Yea Tes 1972

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Environment and Planning, 1972, volume 4, pages 395-414

The congruence between housing space, social space,


and community space, and some experiments
concerning its implications

M Yeates
Department of Geography, Queen's University, Kingston, Ontario, Canada.
Received 20 March 1972; revised version 30 May 1972

Abstract. Housing space, social space, and community space are each defined simply and discussed
with respect to two concepts: housing space is defined in terms of type and quality for individual
housing units; social space is defined in terms of social status and stage in the life cycle for a family
unit. Contrasting with these micro-scale definitions are the constructs of community space—family
status and social status—which describe census units at the macro-social level. The congruence
between housing space and social space for a sample of homes and the families residing within them
in Kingston and Winnipeg, Canada, is then determined statistically with the use of multivariate
multiple regression procedures. The varying degrees of incongruence for a number of social groups
are then analysed subjectively under three defined housing constraints. Finally, the representativeness
of the two samples is examined, and directions for further research are outlined.

Urban areas are fundamentally collections of people who create, react to, and change
their environment. Buildings and other physical structures imbue the city with a
certain form and shape, but it is the people who give it life and colour and, in the
end, have the power to decide its fate. A vital element in this fabric of people
interwoven with structures is the residential pattern of the city for it is their home
which commonly reflects or forms the social, economic, and psychological image of
the family (Cooper, 1971a). The residence chosen should thus reflect the 'aspirations'
of the household (Timms, 1971, p.97) and the degree to which it does not provides
an indicator of the 'stress' (Brown and Moore, 1971) borne by the family involved
and, in the aggregate, by society as a whole. This paper is therefore concerned with
developing a methodology for determining the 'aspirations' of families with respect to
housing in order to determine the degree to which families are incongruent in terms
of their domicile (Michelson, 1970, pp.26-27). Operationally, the study involves an
analysis of a sample of houses in two Canadian cities: Winnipeg and Kingston.
Congruence is determined statistically, and the extent of incongruence for various
social groups is examined with respect to three constraints which are postulated on
the basis of a priori information.

1 Micro- and macro-social models


Although this study is at the micro-social level (Timms, 1971, p.85), the connection
between this approach and its apparent complement at the macro-social level has
been enunciated clearly by Rees (1970, p.313). Rees is basically concerned with
discerning the existence of spatial order in the social geography of cities. He suggests
that the type of community (community space) in which a family resides is related
to the social status and stage in the life cycle of the family (social space) and the
type and quality of housing that the family requires and can afford (housing space).
Furthermore, Murdie (1969), among others (Timms, 1971, pp.229-244), has
provided strong empirical evidence for the suggestion (Berry, 1965) that a position in
community space implies a distinctive location in real physical space. The units
involved in the calibration of the properties of social and housing space are at the
396 M Yeates

micro-social level (families and residential units), while those of community and
physical space are at the macro-social level (census tracts, enumeration areas). Each
one of these four typologies is defined in a particular way, and these definitions are
crucial to this study.
1.1 Community space
The community space typology locates data aggregates, which have discrete locations
in urban areas, with respect to two dimensions or axes. These axes are derived from
the repetitive empirical conclusions of factorial urban ecology which emphasize the
existence in individual studies of some or all of four major dimensions of urban
social differentiation—socioeconomic status, famility, ethnicity, and mobility (Timms,
1971, pp.56-58; Berry, 1971, pp.220-313). The conclusions from these studies
support the generality of the social area constructs defined by Shevky and Bell (1955)
and Shevky et al. (1949), which are social status, family status (or urbanization), and
segregation. The community space model in this analysis incorporates only the first
two constructs, for, while ethnicity is of some limited importance in social
differentiation in Winnipeg, it is of negligible importance in Kingston. Furthermore,
in neither urban area do variables pertaining to mobility prove to be of sufficient
independent importance to constitute a separate dimension, for they are not high
growth cities in the Canadian urban scene. However, further studies in other urban
areas should take these factors into account.
1.2 Social space and housing space
In a similar fashion, the social space typology defines the position of an individual or
family in terms of two properties—social class and stage in the life cycle. Social class
is usually specified operationally in terms of occupation, education, and income, and
the number of classes varies in different studies (McDavid and Harari, 1968, p. 104).
In this study, the social class axis is viewed as a continuum along which families or
individuals are arranged, and thus 'high class' means simply high on the defined scales
for the Winnipeg and Kingston samples of families. Porter (1965) has demonstrated
the existence in Canada of a hierarchy of families according to socioeconomic status,
and the impact of this variable on residential location is clear from the observation
that families of a given class usually move to and stay within an area of equivalent
social status within the city (Simmons, 1968). Furthermore, it is evident that social
class, and particularly level of education, is closely related to the quality of housing
demanded (Feldman and Tilly, 1960; Tilly, 1961).
Differing housing requirements as individuals progress through the life cycle is also
a major factor causing people to change and select particular houses and residential
locations, and can be considered a second independent dimension of a family's social
space. Rossi (1955, p. 175) suggests that this is particularly related to living space,
for, in his study, "among the three out of every five whose moves were undertaken
voluntarily, the most frequently encountered motive for the shift lay in dissatisfaction
with the amount of room available in their old dwellings". Krieger (1969), following
the work of Erikson (1968) and Maslow (1968), suggests that the life cycle should be
the single most important social indicator used for developing social policy in urban
space. An individual passes through many stages in his life, which Erikson has
categorized as: infancy, early childhood, play age, school age, adolescence (teenage),
young adulthood (marriage), adulthood (parents), and old age. As the average length
of life continues to increase, and the birth rate to decrease, to these 'eight ages' may
be added late adulthood (with children having left home), sometimes referred to as
the 'swinging fifties'.
Although each of these stages gives rise to different problems to which social
policy must be directed, the life cycle is important with respect to residential location
Congruence between housing space, social space, and community space 397

