Version classiqueVersion mobile

Trajectoires familiales et espaces de vie en milieu urbain

 | 
Francine Dansereau
, 
Yves Grafmeyer

1. Mobilités, réseaux familiaux, contextes d'habitat

View from the door: a GIS study of neighborhood environment1

John R. Miron

Texte intégral

  • 1 This research has been funded by the Social Sciences and Humanities Research Council of Canada (41 (...)
  • 2 In this paper, «type of dwelling» refers to structural form of the dwelling. Three dwelling types (...)

1Studies of the significance, use, or adequacy of housing and home typically focus on 1) the internal layout, construction, and tenure of dwelling, and 2) the neighborhood or locational context of housing. The Census of Canada is a useful source of information on the former; at least for households in private dwellings (that is, not in collective accommodation). Keeping in mind that questionnaire goes to only a small percentage of all respondents, households in private dwellings generally have been asked to report on type of dwelling occupied2, tenure, period of construction, and state of repair. In marked contrast however, the Canadian census does not ask households anything about their neighborhood environment. Let us clarify the nature of the conundrum with a simple illustrative question about neighborhood. Suppose we want to know just how many Canadians live in homogeneous, modern, low-density suburbs. In terms of census data, this question is «simple» because it implicitly characterizes neighborhood environment in terms of dwellings in the vicinity (about which census data can say something) rather than in terms of nonresidential land use nearby (about which the census is silent).

2To answer such a question, the best that many scholars can do presently is to look to published census counts of dwellings aggregated to a particular geographic lavel: e. g., the Census Tract (CT). If a CT contains mainly single detached dwellings built since 1970, we might then conclude that it is a contemporary, homogeneous, low-density suburb. Classifying each CT in 1991 by the neighborhood category that it exemplifies, we could then enumerate the population in each category of neighborhood. This method presumes that each CT is a homogeneous area; a belief maintained by Statistics Canada, the agency charged with conducting the census. However, there are two reasons for concern about the appropriateness of assigning each CT to a category of neighborhood. First, a CT is typically large; on average including about 1,500 private dwellings. The entire area of Canada is partitioned into only 5,883 census tracts in 1991. While perhaps well suited to some other definition of neighborhood, 1,500 dwellings is typically more than a resident might expect to see when they open their front door. When a CT includes two or more different categories of neighborhood, such information is lost in a procedure that assigns each CT to just one category. A second and related argument is that a CT in practice will be heterogeneous necessarily because of the way in which it is defined. Census tracts were initially laid out for each of Canada s large cities in the 1961 Census. Prominent landscape features (e. g., highways, roads, rivers or streams, lakefronts, municipal boundaries) were used to outline CTs. As a result, most CTs are bounded on at least one side by an arterial street. However, the use of arterial streets is problematic in modern market economies. Sites along arterial streets, perhaps first developed as low-density residential, may become valued by retail and commercial users, as well as for high-density residential development. Sites in the interior of the CT, away from arterial streets, are more likely to remain low-density residential. As a result, some CTs have become bifurcated: a low-density core set inside a higher density edge. Put differently, differences in neighborhood experience within a CT may be as great as differences among CTs.

  • 3 A blockface is defined to be one side of a street between two consecutive critical points: a criti (...)

3However, Statistics Canada also provides summary data for still-finer geographic areas in machine-readable format. This includes, since 1961, rich sets of cross-tabulations for each Enumeration Area (EA) in Canada; albeit cross-tabulations that have varied in form and content from one census to the next. An EA typically consists of only about 300 households, and is defined by Statistics Canada, quite atheoretically, to be the area assigned to one Census Representative (that is, surveyor). There were 45,995 EAs across Canada in 1991. Statistics Canada makes no representation as to the homogeneity of such areas, and certainly does not refer to them as neighborhoods. However, in principle, the EA must be at least as homogeneous as a CT, since EAs are drawn so as to partition each CT. With the 1991 census, Statistics Canada also made available for the first time a machine-readable file of counts of persons and private households at the blockface level for part or all of each of Canada’s 44 largest urban areas3. The file includes 463,822 blockfaces that each had at least one private household.

4A concept of neighborhood can be superimposed on census EA and blockface data to give new indicators of local environment. This paper employs the notion that a neighborhood is the area typically seen by residents stepping out from the front door of their dwelling. Fine geographic scale is the principal advantage of EA and blockface data. Fine geographic scale also poses a disadvantage. However defined, a neighborhood is generally thought to include more than one blockface. Also, what can be seen from one’s own front door may well include more than one EA: particularly in higher-density neighborhoods. Hence, empirical descriptions of neighborhood that are based on EA or blockface data must take into account the housing stock characteristics of nearby EAs or blockfaces.

5The purpose of this paper is to explore what circle aggregations of EA and blockface data can tell us about neighborhood environment across Canada in 1991. As a first step in evaluating neighborhood environment in this way, this paper uses only dwelling attribute counts to characterize neighborhood. Each EA is assignated to a neighborhood category on the basis of all housing found within the circle. Categorization of neighborhood is based on most-common combination of dwelling type and period of construction found within the circle, and the homogeneity of the stock.

I – Neighborhood differentiation: an explanation

6The focus of this paper is on that part of the built environment which is residential in nature as well as being visible from the street. Why might we expect neighborhoods to differ in this specific regard? Central to the explanation offered here are the well-known ideas that housing is both costly to construct and durable. These ideas have immediate consequences. First, the housing stock that individuals see around them today is in part a legacy of market conditions at the time the housing was constructed. Second, while some households find it advantageous to renovate, make additions to, or otherwise alter their dwelling as their needs change, others find it advantageous simply to relocate.

  • 4 Bossons (1993) advances the argument that modern planning is usefully perceived as an outcome of a (...)

