Racial and Ethnic Residential Patterns in Canada

26
Sociological Forum, Vol. 18, No. 4, December 2003 ( C 2003) Racial and Ethnic Residential Patterns in Canada Eric Fong 1,3 and Rima Wilkes 2 The aim of this study is to understand how city factors explain racial and ethnic residential patterns in contemporary multiethnic cities. We examined residential patterns among 17 groups in 12 Canadian cities. The results suggest that we should be cautious in taking factors derived from literature based largely on European experiences at the beginning of the last century and applying them uniformly to different groups in Canadian multiethnic cities. Our results show that the residential segregation of different racial and ethnic groups is affected by different sets of city factors. In addition, most factors suggested in that literature do not relate to the segregation patterns among visible minority groups. KEY WORDS: race; ethnicity; residential patterns. INTRODUCTION Racial and ethnic residential segregation patterns are a window through which sociologists can observe the racial and ethnic relations of a society. Different residential locations are usually associated with particular physi- cal conditions, social characteristics, and quality of local institutions. These neighborhood differences in turn affect the life chances of individuals living there. If groups are concentrated in undesirable neighborhoods, their social and economic well-being will be hampered. Racial and ethnic residential pat- terns also reflect the potential for interaction among groups. Sharing neigh- borhoods facilitates cross-cultural interpersonal contact, which can improve 1 Department of Sociology, University of Toronto, 725 Spadina Avenue, Toronto, Ontario M5S 2J4, Canada. 2 Department of Sociology and Anthropology, University of British Columbia, Vancouver, British Columbia, Canada V6T 1Z1. 3 To whom correspondence should be addressed; e-mail: [email protected]. 577 0884-8971/03/1200-0577/0 C 2003 Plenum Publishing Corporation

Transcript of Racial and Ethnic Residential Patterns in Canada

Page 1: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Sociological Forum, Vol. 18, No. 4, December 2003 ( C© 2003)

Racial and Ethnic Residential Patterns in Canada

Eric Fong1,3 and Rima Wilkes2

The aim of this study is to understand how city factors explain racial andethnic residential patterns in contemporary multiethnic cities. We examinedresidential patterns among 17 groups in 12 Canadian cities. The results suggestthat we should be cautious in taking factors derived from literature basedlargely on European experiences at the beginning of the last century andapplying them uniformly to different groups in Canadian multiethnic cities.Our results show that the residential segregation of different racial and ethnicgroups is affected by different sets of city factors. In addition, most factorssuggested in that literature do not relate to the segregation patterns amongvisible minority groups.

KEY WORDS: race; ethnicity; residential patterns.

INTRODUCTION

Racial and ethnic residential segregation patterns are a window throughwhich sociologists can observe the racial and ethnic relations of a society.Different residential locations are usually associated with particular physi-cal conditions, social characteristics, and quality of local institutions. Theseneighborhood differences in turn affect the life chances of individuals livingthere. If groups are concentrated in undesirable neighborhoods, their socialand economic well-being will be hampered. Racial and ethnic residential pat-terns also reflect the potential for interaction among groups. Sharing neigh-borhoods facilitates cross-cultural interpersonal contact, which can improve

1Department of Sociology, University of Toronto, 725 Spadina Avenue, Toronto, Ontario M5S2J4, Canada.

2Department of Sociology and Anthropology, University of British Columbia, Vancouver,British Columbia, Canada V6T 1Z1.

3To whom correspondence should be addressed; e-mail: [email protected].

577

0884-8971/03/1200-0577/0 C© 2003 Plenum Publishing Corporation

Page 2: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

578 Fong and Wilkes

understanding and may lower prejudice or stereotyping among the groupsinvolved (Fong and Isajiw, 2000; Jackman and Crane, 1986; Sigeleman et al.,1996).

Most studies of residential segregation have focused on the effect ofthe city ecology on the level of residential segregation between groups. Thisapproach originated from the urban ecological studies of the early ChicagoSchool. The city is viewed as a natural phenomenon rather than an arti-fact of human creation. Groups residing in the city are sorted into locationsaccording to their socioeconomic and demographic backgrounds as well asthe urban environment. As tremendous changes that Canadian cities haveundergone in the last few decades, these changes provide a unique opportu-nity to understand racial and ethnic residential patterns from a comparativepoint of view.