in at least three ways. First, as an individual passes through the life cycle, the type of
house and the physical location that is optimally required to cater for the needs of
that particular stage varies. Second, most of the residential location decisions are
made by the heads of the family with respect to their perception of the requirements
of the family. These requirements may be difficult to reconcile or compromise
because of the variations in stage within the life cycle in which the family members
find themselves. For example, the location and space requirements of a teenager
differ considerably from those in childhood, and great 'stress' may be created within
a family when a teenager who wishes to be close to the wide array of experiences of
big city life finds himself physically situated in a suburb of a big city. Third, the
length of each stage varies—five stages are crammed into the first twenty years of life,
and three throughout the next fifty. As a result the decision makers in the family
may become less responsive to life cycle changes. This is frequently evidenced by a
resistance to residential adjustment in later years due to sentimental attachment to a
house and location that is no longer contiguous with the family's requirements. In
this particular study these various aspects are examined in whole or part as we attempt
to calibrate a family's position in the life cycle and examine its interrelationship with
its location in community space and housing space.
The housing space model comprises two independent dimensions (or axes) with
respect to the physical characteristics of the house. Each dwelling unit can be
described with respect to the type of home and its value and quality. Type refers to
whether the unit is an apartment, row house, detached single family dwelling, and so
forth. The inference is that the type of dwelling unit that a family occupies is very
much related to its stage in the life cycle (Lansing et al., 1969, pp.21-28). For
example, after a thorough analysis of the available evidence, Michelson (1970, p.101)
suggests that parents with young children overwhelmingly desire single family housing,
and Lansing et al. (1969, p.26) note a very positive effect of the life cycle on the
percent of those moving into new homes who move into single family homes (table 1).
In contrast, childless adults do not have the same internal and external space
requirements and, as a consequence, tend to live in apartments with more central
locations.
The second axis of the housing space typology—the value and quality of the
home—is related to the social status of the inhabitants. In particular, education,
which is an important determinant of social status, influences tastes, judgement and
the lifestyle (Tilly, 1961). Feldman and Tilly (1960) indicate that education is a
better determinant of housing preference than income, and the life style led by the
family is an illuminating derivative of this variable. This aspect of social status is
particularly important when the value and quality of homes inhabited by workers of
Table 1. The effect of the life cycle on the percentage of those moving into a new home who
move into single family housing (after Lansing et al., 1969, p.26).
Stage in life cycle Those who move into new single family homes
as percentage of those who move into new
homes in each stage of the family life cycle
Young, single 6
Young, married, no children 27
Married, youngest child under 5 70
Married, youngest child 5-18 84
Over 45, married, no children 38
under 18
Over 45, not now married 15
(single, widowed, divorced)
All families 51
398 M Yeates

comparable incomes but different educational levels are compared. White collar
workers live in more expensive homes and maintain their homes better than blue
collar workers of similar income levels because they place a higher utility on the
quality of their residence (Duncan and Duncan, 1957; Wheeler, 1968).

2 The hypotheses
The typologies can, therefore, be linked conceptually into a single framework, for a
family's position in social space should have a direct housing space counterpart, and
these two should be linked via the social area constructs of social and family status
to community space, the units of which have distinct locational characteristics (Berry
and Rees, 1969). These associations can be expressed symbolically as:
^ = Fx(Sk) (1)
Cf =F2(Ht,Sk) (2)
where C7- = the position of the /th census tract in community space; Hi = the
position of the /th house in housing space; and Sk = the position of the /rth family
in social space. In situations of this kind where the symbols refer to vectors a
common procedure is to evaluate the interrelationship statistically using univariate
multiple regression techniques. In this case, however, the symbols refer to matrices,
not vectors, and so multivariate regression techniques are required (Anderson, 1958).
Thus, if there are n observations and m variables (axes or dimensions) for each
matrix
Hn,m = (S)n,mBmym (3)

and
=
^w,m \H ,S)n2m^2m,m (4)

where B is a matrix of regression coefficients. The hypothesis can therefore be


rephrased as whether B =£ 0; in other words, we are seeking to reject the null
hypothesis that the matrix B is a zero matrix. Anderson (1958, p.196) indicates that
this hypothesis can be tested using the F distribution with 2m and 2(rc*— 1) degrees
of freedom, where n* = n — 2m. It should be noted, however, that as the axes of the
matrices are hypothesized to be orthogonal, half of the appropriate elements in the
B matrix should be zero.

3 Calibration of the matrices


3.1 The macro-social models
The axes of the community space models have been calibrated from similar sets of
variables for Winnipeg, using 1961 census tract data (Nicholson and Yeates, 1969),
and Kingston, using 1961 enumeration area data (Hallsworth, 1971). These calibrations
involved the principal components method (Rummel, 1970, p.l 12; Rees, 1971, p.221),
in which the unrotated matrices of factor loadings gave very clear descriptions of the
major underlying dimensions in both cases. The first two dimensions in each study,
which accounted for 55% of the variation in the original data matrix in the Winnipeg
case and 43% in the Kingston study, describe dimensions corresponding closely to the
social status and family status constructs of the social area analysts. The standardized
component scores of the 84 census tracts and 94 enumeration areas on these dimensions
provide the necessary calibrations for locating the position of these units in community
space. The other significant dimensions in the Winnipeg study described clusters of
different ethnic groups, but in the Kingston case they did not appear to consist of
easily interpretable common clusters of variables. This may well be due to the
Congruence between housing space, social space, and community space 399

different levels of aggregation involved; hence metropolitan Winnipeg, which had a


