Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Consumer spatial behaviour and its relation to social class and family status in metropolitan Vancouver,… Gayler, Hugh James 1974

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1974_A1 G39.pdf [ 9.6MB ]
Metadata
JSON: 831-1.0093215.json
JSON-LD: 831-1.0093215-ld.json
RDF/XML (Pretty): 831-1.0093215-rdf.xml
RDF/JSON: 831-1.0093215-rdf.json
Turtle: 831-1.0093215-turtle.txt
N-Triples: 831-1.0093215-rdf-ntriples.txt
Original Record: 831-1.0093215-source.json
Full Text
831-1.0093215-fulltext.txt
Citation
831-1.0093215.ris

Full Text

CONSUMER SPATIAL BEHAVIOUR AND ITS RELATION TO SOCIAL CLASS AND FAMILY STATUS IN METROPOLITAN VANCOUVER, CANADA by HUGH JAMES GAYLER B.A,, University of Leicester, 1963 M.A., University of London, 1966 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in the Department of Geography We accept this thesis as conforming to the requir standard J THE UNIVERSITY OF BRITISH COLUMBIA January 1974 I n p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the r e q u i r e m e n t s f o r an advanced degree a t t h e U n i v e r s i t y o f B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y p urposes may be g r a n t e d by the Head o f my Department o r by h i s r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l n o t be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . Department The U n i v e r s i t y o f B r i t i s h Columbia Vancouver 8, Canada Abstract This study is concerned with the relationship between consumers' social class and family status characteristics and some spatial aspects of their shopping behaviour. In the past much of our understanding of how consumers behave spatially has been derived from studies of r e t a i l structure, and thus i t has to be assumed that consumers behave in the same fashion; and where studies have looked at variations in behaviour they have concentrated on differences in r e t a i l demands or differences in the single, personal attributes (e.g. income or ethnic origin) which influence behaviour. Many of these personal attributes are interrelated, and in this study two such interrelationships are identified - social class and family status. A sample survey of consumers was undertaken in Metropolitan Vancouver, Canada; and using factor analysis i t is possible to identify, and measure consumers according to, three dimensions (or interrelationships) among the socio-economic and demographic attributes asked for in the survey - social class and older and younger family status (two similar dimensions result from respondents being asked to identify their children by age group). Social class and family status are then related to two aspects of consumer spatial behaviour - travel behaviour, and in particular the distance and frequency travelled to shop for various goods^ and secondly the specific department store and i t s location chosen by the consumer. It is found that the higher the social class group, the greater the frequency shopping goods were purchased and the greater i i i the distance travelled for groceries and dress. But for goods required less frequently and/or lacking a specialty nature (furniture, appliances and footwear) the differences in distance are not significant. On the family status dimension the low group (small families, invariably older and without children) often travel significantly shorter distances than other groups; the former reside mostly in the older areas, close to major shopping centres, and/or tend to shop at the nearest centre. Department-store preference is found to vary according to area. A significant polarisation of preference by social class is seen in downtown Vancouver, but in the outer suburbs, where the same firm has decentralised to widely differing social areas, similar allegiances are not found. Different department-store firms do not attract one particular family status group more than another. Differences are, however, related to geographic location with the downtown Vancouver and suburban stores attracting a significantly higher proportion of low and medium-high group consumers respectively. iv Contents Page T i t l e i Abstract I i Contents iv Tables v i i Figures ix Acknowledgements x Chapter I Introduction 1 Statement of the problem 2 The Data 6 The Analysis 6 Chapter II Theoretical and Empirical Background to Consumer Spatial Behaviour 8 Introduction 8 The Location and Structure of Service Activities 8 1. Central Place Theory 9 2. Retail Gravitation Theory 12 Variations in Retail Structure and Consumer Spatial Behaviour 14 1. Modifications of Central Place Theory 15 2. Differentiation in Retail Functions and Structure 18 3. Social and Economic Influences on Spatial Behaviour 21 Differences in Consumer Attitudes 29 Summary 31 Chapter III Concepts of Social Class 32 Introduction 32 The Nineteenth Century Reformers 34 Research into the concepts of social class 38 Summary 45 Chapter IV Measuring Social Class and Family Status 46 Introduction 46 Real Variables 47 Proxy Variables 49 Methodology for Using Proxy Variables 50 1. Early Multiple Factor Designs 50 2. Occupation Groupings 54 3. Multivariate and multi-dimensional techniques 56 Summary 73 V Page Chapter V Metropolitan Vancouver : Case Study 75 Introduction 75 Research Design 75 The Data 78 1. Sample Survey 78 2. Socio-economic and demographic variables 81 3. Shopping variables 84 Methodology for grouping socio-economic and demographic variables 86 1. Correlation Matrix 86 2. Unrotated Factor Matrix 89 3. Orthogonally Rotated Factor Matrix 91 4. Factor scores 94 Summary 100 Chapter VI Social Class, Family Status and Consumer Travel Behaviour 101 Introduction 101 Research Hypotheses 102 Variations in the Spatial Behaviour of Retailing and Consumers 104 1. The spatial pattern of retailing 104 2. Aggregate travel behaviour 108 3. Variations in distance by social class and family status groups 117 4. Frequency of purchase by social class groups 120 Analysis of Variance tests 122 1. Social Class 123 2. Older Family Status 129 3. Younger Family Status 131 Summary 133 Chapter VII Social Class, Family Status and Store Preference 135 Introduction 135 Methodology and Research Hypotheses 135 Social class and Department-Store Preference in Downtown Vancouver 142 1. Dress 143 2. Furniture 146 3. Appliances 147 4. Footwear 148 Social class and department-store preference in Metropolitan Vancouver 148 Family Status and Department-Store Preference 153 1. Downtown Vancouver 153 2. Metropolitan Vancouver 156 3. Downtown versus suburban preference 158 Summary 164 v i Page Chapter VIII Summary and Conclusions 166 Approaches to Consumer Spatial Behaviour 166 Social Class and Family Status 168 The Data 169 Consumer Travel Behaviour 170 Store Preference 171 Retrospect and Prospect 172 Bibliography 175 Appendix A University of British Columbia, Faculty of Commerce and Business Administration, Buying Habits Survey, June 1964 184 Appendix B Blishen's Occupation Class Scale 189 Vita V l l Tables 5.1 Matrix of Correlation Coefficients 5.2 Unrotated Matrix of Factor Loadings 5.3 Othogonally Rotated Matrix of Factor Loadings 5.4 Frequency Distribution of Factor Score Classes 6.1 Mean and standard deviation of distance in miles by Social Class and Family Status for purchases made in Metropolitan Vancouver, 1944 6.2 Number and Proportion of Consumers in each Social Class purchasing Shopping Goods in Metropolitan Vancouver over a one year period 1963-64 6.3 Analysis of variance test for differences in mean distance for various products by social class 6.4 Analysis of variance test for differences in mean distance for various products by older family status 6.5 Analysis of variance test for differences in mean distance for various products by younger family status 7.1 Frequency of consumers by social class and department-store preference for dress in Downtown Vancouver 7.2 Frequency of consumers by social class and department-store preference for furniture in Downtown Vancouver 7.3 Frequency of consumers by social class and department-store preference for appliances in Downtown Vancouver 7.4 Frequency of consumers by social class and department-store preference for footwear in Downtown Vancouver 7.5 Frequency of consumers by social class and department-store preference for dress in Metropolitan Vancouver 7.6 Frequency of consumers by social class and department-store preference for furniture in Metropolitan Vancouver 7.7 Frequency of consumers by social class and department-store preference for appliances in Metropolitan Vancouver 7.8 Frequency of consumers by social class and department-store preference for footwear in Metropolitan Vancouver 7.9 Frequency of consumers by older family status and department store preference for dress in Downtown Vancouver Page 88 90 92 97 118 121 124 130 132 144 144 145 145 150 150 151 151 154 v i i i Page 7.10 Frequency of consumers by younger family status and department store preference for dress in Downtown Vancouver 154 7.11 Frequency of consumer's by older family status and department store preference for dress in Metropolitan Vancouver 155 7.12 Frequency of consumers by younger family status and department store preference for dress in Metropolitan Vancouver 155 7.13 Frequency of consumers by older family status and location of department-store preference for dress in Metropolitan Vancouver 159 7.14 Frequency of consumers by younger family status and location of department-store preference for dress in Metropolitan Vancouver 159 7.15 Frequency of consumers by older family status and location of department-store preference for furniture in Metropolitan Vancouver 160 7.16 Frequency of consumers by younger family status and location of department-store preference for furniture in Metropolitan Vancouver 160 7.17 Frequency of consumers by older family status and location of department-store preference for appliances in Metropolitan Vancouver 161 7.18 Frequency of consumers by younger family status and location of department-store preference for appliances in Metropolitan Vancouver 161 7.19 Frequency of consumers by older family status and location of department-store preference for footwear in Metropolitan Vancouver 162 7.20 Frequency of consumers by younger family status and location of department-store preference for footwear in Metropolitan Vancouver 162 i x Figures Page 2.1 Basic Interactions of a Topographical Model of Consumer Space Preferences 22 5.1 Metropolitan Vancouver : Major Urban Areas 1964 77 5.2 D i s t r i b u t i o n of factor scores 96 6.1 Consumer s p a t i a l behaviour f or groceries i n Metropolitan Vancouver, 1964 109 6.2 Consumer s p a t i a l behaviour f or dress i n Metropolitan Vancouver, 1964 110 6.3 Consumer s p a t i a l behaviour f or f u r n i t u r e i n Metropolitan Vancouver, 1964 111 6.4 Consumer s p a t i a l behaviour f or appliances i n Metropolitan Vancouver, 1964 112 6.5 Consumer s p a t i a l behaviour f o r footwear i n Metropolitan Vancouver, 1964 113 X Acknowledgments I would l i k e to thank my supervisor, Dr. Walter G. Hardwick of the University of B r i t i s h Columbia, who has over a number of years inspi r e d and guided my research and provided encouragement and valued advice. Other members of the University have also provided invaluable assistance, including Professors John Chapman, Jim Forbes, Gary Gates and Roger Leigh. Thanks are also due to the Canadian Council on Urban and Regional Research whose support f o r a larger research project by Dr. Hardwick enabled the data on shopping behaviour to be c o l l e c t e d . The University of B r i t i s h Columbia and the Central Mortgage and Housing Corporation generously supported me while the research was being c a r r i e d out, I would also l i k e to thank Mrs. P h y l l i s Riesberry, who so expertly attended to typing matters, and Mr. John Bennett who provided cartographic assistance. Last, but not least, I owe a r e a l debt to my wife, Fran, who gave support yet endured l o s t attention. 1 Chapter I Introduction Research into consumer behaviour has been the preoccupation of various disciplines in the social sciences, including geography, sociology, economics and psychology and in a practical sphere by schools of marketing and business administration. Consumer behaviour refers to a broad spectrum of events including the processes of decision-making on the part of the consumer, the nature of the journey to shop, the actual selection and purchase of goods and services, and the ways in which consumers influence, or are influenced by, the individuals and organisations providing these goods and services. The interest of the geographer in this subject stems from the fact that consumer behaviour has an important spatial component. Consumer spatial behaviour is defined as the ways in which consumers relate to the distribution of the various goods and services over space. It can be seen that this is a broad f i e l d of study and therefore the present enquiry concentrates on furthering our understanding in two areas - f i r s t l y , travel behaviour, and in particular the distance that is travelled to shop, and secondly the preference for various stores and their locations. Within these two areas of study attention is focussed 2 on the variations that occur and the relationship between these and certain socio-economic and demographic characteristics of the consumers themselves which are organised in terms of their social-class and family-status positions. Statement of the Problem The next chapter shows that there are three approaches to i the study of consumer spatial behaviour, and the differences between them reflect the in-depth nature to which the analysis is taken together with a changing focus within the discipline of geography i t s e l f . The f i r s t approach is one that does not involve the consumer directly, but makes assumptions about the ways in which he behaves on the basis of the spatial distribution of the goods and services that he uses.''' Interest in the spatial distribution of goods and services (central place theory) reflects the preoccupation of geographers in the last twenty years or so in making the discipline more scientific in i t s overall methodology; their aim was not merely to explain spatial patterns of phenomena, but to make generalisations about those patterns which then have value for the purposes of prediction and application elsewhere. However, generalisations have the inherent weakness of being average or aggregate situations; and generalisations about the ways consumers relate to those spatial patterns have the added weakness of being merely generalised assumptions since the consumer has not been interviewed. It is assumed that a consumer maximises his u t i l i t y by shopping at the nearest location where a good or service is provided 1. Berry, B. J. L., Geography of Market Centres and Retail Distribution, Englewood C l i f f s , N. J., Prentice Hall, 1967. 3 (economic man); t h i s i n turn assumes that a l l locations are equal and the consumer has complete knowledge of the opportunities a v a i l a b l e . The second approach represents a desire to get away from the analyses of purely s p a t i a l arrangements of phenomena and to look at the decision-making processes of the people who r e l a t e to such s p a t i a l arrangements. This behavioural approach has resulted i n events or people being more disaggregated. The consumer i s now interviewed, | allowing f or e a r l i e r assumptions about his journey to shop to be put aside; moreover, i t i s seen that consumers vary i n t h e i r shopping behaviour, r e f l e c t i n g therefore differences i n knowledge, need and inte r p r e t a t i o n s of what constitutes maximum u t i l i t y . The nature of thi s v a r i a t i o n i s commonly analysed with respect to either the type of 2 store or shopping centre v i s i t e d , or consumer a t t r i b u t e s which are l i k e l y to contribute to d i f f e r e n t demands being made on the r e t a i l 3 structure, e.g. income and ethnic group. Research c a r r i e d out over the l a s t few years has allowed us to be both more precise and more accurate i n the generalisations we make about consumers : instead of the one concept of economic man, we can make generalisations f o r each group f o r whichever way we disaggregate. 2. See E l i o t Hurst, M. E. and S e l l e r s , J . , "An Analysis of the S p a t i a l D i s t r i b u t i o n of Customers around Two Grocery R e t a i l i n g Operations", Professional Geographer, 21, 1969, 184-90. 3. See Ho l l y , B. P. and Wheeler, J . 0., "Pattern of R e t a i l Location and the Shopping T r i p s of Low-Income Households", Urban  Studies, 9, 1972, 215-20^ Murdie, R. A., " C u l t u r a l Differences i n Consumer Tra v e l " , Economic Geography, 41, 1965, 211-33. A t h i r d , and very recent, approach i s to extend this i n -depth analysis of the consumer by examining why consumers with varying a t t r i b u t e s make d i f f e r e n t demands on the r e t a i l structure. This involves a very extensive questioning of the consumer and measuring of his a t t i t u d e s and perceptions towards the shopping environment. The present study focusses on the second approach above and i s concerned with f u r t h e r i n g our understanding of how v a r i a t i o n s i n consumer a t t r i b u t e s are rel a t e d to some aspects of s p a t i a l behaviour. One problem that i s seen i s that so many consumer studies have been concerned with a sing l e personal a t t r i b u t e ; yet f o r consumers with the same value on a p a r t i c u l a r a t t r i b u t e there can e x i s t considerable v a r i a t i o n i n the ways they behave.~* This i s possibly the r e s u l t of differences i n rel a t e d personal a t t r i b u t e s which influence shopping behaviour. Studies of r e s i d e n t i a l behaviour ( s o c i a l area analysis) have shown that many of these a t t r i b u t e s are i n t e r r e l a t e d and a number of i n t e r r e l a t i o n s h i p s (or dimensions of behaviour) are revealed, the 6 three most important being s o c i a l c l a s s , family status and e t h n i c i t y . These i n t e r r e l a t i o n s h i p s do not merely reduce the number of variables 4. Taylor, S. M., S p a t i a l Perspectives at the Consumer-Store Interface, unpublished M. A. th e s i s , University of B r i t i s h Columbia, 1972. 5. Coleman, R. P., "The S i g n i f i c a n c e of S o c i a l S t r a t i f i c a t i o n i n S e l l i n g " i n Marketing ; A Maturing D i s c i p l i n e , Proceedings of the Winter Conference of the American Marketing Ass o c i a t i o n , 1960, Chicago, American Marketing Association, 1961, 171-84. 6. Shevky, E. and B e l l , W., S o c i a l Area Ana l y s i s , Stanford, Stanford University Press, 1955j Murdie, R. A., F a c t o r i a l Ecology  of Metropolitan Toronto, 1951-61, Research Paper No. 116, Chicago, Un i v e r s i t y of Chicago, Department of Geography, 1969. (or single a t t r i b u t e s ^ but independent from one another represent c e r t a i n forces operating on society and i n f l u e n c i n g the many ways in which people behave. The s p e c i f i c purpose of t h i s study i s to examine the r e l a t i o n s h between some aspects of s p a t i a l behaviour and two of these i n t e r r e l a t e d 7 sets of personal a t t r i b u t e s - s o c i a l class and family status. S o c i a l Class S o c i a l class i s one of the most important bases on which a society i s d i f f e r e n t i a t e d and r e f e r s to the family background and upbringing of i n d i v i d u a l s and t h e i r values, motivations and needs. Research i n sociology has shown that i t i s pertinent to group people into broad s o c i a l classes since s i m i l a r i t i e s i n these a t t r i b u t e s lead 8 to s i m i l a r i t i e s i n overt behaviour. The nature of s o c i a l c l a s s i s b r i e f l y discussed i n Chapter I I I ; and Chapter IV i s devoted to the problems of measurement. Since s o c i a l c l a s s constitutes a myriad of d i f f e r e n t (real) v a r i a b l e s , many of which are d i f f i c u l t to measure, i t i s more common to use a few proxy v a r i a b l e s , such as income and occupation. These variables are easier to measure and represent a much larger number of r e a l v a r i a b l e s . A number of methods have been used to measure and assign people to a s o c i a l c l a s s . The one that i s developed here i s factor a n a l y s i s , since two of i t s major benefits are 7. Three dimensions ( s o c i a l c l a s s , family status and e t h n i c i t y ) account for the major proportion of the variance i n population data i n many North American c i t i e s . E t h n i c i t y i s omitted i n t h i s study because of the size of the sample and the nature of the questions asked and the lesser importance of e t h n i c i t y i n Metropolitan Vancouver compared to other North American c i t i e s . 8. Gordon, M. M., S o c i a l Class i n American Sociology, Durham, Duke University Press, 1958. 6 the reduction of data and the identification of underlying interrelationships. Family Status This refers to the nature of the family in terms of i t s position in the l i f e cycle, and takes into account variables such as the size of family and the number and ages of the children. Family status is measured in the same way as social class by applying factor analysis to a relevant body of socio-economic and demographic data about a population. The Data The data used in the present study were obtained from a random sample of the population of Metropolitan Vancouver. Some 680 home interviews were made, and questions were asked concerning the socio-economic and demographic characteristics of the household and the location and store where purchases were made for a number of different goods. Providing the sampling procedure and answers are correct data of this nature are invaluable; i t is otherwise impossible to find out how consumers vary both in their personal attributes and in their shopping behaviour. The nature of the data obtained is examined in Chapter V, and in the second part of the chapter the socio-economic and demographic attributes of respondents are organized into social class and family status groups using factor analysis. The Analysis Two aspects of consumer spatial behaviour, travel behaviour and store preference, are examined in Chapter VI and VII respectively, and i t is seen whether the various social class and family status groups differ significantly in the ways they behave. In each case 7 research hypotheses are stated and developed, showing the background for the rejection of the null hypothesis. Consumer Travel Behaviour In this chapter we are concerned with the distance that consumers travel between their residence and the store where a good is purchased, and to a lesser extent with the frequency that certain goods are purchased. It i s hypothesised that: 1. the higher the social class group, the greater the distance that is travelled to shop and the greater the frequency that goods are purchased. 2. the higher the family status group, i.e. the larger the family size, the shorter the distance that i s travelled to shop. Store Preference The second example focusses on the choice of department store for the four shopping goods asked for in the questionnaire. (The four department-store firms account for approximately one half of a l l shopping-good purchases in Metropolitan Vancouver.) It is hypothesised that: 1. among the downtown Vancouver department stores there are differences in their social class appeal. 2. over the Metropolitan area as a whole the differences in the social class appeal of department stores are not significant since the same department store is found in different social areas. 3. family-status differences are more likely to occur between suburban and downtown department stores rather than between individual stores in the one area. 8 Chapter II Theoretical and Empirical Background to  Consumer Spatial Behaviour Introduction There are three approaches to the study of consumer spatial behaviour, and each one is briefly discussed here in order to show how our understanding of that behaviour has developed, the problems of the various approaches and the relevance of the present research. The f i r s t approach examines the location and structure of service a c t i v i t i e s , allowing us to make assumptions about the ways in which consumers relate to those services. A second approach, on the other hand, allows a more in-depth analysis of shopping patterns to be made. The consumer is interviewed, and i t is possible to identify variations in shopping patterns and relate these variations to differences in the consumers themselves, such as their residential location or socio-economic attributes. A third approach extends this in-depth analysis by examining why consumers with varying attributes make different demands on the r e t a i l structure. This involves an understanding of the differences in consumer attitudes and perceptions towards the shopping environment. The Location and Structure of Service Activities Assumptions about how consumers behave spatially have resulted from the development of two models relating to retailing i t s e l f - central place theory and r e t a i l gravitation theory. 1. Central Place Theory Central place theory implies a c e r t a i n order to the d i s t r i b u t i o n of services and service centres either within the c i t y or the larger 1 region. I t states that the location, number, size and spacing of these centres are based on c e r t a i n assumptions of r a t i o n a l i t y on the part of r e t a i l and consumer behaviour. This r a t i o n a l i t y i s based on three underlying concepts. F i r s t l y , each good found at a c e n t r a l place or shopping centre has a range, a s p a t i a l concept which denotes the area from which consumers w i l l come to purchase a p a r t i c u l a r good. Each good w i l l have a d i f f e r e n t 1. Central place theory was i n i t i a l l y conceived by Walter C h r i s t a l l e r i n the 1930's and the introduction of his work into the English language can be seen i n Ullman, E., "A Theory of Location for C i t i e s " , American Journal of Sociology, 46, 1941, 835-64. A complete t r a n s l a t i o n was not made u n t i l 1957 and was published i n Baskin, C., The Central Places of Southern Germany, Englewood C l i f f s , N.J., P r e n t i c e - H a l l , 1966; a t r a n s l a t i o n of C h r i s t a l l e r , W., Die Zentraler Orte i n Suddeutschland, Jena, Fischer, 1933. The l i t e r a t u r e on c e n t r a l place theory i s extensive. The following are some of the more important works which set down and discuss i t s major concepts: Berry, B. J . L. and Garrison, W. L., "The Functional Bases of the Central Place Hierarchy", Economic  Geography, 34, 1958, 145-54; Berry, B. J . L. and Garrison, W. L., "A Note on Central Place Theory and the Range of a Good", Economic Geography, 34, 1958, 304-11; Berry, B. J . L. and Garrison, W. L., "Recent Developments of Central Place Theory", Papers and Proceedings of the Regional Science Association, 4, 1958, 107-20; Garrison, W. L., et a l . , Studies i n Highway  Development and Geographic Change, S e a t t l e , University of Washington Press, 1959; Carol, H., "The Hierarchy of Central Place Functions within a C i t y " , Annals of the A s s o c i a t i o n of  American Geographers, 50, 1960, 419-38; Berry, B. J . L., Geography  of Market Centres and R e t a i l D i s t r i b u t i o n , Englewood C l i f f s , N.J., Prentice H a l l , 1967; Szumeluk, K., Central Place Theory :  A Review, Working Papers No. 2, London, Centre for Environmental Studies, 1968, V o l . l j Marshall, J . U., The  Location of Service Towns, Geography Research Publications no. 3, Toronto, University of Toronto Press, 1969. range, r e f l e c t i n g differences i n purchase pric e s , t r a v e l costs and the frequency that a good i s demanded. The upper l i m i t of t h i s range i s the maximum possible extent of sales of a p a r t i c u l a r good from a c e n t r a l place. Beyond t h i s l i m i t the t o t a l costs and f r i c t i o n involved i n the purchase of the good are so great that demand declines r a p i d l y . Moreover, competition from closer centres i s such that the good can be obtained from there more cheaply. The second concept, threshold, i s an economic consideration r e l a t i n g to the f i n a n c i a l v i a b i l i t y of a good and competition from other o u t l e t s . This i n turn influences the spacing of the outlets and the minimum distances that consumers have to t r a v e l . In order to survive a r e t a i l o u t l e t must have a minimum number of sales which w i l l cover operating costs and allow s u f f i c i e n t p r o f i t to s a t i s f y the owners or shareholders and to cover c a p i t a l investment. This minimum sales threshold i s more e a s i l y measured i n terms of the population necessary to support such a good, and the l a t t e r w i l l vary according to the s i z e and nature of the r e t a i l outlet concerned. A t h i r d concept r e f e r s to the h i e r a r c h i c a l s p a t i a l structure 2 of c e n t r a l places i n a region or c i t y . Those functions with the smallest 2. Brush, J . E. "The Hierarchy of Central Places i n Southwestern Wisconsin", Geographical Review, 43, 1953, 380-402; Berry and Garrison, "Functional Bases of the Central Place Hierarchy"; Berry, B.J.L., Barnum, H.G. and Tennant, R.J., " R e t a i l Location and Consumer Behaviour", Papers and Proceedings of the Regional  Science A s s o c i a t i o n , 9, 1962, 65-106; Carol, H., op. c i t . ; Berry, B.J.L., Commercial Structure and Commercial B l i g h t , Research Paper no. 85, Chicago, University of Chicago, Department of Geography, 1963; Johnston, R.J., "The D i s t r i b u t i o n of an Intra-Metropolitan Central Place Hierarchy", A u s t r a l i a n Geographical Studies. 4. 1966. 19-33. thresholds w i l l enter the central place hierarchy on the lowest level. In any area these lowest-order centres w i l l be the most numerous, and thus have the shortest distance between them. Moreover, they w i l l carry the smallest total number of functions. Successive higher order centres in the hierarchy w i l l not only include new goods with larger thresholds but also the goods found in lower-order centres. Thus, with each higher-order centre there w i l l be an increase in the total number of functions and the distance between similar centres and a decrease in their frequency of occurrence. One basic disadvantage of central place theory is that i t only makes very broad generalisations about the nature of r e t a i l functions and centres and the ways in which consumers relate to them. Reliance is made on data gathered from published s t a t i s t i c a l sources, e.g. the census for population distribution, a city directory for -the location, number and type of service functions. This approach leads to a highly aggregated situation : a centre may be considered from the point of view of the total number of functions, rather than any breakdown by type. If type of function i s taken into account, a l l stores of the one type are invariably considered in aggregate. Likewise, consumers are aggregated since there is no information concerning their actual residential location, where they shop and attributes, attitudes and so on about them. In the absence of such information central place theory assumes, for example, that the consumer behaves rationally, has complete knowledge of available shopping oppprtunities and seeks to minimise the frictions of distance by shopping for a good at the nearest centre that can economically support i t . It is shown later that through further observation and interviewing the assumptions of central place theory concerning consumer behaviour often represent an average or majority situation at a macro level. However, i t i s possible to disaggregate both r e t a i l functions and consumers and identify distinct variations from some average pattern. Various stores selling the same type of good differ in their range, threshold and hierarchical position, reflecting, for example, different consumer preferences and entrepreneurial actions. Variations in consumer preferences in turn reflect different needs, desires and a b i l i t i e s to satisfy both of these and different perceptions, for example, of available shopping opportunities and what constitutes f r i c t i o n . 2. Retail Gravitation Theory Further assumptions concerning the ways in which consumers behave over space have been related to notions of Newtonian physics, which states that a force between two masses is proportional to the size of those masses and inversely proportional to the f r i c t i o n , or distance, 3 between them. In establishing the hinterland boundary between two ci t i e s , Reilly stated that the ci t i e s attract r e t a i l trade from an intermediate area in direct proportion to their population and in inverse proportion to the square of the distances from the citi e s to 4 the immediate area. It is unrealistic, however, to talk of a fixed hinterland boundary since i t i s never assumed that consumers on either side of 3. Dodd, S. C., "The Interactance Hypothesis : A Gravity Model f i t t i n g Physical Masses and Human Groups", American Sociological Review, 15, 1950, 245-56; Isard, W., Methods of Regional Analysis, Cambridge, Mass., M.I.T. Press, 1960, 493-568; Olsson, G., Distance and Human Interaction : a Bibliography and Review, Bibliography Series no. 2, Philadelphia, Regional Science Research Institute, 1965. 4. Reilly, W. J., The Law of Retail Gravitation, New York, Reilly, 1931. the boundary always go to the same centre."3 Instead of a fixed boundary for a trade area there are a series of concentric zones, representing a f a l l off in the proportion of consumers as one moves from one r e t a i l centre into areas where there is increasing competition from other centres. Not only is there a f a l l off, but i t s rate varies according to the type of r e t a i l trade concerned. It i s also suggested that Reilly's model best holds up in rural areas where the number of centres is smaller and the frictions of distance are greater. In an urban area the consumer has a much wider choice of r e t a i l centres, many of which w i l l be within a distance he is prepared to travel. The result is a series of isoprobability lines around each shopping centre, overlapping with isoprobability lines from other centres. It is possible for any residential area to measure this overlap and compare probabilities for shopping at the various centres. Retail gravitation theory has been used extensively for locating r e t a i l functions in relation to the distribution of population, and thus the source of potential consumers. But there are restrictions in how far one can go in using i t to understand how consumers behave. In an intra-urban context i t has been shown that the error in estimating a consumer's 6 choice of centre increases as the distance between two centres decreases. Indeed, the amount of cross-traffic between close pairs of centres precludes establishing any hinterland isoprobability lines. It is possible to construct gravity models for different products, different types of 5. Huff, D. L., Determination of Intra-Urban Retail Trade Areas, (Real Estate Research Program, Graduate School of Business Administration, University of California, Los Angeles), Los Angeles, U.C.L.A., 1962. 6. Bucklin, L. P., "Retail Gravity Models and Consumer Choice", Economic Geography, 47, 1971, 489-97. centre and d i f f e r e n t groups of consumers i n order to provide more i n -depth a n a l y s i s . But i t i s s t i l l assumed that the l e v e l of i n t e r a c t i o n i s a function of mass (urban population or shopping centre size) and some measure of f r i c t i o n between r e s i d e n t i a l area and shopping centre. Research elsewhere has shown that there are many other f a c t o r s which influence shopping centre choice. V a r i a t i o n s i n R e t a i l Structure and Consumer S p a t i a l Behaviour The second approach to the study of consumer s p a t i a l behaviour i s both a reaction to the f i r s t one and a genuine desire to understand more f u l l y the v a r i a t i o n s that are seen i n r e t a i l structure and the ways i n which consumers r e l a t e to i t . Interviewing the consumer according to some random sampling procedure means that we no longer have to assume how he behaves. We are also able to disaggregate consumers and group them according to either various facets of t h e i r actual behaviour or various socio-economic and demographic a t t r i b u t e s that they have. I t i s then possible to r e l a t e v a r i a t i o n i n these d i f f e r e n t a t t r i b u t e s to v a r i a t i o n i n the ways consumers a c t u a l l y behave. The one disadvantage of t h i s approach i s that questionnaire construction, sampling design and interviewing i s a lengthy and perhaps costly procedure compared with c o l l e c t i n g data from published sources. This section examines the nature of the research that has been undertaken on v a r i a t i o n s i n r e t a i l structure and consumer s p a t i a l behaviour. F i r s t l y , there have been studies to test the assumptions that c e n t r a l place theory makes about the consumer. Secondly, differences i n r e t a i l functions of the same type have been examined. T h i r d l y , the s p a t i a l behaviour of various socio-economic and demographic groups of consumers has been studied to see i f there are recognisable between-group diff e r e n c e s . 