Among these changes, the most obvious has been the drastic increasein “visible minorities” as a result of the change in immigration policies inthe 1970s (Badets and Chui, 1994).4 Canada was previously a racially ho-mogeneous society. The majority of the population, about 73% in 1971, wasof either British or French origin. The Asian and Black populations con-sisted of only 1% and 0.2% respectively (Kalbach, 1987). However, whenimmigration criteria changed to selecting potential immigrants based onhuman and financial capital, large numbers of immigrants were recruitedfrom nontraditional sending countries, such as those in Asia and the WestIndies. By 1991, the Black population had increased to 4% and the Asianpopulation to 16% of the total population (Statistics Canada, 1993b). Un-derstanding the residential patterns of these racial and ethnic groups posesnew challenges. Most studies of residential patterns are based on the ex-periences of European groups who arrived in North America at the be-ginning of the last century. Given the increase in the racial diversity ofcities, applying the understanding of group residential segregation patternsdrawn from European experiences to other racial groups remains to betested.

Within this diverse population, the socioeconomic resources of racialand ethnic groups differ substantially. In 1991, the average household in-come of Blacks in Canada was about $8000 lower than that of easternEuropeans, $9000 lower than that of northern and western Europeans, and$10000 lower than that of southern Europeans. Conversely, the averagehousehold income of Asians was $2000 above that of eastern Europeans,$1000 above that of northern and western Europeans, and the same as thatof southern Europeans (Fong, 1997). Since most studies are based on the

4“Visible minorities” is a term commonly used in Canadian society. It refers to all racial andethnic groups with darker skin color, including Asian groups and Blacks.

Page 3: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 579

perspective developed from the early experiences of European immigrantswho arrived in North America with limited resources, the diverse socioeco-nomic differences of today’s new immigrant groups pose new challenges tounderstanding their residential patterns.

Another major change that Canadian society has experienced in thelast few decades, which may affect urban residential patterns, is urban trans-formation as the result of globalization: an urban phenomenon that othermajor developed countries also have experienced (Ray, 1996; Rutherford,1996; Sassen, 1991). Globalization has lead to urban economic restructuringamong North American cities. Subsequently, cities have developed their ownniches for survival in the global economic system. Some have emerged as fi-nancial centers, some maintain a manufacturing base, and some concentrateon government activities.

The functional specialization of cities has a substantial impact on theirlocal communities, and disproportionately affects racial and ethnic groups(Frey and Farley, 1996). Studies have documented that a higher proportionof finance-related employment affects the economic opportunities of visibleminorities and immigrants (Logan and Molotch, 1987; Wilson, 1987). For var-ious reasons, these groups usually have lower education and technical skills.As manufacturing jobs in some cities have been replaced by finance-relatedjobs that typically require high levels of skill and knowledge, the economicachievement of visible minorities and immigrants is seriously affected, whichin turn may affect their ability to improve their residential locations (Wilson,1987). Urban functional specialization thus requires the study of residentialpatterns to go beyond group differences to explain residential patterns. Thediverse racial and ethnic composition in major Canadian cities provides anopportunity to test how urban functional specialization affects groups withdifferent levels of socioeconomic resources.

Finally, a major societal development in recent decades that may affectresidential patterns is the independence movement in Quebec. The “QuietRevolution” of organized labor, francophone businessmen, academics, andthe middle class in the 1960s perceived Quebec’s independence as the onlyway to preserve the French–Canadian nation (Salee and Coleman, 1997).The movement blossomed in later decades. It resulted in two referendumsto determine the relationship between Quebec and Canada.

During the course of the political struggle about independence, linguis-tic orientation became an important issue for the Quebec independencemovement (Denis, 1999). The strong linguistic orientation toward Frenchcreated conflict and discomfort among Anglophones in Quebec, who repre-sent a large proportion of the population of non-French origin. This politicalconflict subsequently increased the residential segregation among racial andethnic groups within Quebec—a situation that provides a good opportunity

Page 4: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

580 Fong and Wilkes

to understand how the larger political context shapes residential patternsamong groups.