population of almost half a million in 1961 and was ten times larger than Kingston,
is represented by over eighty census tracts in 1961, whereas Kingston is represented
by, appropriately but uselessly, one tenth that number. These different levels of
aggregation may well result in data units being succinctly described by just one or two
variables in the Kingston case, due to the small size of the enumeration areas.
The distribution of these data units in community space and their location in
physical space are indicated in figure 1. For clarity of mapping, the census units
have been grouped into four main regions as a result of the division of each scale into
two classes, high (H) implying above average and low (L) below average. The physical
space maps reveal some interesting general patterns: (i) the high social status areas in
both cases are to the south and west of their respective central business districts,
whereas the low status areas are in the center of the city and to the north; (ii) in
both cases the high family status areas are on the periphery of the urban areas
surrounding the low family status areas in the center of the city; and (iii) a certain
element of concentricity is evident with respect to family status in each instance, and
a fledgling sectoral pattern of social status can be discerned.
3.2 The micro-social models
Every family and every house (dwelling unit) has a distinct position in housing space
and social space, and there are many individual characteristics of houses and families
that are summarized by the axes of these models. There are two main problems
concerned with the calibration of these particular matrices. The first is that, as data
of the required kind are not available through census reports because of disclosure
rules, direct interview/observation techniques must be used on a sample of dwelling
units and families. Sampling of this kind is an expensive procedure, and so the sizes
of the samples are subject to budgetary limitations. As this is essentially an
exploratory study, the sizes of the samples are small. The Winnipeg sample, which
was taken in August 1970, was a random survey of 90 families and residential units,
or about 1 in 1000 of the families in the metropolitan area. The Kingston sample,
which was taken in May 1971, was a systematic clustered sample of 175 families and
their respective homes, which represents about 1 in 100 of the families in the city.
The second problem concerns the mix of quantitative measures that define the
variables used to determine the axes of the models. The interview/observation sheet
is short and the samples used contain complete information for all the questions.
From this questionnaire/observation sheet fifteen variables, the maximum number of
meaningful characteristics that could be obtained from the sheet, are derived (table 2),
some of which involve rank-order measurement, and others an interval scale. This mix
of measurement scales creates methodological problems when the association between
the variables is to be analysed. Following several experiments concerning different
measures of association, it was decided to use the Pearson product-moment correlation
coefficient because of its greater generality and the fact that it replicated fairly well
some of the other measures investigated (for example, Spearman's and Kendall's rank-
order coefficients).
3.3 Analysis of the data
Thus fifteen variables for each sample of residential units and families comprise the
data from which the axes of the housing and social space models are determined, and
the position of the units along these axes calibrated. If each of these models and
each of these axes are individually distinguishable, the data should reveal this through
principal components analysis. Accordingly, the 90 x 15 and 175 x 15 data matrices
were so analysed to determine whether the observation units can, in fact, be
calibrated clearly along four dimensions consisting of 'quality' of home, 'type' of
400 M Yeates

Family status

CW
1 45« • 58
,60
1 «46
. 48.
• 17 84 •81
79«-1/
•80
43 • •66 . ? 2

1 ©23 *65
•39 # 6 2
28 «61
67
74 • •83 «.63 .60
•20 82» 44 * •
#
, »73 • 7__•
7
«85 •eg
•22 ,30 • 26 -40 78 .86
1 «38 31
62« 17«c§>
87« i?«54
•37 42- 14. .13 , • *1 1^56
8# • . 15 68 27 , 5 3 -fc70
•33 41 • { «49
19»#35 .32 50*7 #
16
•25 .3 7 6 . -B9
•34
• 21 •24
•9 •6
•36 • 4

•5

• 20
11. • 10

12e
•18

Figure 1. Community space (C) and physical space (P) diagrams for Winnipeg (W) and Kingston
(K), 1961. Note that the physical space map for Kingston (PK) is transformed so that the area of
each enumeration unit is proportional to its population density, whereas the map for Winnipeg is
not transformed.
Social status

^
£ in. co
JI
M
u *- •
0.*s |
= 2. 00 J
en o
CO •
*"• oo en

•52
en •
en
• «> o>
o O •
'i o
• CD
in
«-•
5 -•
CM*
• '•
o r-
CD, in
cn en
<• CO ©
o • ••
•* s • » •• CD. m
^t ID CD ° 0 0 .2 o
in "00 • 2>
°.
co« •*
- en. >
P0 •
in* •<-, CM* • • 00 CO

• CO • CO °°
CO. CO • o
CM -00
00 •
o » co •
CD 00 CM
CM co TJ- 'in
5- co*.
CO CO


CO«

5

• o

en

i
402 M Yeates

Table 2. Variables derived from questionnaire concerning housing space and social space.
Variable Variable Question Measurement
number name number

1 Education, 6 number of years spent in


male school (4- college), for
example 12 means grade 12
2 Education, 6 number of years spent in
female school (4- college), for
example 12 means grade 12
3 Occupation,3 6 rank order:
male profession and technical 1 high status
managerial and proprietary 2
sales 3
clerical 4
craftsmen 5
production operatives 6
labourers 7 low status
unemployed 8
retired 9
housewife 0
4 Occupation, 6 rank order: see scale for
female variable 3
5 Income, 10 rank order:
male >$20000 1
(head of $14000-$20000 2
household) $10000-$14000 3
$ 6000-$10000 4
$ 4000-$ 6000 5
< $ 4000 6
6 Income 10 rank order: see scale for
total variable 5
(all earners)
7 Age of heads 5 rank order:
of family older than 50 1
younger than 30 2
between 30 and 50 3
8 Family size 3 number of persons residing
in the home
9 Family age 4 median age of the children
10 Home 9 rank order:
ownership renting 1
mortgage 2
owning 3
11 Trice' of the 9 either
home (1) rent per month
(2) mortgage payments 4-
property tax per month
or (3) an estimate of the
rent per month that could
be obtained
12 Space ratio number of rooms/number
of persons in familyb
Congruence between housing space, social space, and community space 403

Table 2 (contd)
13 Living conditions 2 rank order:
luxurious 10
comfortable 20
adequate 30
run-down 40
blighted 50
14 Type of home 1 rank order:
high rise apartment 1
low rise apartment 2
row house 3
town house 4
duplex 5
single family 6
15 Lot size 1 rank order:
no individual lot or private 10
yard
small lot and miniscule 20
yard
small lot and yard 30
medium lot and yard 40
large lot and yard 50

Occupation ranking based primarily on Blishen (1967).