15 1. Modifications of Central Place Theory (a) Nature of shopping t r i p The central place model assumes the shopping t r i p i s made from home; moreover, each good i s purchased bn a different t r i p . However, three types of t r i p can be recognised, and they vary from the point of view of the distance consumers are prepared to t r a v e l . ^ F i r s t l y , there are single-purpose t r i p s which are undertaken frequently, mostly for convenience goods, where minimising the f r i c t i o n s of distance i s an important consideration. Secondly, multi-purpose t r i p s are concerned with the purchase of more than one good; they are undertaken less frequently and are usually made to higher-order centres at greater distances from the home. Although the choice of centre may be made because a good i s not found at any lower-order centre, consumers w i l l often purchase other goods, even convenience goods, which are found i n centres closer to home. On the other hand, the low frequency that the goods may be demanded and the u t i l i t y involved i n making more than one purchase may be reason enough to by-pass lower-order centres carrying the same goods. The t h i r d type of shopping t r i p i s one where shopping i s combined with some other function, for example the journey to work or workplace location. The distance travelled i s invariably greater than for purely shopping t r i p s , and the range and threshold of stores and centres so involved i s consequently larger. (b) Hierarchy and travel behaviour Central place theory assumes that the greater the frequency of a good, the shorter the distance consumers travel for i t s purchase. 7 . Garrison, W. L., et a l . , op. c i t . ; Nystuen, J . D., "A Theory and Simulation of Intra-Urban Travel", i n Garrison, W. L. and Marble, D. F. (eds.), Quantitative Geography, Part 1 (Economic and Cultural Topics), Studies i n Geography no. 13, Evanston, 111., Northwestern University Department of Geography, 1967, 54-83. However, variations have been identified in this positive relationship. In a study in rural Iowa functions were ranked according to frequency, or their order-of-entry into the central place hierarchy, and also by the mean distance that consumers travelled to maximum (where most purchases were made) and nearest (the nearest location where a purchase was made) towns. When the rankings of frequency and maximum distance travelled to nearest towns were compared, they only corresponded well for low-distance travelled goods and services such as groceries, car service, church and barber. Car sales and furniture, for example, had a low order-of-entry (high frequency) but generated long trips and variable ranges of trip distance. On the other hand, television repairs and men's clothing were less frequent, but they generated shorter distances. A further study was designed to compare the assumptions of central place theory on consumer spatial expenditure patterns for groceries 9 with a new model which would better explain actual spatial behaviour. It was shown that one of the traditional concepts of central place theory, the patronizing of the nearest place where a particular good is available, only explained 34.6 per cent, of journeys for farm households and 33.3 8. Golledge, R. G., Rushton, G. and Clark, W. A. V., "Some Spatial Characteristics of Iowa's Dispersed Farm Population and their Implications for the Grouping of Central Place Functions", Economic Geography, 42, 1966, 261-72. This paper contrasts very clearly with that by Berry, Barnum, and Tennant, op. c i t . , where consumer space preferences were inferred from a study of the location of functions in rural Iowa. 9. Rushton, G., Golledge, R. G. and Clark, W. A. V., "Formulation and Test of a Normative Model for the Spatial Allocation of Grocery Expenditures by a Dispersed Population", Annals of the  Association of American Geographers, 57, 1967, 389-400. per cent, for non-farm households. On the other hand, when the nearest place was changed to exclude those places having a population less than 1,200, the level of explanation rose to 52.4 per cent, and 44.4 per cent, respectively, suggesting that consumers were attracted more by the size of centre than whether i t was the nearest centre offering a particular good. A model was set up to measure the attractiveness of each town, based on the actual number of consumers v i s i t i n g a town of a certain size from a given distance range against the possible number of consumers living within this distance range. A comparison was then made between the most attractive towns within 25 miles of the respondent's residence and the town he actually selected for his maximum grocery purchase. The model explained approximately 60 per cent, of the journeys undertaken. These modifications of the spatial orderliness assumed in a central place-model were extended by looking at patterns of consumer spatial behaviour and the r e t a i l hierarchy on an intra-urban l e v e l . ^ It was found that a substantial proportion of consumers did not frequent the nearest centre where a good was available. Moreover, the range of a good varied with the size of shopping centre where the good was found. Consumers were prepared to travel further to purchase a good in the Central Business District or other high-order centres than to purchase the same good at a lower-order centre. (c) Preference in retailing More recent work shows that while spatial behaviour is influenced by the location of r e t a i l opportunities available to the consumer, i t is d i f f i c u l t to obtain a general theory of behaviour because 10. Clark, W. A. V., "Consumer Travel Patterns and the Concept of Range", Annals of the Association of Geographers, 58, 1968, 386-96. the nature of r e t a i l opportunities, in terms of their type and location, 11 w i l l vary from one area to another. In order to overcome this i t has been shown that consumers have a series of spatial choices for different products : they w i l l compare and evaluate alternative spatial opportunities against some personal preference function which w i l l be independent of the exact nature of r e t a i l opportunities. A measure of preference was obtained by relating the size of towns available to the consumer within certain distance ranges to the size of town chosen for a particular purchase. A comparison between consumer spatial behaviour in Iowa and Michigan showed that i f behaviour is viewed as the rate of decline of trips with distance, the different nature of settlement in the two states results in different spatial behaviour. However, the space preference structures of the two states for grocery and clothing purchases, in terms of town size and distance travelled, are essentially the same. Although more verification is needed, the method points to a conceptually sound way of analysing and understanding consumer spatial behaviour. Moreover, the way in which preference structures are defined can be expanded to include other elements which affect consumer spatial choice. 2. Differentiation in Retail Functions and Structure In the past much of our understanding of r e t a i l functions and 11. Rushton, G., "Analysis of Spatial Behaviour by Revealed Space Preference"', Annals of the Association of American Geographers, 59, 1969, 391-400; Rushton, G., "The Scaling of Locational Preferences", in Cox, K. and Golledge, R. G. (eds.), Behavioral  Problems in Geography, Studies in Geography no. 17, Evanston, 111., Northwestern University Department of Geography, 1969, 197-227; Rushton, G., "Preference and Choice in Different Environments", Proceedings of the Association of American  Geographers, 3, 1971, 146-50. structure assumed an u n d i f f e r e n t i a t e d good and a r e t a i l o utlet only d i f f e r e n t i a t e d by the size of centre i n which i t was located. Observation alone shows t h i s l i n e of reasoning to be somewhat lim i t e d : consumers have very d i f f e r e n t demands and r e t a i l e r s react to these i n very d i f f e r e n t ways. The whole f i e l d of market segmentation has thus become an important 12 factor i n business studies; adding a s p a t i a l component gives further insi g h t s into the ways i n which consumers can be d i f f e r e n t i a t e d according to the r e t a i l f a c i l i t i e s they use. Many goods, for example c l o t h i n g and furniture^ include specialty items which are i n less demand o v e r a l l , and are more expensive and/or represent more sophisticated tastes. In a study of specialty r e t a i l i n g i t has been shown that stores i n t h i s category draw a disproportionate 13 share of consumers from the higher status areas of the c i t y . Furthermore, some spe c i a l t y stores are found i n these areas, whereas c e n t r a l place theory would suggest a downtown location where the high threshold needed w i l l have the benefit of maximum a c c e s s i b i l i t y to the t o t a l market area. Studies of s p e c i a l i s t services and middle cl a s s consumers i n the Outer Metropolitan area around London have indicated that contrary to ce n t r a l place theory a p o l a r i s a t i o n of s p e c i a l i s t services i s taking place between c e r t a i n Outer Metropolitan centres and the inner area Central 14 Business D i s t r i c t s (CBDs). The former i s a r e f l e c t i o n of car ownership, 12. Engel, J . E., K o l l a t , D. T. and Blackwell, R. D., Consumer Behavior, New York, Holt, Rinehart and Winston, 1968; Johnson, R. M., "Market Segmentation : A Strategic Management Tool", Journal of Marketing Research, 8, 1971, 13-18; and Lessig, V. P. and T o l l e f s o n , J . 0., "Market Segmentation through Numerical Taxonomy", Journal of Marketing Research, 8, 1971, 480-87. 13. Leigh, R., Specialty R e t a i l i n g : A Geographical Analysis, B. C. Geographical Series no. 6, Vancouver, Tantalus Research Ltd., 1965. 14. S c h i l l e r , R.K., "Location of S p e c i a l i s t Services", Regional Studies, 5, 1971, 1-10; S c h i l l e r , R.K., "The Measurement of the Attractiveness of Shopping Centres to Middle Class Luxury Consumers", Regional Studies, 6, 1972, 291-97. 20 population dispersal and a resident population with high incomes. The latter reflects a large commuter population, tourists and international visitors. These CBD's do not attract equally from a l l parts of the Metropolitan area. Moreover, many of the Outer Metropolitan centres contain more specialist services than the larger regional centres in the same area. Even for non-specialty goods i t has been shown that the CBD's hinterland is segmented. The purchase of clothing from two downtown Vancouver department stores indicates that the stores draw their customers 15 from different social areas of the city. A number of recent studies have shown that variations also occur amongst convenience goods. It may be thought that since convenience goods are more a matter of necessity rather than taste, are required more frequently and are located closer to the home, they would not have the same variations as shopping goods. Gasoline retailing and the ways in which consumers relate to gas stations and car washes are influenced f i r s t of a l l by a number of site characteristics (e.g. position of pumps with respect to t r a f f i c flow, entrance, t r a f f i c lights, etc.) and secondly by entrepreneurial decision-making and the extent to which this takes 16 account of consumer preference. It was also found that the market was segmented. For example, some consumers would travel further in order to purchase gas at a discount operation, and differences were found between the purchasing, habits of male and female motorists. A comparison of a 15. Hardwick, W. G. and Leigh, R., Geography of Central Retailing, unpublished discussion paper, Department of Geography, University of British Columbia, 1965. 16. Claus, R. J., Spatial Dynamics of Gasoline Service Stations, B. C. Geographical Series, no. 10, Vancouver, Tantalus Research Ltd., 1969, Claus, R. J. and Hardwick, W. G., The Mobile  Consumer : Automobile-Oriented Retailing and Site Selection, Toronto, Collier-Macmillan, 1972. discount and non-discount supermarket located close together revealed the former to have a much more extensive hinterland, suggesting a total market segmented according to price-conscious or convenience-conscious 17 consumers. 3. Social and Economic Influences on Spatial Behaviour A further aspect to the differentiation of consumer spatial behaviour is the possible relationships between that behaviour and the various attributes which allow us to describe and group consumers. It is not sufficient to think of consumers only differentiating by what they specifically purchase, where they purchase i t , where they live and how far they travel. A fundamental part of social s c i e n t i f i c research has been to show that differences in overt behaviour are related to differences in people's personal situations (family background, income, age, etc.); consumer spatial behaviour is no exception. Huff developed a topographical model in which he identified 18 some twenty-five elements that affect consumer decision-making. On the basis of personal deductions he investigated the connections and relations among the elements and examined the relative degree of interdependence of each of the elements (Figure 2.1). The f i r s t two elements, stimulus situation and physiological drive are respectively the physical, social and cultural objects and processes which influence a consumer at a given point in time and a physiological condition producing 17. Elio t Hurst, M. E. and Sellers, J., "An Analysis of the Spatial Distribution of Customers around Two Grocery Retailing Operations", Professional Geographer, 21, 1969, 184-90. 18. Huff, D. L., "A Topographical Model of Consumer Space Preferences", Papers and Proceedings of the Regional Science Association, 6, 1960, 159-73. 22 S T I M U L U S S I TUAT ION P H Y S I O L O G I C A L DR IVE R E P U T A T I O N P OF S O U R C E P E R S O N A L A M E N I T I E S P B R E A D T H O F P , g » M E R C H A N D I S E SERV ICES RENDERED P PR ICE OF P R O D U C T P, T R A N S P O R T M O D E P ^ T R A V E L T IME P j 3 T R A V E L COST P. P A R K I N G COST P j j l P 2 1 M O V E M E N T I M A G E R Y OVERT B E H A V I O U R P . G E O G R A P H I C A L 5 L O C A T I O N p . ETHICAL A N D M O R A L 6 CODE P 7 E THN IC AFF I L IAT ION P g I N C O M E P , P E R S O N A L I T Y P 1 0 S E X P 1 2 A G E P n E D U C A T I O N p M E N T A L S Y N T H E S I Z I N G ABIL ITIES Figure 2.1 Basic Interactions of a Topographical Model of Consumer Space Preferences (Source : see footnote 18) some kind of need. These influence a third element, desideratum, which is the readiness on the part of the consumer to secure some object or satisfy some need. The way in which this readiness is translated into overt behaviour is through a value system, a composite of elements, such as income, ethnicity, geographical location and occupation, which condition consumer behaviour. Recent research has tended to look at many of these elements (especially income, age and ethnicity) individually and to relate variations within each one to differences in consumer spatial behaviour. A further section shows that various elements are interrelated, and i t is perhaps more relevant to study spatial behaviour from the standpoint of some generalised measure, (a) Income Probably the largest number of studies identify variations in consumer spatial behaviour and r e t a i l structure which result from differences in consumer income levels. It has been shown that low income groups travel shorter distances to shop than higher income groups, reflecting differing demand patterns and also the severe economic 19 constraints placed on mobility for those households with low incomes. The shorter distances by low-income households contribute to a persistence of old, small r e t a i l outlets with localised trade areas close to the low-income residential areas. A further study contrasted the nature of retailing and consumer behaviour of low order shopping centres in two differing income 20 areas of the city. In spite of the comparable size of the two areas, 19. Holly, B. P. and Wheeler, J. 0., "Patterns of Retail Location and the Shopping Trips of Low-Income Households", Urban Studies, 9, 1972, 215-20. 20. Davies, R. L., "Effects of Consumer Income Differences on the Business Provisions of Small Shopping Centres", Urban Studies, 5, 1968, 144-164. the low-income area contained a greater variety of functions, while the high-income area had a larger number of actual business establishments. In the low-income area there were fewer specialised stores; a higher proportion of stores carried out two or more functions. It was also found that for each function common to both areas there was a lower threshold value for the functions of the low income area. Moreover, i t was shown that higher threshold values in high income areas and the absence or lower incidence of many functions result in high income consumers travelling greater mean distances to shop. A study of the r e t a i l structure of Chicago revealed that there are marked differences in the hierarchy of shopping centres between 21 higher and lower income areas of the city. This in part reflects different demands from the two groups as well as varying spatial patterns of behaviour. It was found that in higher income areas a l l four levels of centre - major regional centres, smaller shopping goods centres, community centres and neighbourhood centres - are present, whereas in lower income areas major regional centres and community centres are absent. This suggests that since incomes are lower in these areas there is less demand (i.e. too small a threshold) for the specialised stores which characterise a major regional centre. Similarly, there is less demand for the wider range of convenience good stores which distinguish community centres from neighbourhood centres, (b) Age Huff in his topographical model states that from the point of view of linkages between the twenty-five different elements the most 22 important one is age. In a non-spatial sense i t has been shown that 21. Berry, Commercial Structure and Commercial Blight, 4-5, 60-61, 126-42. 22. Huff, "A Topographical Model of Consumer Space Preferences". different age groups have varying demands and a b i l i t i e s to pay for goods 23 and services. Compared with older families younger families have more demands made on their income for goods and services associated with housing, children and setting up a home (consumer durables such as appliances and furniture). Since there is also a positive relationship between age and income, younger families are particularly worse off for purchasing goods and services, such as convenience goods, which are essential to a l l age groups. Age also has a spatial component. F i r s t l y , older people are more likely to be physically restricted than younger people in terms of where they can shop and the mode of transportation they can use. Secondly, factorial ecological studies have demonstrated that within metropolitan areas age of families can be arranged in approximately a concentric fashion with the proportion of older families declining 24 outwards from the centre of the city. This age difference influences travel behaviour with respect to the hierarchy of shopping centres in a city. In a comparison of downtown and suburban shopping centres i t was found that a significantly higher proportion of people in the 50-64 year old group shopped downtown compared with people in the 18-34 and 35-49 year old groups. 23. See for example, Zwick, C , "Demographic Variation : Its Impact on Consumer Behaviour", Review of Economics and Statistics, 39, 1957, 451-56. 24. For example, Murdie, R. A., Factorial Ecology of Metropolitan Toronto, 1951-61, Research Paper no. 116, Chicago, University of Chicago Dept. of Geography, 1969; Rees, P. H., The Factorial Ecology  of Metropolitan Chicago, unpublished M.A. thesis, University of Chicago, 1968. 25. Jonassen, C. T., The Shopping Center Versus Downtown, Columbus, Ohio, Bureau of Business Research, College of Commerce and Administration, Ohio State University, 1955, 70, 81-82, 92. (c) E t h n i c i t y E t h n i c i t y , which constitutes a myriad of c u l t u r a l values stemming from race, language, r e l i g i o n and former geographic location, i s also related to differences i n overt behaviour. Various groups may have d i f f e r e n t r e t a i l demands and whether integrated or segregated a c i t y may contain two or more r e t a i l structures. A growing i n t e r e s t i n studies of Black America has highlighted many of the v a r i a t i o n s i n consumer behaviour between Whites and Blacks. I t has been demonstrated that predominantly White and Black areas could 2 both be low income but the nature of r e t a i l land use varied considerably. Black areas contained stores which r e f l e c t e d a greater concern for personal appearance, catered for Black pastimes or had an external appearance which i d e n t i f i e d with Black values. Further research has shown the e f f e c t on a neighbourhood commercial ribbon of changes i n the 27 r a c i a l composition of the trade-area population. As the proportion of Blacks increases the neighbourhood centre shows less emphasis i n supplying goods and more emphasis i n supplying s o c i a l services, contributing to a f u n c t i o n a l and physical b l i g h t and the greater preference for larger shopping centres. In a Canadian context i t was found that Old Order Mennonites tend to shun modern e l e c t r i c a l appliances and fashion-conscious goods i n favour of more simple goods, supplied by t h e i r own group at a l o c a l 28 l e v e l . But where Old Order Mennonites and "modern" Canadians have 26. Pred, A., "Business Thoroughfares as Expressions of Urban Negro Culture", Economic Geography, 39, 1963, 217-33. 27. Rose, H. M., "The Structure of R e t a i l Trade i n a R a c i a l l y Changing Trade Area", Geographical A n a l y s i s , 2, 1970, 135-48. 28. Murdie, R. A., " C u l t u r a l Differences i n Consumer Travel", Economic Geography, 41, 1965, 211-33. similar demands, e.g. banking and doctor and dental services, there are no significant differences in distance travelled between the two groups. In a comparison of French and English Canadian consumer behaviour i t was found that for services demanding much verbal contact consumers would tend to by-pass local services i f they were not available in their own i 2 9 language. An awareness of r e t a i l and consumer variations has resulted from the cultural geographers' interest in minority groups and the variations that result in urban landscape. Two recent studies show how the Greek and Italian groups in Vancouver, although numerically small, have different shopping demands from Canadians which result in their own stores and a willingness to travel longer distances for specialised 30 products not obtainable at closer Canadian stores. (d) Interrelationships of socio-economic attributes Research in sociology and marketing suggests that i t is more meaningful to differentiate consumer behaviour according to more generalised socio-economic measures rather than any single attribute. Many of these attributes are interrelated, and the way in which individual consumers score on some index representing these interrelated attributes allows the consumers to be placed in f a i r l y distinct and 29. Ray, D. M., "Cultural Differences in Consumer Travel Behaviour in Eastern Ontario", Canadian Geographer, 11, 1967, 143-56. 30. Grant, K. F., "Food Habits and Food Shopping Patterns of Greek Immigrants in Vancouver", in Minghi, J. V. (ed.), Peoples of the  Living Land : Geography of Cultural Diversity in British  Columbia, B. C. Geographical Series no. 15, Vancouver, Tantalus Research Ltd., 1972, 125-44; Gale, D., "The Impact of Canadian Italians on Retail Functions and Facades in Vancouver, 1921-1961", in Minghi, J. V. (ed.), op. c i t . , 107-24. ranked groups. The groups may convey a wider meaning than merely the sum of individual attributes. The combined scores for interrelated attributes such as education, income and occupation have been organised into groups implying socio-economic status or rank. But the groups are not just the result of classification. It i s shown in Chapters III and IV below that the groups, or social classes, exemplify a wide range of variation in values and attitudes, and there are also the notions of conflict and superiority-inferiority between groups. A problem associated with generalised groups is that they are less homogeneous in their composition than, for example, individual income groups. However, this weakness i s compensated by the fact that these generalised groups have an internal consistency with respect to many forms of overt behaviour. Furthermore, Coleman cites the weakness 31 of using a single attribute to differentiate consumer behaviour. Three families can each have a gross income of $8,000, but since the heads of household have widely differing occupations (a young lawyer, an insurance salesman and a welder), i t can be expected that many other values and behavioural patterns w i l l d i f f e r . Using interrelated personal attributes to differentiate overt behaviour has not been widespread in geographic research. Indeed, the only major area where such attributes have featured is in the fac t o r i a l -ecological studies of aggregate residential behaviour. Other disciplines, 31. Coleman, R. P., "The Significance of Social Stratification in Selling , in Marketing : A Maturing Discipline, Proceedings of the Winter Conference of the American Marketing Association, 1960, Chicago, American Marketing Association, 1961, 171-84. 32 however, have used such measures to differentiate consumer behaviour, and the present study w i l l extend the work by relating some spatial perspectives of consumer behaviour to variations in interrelated personal attributes. Differences in Consumer Attitudes The third approach to consumer spatial behaviour as yet has been subject to far less intensive research. Its aim i s to understand better why consumers behave in the way they do. This type of research is a natural outgrowth of the second approach which was discussed above : in looking at variations in r e t a i l a c t i v i t i e s and consumer behaviour and how these are related to differences in consumer attributes, i t must s t i l l be assumed, for example, why low income consumers travel shorter distances or frequent different stores compared to high income consumers. To be able to answer these questions i t i s necessary to interpret consumer attitudes and their perceptions of the shopping environment around them. Furthermore, a different type of questionnaire must be constructed : questions must now e l i c i t opinions, rather than factual replies, and therefore there is the added problem of interviewer and interviewee bias. Most work so far has been concerned with grouping consumers according to store preference or attitudes about particular stores. The c r i t e r i a upon which consumers judge stores are needless to say 32. Martineau, P., "Social Classes and Spending Behavior", Journal of  Marketing, 23, 1958, 121-30; Carman, J. M., The Application  of Social Class in Market Segmentation, Research Program in Marketing, Graduate School of Business Administration, University of California, Berkeley, Institute of Business and Economic Research, 1965; Engel, Kollat and Blackwell, op. c i t . , 309-24. extremely numerous, extending from factual answers concerning stores to consumers' perceptions of the wide range of factors which contribute to store image, to the environmental conditions and cultural factors which colour a consumer's perception. Moreover, different types of store and product are judged according to different c r i t e r i a . One study defines 26 c r i t e r i a , or variables, on which to judge consumer 33 preference; a further study on clothing store preference, on the other hand, identifies 172 variables on the basis of what sample respondents 34 mentioned. The large number of variables presents no handling problem, but there is the dilemma that consumers are not conscious of a l l the c r i t e r i a on which their image of a store is based. A major aspect of this research is reducing the large amount of data into a smaller number of underlying dimensions, or interrelationships of consumer attitudes. Not only does this simplify the problem but affords some comparison with other work. Using factor analysis i t was found that the 172 c r i t e r i a used in shopping for clothing could be grouped according to ten underlying dimensions, and on the basis of factor 35 loadings names could be given to eight of these. From factor score similarities i t was then possible to place the sample respondents into six groups - the young, fashion-conscious shopper, the bargain-hunter, 33. Downs, R., "The Cognitive Structure of an Urban Shopping Centre", Environment and Behaviour, 2, 1970, 13-39. 34. Taylor, S. M., Spatial Perspectives at the Consumer-Store Interface, unpublished M.A. thesis, University of Br i t i s h Columbia, 1972. 35. Ibid., 100-103. The named factors identified are boutiqueness, cheapness, r e l i a b i l i t y , convenience, exclusiveness, display, security and noisiness. 31 the older specialty store customer, etc. It was then possible to relate the ten factors to the respondents' socio-economic and demographic characteristics, thereby giving greater depth to various aspects of the second approach that were discussed. Conceptually and methodologically this type of research plays an important role. It i s possible for retailers to have some precision about the type of consumer they attract, and what qualities they must change in order to attract other consumers. We also have a greater understanding of how consumers and r e t a i l stores can be classified. Summary This chapter has identified three basic approaches to the study of consumer spatial behaviour. The f i r s t makes assumptions about the ways in which consumers behave, based on the location, size and spacing of r e t a i l a c t i v i t i e s . However, i t is only possible to make generalisations : i t must be assumed that a l l consumers behave alike with respect to the one good and that a l l r e t a i l outlets selling the one good are the same. An approach where the consumer is interviewed can test these generalisations and examine the nature of variation from them. Consumers are very different in their demands, and i t is possible to relate variations in r e t a i l activities to differences in consumer attributes. A further approach lends greater depth to the analysis by examining underlying consumer attitudes towards different types of purchases or stores. The present study extends the second approach by analysing some aspects of consumer spatial behaviour with respect to interrelated, rather than individual, personal attributes. The next two chapters examine the nature of one of these interrelationships (social class) and the problems associated with the measurement of such groupings of attributes. Chapter III Concepts of S o c i a l Class Introduction S o c i a l c l a s s , which constitutes one of the two sets of i n t e r r e l a t e d personal a t t r i b u t e s being considered i n t h i s d i s s e r t a t i o n , i s a subject that has produced considerable debate and c o n f l i c t i n both the academic and the p r a c t i c a l sphere.''' Since i t i s not a f a m i l i a r subject i n geographic research, some of the major developments are very b r i e f l y discussed i n t h i s chapter. Class can be defined in two quite d i s t i n c t ways. F i r s t l y , i t can mean mutually exclusive groups, which may be r e a d i l y i d e n t i f i a b l e , such as White and Black and Roman Catholic and Protestant. Membership of such groups can be either hereditary or by choice, and each group has some form of d i s t i n c t i v e behaviour or mores. The second concept 1. Works which give a good t r e a t i s e of the development of ideas on s o c i a l class include Gordon, M. M., S o c i a l Class i n American  Sociology, Durham, Duke University Press, 1958; K a r l , J . A., The American Class Structure, New York, Rinehart and Co., 1957; Tumin, M. M., S o c i a l S t r a t i f i c a t i o n : The Forms and  Functions of Inequality, (Foundation of Modern Sociology S e r i e s ) , Englewood C l i f f s , N. J . , Prentice H a l l , 1967; Tumin, M. M. (ed.), Readings on S o c i a l S t r a t i f i c a t i o n , Englewood C l i f f s , N. J . , Prentice H a l l , 1970; Laumann, L. 0. (ed.), S o c i a l S t r a t i f i c a t i o n , New York, Bobbs-Merrill, 1970. of class, and the one that this chapter addresses i t s e l f to, refers to the horizontal stratification of a population according to a variety of forms of economic, social and p o l i t i c a l behaviour. These various forms of classification are invariably reduced to five or six broad categories, or social classes, which researchers have attempted to identify, analyse and measure and to use as a basis for differentiating other behavioural patterns (such as in the case here of the consumer spatial behaviour). Since so many factors go into making up social class, i t is small wonder one writer comments that the proliferation of professional monographs on the subject has failed to produce a substantial body of complementary and comparable research which results in a cumulative 2 knowledge about the phenomena themselves. Furthermore, working against the notions of social class in North America is the commonly-held and popular belief that class divisions do not exist at a l l . This results from the idea that everyone has an equal chance to attain a certain position in society, i.e. unlimited social mobility. Two areas of interest brought attention to the ideas of social class, and these are expanded in this chapter. F i r s t of a l l there were the writings of people who could be called social and p o l i t i c a l reformers, and who were concerned with the general theme of subservience and degradation of one group compared to another. The second area of interest is later and connected with the development of sociology as 2. Gordon, M. M., op. c i t . , 4. a d i s c i p l i n e . S o c i a l classes emerge from attempts to explain differences in i n d i v i d u a l behaviour and group r e l a t i o n s h i p s . The studies undertaken seek close t i e s with s o c i a l theory, o b j e c t i v i t y i n t h e i r approach and p r e c i s i o n i n their analysis and measurement of s o c i a l c l a s s . The Nineteenth Century. Reformers A d e f i n i t i o n of s o c i a l c l a s s i n the period before the I n d u s t r i a l Revolution presents fewer problems than i n the period since then. Classes could be r e a d i l y i d e n t i f i e d by t h e i r wealth and power. The r o y a l f a m i l i e s and r u l i n g n o b i l i t y , the Church, the merchants, the craftsmen and t h e i r guilds, the labourers and landless peasants each played a d i s t i n c t r o l e i n maintaining a s o c i a l and p o l i t i c a l hierarchy i n any society. Upward mobility was v i r t u a l l y impossible since the group above could c o n t r o l i t s membership and block any competition that would come about through increased s i z e . I n d u s t r i a l i s a t i o n i n the nineteenth century, on the other hand, upset the old order. The middle class of i n d u s t r i a l i s t s , craftsmen and businessmen had a rapid r i s e i n economic fortunes and were able to win for themselves s o c i a l and p o l i t i c a l power at the expense of the old r u l i n g classes. Meanwhile there was increased p o l i t i c a l freedom for more people, improved means of communication, better education, more l e i s u r e time and better health and l i v i n g conditions. However, these improvements were much slower coming to the labouring classes. For these people the I n d u s t r i a l Revolution had the e f f e c t of s u b s t i t u t i n g one r u l i n g c l a s s f o r another and t r a n s f e r r i n g them from the deprivations 3 of a r u r a l existence to the squalor of an i n d u s t r i a l town. I t was against t h i s background that two c l a s s i c works on s o c i a l c l a s s were written i n the l a t t e r half of the nineteenth century. The f i r s t viewed society as being a number of forms of production, including a g r i c u l t u r e , industry and the s k i l l e d trades, and within any one form of production i t was possible to i d e n t i f y d i f f e r e n t classes 4 on the basis of the r o l e that people played. In f a i r l y s i m p l i s t i c terms i t was a c a p i t a l i s t minority, owning the means of production and maximising p r o f i t s and an exploited p r o l e t e r i a t majority providing the labour which was e s s e n t i a l for the former group to make a p r o f i t . However, one of the major c r i t i c i s m s that can be made of such class d i v i s i o n s i s that they are too r i g i d and ove r - s i m p l i f i e d . Cer t a i n l y by the lat e nineteenth century i n western Europe and North America one cannot t a l k of clear-cut d i v i s i o n s between c a p i t a l i s t s and workers. Improvements i n educational opportunities, f o r example, enhanced s o c i a l mobility and allowed people to enter sectors of the economy from which they were formerly barred. Moreover, the number of classes that could be i d e n t i f i e d i n any society was never as few as the two major ones above. The development of a sophisticated i n d u s t r i a l economy resulted i n a growing middle cl a s s of businessmen and professional workers. Through the services they 3. A b r i e f comparison of the l i v e s of the various s o c i a l groups i n the nineteenth century c i t y can be seen i n Mumford, L., The City i n History, London, Seeker & Warburg, 1961, 508-48. 4. Marx, K., Das C a p i t a l , 1967-83; translated by Untermann, E., C a p i t a l , Chicago, C. H. Kerr & Co., 1909. 36 offered these people could also dictate terms to the working class. Yet the extent to which people could afford these services, and thus the prosperity of this middle class, often depended on economic forces outside the control of these two groups. The middle class could not be easily labelled either exploiters or exploited. Marx's major contribution to sociology and the study of social class was to provide an intellectual stimulus to both the discipline and the subject area when they were new in the academic world. While the rig i d i t y and oversimplification of Marxist thought did not lend i t s e l f well to later social class studies, the emphasis on economic power as a means of determining status level has been expanded and refined. The second important work on social class resulted from a 5 study of poverty in London. From work by others and personal communications (with Booth) i t was seen how l i t t l e was known about the extent of poverty in London, and using personal interviews and more especially unpublished materials an attempt was made to describe and categorise urban poverty. In order to reduce the overwhelming mass of data, households were assigned to one of eight classes on the basis of occupation, income and the size and type of family dependent on this income.