Despite the importance and growing complexity of studying racial andethnic residential patterns, and the ability of multivariate analysis to evalu-ate the relative importance of various social changes, most studies of Cana-dian residential segregation are either descriptive in nature or provide bi-variate analysis (Balakrishnan, 1976, 1982; Balakrishnan and Kralt, 1987;Balakrishnan and Selvanathan, 1990; Darroch and Marston, 1971; Kalbach,1990).5 No study has attempted to provide a systematic, multivariate causalanalysis of residential segregation patterns at the metropolitan level inCanada. The major limitation is the small number of metropolitan areas. Inthe 1991 census, there were only 39 census metropolitan areas in Canada asdefined by Statistics Canada. This small number limits the degree of freedomwhen using multivariate causal analysis. As well, many of these metropolitanareas are characterized as having homogeneous racial and ethnic composi-tion.6 Excluding these racially and ethnically homogeneous cities, furtherconstrains the degree of freedom available for multivariate causal analysis.Consequently, little is known about how the various social changes discussedand factors mentioned in the literature affect the residential segregationpatterns in Canadian metropolitan areas.

Another limitation is that most studies on Canadian residential pat-terns have focused on the settlement patterns and residential segregationof various European ethnic groups (Balakrishnan, 1976, 1982; Darroch andMarston, 1971; Kalbach, 1990; Lieberson, 1970). Comparison of residen-tial segregation among racial groups in Canada is limited. One reason isthat Canada was racially quite homogeneous until the change in immigra-tion policy about four decades ago (Badets and Chui, 1994). In addition,census tract data released to the public provided limited information aboutminority ethnic groups in the 1980s. The 1981 census tract level data onlyshows information about British, French, and “other.” The 1986 data pro-vides distribution of only four visible racial and ethnic groups at the tractlevel.

For this study, we developed multivariate causal models to understandracial and ethnic residential patterns in Canadian cities using a new approach,developed by White et al. (2003), to overcome the problem of the small num-ber of metropolitan areas in Canada. This paper contributes to the existingliterature in three ways. First, it is the first study based on Canadian datato delineate the relative importance of various social factors to residential

5A few papers provide multivariate analysis at the municipal level, not at the metropolitan level.The limitations of using municipality as the unit of analysis are discussed in Fong (1996).

6In 26 census metropolitan areas, more than 75% of the total population is composed of thecharter groups (i.e., British and French).

Page 5: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 581

patterns at a multivariate level. Second, studying Canadian residential pat-terns allows us to evaluate how factors derived from American data can beapplied to other countries. This comparative perspective can refine our un-derstanding of how these factors affect residential segregation among groups.Third, the Canadian data allows us to study various European ethnic groups.The study provides a unique opportunity to disentangle the residential seg-regation patterns of 17 racial and ethnic groups in cities of diverse racialand ethnic composition, varied socioeconomic resources among new immi-grant groups, and specialized economic and social functions, influenced bythe larger political context.

The paper is divided into four sections. The first reviews factors sug-gested in the literature that affect racial and ethnic residential segregation.The second gives a brief overview of the history of immigration in Canadato provide a context in which to interpret the results reported in the sub-sequent sections. The third section discusses the dissimilarity index of thevarious European ethnic groups and visible minority groups on the basis of1991 census. The final section presents the multivariate analysis.

LITERATURE REVIEW

Drawing from the ecological study of cities, most studies of racial andethnic residential segregation focus on three sets of ecological factors. Thefirst set is related to the demographic composition of the city, which usu-ally refers to the total city population and the proportion of visible minoritypopulation in the city. Larger cities are usually associated with higher levelsof residential segregation. As Fischer (1975) elegantly argued, a larger cityencourages larger structural differentiation and results in subgroup varia-tions. In addition, a larger city is more likely to attract a larger populationof diverse cultural and social interests that reaches the critical levels neededto foster subgroup activities and institutions. In turn, it creates a situationof “institutional completeness” that further reinforces the differentiation(Breton, 1964). This dynamic is supported by recent studies based on the1990 census (Farley and Frey, 1994).