7It is helpful, therefore, to think of neighborhood both as an economic choice made by consumers and as an outcome of public planning. Individuals can be thought to choose their neighborhood, either explicitly (when they leave one neighborhood and move to another) or implicitly (when they choose not to move). From an economic perspective, an individual chooses the neighborhood and dwelling which best advances their aspirations within the constraints of their budget. But, in this, choice is not neutral in its effects. By moving, individuals may well cause the price of housing service to rise in the new neighborhood, and fall in the old one. In turn, this can have implications for housing maintenance and renovation, the price of housing stock, and site redevelopment. In addition, while such decisions can be thought to be private undertakings by the household, they may also have externality effects on neighborhood quality, and hence on housing stock investment in nearby properties. Public planning can be thought of as the political side of land use allocation that complements the market (economic side) in internalizing such externalities. The efforts of planners, here seen to be responding to the demands of the electorate, have been significant in shaping urban residential environments4.

8Over the past half century, there have been major changes in market conditions. Market conditions have changed in response both to supply-side and demand-side factors, and against a backdrop of rapid population growth. While it is difficult to unravel the separate contribution of each factor, the consequent changes in housing stock and neighborhood are readily identified. One important story was the growing affluence of Canadians, at least up to the late 1970s. After the deprivations of the great depression and the second world war, the small, storey-and-a-half «wartime housing» built in the late 1940s and early 1950s was a fresh and optimistic start: see McKellar (1993, p. 140). In the 1950s and early 1960s, the ranch bungalow and split level dwelling, typically with three bedrooms, took over as Canadians grew more affluent. In the 1970s, these were themselves largely replaced by the two-storey four-bedroom house with family room and two-car garage. Then, in the mid to late 1980s, the «monster» home, typically with five or more bedrooms and three-car garage, came into its own for those who could afford it. A second story concerns the rise of the nontraditional household. Typically smaller in size and less affluent, such households were attracted to smaller, less-costly dwellings. In larger urban areas, new construction technologies led to the apartment boom of the 1960s and 1970s; a boom that originated in the rental market but eventually spread to condominium ownership in the 1970s. A third important story was the effect of the baby boom generation, that began to spill out of parents’ homes in the mid 1960s. By the 1970s, rapidly rising land prices in Canada’s major urban areas led to the emergence of «cluster dwellings» (semi detached, row, and link housing) widely seen to be starter housing for young couples. A fourth story was the revitalization of inner city areas (gentrification, redevelopment, intensification, and re-use) that got underway in several Canadian cities beginning in the 1960s. The fifth story was the changing nature of assisted housing construction in Canada. In the 1950s and 1960s, housing agencies saw high-rise, large-scale buildings as a new technology that was cost-effective in delivering dwellings to needy households. After the 1960s, concerns about the suitability of high-rise accommodation for families, and the geographic concentration of low-income households led public housing managers to prefer low-rise, small-scale social housing.

9There have also been major changes in the nature of town planning over this period. In general, planning practice became more restrictive as communities experienced growing pains. Smith and Moore (1993) characterize the changes as 1) a switch from the reactive planning in the 1950s, wherein the goals were to enable CBD development and eliminate blight in the metropolitan core and to prevent inefficient sprawl in the suburbs, to the proactive planning practised in recent decades, 2) the decline of expert-based master planning and the emergence of advocate-based community participation, 3) emergence of the neighborhood unit and corporate suburb, and 4) increasing emphasis on social mix, if not at the neighborhood scale, then at a broader local scale: «local» here taken to mean a collection of adjacent neighborhoods. To this list can he added a change in planning focus from what is happening on the urban fringe to issues of redevelopment and re-use of land within existing neighborhoods.

II – Method

  • 5 Hereinafter, the term «kind» is used to indicate a breakdown by dwelling type or period of constru (...)

10The method employed in this study is to traverse a map of Canada stopping at each EA centroid, draw a 400 meter circle, count dwellings (grouped by EA or blockface) by kind within the circle, and then assign the circle to a neighborhood category on this basis5. Because of the limitations of census data, nonresidential land uses are ignored in this assignment. In this study, proximity is measured by the straight-line distance between centroids: in other words, a household is deemed to lie within the circle if its group (blockface, or EA) centroid lies within the circle. In turn, each EA centroid can then be thought of as a point from which to reconnoiter the nearby housing stock. Since EAs each include roughly the same number of households, this procedure can be thought to generate a representative sample of neighborhood environments.

1. Categorization

  • 6 A similar distinction is drawn by Goldsteen and Elliot (1994, p. 128-136).

11How should we categorize neighborhood? Any answer to this question is severely constrained by the nature of available Census data. We have already noted, for example, that Census data tell us nothing about nonresidential land use in the vicinity. Further, Census data on residential land use are limited to aspects of what can be seen, rather than how the observer interprets what is seen; that is, when they focus on perception rather than satisfaction6. In part, complexity arises because perception itself is a function of what the resident chooses to observe. Further, even if all residents choose to view a neighborhood in the same known way, the range of characteristics that could he observed is virtually unlimited. The 25 case studies of urban neighborhood in Urhahn and Bobic (1994, p. 34-85) nicely illustrate the multidimensional nature of neighborhood view.

  • 7 Prior to the 1971 Census, type of dwelling was recorded by the census taker.

12Goldsteen and Elliott (1994, p. 109-205) present a typology of physical space using 5 design elements (points, lines, planes, masses, and spaces) and 8 visual qualities (light and dark, color, texture, figure-ground relationships, grids and modules, balance and symmetry, scale, and proportion). Viewed in terms of this broad set of concepts, census data give us scant information. Nonetheless, recent censuses in Canada provide two pieces of information about each dwelling that could be thought to be basic descriptors of what people see from their front door. First, households are asked to identify their type of dwelling: e. g., single detached dwelling, apartments in building of five storeys or more, any other dwelling7. Counts of dwelling by type within a neighborhood tell us something about planes and masses, including height, density, and spacing of buildings. Second, households are asked about period of construction of their dwelling. This can tell us something about texture (in the sense of the sheathing of buildings and maturity of trees) and street width and alignment.

  • 8 Note that prior to 1946, there were few apartment buildings more than five storeys tall. Much of t (...)