^ The groups in ascending order or prestige were as follows:-5. Booth, C., Life and Labour of the People, London, Williams & Norgate, 1889, vol. I; Labour and Life of the People, London, Williams & Norgate, 1891, vol. II; Li f e and Labour of the People in London, London, Macmillan & Co. Ltd., 1892-97, 9 vols.; L i f e  and Labour of the People in London, London, Macmillan & Co. Ltd. 1902-3, 17 vols. 6. Booth, C., Life and Labour of the People, I, 33. 37 1. A group, which was d i f f i c u l t to enumerate because they lived and slept rough, consisting of beggars, street sellers, petty criminals, prostitutes, etc. 2. Those people dependent on casual labour, largely because they were physically and mentally incapable of holding full-time employment. 3. Seasonally employed people who when working were able to earn high wages. 4. Regularly employed people in unskilled trades who in a year could earn as much as those people in 3 above. 5. Regularly employed people, again in unskilled trades but where standard of living was improved by children working and living at home. 6. Highly-paid manual workers in jobs demanding s k i l l and responsibility. 7. White-collar c l e r i c a l workers perhaps earning less than those in 6 above but having more prestigous work and greater job security. 8. An undifferentiated group which Booth was not concerned with, constituting royalty, landowning nobility and business and professional men. This extensive analysis of London's population has had l i t t l e impact on the later work of sociologists.'' Yet from the point of view of social class he raised a number of issues, and inferred others, which were to be expanded upon, without reference, by later authors. Booth recognized that the number of classes and the dividing lines 8 between them were chosen ar b i t r a r i l y . This reflected f i r s t l y a pragmatic approach, designing classes to f i t the problem, and secondly 7. Reid, A. & Elman, R. M. (ed.), Charles Booth's London, New York, Partheon Books, 1968. This work is a series of extracts from Booth's original work and is offered without a comment. A second work, Pfantz, H. W. (ed.), Charles Booth on the City :  Physical Pattern and Social Structure, Chicago, University of Chicago Press, 1967, c r i t i c a l l y analyses Booth's work and in doing so recognises the contribution he made to an understanding of the city as a spatial pattern and social structure. 8. Booth, Life and Labour of the People, I, 61. the perspective on the status system of the community from a member of the upper-middle class. Since the problem under discussion was relative poverty and standards of living, the eight-class structure was essentially based on income. However, numerous references were made to the fact that the income variable alone could not be used to designate a family's social class, especially when comparing the lower-middle class and the skilled and highly-paid working class. In the eyes of both classes i t was found that lower-middle class occupations carried 9 more status, yet were often less well paid. Meanwhile i t was recognised that many occupations did not correspond closely to a particular social c l a s s . ^ The status position varied according to whether employment in any one occupation was irregular, seasonal or regular, the degree of s k i l l that might be required, the length of time one had been employed in a particular job and the status of the person whom the work was performed for. Research into the concepts of social class The development of social class concepts has been very closely tied to the emergence of sociology as a discipline. F i r s t of a l l there has been a move from empiricism to seeking and testing theories concerning individual and group behaviour. In doing so i t has been necessary to make generalisations about behaviour, and where similarities exist, regardless of whether there is any social interaction, individuals have been grouped together. Moreover, rather than use 9. For example, Booth, Li f e and Labour of the People in London, 1902-3, "Religious Influences", VII, 399. 10. Ibid., "Poverty Series", III, 260. social classes to f i t a particular problem under discussion, as Marx and Booth did, sociologists have been interested in establishing generalised and hopefully universal, social classes as one means of understanding how a community functions. One important early work was responsible for refining and expanding upon Marxist thought on social class. Weber, like Marx, maintained that ownership and control over property were the underlying factors in determining the life-chances of an individual. However, a larger number of classes were now identified, and social class was based on more than one (economic) dimension. Two other dimensions were identified. F i r s t l y , these were prestige differences which generated status groupings. Individuals grouped together as a means of and as a result of engaging in certain l i f e styles, and in return expected varying degrees of privilege which usually, although not necessarily, resulted in status groups and economic classes becoming synonymous. The third dimension was power which grouped i t s e l f by p o l i t i c a l parties. A hierarchy of parties reflected the varying a b i l i t i e s of followers to recruit members and in turn influence events in the community. The party with the most power need not necessarily carry the most prestige or economic standing. This multi-dimensional approach to social class formed the basis of a number of studies in North America. One such study undertaken in "Middletown", a small Mid-Western industrial city, set 11. Henderson, A. M. & Parsons, T. (trans.), Max Weber : The Theory of  Social and Economic Organisation, New York, Oxford University Press, 1947; Gerth, H. & M i l l s , C. W., From Max Weber, New York, Oxford University Press, 1958; Bendix, R., Max Weber :  An Intellectual Portrait, New York, Doubleday & Co. Inc., 1960. out to analyse the functioning of a community.'1''' In the f i r s t analysis only two classes were identified : a working class who were responsible for making goods and performing services and a business class who sold 13 or promoted goods, services and ideas. This latter group also included the upper echelons of a service population, such as lawyers, doctors and engineers. Gradations within each group and overlap between the groups were recognised, but the simplistic division was upheld because of i t s overall functional significance. A difference in power relationships was an underlying factor in the twofold stratification, with the business class holding a position of dominance over the working class. This dominance in part reflected differences in economic power, measured in terms of income, expenditures and wealth. In a further analysis the twofold divisions was expanded to s i x , ^ as follows:-1. A small group of wealthy families, engaged in industry, business and law and exerting most p o l i t i c a l control. 2. A larger but s t i l l relatively exclusive group, consisting of less well-paid professional people, small manufacturers, shopowners, etc. 3. Lower-paid white-collar workers, e.g. clerks, salesmen and the lower echelons of the c i v i l service. 4. Blue-collar workers in positions of s k i l l or trust. 5. Semi-skilled and unskilled workers, comprising the bulk of the population. 6. Unskilled people incapable of holding regular employment. 12. Lynd, R. S. & Lynd, H. M., Middletown, New York, Harcourt, Brace & Co., 1929; and Middletown in Transition, New York, Harcourt, Brace & Co., 1937. 13. Middletown, 23-24. 14. Middletown in Transition, 458-60. It was found that built into these various groupings was a hierarchy of status, power, the part that each group played in the community and patterns of social behaviour. Whilst income was a factor in giving status and power i n i t i a l l y , i t s importance was played down once a community had established i t s e l f . Perhaps this was because during the height of the Depression, when the second Middletown study was undertaken, small manufacturers in group 2 and semi-skilled workers in group 5, for example, may have both found themselves unemployed and financially destitute; but this did not mean that both groups were reduced to the same level in terms of status and power. Furthermore, there was increasing overlap in terms of income between the menial office worker and the skilled and semi-skilled worker. Instead of income, the underlying variable used to denote social class (and i t s status and power ramifications) in Middletown was the nature of occupations. Similar social class categories were found in the extensive research undertaken on the structure and social l i f e of a small New 15 England City ("Yankee City"). Unlike the Middletown studies class was now defined as a status group where the ranking was determined by the evaluations of members of the community; and for an individual to be put in a certain class he must f i r s t of a l l participate in the activities and associations of that class and secondly must be accepted 15. Warner, W. L. & Lunt, P. S., The Social L i f e of a Modern Community, New Haven, Yale University Press, 1941, "Yankee City Series", I; The Status System of a Modern Community, New Haven, Yale University Press, 1942, "Yankee City Series", II; Warner, W. L & Srole, L., The Social Systems of American Ethnic Groups, New Haven, Yale Univ. Press, 1945, "Yankee City Series", III; Warner, W. L. & Low, J. 0., The Social System of a Modern  Factory, New Haven, Yale University Press, 1947. by i t s members. The variables adopted to measure social class, or more correctly prestige ratings, were the ones that Warner believed Americans used to judge each other's social standing, and included occupation, source of income, type of house and quality of residential 16 area. The following six social classes resulted from the groups that the residents of "Yankee City" could i d e n t i f y . ^ They included:-1. Upper-Upper class - the oldest established families in the community, possessing wealth, status and power. 2. Lower-Upper class - recently-acquired wealth and striving to establish themselves in the community. 3. Upper-Middle class - professional people, businessmen higher managerial staff and young people likely to reach these positions within ten to fifteen years. 4. Lower-Middle class - the lowest-paid white-collar workers. 5. Upper-Lower class - skilled and semi-skilled workers. 6. Lower-Lower class - the unskilled worker, especially unassimilated immigrants and sporadically-employed people. For the f i r s t time an elaborate, yet objective, method to measure social class was set up (once the variables had been subjectively chosen). However, an important problem is the extent to which social class studies carried out in a small urban centre in New England are applicable elsewhere. In a small community members of one group are more aware of the behaviour of other groups than in a larger community 16. This is discussed in more detail in Chapter IV below which deals with the problems of measuring social class. 17. Warner, W. L., Social Class in America : The Evaluation of Status, Chicago, Social Science Research Associates Inc., 1949. where physical separation is more emphasised. Therefore, a social class methodology resulting from status groups identified by residents may lead to different types or numbers of groups in another area or a larger centre. A further problem that has been raised questions whether one can talk about distinct status groups. A number of researchers have argued that instead of groups individual status positions constitute a status continuum. While the continuum may be divided up arbitrarily for purposes of convenience, there is no theoretical justification 18 for doing so. It is argued that any individual is not able to place himself in a group, nor can he see divisions between groups. While he recognises a status hierarchy, i t is merely a scale on which he places himself and others he knows above and below him. These notions are borne out by empirical research. For example, where respondents are asked to identify groups, the number varied from three to seven. Among those respondents using the same number of groups there is no agreement on the division between the groups. The results of the status continuum research do not prove, however, that there are no social classes. The nature of the questioning often asks respondents to rank, rather than group, individuals. Moreover, even where grouping is asked for, there is l i t t l e attempt on the part of the researchers to make generalisations about the individual 18. Cox, 0. C , Caste, Class and Race, New York, Doubleday & Co. Inc., 1948; Lenski, G. E., "American Social Classes : S t a t i s t i c a l Strata or Social Groups?", American Journal of Sociology, 58, 1952, 139-44; Cuber, J. F. & Kenkel, W. F., Social Stratification  in the United States, New York, Appleton-Century-Crofts, Inc., 1954; Hetzler, S. A. "An Investigation of the Distinctiveness of Social Classes", American Sociological Review, 18, 1953, 493-97. results. The apparent conflict between social class and status continuum can perhaps be resolved : the degree of class consciousness may vary considerably from one area to another and between individuals in the same area. Class membership may be identified where there are barriers against upward social mobility. Where such barriers are weak or do not exist, social mobility and interaction are sufficiently extensive to remove the sense of class membership from people's minds: hence, the American dream of a classless society since everyone has the same opportunity, supposedly, to advance himself. However, since economic, social and p o l i t i c a l inequalities exist, a hierarchy of classes in some form must exist also. Much of the work on social class has been done in small communities or has been based on a relatively few number of respondents. Porter's extensive study of social class, on the other hand, not only gives an invaluable Canadian perspective, but views the problem on a 19 national, or macro, level. At this scale Porter sets out to disprove the ideas that Canada is a classless society (a feature more common of pioneer days, i f i t even existed then), or an essentially middle-class society lacking the great extremes of wealth and poverty. It is shown that on the basis of income, education, ethnic status and the development of power and economic elites, inequalities exist in society which result in feelings of inferiority and superiority to others and perhaps desires to attain upward social mobility. This work, however, does not establish social classes per se since the data collection is almost entirely from published s t a t i s t i c a l 19. Porter, J., The Vertical Mosaic : an Analysis of Social Class and  Power in Canada, Toronto, University of Toronto Press, 1965. sources and the methodology to enable social classes to be differentiated i s not available. Instead, s t a t i s t i c a l classes are used, based on the manner in which each variable is recorded and classified. From the interrelationships of these different variables i t i s s t i l l possible to talk about e l i t e middle-class and lower-class groups in an extensive and very generalised way. Summary Social class has been analysed with respect to highlighting the need for social and p o l i t i c a l reform and as a means of understanding the structure and behaviour of a community. It refers to the horizontal, and more specifically the hierarchical, stratification of a population where one group carries more or less status or prestige than another group and a desire exists for upward social mobility. Membership of a particular group implies some degree of similarity in terms of socio-economic attributes and values with other members of the group and social interaction mainly with people similar to oneself. However, a problem which presents i t s e l f throughout this study is whether broad social classes can be identified in the community, or whether they are merely a convenient way of classifying a population which cannot be differentiated in any other way. The weight of theoretical and empirical evidence suggests that five or six broad categories can be seen, and these can be used in subsequent research where one needs to differentiate a population. The problem now remains to measure social class and assign a sample population to one. Chapter IV Measuring Social Class and Family Status Introduction This chapter is concerned with the adoption of a methodology which w i l l allow consumers to be classified into social class and family status groups. It is shown that measuring social class, especially, is a problem area that is thwart with subjectivity. The f i r s t part of the chapter contrasts the two different types of variables which are used to measure social class. Since social class differences result from variations in values, beliefs, attitudes and motivations, measurement of these variables (real variables) w i l l allow a population to be classified. However, the use of such variables is problematical since they themselves are d i f f i c u l t to measure. A second way is to use proxy variables. Differences in attitudes, values and so on reflect and/or result in differences in variables such as education, income, occupation and family size. While i t is debatable whether proxy variables measure social class or merely indicate socio-economic status, they are easier than real variables both to measure and to obtain. Proxy variables are used in this dissertation, and the second part of the chapter is concerned with an evaluation of the various methods that can be used to organise these variables and classify individuals. The study focusses on the development of multi-variate s t a t i s t i c a l techniques, especially factor analysis, since i t is possible to use proxy variables to obtain more than just social class; other interrelated variables, including family status and ethnicity, may also be identified. Real Variables Since social classes reflect differences in values, beliefs, attitudes, motivation and behaviour^ there have been attempts by sociologists to measure these variables directly, giving weights to particular answers and constructing scales on the basis of scores derived. One of the f i r s t scales to be constructed was by 1 F. Stuart Chapin. He postulated that social class could be represented by four factors, including cultural possessions, material possessions, effective income and participation in group activity in the community. The reason for choosing these four factors is nowhere explained and their relationship to social class attitudes, values, etc. is likewise unknown. The reader is l e f t to assume that the answer to a particular question w i l l be indicative of the social status of the respondent. Material possessions consisted of l i t t l e more than an inventory of household furniture and appliances; cultural possessions related to musical instruments, books, records, etc. The presence of any of these in the home was given an arbitrary weight, and social status was ranked according to overall score. Participation in the group activity of the community was weighted highest in the case of someone being an officer 1. Chapin, F. S., "A Quantitative Scale for Rating the Home and Social Environment of Middle Class Families in an Urban Community", Journal of Educational Psvcholoev. 19. 1928. 99-111. in an organisation through to no weight at a l l for someone who was not a member of anything. Again i t was assumed that the highest weight carried the highest social class standing. It was found, moreover, that the weight given to material possessions correlated so well with the combined weights of a l l four factors that the former could be taken as a social class index. A similar scale was developed where school children answered questions dealing with material and cultural possessions in the home, the occupation and income of parents and the outside interests of 2 parents and children. The answers served a twofold purpose. They could relate to the scholastic and eventual occupational achievement of the child, while providing a useful means of distinguishing the social class of the home. These two works are sufficient to point to the problems in using such scales. While subjectivity w i l l always remain with regard to the variables chosen, subjective judgments were also made for so many of the answers, often on the basis of scanty or non-existent evidence. Sims, for example, asked a question concerning attendance at concerts, assuming no attendance to imply a lower social class position than frequent attendance. Although, intuitively, the implication was probably f a i r l y sound, there was a lack of conceptual and empirical evidence to reinforce his ideas. 2. Sim, V. M., The Measurement of Urban Home Environment, Minneapolis, University of Minnesota Press, 1936. The social class scales seen above do not lend themselves very well for use by other social scientists. Because of changing fashions, rising standards of living and rapid social mobility, some of the detailed questions that are asked either change in their suitability or become irrelevant. Therefore, there is a constant need for reassessment and updating the questions. For example, taking a European holiday was once a means of distinguishing an upper or upper-middle class person from the remainder of the population. Today with inclusive tours and charter flights the same assumption i s no longer true. On the other hand the proxy variables that have been used to designate social class are for the most part as relevant today as they were f i f t y or more years ago. Although factors such as occupation, income and type of residential area change in nature over time, they are as applicable for differentiating people in late twentieth-century North America as they were in late nineteenth-century London. Proxy Variables There are two contrasting viewpoints concerning the use of proxy variables such as income and occupation, for establishing social classes. One is that since so many (real) variables go into making up social class and the differences between classes are so varied and subtle, the use of a few proxy variables cannot give us 3 a complete picture of social class. 3. It is argued in much American literature that the use of proxy variables, such as income and occupation, results in the establishment of an individual's socio-economic status rather than his social class : social class is said to constitute many other factors, including differences in power and kinship ties. However, d i f f i c u l t i e s in measuring many of these have resulted in socio-economic variables becoming the major determinants of social class position. A second viewpoint, which is stronger and also more re a l i s t i c , is that proxy variables not only measure structural differences in society, but represent a large number of real variables affecting social class, many of which may, in fact, are impossible to measure. Occupation, for example, expresses a whole range of status-conscious and different class situations, in one's family background and upbringing, educational opportunities, personal motivations and response to other societal influences such as the mass media, peer groups and organisations to which one belongs. It would be a d i f f i c u l t task to measure a l l these influences on occupation. F i r s t of a l l they are not known; secondly, i t would have to be assessed, since not a l l influences are of the same importance, what weights would be placed on the many variables. Methodology for Using Proxy Variables The remainder of this chapter is concerned with the methodology for using proxy variables both to assign individuals to a social class and to reveal other interrelationships amongst personal attributes. 1. Early Multiple Factor Designs One of the most common ways of measuring social class using proxy variables is to adopt the techniques designed by W. Lloyd Warner 4 and his associates for their "Yankee City" studies. He conceived a two part system : Evaluated Participation (E.P.) where informants would be asked to rate their position and the position of other individuals in the community, and secondly an Index of Status Characteristics (I.S.C.) which would u t i l i s e the E.P. information to provide an easier system for placing individuals into a particular class. In the E.P. design an informant was required to evaluate himself and others by social class, and a number of rating techniques were used for this end including: (a) Matched agreements. The researcher would see to what extent there was agreement between informants on the same subject. (b) Symbolic placement. The researcher would extract information which might have a status connotation, e.g. 'the wrong side of the tracks'. (c) Status reputation informants frequently make remarks about personal traits of others which might have a status connotation, e.g. a leader or a 'down and out'. (d) Comparison. Informants would be asked to rate a person superior, equal or inferior to others who had already been assigned to a social class. (e) Simple Assignment. Informants would be asked to assign an individual to a social class without reference to others. (f) Institutional membership. Informants gave information about the institutions to which people belong. The status position of these institutions had already been determined. Warner's E.P. design has been c r i t i c i s e d on a number of counts."' It was not clearly stated how the informants were chosen, 4. Warner, W. L., Social Class in America : the Evaluation of Status, Chicago, Social Science Research Associates Inc., 1949. 5. Gordon, M. M., Social Class in American Sociology, Durham, Duke University Press, 1958, 104-108. except they were shown to be biased in favour of a professional upper-middle class group. These people, i t might be argued, were the best informed in the community, but i t did not mean to say that they knew more about the people they had to assess than informants in a wider based sample. The informants could not agree as to the number of status groups which could be recognised, and i t must be asked whether the f i n a l figure of six was the number that the researchers had established beforehand. The E.P. design was not only cloaked in a good deal of mystery; i t s results suffered from so much conflicting subjective judgment as to make i t s use doubtful. The Index of Status Characteristics (I.S.C.), on the other hand, was an attempt to develop a quicker and less costly measuring technique. It was a combination of factors which with certain weights was found to correlate highly with the E.P. determined statuses. It was hypothesised that six factors - occupation, source of income, house type, dwelling area, income and education - correlated highly with the E.P. determined statuses. The f i r s t four factors only were used since they correlated almost as highly as a l l six (0.972 as opposed to 0.974). A seven point scale was constructed for each factor, from 1 representing the highest status through 7 - lowest status. A score for any individual could be obtained by multiplying each weight by his rating on the seven point scale. A total of 12 would denote highest status, 6 84 - lowest status. Warner's Index of Status Characteristics has one major point in i t s favour : the methodology involved is easy for other social scientists to grasp, and providing one accepts the fact that the Index 53 can be applied in another community, i t is a convenient way of determining social class. Criticism has been levelled at Warner because two of the variables, house type and dwelling area, depend on the subjective judgment of the interviewer. For example, what distinguishes a good 7 from a f a i r house? Moreover, whilst more objective variables could have been chosen, such as size of house and density of housing area, Warner did overcome interviewer bias to a large degree by setting up beforehand 8 a check l i s t of the seven ratings for each variable. A number of other studies since the 1930's have been designed to divide the community into social class groups. These included the 9 designation of status both by the researcher and the informant. 6. Warner, op. c i t . , .41. Index of Status Characteristics (I.S.C.) Status Characteristic Ranking Weight Weighted Rating 1-7 Occupation x 4 = Source of Income x 3 = House Type x 3 = Dwelling Area x 2 = Grand Total = Totals: 12-21 (Upper-Upper and Lower-Upper), 22-37 (Upper-Middle), 38-51 (Lower-Middle), 52-66 (Upper-Lower) and 67-84 (Lower-Lower) . 7. Gordon, M. M., op. c i t . , 111. 8. Warner, W. L., op. c i t . , 143-54. 9. Dollard, J., Caste and Class in a Southern Town, New Haven, Yale University Press, 1937; West, J., Plainsville, U.S.A., New York, Columbia University Press, 1945; Hollingshead, A. B., Elmtown's Youth, New York, J. Wiley, 1949; Kaufman, H. F., Prestige Classes in a New York Rural Community, Ithaca, N.Y., Cornell University Agricultural Experiment Station, 1944; and Laumann, E. 0., Prestige and Association in an Urban Community, New York, The Bobbs-Merrill Co. Inc., 1966. However, the value of these studies lies more in their in-depth analyses of American communities than in their refinements of the means of measuring social class and presenting a methodology that is easily adaptable for other disciplines. 2. Occupation groupings A recurring theme in social class research is to use a single proxy variable, occupation, rather than a composite of variables, to denote social class. The attainment of a certain occupation w i l l imply a particular family background, education and so on, and once attained an occupation w i l l generate certain attitudes, beliefs and l i f e styles. Occupations can be divided into functional groups in a number of ways, including prestige rankings by respondents and by the researcher himself and a ranking based on the average income and years of schooling in each occupation. Prestige ranking of occupations by respondents can be seen best in the North-Hatt S c a l e . ^ In this research, carried out in many parts of the United States, respondents were asked to evaluate each of 90 occupations on the basis of a five-point rating scale, from excellent to poor. For each occupation a score was devised from the percentage of replies in each rating category. Excluding "Don't Know" answers, the scoring theoretically varied from 100 points for an occupation receiving only "excellent" ratings to 20 points for one receiving a l l "poor" ratings. It was found that despite variations in the evaluation 10. North, C. C. and Hatt, P. K., "Jobs and Occupations : a Popular Evaluation", in Wilson, L. and Kolb, W. L. (eds.), Sociological  Analysis, New York, Harcourt, Brace & Co., 1949, 464-74. of occupations according to size of place and age, education, and economic level of the respondent, there was basic agreement between the different parts of the United States. A second type of rating is the "common-sense" functional classification where the researcher divides occupations according to his own prestige rating. One of the most common of these was proposed 11 by Edwards. Again, i t had nationwide applicability, and for this reason was adopted by the U. S. Bureau of the Census. Six groups were identified, including professional persons, proprietors, managers and o f f i c i a l s , clerks and kindred workers, skilled workers and foremen and semi-skilled and unskilled workers. While being functionally convenient such a scheme has l i t t l e value as a social class index. Any one occupation t i t l e can convey a wide range of prestige. Professional workers, for example, w i l l include a University President and a chorus-g i r l . Moreover, the one occupation w i l l often have a wide salary range which w i l l frequently result in a variation in l i f e styles and hence prestige levels. Both the North-Hatt and Edwards scales suffer from subjective judgment. However, this is unavoidable so long as social class is measured by how a group or an individual perceives and evaluates occupations. A more objective method of using occupations to measure social 12 class was developed by Blishen. Using the 1951 Census of Canada, he constructed a rank order of 343 occupations with a score based on 11. Edwards, A.M., A Social-Economic Grouping of the Gainful Workers of the United States, Washington D.C., U.S. Government Printing Office, 1938. 12. Blishen, B. R., "The Construction and Use of an Occupational Scale™, Canadian Journal of Economics and P o l i t i c a l Science, 24, 1958, 519-31. the average income and average number of years of schooling in each. However, the range of occupations was arbitrarily divided into seven classes, and the sizes of the class intervals, in terms of range of scores, were unequal. This was done not because seven classes could necessarily be identified in Canada, but so that occupations which Blishen considered to have different prestige levels would not be included in the same class. He envisaged two uses to which such a scale could be put including a comparison of the class distribution in various areas of Canadian society, and secondly to compare the actual behaviour of people in the different groups. The method of using average income and years of schooling to derive an occupation score results in one major weakness. An occupation such as farmer, for example, can mean anything from a large and wealthy dairy! farmer to a poor, subsistence farmer in some outlying area of bush. Therefore, the finer the occupation breakdown, the more accurate w i l l be the resulting social class index. This can be obtained by using occupation, not as a single variable, but in conjunction with other socio-economic variables. It is this multivariate approach which is discussed in the next section. 3. Multivariate and multi-dimensional techniques Warner's Index of Status Characteristics which was discussed earlier, was constructed using a multivariate approach. However, the interrelationship between variables chosen meant that they could be combined and reduced to the one standardised ccore (a uni-dimensional scale). Scores were then grouped arbitrarily to give social classes. In more recent studies, on the other hand, i t has been shown that social class is only one dimension of social behaviour. An analysis of a much larger number of socio-economic and demographic variables, especially the extensive coverage given by the Census, has shown that other dimensions of behaviour can be revealed which are largely independent of social class. These include, for example, family status (or l i f e cycle) variables, such as household size and number of children, and the ethnic composition of the household. The next sections w i l l look at the development of a multi-dimensional approach and the methods used to classify data. (a) Social Area Analysis The impetus behind a multi-dimensional approach to social behaviour has come with the development of social area analysis and the improvements in computer technology which have allowed the speedy processing of enormous amounts of urban data. Social area analysis was developed f i r s t by Shevky and his associates as a means of examining and 13 classifying urban social structure. For the basic unit of analysis Shevky chose the census tract, maintaining that the tracts were intended to be as socially homogeneous as possible. It is more feasible, however, that the Bureau of the Census chose tracts as a convenient basis for collecting data. Moreover, even i f the tract was socially homogeneous in the f i r s t place, this is often lost by the time of the next census. Such loss results both from the physical movement of population and change affecting one part of the static population more than another. 13. Shevky, E. and Williams, M., The Social Areas of Los Angeles, Berkeley, University of California Press, 1949; Shevky, E. and Bel l , W., Social Area Analysis, Stanford, Stanford Univ. Press, 1955; and Bell, W., "Social Areas : Typology of Urban Neighbourhood in Sussmann, M. B., Community Structure and Analysis, New York, Thomas Cromwell, 1959, 61-92. 58 Shevky and B e l l have received considerable c r i t i c i s m , e s p e c i a l l y with respect to the a r e a l unit chosen and the lack of a t h e o r e t i c a l basis 14 as to why r e s i d e n t i a l areas i n c i t i e s d i f f e r from one another. But more pertinent to the discussion here are the concepts underlying the multi-dimensionality of the v a r i a b l e s . Shevky and B e l l hypothesised that s o c i a l areas can be examined using three broad dimensions, representing d i s t i n c t processes changing the character of modern society.''""' These dimensions were termed s o c i a l rank (socio-economic v a r i a b l e s ) , urbanisation (family variables) and segregation (ethnic composition), ( i ) S o c i a l Rank Technological change i s seeing a changing d i s t r i b u t i o n of s k i l l s where manual work i s becoming less important compared with white c o l l a r jobs, p a r t i c u l a r l y those needing professional t r a i n i n g and administrative and technical a b i l i t y . In general, the more a p a r t i c u l a r s k i l l i s demanded and/or the more d i f f i c u l t i t i s to a t t a i n , the more prestige that the job w i l l carry and the more economic and s o c i a l gain that can be expected. To a t t a i n a p a r t i c u l a r s k i l l w i l l require a c e r t a i n background ( i . e . education and family upbringing) and once attained w i l l allow a c e r t a i n kind of l i f e s t y l e . The variables suggested by Shevky and B e l l to denote s o c i a l rank are occupation, years of schooling, value or rent of home, persons per room and c h a r a c t e r i s t i c s concerning heating and 14. Hawley, A. H. and Duncan, 0. D., " S o c i a l Area Analysis : A C r i t i c a l A p praisal", Land Economics, 33, 1957, 337-45. Using ethnic and occupation data the authors show the d i f f i c u l t y of obtaining homogeneous units and conclude that characterising i n d i v i d u a l s on the basis of t h e i r r e s i d e n t i a l locations alone i s subject to a large e r r o r . They maintain also that attempts to explain the t h e o r e t i c a l constructs of s o c i a l areas are no more than a p o s t e r i o r i j u s t i f i c a t i o n f o r the indices that have been chosen beforehand. 