Another demographic aspect discussed in the literature is the city’sproportion of visible minorities. Previous research suggests that a largerproportion of visible minorities increases the awareness of racial and ethnicdifferences. Lieberson’s study (Lieberson, 1980) of American cities from1880 to 1960 documented that residential segregation rose as the size of theBlack population increased. However, other research has shown an oppositeeffect of visible minority size on residential segregation (Massey and Denton,1987). When the size of a visible minority group increases, existing ethnic

Page 6: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

582 Fong and Wilkes

enclaves may no longer be able to accommodate all members. Spillover ofsome of the group members to adjacent nonethnic neighborhoods may occurand may lead to a reduction in segregation. Thus, the relationship betweenvisible minority size and segregation level needs to be tested.

The second set of ecological factors suggested to be related to racialand ethnic residential segregation patterns is the structural context of thecity. The salient structural context usually discussed includes the effects ofthe city economy, the age of the city, and its functional specialization. Aweak economy can influence residential segregation patterns. During timesof economic uncertainty, households are less likely to spend large amounts ofmoney to improve their housing by moving to better neighborhoods (Fong,1996). Research in the United States also showed that the age of the city, an-other city structural context, can affect ethnic and racial residential patterns.The ecology of older cities is more likely to facilitate residential segregation,with working-class neighborhoods in close proximity (Massey and Denton,1987). Younger cities tend to have lower residential segregation levels. Leg-islation passed in recent decades in the United States, such as the Fair Hous-ing Act, discourages housing discrimination against minority groups (Farleyand Frey, 1994). Similarly in Canada, the multiculturalism policy adopted inthe late 1970s, which promoted group equality and attempted to alleviatesources of discrimination, may have encouraged spatial integration amonggroups.

In addition, as Farley and Frey (1994) pointed out, functional special-ization as a significant part of city structural context can relate to residentialpatterns in different ways. First, the economic base of a city reflects to alarge extent the socioeconomic background of its residents, which may inturn influence residential patterns. Second, different economic bases areusually associated with different types of housing, which may also affectsegregation levels. Specifically, we expect that cities with higher proportionsof manufacturing and government jobs will be more likely to decrease seg-regation between European and visible minority groups. Since manufac-turing jobs have been the backbone of minority groups, and governmentsusually hire a high proportion of visible minorities, a larger proportion ofmanufacturing and government jobs suggests a higher proportion of visibleminorities with economic security, which increases their likelihood of im-proving their residential locations. Cities with a higher proportion of retiredpeople will have a higher level of residential segregation among groups,because retired individuals are more likely to live in neighborhoods withresidents of similar lifestyle and background. Finally, cities with a higherproportion of individuals involved in trade are likely to have lower segrega-tion among groups. Since trade usually involves interacting with individuals

Page 7: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 583

from different backgrounds or even cultures, such opportunities can increaseunderstanding among groups and subsequently reduce barriers to residentialintegration.

The final set of ecological factors suggested in the literature is the eco-nomic difference between two specific groups, measured by the differencein their average household income. Research has suggested that becausehousing values vary among neighborhoods, groups that do not share neigh-borhoods may have different economic resources (Fong and Shibuya, 2000;Fong and Wilkes, 1999; Massey, 1985).

In addition to the three sets of city factors, research has also suggestedthat there are effects of specific race and ethnicity for various cultural andhistorical reasons. Among European groups, we expect that the influx ofrecent immigrant groups, such as Greek and Spanish, will encourage theirethnic members to cluster, which in turn will affect their residential patterns.However, some groups (e.g., Italian and Ukrainian) have higher proportionsof Canadian-born population than the more recently arrived southern Euro-pean groups, and research findings suggest that their Canadian-born mem-bers have low levels of ethnic retention (Isajiw, 1990). They are more likelyto disperse and share neighborhoods with other groups. Although Blacks inCanada do not have the same hypersegregation from Whites as their counter-parts in the United States, research has suggested that all visible minorities,including Blacks, experience higher levels of segregation from other groupsthan do European groups, even when controlling for other factors (Fong,1997). To explain such patterns, some studies have pointed out that visi-ble minorities may experience discrimination in the housing market (Henry,1994, Krysan and Farley, 2002). Less residential choice and unequal accessto housing information because of discrimination may contribute to theirhigher level of residential segregation. In addition, visible minority groups,mainly Asians in this case, because of their recent arrival and strong culturalvalue of family, may prefer to stay close to each other (Min, 1995).