13The categorization employed here uses 9 kinds of private dwelling: 3 types of structure (single detached, apartment in building of five storeys or more, other dwelling) by 3 periods of construction (pre 1946, 1946-1970, and 1971-1991). These are displayed on the left side of Table 1. Within each circle, the most common of the 9 kinds of dwelling is identified, and the cicle is further categorized by whether or not it is homogeneous (that is, whether the most common kind forms a majority of all private dwellings in the circle)8. Hence, there are 18 neighborhood categories in all, as shown in Table 1. Note that heterogeneity and homogeneity, as used in this paper, refer to the mix of housing stock only: not to other kinds of mix (e. g., social class, income, tenure, race or ethnicity). Categorizations based on type of dwelling and period of construction are nothing new in thinking about neighborhood environment. Bourne (1993, p. 279-280) uses a similar typology (although Bourne also differentiates neighborhood by size of urban area). What is novel about the categorization employed in the present paper is that it also takes into account the homogeneity of housing stock within the neighborhood. In what follows, I argue that homogeneity and heterogeneity of the housing stock are valuable to understanding how neighborhood experience varies across Canada.

Most common kind of dwelling

Homogeneous (most common kind of dwelling is majority of all dwellings)

Heterogeneous (most common kind of dwelling is not majority of all dwellings)

Single detached dwelling
Pre 1946
1946-1970
1971-1991


16
17
18


7
8
9

Apartment, building of 5 storeys or more
Pre 1946
1946-1970
1971-1991


10
11
12


1
2
3

Other dwelling
Pre 1946
1946-1970
1971-1991


13
14
15


4
5
6

Source: See text for details
Table 1
Eighteen neighborhood allegories

14How does our discussion of neighborhood differentiation above play out in the context of this scheme of categorization? Hypothetically, suppose that a developer built a neighborhood in 1901 which consisted of 350 single detached dwellings (that, conveniently, fill a circle of 400 meters radius). See Table 2. Assume that this neighborhood is one EA, and that there are no other EAs within the circle. In terms of Table 1, this neighborhood starts out as category 16 (homogeneous, mainly single detached, built prior to 1946). Now, let the decades pass, and assume that during the 1920s through 60s, a substantial number of these now-old homes are converted to more-affordable duplexes and triplexes (that is, «other dwelling» according to our categorization scheme); by 1971, we might well classify this area as category 7 (heterogeneous, mainly single detached, built before 1946). Heterogeneity is here enhanced also because some of the old homes are demolished to make way for newer single detached housing. Now, suppose that, with gentrification in the 1970s and 80s, some of the converted duplexes and triplexes are deconverted back to single detached dwellings. In Table 2, there has been enough deconversion by 1991 to return this area (barely) to category 16. Of course, this is just one possible scenario for the historical development of this neighborhood; nonetheless, any scenario serves to emphasize the importance of conversion, deconversion, demolition, and new construction in shaping neighborhood change.

Source: See text for details
Table 2
Categorization and housing stock in hypothetical neighborhood, 1990-1991

15Also important here is the notion that heterogeneity can indicate redevelopment activity. Although redevelopment can sometimes occur on a massive scale (e. g., a project to construct a large cluster of apartment buildings), it is more commonplace to redevelop properties one or two at a time; hence redevelopment generally leads to increased heterogeneity. However, two important caveats come to mind. The first caveat is that, using our categorization, the link between heterogeneity and redevelopment will be most marked for neighborhoods that start out as mainly single detached housing. Suppose in Table 2 that our builder had instead constructed 350 semi-detached houses in 1901. Then, the neighborhood would have been initially in category 13 (homogeneous, other dwellings, pre 1946). If over the years these houses were flatted (that is, converted to duplexes and triplexes), the categorization would still remain the same. If, alternatively, the builder had constructed 350 apartment dwellings in 1901, we might expect to see relatively little conversion activity down through the years; it is not easy to convert an apartment building to single detached or other dwellings. The second caveat is that our categorization is blind to renovation of the housing stock. In Table 2, the purchaser who buys an original house in 1971, guts it, and rebuilds inside the original structure, and perhaps puts on a new addition is nonetheless still living in a 1901 house. Only if renovation changes the structural type of the dwelling is it captured in our categorization scheme.

2. Enumeration

16How do we count the dwellings by kind within each circle? Since such counts are available at the EA level, we could simply add up the counts for all EA centroids lying within the circle. This is the EA-aggregated count (or «all-in all-out» method). Prior to the 1991 Census, this is the only option available to scholars. With the advent of the 1991 census, an alternative approach is to make use of the blockface file which includes a count of private dwellings on each blockface. On the assumption that the housing in an EA is homogeneously distributed, EA counts of dwelling by kind can be apportioned to each EA circle in the vicinity. This yields what is referred to in this paper as a «blockface-aggregated count». However, such counts are possible only for 44 principal urban areas across Canada.

  • 9 All EAs shown are part of Federal Electoral District 78.

17To illustrate the method, consider Map 1 which shows streets, and 1991 blockface and EA centroids (labeled) in southwestern Scarborough, Ontario9. Also shown in this Map is a 400 meter cicle drawn around the centroid of EA 212. Note how much smaller this cicle is compared to Census Tract 338, of which EA 212 is part. Also shown within the circle are pertinent EA boundaries. The cicle encloses just two EA centroids (212 and 219); however, it includesin addition to all blockfaces in 212-some blockfaces from five other EAs: 208, 210, 211, 219, 254. Of course, we are using the circle around EA 212 in particular simply to illustrate the method. In fact, the method employs a circle around each EA centroid. Hence circles may well overlap: dwellings at one blockface or EA may well contribute to neighborhood attribution for more than one circle.

18Why use circles of 400 meters radius? In principle, we want to approximate the area visible to residents from their front door. However, in practice, visibility will depend on several factors: e. g. nature of the built form, vegetation coverage, topography, and street alignment. Further, a resident may well ignore more-distant land uses that, while observable, are different from the built form that is nearby. In ibis study, I experimented with circles of 200, 300, 400, and 500 meters. In general, buildings that are 500 meters away are not readily observed (at least from ground level) unless they are high-rise apartment buildings. This suggests using a radius of not more than 400 meters. However, the use of circles smaller than 400 meters is problematic because our housing data are EA-aggregated or BF-aggregated. If we use a radius of 300 meters or less, the number of circles that would not include any blockfaces, or just one EA centroid increases substantially. Therefore, in this study, because the census data are already aggregated, we can do no better than use 400 meter circles.