15. Shevky, E. and B e l l , W., op. c i t . , 3-19. plumbing in the home. Thus, the social rank dimension is related to 16 the underlying theories of social class seen above, ( i i ) Urbanisation The second process affecting modern society Shevky and Bell 17 termed the changing structure of production activity.' The t i t l e of urbanisation demonstrated the fact that the primary means of production were becoming relatively and often absolutely less important in the industrially-advanced societies and economic activity was increasingly concentrated in the larger centres. Moreover as societies advanced economically there was f i r s t of a l l a decline in the death rate, followed later by a f a l l in the birth rate. This reflected improved methods of food supply, birth control and combatting disease. It was hypothesised that these broad economic and demographic changes influenced the size and nature of the family. It was no longer necessary to have a large number of children. Rising living standards meant that children were less needed to supplement family income and support parents who were no longer employed. Meanwhile, a growth in light industrial, service and c l e r i c a l occupations resulted in a demand for labour which could only be satisfied by encouraging housewives to work. For the most part the size of family has become a matter of preference, and is often inversely related to the preference of the housewife to work and contribute materially 16. Since social class refers to groups in a hierarchical position according to status or prestige, the terms social class and social rank appear synonymous. However, because of differences in the way factor scores are grouped and named the term social class w i l l be used in subsequent analyses in this thesis (see Chapter V, footnote no. 6). 17. Shevky, E. and Bell, W., op. c i t . , 10. to the well-being of the family. The variables which represent this dimension include age and sex composition, persons in the household and percentage of adult women employed in the labour force. Because of the nature of the variables chosen, the term family status is often used instead of urbanisation. ( i i i ) Segregation The third independent dimension relates to the changing 18 composition of the population. As societies advance economically and livi n g standards increase, mobility increases. Shevky and Bell note three different concomitants of increased mobility. F i r s t l y , there is redistribution of population in space, resulting from internal migration from areas of declining opportunities to areas that are expanding. Secondly, not a l l age and sex groups migrate to the same extent. In areas where there has been a high in or out migration, the composition of the population w i l l vary most from the national pattern. Finally, in a country like the United States which has depended heavily on immigration from abroad there has been a tendency for immigrants to associate residentially with groups of similar origin, especially those groups who have arrived recently, do not speak English and/or are non-White. However, Shevky and Bell's index of segregation only considered the last aspect of increased mobility and took account of variables such as race, birthplace and citizenship. (iv) The Analysis Shevky and his associates had d i f f i c u l t y trying to obtain data 18. Ibid., 14. to f i t the three underlying dimensions above. This was because they had to rely on census-tract material where many variables were not consistent 19 from one census date to the next. Few variables were in fact chosen for the different dimensions. The index of social rank included a combined standardised score of only occupation ratio and education. The index of urbanisation combined a f e r t i l i t y ratio (number of children under 5 years per 1,000 females age 15-40) with women in the labour force and the proportion of single family dwelling units. The standardised scores of these indices ranged from 0 to 100 and were evenly divided into four groups. The index of segregation considered the proportion of subordinate people in a census tract. Subordinate represented Negroes, Orientals, East and South Europeans, Latin Americans and French Canadians. There was a twofold division here, high and low, depending on whether the proportion of subordinate people in the tract was above or below the proportion in the urban area as a whole. Although criticism has been raised as to the contribution of social area analysis to urban social theory, l i t t l e argument has been raised against the basic multi-dimensional classificatory approach. Further support for social area analysts has come as high-speed computers have allowed the fast execution of multivariate s t a t i s t i c a l techniques. Instead of relying on theory or substantive work to select a limited number of variables for a multi-dimensional approach to social behaviour, i t i s possible, using a technique such as factor analysis for example, to work with a much larger number of census characteristics and isolate 19. Ibid., 54-58. 62 those dimensions which explain as much of the variation in the data 20 as possible. Simmonds, for example in his r e t a i l studies in Toronto used 75 variables to represent patterns of socio-economic and demographic variation in the city; and in applying a .factor analysis seven factors, 21 or dimensions, were identified. The technique of factor analysis is explained in more detail with respect to the present author's work below. But here i t can be stated briefly that factor analysis is a technique that reduces a large number of variables to a smaller number of distinct factors or dimensions. In social area studies principal components factor analysis has invariably revealed at least three underlying dimensions which account for most of the variance in the data. These are identified as social rank (e.g. education, income and occupation), urbanisation or family status (persons per family and children under 15) and ethnicity (percentage of the population of a certain country of origin). The s t a t i s t i c a l technique, therefore, lends considerable support to the conceptual frameworks set up earlier by social area analysts such as 20. The literature on recent social area (factorial ecological) studies is extensive. Some of the more important works include Sweeter, F. L., "Factorial Ecology : Helsinki, 1960", Demography, 2, 1965, 372-85; Berry, B. J. L. and Rees, P.. H., "The Factorial Ecology of Calcutta", American Journal of  Sociology, 74, 1969, 445-91; Murdie, R. A., Factorial Ecology  of Metropolitan Toronto, 1951-61, Research Paper no. 116, Chicago, University of Chicago Department of Geography, 1969; Robson, B. T., Urban Analysis : a study of city structure  with special reference to Sunderland, Cambridge, University of Cambridge Press, 1969; and Berry, B. J. L. (ed.), "Comparative Factorial Ecology", Economic Geography, 47, 1971. 21. Simmonds, J. W., Toronto's Changing Retail Complex : A Study in Growth and Blight, Research Paper no. 104, Chicago, University of Chicago Department of Geography, 1966, 50-51, 113-21. Shevky and B e l l . The abi l i t y to use multivariate techniques then raises the question of which technique is to be used, or is best suited, to group people according to social classes, family status groups and so on. Three techniques are discussed here. (b) Cluster Analysis The f i r s t technique, cluster analysis, is concerned with grouping a number of cases (observations), each having certain objects (variables) into subsets of cases with similar objects. Each object w i l l be characterised by an attribute score. Furthermore, the researcher has no 22 prior knowledge of how the cases are likely to group. The purpose of the cluster analysis is to assign cases to groups so that there w i l l be maximum likeness within groups and difference between groups. This can be done f i r s t l y by prespecifying the number of groups desired, and since the total number of cases is known, the number of cases that each group w i l l contain. Euclidian distances are calculated between each case in 'n' object space. The pair with the smallest distance is chosen as the node of the f i r s t cluster (centroid), and the average is calculated. The next case with the closest f i t to this average is computed and added. This is continued unt i l the prespecified number of cases is grouped. Then the procedure is repeated with the ungrouped cases that remain until the second and subsequent groups are formed. " 22. Green, P. E., Frank R. E. and Robinson, P. G., "Cluster Analysis in Text Market Selection", Management Science, 13, 1967, Series B, 387-99; Morrison, D. G., "Measurement Problems in Cluster Analysis", Management Science, 13, 1967, Series B, 775-80; and Frank, R. E. and Green, P. E., "Numerical Taxonomy in Marketing Analysis : A Review Article", Journal of Marketing  Research, 5, 1968, 83-98. 64 23 A second method is to group the cases individually. The shortest distance between two cases in the n by n matrix (where n is the total number of cases) is computed and grouped f i r s t , and with the matrix reduced to n-1 by n-1 the operation is repeated. Computation and grouping can proceed until a l l n cases have been grouped together. However, i t is desirable to terminate the grouping at some stage where the number of groups has become small yet the loss of detail resulting from grouping has not become too great. A disadvantage in using cluster analysis is that i t is a Euclidian (or geometric), rather than a s t a t i s t i c a l , design. There is no indication, for example, which variables or objects are the most important in explaining the overall variance. There is no analysis of variance to check whether there is a significant within-group similarity and between-group difference. Furthermore, the technique does not take account of correlations between the variables. We cluster in 'n' object space with orthogonal axes. Yet there is likely to be strong correlations between many of the variables. Spence overcame this problem by f i r s t reducing the variables to independent constructs, or underlying dimensions, using principal components analysis. However, this technique in i t s e l f i s a grouping procedure. Since i t is st a t i s t i c a l l y more sound, then i t must be asked, why not use i t in the f i r s t place. The validity of the technique of cluster analysis as a whole has been questioned. It has been suggested that despite a large body of 23. Spence, N. A., "A Multifactor Uniform Regionalisation of British Counties on the Basis of Employment Data for 1961", Regional Studies, 2, 1968, 87-104. literature on the technique in the l i f e and behavioural sciences, cluster analysis has proved very l i t t l e , and i t may in fact be a mistake to promote i t in other disciplines (in this case marketing) when there 24 are other better techniques. Earlier, Greer, Frank and Robinson using fourteen socio-economic variables had reduced 88 cities to 18 clusters where the clustered c i t i e s would have similarities in test marketing situations. Shuchman notes that the clusters link such cit i e s as New Orleans and Minneapolis, Atlanta and Seattle and Tacoma and E l Paso, and goes on to say 'as one who has spent more than a decade studying consumer responses to marketing offers and who is a l l too cognizant of the great regional differences in consumer buying attitudes and habits which exist in the United States, I can scarcely believe that the authors are serious in suggesting that they have produced clusters which have any meaning in the context of test market selection. Criticism i s levelled at the editors of the Journal for not demanding that the authors carry out an analysis of variance test on their clustering. 1 (c) Discriminant Analysis A more statistically-oriented technique that has been used in 25 measurement and grouping procedures is discriminant analysis. In design i t is similar to regression analysis. There is a relationship 24. Shuchman, A., in a letter to the Editor, Management Science, 13 1967, Series B, 688-91. This letter is a reply to the a r t i c l e by Green, P. E., Frank, R. E. and Robinson, P. G., op c i t . 25. Cooley, W. W. & Lohnes, P. R., Multivariate Procedures for the Behavioural Sciences, New York, J. Wiley, 1962, 116-50; King, W. R. "Structural Analysis and Descriptive Discriminant Functions", Journal of Advertising Research, 7, 1967, 39-43; King, L. J., "Discriminant Analysis : a Review of Recent Theoretical Contributions and Applications", Economic  Geography, 66, 1970, 367-78. whether causal or not between independent and dependent variables, and i t is also a predictive tool. However, the basic difference between the two techniques depends on the questions that are being asked and the unit of analysis (census tract, individual, etc.) that is being used. In regression analysis i t can be asked what is the probability that an individual X^ w i l l belong to Group I, whereas a discriminant test w i l l ascertain the probability that an individual X^ w i l l belong to Group I as opposed to Group II or Group III. Also, given two well-defined groups, discriminant analysis can be used to test whether the between-group differences, based on a number of independent variables, are significant or not. In a regression analysis the Y (or dependent variable) is a quantitative measure of some form, but in discriminant analysis this measure is qualitative, i.e. whatever constitutes Group I .... Group N. With these notions of probability of membership and significance of the group, one has to establish beforehand very concisely what the groups are and, conceptually, the boundaries between groups. So far the technique has been l i t t l e used outside the l i f e and behavioural sciences. However, in a study of shopping patterns in the San Francisco-Bay area, discriminant analysis was used to establish the probability that an individual living a certain distance from two shopping centers (Downtown Berkely and El Cerrito Plaza) and having certain socio-economic and demographic attributes would shop at one or 26 other of the two places. Conceptually, the two groups are clearly 26. Bucklin, L. P., Shopping Patterns in an Urban Area, (Research Program in Marketing, Graduate School of Business Administration, University of California, Berkeley), Berkeley, Institute of Business and Economic Research, 1967, 70-118. defined and have distinct 'boundaries'. Given a sample of customers with certain attributes who shop at one or other of the centres, i t is possible putting in distance and attribute variables to establish spatial and socio-economic boundaries. With new customer data discriminant scores can be established, indicating the probabilities of customers going to the different centres. In the present research discriminant analysis is not a feasible technique, at least not at this stage. Although we want to assign people to groups (e.g. social classes), no groups have been prespecified. If somehow we could have prespecified groups, i.e. by using Warner's classificatory system, we would face a very complex computational problem. In the work of Bucklin the groups had very fixed 'boundaries', whereas the boundaries between social classes or family status groups cannot be easily defined beforehand. To the author's knowledge discriminant analysis has not been used where groups have not been established beforehand. If one did establish social classes, for example, beforehand, there would be no point in using discriminant analysis : the work would already be done. (d) Factor Analysis It was mentioned briefly above factor analysis has been the multivariate s t a t i s t i c a l technique which has reinforced the multi-dimensional approach of social area analysts. The technique i t s e l f was 27 developed in the early 1900's by psychologists, but i t was not until 27. Royce, J. R., "The Development of Factor Analysis", Journal of  General Psychology, 58, 1958, 139-64. the 1950's that i t was used widely by other social scientists. Its basic role is to handle and reduce to more meaningful terms an unwieldy 28 amount of data by seeking distinct patterns or relationships. From the point of view of scientific method factor analysis can be used in a large number of inductive and deductive problems, such as exploring areas with l i t t l e a priori theory, establishing relationships, defining underlying dimensions, classifying and ordering data. The approach here is essentially deductive. It has already been hypothesised, based on established theory and previous substantive research, that certain underlying dimensions representing social behaviour w i l l be revealed. Factor analysis is used here to reveal these dimensions and to rank and group individuals on them. This can be done in two ways. F i r s t l y , individuals can be grouped according to some similarity of profile, for example, low on income, low on education, high on household numbers, etc. (Q - analysis). Secondly, patterns of intercorrelated variables can be distinguished. For example, individuals with a high education score tend to have high incomes and high status jobs. This pattern of variation, or dimensional approach, i s called R - analysis and is used more frequently. In seeking underlying dimensions and ranking individuals on them factor analysis follows a series of procedures which w i l l be 28. Texts on factor analysis vary considerably in their mathematical complexity. Two papers of value in understanding the conceptual elements are Cattell, R. B., "Factor Analysis : An Introduction to Essentials", Biometrics, 21, 1965, 190-215, 405-35; and Rummel, R. J., "Understanding Factor Analysis", Journal of Conflict Resolution, 11, 1967, 444-80. More mathematical texts include Cooley, W. W. and Lohnes, P. R., op. c i t . , 151-85; and Harman, H., Modern Factor Analysis, Chicago, University • of Chicago Press, 1960. b r i e f l y discussed below. ( i ) Correlation Matrix A c o r r e l a t i o n c o e f f i c i e n t indicates the r e l a t i o n s h i p between any two var i a b l e s , and can have a twofold importance. F i r s t of a l l i t w i l l indicate the usefulness of the d i f f e r e n t variables that have been incorporated into the factor a n a l y t i c model. Normally variables are chosen where there i s a p r i o r i reason to believe that they are related to some degree with some of the other variables chosen. However, factor analysis i s used as a search technique where previous theory and research i s weak. Therefore, the c o r r e l a t i o n matrix w i l l indicate which variables are poorly associated with the others and which can perhaps be eliminated from further a n a l y s i s . In addition to the degree of association between variables i t i s possible to test whether each c o e f f i c i e n t i s s t a t i s t i c a l l y s i g n i f i c a n t or not at various confidence l e v e l s . Variables can be rejected i f they do not r e l a t e s i g n i f i c a n t l y to other variables i n the matrix. Secondly, the c o r r e l a t i o n matrix can show the j u s t i f i c a t i o n for continuing with the factor analysis. The analysis i s designed to uncover patterns of highly i n t e r r e l a t e d v a r i a b l e s . Thus, as Rummel states, ' i f v a r i a b l e A i s highly correlated with both B and C, and i f B and C are highly correlated with each other, then A, B, and C form a c o r r e l a t i o n c l u s t e r . If A, B and C are not correlated with other v a r i a b l e s , then i 2 9 they form an independent pattern that factor analysis w i l l delineate. The c o r r e l a t i o n matrix provides the key for continuing with the factor a n a l y s i s . 29. Rummel, op. c i t . , 461. ( i i ) Unrotated Factor Matrix The next function of the factor analysis is to uncover dimensions or factors which account for decreasing amounts of variation in the data. The number of factors identified represents the number of independent or uncorrelated patterns of relationships between the variables. To uncover the f i r s t dimension a factor is f i t t e d to the data to account for the greatest decrease in the total variance, and each succeeding factor is f i t t e d to reduce best the remaining variance. Each factor involves a l l the variables, and the degree to which a variable is involved in an unrotated factor pattern is indicated by a factor loading. Loadings on a factor can be interpreted in much the same way as correlation coefficients. The loading squared X 100 equals the percentage variation that a variable has in common with the unrotated factor. In comparing the factor loadings for a l l variables and factors i t is possible to see which variables are most involved in a particular pattern. A further procedure in the unrotated factor matrix shows the communality of each variable. This gives the proportion of a variable's total variation which is accounted for by the three factors together and is calculated by squaring and then summing the variable's loadings on each factor. By subtracting the communality from 1.0 a measure of uniqueness can be obtained. This shows the degree to which a variable is unrelated to the others, and therefore the degree to which data on that variable cannot be predicted from data on the other variables. For each factor the sum of the squared loadings is termed the eigenvalue. This in turn shows the degree of variation accounted for by each pattern, with most variation found in the f i r s t factor and least variation in the last factor to be identified. The unrotated factor matrix, however, does not identify independent clusters of interrelated variables. Whilst reducing the variance the factors may have been projected between the clustering of variables, i f any exist. ( i i i ) Rotated Factor Matrix In order to get a better f i t between the factors and the clusters of interrelated variables, a further step is taken whereby the factor axes are rotated to a simple structure solution. This implies that each factor has been rotated until each represents a distinct cluster of interrelated variables. There are two basic ways in which axes may be rotated. F i r s t l y , there are the variety of oblique rotations, these have the advantage of uncovering clusters irrespective of their correlation. However, they are conceptually more d i f f i c u l t to use and have not been programmed for most computers. The second, and more usual, rotation is orthogonal (right angle). Its one weakness is that i t must be assumed that the factor patterns are uncorrelated with one another, whereas invariably some degree of association exists between a l l the variables and also between a l l variables and factor axes. The assumption of non-correlation results because orthogonal axes are a l l fixed to the origin, and each axis is at right angles to every other axis. The axes are rotated as a group unt i l they are maximally aligned with the separate clusters of variables. The result is that instead of moderately high or low loadings, as in the unrotated factor matrix, there is a tendency for each factor to show much higher and lower loadings, showing a much better picture of underlying interrelationships. In social area studies i t is at this stage that the three or more factors relating to social rank, family status and ethnicity can be identified and named,"^although the more correlated the separate clusters become, the less the orthogonal rotation can discriminate between them, (iv) Factor Scores The f i n a l step in the computation is the establishment of a factor score for each of the observations on each of the factors identified. Each variable is given a loading (weight) in proportion to i t s involvement in a factor pattern, the more involved the variable, the higher the loading. To derive a factor score for any of the cases on a particular pattern the case's data on each variable is multiplied by the pre-established loading. The sum of the loading X data products for a l l variables gives the factor score. The value of obtaining such a score is that i t is then possible to rank or order the various cases on each underlying dimension. However, i f one is going to work with the results of such an analysis, i t becomes necessary to group the ranked data on each dimension into a manageable number of classes. The number of classes chosen should not merely reflect convenience or the nature of the particular problem under discussion but should attempt to correspond f a i r l y closely with generalised, but distinguishable, groups that previous substantive research (e.g. Warner's "Yankee City" studies) has identified. James Carman, who has used factor analysis to distinguish social classes, 30. For example, Murdie, R., op. c i t . , 76-115. 73 31 has discussed some of the issues involved in grouping factor scores. If classes are said to exist, then a frequency distribution of the factor scores on that dimension should indicate a number of clusters of factor scores with the boundaries between classes drawn somewhat arbitrarily between the clusters. Four reasons are suggested for why clusters may not appear. F i r s t l y , theory could be suspect and subcultures do not exist within a society. However, in view of the substantive research that has been carried out, this can be dismissed. Secondly, factor scores are not sufficiently discriminating. Thirdly, individualism is such that people do not f i t well into classes. Finally, social mobility may result in a blurring of class boundaries. These problems concerning the clustering of factor scores are discussed in more detail in Chapter V with respect to the present author's socio-economic and demographic data on consumers in Metropolitan Vancouver. Summary This chapter has looked at a number of problems concerning the methodology to be adopted for measuring social class and family status. F i r s t l y , social class, i t s e l f , i s measured in terms of real or proxy variables, but proxy variables are used here since they are 31. Carman, J. M. , The Application of Social Class in Market Segmentation, (Research Program in Marketing, Graduate School of Business Administration, Univ. of California, Berkeley), Berkeley, Institute of Business and Economic Research, 1965, 48-59. The variables that were used in the factor analysis included occupation, education, income, property value or rent, race, stage in the l i f e cycle and family size. Three dimensions were identified which were termed the cultural factor, a racial discrimination factor and a l i f e cycle factor. However, rather than a multi-dimensional approach, Carman finished up with a single dimensional approach, for only the scores on the cultural factor were grouped and used in the ensuing analysis of consumer buying habits. 74 more e a s i l y obtainable, yet depending on the variables used s t i l l convey the e s s e n t i a l s o c i a l c l a s s meaning. The second part of the chapter discussed the various methodologies involved i n using proxy v a r i a b l e s , including Warner 1s multiple factor design, the use of occupation alone to denote s o c i a l c l a s s and the development of multivariate s t a t i s t i c a l techniques. The benefit of multivariate a n a l y s i s , and es p e c i a l l y factor analysis i s that i t i s possible to reveal more than one independent dimension, or set of i n t e r r e l a t e d v a r i a b l e s ; these include besides s o c i a l c l a s s , family status (or l i f e cycle) and ethnic v a r i a b l e s . I t i s t h i s approach, using factor analysis on the socio-economic and demographic data f o r Metropolitan Vancouver, which i s discussed i n the next chapter. Chapter V Metropolitan Vancouver : Case Study Introduction This chapter sets out by briefly looking at the overall research design. This is followed by a discussion of the way in which the sample was chosen and the survey method following this and the nature of each of the socio-economic, demographic and shopping variables asked for in the survey questionnaire. The f i n a l part of the chapter is specifically concerned with the use of factor analysis, discussed in Chapter IV above, for establishing social classes and family status groups from the socio-economic and demographic data. Research Design The actual analysis undertaken developed from a series of assumptions, or working hypotheses, which were outlined in Chapter I and are expanded upon in Chapters VI and VII when two examples of consumer spatial behaviour are examined. These hypotheses are formulated in order to direct research into areas where further study i s needed; however, no study i s done in isolation, and i t is possible to justify such hypotheses on the basis of previous research. The key to successful hypothesis testing is the nature of the data available, and since the present study i s concerned with the relationship between social class, family status and some aspects of consumer spatial behaviour i t is necessary to interview the consumer. This type of research is more time consuming and often more costly compared to the approach where published s t a t i s t i c a l sources give the location and nature of r e t a i l activities and consumer behaviour is then assumed> and the result is that data are not always so readily available. This author is fortunate in being able to use data obtained from a random sample of consumers interviewed in their homes over the Metropolitan 1 Vancouver area, and the strength of the data in carrying out the present research is discussed in the next section. The next part of the analysis examines the socio-economic and demographic information about consumers that was obtained from the interviews. It was stated in Chapter II above that research on variations in consumer spatial behaviour has tended to stress the relationship between these variations and differences in individual consumer attributes. Many attributes are interrelated, and the more generalised measures that result, such as social class, have identity in terms of differentiating very many types of behaviour, and may be more meaningful than using a single attribute. Using factor analysis, the socio-economic and demographic data are examined in order to uncover groups of interrelated attributes. The consumers themselves are then grouped on each set of interrelated attributes, and i t is then seen whether there are significant between-group differences in the ways consumers behave spatially. 1. See Figure 5.1. The definition of Metropolitan Vancouver in this dissertation conforms with the coverage of the volume, Census of Canada 1961, Vancouver: Population and Housing Characteristics  by Census Tract, Bulletin CT-22, Ottawa, Dominion Bureau of Statistics, 1963, and includes the Cities of Vancouver, North Vancouver, New Westminster, Port Moody, Port Coquitlam and White Rock, the District Municipalities of West Vancouver, North Vancouver, Burnaby, Coquitlam, Surrey, Delta and Richmond and the University Endowment Area. 78 Two examples of consumer spatial behaviour are subsequently studied in Chapters VI and VII. The f i r s t examines the distance that consumers travel to purchase the five goods asked for in the interview (groceries, dress, furniture, appliances and footwear) and relates this to the social class and family status position of the consumer. The second example looks at the department store preferences of consumers for the four shopping goods and analyses whether there is a relationship between social class and family status position and the choice of department-store firm and i t s location with respect to either downtown Vancouver or a suburban shopping area. The Data 1. Sample Survey The sample survey of consumers in Metropolitan Vancouver was undertaken in two parts, the City of Vancouver (except for the West End) 2 in March 1964 and the remainder of the Metropolitan area in June, 1964. Within the City of Vancouver every block was numbered and every f i f t i e t h block was selected for sampling. The interviewing plan involved a systematic sample with every seventh household on the block being called upon. A random start was made on each block, with the f i r s t household to be interviewed being from the f i r s t to the seventh household counting clockwise from the north-west corner of the block. After the f i r s t household was randomly chosen every seventh household was interviewed. 2. This survey was undertaken as part of a research project on the changing emphasis of r e t a i l functions in the Vancouver core which was directed by Dr. Walter G. Hardwick, Department of Geography, University of British Columbia. The project was financed by the Canadian Council on Urban and Regional Research. Interviewing of consumers was carried out by six senior undergraduates in the Faculty of Commerce and Business Administration. A check on interviewer r e l i a b i l i t y can be made by interviewing the same respondent again but with a different interviewer; this, however, was not done. If a block chosen was a school, park, playground, institution or entirely commercial enterprise, then the next block was substituted for the purpose of the sample. No interviews were made in any form of institution (e.g. a private hospital). If an apartment block or a house divided into suites was encountered, each suite was treated as a separate household. Every seventh suite was interviewed, beginning on the lowest floor and counting clock-wise on each floor. In each case questions were put to an adult female member of the household, since i t was considered by the organisers of the questionnaire that women most likely did more household shopping than men and could more easily remember where and when purchases 3 were made. If there was no reply from any household contacted, the interviewer made one r e c a l l . In a number of instances substitution of the next household was made. This occurred i f there was no adult female in the household to interview, i f there was no one at home after the second c a l l and i f a household member refused to answer the questionnaire. Outside the City of Vancouver a different interviewing procedure was followed. Whilst the City has for the most part a uniform grid-iron street system, this i s less evident in other parts of the Metropolitan area where the block system is disrupted by physical features and non-residential or vacant land. Here a l l census tracts and enumeration areas were listed in numerical order. Every eighth enumeration area was selected, and within each one every fortieth house, clockwise by blook, was chosen. A pilot survey indicated that sampling by interview would be 4 more beneficial than a mailed questionnaire. The former i s more time 3, Details of the questionnaire can be seen in Appendix A below. 4. Comments here reflect Moser, C. A. Survey Methods in Social Investigation, London, Heinemann, 1958, 175-209. 80 consuming and expensive, especially i f the interviewers have to be paid. Invariably, the size of the sample here is smaller than i f one was conducting a mailed questionnaire. The present data were based on 681 household interviews. There were 303 interviews in the City of Vancouver, out of a total of 118,405 households according to the census of 1961 (slightly over one quarter of one per cent); outside the City 378 interviews were conducted, out of a total of 110,193 households (slightly over one third of one per cent). In an interview i t was found that a reluctant interviewee could be persuaded to answer the questionnaire, whereas a mailed questionnaire might be put aside and forgotten, even after follow-up reminders had been sent out. Providing the interviewers have been properly instructed in interviewing techniques and background relating to the questions, they are in a valuable position to help respondents in a number of ways. In a mailed questionnaire a question that is not understood or is too vague may result in the respondent giving no answer at a l l or an inaccurate one, whereas i f an interviewer is present the d i f f i c u l t y can more readily be eliminated. This is reflected in the much higher percentage returns that are usually obtainable in interviews as opposed to mailed questionnaires. In the present survey only four out of 685 questionnaires were incomplete or obviously inaccurate. Certain questions may be f a i r l y sensitive, e.g. asking a person's income or the ethnic background of the household, and i f l e f t to a mailed questionnaire such questions may result in a non-response. In an interview problems such as these can often be overcome by stressing the confidential nature of the questionnaire and interesting the respondent in the purpose of the study and perhaps i t s academic and/or practical results. 81 There i s also the problem, however, that the interviewer, unwittingly, w i l l introduce biases into a survey's results. This can often be done when a respondent f a i l s to understand a question, and rather than the interviewer merely giving a point of cl a r i f i c a t i o n , an opinion or other remark may be forwarded which results in a respondent giving an incorrect or biased answer. Biases of this nature can usually be avoided i f the information sought after is factual rather than opinionated. The present survey in fact did not seek any opinions. 2. Socio-economic and demographic variables The socio-economic and demographic variables asked for in the questionnaire are ones that have been used in previous research to denote differences in social class and family status characteristics of individuals. The number of variables included i s small in comparison, for example, to the number which are available in the census and are incorporated in social area analysis. This fact need not seem a disadvantage : a better response is likely on the part of the respondent when asked as few questions as possible, especially i f he or she is not paid for the time and trouble and the questions, when not backed by any legal authority, may be regarded as an invasion of one's privacy. The variables used in this dissertation include:-(a) Size of household. A household constitutes a person or group of persons occupying a living area which i s structurally separated from other households by a private entrance. A person is a member of the household i f he normally sleeps there. The number was coded here from one through seven. More than seven members in a household was coded as seven. (b) Number of children age 10 - 19. 