Number of rooms excludes bathrooms.

home, stage in the life cycle, and social status of the family. Although the first four
dimensions in each case cannot be interpreted clearly in an unrotated form, a varimax
rotation of the four factors yields a very clear structure that can be interpreted with
some ease (table 3).
The first rotated dimension for both Winnipeg and Kingston, accounting for
20-2% and 23-2% of the variation in the original data matrices respectively, comprises
the income of the family, the living condition (quality) of the house, and the price of
the home, which together describe the 'value and quality of the home'. The analysis
emphasizes the important fact that the living condition of the home is related to the
price (or rent) paid which is determined in large part by the income of the family.
Thus the quality of the home, which influences particularly the health of young
children (Wilner et al., 1962; Martin, 1967) and the important 'socializing' effect of
property in a neighborhood (Smith, 1970, pp.76-83) is related to the price that can
be paid by the family—a point which is also recognized by Muth (1967, pp.15-16)
with respect to substandard housing. In a social policy context the quality of
housing in which a family resides may well be susceptible to improvement through
such methods as income support (for example, a negative income tax) providing that
certain irrationalities in the housing market with respect to the financing of poor
quality housing can be resolved (Grigsby, 1967).
The second rotated dimension, which accounts for 16-4% and 18*3% of the
variation in the original data matrices, is composed primarily of variables pertaining
to the lot size, the type of the home and the status of the tenancy—variables
commonly grouped together in the residential preference literature (Menchik, 1971,
p.27). In general the loadings indicate that, for example, single family homes tend to
be located on lots of reasonable size and are owned outright or are being purchased
through mortgages. On the other hand, apartments and row houses that have little
exterior space are generally rented. Thus these two dimensions define the basic axes
of housing space, and each dwelling unit can be calibrated with respect to these
404 M Yeates

dimensions by calculating the standardized factor scores and its position determined
in housing space. Perusal of figure 2 indicates that the nature of Kingston's housing
space differs radically from that in Winnipeg. Whereas all types and qualities of
housing seem to be represented in the Winnipeg sample, high quality apartments, low
quality apartments, and low quality single family housing are less evident in the
Kingston case. The implication of this distribution is either that such housing is
scarce in Kingston, or that the sample is biased, and the latter implication is the more
plausible. This bias does not, however, affect the analysis adversely with respect to
contiguity, though it does adversely affect any statements that are made with respect
to Kingston as a whole.
The axes of the social space model are determined with equal clarity (figure 3).
The third dimension (table 3), consisting almost exclusively of variables indicating the
size of the family, the space it occupies, the median age of the children, and the age
of the parents, comprises a melange of characteristics which together describe the
stage in the life cycle of the family. At one end of the scale are childless adults: they

Table 3. Winnipeg and Kingston3 samples: varimax rotation loadings1


Variable Wealth and Type of life Social
quality of home cycle status
home
Total income 0-8936
0- 7612
Income, male 0-8484
0-8126
living condition 0-6531
0-7167
Price of home 0-6300
0-8009
Lot size 0-8998
0-6137
Type of home 0-8587
0- 7239
Home ownership 0-7501
0-8612
Family size 0-7701
0-9111
Space ratio 0-6521
0-7163
Age of family heads 0-5998
0- 7477
Family age 0-5314
0-8177
Education, male 0-7859
0- 7239
Education, female 0-7113
0-8225
Occupation, female 0-6232
0- 7436
Occupation, male 0-5884
0- 6502
Proportion of variance (%) 20-24 16-39 12-16 13-98
23-21 18-33 16-93 16-30
Factor concentration (%) 17-44 85-88 91-00 88-32
68-80 89-72 94-45 88-98
a
b
Figures pertaining to Kingston are italicized; those pertaining to Winnipeg are upright.
All signs and loadings < 0-5 excluded.
Congruence between housing space, social space, and community space 405

Type of home
oTT
• C
•5 I =
CO

HW •
I
C O

o) g> i
• •

• • •
• ,_ a x j
o .2 I

• •* • g«tl
• • • z> a®
o
•• 4. °>co
2 OCT
• *• •• • E ><D
O <r"D
• *• • •• * • •••••• " <
• •
• •
1 • • • • ••
1i •

% • •
• • • • • •
• • •

1 • •
1 • • CD 5"°
E °£

• m

APARTMENTS
Rent
ITOWN HOUSESh—SINGLE FAMILY HOUSES -
1 Mortgage l Own
ill
- No External —hSome External —f— Small Lot 1 Medium Lot H- Large Lot -
Space Space

Type of home


• "IS
HK •• #
•• • ••
• x "Si
• •• .1^1
• •
• * • .

1 • ••«•• • •
I •
. •
• • •
1 • • -• »• •• •
1 * • :• • •
. •••*
•• • < **.
• * ••
••i . *

• * ••• • •

if'*
• : • I—I CDO

T
CD
OO
$QC

1 •
1 •

-APARTMENTS TOWN HOUSES *- -SINGLE FAMILY HOUSES


No External - ; RentExternal
- Some • •+- — Mortgage —
. Medium ;—;—Own
Lot +- Large• Lot H
Space Space

Figure 2. Housing space in Kingston (HK) and Winnipeg (HW).