Map 1
Southwest section of Scarborough, Ontario, 1991

Source: Culumns (1) through (6) assembled from machine-readable 1991 census data files. Culumns (7) and (8) computed by author. See the text for details. Neighborhood categories listed in Table 1.
Note: All EAs located in Federal Electoral District 78.
Table 3
EA and blackface data for 400 meter around EA 212, southwestern Scarborough, 1991

19The calculations for EA-aggregated counts and BF-aggregated counts in the Map 1 circle are shown in Table 3. Columns (1) through (6) give counts for each of the six EAs that intersect the 400 meter circle. Columns (7) and (8) show the EA-aggregated and blockface-aggregated counts: here, column (7) is the sum of columns (5) and (6). The first two rows give counts of private households from the blockface file: row 1 is the sum of households on blockfaces whose centroids lie within the 400 meter circle; row 2 is the sum of households within the EA. The next nine rows of data are the counts of private dwellings by kind within the EA. Column (8) is the sum of columns (1) through (6), each weighted by the corresponding ratio of rows 1 to 2. The final row gives the neighborhood assignment: column (7) indicates category 9 (heterogeneous, single detached, 1946-1970); column (8) indicates category 7 (heterogeneous, single detached, pre 1946).

  • 10 «Lie within it» taken here to mean that the dwelling centroid lies within the circle. In the case (...)

20Before proceeding, let us be clear about what this method does, and does not, do. If, in contrast, our method had used a separate centroid for each of Canada’s 10 million dwellings, we could say that each circle would count precisely those dwellings that lie within it10. This is not, however, what the EA and blockface aggregation methods do. Instead, they each assume that a group (blockface or EA) of dwellings, not each dwelling individually, is entirely in or out of the circle. Thus, it is possible that a small variation in the location of a group centroid near the edge of a circle could produce a large change in the count of dwellings for that circle. A second, and related, concern is that, in rural or remote areas, an EA may be considerably larger in area than a circle of 400 meters radius. In the EA aggregated method, the circle for such an EA would include its own centroid, but the count of dwellings would he substantially larger than what is actually sited within the circle. If we wanted to count dwellings across Canada, the grouping problem would be severe. However, what we want to do here is something different; we want to characterize the neighborhood in which households find themselves. Here, the lumpiness of blockface and EA aggregation is less important.

III – Empirical results: Scarborough

21To better understand the method, this paper begins with a case study of the City of Scarborough, a suburban community (187 km2 in area, with a population in 1991 of 525 thousand persons) that is immediately adjacent to, and east of, the City of Toronto. See Map 2. At present, with the exception of its northeast corner, Scarborough is completely built up; now consisting of remnants of scattered historic villages surrounded by tracts of postwar housing, light industry, and some suburban office park development.

Map 2
Neighborhoods (EA aggregation) Scarborough, 1991

  • 11 The neighborhood assignments in this map take into account the housing characteristics of centroid (...)

22Map 2 summarizes the neighborhood structure of Scarborough using EA aggregation11. The Map shows periods of construction as circles in the background. To better emphasize new construction, these circles are drawn in layers; circles with mainly new housing are drawn on top of circles with mainly older housing. Map 2 illustrates the sequencing of development in Scarborough, from a few old areas mainly in the southwest corner adjacent to the City of Toronto, through suburban tract developments that spread north and east, to the present-day redevelopment in older suburbs along Kingston Road, an arterial street that parallels the Lake Ontario shoreline. Superimposed on Map 2 are symbols indicating heterogeneity/homogeneity and principal type of dwelling. A grey symbol (o) is drawn on top of circles where none of the 9 kinds of housing constitutes at least 50% of the housing stock in that circle. These are the heterogeneous areas. Where one kind of housing is in the majority, a different symbol is employed depending on the principal type of dwelling. These latter categories of neighborhood are thus labeled homogeneous.

23As with many postwar suburban areas, Scarborough was developed as a sequence of land subdivisions within which housing was distinctly clustered by kind. The story, a familiar one, was repeated from suburb to suburb across Canada. In the 1940s and 1950s, small-scale subdivision gradually gave way to large tract developments: typically in the form of single detached housing. The 1960s saw the advent of the apartment boom: particularly along arterial roads and near subway stops. Then, in the 1970s and 1980s, planners actively encouraged developers to experiment with alternative development schemes that avoided some of the pitfalls of earlier developments. In Scarborough for example, planners worried about the large tracts of homogeneous development that had been built in the 1960s.

24Scarborough planners encouraged builders to mix «pockets» of low-rise housing (row housing, semi detached, and single detached) in their developments. Map 2 evidences the consequences of this planning activity. In the southern (older) half of Scarborough, Map 2 evidences vast tracts of homogeneous housing: commonly single detached, although pockets of apartment buildings have become more common in the last two decades. In the northern (newer) half of Scarborough, there are more pockets of «other» housing (typically semi detached and row housing). As well, pockets tend to be smaller in area in the newer half; reflecting the tendency of planners in recent decades to induce more social mix. The result in newer Scarborough is more heterogeneity viewed at a gross geographic scale, but more homogeneity when viewed at a fine scale.

Source: Calculations by author. See text.
Notes: EA aggregation used. «Same» is same neighborhood category. «Het-hom» is otherwise equivalent but neighbor opposite in homogeneity. Homogeneous neighborhood categories instanced in fewer than 10 EAs are not shown here. Table omits EAs with no or few private dwellings: 37 in total.
Table 4
Nearby neighborhoods, Scarborough, 1991

  • 12 Here, «south» includes all Scarborough neighborhoods with a centroid UTM Northing smaller than 484 (...)