Taken together, four sets of factors suggested in the literature may affectracial and ethnic residential segregation in cities: demographic composition,structural contexts, economic differentiation between groups, and specificracial and ethnic group effects. From the discussion, we draw the followinghypotheses:

(1) Demographic composition. A city with a larger population or alarger proportion of visible minority residents is associated with ahigher level of racial and ethnic residential segregation.

(2) Structural context. Higher levels of racial and ethnic segregation areassociated with a weak economy, a higher proportion of old housing,

Page 8: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

584 Fong and Wilkes

a lower proportion of manufacturing jobs, a lower proportion ofgovernment jobs, a higher proportion of retired population, or alower proportion of retail-related jobs.

(3) Economic differentiation between groups. Larger economic differ-ences between two groups are related to higher levels of racial andethnic segregation.

(4) Specific racial and ethnic group effects. Visible minorities are morelikely to experience higher levels of racial and ethnic segregation.

However, since most results of the relationships between these factors andresidential segregation are drawn from American data, the applicability ofthese factors to racial and ethnic residential segregation in Canada remainsto be tested.

HISTORY OF SELECTED IMMIGRANT GROUPS IN CANADA

The majority of early immigrants to Canada were European. At the timeof Confederation in 1867, the two major immigrant groups were British andFrench (Kalbach, 1990). The settlements of these two groups remained fairlyseparate in the nineteenth century. The French were concentrated in Quebecand the English in Ontario. Dutch and German immigrants formed only asmall percentage of the population.

Subsequent European groups were recruited in response to specificlabor demands at different periods of time. At the end of the nineteenthcentury, Germans, Ukrainians, Poles, and Dutch arrived in waves to settlein Manitoba, Saskatchewan, Alberta, and British Columbia (Avery, 1995).Most of them participated in agriculture, the dominant industry at that time(Avery and Fedorowitz, 1982; Gerus and Rea, 1985). A few decades later,at the beginning of the twentieth century, central and eastern Europeanagricultural immigrants were recruited for work by the Canadian PacificRailway and the Canadian National Railway (Prymak, 1991). In the 1950s,large numbers of immigrants from eastern and southern Europe were hiredto fill agricultural and manual jobs as the Canadian economy expanded(Avery, 1995).

Like the European groups, non-European groups have a long history ofimmigrant waves, but these waves have been tempered by racist immigrationlaws, which discouraged or prohibited the immigration of non-European andnon-White individuals (Taylor, 1991). Most Chinese first arrived in BritishColumbia in response to the discovery of gold in the Fraser River. When goldmining declined, 17,000 Chinese immigrants came to work on the CanadianPacific Railway between 1881 and 1884 (Li, 1998). However, in 1885 when

Page 9: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 585

that work was completed, the federal government imposed a head tax of$50 on every Chinese person who came to Canada. The tax was raised a fewtimes in subsequent years. By 1923 the Chinese Immigration Act was passed,and Chinese immigration to Canada came to a complete stop. The Act wasnot repealed until 1947 (Li, 1998).

Other groups such as Asian Indians and Blacks were also subject toimmigration restrictions in Canada (Anderson and Frideres, 1981; Johnston,1984). During the early 1900s, Asian Indians were required to pay $200upon arrival in Canada, whereas European immigrants only needed $25(Johnston, 1984). During the nineteenth century, many Black migrants fromthe United States were turned back at the border (Bolaria and Li, 1985). Asthese visible minority groups faced an overtly discriminatory environmentin Canada, it is not surprising to find that European immigrants far out-numbered their non-European counterparts throughout most of Canadianhistory.