406 M Yeates

Stage in the life cycle

CHILDLESS ADULTS—^SINGLE CHILD MEDIUM SIZED hLARGE — I


FAMILY FAMILY FAMILY
i Less Than One Person PerRoom . More Than One Person Per Room
. Young Children _ Some Teenagers |
-+-

Stage in the life cycle

SK
111
>-P
cr a.
p _•
Ml .2

I roco
P-=
• •

• • •• • •
lO w

pi ro
p
PQ ©
£
UJ O

II
CHILDLESS ADULTS—^SINGLE CHILD ^- MEDIUM SIZED l-LARGE — I
FAMILY k j FAMILY^ n FAMILY
-Less Than One Person ~
Per Ro
Room . _|. More Than Cme.person Per Room
Young Children _Some Teenagers .

Figure 3. Social space in Kingston (SK) and Winnipeg (SW).


Congruence between housing space, social space, and community space 407

may be older than 50 (children left home) or younger than 30, they have one or more
rooms per person, and of course, the family size is small (two or less). At the other end
of the scale are adults with children: the adults are usually between 30 and 50 years
old, they tend to have a number of children; as the number of children increases so
does the median age of the children, and the number of rooms per person decreases.
The fourth dimension defines a factor which can be labelled, without question,
social class. It consists almost entirely of variables relating to years of education and
rank of occupation for both male and female heads of households. The importance
of the education component of this factor in the analysis of contiguity has been
emphasized by Michelson (1970, p. 130) who suggests that "... the percentage of income
that people will spend on good housing varies primarily according to their education".
In other words, the higher the level of education (and therefore amenity expectation)
a person has, the more likely he is to require comfortable or even luxurious
surroundings. With respect to type of home there is some evidence to suggest that
families at the lower end of the socioeconomic scale may seek out detached homes
to answer the problems of dirt, overcrowding, and safety (Clark, 1966; Rainwater,
1966). Thus one would expect a strong relationship between the house quality and
social status axes, and perhaps a weak relationship between social status and house type.
Thus it is apparent that the axes of both the housing and social space models for
Winnipeg and Kingston can be determined with a high degree of clarity using principal
components analysis. A varimax rotation of the first four factors provides an easily
interpretable solution, and thus the associated factor scores can be used for scaling
the dwelling and family units along each axis (Rummel, 1970, p.30). Additional
experiments with the data indicated that the extraction and rotation of five, six, or
more significant factors simply resulted in the disaggregation of these four basic
dimensions.

4 The results
The first hypothesis to be tested, which is expressed in equation (3), is the more
important relationship and will consume the bulk of the ensuing discussion. The
second major hypothesis, expressed in equation (4), is interesting in that, should it
exist, it ties the housing and social space models to real geographical space, and it
casts some light on the veracity of the samples.
4.1 The relationship between housing and social space
The relationship expressed in equation (3) can be interpreted in terms of congruence.
The implication is that the position of an individual or family in social space should
be reflected in its position in housing space, recognizing, of course, the rather limited
definition of social space used in this analysis. Thus congruence in this deterministic
social space context can be tested statistically in the aggregate by evaluating whether
the matrix of regression coefficients B =£ 0. The matter of congruence can also be
approached in a social context by deciding on certain minimum housing requirements,
based on a priori empirical evidence, expressing these as constraints, and then
determining the degree to which the different groups are congruent with respect to
these constraints. This second approach to the notion of congruence is non-statistical
and it is used to indicate the degree of incongruence among various social groups. As
a consequence, it could lead toward a clearer understanding of housing deficiency
according to both real and perceived needs.
4.2 The statistical analysis of congruence
The hypothesis to be tested statistically with respect to congruence and social space
is that the matrix Bamtm in equation (3) is significantly different from a zero matrix
of the same order. In this case m = 2 as there are two axes for each model, and n
408 M Yeates

(the sample size) is 175 for Kingston, and 90 for Winnipeg. The coefficient a of m
is unity, for there is only one independent variable, the matrix S. The method used
for estimating the regression coefficients in this multivariate linear regression model is
described by Anderson (1958, pp. 179-187).
The matrices of regression coefficients (table 4), which indicate the effect of one
axis in the independent matrix (S) on one in the dependent matrix (H) with all the
other axis vectors held constant, are extremely interesting. The regression coefficient
matrices do, in the aggregate, test as being significantly different from a zero matrix
for both Winnipeg and Kingston. The F statistic in the Winnipeg case is calculated to
be 22-61 and in the Kingston example 25-94, and when these are compared with the
significant points of F 4 1 7 4 and F 4 3 4 4 respectively, they indicate that the null hypothesis
in both instances should be rejected and the research hypothesis (B ¥= 0) inferred.
The individual regression coefficients are, however, of greater interest.
The coefficients relating house quality to social status are, as predicted, positive
and significant at the 1% confidence level. The magnitude of the coefficient for the
Kingston sample is larger than that for Winnipeg, and in both cases the important
interrelationship between quality of residence and social status is emphasized. The
association between stage in the life cycle and house type is not as clear, particularly
in the case of Kingston where the coefficient tests as being significant at the 5% level
but not at the 1% level as in the Winnipeg sample. Though both coefficients in
essence support the contention that stage in the life cycle is related to type of home
inhabited, the rather low value of the coefficient for the Kingston sample requires
some explanation.
The reason for this low value can be discussed with respect to the intriguing
coefficient relating social status to type of home in the Kingston sample. This
coefficient indicates that there is a tendency for low social status families to live in
single family residences, and, conversely high social status families to live in
apartments. While this interpretation tends to support the findings of Rainwater
(1966) and Clark (1966), mentioned previously, it tends to contradict those of
Lansing et al. (1970, pp.118-119) who suggest that the higher income groups
overwhelmingly prefer to live in, and in fact do reside in, single family dwellings.
Although this may be true, our study has isolated income with quality of home, and
excluded it from status determination. Thus the students, who comprise one-seventh
of the sample and one-eighth of the population of Kingston, may overemphasize this
relationship, for they are classified as generally above average status yet they tend to
live in apartments wherever possible. Furthermore, a number of these students are
married with young children, yet they live in row houses or multi-family rental units
due to the transitory nature of their residence in Kingston, and as a consequence
adversely affect the life cycle-type of home coefficient.
4.3 Variations in congruence among various social groups
If the position of the family in social space can be regarded as the predicted position