25This pattern can be described empirically using Table 4. Suppose we define the set of nearby neighborhoods (hereinafter «nearbys») to be those whose centroids lie within a radius of 1 km of an EA centroid. In southern Scarborough, again using EA aggregation, there are 148 heterogeneous neighborhoods and another 168 that are homogeneous. The homogeneous neighborhoods have on average 12.7 nearbys each; see panel (a)12. Around these homogeneous neighborhoods, 34 % of nearbys fall into the same category, 13% are otherwise similar but heterogeneous, and the remaining 53 % are in some other neighborhood category. The remaining rows in panel (a) show detailed breakdowns for each homogeneous category. Panel (b) shows corresponding results for the northern half of Scarborough. The impact of planning is evidenced by comparing two rows from each panels (a) and (b). In homogeneous areas of single detached housing (built 1946-1970) in older Scarborough, 38% of nearbys are in the same category: indicative of the large tracts of such housing. However, in newer Scarborough, only 29% of the nearbys to homogeneous single detached housing (built 1971-1991) are of the same category: indicating that tracts of such housing are typically smaller.

26There is a second important difference between older Scarborough and newer Scarborough evident by comparing panels (a) and (b) of Table 4. In older Scarborough, category 11 (homogeneous, apartment, 1946-1970) neighborhoods are less likely to have similar nearbys than are the more-modern apartment neighborhoods in newer Scarborough: 38% for the former versus 51 % for the latter. Why the difference? In the 1950s and 1960s, apartment buildings were commonly smaller in scale and often undertaken each as a single project. In the 1970s and 1980s, however, developers became more interested in constructing apartment buildings in large clusters. For developers, clustering renders economic the provision of appurtenant facilities and services (e. g., games room, concierge) for which consumers are willing to pay. On the demande side, the fastest-growing market for apartments was seen to be the nontraditional household that had little time to spend on household activities and hence wanted facilities and services that could substitute for their own time. This massing of apartment buildings also had the support of planners who saw the possibilities for an improved community environment and easier provision of public services.

27A third important difference between the two Scarboroughs is in the nature of a heterogeneous neighborhood. In newer Scarborough, heterogeneous neighborhoods are typically found at the boundary of homogeneous neighborhoods. This is the notion of heterogeneous neighborhood as a spatial transition zone: a tool used by planners to buffer neighbors thought to be incompatible. In older Scarborough, however, heterogeneous neighborhoods do not form this kind of «thin line» between homogeneous areas. Instead, as seen in Map 2, heterogeneous neighborhoods are more likely to be areas undergoing transition over time from one dominant kind of housing to another: e. g., gentrification, infilling, or other redevelopment. This area attracts renovators because the Toronto region is growing quickly, and because older Scarborough is accessible to downtown Toronto, is largely residential, and has good services, schools, and other amenities.

Source: Calculations by author. See text.
Note: Neighborhood categories instanced in fewer than 10 EAs are not shown here.
Table 5
Estimated population, private dwellings, and Enumeration Areas by neighborhood category, Scarborough, 1991

28Table 5 presents counts by neighborhood type. In each case, these are aggregations of EA-level counts (not 400 m circle counts) for all included EAs. Unlike 400 m circle data which include centroid data each time the centroid lies within a circle, there is no double counting in Table 5; each column totals to the corresponding Scarborough total of population, dwellings, or EAs. In principle, all three counts are of interest. However, since this paper emphasizes the view from the door, preference is given here to population counts. It should be added that these are counts of all individuals in the EA: that is, the counts include both residents of private dwellings and collective dwellings. Hence, population counts here include more people than would be counted in private dwellings alone.

29How many people live in homogeneous neighborhoods within Scarborough? In short, a lot. See Table 5. Using the EA-aggregation method, 327 thousand people in Scarborough (that is, 62% of the population) are estimated to live in homogeneous neighborhoods: that is, are counted in panel (a). Most of these (56 %) are in neighborhoods dominated by single detached dwellings; another 33 % are in apartment-dominant neighborhoods. If we use instead the blockface aggregation method, the story is similar; although here the counts of homogeneous neighborhoods are modestly lower on average.

IV – Empirical results: Toronto cma

30Now, let us move up to the scale of the larger region: here the Toronto CMA. In terms of the history of housing construction, the Toronto CMA consists of three distinct tiers. See Map 3. The central city area (City of Toronto), originally founded in 1793, was largely built up by 1945. According to the 1991 census, 635 thousand people live in the City. The City is surrounded by a horseshoe-shaped tier that I call the Old Suburbs (Etobicoke, East York, North York, Scarborough, and the city of York) that prior to 1945 consisted mainly of scattered villages, and farm houses, which were then largely built over between 1946 and 1970. Only the northeast section of Scarborough remained to be built after 1970. Approximately 1,6 million people lived in these older suburbs in 1991. The third tier («New Suburbs») is formed by communities in the neighboring regional municipalities (Durham, Halton, Peel, and York). Again, these areas consist of what had been distinct towns and villages surrounded by farmland that have, since 1970, been increasingly built over as suburbs within the Toronto. Approximately 1.5 million people lived in the New Suburbs in 1991.

Map 3
EAs by earliest period of construction, Toronto arm, 1991

  • 13 See Smith and Moore (1993). The Metropolitan Toronto Planning Board is a second-tier planning juri (...)

31Between 1946 and 1970, much of the new housing construction took place in a dense belt of development adjacent to the city as built up in 1945. Some redevelopment of existing older areas did occur, notably along the subway lines, but the level of construction was relatively modest. At the same time, the pattern of suburban development was dense. With the exception of the northeast section of Scarborough, the land area of the Municipality of Metropolitan Toronto had largely been filled by 1971, the blank parts of Map 3 often containing park, commercial, office, and industrial land uses. In part, the reason for this dense pattern of development was that the Metropolitan Toronto Planning Board (as it was then known) had jurisdiction over land just outside the Municipality of Metropolitan Toronto and used its powers to encourage orderly land development and specifically to discourage development north of Metropolitan Toronto13. Extraterritorial jurisdiction came to an end in the early 1970s with the creation of regional governments in Halton, Peel, York and Durham.

32After 1971, metropolitan development changed in two main ways. First, suburban development became more scattered at a regional scale. The surrounding regional municipalities each had their own ideas about where and how to encourage development. Between 1971 and 1991 development occurred around a set of nodes that ranged from Mississauga and Brampton in the west to Woodbridge, Richmond Hill and Markham in the north, and Pickering and Whitby in the east. Second, at the same time, redevelopment in older areas became more widespread. By 1991, the result was much heterogeneity: many neighborhoods within the City of Toronto include substantial amounts of housing of various types and periods of construction.