This discriminatory environment began to change in the 1970s, as re-flected in the change in immigration laws in 1976. Individuals had previouslybeen accepted according to country of origin; now recruitment was based onuniversal, objectively evaluated factors such as education, occupation, andlanguage ability. This system, in combination with immigration sponsorshipprograms for family members, has transformed the demographic composi-tion of Canada. In the late 1980s and 1990s, the top five immigrant groupsarriving in Canada all were Asian (Driedger, 1996), which led to a substantialincrease in visible minority population.

Today, the distribution of these racial and ethnic groups reflects theirhistorical settlement patterns and experiences. The French represent a largepercentage in Quebec, while the English are in Ontario. Most otherEuropean groups settled in cities in the west.7 Disproportionately, visibleminorities, the majority of them immigrants, chose to reside in Canada’sthree largest cities, Toronto (Ontario), Montreal (Quebec), and Vancouver(British Columbia). As the immigrant population has risen, the visible mi-nority population has drastically increased in these cities. For example, in1991, the population of Toronto was 30% visible minority, and Vancouver’swas 33%.8

7Proportion of Europeans in selected metropolitan areas: Montreal: 85.51%; Ottawa: 79.43%;Toronto: 61.64%; Hamilton: 76.19%; St. Catharines: 77.63%; Kitchener: 71.17%; London:81.32%; Windsor: 78.6%; Winnipeg: 68.5%; Calgary: 65.8%; Edmonton: 63.5%; Vancouver:61.9%. The information on Europeans is estimated from the ethnic question of the 1991 PUMFindividual file. The percentage represents only groups included in the analysis.

8Proportion of visible minorities in other included metropolitan areas: Montreal: 6.5%; Ottawa:7.9%; Hamilton: 7.7%; St. Catharines: 3.2%; Kitchener: 9.2%; London: 5.5%; Windsor: 8.6%;Winnipeg: 14.5%; Calgary: 20.8%; Edmonton: 33.3%. The data is based on the visible minorityindicator from the 1991 PUMF individual file.

Page 10: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

586 Fong and Wilkes

DATA

This paper is based on the 1991 Profile file, Parts A and B, of theCanadian Census (Statistics Canada, 1993a,b). We use 12 of the 15 largestcensus metropolitan areas in Canada, except Halifax, Quebec City, andVictoria. Halifax and Quebec City are excluded because very small num-bers of Asians, eastern Europeans, and southern Europeans live in thesemetropolitan areas. Similarly, Victoria is not included because its popula-tion includes few Blacks, an important comparison group in our study.

We selected 17 major racial and ethnic groups in our study. They in-clude the 2 charter groups (British and French), 10 other European groups(German, Dutch, Polish, Hungarian, Ukrainian, Portuguese, Spanish, Greek,Italian, and Jewish), and 5 visible minority groups (Asian Indian, Chinese,Filipino, Vietnamese, and Blacks). To ensure that any group size is not sosmall that the segregation reflects only the randomness of its spatial distri-bution, we adopted Langberg and Farley’s approach (1985). This approachhas been used in various studies of ethnic groups of small size (Fong, 1996;Massey and Denton, 1992). We computed segregation between two groupswhen their population was at least 10 times the number of tracts in the city.9

The advantage of this approach is to ensure the reliability of the calculatedindices. As well, the results allow direct comparison with previous studies.

We used the index of dissimilarity to measure the residential segregationlevels of groups. The index is widely used in studying the spatial distributionof groups. It shows the proportion of group members required to changetracts to achieve an even distribution in the city. Census tracts are geosta-tistical areas designated by Statistics Canada, which are intended to reflecttheir homogeneous socioeconomic composition. Tract boundaries usuallyfollow physical features, such as major roads. Census tracts usually have anaverage population of about 4000.

The index of dissimilarity is defined as

D= 12

k∑i=1

∣∣∣∣Xi

X− Yi

Y

∣∣∣∣where Xi is the population of group X in tract i , and X is the population ofgroup X in the city; and Yi is the population of group Y in tract i, and Y isthe population of group Y in the city. The index ranges from 0 to 1. It canbe interpreted as the level of uneven distribution between two groups. The

9The following groups in specific cities are not included: Portuguese in St. Catharines, Jews in St.Catharines and Kitchener, Blacks in St. Catharines, Filipinos in St. Catharines and Kitchener,Greeks in Edmonton, Vietnamese in St. Catharines, Spanish in St. Catharines and Winnipeg.The specific group population in each city is available upon request.