Table 4. Matrix of regression coefficients for equation (1), Winnipeg and Kingston3.
Social space Housing space
quality type
Social status 0-679 -0-124
0-852 -0-481
life cycle -0-091 0-447
-0-097 0-226
a
Figures pertaining to Kingston are italicized; those pertaining to Winnipeg are upright.
Congruence between housing space, social space, and community space 409

of that family's residence in housing space, then the distance between the house that
that family occupies in housing space and the position of the family in social space
will indicate the degree of congruence. Though it has been suggested in the previous
section that a degree of congruence in both cases exists, nevertheless the distances
between the positions of the sample observations in many instances are great. As a
degree of incongruence is to be expected, due, for example, in some cases to inertia
(Lansing and Mueller, 1967, pp. 198-208), it is important to distinguish the degree
of incongruence with respect to housing requirements which are stated as constraints.
The first constraint is with respect to stage in the life cycle and type of housing
inhabited. As indicated previously, there is considerable evidence to suggest that
families with children have particular housing requirements—a necessity for ease of
access to outdoor space and separation from neighbors, which, given the housing
stock available today, virtually implies single family housing of one type or another.
These requirements are not based on what people might like, but what is thought to
be needed for parental control of children and the healthy development of the child.
Other housing requirements, for example, self-contained apartment units for the aged,
are less clearly defined but could, nevertheless, be included. Thus the major constraint
with respect to type of housing is that families with children should live in dwelling
units that have direct entry and some clearly defined external space (Cooper, 1971b).
This constraint is drawn in figure 4 as CI, and the line separates multi-family dwellings
on the left from duplexes, row houses, and single family dwelling units on the right.
Due to the standardization of the data, this same line also divides families without
children (or with one young child) on the left, from families with children on the right.
A second constraint involves housing condition, price, and quality. As has been
suggested previously, there is some evidence to indicate that blighted housing
conditions adversely affect the health of young children, which in turn, of course,
influences their performance in school as well as their general physical development,
and that they also contribute to neighborhood disintegration. Also, much of the
literature in human ecology suggests that many social pathologies are associated with
blight and overcrowding (Burgess and Bogue, 1964, pp.591-615), though this cannot
be construed as implying that inferior housing conditions cause social pathologies.
Perhaps, what is important is that in a wealthy society such as that found in North
America, blighted and run-down housing conditions are hard to countenance,
particularly if they can cause poor health (psychological and physical), and may be
related to certain social pathologies. Thus the distinction between blighted and run-
down housing in contrast with those of adequate condition or better is regarded as a
constraint for it is proposed that no family should reside in these grossly inferior
conditions. This constraint is drawn in figure 4 as line C2, and it distinguishes run-
down and blighted housing and families of low social status below the line from
adequate or better quality housing and higher status families above.
The third policy constraint is much more tentative than the previous two, and
could well be interpreted as class bias. It is, however, based on the observation that
the higher the level of education of the heads of the household the greater the demand
and willingness to pay for comfortable conditions. As the breakpoint in taste appears
to be between blue collar and clerical workers with respect to single family housing,
this is regarded as the locus of line C3 in figure 4, and can be interpreted as a 'taste',
or, perhaps, a life-style constraint. Thus it is proposed that families with children
living above C3 will not wish to occupy residence below this line as they will not suit
their self-perceived style of living. The constraint has not been extended to individuals
and families without children, for it is not clear that they are as concerned with the
environment internal to the residence, though they should not live in blighted
conditions.
410 M Yeates

Stage in life cycle, type of home

Stage in life cycle, type of home


JC1 • • wl

IV

y#^
•a
- ^» _ w
\ ^ : _C3 J
^JAV^
v
*^ "^rvl C2 1
^ 5 ^ $ T ^ K } yt r i
I
\ "

Figure 4. The congruence between social space and housing space for Winnipeg (W) and Kingston
(K), and the location of the policy constraints.
Congruence between housing space, social space, and community space 411

In accordance with these assumptions, figure 4 distinguishes five coincident


categories for both housing space and social space, and the descriptions of each are
indicated by the captions on the diagram and also in table 5. For example, I refers
to an area which, in terms of social space (S) can be described as 'low social status
childless adults', and in terms of housing space (H) as 'blighted or rundown
multi-family dwellings'. The congruence and incongruence of the social groups with
respect to these housing categories is revealed in table 5, though inferences from this
data to the urban area as a whole cannot be made due to the inadequacy of the
sample size. Thus, in Winnipeg, one family in social space II (S II) has a dwelling in
housing space I (H I), six of S II are in H II, one of S II is in H III, one of S II is
in H IV, and two of S II are in H V. If one takes a deterministic view, as discussed
in the previous section, the elements on the diagonals would indicate the degree of
housing congruence with respect to social space.
If, however, one measures congruence in terms of housing requirements and what
would be generally acceptable according to the defined constraints, the following
guidelines are suggested:
(a) S I should not be located in H I or H II for, given the constraint that no group
should live in blighted housing, these childless adults should be in H III, H IV, or H V;
(b) S II should not be in H I, H II, or H IV for, given the constraint that families
with children should not live in blighted residential units or apartments, this group
should be in either H III or H V;
(c) S III should not be in H I, H II, H IV, or H V for, given the assumption that
families with children should not live in blighted residential units or apartments, and
that this particular group has tastes that cannot be sufficed by merely adequate
dwellings, this group should be located in H III;
Table 5. The congruence of various social groups with respect to housing space; Winnipeg and
Kingston3.
Social space Housing space
1 II III IV V total
(blighted or (blighted or (adequate (adequate (adequate,
rundown, rundown, or better, or better, single family)
multi-family) single family) single family) multi-family)
I (low status, 1 1 0 0 0 2
childless) 1 1 0 0 0 2
II (low status, 1 6 1 1 2 11
with 10 6 1 1 9 27
children)
III (middle class 1 6 26 5 4 42
or higher 0 0 36 4 7 47
status with
children)
IV (lower middle 0 1 7 8 1 17
class or higher 5 1 13 9 7 35
status
childless)
V (lower middle 1 5 5 2 5 18
class, with 3 6 14 5 36 64
children)
Total 4 19 39 16 12 90
19 14 64 19 59 175
Figures pertaining to Kingston are italicized; those pertaining to Winnipeg are upright.
412 M Yeates