82 (c) Number of c h i l d r e n age under 10. A question concerning the number of c h i l d r e n i n a household has always been considered important i n shopping surveys. Children need c e r t a i n products which often do not concern adults. Moreover, of p a r t i c u l a r relevance i n the present d i s s e r t a t i o n i s the f a c t that children contribute l i t t l e or nothing to the family income. Thus the s i z e of family frequently influences the decision whether or not various products can be afforded. Even a f t e r a decision to purchase has been made c h i l d r e n often influence where the purchase i s made. A large family e s p e c i a l l y with young chi l d r e n w i l l f i n d the shopping process more of an inconvenience than a smaller family e s p e c i a l l y where the c h i l d r e n are older and can be l e f t at home. The larger and younger family w i l l tend to minimise those parts of the shopping process which involve time and money such as t r a v e l and parking costs. The two categories of children's age ( i n addition to the size of household) were chosen so that shopping behaviour could be d i f f e r e n t i a t e d according to the type of family. A f t e r the age of nineteen a c h i l d was recorded as an adult member of the household, since at t h i s time he would have graduated from high school and most l i k e l y c ontributing to the family income. The number of children i n each category was coded from zero through seven. (d) Occupation of household head. Interviewers included a b r i e f two or three word de s c r i p t i o n with each occupation. This was done to avoid any ambiguity and to help c l a s s i f y occupations f o r further a n a l y s i s . For example, secretary by i t s e l f was considered inadequate. I t could r e f e r to an executive secretary of a large f i r m or a c l e r i c a l secretary doing typing and f i l i n g i n a small 83 business. These two occupations refer to widely varying levels of prestige, education needed and income. Care had to be taken in recording the occupation of the household head since in most cases the wife was interviewed and details were asked of the husband's occupation which was not always too clear. Where the household did not constitute a family, for example two unmarried g i r l s sharing an apartment, the person answering the questionnaire acted as a substitute for the household head. If the household head was retired or unemployed, the occupation of most recent employment was recorded. The Blishen scale (seen in Chapter IV above) was used in order 5 to collapse the various occupations into some quantifiable expression. Some 343 occupations had previously been ranked according to a score based on the average income and average number of years of schooling in each occupation. The scores were then grouped into seven classes. In the present survey the occupations stated were assigned a number from one through seven, with the higher the number implying occupations with lower average income and years of schooling. The problems outlined in Chapter IV of using occupation data based on average income and education were overcome here by including income and education as distinct variables in the analysis. (e) Education of household head. The respondent was asked when the head of the household lef t school, and this was categorised in one of five areas - before grade 8, during high school, graduated from high school, during university and 5. Blishen, B. R., "The Construction and Use of an Occupational Scale , Canadian Journal of Economics and P o l i t i c a l Science, 24, 1958, 519-31. For detailed breakdown see Appendix B. graduated from university. As with occupation, i f the household was a non-family one the answer given applied to the person answering the questionnaire. For the purposes of s t a t i s t i c a l analysis the five categories were coded from one (before grade 8) through five (graduated from university). (f) Household income. Asking people about their income is perhaps the most sensitive question in a survey of this nature. In this questionnaire the problem was partially off-set by not asking for an exact figure but letting the respondent place the household in one of the income categories. These were coded from one through seven, from under $1,000 per annum, $1,000 -$3,000, $3,000 - $5,000, $5,000 - $7,000, $7,000 - $9,000, $9,000 -$11,000 to over $11,000. The wife very often does not know her husband's exact income, and categories such as these are the best approximation. 3. Shopping variables In the f i r s t part of the survey carried out in the City of Vancouver, with the exception of the West End, respondents were asked about one convenience good (groceries) and two shopping goods (women's clothing (dress) and furniture). The response rate was low in the case of dress and furniture because many people had made no purchases within the time period of one year asked for in the questionnaire. When the survey was extended to include the West End of Vancouver and the remainder of the Metropolitan area two more shopping goods were included (appliances and footwear). (a) Groceries. Respondents were asked to refer to a major purchase, including canned and packaged goods, dairy and vegetable produce and meats, and not a casual purchase of a quart of milk or a loaf of bread. Respondents were asked where they last bought groceries, giving the name of the store and a block intersection for the address. It was also noted whether this was the store where groceries were usually purchased, and i f not the name and address of this store was also indicated. In the majority of cases the one store applied to both questions. Where this was not the case the analysis of shopping behaviour, seen in Chapter VI below, used the location of the grocery store normally visited. In this way eccentric occurrences, for example a grocery purchase last made at an unusually large distance from the home whilst perhaps en route from v i s i t i n g friends, could be eliminated from the analysis, (b) Shopping goods. For the four shopping goods, dress, furniture, appliances and footwear, respondents were asked i f they had made a new purchase (in a store, not by telephone) within the last year, and i f so to give the name and address of the store where they had last made such a purchase. An arbitrary time period was introduced because respondents could very easily forget where they had last bought a good. Moreover, this limit would reduce, although not eliminate, the problem of a purchase being made from a previous home. Only i f a purchase was made well outside the Metropolitan area and i t s adjacent municipalities was i t excluded from the analysis. A brief description of the shopping goods is as follows: (i) Dress. This referred to a one or two piece dress only. ( i i ) Furniture. This included chairs, living room, bedroom, kitchen and dining room furniture, but excluded small items such as table lamps and appliances modelled to resemble funiture (e.g. stereo and television sets). ( i i i ) Appliances. Only appliances that were power operated were included here (e.g. washing machines, freezers and toasters). (iv) Footwear. This referred only to normal street footwear. Methodology for grouping socio-economic and demographic variables The f i r s t part of the analysis organised the socio-economic and demographic variables, described above, into a smaller number of groups of interrelated variables. A score for each household for each group of variables was then obtained so that the households themselves could be grouped in a rank order. Factor analysis, which was described in some detail in Chapter IV above, was the technique used here to reduce the variables to a smaller number of underlying factors and obtain a score for each household on each factor. The data used in the factor analysis comprised m x n order matrices, where the number of variables (m) is six and the number of observations (n) is 681, the number in the sample survey of households in Metropolitan Vancouver. The University of British Columbia's principal components1 factor analysis programme with varimax rotation, using an I.B.M. 7044 computer, performed the analysis. The data for each variable were standardised to common units of measurement. Standard scores around a zero mean, having a variance of one, were calculated with the formula: z = x - * s where z = the standard score x = value to be standardised x = mean value of x s = standard deviation 1. Correlation matrix The f i r s t part of the factor analysis established a matrix of correlation coefficients, showing the degree of association between the six variables (size of household, number of children age 10-19, number of "children age under 10, occupation of household head, education of household head and income per annum of adult members of the household). This is seen in Table 5.1 below. One important aspect of the correlation matrix is that i t reveals whether i t is worthwhile to continue with the factor analysis, i.e. whether a pri o r i reasoning and the choice of variables were correct. It can be seen in Table 5.1 that two thirds of the correlation coefficients are s t a t i s t i c a l l y significant, a l l but one of them at the 99 per cent confidence limit, and i t was considered worthwhile to continue the analysis. Two correlation clusters are seen from Table 5.1. A positive relationship exists between the size of household and the number of children age under 10 (0.58). The second cluster includes occupation, education and income. Because of the coding of occupation from one (highest status occupations) through seven (lowest status occupations) the relationship between occupation and education and also between occupation and income are thus inverse ones (-0.55 and -0.47 respectively). A positive relationship of 0.47 was found between education and income. The low correlations that were found in this second cluster may be attributed to the fact that a parametric technique based on continuous variables was used here with discrete variables. In the present analysis one discrete variable i s used instead of a much larger number of continuous variables. For example, an income figure may have been coded four, whereas i f the actual figure, a continuous variable, had been used the figure would have been between $5,000 and $7,000. The fewer discrete variables tend to blur the relationship between two characteristics. In the present survey two observations could have the same discrete variable 88 TABLE 5.1 MATRIX OF CORRELATION COEFFICIENTS Variable 1. 2. 3. 4. 5. 6. 1. Size of household 1.00 2. No. of ch i l d r e n , 10-19 0.68** 1.00 3. No. of c h i l d r e n , under 10 0.58** -0.03 1.00 4.. Occupation of household head -0.07 -0.05 •0.04 1.00 5. Education of household head 0.09* 0.11** • -0.01 -0.55** 1.00 6. Income per annum of adults 0.29** 0.16** 0.20** -0.47** 0.47** 1.00 * S i g n i f i c a n t at ** S i g n i f i c a n t at the .05 the J01 confidence confidence l i m i t l i m i t Source: Sample Survey of Consumers i n Metropolitan Vancouver, 1964. for education and an income variable of four ($5,000 - $7,000) and five ($7,000 - $9,000). Continuous variables for income, on the other hand, if known, could have been $6,900 and $7,150 or $5,001 and $8,999 respectively. The former pair of income variables would have led to an improved correlation coefficient, the latter to a lower correlation coefficient. 2. Unrotated Factor Matrix The next part of the factor analysis was the computation of an unrotated factor matrix. Here, dimensions or factors were uncovered which accounted for decreasing amounts of variation in the data, while the number of factors revealed the number of independent or uncorrelated patterns of relationships between the variables. The principal components analysis was designed so that the f i r s t factor to be extracted accounted for the greatest decrease in total variance from the correlation matrix. From the residual matrix a second factor was extracted, accounting for the second largest decrease in total variance. The analysis continued until the total variance of the correlation matrix was accounted for. Table 5.2 shows that three factors were extracted from the correlation matrix. Each of the variables was involved with a l l three factors, and the relationship between the variables and the factors in the unrotated matrix is indicated by means of factor loadings. These loadings are like correlation coefficients in that they take on values between -1.0 and +1.0. The communalities given for each variable result from the sum of squared factor loadings across every row of the matrix. They show, for example, that taking the three factors together less of the total variance is explained in the case of occupation, education and income compared to the three family variables. The eigen-values at the foot of each column equal the sum of the squared loadings on each factor 90 TABLE 5.2 UNROTATED MATRIX OF FACTOR LOADINGS Factors Communalities Variable I II III 3 Factors 1. Size of household 0.718 0.657 -0.029 0.948 2. No. of ch i l d r e n , 10-19 0.537 0.466 -0.682 0.970 3. No. of ch i l d r e n , under 10 0.434 0.468 0.743 0.959 4. Occupation of household head -0.607 0.581 -0.038 0.707 5. Education of household head 0.620 -0.568 -0.087 0.714 6. Income per annum of adults 0.741 -0.298 0.128 0.655 Variance accounted f o r (%) or Eigenvalue 38.2 27.0 17.4 82.6 and show the proportion of total variance in the data which is accounted for by each factor. Table 5.2 shows that 82.6 per cent of the total variance in the data is accounted for by the three factors taken together, and the factors were extracted in order of decreasing magnitude of variance reduction - 38.2, 27.0 and 17.4 per cent. One of the problems in an unrotated factor matrix is that independent clusters of interrelated variables are usually not identified. Whilst reducing the variance the factors may have been projected between clusterings of variables, i f any exist. Table 5.2 shows that apart from income the variables do not correlate well with only one factor. Moreover the matrix is not unique. It is possible to obtain a very large number of matrices with a series of mathematical transformations. Therefore, i t is not possible at this point to give any name to the factors which have been uncovered. 3. Orthogonally Rotated Factor Matrix Giving some meaningful identity to the factors was accomplished by rotating them orthogonally u n t i l each factor identified a distinct cluster of interrelated variables. Moreover, each of the six variables showed a high correlation with (or high loading on) one factor and a lower correlation with the other two factors. Table 5.3 shows that three distinct clusterings emerged from the data. Occupation, education and income loaded highly on Factor I, with correlations of -0.841, 0.837 and 0.754 respectively, while the other three variables, indicated in parentheses, were close to zero. Size of household and the number of children age 10-19 loaded highly on the second factor (0.770 and 0.980 respectively), while the remaining variable, the number of children under 10, loaded highly on the third factor (0.979). It was noted in the section on the rotated factor matrix in 92 TABLE 5.3 ORTHOGONALLY ROTATED MATRIX OF FACTOR LOADINGS Factors I II III Social Older Younger Communalities Class Family Family (3 Factors) Variable Status Status 1. Size of household (0.093) 0.770 (0.590) 0.948 2. No. of children, 10-19 (0.066) 0.980 (-0.082) 0.970 3. No. of children, under 10 (0.032) (0.024) 0.979 0.958 4. Occupation of household head -0.841 (0.027) (0.007) 0.707 5. Education of household head 0.837 (0.066) (-0.091) 0.714 6. Income per annum of adults 0.754 (0.146) (0.255) 0.655 Variance accounted for, or Eigenvalue (7») 38.2 27.0 17.4 82.6 93 Chapter IV above that the more correlated the separate clusters became, the less the orthogonal rotation could discriminate them. Furthermore, an orthogonal rotation assumes that the factor axes are fixed to an origin and each axis is at right angles to each other. The weakness of the orthogonally rotated matrix seen in Table 5.3 is that the factor patterns do correlate with one another. The variables that load highly on Factor I do not have zero correlations with Factors II and III. In the case of size of household i t loads highly on the second and third factors. This lack of orthogonality is inevitable, since the correlation matrix (Table 5.1) revealed that some measure of association existed between a l l the variables. The correlation between the number of children age 10-19 and the number of children age under 10, for example, was very small and not s t a t i s t i c a l l y significant anyway, but both variables were correlated significantly with the size of household. The two variables on numbers of children loaded on to different factors, and thus the size of household variable could then be expected to load highly on both factors. The orthogonally rotated matrix of factor loadings allows the researcher to give some identity, albeit subjective, to the factors which have been uncovered. In Table 5.3 the f i r s t factor, where occupation, education and income loaded highly on to i t , was termed Social Class.^ The second and third factors identified in Table 5.3 6. Problems associated with the interpretation and measurement of social class are discussed in Chapters III and IV. The term social class is used here (and in Chapters VI and VII below) whereas in social area studies the normal terminology is social rank. Social classes need not imply any ranking or status connotations but merely differences between groups of people. However, the way in which social class is measured invariably makes any mention or ranking unavoidable. Social class is s t i l l used rather than social rank. 94 included demographic variables relating to size of household and numbers of children in two different categories. Because of the way in which the two variables on children's age loaded on to different factors, Factors II and III were termed Older Family Status and Younger Family Status respectively. 4. Factor scores The whole object of the factor analysis i s to establish a ranking for the 681 households in the data base. Three factors were identified in the analysis, and factor scores on each of the factors were obtained by multiplying the rotated factor loading of each of the six variables relating to that factor. For each observation (household) the sum of the loading multiplied by data products for the six variables gave the factor score.^ This operation was repeated for each of the 6. (cont'd). On the basis of factor scores in the present analysis, respondents are grouped into four social classes which can be named and where differences in various forms of behaviour can be identified. On the other hand, social rank may mean as many as eight groups; t i t l e s are invariably not given to the groups, and the large number makes i t d i f f i c u l t to recognise differences between them. 7. The calculation of one of the 681 factor scores on the social class dimension (Factor I) is shown as:-Variable Measurements Loading of Column (2) of a variable on multiplied particular Factor I* by Column household (3) (1) (2) (3) * (4) 1. Size of household 4 0.093 0.372 2. No. of children (10-19) 1 0.066 0.066 3. No. of children (under 10) 1 0.032 0.032 4. Occupation of household head 3 -0.841 -2.523 5. Education of household head 3 . 0.837 2.511 6. Income per annum of adults 5 0.754 3.770 Factor score = 4.228 * See Table 5.3 three factors, and the distribution of factor scores i s presented in Figure 5.2. The scores on each factor were then grouped into a number of classes so that an analysis of variations in consumer spatial behaviour can be made in the following two chapters. One very common method of grouping i s to draw arbitrary boundaries between the factor scores at regular intervals. This i s done in social area analysis, but such an arbitrary decision i s less relevant here. It is possible to identify a smaller number of groups which are distinguishable from one another in their overall make-up and the ways in which they behave. Secondly, there i s no reason to believe that such groups can be defined by regular interval factor scores. The method that was adopted in the present study aimed at identifying a manageable number of clusters of scores on each factor and drawing boundaries between the clusters. However, the frequency distribution of factor scores (Figure 5.2) reveals a large number of small clusters. This results in part from the lack of orthogonality between the various factor patterns : a number of observations have the same occupation, education and income (these are variables which load high on Factor l ) , but they have different scores on Factor I because of differences between the observations in the data on the other three variables. Although the other three variables load very low on this factor, they s t i l l contribute to the overall factor score. This argument i s supported by Carman who found that one reason why factor scores do not cluster well i s that they 8 are not sufficiently discriminating. 8. Carman, J. M., The Application of Social Class in Market Segmentation, Qlesearch Program in Marketing, Graduate School of Business Administration, University of California, Berkeley,) Berkeley, Institute of Business and Economic Research, 1965, 54. 5 E -1 -2 -3 U P P E R / UPPER MIDDLE 4 LOWER MIDDLE u u u w a JE:::::::. UPPER ::: : : LOWER • LOWER LOWER HIGH MEDIUM 0 LOW - 4 HIGH EE:: MEDIUM 96 1 LOW •••••••••• 10 Observations SOCIAL C L A S S (FACTOR I) OLDER FAMILY STATUS YOUNGER FAMILY STATUS (FACTOR H) (FACTOR ID) Figure 5.2 Distribution of factor scores 97 TABLE 5.4 FREQUENCY DISTRIBUTION OF FACTOR SCORE CLASSES Group FACTOR I (SOCIAL CLASS) Factor Score Frequency No. Per cent Upper/Upper Middle Lower Middle Upper Lower Lower Lower +2.40 - +6.50 -0.15 - +2.39 -1.80 - -0.16 -4.00 - -1.81 95 13.95 215 31.57 263 38.61 108 15.85 TOTAL 681 100.00 FACTOR II (OLDER FAMILY STATUS) High Medium Low + 1.30 - +6.80 -1.05 - +1.29 -3.30 - -1.06 124 343 214 18.21 50.37 31.42 TOTAL 681 100.00 FACTOR III (YOUNGER FAMILY STATUS) High Medium Low + 1.40 - +6.30 -0.70 - +1.39 -2.00 - -0.71 127 294 260 18.65 43.17 38.18 TOTAL 681 100.00 The boundaries and number of groups seen i n Figure 5.2 r e f l e c t both the major clusterings of factor scores and the size of the sample. The boundaries were set so that observations with the same data on the high loading variables were not s p l i t between groups. The nature of the groups on each of the factors i s as follows: (a) Factor I (Social Class) The frequency d i s t r i b u t i o n of factor scores (Table 5.4) was divided into four groups: Upper and Upper-Middle. The size of the sample was such that the number of respondents who could have been c l a s s i f i e d as upper class was very small indeed, and i t was therefore decided to combine this group 9 with the upper-middle class. Included here are people having professional, technical or managerial occupations that carry prestige and involve considerable t r a i n i n g i n many instances. A l l have completed Grade 12 and many have completed university. In 1964 a l l incomes were above $7,000 per annum. Lower Middle. This group included the less prestigous white c o l l a r jobs, such as bank clerks and commercial t r a v e l l e r s , and positions of r e s p o n s i b i l i t y i n blue-collar jobs. Few have University t r a i n i n g , the majority f i n i s h i n g their education at Grade 12. The majority of incomes l i e between $3,000 and $7,000 i n 1964. 9. Warner, W. L., S o c i a l Class i n America : the Evaluation of Status, Chicago, S o c i a l Science Research Associates Inc., 1949, 41, recognized i n his work six s o c i a l class groups i n the community. The very small percentage i n the upper class was further subdivided according to an establishment group (upper-upper) and a nouveau-riche group who were aspiring to t h i s f i r s t . p o s i t i o n (lower-upper). 99 Upper-Lower. The majority of people in this group have skilled or semi-skilled manual jobs. Most have some high school education, but few have completed Grade 12. Incomes were mostly between $3,000 and $5,000 per annum. Also included in this group were unskilled workers with better educational background and/or higher incomes. Lower-Lower. This group was made up of semi-skilled and unskilled workers, many of whom were unemployed or only worked for part of the year. The majority finished school at Grade 8, and few earned more than $3,000 per annum in 1964. (b) Factor II (Older Family Status) Table 5.4 shows the breakdown of the grouped scores on the second factor. These groups are: High: Households of five or more people with two or more children age 10-19 and up to two children under the age of 10. Medium: Households of three or four people with no more than two children of any age. Low: Households of up to three people but without any children. (c) Factor III (Younger Family Status) Similarly, Table 5.4 shows the breakdown of the grouped scores on the third factor. These groups are: High: Households of five or more people with two or more children under 10 and up to two children age 10-19. Medium: Households of three to five people with no more than one child age under 10 but two or more children age 10-19. Low: Households of up to three people with either no children or one child age 10-19. 100 S ummary This chapter outlined the basic research design governing the collection of data and i t s analysis, and then discussed the sampling and interviewing techniques adopted and the nature of the socio-economic, demographic and shopping variables asked for in the survey. The second part of the chapter was concerned with the use of factor analysis to uncover a number of interrelationships amongst the socio-economic and demographic attributes and to rank and group the respondents in each one. Three interrelationships, or underlying dimensions, were revealed which were termed social class and older and younger family status. Factor scores were calculated for each respondent on each dimension, and the scores were then ranked and arbitrarily grouped. Underlying the grouping of respondents in this way i s the notion that differences in behaviour are evident between groups. Previous work has suggested that variations in social class and family status are related to a variety of differences in behaviour, and in the next two chapters i t i s seen to what extent these variations can be related to two examples of differences in consumers' shopping patterns. 101 Chapter VI Social Class, Family Status and Consumer Travel Behaviour  Introduction The f i r s t of two chapters, analysing some aspects of consumer behaviour in Metropolitan Vancouver, focuses on the relationship between social class and family status characteristics of the consumer and their travel behaviour for purchasing five goods - groceries, dress, furniture, appliances and footwear. Two aspects of travel behaviour are considered here, the variations in distance that consumers travel between place of residence and the r e t a i l outlet and briefly the different travel frequencies of the various social class groups. In the f i r s t part of the chapter two research hypotheses are stated and developed, relating to the travel behaviour of the social class and family status groups respectively. In the analysis that follows there is a brief discussion of the spatial pattern of retailing and i t s relationship with the distribution of the various social class and family status groups and their aggregate travel behaviour in Metropolitan Vancouver. Aggregate behaviour i s then broken down according to frequency and/or mean distance travelled by the various social class and family status groups. One-way analysis of variance tests are performed to see whether or not the differences between mean distances for the various groups are st a t i s t i c a l l y significant and the i n i t i a l hypotheses are supported. The survey questionnaires are examined 102 further to suggest reasons for the v a r i a t i o n s i n mean distances that are displayed. Research Hypotheses I t i s hypothesised that: 1. the higher the s o c i a l c l a s s p o s i t i o n of the consumer, the  greater the frequency that a p a r t i c u l a r good i s purchased  and the greater the distance t r a v e l l e d to shop; and 2 . the higher the family status, i . e . the larger the family  s i z e , the shorter the distance that consumers go i n order  to shop. 1. Underlying the f i r s t hypothesis i s the notion that d i f f e r e n t s o c i a l c l a s s groups have varying tastes within the one product type which no one store, except perhaps a department store, can cater f o r . The higher the s o c i a l c l a s s group, the more s p e c i a l i s e d , expensive and sophisticated tastes become. However, specialised-product stores serve a minority of the population, and thus i n theory they have the greatest range and threshold and t h e i r customers incur the longest distances to shop.''" A second notion suggests that the higher the s o c i a l c l a s s group, the greater the mobility of the i n d i v i d u a l . Studies have shown that upper and middle cl a s s groups make greater use of the private automobile, which compared with public transportation i s f a s t e r and more convenient. Leigh, R., Specialty R e t a i l i n g : A Geographic A n a l y s i s , B. C. Geographical S e r i e s , no. 6, Vancouver, Tantalus Research Ltd., 1965, shows that there i s a tendency for such stores to locate either i n , or close to, the downtown area and thus be i n t r a d i t i o n a l l y the most accessible l o c a t i o n to the t o t a l Metropolitan market, or i n an upper middle cl a s s suburban area and therefore c l o s e l y adjacent to at least one segment of the market. 103 and encourages longer distances to be travelled. The higher social class groups constitute mostly white-collar workers whose place of employment is more closely associated with service centres in cit i e s than is the case for blue-collar workers. Therefore, much shopping activity may originate from the place of work rather than the place of residence, thus accounting for longer distances between home and shop and the by-passing of intervening opportunities. Meanwhile, upper and lower middle-class housewives generally have more freedom than their lower-class counterparts to shop at greater distances because they have less financial need to be engaged in full-time 3 employment. 2. In the second hypothesis i t is suggested that distance travelled is affected by the size of family, since the needs of children tend to restri c t adult l i f e - s t y l e s . Households consisting of one or two persons, thus containing no children, have more time, and in some cases more money, available than larger households to engage in shopping a c t i v i t i e s . It i s argued that where there are children in the family, small families have a greater degree of mobility than large families. Moreover, families where the children are older, or at least at school, are more mobile than those who have children of ,pre-school age. 2. See for example, Oi, W. Y. and Suldiner, P. W., An Analysis of Urban Travel Demands, Evanston, 111., Northwestern University Press, 1962; Meyer, J. R. , Kain, J. F. and Wohl, M., The Urban  Transportation Problem, Cambridge, Mass., Harvard University Press, 1965. 3. Kalbach, W. E. and McVey, W. W., The Demographic Bases of Canadian Society, Toronto, McGraw H i l l , 1971, 214-36. From a study based on the 1961 Census of Canada i t was shown that where the income of the husband was less than $3,000 per annum 26 per cent, of housewives participated in the labour force. In each successive category of income of husband the percentage of working wives declined; where the husband earned more than $10,000 per annum only 9.4 per cent, were employed. One factor, however, can upset such a hypothesis. The grouping on the two family status dimensions does not take account of the age of the respondent. A high proportion of the one and two person households constitute older people whose families have grown up and le f t home. Whilst this group has the potential for being very mobile, other factors are brought to bear such that i t becomes invariably the least mobile section of the population. The physical incapacities of old age and the reduction of income that accompanies retirement from full-time work place restrictions on personal mobility : for example, older people give up ownership} or reduce their use, of an automobile. Variations in the Spatial Behaviour of Retailing and Consumers Variations in the distances that consumers travel to shop cannot be appreciated without brief reference to the spatial pattern of retailing in Metropolitan Vancouver and i t s relationship, in terms of both association and interaction, with consumer residence. This section shows the development of a r e t a i l hierarchy and i t s association with the general location of the various social class and family status groups. Secondly, i t is shown how consumers relate in aggregate to the r e t a i l hierarchy, and this i s then broken down according to frequency and/or distance travelled by the various groups. 1. The spatial pattern of retailing The spatial pattern of retailing i s seen as an arbitrary hierarchical arrangement of service centres (central places) varying in their size and frequency from a single, very large downtown shopping area (Vancouver) to an increasing number of planned and unplanned regional centres, community and neighbourhood centres. The distribution and nature of service centres reflect the various concepts of central place theory (see Chapter II), and thus the changing distribution and density of population, as well as the effects of inertia and land-use zoning. Downtown Vancouver developed as the major service centre in the Metropolitan area from the time Vancouver replaced New Westminster, a former provincial capital, as the focal point of commercial activity in British Columbia. In the f i r s t half of the twentieth century residential 4. Figure 6.1 shows the distribution of a l l service centres. Information on the size of r e t a i l centres was obtained from the Planning Office of the Greater Vancouver Regional District and was assembled from municipal sources over the period from 1961 to 1968. For the purposes of mapping and analysis the following grouping of centres was made: (i) Neighbourhood centres : from approximately 10,000 sq. f t . of r e t a i l space (the corner grocery store, gas station, etc.) to 150,000 sq. f t . (a group of convenience-type stores including a large supermarket). ( i i ) Community centres : between about 150,000 sq. f t . and 400,000 sq. f t . , containing a mixture of convenience and shopping goods stores and including some specialty stores and small discount department stores. ( i i i ) Regional centres : between about 400,000 sq. f t . and 1.5 million sq. f t . , containing a concentration of the types of store found in the community centre and, in addition, the fully-fledged department store in some centres. (iv) Planned regional centres : these numbered three at the time of the survey (1964) and are distinguished by their off-street location and single ownership. A l l the centres contain one or more department stores. Included are Park Royal, West Vancouver (425,000 sq. f t . ) , Oakridge, Vancouver (495,000 sq. ft.) and Brentwood, Burnaby (350,000 sq. f t . ) . (v) Downtown Vancouver : a total of 3.9 million sq. f t . from Bute and Robson on the west to Main Street on the east and from False Creek to Burrard Inlet. 106 and commercial development grew outwards from the Vancouver core along a series of public transportation routes, to the south and east into the area constituting the present city of Vancouver and parts of Burnaby and Richmond and to the north by ferry to localities on the north side of 5 Burrard Inlet. The downtown area was the focal point of these transportation routes which helped i t to overcome i t s increasingly spatially eccentric position. In terms of numbers of stores and functions the area was dominant over a l l others. The system of unplanned regional and community centres was largely tributary to the Vancouver core, developing along the inter-urban railway and street-car routes. They were considerably smaller and catered for the more immediate needs of the local region or community. The one exception to this was New Westminister. Although smaller than Vancouver i t maintained i t s distinctiveness through distance from i t s larger neighbour and i t s a b i l i t y to serve the numerous small agricultural and industrial communities of the Fraser Valley. In the post-war period the pattern of r e t a i l development has reflected changes in the nature of residential and commercial development. Population increase and the availability of land have meant increasing peripheral development in a much more dispersed manner in areas such as the North Shore, Richmond, Burnaby and Surrey. Meanwhile, the increased population in these areas and the expanding highway orientation of economic activity have led to new focal points and a lessening of the 5. Hardwick, W. G., "Vancouver : The Emergence of a 'Core-Ring' Urban Pattern," in Gentilcore, R. L. (ed.), Geographical Approaches  to Canadian Problems, Scarborough, Ontario, Prentice Hall of Canada, 1971, 112-18. dependence on the Vancouver core. Retail developments in these peripheral areas show essentially two patterns. F i r s t l y , many of the old regional and community centres along public transportation routes (e.g. the regional centres in Richmond and South Burnaby) increased their total number of functions and took on new functions as surrounding areas increased in population. Also, many neighbourhood centres which constitute a general store and a gas station have expanded to include, for example, a supermarket and a drug store. Secondly, new, planned regional shopping centres have been built adjacent to good highway locations (e.g. Brentwood in Burnaby and Park Royal in West Vancouver), providing a range of services from department stores and specialty shops to convenience-good stores and designed not only to serve new suburban areas but to offer an alternative to shopping in downtown Vancouver or at existing centres. The overall distribution of r e t a i l centres reflects the density of population. This and the general spatial arrangement of the various social class groups in Metropolitan Vancouver have resulted in no one group being at a disadvantage in terms of distances that have to be travelled to shop (i.e. shopping at the nearest outlet selling a particular good). Vancouver is divided into two dominant social areas with the higher social class groups predominating on the west side of the city and the lower groups on the east side.