33In what kinds of neighborhoods do CMA residents now live? Unfortunately, the size of Toronto does not allow us to present the CMA-wide equivalent of Map 2. Let us instead look at Table 6 which summarizes numerically the results of applying EA-aggregation. Consider first the left-hand column which totals all CMA residents. Overall, almost 60 % of CMA residents in 1991 lived in homogeneous neighborhoods. Among these, almost 60 % lived in neighborhoods built between 1971 and 1991; only 10% lived in neighborhoods that pre-date 1946. Further, over half of the residents of homogeneous areas lived in neighborhoods that consisted mainly of single detached dwellings.

Source: Calculations by author. See text.
Notes: EA aggregation used here. «Older Suburbs» includes cities of Etobicoke, North York, Scarborough, York and the borough of East York. <«Newer Suburbs» includes municipalities in regions adjacent to the Metropolitan Toronto: that is, Peel, Hal ton, York and Durham.
Table 6
Estimated population by neighborhood category, Toronto CMA, 1991

34These findings conjure up an image of the geographic pattern of homogeneous neighborhoods across the CMA. After all, the housing stock of the City of Toronto includes many row houses, duplexes, and small apartment buildings (and relatively few single detached dwellings) and, as has just been noted, has been home to much redevelopment in the past couple of decades. We might therefore expect to find that neighborhoods in the City of Toronto are more heterogeneous than elsewhere in the CMA. The three right-hand columns of Table 6 support this assertion. Only 42 % of City residents live in homogeneous neighborhoods compared with 62 % in the older and New Suburbs. Further, only 5% of City residents live in homogeneous neighborhoods of single detached housing, all pre-1946, and concentrated in just a few affluent districts: Rosedale, Forest Hill, Baby Point, and North Toronto. Many more pre-1946 homogeneous neighborhoods are in the «other dwelling» category, and are to be found in two broad areas extending a couple of kilometers east and west of the downtown core, and lying mainly south of Eglinton Avenue.

35And the City of Toronto has numerous clusters of homogeneous apartment neighborhoods. These clusters mirror the two principal planning initiatives of the day. Clusters built between 1946 and 1970 are found mainly near a few subway stations (e. g., Yonge-Bloor, Davisville, and Eglinton) in the northern half of the city. Largely built during the apartment boom of the 1960s, these clusters are in part the result of aggressive marketing of land near newly-opened subway lines by the Metro government. In contrast, clusters built since 1971 are more typically concentrated in the southern half of the City: along the Lake Ontario waterfront, and throughout the downtown core. Much of this development came as a result of incentives given by the City of Toronto to induce more housing in the downtown area.

36As has already been evidenced in our study of Scarborough, one key difference between suburbs developed before 1971 and those after is a consequence of the apartment boom of the 1960s. This difference is also evidenced in the CMA as a whole. As seen in right-hand columns of Table 6, there are more people in homogeneous apartment neighborhoods, and fewer in homogeneous single detached dwelling neighborhoods, in the Old Suburbs (built about the same time as the apartment boom) compared with the New Suburbs (that were built up afterwards).

37Additional insights can be made by constructing a table of spatial association akin to Table 4. In Table 7, such results are paneled separately for neighborhoods within the City of Toronto, elsewhere in the Municipality of Metropolitan Toronto (Old Suburbs), and the remainder of the Toronto CMA (New Suburbs). How do these findings compare with those reported for Scarborough above?

Source: Calculations by author. See text.
Notes: See notes to table 4. Table omits EAs with no or few dwellings: 400 in total.
Table 7
Neraly neighborhoods, Toronto CMA, 1991

38In one respect, the New Suburbs are like newer Scarborough. In both areas, post-1970 apartment neighborhoods are in large clusters. In the New Suburbs, for instance, these neighborhoods average 17.7 nearbys, and 55 % of these are in the same category (versus 16.8 nearbys, and 51%, for newer Scarborough). In contrast, the density of development of single detached and other dwellings (as measured by the count of nearbys) is much lower in the New Suburbs than in new Scarborough.

Map 4
Post-1970 neighbourhoods in northeastern Toronto, 1991

39In the case of Scarborough above, we saw evidence of the impact of planning after 1970 in mixing neighborhoods at the local scale. However, in the New Suburbs, the impact is different. Earlier, we saw how urban planners after 1970 undertook to fill in the small remaining area of northeastern Scarborough with development that mixed dwelling types at the local scale. However, this practice was not followed by planners in the New Suburbs. Panel (c) of Table 7 confirms that 1971-1991 single detached subdivisions are very large in the New Suburbs; 64% of the nearbys for such neighborhoods are of the same category, compared with just 30% in the Old Suburbs. Map 4 provides further evidence of this difference; note how the clusters of single detached neighborhoods that postdate 1970 are much larger in Markham, Richmond Hill, and Vaughan than in Scarborough. These differences between northern Scarborough and the New Suburbs serve to emphasize how different planning practice can be from one jurisdiction to the next. It should also serve to make us cautious about the interpretation of patterns at the national level.

V – Empirical results: Canada

40Now, let us move up to the scale of Canada as a whole. Assigning each of Canada’s 45,995 EA centroids in 1991 to a neighborhood type, based on EA-aggregation, we get the results summarized in the left hand column of Table 8. Just under one-half of all Canadians live in homogeneous neighborhoods, and almost one-half of these (6.128 million) live in category 18 (homogeneous, single detached, 1971-1991). In contrast, very few Canadians (0.828 million in all) live in homogeneous neighborhoods consisting mainly of apartment buildings.