(d) S IV should not be in H I or H II for, given the constraint that no group should
live in blighted housing, these childless adults should be in H III, H IV, or H V;
(e) S V should not be in H I, H II, or H IV for, given the assumption that families
with children should not live in blighted residential units or apartments, this group
should be located in H III or H V.
Following these guidelines, 38-9% of the families in the Winnipeg sample and 28-6%
of the families in the Kingston sample are incongruent with respect to the housing in
which they reside.
More interesting, however, is an examination of the degree of housing incongruence
for each of the five social groups. From table 6 it can be observed that the lowest
social groups with or without children, as would be expected, are least congruent
with respect to the defined social policy constraints in both Winnipeg and Kingston.
Childless adults are the most congruent, indicating that the housing market in both
urban areas caters for this group reasonably well, whereas families with children who
can afford to live in duplexes, row houses, or single family units are on the average
30% incongruent for both samples combined. Most interesting (from the viewpoint
of the politician) is the degree of incongruence among the higher social groups, the
percentages being far beyond what one might expect purely on the basis of personal
preferences (for example, 10%).
4.4 The relationship between community space and housing and social space
The hypothesis to be tested statistically is that the matrix i? am>m in equation (4) is
significantly different from a zero matrix of the same order. In this case m and n
are the same as before, but the coefficient a = 2, for there are two independent
variables, the matrices H and S. The method used for estimating the regression
coefficients in the multivariate regression model is described by Anderson (1958,
pp. 179-187). Using the previously explained derived factor score matrices for C, H,
and S, the F statistic is calculated to be 4-64 in the Winnipeg sample and 17-14 for
Kingston. Both these statistics are significant at the 1% confidence level. Thus it is
encouraging to note that, even though the sizes of the samples are small, and there is
ten year interval between the surveys and the census, they do relate to the general
characteristics of the urban areas as summarized by the community space models,
notwithstanding the considerable change that can occur in the social geography of an
Table 6. Proportion of families in each social group that are incongruent with respect to the
housing space in which they reside: Winnipeg and Kingston.
Social group Winnipeg Kingston
proportion sample proportion sample
S I (low status, 100% 2 100% 2
childless)
SII (low status, with 72-7 11 62-9 27
children)
S III (middle class or 38-1 42 23-4 47
higher status
with children)
S IV (lower-middle 5-9 17 17-1 35
class or higher
status, childless)
SV (lower-middle 44-8 18 21*8 64
class with
children)
Total 38-9 90 28-5 175
Congruence between housing space, social space, and community space 413

urban area in a decade (Haynes, 1971). The differing magnitudes of the F statistics
are related t o t h e different sample sizes and the fact that census tracts, in which the
internal variation is quite large, are used in t h e Winnipeg example, whereas enumeration
areas, in which the internal variation is m u c h smaller, are used in the Kingston case.
Regardless of these shortcomings, it is clear t h a t i m p o r t a n t implications concerning
housing can be analysed in the context of social policy through this t y p e of analysis.
The generality of t h e study could be improved immeasurably b y :
(a) increasing t h e size of the samples so t h a t more significant statements can be m a d e
with respect to urban areas as a w h o l e (Blalock, 1960, p p . 1 6 5 - 6 7 ) ;
(b) changing the sample design t o one stratified according t o house t y p e and social
group of t h e inhabitants (Michelson, 1969);
(c) improving and adding t o the quality of data t h a t can be gleaned from t h e
questionnaire by increasing the range of direct questions and incorporating information
obtained from assessment rolls;
(d) trying to determine the life style of the inhabitants in order t o assess t h e
importance of this variable in housing congruence (Michelson and Reed, 1970);
(e) attempting to assess the change in housing contiguity through a reconstruction of
past housing space and social space models for different urban areas through a
detailed examination of assessment rolls, city directories, and electoral enumeration
lists (Kirkland, 1969).
As the socioeconomic structure of t h e Canadian population changes, so will the housing
t h a t is required, and a large-scale analysis of t h e t y p e outlined in this paper is necessary
t o m o n i t o r these changes as a prelude to an evaluation of social policy alternatives.

