente tout dabord une definition conceptuelle de Iunitt de voisinage, definition
approprite ti Iexamen et la prevision des rtponses comportementales des individus face aux
changements de leur environnement immediat. La notion de voisinage est perque par les
personnes qui vivent et/ou investissent dans un espace habitable donne. Le voisinage est defini
en termes despace dexternalitt, cest-a-dire Iespace dans lequel tout changement
environnement cause par un tiers est pequ comme une attente au bien-ttre dont un individudonne
jouit dans Iespace considere. Ensuite, des algorithmes sont tlabores lesquels les perceptions
individuelles peuvent etre curnulees et leur differences quantifiees selon trois dimensions scalaires:
leur congruence, her generalite et leur concordance. Ces trois echelons expriment le degre de
consensus sur les limites des espaces dexternalite: I ) par comparaison a des limites pre-
determintes ou officielles, 2) entre individus dans un espace donne 3) entre differents types
despaces. Ensemble, ils traduisent la mesurabilite de la notion de voisinage pour un espace
donne. Cet article avance quil est possible detablir des specifications operationnelles et
gtographiques du voisinage a partir dinvestigations empiriques dkcoulant de ces algorithmes.
Enfin. a la lumitre de cette thtorie, les conceptions traditionnelles du voisinage et leurs
implications politiques sont discuttes sous un nouveau jour et de multiples axes de recherche
empirique sont proposes.
Z u Beginn der Abhandlung wird eine Definition des Begriffs Nachbarschaft gegeben, anhand
derer die Verhaltensweisen des einzelnen als Reaktion auf Veranderungen in seiner sozialen
Umgebung untersucht und vorausbestimmt werden konnen. Ausgangspunkt sind die
Wahrnehmungen der Menschen, die in einem gegebenen Bezirk leben und/oder darin eine aktive
Rolle Spielen. Nachbarschaft wird definiert als aukrlicher Wahrnehmungsraum, d.h. der
Bereich, in dem der einzelne durch andere bewirkte Umgebungsveranderungen als eine
Veranderung seines Wohls wahrnimmt, das er in einem gegebe nen Bezirk genieBt. Als nachstes
werden Verfahren entwickelt, mit deren Hilfe einzelne Wahrnehmungen aggregiert und die
Unterschiede innerhalb dieser Wahrnehmungsfolge mittels dreier skalarer GroBen quantifiziert
werden konnen: Kongruenz, Allgerneingiiltigkeit, Ubereinstimmung. Diese drei GroOrn
geben an, in welchem MaBe Grenzen von auBerlichen Wahrnehmungsraumen miteinander
ubereinstimmen, 1) wenn sie mit einer vorher bestimmten oder offiziellen Grenze verglichen
werden, 2) ganz allgemein bei beliebigen Individuen fur einen gegebenen auBeren
Wahmehmungsraum, 3) ganz allgemein bei verschiedenen Arten von auOeren
Wahrnehmungsraumen. Alle drei GroBen zusammen dienen als quantitativer Indikator fiir das
AusmaO, bis zu dem eine bedeutungsvolle Nachbarschaft innerhalb bestimmter Grenzen
definierbar ist. Weiterhin werden Vorschlage gemacht, wie operationale, georgraphische
Spezifikationen fur den Begriff Nachbarschaft auf der Basis empirischer Untersuchungen
gefunden werden konnen, die sich aus den genannten Verfahren ergeben. AbschlieBend werden
neue Einsichten in traditionalle Vorstellungen von Nachbarschaft und die Bedeutung fur
Untersuchungsmethoden vor dem Hintergrund der Theorie diskutiert. Dariiber hinaus werden
zahlreiche Leitlinien fur die empirische Forschung gegeben.
George C Galster 263
Primeramente, la ponencia presenta una definici6n conceptual de la vecindad, que es adecuada
para examinar y pronosticar la respuesta reflejada en el comportamiento de 10s individuos, frente
a cambios en el medio ambiente que les rodea. Se basa en las percepciones de personas que viven
y/o invierten en una localidad especifica. La vecindad se define en tkrminos de espacio
exteriorizado: el area en la que 10s cambios ambientales iniciados por otros se percibe alterarhn
el bienestar que cualquier individuo obtiene de la localdad especifica. Luego, se desarrollan
algoritmos por medio de los cuales se pueden agregar las percepciones individuales, y se pueden
cuantificar diferencias en esas percepciones en dimensiones de tres escalas: congruencia,
generalidad y concordancia. Estas dimensiones miden el graso de consenso en 10s linderos de
espacios exteriorizados: 1) cuando se comparan a algdn lindero determinado anteriormente u
oficial, 2) para 10s individuos de una exteriorizaci6n dada, y 3) para 10s diversos tipos de
exteriorizaciones. En combinaci6n, estas dimensiones proveen un indicador cuadtitativo del punto
hasta el cual se puede definir una vecindad significativa con un cierto espacio. Se sugiere dmo
las especificaciones geogrhficas, operativas de una vecindad podrlan basarse en investigaciones
empfricas haciendo us0 de estos algoritmos. Finalmente, se discuten nuevos aspectos de las ideas
tradicionales de la vecindad, y las implicaciones para la politica, como consecuencia de la teorfa,
y se sugieren numerosas posibilidades para la investigaci6n empirica.