** The more dispersed peripheral areas are similarly segregated with the higher social class 6. Bell, L. I., Metropolitan Vancouver ; an Overview for Social Planners, Vancouver, Community Chest and Councils of the Greater Vancouver Area, 1965. groups concentrating in North and West Vancouver, Richmond and Delta, while the lower class groups are found to the east in Burnaby, Coquitlam and Surrey. On the other hand, family status groups do vary in their association with r e t a i l distribution. The low group, the single and two-member households, live mainly in the higher density core areas, while larger, and especially younger, families are more dominant in the peripheral suburbs. The low density and dispersed nature of peripheral development mean that longer average distances are travelled to shop compared to the core areas. 2. Aggregate travel behaviour Distance separation between residence and the store chosen for the five goods asked for in the survey shows the interaction rather than the association between the consumer and the r e t a i l hierarchy (Figures 6.1 - 6.5). The spatial patterns of movements are of two types the f i r s t concerns a convenience-good item such as groceries and the second relates to the four shopping goods - dress, furniture, appliances and footwear. (a) Groceries Groceries are items which are required frequently, and in theory consumers are not prepared to travel unnecessarily long distances for their purchase. The range and threshold of these stores are small, and the stores are found in a l l levels of the r e t a i l hierarchy. The numbers and types of grocery stores vary according to level in the hierarchy. In downtown Vancouver the sole grocery outlets, apart from specialty stores for items such as meat and f i s h , are the food floors of the major department store chains. Many of the suburban department stores in the planned and unplanned regional shopping centres also have 114 food floors, but there is competition at this level from one or more of the major supermarket chains. Community shopping centres may have more than one supermarket, but many neighbourhood centres have no supermarket at a l l , especially on the rapidly-growing fringes of the Metropolitan area where improved r e t a i l f a c i l i t i e s invariably lag behind residential development. Specialty stores and/or small corner grocery stores are found in most centres, with their number reflecting the size and importance of the centre concerned. Figure 6.1 shows that the majority of the sample population travel to the nearest centre to make their major grocery purchases; and many of those who travel to a more distant centre, especially in the rural-fringe area, do so in order to shop at the nearest supermarket and take advantage of i t s lower prices and greater selection of goods. In other instances where the nearest centre is by-passed consumers normally travel to a supermarket in a larger, or higher-order, centre. The longest distances to purchase groceries are made to the planned and unplanned regional shopping centres. Moreover, these centres have the highest proportion of shopping trips which have by-passed intervening opportunities. Downtown Vancouver, at the top of the r e t a i l hierarchy, is similar to the regional centres and does not attract consumers from the whole of the Metropolitan area as would be expected. At the other end of the r e t a i l hierarchy the shortest distances are made to neighbourhood centres, and the number of instances where intervening opportunities are by-passed in favour of a more distant neighbourhood centre are very small, (b) Shopping goods The four shopping goods - dress, furniture, appliances and footwear - present a somewhat different picture from a convenience good such as groceries. F i r s t l y , shopping goods require a larger range and threshold in order to be viable. Their entry into the r e t a i l hierarchy is rarely lower than the community centre level ; in fact, in Figures 6.2 - 6.5 above the neighbourhood centres were omitted altogether.^ The location of shopping goods, therefore, results in longer journeys for the consumer even i f the nearest available outlet is used. Secondly, shopping goods are not demanded as frequently as convenience goods (see Table 6.2). This should result in the frictions imposed by distance not being as great as those for convenience goods, and the greater distances travelled to higher order centres are off-set by the greater opportunities available for comparison shopping. Figures 6.2 - 6.5 reveal similarities in the shopping patterns for the four goods. Downtown Vancouver with the largest concentration of stores and widest range of store types is by far the most frequented centre; i t is the focal point of the highest density population area and i t also attracts a large, daytime office population. However, i t s attraction in relation to other centres reflects the core-ring pattern 8 of urban development. Downtown Vancouver is clearly dominant as the choice for Vancouver's consumers, but in the surrounding municipalities of Burnaby and Richmond and on the North Shore the majority of shopping trips are made to other centres. Further away in New Westminster, Delta and Surrey fewer than five per cent of shopping trips are to downtown Vancouver. 7. The only neighbourhood centre of any importance to feature in the desire line movements was downtown White Rock; although this area was classed as a neighbourhood centre in terms of i t s size, i t does offer a number of community centre functions. 8. Hardwick, op. c i t . , 112-118. 116 The planned and unplanned regional shopping centres exhibit more l o c a l i s e d hinterlands, but considerable v a r i a t i o n e x i s t s between them from the point of view of t r i p frequency and distances t r a v e l l e d by consumers. The lowest frequency and shortest distances t r a v e l l e d apply not to the smallest centres but to those centres which lack a department store and are l i t t l e more than a concentration of community and neighbourhood services along an a r t e r i a l highway (e.g. Whalley i n Surrey and Lonsdale Ave. i n North Vancouver). Amongst the centres with department stores (the planned centres, Richmond, South Burnaby and New Westminster) only one has an extensive hinterland which overlaps considerably with other regional centres; the centre i n South Burnaby consists l a r g e l y of the p r i n c i p a l department-store o u t l e t for Simpson Sears (there i s no o u t l e t i n downtown Vancouver) and t h i s store a t t r a c t s consumers over a considerable area from the east side of the North Shore through to Richmond and Surrey. Community shopping centres receive only a very small proportion of t o t a l shopping-good t r i p s and t h e i r hinterlands are even more l o c a l i s e d . Two exceptions i n the pattern of l o c a l t r i p s are seen. Some community centres contain s p e c i a l t y stores which a t t r a c t consumers over much longer distances, for example the t r i p s to centres on the west side of Vancouver f o r dress (Figure 6.2). Secondly, consumers w i l l t r a v e l long distances to those centres which have stores noted, for example, for t h e i r bargains or good a f t e r - s a l e s s ervice. These are important considerations when buying expensive items such as f u r n i t u r e and appliances, and many of the long journeys to community centres seen i n Figures 6.3 and 6.4 are of t h i s nature. I t has been shown above that v a r i a t i o n s e x i s t i n aggregate travel behaviour; the distance travelled and the number of intervening opportunities that are by-passed vary according to the size of the centre that is frequented. The analysis in this chapter is concerned with whether these variations in distance travelled affect a l l consumers alike or whether one particular social class or family status group of consumers travel longer distances than another. 3. Variations in distance by social class and family status groups. The measurement of distance to shop is made between place of residence and store, although i t is recognised that some journeys originate from place of work or some other location. Distance is also measured over the shortest land routes, rather than along the desire lines shown in Figures 6.1 - 6.5, since the latter masks the circuituous journeys that are often necessary to cross Burrard Inlet and the arms of the Fraser River. For the purposes of analysis the mean and standard deviation of distance are calculated for each social class and family status group (Table 6.1). Table 6.1 shows that the different social class groups travel varying distances to shop, in spite of the similarities of the groups in their spatial association with the r e t a i l hierarchy. The nature of the variation for some products would tend to support the hypothesis that the higher the social class group, the greater the distance travelled to shop. Upper/upper-middle class consumers travel the longest and lower-lower class consumers travel the shortest mean distances in a l l but one case; both of the exceptions are for appliances. The variation between social class groups in relative terms is greatest for groceries with upper/upper-middle class consumers travelling on average more than twice the distance of their lower-lower class counterparts. For the four Table 6.1 Social Class Mean and standard deviation of distance in miles by Social Class  and Family Status for purchases made in Metropolitan Vancouver, 1964 Groceries Dress Mean Std. D. Mean Std. D. Furniture Mean Std. D. Appliances Mean Std. D. Footwear Mean Std. D. Upper/Upper-Middle 2.11 2.34 5.42 4.11 5.15 4.18 6.82 4.32 5.71 4.48 Lower-Middle 1.70 2.39 4.65 3.58 4.97 4.07 9.02 7.14 4.95 4.15 Upper-Lower 1.48 1.84 4.06 3.38 4.96 4.39 5.06 4.85 4.93 4.41 Lower-Lower 0.93 1.20 3.62 3.15 2.66 2.40 7.45 7.86 4.47 3.77 Older Family Status Low 1.09 1.56 3.92 3.50 Medium 1.70 2.23 4.45 3.50 High 1.89 2.15 5.23 3.91 3.72 3.87 7.61 9.16 4.81 4.45 5.07 3.95 7.38 5.38 5.06 4.19 5.73 4.77 6.39 3.67 5.26 4.28 Younger Family Status Low 1.01 1.34 3.72 3.38 3.94 3.65 7.29 9.11 4.32 4.10 Medium 1.90 2.47 4.93 3.74 5.32 4.51 7.03 5.10 5.35 4.36 High 1.81 1.98 4.55 3.43 5.26 3.75 7.60 4.77 5.28 4.19 Source : Sample Survey of Metropolitan Vancouver, 1964 (see Appendix A). shopping goods the relative differences between groups are less. These result from the absence or weakness of these goods at the neighbourhood and community shopping centres respectively and the need for a l l consumers to travel greater distances. Furthermore, shopping goods are demanded less frequently and therefore considerations of distance may play a smaller part with a l l consumers. A decline in distance travelled for dress and footwear is related to social class position, but for furniture the distance for three of the groups is virtually the same and for appliances no trend is revealed at a l l . The hypothesis which states that the higher the family status position, i.e. the larger the family size, the shorter the distance travelled to shop is not supported (Table 6.1). In only one instance, for appliances on the older family status dimension, does the low group have the highest figure, and the degree of within-group variance probably precludes this difference from being significant. Consumers in the low group travelling the shortest distances reflects the spatial association between this group and the underlying structure of r e t a i l activity; the one and two member households are found mainly in the older and denser core areas of Metropolitan Vancouver and thus in closer proximity to the major shopping centres. A further inference on shopping distance is the age of the household head. The low family status groups are likely to contain a higher proportion of old persons than young persons simply because the span of years between marriage, and/or setting up home, and having children (and thus moving into the medium and high family status groups) is much smaller than between the last child leaving home and the termination of the household by death or old age. For reasons cited earlier older people are more likely to travel short distances in order to shop. In rejecting the research hypothesis i t must be noted that the tendency for family size and age of children to restrict distances travelled is not conclusively disproved. It is just that the residential location of families in the low group and the age of the household head are sufficiently strong influences to upset the hypothesis, and moreover i t is not possible to isolate these influences. However, a comparison of the older and younger family status dimensions in Table 6.1 indicates that apart from appliances the group travelling the longest distances to shop is the high group on the older family status dimension but the medium group on the younger family status dimentsion. Because of the differences in the way in which the variables load on to Factors II and III (see Chapter V) there are many instances of households being found in one group on one dimension and in a different group on the other. The low group on both dimensions contains the small, predominantly childless, households, but at the other end of the scale the high group on the younger family status dimension has a larger proportion of young children. The group, therefore, which travels the longest distance to shop is in each case the one with the more grown-up families. However, the differences between the medium and high groups are quite small and represent a trend rather than a significant variation. 4. Frequency of purchase by social class groups The f i r s t research hypothesis above stated that the higher the social class group, the more frequently goods are purchased. A convenience good such as groceries is not included since i t is a necessity and is bought f a i r l y regularly by everyone. However, a shopping good is less of a necessity, and whether or not one is bought reflects the a b i l i t y to Table 6.2 Number and Proportion of Consumers in each Social Class  purchasing Shopping Goods in Metropolitan Vancouver  over a one year period, 1963-64. Upper/ Upper-Middle Lower Middle Upper Lower Lower Lower Total T o t a i a 94(100%) 214(100%) 263(100%) 108(100%) 679(100%)b Consumers purchasing:-Dress 78(83%) 155(72%) 166(63%) 50(46%) 449(66%) Furniture 33(35%) 75(35%) 79(30%) 19(18%) 206(30%) c Total 69(100%) 134(100%) 134(100%) 41(100%) 378(100%) Consumers purchasing:-Appliances 21(30%) 41(30%) 31(23%) 7(17%) 100(26%) Footwear 52(75%) 106(79%) 97(72%) 25(61%) 280(74%) a - Total for the Metropolitan Vancouver area b - Total sample size was 681 but two questionnaires were incomplete c - Purchases of appliances and footwear were asked only in the West End of Vancouver and municipalities other than the City of Vancouver 122 afford the good and differing perceptions regarding a need for the good. Table 6.2 shows that the hypothesis is supported. Over a one year period 66 per cent of the respondents in the sample survey purchased one or more items of dress, for example; but this figure varied within each social class from 83 per cent for the upper/upper-middle class group to 72 per cent for the lower-middle, 63 per cent for the upper-lower and 46 per cent for the lower-lower class groups. For furniture, appliances and footwear the proportions for the two middle class groups are either the same or reversed, but in each case the upper-lower and lower-lower class groups rank third and fourth respectively. Analysis of variance tests In the f i n a l part of the chapter one-way analysis of variance tests are carried out on the data shown in Table 6.1 to see whether there are significant differences in distances travelled between the various social class and family status groups for each of the product types. In the analysis of variance test the null hypothesis states that there is no difference between mean values : that, in fact, the various sample 9 means come from the one underlying population. This is tested by partitioning the variance of the aggregated sample observations into between-group and within-group variation. If the null hypothesis is to be rejected, there must be significantly greater variation between the groups than within them. The F ratio in Tables 6.3 to 6.5 gives the relationship between the mean squares of the two types of variation and whether this i s significant or not in terms of the F distribution at certain degrees of freedom. 9. Blalock, H. M., Social Statistics, New York, McGraw H i l l , 1960, 242-71; Walker, H. M. and Lev, J., S t a t i s t i c a l Inference, New York, Holt, Rinehart and Winston, 1953, 196-229. 1. Social Class Table 6.3 shows that the research hypothesis discussed above, that distance to shop w i l l increase with the prestige accorded the particular social class, is accepted as s t a t i s t i c a l l y significant in the case of groceries and dress. Closer examination of the questionnaires themselves shows that the significantly greater distances of the higher social classes (Table 6.1) reflect certain r e t a i l preferences which involve longer distances, rather than simply an ab i l i t y and desire to travel longer distances. It was found that for the purchase of groceries there was a greater tendency for upper/upper-middle and lower-middle class consumers to by-pass neighbourhood and community shopping centre supermarkets and travel longer distances to grocery outlets in downtown Vancouver and the regional centres. This invariably implies by-passing one of the major supermarket chain stores in favour of the food floors in department stores. The development of, and greater preference by certain social class groups for, department-store grocery purchasing i s an interesting feature. Woodward's store in downtown Vancouver had in fact the f i r s t large-scale supermarket in the Metropolitan area:, opening before the Second World War and long before the national supermarket chains began developing their large suburban stores. Later, the downtown Eaton's and Hudson's Bay stores also established food floors. As department store operations have decentralised along with the growth of suburban population and the development of planned regional centres, Woodward's at least has extended i t s supermarket operations. By 1964 the firm had established supermarkets in i t s stores in New Westminster, Park Royal in West Vancouver and Oakridge in Vancouver. It may be that the establishment of Table 6.3 Analysis of variance test for differences in mean distance for various products by social class Product Groceries Dress Furniture Appliances Footwear Between Groups Sum of Squares 7.0 138.6 94.8 281.7 33.8 Degrees of Freedom 3 3 3 3 3 Mean Square 25.6 46.2 21.6 93.9 11.2 Within Groups Sum of Squares 2776.1 5678.2 3405.3 3488.7 5112.6 Degree of Freedom 675 446 204 96 280 Mean Square 4.1 12.7 16.6 36.3 18.2 F ratio 6.24 ** 3.63 * 1.89 2.58 0.62 * significant at the .05 level ** significant at the .01 level department store food floors was an innovation which found greater favour with middle-class consumers while lower-class consumers were more tied to the traditional, smaller and closer grocery store. This allegiance seems to have persisted in spite of the fact that department-store supermarket operations today have l i t t l e to distinguish themselves from the major supermarket chains, such as Canada Safeway and Super Value, and the latter are more numerous and better located. In contrast, the lower-lower class consumers, who travel the shortest mean distance (Table 6.1), are more likely than other consumers to v i s i t the nearest supermarket or grocery store outlet. The make-up of this group probably lends i t s e l f to a less mobile state. It includes a higher proportion than any other group of people who do not have private transportation, and in many instances these people are old (indicated in the survey data by their being retired from regular employment) and living on very low incomes. Those few consumers who regularly make their major grocery purchases at a corner store rather than a more distant supermarket are solely in this group. Even within the one shopping centre there were examples of a corner store being preferred over a nearby supermarket, i f the former was nearer to place of residence. Three features underlie the significant variations in mean distances travelled by the different social class groups to purchase items of dress. F i r s t l y , upper/upper-middle and lower-middle class consumers travel longer distances in order to shop for more specialised merchandise. This i s recognised from the sample survey where a specialty shop with single location i s frequented rather than a chain dress shop with more than one location. Single-location specialty dress shops are mainly located in downtown Vancouver or on the west side of Vancouver and on the North Shore and thus in close proximity to the largest share of the market. However, these shops also encourage long journeys on the part of upper/upper-middle and lower-middle class consumers who happen to live in generally less prestigious areas elsewhere. A second feature noted about these groups i s the greater incidence of by-passing an intervening opportunity or closer r e t a i l outlet (in some other direction) in favour of a more distant outlet. This does not merely involve by-passing any r e t a i l outlet selling items of dress, but especially one or more branches of a particular chain or department store in favour of a more distant branch. Invariably, this means that department stores and chain stores in nearby regional centres are avoided in favour of the same store in downtown Vancouver. Two factors may underlie the greater preference by upper/upper-middle and lower-middle class consumers for shopping in downtown Vancouver stores rather than stores closer to home. F i r s t l y , the downtown stores are larger and not only carry a wider selection of goods but are likely to carry more quality merchandise. The department stores can in fact contain a series of specialty shops as well. Secondly, i t has been shown that a high proportion of shopping trips in downtown Vancouver originate from the workplace rather than from the home. Downtown Vancouver has the largest concentration of jobs in the Metropolitan area, mostly within a few blocks of a l l the stores; but more important i t has an even larger concentration of the white-collar jobs that upper/upper-middle and lower-middle class people undertake. 10. Hardwick, W. G. and Leigh, R., Geography of Central Retailing, unpublished discussion paper, Department of Geography, University of B r i t i s h Columbia, 1965. 127 The third feature underlying the longer journeys by upper/ upper-middle and lower-middle class consumers i s their greater preference for the Hudson's Bay Co. department store in downtown Vancouver. This firm resisted locating in the planned regional shopping centres and other suburban locations until the latter part of the 1960's, and thus at the time of the survey had a hinterland which theoretically encompassed the Metropolitan area. As far as dress purchases are concerned the Hudson's Bay Co. store i s the most important downtown store for consumers in the surrounding areas of Burnaby, Richmond and the North Shore. Whether or not this store does attract a significantly higher proportion of these two social class groups w i l l be examined in the next chapter. The f i r s t research hypothesis i s not supported with respect to the other three shopping goods (Table 6.3). For furniture the survey data suggest that variations exist in the nature of the store where the good is purchased, but for three of the social class groups there i s virtually no difference in the distances that are travelled. The fourth group, the lower-lower class consumers, on the other hand, travelled considerably shorter distances to shop; 't' tests for the difference between two means found that the mean distance for this group i s significantly lower than for the other three groups. This in part reflects the close proximity of lower-lower class consumers living on the east side of Vancouver to the-stores they frequent in the downtown area; a second feature concerning this group i s that, unlike the other three groups, there i s a greater tendency to v i s i t centres nearby rather than by-pass them in favour of regional centres (and downtown Vancouver) further away. 128 Table 6.3 shows that for appliances the null hypothesis only just f a i l s to be accepted at the 0.5 level. However, acceptance would not imply support for the research hypothesis. There is no trend in distance travelled by social class (Table 6.1); and 't' tests for the difference between two means indicate that lower-middle class consumers travel significantly longer distances than upper/upper-middle and upper-lower class consumers. The significant differences result from particular residential locations and shopping centre choices. The survey data show that upper/upper-middle and upper-lower class consumers reside predominantly in Vancouver's West End - North Shore and the Burnaby -New Westminster - Surrey areas respectively and engage in more 'local' journeys to regional centres or downtown Vancouver compared to lower-middle class consumers; the latter have a more diffused residential location and make a higher proportion of long distance trips, e.g. from Richmond and Coquitlam to downtown Vancouver. The order by social class of mean distances travelled to purchase footwear i s similar to that for dress (Table 6.2). But the difference between groups i s not significant and could have occurred by chance. Unlike items of dress, footwear i s a product which does not generate much specialty store trade,''"''" and there i s , therefore, less likelihood of certain groups travelling greater distances to shop than other groups. In summary, significant between-group differences in distances travelled to shop are evident only for groceries and dress. Following an examination of the survey data i t i s suggested that for these two goods the various social class groups have different r e t a i l preferences, from 11. Leigh, op. c i t . , 61. 129 the point of view of store type and location, which necessitate different distances to be travelled. 2. Older Family Status The research hypothesis that size of family ( i . e . number of children) restricts the distance that consumers w i l l travel was not borne out by Table 6.2 above. In fact i t was shown that in a l l cases except one i t was the low group predominantly the one and two member households without any children, which travelled the shortest distances to shop. Table 6.4 shows that these variations between the different family status groups are highly significant in the case of grocery purchases and s t i l l significant at the .05 level for dress and furniture. These significant variations in part reflect the differences in residential location between the various groups. It was stated above, in terms of spatial association between residences and stores, that the low group predominate in the older and more densely populated areas of Metropolitan Vancouver and thus live in closer proximity at least to the nearest available store. Meanwhile, the medium and high groups are located principally in the lower density suburbs; not only are greater distances made necessary through a more diffused r e t a i l hierarchy, but frequently consumers travel greater distances than would seem necessary because neighbourhood, community or regional r e t a i l f a c i l i t i e s lag behind residential development u n t i l such time as population size i s sufficient enough to support the desired f a c i l i t y . The nature of interaction between residence and store for groceries also varies by family status group and further contributes to the low group travelling the shortest distance. The survey data show that consumers in the low group are more likely to use the nearest supermarket outlet. Within the low group i t was found that those consumers in the Table 6.4 Analvsis of variance test for differences in mean distance for various products by older family status . Product Groceries Dress Furniture Appliances Footwear Between Groups Sum of Squares 66.6 83.0 106.2 20.4 6.4 Degrees of Freedom 2 2 2 2 2 Mean Square 33.3 41.5 53.1 10.2 3.2 Within Groups Sum of Squares 2785.2 5733.8 3393.9 3750.1 5140.0 Degrees of Freedom 675 447 205 97 281 Mean Square 4.1 12.8 16.5 38.6 18.3 F ratio 8.08 ** 3.24 * 3.21 * 0.26 0.18 * significant at the .05 level ** significant at the .01 level o lower-lower social class are even more immobile, travelling an average of 0.78 miles to shop for groceries compared to the overall mean for the low group of 1.09 miles. The data show that this particular sub-group is more closely tied to the nearest grocery outlet and also there is the highest incidence here of shopping at a corner grocery store or small supermarket rather than a larger supermarket a few blocks further away. In contrast there is no evidence to suggest that consumers in the low group shop for dress and furniture at the nearest outlet more than the other two groups. For these two goods the significant between-group differences in distances travelled relate solely, i t would seem, to variations in residential location. Table 6.4 shows that for appliances and footwear the within-group variance is considerably greater than between-group variance, and therefore i t is assumed that the different sample means come from the one underlying population. It must be remembered that in the survey questions relating to appliances and footwear were not asked in the city of Vancouver (except for the West End). The low group i s thus underrepresented compared to the sample taken from the Metropolitan area as a whole, and the mean distances travelled by consumers in the low group (Table 6.1) are higher than would have been the case had the whole Metropolitan area been included. 3. Younger Family Status Table 6.5 shows that the between-group variation in distances travelled remains significant for groceries and dress and only just f a i l s to ba significant for furniture. Again, appliances and footwear f a i l to show up any significant variation. It was mentioned above that the composition of the low| medium and high groups varied on the two family Table 6.5 Analysis of variance test for differences in mean distance for various products by younger family status Product Groceries Dress Furniture Appliances Footwear Between Groups Sum of Squares 118.2 131.9 91.4 5.3 59.1 Degrees of Freedom 2 2 2 2 2 Mean Square 59.1 65.9 45.7 2.6 29.5 Within Groups Sum of Squares 2732.1 5684.9 3408.7 3765.2 5087.4 Degrees of Freedom 674 447 205 97 281 Mean Square 4.1 12.7 16.6 38.8 18.1 F ratio 14.6 ** 5.2 ** 2.75 0.07 1.63 * significant at the .05 level ** significant at the .01 level to status dimensions: on the younger family status dimension there tended to be a wider age differential between the groups. This difference leads to a sharper distinction between the groups, in comparison to those on the older family status dimension, in terms of core versus periphery residential location and the varying distances that have to be travelled to shop. This is a factor in the between-group variation in mean distance travelled for groceries and dress purchases being even more significant than was the case on the older family status dimension. Summary The purpose of this chapter was to examine the relationship between the frequency and/or distance travelled to purchase the five goods asked for in the sample survey and the social class and family status position of the consumer. The analysis focused on two research hypotheses : the higher the social class group, the greater the frequency and distance travelled to shop; secondly, the higher the family status group ( i . e . the greater the family size) the less distance that is travelled. Spatial association between the various social class and family status groups and the r e t a i l hierarchy showed that no one social class group is closer than another to the various types of shopping centres; however, the low family status group, located in the older and more densely populated core areas, is situated more closely than the other two groups to a l l types of centre. Actual interactance between home and store, meanwhile, revealed variations in distances travelled according to size of centre. The analysis that followed examined whether those variations in distances travelled affected a l l consumers alike or whether particular social class and family status groups travelled significantly longer distances. The f i r s t hypothesis was supported from the point of view of social class and the frequency of purchase for the four shopping goods and the distance travelled to purchase groceries and dress; the significant differences reflect differences in r e t a i l preference which require upper/upper-middle and lower-middle class consumers to travel longer distances. The second hypothesis was not supported; interactance reflects spatial association and the low family status group travel significantly shorter distances for the three products which were asked for over a l l of Metropolitan Vancouver. Moreover, the consumers in the low group, which contains a high proportion of old people, are more closely tied to the nearest r e t a i l opportunity, especially for a good purchased as frequently as groceries. 135 Chapter VII Social Class, Family Status and Store Preference  Introduction The choice of store made by a consumer i s a second aspect of social class, family status and consumer spatial behaviour. The previous chapter analysed distance travelled with respect to social class, family status and types of goods purchased. This chapter examines whether there are significant differences between the various social class and family status groups and their store preference for the different products. Various marketing factors, such as the price and nature of products, store atmosphere and level of service, help to give a store a certain "personality" which consumers relate to with varying degrees of intensity. That intensity can be associated with the character and make-up of the consumers themselves. One type of consumer w i l l find a particular store a pleasing place to shop and w i l l go there regularly; at the other extreme there w i l l be consumers who feel uncomfortable in the same store, and others with prior knowledge of what the store i s like who w i l l never go there at a l l . Methodology and Research Hypotheses The anlaysis in this chapter includes the four shopping goods - dress, furniture, appliances and footwear - and the purchasing 136 1 which is undertaken at department stores in Metropolitan Vancouver. Department stores are chosen since they account for a major proportion of shopping goods sales and do not present any of the classificatory problems that are involved with the more numerous, small stores selling these goods. The survey data reveal that the four department store companies - Hudson's Bay, Eaton's, Woodward's and Simpson Sears -account for 50 per cent of the purchases for dress and footwear, 53 per cent for furniture and 65 per cent for appliances. There is no evidence from the present data to suggest that one social class or family status group prefers department store purchasing more than another. Moreover, i t has been shown that department stores represent a type of retailing and have 2 something in common that finds favour with a l l consumers. The analysis again centres around the testing of a number of research hypotheses which are stated and developed at this time. It is hypothesised that: 1. the department stores in downtown Vancouver attract customers  from one social class group more than another. Hudson's Bay  has a larger share of higher status consumers than Eaton's or Woodward's, and Woodward's has a larger proportion of  lower status consumers. 2. in Metropolitan Vancouver as a whole the decentralisation of  department stores into varying social class areas results in  no department-store firm having a particular social class 1. Grocery purchases are excluded since the major supermarket chains account for the majority of purchases for each social class and family status group, and there i s no evidence to suggest that any group prefers the stores of one particular chain more than another. 2. Rich, S. U., Shopping Behavior of Department Store Customers, Boston, Division of Research, Graduate School of Business Administration, Harvard University, 1963. appeal. 3. family status differences exist between downtown and suburban  department stores, rather than between individual department  stores in the one area. 1. It has been argued that a consumer frequents a store not merely because he needs a particular product but because the store has an 3 image most like the one the consumer has of himself. Two factors underlie this selective process. F i r s t l y , the retailer generates a certain store image by promoting certain types of product, arranging and advertising them in a particular way and adopting certain selling techniques, and by so doing identifies with a particular group of consumers. Secondly, the consumer has a particular image of himself, reflecting a series of values, beliefs and attitudes, which result in him identifying more readily with one store than with another. This identity i s related to the social class position of both the consumer and the store. Martineau's studies point out that whilst an upper-middle class consumer may have the income to v i s i t various stores for a particular product, some stores w i l l be avoided because of the associations they have with lower-status consumers e.g. garish advertising or uneducated sales personnel. Similarly, lower class consumers w i l l avoid certain stores which exhibit higher status connotations, e.g. products not showing the price or sales personnel creating the atmosphere that the consumer does not f i t i n . 