Acknowledgements. The author would like to thank the Canada Council and Queen's University
for support during his sabbatical leave 1971 -1972 in the course of which this paper was written.
Thanks are also due to Mr. A. Hallsworth who undertook the Kingston survey, and Dr. G. Carey
who commented critically on an earlier draft.
References
Anderson, T. W., 1958, Introduction to Multivariate Statistical Analysis (John Wiley, New York).
Berry, B. J. L., 1965, "Internal structure of the city", Law and Contemporary Problems, 30,111-119.
Berry, B. J. L., 1971, "Comparative factorial ecology", Economic Geography, 47 (Supplement,
special edition).
Berry, B. J. L., Rees, P., 1969, "The factorial ecology of Calcutta", American Journal of Sociology,
74,445-491.
Blalock, H. M., 1960, Social Statistics (McGraw-Hill, New York).
Blishen, B. R., 1967, "A socio-economic index for occupations in Canada", Canadian Review of
Sociology and Anthropology, 4, 41-53.
Brown, L. A., Moore, E. G., 1971, "The intra-urban migration process: a perspective" in Internal
Structure of the City, Ed. L. Bourne (Oxford University Press, New York).
Burgess, E. W., Bogue, D., (Eds.), 1964, Contributions to Urban Sociology (University of Chicago
Press, Chicago).
Clark, S. D., 1966, The Suburban Society (University of Toronto Press, Toronto).
Cooper, C , 1971a, "The house of symbol of self, Institute of Urban and Regional Development,
Working Paper number 120, University of California, Berkeley.
Cooper, C , 1971b, "Resident dissatisfaction in multi-family housing'', Institute of Urban and
Regional Development, University of California, Berkeley, Working Paper number 160.
Duncan, O. D., Duncan, B., 1957, "Residential distribution and occupational stratification", in
Cities and Society, Eds. P. K. Hatt, A. J. Reiss (The Free Press, New York).
Erikson, E., 1968, Identity: Youth and Crisis (W. W. Norton, New York).
Feldman, A. S., Tilly, C , 1960, "The interaction of social and physical spaces", American
Sociological Review, 25, 877-884.
Grigsby, W. G., 1967, "Home finance and housing quality in ageing neighborhoods", in The
Economic Problems of Housing, Ed. A. A. Nevitt (Macmillan, New York), pp.105-122.
Hallsworth, A., 1971, Housing Space, Social Space, and Social Areas in Kingston, Unpublished
Masters Thesis, Department of Geography, Queen's University, Kingston, Ontario, Canada.
414 M Yeates

Haynes, K. E., 1971, "Spatial change in urban structure: alternative approaches to ecological
dynamics", Economic Geography, 42 (Supplement), 324-335.
Kirkland, J. S., 1969, Housing Filtration in Kingston, 1953-68, Unpublished Masters Thesis,
Department of Geography, Queen's University, Kingston, Ontario, Canada.
Krieger, M. H., 1969, "The life cycle as a basis for social policy and social indicators", Working Paper
number 106, Institute of Urban and Regional Development, University of California, Berkeley.
Lansing, J. B., Clifton, C. W., Morgan, J. N., 1969, "New homes and poor people", Institute for
Social Research, The University of Michigan, Ann Arbor, Michigan.
Lansing, J. B., Morans, R. W., Zehner, R. B., 1970, "Planned residential environments", Institute for
Social Research, The University of Michigan, Ann Arbor, Michigan.
Lansing, J. B., Mueller, E., 1967, "The geographic mobility of labour", Institute for Social Research,
The University of Michigan, Ann Arbor, Michigan.
Martin, A. E., 1967, "Environment, housing and health", Urban Studies, 4, 1-21.
Maslow, A., 1968, Toward a Psychology of Being (Van Nostrand, Princeton, NJ).
McDavid, J. W., Harari, H., 1968, Social Psychology (Harper and Row, New York).
Menchik, M. D., 1971, "Residential environmental preference and choice: some preliminary
empirical results relevant to urban form", RSRI Discussion Paper number 46, Philadelphia.
Michelson, W., 1969, "Analytic sampling for design information: a survey of housing experience",
Research Paper number 21, Centre for Urban and Community Studies, University of Toronto,
Toronto.
Michelson, W., 1970, Man and His Urban Environment: a Sociological Approach (Addison-Wesley,
Reading, Mass.).
Michelson, W., Reed, P., 1970, "The theoretical status and operational usage of life style in
environmental research", Research Paper number 36, Centre for Urban and Community Studies,
University of Toronto, Toronto.
Murdie, R. A., 1969, "The factorial ecology of metropolitan Toronto, 1951-1961: an essay on the
social geography of the city", Research Paper number 116, Department of Geography, University
of Chicago, Chicago.
Muth, R. F., 1967, "Slums and poverty", in The Economic Problems of Housing, Ed. A. A. Nevitt
(Macmillan, New York), pp.12-26.
Nicholson, G. T., Yeates, M. H., 1969, "The ecological and spatial structure of the socio-economic
characteristics of Winnipeg", Canadian Review of Sociology and Anthropology, 6, 162-178.
Porter, J., 1965, The Vertical Mosaic (University of Toronto Press, Toronto).
Rainwater, L., 1966, "Fear and the house-as-haven in the lower class", Journal of the American
Institute of Planners, 32, 23-31.
Rees, P. H., 1970, "The factorial ecology of metropolitan Chicago", in Geographic Perspectives on
Urban Systems, Eds. B. J. L. Berry, F. W. Horton (Prentice-Hall, Englewood Cliffs, NJ).
Rees, P. H., 1971, "Factorial ecology: an extended definition, survey, and critique of the field",
Economic Geography, 47 (Supplement), 220-233.
Rossi, P. H., 1955, Why Families Move (The Free Press, New York).
Rummel, R. J., 1970, Applied Factor Analysis (Northwestern University Press, Evanston, Illinois).
Shevky, E., Bell, W., 1955, Social Area Analysis (Stanford University Press, Stanford).
Shevky, E., Bell, W., Williams, M., 1949, The Social Areas of Los Angeles: Analysis and Typology
(University of California Press, Berkeley).
Simmons, J. W., 1968, "Changing residence in the city: a review of intra-urban mobility",
Geographical Review, 58, 622-651.
Smith, W. F., 1970, Housing: The Social and Economic Elements (University of California Press,
Berkeley).
Tilly, C , 1961, "Occupational rank and grade of residence in a metropolis", American Journal of
Sociology, 67, 323-330.
Timms, D. W. G., 1971, The Urban Mosaic: Towards a Theory of Residential Differentiation
(Cambridge University Press, Cambridge).
Wheeler, J. O., 1968, "Residential location by occupational status", Urban Studies, 5, 24-32.
Wilner, D. M., Walkley, R. P., Pinkerton, T., Taybeck, M., 1962, The Housing Environment and
Family Life (The Johns Hopkins Press, Baltimore).

p © 1972 a Pion publication printed in Great Britain

You might also like