41Comparing the left hand column each of Table 6 and Table 8 permits us to look at the Toronto CMA in the context of the rest of the country. The Toronto CMA is 14% of the population of Canada. Residents of Toronto are slightly more likely than Canadians overall to live in a homogeneous neighborhood. However, this overall similarity is misleading; the Toronto CMA, along with other large cities in Canada, is not like much of the rest of the nation. The differences are most marked in category 12 (homogeneous, apartment, 1971-1991), where Toronto accounts for 336 of the 553 thousand such residents across Canada, and category 13 (homogeneous, other dwelling, pre 1945), where Toronto accounts for 142 of the 503 thousand such residents nationally. In contrast, the Toronto CMA accounts for only 61 of the 668 thousand Canadians living in category 16 (homogeneous, single detached, pre 1946). Such data suggest that, even among the largest cities, there are important differences. Montréal, for example, has substantially more neighborhoods of category 13 (homogeneous, other dwelling, pre 1946) and 14 (homogeneous, other dwelling, 1946-1971) than are found in Toronto, and substantially fewer heterogeneous neighborhoods. This does not necessarily mean, however, that Montréal has experienced less redevelopment than Toronto. We need to keep in mind that single detached housing makes up a smaller proportion of the Montréal housing stock, and the caveat mentioned earlier that the link between heterogeneity and redevelopment is best evidenced in neighborhoods that start out as mainly single detached housing. Vancouver is different again. With some significant exceptions, most notably some densely-clustered neighborhoods in the West End that are mainly category 11 (homogeneous, apartment, 1946-1970) and in Kitsilano that are mainly category 15 (homogeneous, other dwelling, 1971-1991), neighborhoods in Vancouver are for the most part heterogeneous.

Source: Calculations by author. See text.
Note: EA aggregation used here.
Table 8
Estimated population by neighborhood category, Canada, 1991

42In this paper, the Toronto case study hints that there are important distinctions among cities, towns, and villages. In the right hand columns of Table 8, EAs are further subdivided based on the kind of area in which they are found. I follow Census usage here. The Census distinguishes among larger urban areas (Census Metropolitan Areas and Census Agglomerations), other urban areas, and rural areas. Further, CMA/CA areas are subdivided into an urbanized (continuously built-up) core, self-standing urban communities in the metropolitan fringe, and the remaining rural fringe. Residents of the urbanized core-that is, the continuously built-up areas of Canada’s larger cities-considered alone make up 2/3 of Canada’s population, yet account for aver 3/4 of all residents in homogeneous areas. And, even this understates the differences when we make comparisons by neighborhood category. Looking down Table 8, note that there are only two homogeneous categories in which the urbanized core is not overwhelming: category 16 (homogeneous, single detached, pre 1946) and category 18 (homogeneous, single detached, 1971-1991). These patterns repeat themselves if we look instead at heterogeneous neighborhoods.

43What about neighborhoods in the rest of Canada? In the urban and rural fringes of CMAs and CAs, about one-hall of neighborhoods are homogeneous, and these are mainly single detached built since 1971. For the most part, the latter are found in rural areas, and take the form of exurban developments. We can make clear the effects of proximity to larger cities if we now look at the two right-hand columns of Table 8. In the urban and rural areas that are beyond the CMAs and CAs, only about 1/4 to 1/3 of neighborhoods are homogeneous (many of which are new single detached). We can summarize these patterns as follows. Outside the urbanized cores of Canada’s larger cities, there is only one common dwelling type (single detached), and much heterogeneity. Where there is homogeneity, in the smaller towns and rural areas, it largely takes the form of new development in rural areas remote from cities (1.095 million persons), exurban development nearer to cities (1.048 million), or pre-1946 rural area settlements that have not undergone redevelopment (0.301 million). Within the latter, the densest concentrations of category 16 (homogeneous, single detached, pre 1946) are to be found in the farming regions of southwestern Ontario well away from the larger cities.

Source: Calculations by author. See text.
Notes: See notes to table 4. Table omits EAs with no or few private dwellings: 5, 963 in total.
Table 9
Nearby neighborhoods, Canada, 1991

44What additional insights can be gained by constructing a table of spatial association for Canada akin to Tables 2 and 5? In Table 9, results are paneled separately for the five subdivisions of EAs identified in Table 8. Three differences stand out. First, only in the urbanized cores of CMA/CAs (and to a much lesser extent in the urban fringe of CMA/CAs and other urban areas) are neighborhoods in close spatial proximity. In contrast, the number of other neighborhoods within one kilometer is negligible in the rural fringe and other rural areas. Second, the clustering of apartment buildings is a pattern common to both 1946-1970 and 1971-1991; in each period, about one-third of nearbys are in the same category. Third, the spatial pattern of development is much more homogeneous in larger cities. In the homogeneous, single detached, post 1970 category, 54 % of the nearbys are of the same category in the urbanized cores and urbanized fringes of CMA/CAs; compared to only 21 % in other urban areas. A similar story is evidenced if we compare the siting of homogeneous neighborhoods of single detached housing after 1970 between the rural fringes of CMA/CAs and other rural areas. In the former, 69% of nearbys are in the same category, compared with only 33 % in the latter.

Conclusions

45This paper has considered a typification of neighborhood based on the characteristics of the housing stock found nearby. The method makes use of data for 45,995 EAs available from the 1991 Census of Canada. In effect, each EA centroid is treated as a point from which to reconnoiter the housing stock within a radius of 400 meters. In all, 18 categories of neighborhood are used based on principal type of dwelling (type of dwelling by period of construction), and degree of homogeneity. This paper finds that about one-half of all Canadians overall live in homogeneous neighborhoods, hut that homogeneity varies substantially with urban status and size of community.

46The purpose of this study has been to examine the feasibility of using geocoded census data to look at neighborhood experience. This is problematic for several reasons. First, we do not have data on individual dwellings; we know only about EA aggregations. Hence, our estimates of the characteristics of housing stock within a given circle is crude. Second, we are limited to the small set of housing characteristics covered in the census. Third, we simply do not know anything about nonresidential land use nearby.

47In spite of these evident limitations, there is much that remains to be done in exploring this approach. First, we should incorporate other aspects of dwelling into the neighborhood categorization: for example, dwelling size and state of repair. Second, we should include variables that measure social mix in the neighborhood: e. g, age of residents, presence of children, income, and immigration status. Third, we might consider how the neighborhood experience differs among social groups. Fourth, we might consider how the neighborhood experience has changed in recent decades.