3. Martineau, P., "The Personality of the Retail Store", Harvard  Business Review, 36, 1958, 47-55; Martineau, P., "Social Classes and Spending Behaviour", Journal of Marketing, 23, 1958, 121-30. These two papers are part of the extensive research on consumer behaviour carried out by the author whilst Director of Research and Marketing for the Chicago Tribune. 138 Martineau suggests that these status differences between 4 stores do not necessarily apply to a l l types of purchases made. He stated that middle class people, for example, had no hesitancy in buying appliances in discount or bargain stores because they were dealing with a few well-known brands where i t would be hard to make a mistake. But in the case of furniture they were more likely to go to a status-conscious store because tastes in furniture were more subtle, brand names were less well known and the support of the store's own tastes would prevent a social mistake being made. This relationship between social class and store preference has been questioned. It has been found that retailers believe their stores have universal appeal;"* they are located in the downtown area and are exposed to a l l the mass media, and thus i t is assumed that these factors would automatically result in the store being accepted by everybody in a l l parts of the city. However in spite of the claims made i t was found that the stores sampled had in fact segmented markets. Rich and Jain suggest that with higher standards of living, including increases in discretionary income, leisure time, opportunities for higher education and social benefits and movements to suburbia, i t may not be effective to use social class and l i f e cycle to explain 6 consumer behaviour. A number of aspects of consumer behaviour were tested, and social class and l i f e cycle were found not to be of significant importance in some cases of market segmentation. However, 4. Martineau, "Social Classes ", 122. 5. Martineau, "The Personality ". 6. Rich, S. U. and Jain, S. C., "Social Class and L i f e Cycle as Predictors of Shopping Behaviour", Journal of Marketing  Research, 5, 1968, 41-49. 139 in the case of store preference social class was s t i l l an important factor in differentiating behaviour.^ The specific hypothesis that there is a relationship between social class and preference for the downtown Vancouver department stores is further supported by research carried out on the r e t a i l structure 8 of downtown. It was found that two sub-areas are identifiable within the downtown area. The one centring on Hastings Street and the Woodward's store on the east side of downtown exhibits the types of goods and store personalities which would appeal to lower status consumers. The other further west on Granville Street, focussing on the Hudson's Bay store, characterises a higher status shopping area with higher-priced goods and stores with a more exclusive personality. The Eaton's store and nearby specialty stores are juxtaposed between these two 9 sub-areas. It was found that the hinterlands of these two sub-areas also varied. There i s no even distribution across the Metropolitan area. Leaving aside those areas in the Fraser Valley, such as Surrey, where few consumers originate, i t was found that the Granville Street area draws more heavily on the west side of Vancouver and the North Shore of Burrard Inlet than on the east side of the city. On the other hand, the Hastings Street area draws customers more evenly across the city and is 7. This view is confirmed in Levy, S., "Social Class and Consumer Behaviour", in Newman, J., (ed.), On Knowing the Consumer, New York, J. Wiley, 1966, 146-61; and Engel, J. F., Kollat, D. T. and Blackwell, R. D., Consumer Behavior, New York, Holt, Rinehart and Winston, 1968, 263-343. 8. Hardwick, W. G. and Leigh, R., Geography of Central Retailing, unpublished discussion paper, Department of Geography, University of British Columbia, 1965. 9. Since the survey date (1964) Eaton's has moved i t s department store from Hastings Street to Granville Street, across from the Hudson's Bay Co. store. 140 more strongly represented than the Granville Street area on the east side of the city. These differences in hinterland reflect a basic social class difference in the nature of residential areas. Vancouver is sectorally divided with the west side of the city, and much of the North Shore, being a higher status area than the east side. The present analysis w i l l show whether hinterland differences in store preference are in fact social class differences, or whether, for example, the Granville Street area is more preferred by west side consumers merely becuase i t i s nearer than Hastings Street to where they live and work. 2. The second hypothesis suggests that there is no significant relationship between social class and department store preference over Metropolitan Vancouver as a whole. F i r s t l y , Hudson's Bay Co. at this time (1964) had no suburban outlets. Many consumers living in suburban areas presumably would have frequented this store, had i t have had a suburban location or had they wished to shop downtown. In considering the frictions imposed by distance, another department store, which i s closer, i s chosen instead; and, therefore, should these consumers have been predominantly from one or two social class groups, their choice of an alternative store may result in a blurring of that store's overall social class image. Secondly, Eaton's and Woodward's have both decentralised their a c t i v i t i e s . Both stores are found at the Park Royal shopping centre and in New Westminster; Eaton's also has a store at the Brentwood shopping centre in Burnaby, and Woodward's are at the Oakridge shopping centre in South Vancouver. It was shown in Chapter VI above that these regional shopping centres have more localised hinterlands than a 141 downtown area. Since the hinterland areas concerned vary in social class terms, i t i s unlikely that the two stores w i l l appeal to the same group over the whole Metropolitan Vancouver. Thirdly, the suburban department stores are smaller than their downtown counterparts, and there is perhaps less opportunity for the stores to distinguish themselves from one another. For example, the merchandise carried w i l l be more limited in suburban stores and w i l l cater more to general than specialised tastes. Finally, there i s a higher proportion of young families in the suburban areas compared to the City of Vancouver i t s e l f . Younger people invariably have had less time or less need to build up associations or prejudices about various stores. 3. The third hypothesis states that there are significant family-status differences between downtown and suburban department stores. This reflects, f i r s t l y , the differences in residential location of the various family status groups : the majority of trips to downtown Vancouver stores are from those inner areas where the proportion of consumers in the low group, the one and two member households, i s highest. Secondly, i t has been found that older people, especially those without children, are more likely than younger people to travel 10 downtown from outer suburban areas. There is less likelihood of department stores in the one area of Vancouver having variations in the family-status characteristics of their consumers. It has been found that variations in consumers 10. Jonassen, C. T., The Shopping Center Versus Downtown, Columbus, Ohio, Bureau of Business Research, College of Commerce and Administration, Ohio State University, 1955, 70, 81-82, 92. according to family-status characteristics occur principally between regular department stores and discount department stores, with the latter attracting a higher proportion of younger and larger f a m i l i e s . ^ The four Vancouver area department-store firms (Hudson's Bay, Eaton's, Woodward's and Simpson Sears) in fact belong to the former group. The three research hypotheses are tested using the chi-square 12 s t a t i s t i c . It is designed to test the deviation between the observed and expected frequencies of each c e l l in an n x m frequency table where n are the department stores and m the various social class and family status groups. Besides being a non-parametric test, the chi-square st a t i s t i c i s also used where the sample size is small and the exact nature of the underlying population distribution is not known. In some cases there are c e l l s where the expected frequency is too small, i.e less than five. It i s possible to combine cells and obtain a frequency of more than five. Secondly, in a 2 x 2 frequency table Yates' correction i s applied, whereby the frequency in each c e l l i s raised or lowered by 0.5, keeping marginal distributions (and thus expected frequencies) the same and reducing the value of chi-square and therefore the possibility of rejecting the null hypothesis. Social Class and Department-Store Preference in Downtown Vancouver The f i r s t hypothesis states that department stores in the downtown area do not have the same appeal to a l l social classes, and i t is seen that this relationship between social class and department-store 11. Rich, op. c i t . , 108-110. 12. Walker, H.M. and Lev, J., S t a t i s t i c a l Inference, New York, Holt, Rinehart and Winston, 1953, 81-108; Blalock, H. M . , Social  Statistics, New York, McGraw H i l l , 1960, 212-41. preference is s t a t i s t i c a l l y significant with one exception (Tables 7.1-7.4). Moreover, there is a consistency about the nature of store preference. The patronage of Hudson's Bay and Woodward's is polarised with the former attracting a disproportionate share of upper/upper middle and lower middle class consumers, while the latter attracts mostly consumers from the two lower class groups. This is a finding which is supported by research undertaken on store hinterlands and the 13 social nature of residential areas. Eaton's, on the other hand, does not identify i t s e l f with any particular social class or classes. 1. Dress The most highly significant relationship i s seen for women's dress (Table 7.1). Of those persons shopping in downtown stores, more than half in the upper/upper-middle and lower-middle class groups shop at Hudson's Bay, while only slightly more than one sixth of those in the upper-lower and lower-lower class groups shop at the same store. Furthermore, Woodward's is hardly a factor in upper/upper middle class shopping for dress, nor is Hudson's Bay for the lower-lower class. Eaton's, in addition to not identifying with any particular social class or classes, also does not exhibit any directional bias : the highe proportion of any one group shopping for dress at Eaton's is from the lower-lower class group, while the next highest is from the upper/upper-middle class group. Hardwick and Leigh found that the hinterland for dress purchases at Hudson's Bay was more biased than the hinterland for the store as a whole, with a marked concentration of the former on the 13. Hardwick and Leigh, op. c i t . 144 Table 7.1 Frequency of consumers by social class  and department-store preference for dress in Downtown Vancouver Hudson's Bay Eaton's Woodward's Total Upper/Upper-Middle 11 6 3 20 Lower-Middle 23 11 8 42 Upper-Lower 10 9 24 43 Lower-Lower 1 8 11 20 Total 45 34 46 125 Chi square = 27.49 degrees of freedom = 6 significant at the 0.01 level Table 7.2 Frequency of consumers by social class  and department-store preference for furniture in Downtown Vancouver Hudson's Bay Eaton's Woodward's Total Upper/Upper-Middle ^ 1 Q l 25 & Lower-Middle Upper-Lower 5 8 11 24 & Lower-Lower Total 19 18 12 49 Chi square = 12.80 degrees of freedom = 2 significant at the 0.01 level 145 Table 7.3 Frequency of consumers by social class  and department-store preference for appliances in Downtown Vancouver Upper/Upper-Middle  & Lower-Middle Upper-Lower  & Lower-Lower Total Hudson's Bay 9 2 11 Eaton's/Woodward's Total 3 12 10 21 Chi square = 3.82 degrees of freedom = 1 not significant at the 0.05 level Table 7.4 Frequency of consumers by social class  and department-store preference for footwear in Downtown Vancouver Upper/Upper-Middle  & Lower-Middle Upper-Lower  & Lower-Lower Total Hudson's Bay Eaton's Woodward's Total 17 8 * 29 5 6 8 19 22 14 12 48 Chi square = 6.35 degrees of freedom= 2 significant at the 0.05 level western side of the city, especially the downtown area (the downtown 14 worker) and the West End. This implies a higher status consumer which is certainly borne out by Table 7.1. For Woodward's, i t was found that the hinterland for dress purchases had a less noticeable east Vancouver bias than that for the store as a whole. This resulted again because downtown workers, with their more dominant west-side residential locations, frequented the store. However, this similarity between the two stores cannot be extended to similar types of consumers. Table 7.1 suggests that the two stores either attract the same type of downtown worker in significantly different numbers and/or attract different types of downtown worker. ; 2. Furniture Table 7.2 indicates that the relationship between social class and department-store preference for furniture is s t i l l highly significant, thus supporting Martineau's claim that for certain products consumers w i l l identify with stores where they know their tastes w i l l be reinforced. Furniture is a good which is less frequently demanded than dress and i s invariably more expensive, and i t is suggested that before purchase there 16 is a somewhat lengthy shopping around". One may have expected this to lead to greater comparison shopping between department stores and less customer bias towards a particular store. However, should there be comparison shopping, i t i s perhaps confined to the learning process : 14. Ibid., 43-46. 15. Martineau, "Social Classes 122. 16. Hardwick and Leigh, op. c i t . , 57. 147 purchases at department stores which are then made are differentiated according to social class. The two middle-class and two lower-class groups are collapsed to form a higher and a lower status group respectively, since various joint distributions have expected frequencies of less than five. The same social class polarisation exists between Hudson's Bay and Woodward's, but for the higher status group the distribution of consumers between Hudson's Bay and Eaton's is more evenly divided than was the case for dress pruchases. 3. Appliances With the exception of the West End, Vancouver respondents were not asked about appliance and footwear purchases. Thus, in Table 7.3 the number of observations is small; and since many expected frequencies are less than five, the number of social class groups has been reduced and the purchase at Eaton's and Woodwards have been combined. Purchasing appliances does not have the same social class connotations as purchasing dress and furniture. The differences between stores are more likely to focus on the brand name that is carried, rather than notions of quality, style and taste. Meanwhile consumer differences w i l l centre on what they can afford to pay. If social class differences are found between the various stores, i t w i l l perhaps be less a reflection of the good i t s e l f but more one of overall shopping atmosphere and associations one has with that atmosphere. Table 7.3 shows that relationship between social class and department-store preference for appliances is hot s t a t i s t i c a l l y significant, (although i t should be noted that the chi-square sta t i s t i c i s only 0.02 away from the c r i t i c a l limit whereby the null hypothesis would be 148 rejected). Since Hudson's Bay was the only department store to have no suburban outlets at this time, i t i s to be expected that this store would receive the majority of appliance purchases from a largely suburban population. Moreover, i t s attraction i s greater amongst the higher status group. For Eaton's and Woodward's together, on the other hand, a larger number of consumers are attracted from the lower status group. 4. Footwear Table 7.4 indicates that the relationship between social class and department-store preference for footwear, while not as strong as that for dress and furniture, i s s t i l l s t a t i s t i c a l l y significant. The weaker relationship may reflect the nature of the sample population. Since the sample excluded most Vancouver respondents, the reasons for choosing downtown department stores, especially Eaton's and Woodward's which are located in suburban areas, may be for other than social class reasons. As was the case for appliances, Hudson's Bay has the largest share of the market amongst the sample population, and furthermore, i t attracts a much higher proportion of upper/upper middle and lower-middle class consumers than the other two stores. Woodward's attracts more lower-class consumers, while again Eaton's appeal i s more evenly divided between the two status groups. Social Class and Department-Store Preference in Metropolitan Vancouver The second hypothesis states that over the Metropolitan Vancouver area as a whole the various department-stores have no particular social class appeal. The stores of the one firm, especially Eaton's and Woodward's, are located in socially very different areas and attract 149 very different types of consumer. Tables 7.5 - 7.8 show that the hypothesis is supported in a l l but one case. The polarisation of department-store preference seen in downtown Vancouver is no longer exemplified. In suburban areas, in the absence of Hudson's Bay at this time, Eaton's and Woodward's do not identify with middle class and lower class groups respectively. For the most part Woodward's capture a larger share of the market, in terms of number of consumers, and have the majority of consumers in one or both of the middle class groups as well as perhaps the two lower-class groups. Figures 6.2 - 6.5 indicate that the regional shopping centre in south Burnaby, which consists largely of the major Simpson Sears outlet in Metropolitan Vancouver, generates longer journeys to shop than other regional centres, perhaps suggesting among other things a store image which is attractive to a particular group of consumers. The location of Simpson Sears in a generally lower status area and a hinterland which is predominantly east side suggest an appeal to mainly lower status consumers. However, Tables 7.5-7.8 show that this store does not appeal to any one particular group, and certainly not to only lower class consumers. It appeals most to the lower-middle and upper-lower classes. By contrast, i t is relatively unimportant to the groups at either end of the social class scale. For the upper/upper-middle class this i s to be expected since the store i s not located close to where most consumers in this group live. However, the same is not the case for the lower-lower class : the store is located in close proximity to where the majority of lower-lower class consumers li v e . Table 7.5 shows that the relationship between social class and 150 Table 7.5 Frequency of consumers by s o c i a l c l a s s  and department-store preference for dress  i n Metropolitan Vancouver Hudson's Bay Eaton's Woodward's Simpson-Sears To t a l Upper/Upper-Middle 11 12 13 4 40 Lower-Middle 23 24 25 11 • 83 Upper-Lower 10 16 37 13 76 Lower-Lower 1 11 12 2 26 Tot a l 45 63 87 30 225 Chi square = 18.13 degrees of freedom = 9 s i g n i f i c a n t at the 0.05 l e v e l Table 7.6 Frequency of consumers by s o c i a l c l a s s and  department-store preference for f u r n i t u r e  i n Metropolitan Vancouver Hudson's Bay Eaton's Woodward's Simpson-Sears To t a l Upper/Upper-Middle & Lower-Middle 14 18 23 62 Upper-Lower & Lower-Lower 12 26 48 To t a l 19 30 49 12 110 Chi square = 4.26 degrees of freedom = 3 not s i g n i f i c a n t at the 0.05 l e v e l 151 Table 7.7 Frequency of consumers by social class and  department-store preference for appliances  in Metropolitan Vancouver Upper/Upper-Middle  & Lower-Middle Upper-Lower  6c Lower-Lower Total Hudson's Bay Eaton's Woodward's Simspn-Sears Total 9 9 18 7 43 11 10 19 23 13 23 66 Chi square = 6.46 degrees of freedom= 3 not significant at the 0.05 level Table 7.8 Frequency of consumers by social class and  department-store preference for footwear in Metropolitan Vancouver Upper/Upper-Middle Lower-Middle Hudson's Bay Eaton's Woodward's Simpson-Sears Total 6 11 11 17 5 22 4 7 26 57 Upper-Lower 6c Lower-Lower 15 25 54 Total 22 43 52 20 137 Chi square = 6.87 degrees of freedom = 6 not significant at the 0.05 level 152 department-store preference for dress is significant, and thus the hypothesis i s not supported. However, this results from the very strong preferences in downtown Vancouver, especially for Hudson's Bay, affecting the total Metropolitan area picture. If one compares Tables 7.1 and 7.5 i t i s seen that the various social classes have a much weaker allegiance to particular stores in suburban areas. Principally, the three suburban Woodward's stores together appeal to a wider range of consumers than the downtown Vancouver store. A similar situation i s seen for furniture, although the hypothesis is now supported (Table 7.6). While the strong attraction of Hudson's Bay for higher status consumers obviously s t i l l shows up, the other department stores show no strong social class preferences. In the Metropolitan area as a whole Eaton's is favoured more by the higher status group, but i f this difference is significant i t i s overshadowed by the appeal of Woodward's and Simpson Sears being spread more evenly across the two status groups. Department-store preference for appliances also reinforces the hypothesis (Table 7.7). However, Table 7.7 does exhibit a trend in another direction. Previous tables have shown Eaton's having either no special class appeal or to be preferred by higher status consumers. Table 7.7, on the other hand, shows that Eaton's is preferred more by the lower status than the higher status group, and amongst the lower status group i t is the most frequented store. Meanwhile, Woodward's appeals more to higher status consumers. Table 7.8 shows that the relationship between social class and department-store preference for footwear is also not significant. There is an overall tendency for Hudson's Bay and Eaton's on the one hand and Woodward's on the other to appeal to the two higher and lower social 153 classes respectively. But the similar preferences of the various social classes in suburban areas for Eaton's, Woodward's and Simpson .Sears are sufficient for the hypothesis to be supported. In summary, i t is seen that the relationship between social class and department-store preference for the four shopping goods i s significant in the case of downtown Vancouver department stores, but over the Metropolitan area as a whole the distinctive social class appeal of the various stores no longer applies. The f i r s t two hypotheses are both supported, and the analysis now turns to the third hypothesis which examines the relationship between department-store preference and family status. Family Status and Department-Store Preference The third hypothesis states that similar types of department store in the one area cannot be differentiated according to family status. If any differences are found between department stores, i t w i l l more likely be a reflection of store location and the nature of the surrounding residential area. The analysis w i l l therefore look at the preference for downtown and metropolitan area stores, followed by a comparison between downtown and suburban stores. It was shown in Chapter V above that the nature of the data results in two family-status dimensions. These are older and younger family status; and while the divisions into low, medium and high refer to approximately the same family-size groups, the constitution of the two groupings varies by age of children. Both dimensions are included here. 1. Downtown Vancouver Tables 7.9 and 7.10 show that there is no significant Table 7.9 Frequency of consumers by older family status and department-store preference for dress in Downtown Vancouver Hudson's Bay Eaton's Woodward's Total Low 15 12 20 47 Medium 23 14 18 55 High 7 8 8 23 Total 45 34 46 Chi square = 2.22 degrees of freedom = 4 not significant at the 0.05 level 125 Table 7.10 Frequency of consumers by younger family status and department-store preference for dress in Downtown Vancouver Hudson's Bay Eaton's Woodward's Total Low 24 15 23 62 Medium 15 13 18 46 High 6 6 5 17 Total 45 34 46 125 Chi square = 1.23 degrees of freedom = 4 not significant at the 0.05 level Table 7.11 Frequency of consumers by older family status  and department-store preference for dress  in Metropolitan Vancouver 155 Hudson1 s Bay Eaton' s Woodward's Simpson Sears Total Low 15 20 27 2 64 Medium 23 27 42 20 112 High 7 16 18 8 49 Total 45 63 87 30 225 Chi square degrees of = 9.76 freedom = 6 not significant at the 0.05 level Table 7.12 Frequency of consumers by younger family status  and department-store preference for dress in Metropolitan Vancouver Low Medium High Total Hudson's Bay 24 15 6 45 Eaton's Woodward's 27 25 11 63 27 47 13 87 Simpson Sears Total 5 83 17 104 8 38 30 225 Chi square =14.34 degrees of freedom = 6 significant at the 0.05 level relationship between family status and downtown department-store preference for dress. However, Table 7.10 does show one slight trend. When comparing Eaton's and Hudson's Bay, the former's distribution of consumers in the low and medium groups i s proportionately closer than the marginal distribution, while the latter store has a proportionately larger share of consumers in the low group, i.e. households either without children or one child over the age of ten. If department stores in the one area are going to appeal to one particular type of family more than another, i t would most likely have been with dress purchases where these differences would have emerged. Unlike many other products, dress not only carries connotations of social class but has styles which appeal to different age groups. While the latter may be a significant way in which to distinguish the smaller dress store, the same differentiation does not seem to apply to downtown department stores; although i t should be noted that differentiation is d i f f i c u l t when two age groups, the young and old, may be included in the low group. For the other three shopping goods purchased in downtown department stores the expected frequencies in some of the cells are too small to permit a chi-square test to be used, and collapsing groups would in this case have negated the purpose of the analysis. However the distribution of consumers would suggest that for each of the three goods, furniture, appliances and footwear, the downtown department stores do not appeal to any one particular family-status group. 2. Metropolitan Vancouver The organisation of frequency tables for department-store preference over the whole metropolitan area allows us f i r s t l y to see 157 the overall appeal of department-store firms. Secondly, since Hudson's Bay and Simpson Sears at this time (1964) are solely located in downtown Vancouver and suburban areas respectively, there is an opportunity to see whether location, as well as store appeal, is a factor in differentiating department stores according to family status. Table 7.11 shows that the relationship between older family status and department-store preference for dress is not significant. However, there is a tendency for Hudson's Bay to attract a larger proportion of consumers in the low group (i.e. the one and two member households) compared to, for example, Simpson Sears. The regrouping of consumers on the younger family-status dimension (Table 7.12) results in the relationship being significant at the 0.05 level. Hudson's Bay has a larger share of consumers in the low group compared to the other three firms, reflecting, besides a downtown-only location, a hinterland where the majority in this group live. On the other hand, Simpson Sears, without an inner-city location, has the smallest proportion of consumers in the low group; .again this i s a reflection of hinterland, since the south Burnaby store draws the majority of i t s consumers from the outer and newer suburban areas where medium and larger-size families live. Eaton's and Woodward's present an interesting contrast with respect to dress purchases. Both firms have the same number of downtown and suburban outlets, and with the exception of their respective outlets at Brentwood and Oakridge the two stores are located close together. However, Table 7.12 shows that Eaton's appeal is disproportionately high amongst the low group, while for Woodward's the same is true for the medium group. Compared to Hudson's Bay and Eaton's, Woodward's and Simpson Sears are more oriented to families with children, at least as 158 far as dress purchases are concerned. For furniture and appliances the expected frequencies in some cells are again too small to permit a chi-square test to be used. The observed frequencies would suggest, on the other hand, that no relationship exists between family status and department-store preference. Contingency tables were also constructed for footwear, but no significant relationship was found on either of the two family status dimensions; although on the younger family-status dimension a similarity could be seen with Table 7.12 in that Hudson's Bay has a disproportionate share of i t s consumers in the low group. 3. Downtown versus Suburban Preference Table 7.12 above and a comparison between Tables 7.10 and 7.12 suggest that the differences between department stores with respect to family status may be more significant when the stores are arranged by location, rather than seeking differences between the individual stores. It may be that Eaton's and Woodward's attract different family status groups for dress purchases, but in terms of overall appeal for shopping goods the preferences for Hudson's Bay and Simpson Sears, especially, indicate a relationship between family status groups and their residential location. Frequency tables are given for each of the four shopping goods on both family-status dimensions on the basis of whether consumers shopped for a product in downtown Vancouver or at a suburban location (Tables 7.13 - 7.20). Tables 7.13 and 7.14 show that the relationship between location and family status for dress purchases i s highly significant, and the hypothesis is therefore supported. Downtown Vancouver appeals more 'to the low group, while the suburban shopping areas appeal more to families Table 7.13 Frequency of consumers by older family status  and location of department-store preference for dress  in Metropolitan Vancouver Downtown Vancouver Suburbs Total Low 47 17 64 Medium 55 57 112 High 23 26 49 Total 125 100 225 Chi square = 11.65 degrees of freedom = 2 significant at the 0.01 level Table 7.14 Frequency of consumers by younger family status  and location of department-store preference for dress  in Metropolitan Vancouver Downtown Vancouver Suburbs Total Low 62 19 83 Medium 46 58 104 High 17 21 38 Total 125 100 225 Chi square = 21.68 degrees of freedom = 2 significant at the 0.01 level Table 7.15 Frequency of consumers by older family status  and location of department-store preference for  furniture in Metropolitan Vancouver 160 Low Medium High Total Downtown Vancouver 14 28 7 49 Suburbs 18 33 10 61 Total 32 61 17 110 Chi square = 0.13 degrees of freedom = 2 not significant at the 0.05 level Table 7.16 Frequency of consumers by younger family status  and location of department-store preference for  furniture in Metropolitan Vancouver Downtown Vancouver Suburbs Total Low 21 Medium 21 High 7 Total 49 18 31 12 61 39 52 19 110 Chi square =2.19 degrees of freedom = 2 not significant at the 0.05 level Table 7.17 Frequency of consumers by older family status  and location of department-store preference for  appliances in Metropolitan Vancouver Downtown Vancouver Suburbs Total Low 7 7 14 Medium 10 27 37 High 4 11 15 Total 21 45 66 Chi square = 2.71 degrees of freedom = 2 not significant at the 0.05 level Table 7.18 Frequency of consumers by younger family status and location of department-store preference for appliances in Metropolitan Vancouver Downtown Vancouver Suburbs Total Low 5 5 10 Medium 11 32 43 High 5 8 13 Total 21 45 66 Chi square = 2.56 degrees of freedom = 2 not significant at the 0.05 level Table 7.19 Frequency of consumers by older family status  and location of department-store preference  for footwear in Metropolitan Vancouver Downtown Vancouver Suburbs Total Low 23 13 36 Medium 19 59 78 High 6 17 23 Total 48 89 137 Chi square = 17.88 degrees of freedom = 2 significant at the 0.01 level Table 7.20 Frequency of consumers by younger family status  and location of department-store preference  for footwear in Metropolitan Vancouver Downtown Vancouver Suburbs Total ±22. - 26 n 4 3 Mea^um 14 5 1 6 5 8 21 29 Total 48 89 1 3 7 Chi square = 18.12 degrees of freedom = 2 significant at the 0.01 level 163 with children. Underlying this is the fact that the majority of trips into downtown Vancouver are from the City of Vancouver i t s e l f , while trips to suburban shopping centres almost exclusively originate from surrounding suburban areas. Furthermore, the two areas are differentiated according to family status : the majority of consumers in the low group are found in the City of Vancouver, while larger families are more concentrated in the suburban areas. Table 7.15 to 7.18 indicate that there is no significant relationship between location and family status for furniture and appliance purchases and the hypothesis is therefore not supported. Compared to dress both products are purchased less frequently and are probably accompanied by considerable 'shopping around', especially by the younger and more mobile consumer. In spite of this, and although no significant relationship exists between family status and store location, i t can s t i l l be seen that downtown Vancouver has a tendency to attract a higher proportion of consumers in the low group. The relationship between family status and store location is highly significant for footwear (Tables 7.19 - 7.20). However, this does not reinforce the hypothesis since we are not concerned with contrasting family-status/store location relationships (downtown-low family status versus suburban-medium and high family status) : the sample survey for footwear excluded the City of Vancouver (except for the West End) and thus a substantial proportion of consumers in the low family status group. The majority of consumers in the low group from this predominantly outer suburban population shop for footwear at downtown department stores, while consumers with larger families shop overwhelmingly at suburban locations, often closer to home. The dominance of consumers from the low group shopping in downtown Vancouver cannot be explained entirely, for example, by shopping from place of work or the wider choice of downtown stores, since these factors apply to the other groups as well. An important consideration in this context is perhaps that consumers in the low group have the freedom, especially from young children, to travel often greater distances to the downtown area. Moreover, many of the older consumers in this group may be traditionally tied to downtown stores, since they grew up in an era before large-scale suburban shopping expansion took place. Summary An analysis of social class, family status and department-store preference reveals two conflicting notions. F i r s t l y , consumers have different tastes, reflecting different needs or more subtle differences in family background and l i f e style, which result in a preference for stores where those tastes w i l l be reinforced. Secondly, both consumers and retailers may not be aware of these types of variation between stores. Moreover, in order to compete with other firms for the ever-expanding middle-class market, retailers w i l l frequently deny that biases, especially social class ones, occur at a l l . There are variations in the preference of the different social classes for the downtown Vancouver department stores, but similar variations are not revealed amongst the stores over the metropolitan area as a whole. In the downtown area there is a significant polarisation of store preference with Hudson's Bay and Woodward's attracting the middle class and lower class markets respectively. However, the diffusion of department stores into surrounding suburban areas, and more particularly the location of stores of the one firm in, socially, very different residential areas, have resulted in suburban stores not having the same narrow social class appeal of their downtown parents. Department stores in the one area cannot be differentiated according to family status. For dress and footwear significant variations exist between downtown and suburban department stores, reflecting both the different family-status composition of the inner and outer suburban population and a stronger preference amongst low group consumers in these outer areas for the downtown stores. For furniture and appliances, on the other hand, the variations between downtown and suburban stores are not significant, reflecting perhaps the longer journeys that are travelled overall and the wider search activity for products which are more expensive and are required less frequently. Chapter VIII Summary and Conclusions The purpose of this study was to analyse the relationship between a consumer's social class and family status characteristics and two aspects of his spatial behaviour - travel behaviour, in particular the distance and frequency travelled for certain goods, and preference for various department stores. The study exemplifies one of the three approaches to consumer spatial behaviour, and one that is continually refining the many variations in the ways consumers behave and seeking improved generalisations and predictions. However, an area of study that has remained relatively unexplored (at least from the point of view of shopping behaviour) is relating variations in spatial behaviour to consumers' social class and family status characteristics. Approaches to Consumer Spatial Behaviour It was shown in Chapter II above that the various approaches to the study of consumer spatial behaviour reflect a trend in geography away from the analysis of the spatial arrangement of phenomena at a very aggregate level to a more in-depth and more disaggregate analysis of the decision-making which either affects such spatial arrangements or influences the ways people interact with various phenomena. 