Page 11: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 587

Table I. Percentage of Individuals in Major Racial and Ethnic Groups Who Completed Uni-versity Education and Percentage of Individuals With Income in the First Quartile of Income

Distribution, 1991

Completed First quartile Completed university andEthnic group university (%) income (%) first quartile income (%)

British 14.6 30.0 8.5French 14.6 24.7 7.9German 14.3 29.6 7.5Dutch 14.0 29.5 7.1Polish 17.7 23.1 7.6Hungarian 17.2 27.9 8.7Ukrainian 14.9 28.5 8.5Portuguese 3.4 18.8 1.4Spanish 11.9 12.8 3.2Greek 8.5 15.8 3.1Italian 9.6 23.6 4.9Jewish 36.5 37.2 21.3Asian Indian 22.7 18.4 8.4Chinese 23.6 18.0 9.3Filipino 33.5 16.5 9.5Vietnamese 14.4 9.7 4.6Black 9.6 15.3 3.8

Source: Statistics Canada (1991). Census of Canada. Individual Public Use Microdata File.

index is 1 when two groups in the city are distributed completely unevenlyand do not share a single common tract. The index is 0 when two groupsin the city are evenly distributed and share tracts in accordance with theproportion of their population in the city. Segregation levels are consideredlow if the value of the dissimilarity index is below 0.4, moderate if the valueis between 0.4 and 0.6, and usually are considered high if the value is above0.6. Census tracts in which more than 40% of the population is in institutionswere deleted to avoid the possible segregation caused by formal institutionssuch as mental hospitals or military bases.

DISCUSSION

Table I reports the percentage of individuals in racial and ethnic groupswho have completed university education, the percentage whose incomeis in the top quartile of income distribution, and the percentage who havecompleted university education and have income in the top quartile. Thedata is drawn from the Public Use Microdata (Statistics Canada, 1993c).10

10The PUMF individual file does not provide separate information on Asian Indians. We usedSouth Asian instead. South Asians include Bengali, Punjabi, Singhalese, Tamil, Bangladeshi,East Indian, Pakistani, and Sri Lankan. However, East Indians comprise more than 80% ofthe total South Asian population.

Page 12: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

588 Fong and Wilkes

Table II presents the average values of the pairwise segregation index forthe 12 European groups included in the analysis of the 12 census metropoli-tan areas. In other words, each value reported in the table is the averagedissimilarity index score of a specific group with another European groupin the 12 census metropolitan areas.11 The last 2 rows of the table pro-vide summary statistics of the segregation level of a specific group withall other groups included in the table (i.e., average and standard devia-tion). Because of space limitations and the large amount of informationin these two tables, our discussion focuses on three major residential seg-regation patterns revealed in Table II. Specific segregation scores betweentwo groups are mentioned when a particular segregation pattern merits at-tention. We use socioeconomic differences reported in Table I to illustratethe interpretation.

Among all the segregation indices shown in Table II, the charter groups(i.e., British and French) are more likely to share their neighborhoods withwestern Europeans (e.g., German and Dutch). Their segregation levels ingeneral are low. For example, the segregation levels of German vs. British,German vs. French, and German vs. Dutch are 0.20, 0.33, and 0.29 respec-tively. Given that most members of these groups arrived during early im-migration periods, one might conclude that the integration process has hadthe effect of lowering segregation levels over time. However, one mightargue that their low segregation levels are related to their similar socioeco-nomic backgrounds. Among these groups, about 14–15% have completeduniversity education and about 25–30% are in the top quartile of individualincome.