Bibliographie

References

Bourne L. S., «The changing settlement environment of housing», in Miron J. R. (ed.), House, Home and Community: Progress in Housing Canadians, 1945-1986, Montréal, McGill-Queen’s University Press, 1993, p. 271-288.

Goldsteen J. B., Elliott C. D., Designing America: Creating Urban Identity, New York, Van Nostrand Reinhold, 1994.

McKellar J., «Building technology and the production process», in Miron J. R. (ed.), House, Home and Community: Progress in Housing Canadians, 1945-1986, Montréal, McGill-Queen's University Press, 1993, p. 136-154.

Miron J. R., Housing in Postwar Canada, Kington and Montréal, McGill-Queen's University Press, 1988.

Smith P.J., Moore P. W., «Cities as a social responsibility: planning and urban form», in Bourne L. S. and Ley D. (eds), The Changing Social Geography of Canadian Cities, Montréal, McGill-Queen's University Press, 1993, p. 343-366.

Urhahn G., Bobic Μ., A Pattern Image: A Typological Tool for Quality in Urban Planning, Bussem, Netherlands, TROTH Publishers, 1994.

Notes

1 This research has been funded by the Social Sciences and Humanities Research Council of Canada (410-94-0684). The outstanding efforts of my research assistants, R. Basil and C. Ho. are gratefully acknowledged.

2 In this paper, «type of dwelling» refers to structural form of the dwelling. Three dwelling types are considered: single detached dwellings, apartments in buildings of five storeys or more, and all other dwellings.

3 A blockface is defined to be one side of a street between two consecutive critical points: a critical point being an intersecting street, a cul-de-sac, railway, river, or other feature.

4 Bossons (1993) advances the argument that modern planning is usefully perceived as an outcome of a market for public action.

5 Hereinafter, the term «kind» is used to indicate a breakdown by dwelling type or period of construction.

6 A similar distinction is drawn by Goldsteen and Elliot (1994, p. 128-136).

7 Prior to the 1971 Census, type of dwelling was recorded by the census taker.

8 Note that prior to 1946, there were few apartment buildings more than five storeys tall. Much of the older apartment building stock was in the form of three storey walkups.

9 All EAs shown are part of Federal Electoral District 78.

10 «Lie within it» taken here to mean that the dwelling centroid lies within the circle. In the case of a dwelling near the edge of the circle, part of the dwelling (however defined) itself might lie outside the circle.

11 The neighborhood assignments in this map take into account the housing characteristics of centroid that lie just outside Scarborough but within cicles for EA centroids within Scarborough.

12 Here, «south» includes all Scarborough neighborhoods with a centroid UTM Northing smaller than 4848432. The set of neighbors includes neighborhoods north of this line.

13 See Smith and Moore (1993). The Metropolitan Toronto Planning Board is a second-tier planning jurisdiction; cities and boroughs within its jurisdiction had to bring their plans into conformity with the Metro plan. The Municipality of Metropolitan Toronto itself was created in 1953. Prior to that, each city, town and village within the Toronto area did its own planning independently.

Table des illustrations

Légende Source: See text for detailsTable 2Categorization and housing stock in hypothetical neighborhood, 1990-1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-1.jpg
Fichier image/jpeg, 150k
Légende Map 1Southwest section of Scarborough, Ontario, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-2.jpg
Fichier image/jpeg, 271k
Légende Source: Culumns (1) through (6) assembled from machine-readable 1991 census data files. Culumns (7) and (8) computed by author. See the text for details. Neighborhood categories listed in Table 1.Note: All EAs located in Federal Electoral District 78.Table 3EA and blackface data for 400 meter around EA 212, southwestern Scarborough, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-3.jpg
Fichier image/jpeg, 207k
Légende Map 2Neighborhoods (EA aggregation) Scarborough, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-4.jpg
Fichier image/jpeg, 401k
Légende Source: Calculations by author. See text.Notes: EA aggregation used. «Same» is same neighborhood category. «Het-hom» is otherwise equivalent but neighbor opposite in homogeneity. Homogeneous neighborhood categories instanced in fewer than 10 EAs are not shown here. Table omits EAs with no or few private dwellings: 37 in total.Table 4Nearby neighborhoods, Scarborough, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-5.jpg
Fichier image/jpeg, 165k
Légende Source: Calculations by author. See text.Note: Neighborhood categories instanced in fewer than 10 EAs are not shown here.Table 5Estimated population, private dwellings, and Enumeration Areas by neighborhood category, Scarborough, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-6.jpg
Fichier image/jpeg, 243k
Légende Map 3EAs by earliest period of construction, Toronto arm, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-7.jpg
Fichier image/jpeg, 260k
Légende Source: Calculations by author. See text.Notes: EA aggregation used here. «Older Suburbs» includes cities of Etobicoke, North York, Scarborough, York and the borough of East York. <«Newer Suburbs» includes municipalities in regions adjacent to the Metropolitan Toronto: that is, Peel, Hal ton, York and Durham.Table 6Estimated population by neighborhood category, Toronto CMA, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-8.jpg
Fichier image/jpeg, 250k
Légende Source: Calculations by author. See text.Notes: See notes to table 4. Table omits EAs with no or few dwellings: 400 in total.Table 7Neraly neighborhoods, Toronto CMA, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-9.jpg
Fichier image/jpeg, 272k
Légende Map 4Post-1970 neighbourhoods in northeastern Toronto, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-10.jpg
Fichier image/jpeg, 183k
Légende Source: Calculations by author. See text.Note: EA aggregation used here.Table 8Estimated population by neighborhood category, Canada, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-11.jpg
Fichier image/jpeg, 236k
Légende Source: Calculations by author. See text.Notes: See notes to table 4. Table omits EAs with no or few private dwellings: 5, 963 in total.Table 9Nearby neighborhoods, Canada, 1991
URL http://books.openedition.org/pul/docannexe/image/9782/img-12.jpg
Fichier image/jpeg, 334k

Auteur

Centre for Urban and Community Studies - University of Toronto, Canada.

Le texte et les autres éléments (illustrations, fichiers annexes importés) sont sous Licence OpenEdition Books, sauf mention contraire.

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search