167 The f i r s t approach is concerned with the spatial arrangement of r e t a i l functions. No direct information is known about the consumer; but since the size, nature and spacing of functions must reflect the size, needs and distribution of the population, i t is possible to make assumptions about the ways in which consumers relate to the r e t a i l functions. However, while improvements are made to theories concerning spatial structures, l i t t l e insight is gained into how consumers behave. Consumers must be considered equal (economic man) - having the same needs, knowledge of opportunities and desires to maximise some form of u t i l i t y . Moreover, a l l r e t a i l outlets selling the same good are considered equal. It is seen that consumers and r e t a i l outlets are by no means equal. Interviewing the consumer has shown that many assumptions may in fact hold true in the majority of cases, but a minority of consumers often form a distinctive group. This second approach allows such variations to be analysed. One method i s to relate variations in the r e t a i l structure i t s e l f to differences in the ways consumers behave; a second i s to differentiate consumers according to various attributes which influence decision-making in shopping and relate these to differences in spatial behaviour. A third, and very recent, approach involves the examination of consumers' attitudes and perceptions and thus account for different decisions being made. The present study has focussed on furthering our understanding of how differences in some aspects of spatial behaviour are related to consumer attributes. Previous work has concentrated on single attributes such as income or ethnic origin, but i t has been shown that consumers with similar incomes, for example, may vary in their 168 behaviour because of the differences amongst other attributes which influence behaviour such as occupation and age. A solution to this problem is to look at more than a single attribute, and since many of the attributes are interrelated i t is possible to construct a number of integrated measures. These measures are not s t a t i s t i c a l conveniences; they have a real identity and in fact may represent more than the sum of their parts. Two interrelationships were considered here - social class and family status. Social Class and Family Status A substantial body of theoretical and empirical work, especially in geography and sociology, has identified various interrelationships of individuals' socio-economic and demographic attributes and for each one i t is both possible and meaningful to rank and group individuals. People think of themselves as belonging to a particular group or class and may judge others from a standpoint of superior or inferior to their own position. It is very common for people in the same group to have similar backgrounds and goals in l i f e , to behave in similar fashion and to interact more with people in their own group than with others. Social class is one area which is more than simply an interrelationship of a few attributes. It represents one of the most important ways in which society is differentiated and stems from variations in family background and upbringing, l i f e styles, values and motivations. It was shown in Chapter III how social class differences manifest themselves, but the major hindrance to the use of social class as a basis for differentiating various forms of behaviour is the problem of measurement. A large number of real variables constitute social class, and i t is d i f f i c u l t to know which ones to include, the relative importance to attach to each one and how to quantify many of the variables. It was shown in Chapter IV that i t is more r e a l i s t i c to use a smaller number of proxy variables, such as income and occupation; while much of the detail of social class differences is perhaps lost, these variables are easier to measure and use and variations in them s t i l l express social class differences. There are many ways in which social class i s measured, but the most objective i s the use of factor analysis. It reduces a large amount of data by identifying interrelationships, or dimensions, amongst the variables (one of which is social class). Scores are obtainable for each respondent on each dimension, and i t is then possible to rank and group respondents. Family status constitutes a second important way in which society can be differentiated and refers to the position of the family in the l i f e cycle. Again proxy variables, such as family size and age of children, were used, and factor analysis revealed the nature of the interrelationship amongst the variables. The Data The key to differentiating consumers on the basis of their personal attributes, and relating these to various aspects of shopping behaviour, is the availability of good interview data. The present study was fortunate in having a random sample of some 680 respondents carried out in Metropolitan Vancouver; and questions were asked concerning the socio-economic and demographic characteristics of the household and the place of purchase for various goods. Using factor analysis the socio-economic and demographic data were reduced to three 170 dimensions - s o c i a l c l a s s and older and younger family status; the two family status dimensions r e s u l t from d i f f e r e n t questions about age of c h i l d r e n . The respondents were then ranked and a number of groups or classes were i d e n t i f i e d . Consumer Travel Behaviour In the f i r s t of two examples of consumer s p a t i a l behaviour s o c i a l class and family status were r e l a t e d to the distance and frequency that consumers were prepared to t r a v e l to purchase various goods. The analysis centred on two hypotheses - (1) the higher the s o c i a l c l a s s , the greater the distance t r a v e l l e d and the greater the frequency the good i s demanded; and (2) the higher the family status (the larger the family s i z e ) , the shorter the distance t r a v e l l e d . There were wide v a r i a t i o n s between the four s o c i a l classes i n the frequency they demanded the four shopping goods. (Groceries, on the other hand, are more necessary items and v a r i a t i o n s i n frequency are of l i t t l e consequence.) However, there was no universal acceptance of the hypothesis r e l a t i n g to distance. An analysis of variance test found s i g n i f i c a n t differences between the various s o c i a l classes for only grocery and dress purchases. These differences r e f l e c t e d varying r e t a i l preferences. Upper/upper-middle and lower-middle class consumers t r a v e l l e d longer distances than other consumers in order to shop at supermarkets and department-store food f l o o r s i n downtown Vancouver and the regional shopping centres rather than frequent more l o c a l supermarkets. Meanwhile, lower-lower class consumers were more l i k e l y to shop at the nearest grocery o u t l e t . For dress the two middle classes again t r a v e l l e d longer distances s e l e c t i n g s p e c i a l t y stores and stores in downtown Vancouver and the regional shopping centres i n 171 preference to more l o c a l stores. For f u r n i t u r e , appliances and footwear the v a r i a t i o n s i n distance t r a v e l l e d were not s i g n i f i c a n t , r e f l e c t i n g s i m i l a r i t i e s i n either the type of store chosen and/or the l e v e l of the r e t a i l hierarchy at which purchases are made. The second hypothesis was not supported. Indeed, i t was low family status consumers who t r a v e l l e d s i g n i f i c a n t l y shorter distances for three of the goods (groceries, dress and f u r n i t u r e ) . This r e f l e c t s a high proportion of old people who may be unable to t r a v e l greater distances and who i n v a r i a b l y frequent the nearest centre where a good i s a v a i l a b l e . Moreover, there i s a greater l i k e l i h o o d that consumers in the low group w i l l l i v e i n the older parts of the Metropolitan area and thus close r to the major shopping centres. The same s i g n i f i c a n t differences were not found for appliances and footwear since the sample excluded much of the C i t y of Vancouver and thus most of the older parts of the Metropolitan area. Store Preference In the second example s o c i a l class and family status were re l a t e d to a consumer's department-store preference for the four shopping goods. Three hypotheses were stated and developed - (1) the department stores i n downtown Vancouver a t t r a c t customers from one s o c i a l c l a s s group more than another with Hudson's Bay and Woodward's a t t r a c t i n g a higher proportion of high and low status customers res p e c t i v e l y ; (2) the d e c e n t r a l i s a t i o n of department stores into varying s o c i a l c l a s s areas r e s u l t s i n no f i r m having a p a r t i c u l a r s o c i a l class appeal over the whole Metropolitan area; and (3) family status differences e x i s t between downtown and suburban department stores, rather than between i n d i v i d u a l stores i n the one area. A p o l a r i s a t i o n of the downtown Vancouver department stores ( e s p e c i a l l y Hudson's Bay and Woodward's) according to s o c i a l class was s i g n i f i c a n t and reinforced other research which had shown that the same stores had very d i f f e r e n t hinterlands both s p a t i a l l y and socio-economically. However, two of the three downtown Vancouver department stores had at t h i s time located branches i n varying socio-economic parts of the Metropolitan area. Given also the sma.ller nature of suburban store hinterlands, the smaller and less s p e c i a l i s e d stores, a younger and perhaps less t r a d i t i o n a l l y - o r i e n t e d population, the same kinds of allegiances were not found over the Metropolitan area as a whole. Except for two of the department stores for one p a r t i c u l a r product, no evidence was found to support any r e l a t i o n s h i p between family status and a p a r t i c u l a r department-store preference. However, the l o c a t i o n of department stores was an important consideration. When the location of purchases was d i f f e r e n t i a t e d according to downtown Vancouver or suburban areas, i t was found that the two areas at t r a c t e d a s i g n i f i c a n t l y higher proportion of consumers i n the low group and medium and high groups respectively f or two of the products; si m i l a r trends could also be seen f o r the other two products. Retrospect and prospect The present study has shown that from a conceptual and methodological viewpoint i t i s possible to analyse a sample population according to i n t e r r e l a t i o n s h i p s amongst t h e i r personal a t t r i b u t e s and to r e l a t e these to d i f f e r e n t examples of shopping behaviour. The underlying methodology has been used previously with census-tract data to show v a r i a t i o n s i n urban r e s i d e n t i a l patterns according to 173 social class and family status; and in this study the use of individual attributes highlights the value of looking at other variations in urban functions and differences in the ways people interact with them. It is seen that social class and family status are important ways in which to differentiate some aspects of shopping behaviour, reflecting the diverse nature of society and the differences between people in terms of opportunities, values and needs. However, these differences are not static. Changes in technology, standards of living and fashion, together with increasing upward social mobility, alter the ways in which people behave and accentuate or diminish the differences between various groups. It is necessary over time to repeat such a study as this to see the extent to which the relationship between social class, family status and various aspects of shopping behaviour has changed. There is one important practical consideration emanating from the present study. It indicates the segmented nature of many stores' r e t a i l trade areas, not only spatially but also socio-economically, a factor that retailers themselves may not be aware of. An understanding of such segmentation, rather than simply the total trade area population and potential demand, should play an important part in the decision-making surrounding the location of many stores. To not be concerned with this issue may result in a store not realising the scale of patronage that i t should or even f a i l i n g to be economically viable. The present study examined two aspects of shopping behaviour, and the results indicate that further work could examine the relationship between social class and family status characteristics and other aspects of behaviour. More attention could also be given to specific locations within c i t i e s to see whether similar or different patterns of variation occur between the various groups compared to the larger city or metropolitan area. Finally, a similar type of analysis could be carried out in cit i e s smaller than Metropolitan Vancouver to see the extent to which the same hypotheses could be accepted. 175 Bibliography Abler, R., Adams, J.S. and Gould, P., Spatial Organization, Englewood C l i f f s , N.J., Prentice Hall, 1971. Banfield, E.C, The Unheavenly City, Boston, L i t t l e , Brown and Co., 1968. Barber, B., Social Stratification, New York, Harcourt, Brace and Co., 1957. Baskin, C , The Central Places of Southern Germany, Englewood C l i f f s , N.J., Prentice Hall, 1966; a translation of Christaller, W., Die Zentraler Orte in Suddeutschland, Jena, Fischer, 1933. Be l l , L. I., Metropolitan Vancouver : an Overview for Social Planners, Vancouver, Community Chest and Councils of the Greater Vancouver Area, 1965. Bell, W., "Social Areas : Typology of Urban Neighbourhoods", in Sussmann, M. B., Community Structure and Analysis, New York, Thomas Cromwell, 1959, 61-92. Berry, B.J.L., Commercial Structure and Commercial Blight, Research Paper no. 85, Chicago, University of Chicago Department of Geography, 1963. Berry, B.J.L., Geography of Market Centres and Retail Distribution, Englewood C l i f f s , N.J., Prentice Hall, 1967. Berry, B.J.L., (ed.), "Comparative Factorial Ecology", Economic Geography, 47, 1971. Berry, B.J.L., Barnum, H.G. and Tennant, R.J., "Retail Location and Consumer Behaviour", Papers and Proceedings of the Regional  Science Association, 9, 1962, 65-106. Berry, B.J.L. and Garrison, W.L., "Recent Developments of Central Place Theory", Papers and Proceedings of the Regional Science  Association, 4, 1958, 107-20. Berry, B.J.L. and Garrison, W.L., "The Functional Bases of the Central Place Hierarchy", Economic Geography, 34, 1958, 145-54. Berry, B.J.L. and Garrison, W.L., "A Note on Central Place Theory and the Range of a Good", Economic Geography, 34, 1958, 304-11. Berry, B.J.L. and Rees, P.H. , "The Factorial Ecology of Calcutta", American Journal of Sociology, 74, 1969, 445-91. 176 Blalock, H.M , Social Statistics, New York, McGraw H i l l , 1960. Blishen, B.R., "The Construction and Use of an Occupational Scale", Canadian Journal of Economics and P o l i t i c a l Science, 24, 1958, 519-31. Booth, C , Life and Labour of the People in London, London, Macmillan & Co. Ltd., 1902-3. Brush, J.E., "The Hierarchy of Central Places in Southwestern Wisconsin", Geographical Review, 43, 1953, 380-402. Brush, J.E. and Gauthier, H.L., Service Centers and Consumer Trips, Research Paper no. 113, Chicago, University of Chicago, Department of Geography, 1968. Bucklin, L.P., Shopping Patterns in an Urban Area, (Research Program in Marketing, Graduate School of Business Administration, University of California, Berkeley), Berkeley, Institute of Business and Economic Research, 1967. Bucklin, L.P., "Retail Gravity Models and Consumer Choice", Economic  Geography, 47, 1971, 489-97. Carman, J.M., The Application of Social Class in Market Segmentation, (Research Program in Marketing, Graduate School of Business Administration, University of California, Berkeley), Berkeley, Institute of Business and Economic Research, 1965. Carol, H., "The Hierarchy of Central Place Functions within a City", Annals of the Association of American Geographers, 50, 1960, 419-38. Cattell, R.B., "Factor Analysis : An Introduction to Essentials", Biometrics, 21, 1965, 190-215, 405-35. Chapin, F.S., "A Quantitative Scale for Rating the Home and Social Environment of Middle Class Families in an Urban Community", . Journal of Educational Psychology, 19, 1928, 99-111. Chapin, F.S., Contemporary American Institutions, New York, Harper, 1935. Clark, W.A.V., "Consumer Travel Patterns and the Concept of Range", Annals of the Association of American Geographers, 58, 1968, 386-96. Claus, R.J., Spatial Dynamics of Gasoline Service Stations, B. C. Geographical Series no. 10, Vancouver, Tantalus Research Ltd., 1969. Claus, R.J. and Hardwick, W.G., The Mobile Consumer : Automobile-Oriented  Retailing and Site Selection, Toronto, Collier-Macmillan, 1972. 177 Coleman, R.P., "The Significance of Social Stratification in Selling", in Marketing : A Maturing Discipline, Proceedings of the Winter Conference of the American Marketing Association, 1960, Chicago, American Marketing Association, 1961, 171-84. Cooley, W.W. and Lohnes, P.R., Multivariate Procedures for the  Behavioural Sciences, New York, J. Wiley, 1962. Cox, O.C., Caste, Class and Race, New York, Doubleday & Co. Inc., 1948. Cuber, J.F. and Kenkel, W.F., Social Stratification in the United States, New York, Appleton - Crofts - Century Inc., 1954. Davies, R.L., "Effects of Consumer Income Differences on the Business Provisions of Small Shopping Centres", Urban Studies, 5, 1968, 144-164. Dodd, S.C, "The Interactance Hypothesis : A Gravity Model f i t t i n g Physical Masses and Human Groups", American Sociological  Review. 15, 1950, 245-56. Dollard, J., Caste and Class in a Southern Town, New Haven, Yale University Press, 1937. Downs, R., "The Cognitive Structure of an Urban Shopping Centre", Environment and Behavior, 2, 1970, 13-39. Edwards, A.M., A Social-Economic Grouping of the Gainful Workers of the  United States, Washington, D.C., U.S. Government Printing Office, 1938. Eliot-Hurst, M.E. and Sellers, J., "An Analysis of the Spatial Distribution of Customers around two Grocery Retailing Operations", Professional Geographer, 21, 1969, 184-90. Engel, J.E., Kollat, D.T. and Blackwell, R.D., Consumer Behavior, New York, Holt, Rinehart and Winston, 1968. Frank, R.E. and Green, P.E., "Numerical Taxonomy in Marketing Analysis : A Review Ar t i c l e " , Journal of Marketing Research, 5, 1968, 83-98. Gale, D., "The Impact of Canadian Italians on Retail Functions and Facades in Vancouver, 1921-1961", in Minghi, J.V. (ed.), Peoples of the Living Land : Geography of Cultural Diversity  in British Columbia, B.C. Geographical Series no. 15, Vancouver, Tantalus Research Ltd., 1972, 107-24. Garrison, W.L. et a l . , Studies of Highway Development and Geographic  Change, Seattle, University of Washington Press, 1959. 178 Golledge, R.G., Rushton, G. and Clark, W.A.V., "Some Spatial Characteristics of Iowa's Dispersed Farm Population and their Implications for the Grouping of Central Place Functions", Economic Geography, 42, 1966, 261-72. Gordon, M.M., Social Class in American Sociology, Durham, Duke University Press, 1958. Grant, K.F., "Food Habits and Food Shopping Patterns of Greek Immigrants in Vancouver", in Minghi, J.V. (ed.), Peoples of the Living  Land : Geography of Cultural Diversity in British Columbia, B.C. Geographical Series no. 15, Vancouver, Tantalus Research Ltd., 1972, 125-44. Green, P.E., Frank, R.E. and Robinson, P.J., "Cluster Analysis in Test Market Selection", Management Science, 13, 1967, Series B, 387-99. Hardwick, W.G. , "Vancouver : The Emergence of a "Core-Ring" Urban Pattern", Gentilcore, R.L. (ed.), Geographical Approaches to  Canadian Problems, Scarborough, Ont., Prentice Hall of Canada, 1971, 112-18. Hardwick, W.G. and Leigh, R., Geography of Central Retailing, unpublished discussion paper, Department of Geography, University of British Columbia, 1965. Harman, H., Modern Factor Analysis, Chicago, University of Chicago Press, 1960. Hawley, A.H. and Duncan, O.D., "Social Area Analysis : A C r i t i c a l Appraisal", Land Economics, 33, 1957, 337-45. Henderson, A.M. and Parsons, T. (trans.), Max Weber : The Theory of Social and Economic Organisation, New York, Oxford University Press, 1947. Herbert, D.T., "Social Area Analysis : A British Study", Urban Studies, 4, 1967, 41-60. Hetzler, S.A., "An Investigation of the Distinctiveness of Social Classes", American Sociological Review, 18, 1953, 493-97. Hollingshead, A.B., Elmtown's Youth, New York, J. Wiley, 1949. Holly, B.P. and Wheeler, J.O., "Patterns of Retail Location and the Shopping Trips of Low-Income Households", Urban Studies, 9, 1972, 215-20. Horton, F.E., "Location Factors as Determinants of Customer Attraction to Retail Firms", Annals of the Association of American  Geographers, 58, 1968, 787-801. 179 Huff, D.L., "A Topographical Model of Consumer Space Preferences", Papers and Proceedings of the Regional Science Association, 6, 1960, 159-73. Huff, D.L., Determination of Intra-Urban Retail Trade Areas, (Real Estate Research Program, Graduate School of Business ~ Administration, Universtiy of California, Los Angles), Los Angeles, U.C.L.A., 1962. Isard, W., Methods of Regional Analysis, Cambridge, Mass., M.I.T. Press, 1960. Johnson, R.M., "Market Segmentation : A Strategic Management Tool", Journal of Marketing Research, 8, 1971, 13-18. Johnston, R.J.,'The Distribution of an Intra-Metropolitan Central Place Hierarchy", Australian Geographical Studies, 4, 1966, 19-33. Jonassen, C.T., The Shopping Center Versus Downtown, Columbus, Ohio, Bureau of Business Research, College of Commerce and Administration, Ohio State University, 1955. Karl, J.A., The American Class Structure, New York, Rinehart and Co., 1957. Kaufman, H.F., Prestige Classes in a New York Rural Community, Ithaca, N.Y., Cornell University Agricultural Experiment Station, 1944. King, L.J., S t a t i s t i c a l Analysis in Geography, Englewood C l i f f s , N.J., Prentice Hall, 1969. King, L.J., "Discriminant Analysis : a Review of Recent Theoretical Contributions and Applications", Economic Geography, 46, 1970, 367-78. King, W.R., "Structural Analysis and Descriptive Discriminant Functions", Journal of Advertising Research, 7, 1967, 39-43. Laumann, E.O., Prestige and Association in an Urban Community, New York, Bobbs-Merrill, 1966. Laumann, L.O. (ed.), Social Stratification, New York, Bobbs-Merrill, 1970. Leigh, R., Specialty Retailing : A Geographic Analysis, B. C. Geographical Series No. 6, Vancouver, Tantalus Research Ltd., 1965. Lenski, G.E., "American Social Classes ; S t a t i s t i c a l Strata or Social Groups?", American Journal of Sociology, 58, 1952, 139-44. Lessig, V.P. and Tollefson, J.O., "Market Segmentation through Numerical Taxonomy", Journal of Marketing Research, 8, 1971, 480-87. 180 Levy, S., "Social Class and Consumer Behavior", in Newman, J. (ed.), On Knowing the Consumer, New York, J. Wiley, 1966, 146-61. Lynd, R.S. and Lynd, H.M., Middletown, New York, Harcourt, Brace & Co., 1929. Lynd, R.S. and Lynd, H.M., Middletown in Transition, New York, Harcourt, Brace & Co., 1937. Marshall, J.U., The Location of Service Towns, Geography Research Publications no. 3, Toronto, University of Toronto Press, 1969. Martineau, P., "The Personality of the Retail Store", Harvard Business  Review, 36, 1958, 47-55. Martineau, P., "Social Classes and Spending Behavior", Journal of  Marketing, 23, 1958, 121-30. Marx, K., Das Capital, 1867-83; translated by Untermann, E., Capital, Chicago, C. H. Kerr & Co., 1909. Morrison, D.G., "Measurement Problems in Cluster Analysis", Management  Science, 13, 1967, Series B, 775-80. Moser, C.A., Survey Methods in Social Investigation, London, Heinemann, 1958. Murdie, R., "Cultural Differences in Consumer Travel", Economic Geography, 41, 1965, 211-33. Murdie, R.A., Factorial Ecology of Metropolitan Toronto, 1951-61, Research Paper no. 116, Chicago, University of Chicago, Department of Geography, 1969. Nadar, G.A.,"Socio-Economic Status and Consumer Behaviour^ Urban Studies, 6, 1969, 235-45. North, C.C. and Hatt, P.K., "Jobs and Occupations : a Popular Evaluation", in Wilson, L. and Kolb, W.L. (eds.), Sociological Analysis, New York, Harcourt, Brace & Co., 1949, 464-74. Nystuen, J.D., "A Theory and Simulation of Intra-Urban Travel", in Garrison, W.L. and Marble, D.F. (eds.), Quantitative Geography, Part 1 (Economic and Cultural Topics), Studies in Geography no. 13, Evanston, 111., Northwestern University Department of Geography, 1967, 54-83. Olsson, G., Distance and Human Interaction : a Bibliography and Review, Bibliography Series no. 2, Philadelphia, Regional Science Research Institute, 1965. 181 Porter, J., The Vertical Mosaic : An Analysis of Social Class and Power  in Canada, Toronto, University of Toronto Press, 1965. Pred, A., "Business Thoroughfares as Expressions of Urban Negro Culture", Economic Geography, 39, 1963, 217-33. Ray, D.M., "Cultural Differences in Consumer Travel Behaviour", Canadian  Geographer, 11, 1967, 143-56. Reilly, W.J., The Law of Retail Gravitation, New York, Reilly, 1931. Rich, S.U., Shopping Behavior of Department Store Customers, Boston, Division of Research, Graduate School of Business Administration, Harvard University, 1963. Rich, S.U. and Jain, S.C, "Social Class and Life Cycle as Predictors of Shopping Behavior", Journal of Marketing Research, 5, 1968, 41-49. Robson, B.T., Urban Analysis : A Study of City Structure with Special  Reference to Sunderland, Cambridge, University of Cambridge Press, 1969. Rose, H.M., "The Structure of Retail Trade in a Racially Changing Trade Area", Geographical Analysis, 2, 1970, 135-48. Royce, J.R., "The Development of Factor Analysis", Journal of General  Psychology, 58, 1958, 139-64. Rushton, G., "Analysis of Spatial Behavior by Revealed Space Preference", Annals of the Association of American Geographers, 59, 1969, 391-400. Rushton, G., "The Scaling of Locational Preferences", in Cox, K. and Golledge, R.G. (eds.), Behavioral Problems in Geography, Studies in Geography no. 17, Evanston, 111., Northwestern University Department of Geography, 1969, 197-227. Rushton, G., "Preference and Choice in Different Environments", Proceedings of the Association of American Geographers, 3, 1971, 146-50. Rushton, G., Golledge, R.G. and Clark, W.A.V., "Formulation and Test of a Normative Model for the Spatial Allocation of Grocery Expenditures by a Dispersed Population", Annals of the  Association of American Geographers, 57, 1967, 389-400. Rummel, R.J., "Understanding Factor Analysis", Journal of Conflict  Resolution, 11, 1967, 444-80. Salins, P.D., "Household Location Patterns in American Metropolitan Areas", Economic Geography, 47, 1971, 234-48. Schiller, R.K., "Location Trends of Specialist Services", Regional  Studies, 5, 1971, 1-10. Schiller, R.K., "The Measurement of the Attractiveness of Shopping Centres to Middle Class Luxury Consumers", Regional Studies, 6, 1972, Shevky, E. and Bell, W., Social Area Analysis, Stanford, Stanford University Press, 1955. Shevky, E. and Williams, M., The Social Areas of Los Angeles, Berkeley, University of California Press, 1949. Simons, J.W., Toronto's Changing Retail Complex : A Study in Growth and Blight, Research Paper no. 104, Chicago, University of Chicago, Department of Geography, 1966. Sims, V.M., The Measurement of Urban Home Environment, Minneapolis, University of Minnesota Press, 1936. Spence, N.A., "A Multifactor Uniform Regionalisation of British Counties on the Basis of Employment Data for 1961", Regional Studies, 2, 1968, 87-104. Sweeter, F.L., "Factorial Ecology, Helsinki, 1960", Demography, 2, 1965, 372-85. Szumeluk, K., Central Place Theory : A Review, Working Papers no. 2, London, Centre for Environmental Studies, 1968, vol. I. Taylor, S.M., Spatial Perspectives at the Consumer-Store Interface, unpublished M.A. thesis, University of B r i t i s h Columbia, 1972. Tumin, M.M., Social Stratification : The Forms and Functions of Inequality, Englewood C l i f f s , N.J., Prentice Hall, 1967. Tumin, M.M., (ed.), Readings on Social Stratification, Englewood C l i f f s , N.J., Prentice Hall, 1970. Walker, H.M. and Lev, J., S t a t i s t i c a l Inference, New York, Holt, Rinehart and Winston, 1953. Warner, W.L., Social Class in America : the Evaluation of Status, Chicago, Social Science Research Associates Inc., 1949. Warner, W.L. and Low, J.O., The Social System of a Modern Factory, New -Haven, Yale University Press, 1947. Warner, W.L. and Lunt, P.S., The Social Life of a Modern Community, New Haven, Yale University Press, 1941, "Yankee City Series", I. 183 Warner, W.L. and Lunt, P.S., The Status System of a Modern Community, New Haven, Yale University Press, 1942, "Yankee City Series", II. Warner, W.L. and Srole, L., The Social Systems of American Ethnic Groups, New Haven, Yale University Press, 1945, "Yankee City Series", III. West, J., Pl a i n v i l l e , U.S.A., New York, Columbia University Press, 1945. Y u i l l , R.S., "Spatial Behaviour of Retail Customers : Some Empirical Measurements", Geografiska Annaler, 1, 49B, no. 2, 1967, 105-115. Zwick, C , Demographic Variation : Its Impact on Consumer Behavior", Review of Economics and Statistics, 39, 1957, 451-56. 184 APPENDIX A THE UNIVERSITY OF BRITISH COLUMBIA FACULTY OF COMMERCE AND BUSINESS ADMINISTRATION BUYING HABITS SURVEY, June, 1964 Census Tract E.A. or Block Interv iew This is a survey that i s being conducted by students i n Market Research at the University of B r i t i s h Columbia. We wish to obtain some information about buying habits i n the c i t y of Vancouver. The r e s u l t s w i l l be held i n complete confidence. 1. How many people are there l i v i n g i n your household? ( C i r c l e number) 1 2 3 4 5 6 7 or more. 2. Are there any chi l d r e n (19 or under) i n the household? Yes No IF YES, ( c i r c l e number) a. How many are under 10 years of age? 0 1 2 3 4 5 6 7 or more. b. How many are from 10 to 19 years of age? 0 1 2 3 4 5 6 7 or more. 3. a. Where did you l a s t buy groceries? b. How did you t r a v e l to the store? by car by bus walked other D.K. (specify) c. Is that where you usually buy groceries? Yes No D.K. IF NO, d. Do you have a store where you buy most of your groceries? Yes No D.K. IF YES , e. What store is that? Have you bought a new dress in the last year? Yes No D.K. IF YES, b. At what store did you make your last purchase of a new dress? c. How did you travel to the store? by car by bus walked other D.K. (specify) d. Did you look at dresses at any other stores when you shopped for that dress? Yes___ No D.K. IF YES, e. How many stores did you go to look at dresses? 1 2 3 4 5 6 or more D.K. a. Have you purchased a pair of shoes for yourself in the last year? Yes No D.K. b. At what store did you make your last purchase of a pair of shoes? c. How did you travel to the store? by car by bus walked other D.K. (specify) FURNITURE a. Have you purchased any new furniture in the last year? Yes No D.K. IF YES , b. At what store did you make the last purchase? c. How did you travel to the store? by car by bus walked other D.K. (specify) APPLIANCES a. Have you purchased any new appliance in the last year? Yes No D.K. IF YES, b. What appliance was it? c. At what store did you make the last purchase? d. How did you travel to the store? by car by bus walked other D.K. (specify) a. Have you shopped at any shopping districts in the last month? Yes No D.K. IF YES , b. How many times in the last month have you shopped at? c. Where did you go on your last shopping trip? IF YES, Last How many times Place Downtown Oakridge Park Royal Simpson-Sears Brentwood Other (specify) d. Can you remember as c l o s e l y as possible how much you spent the l a s t time you shopped? Yes No 9. 10. IF YES , e. How much was i t ? Which of the people who l i v e i n your home are employed f u l l time? What is their occupation? Check Occupation Husband Wife Relative(s) Other Place of work (company and address) Hus b and Wife Relative(s) Other 11. When did your husband (you) leave school? Before grade 8 A During high school B | j Graduated from high school C | | Attended u n i v e r s i t y D | | Graduated from u n i v e r s i t y E [ [ 12. What is the approximate annual income of your household? Under 1000 A | | 3000 to 5000 C 9000 to 11,000 F 5000 to 7000 D Over 11,000 G | 1000 to 3000 B | | 7000 to 9000 E j | Comments: 13. IDENTIFICATION a. Interviewer b. Date of interview ********** APPENDIX B Blishen's Occupation Class Scale Occupation Group Breakdown * Group 1 Judges A r c h i t e c t s Engineers (chemical, mining, e l e c t r i c a l , c i v i l ) Group 2 S t a t i s t i c i a n s (M & F) Mining Managers A i r P i l o t Construction Managers Manufacturing Managers Accountants & Auditors Farmers (F) Advertising Agents Engineers (mechanical) Finance Managers O f f i c e r s (armed forces) L i b r a r i a n s (M & F) Clergymen Gov't Service O f f i c i a l s A r t i s t s & Teachers of Art Professors (M & F) Chemists & M e t a l l u r g i s t s (M & F) Authors, Editors & J o u r n a l i s t s (M & F) Foremen Communication Lawyers Dentists Physicians & Surgeons Actuaries Stock & Bond Brokers Osteopaths & Chiropractors (M 6c F) Wholesale Trade Manager School Teachers (M 6c F) Designers, c l o t h i n g (M 6c F) Despatchers, t r a i n Real Estate Agents 6c Dealers Veterinarians D i e t i t i a n s (F) A g r i c u l t u r a l , Professional Community Service Workers (M 6c F) Insurance Agents R e t a i l Trade Managers Business Service O f f i c e r s E l e c t r i c i t y , Gas 6c Water O f f i c i a l s Transportation Managers S o c i a l Welfare Workers (M 6c F) Nurses, Graduates (F) * B l i s h e n , B.R., "The Construction and Use of an Occupational Class Scale", Canadian Journal of Economics and P o l i t i c a l Science, 24, 1958, 519-31. APPENDIX B (cont'd) Occupation Group Breakdown * (cont'd) Gr oup 3 Actors (M 6c F) Laboratory Technicians (F) Surveyors Stenographers & Typists (F) O f f i c e Appliance Operators (F) Commercial T r a v e l l e r s Draughtsmen Agents Ticket S t a t i o n Photo-Engr aver s Stenographers Radio Announcers Recreation Service O f f i c e r s Music Teachers Window Decorators (F) Group 4 Bookkeeper & Cashier (M 6c F) Window-Decor ators Undertakers O f f i c e Appliance Operators Foremen, Commercial Telegraph Operators O f f i c e Clerks (F) Doctor, Dentist attendants (F) Foremen, Manufacturing Engravers, except Photo-Engravers Foremen, Transportation Forestry Managers Inspectors, Communication Purchasing Agents Locomotive Engineers Foremen, Mining Nurses, Graduate A r t i s t s Commercial (M 6c F) A r t i s t s 6c Teachers of Art (F) Conductors, Railway R e t a i l Trade Managers (F) Advertising Agents (F) Brokers Agents 6c Appraisers Radio Operators Telegraph Operators (F) Photographers Petroleum Refiners Locomotive Firemen Motion Picture P r o j e c t i o n i s t s Forewomen, Communication (F) Toolmakers Brakemen-Railway Captain, Mates, P i l o t s Personal Service O f f i c e r s Inspectors, Construction Power Stati o n Operator Radio Repairman 191 APPENDIX B (cont'd) Occupation Group Breakdown * (cont'd) Group 5 Patternmakers O f f i c e Clerks Pressmen and Plate P r i n t e r s Baggagemen Inspectors & Graders Motormen Farm Managers Furnacemen Bookbinders (F) Cutters, T e x t i l e Goods Other Pulp & Paper Workers (F) Mechanics, Airplane Telephone Operators (F) Transportation Inspectors Other Ranks, Armed Forces T e x t i l e Inspectors F u r r i e r s Transportation, Storage, Communication Workers (F) Policemen E l e c t r i c i a n s Auctioneers Plumbers Nurses, in t r a i n i n g Paper Box Makers Welders Millmen Paper Makers Music Teachers (F) Machinists Metal R o l l i n g Millmen Dental Mechanics Upholsterers Kn i t t e r s Liquor 6c Beverage Workers Mechanics, Railroad Riggers Compositors Photographers (F) Firemen Linemen 6c Service Men Farmers Logging Foremen Barbers Postmen Sheetmetal Workers Inspectors Inspectors, Metal Products (F) Engineering O f f i c e r s (on ships) Collectors Paint Makers M i l l i n e r s (F) Meat Canners F i t t e r s , Metal Shipping Clerks APPENDIX B (cont'd) Occupation Group Breakdown * (cont'd) Group 6 Sales Clerks (M & F) T e x t i l e Inspectors (F) T e x t i l e Inspectors Miners Bookbinders Personnel Service Workers (F) Brick & Stone Masons Blacksmiths Truck Drivers Chauffeur Forest Rangers Elevator Tenders S a i l o r s Nurses, p r a c t i c a l (M & F) Timbermen Cabinet & Furniture Makers Canvassers (M & F) Opticians (M & F) M i l l i n e r s Service S t a t i o n Attendants Other Personnel Service Workers (F) Waiters Bakers (F) Cutters T e x t i l e Goods (F) S t r u c t u r a l Iron Workers Operators, E l e c t r i c Street Railway Construction Foremen Painters & Decorators Bakers F i n i s h e r s , Wood Spinners T e x t i l e Inspectors (F) Launderers Elevator Tenders (F) K i l n Burners Weavers, T e x t i l e (F) Stationary Engineers Switchmen & Signalmen Religious Workers (M & F) Bus Drivers Porters Weavers F i n i s h e r s , T e x t i l e (M 6c F) Postmen (F) Knit t e r s (F) Firemen, on ships Nuns (F) Tailoresses (F) F u r r i e r s (F) Assemblers, E l e c t r i c a l Equipment (M 6c F) Millwrights Photographic Workers (F) T a i l o r s Teamsters B o i l e r Firemen Labourers, Mines 6c Quarries Cement 6c Concrete F i n i s h e r s Prospectors Butchers Telephone Operators Jewellers 6c Watchmakers (M 6c F) Machine Operators (M & F) M i l l e r s Barbers Stone Cutters Carpenters Glover Makers Dressmakers 6c Seamstresses (F) APPENDIX B (cont'd) Occupation Group Breakdown * (cont'd) Group 7 Cooks Waitresses Housekeepers & Matrons (F) Fishermen Longshoremen Shoemakers Fish, Canners, Curers & Packers (M & F) Ushers Lumbermen S awyer s Messengers Guides Janitors (M & F) Labourers Hotel, Cafe & Household Workers (M 6c F) Charworkers 6c Cleaners Newsboys Hunters 6c Trappers 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0093215/manifest

Comment

Related Items