Although the charter groups and western European groups have lowsegregation levels, they are less likely to share neighborhoods with otherEuropean groups. The dissimilarity indices between any of these groups withany other European groups are higher. The average pairwise segregationlevel is 0.51. Most members of other European groups arrived Canada laterthan the charter groups and western Europeans. There are also differences insocioeconomic resources between the earlier European groups and the otherEuropean groups. Thus, the data seem to support equally the arguments forthe effect of duration in the country and for the socioeconomic effects onsegregation levels.

11There are two reasons for setting up the analysis to compare the dissimilarity index of twospecific groups instead of comparing one specific group with all other racial groups (e.g., allEuropean groups). First, the setup (i.e., comparing two specific groups) is consistent withthe discussion throughout that we compared segregation levels of different specific racialand ethnic groups. Second, given the triangle inequality states |x + y| ≤ |x| + |y|, the actualdissimilarity index between visible minority groups and all European groups is not alwaysthe same as the average dissimilarity index of a specific visible minority group with a specificEuropean group.

Page 13: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 589

Tabl

eII

.U

neve

nD

istr

ibut

ion

Bet

wee

nE

urop

ean

Eth

nic

Gro

ups

inM

ajor

Can

adia

nM

etro

polit

anA

reas

,199

1

Bri

tish

Fren

chG

erm

anD

utch

Polis

hH

unga

rian

Ukr

aini

anPo

rtug

uese

Span

ish

Gre

ekIt

alia

nJe

wis

h

Bri

tish

—0.

288

0.20

30.

276

0.34

60.

371

0.29

70.

589

0.58

30.

545

0.41

70.

663

Fren

ch—

0.33

40.

393

0.41

40.

442

0.39

90.

572

0.57

80.

560

0.47

80.

746

Ger

man

—0.

292

0.36

20.

394

0.31

90.

612

0.59

90.

558

0.45

10.

673

Dut

ch—

0.42

50.

443

0.37

40.

639

0.65

30.

609

0.49

30.

693

Polis

h—

0.43

20.

332

0.57

70.

528

0.56

30.

452

0.68

3H

unga

rian

—0.

428

0.62

50.

609

0.56

70.

508

0.66

4U

krai

nian

—0.

604

0.59

00.

577

0.44

00.

685

Port

ugue

se—

0.59

10.

677

0.54

10.

825

Span

ish

—0.

649

0.59

10.

790

Gre

ek—

0.55

80.

667

Ital

ian

—0.

730

Jew

ish

—A

vera

ge0.

416

0.47

40.

435

0.48

00.

466

0.49

90.

459

0.62

30.

614

0.59

40.

515

0.71

1SD

0.16

70.

173

0.16

50.

158

0.13

10.

127

0.15

40.

097

0.08

40.

095

0.12

30.

100

Sour

ce:1

991

Can

adia

nC

ensu

sP

rofil

efil

e.

Page 14: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

590 Fong and Wilkes

Table II also demonstrates that the most recently arrived Europeangroups, such as Portuguese, Greek, and Spanish, have high levels of resi-dential segregation from all groups. The average segregation levels ofPortuguese, Spanish, and Greek with other European groups are 0.62, 0.61,and 0.59, respectively. The segregation levels of some groups, such as the seg-regation between Spanish and Dutch or between Greek and Portuguese, areabove 0.65. The pattern again seems to highlight the effect of length of timein the country on the residential patterns of groups. Groups recently arrivedin Canada may prefer to stay close together. This interpretation is furtherconfirmed by the general pattern of lower segregation levels for other Euro-pean groups, such as Polish and Ukrainian, most of whose members arrivedbefore the southern Europeans.

Finally, table also clearly demonstrates that Jews experience high levelsof segregation from other European groups. Their segregation levels with allother European groups are above 0.66. Nearly half of those levels are above0.69. At first glance, this peculiarly high residential concentration of Jewsmay reflect attempts by the Jewish community to maintain close physicalproximity. However, it is equally possible to argue that high residential seg-regation of Jews from other groups simply reflects their economic achieve-ments, which have resulted in residential differentiation between Jews andother groups. Of all the groups, Jews have the highest percentage of memberswith income over $60,000 and university completion.

The results in Table II suggest an optimistic picture, as the segregationlevels of European groups appear to be affected by duration in the countryor socioeconomic differences. In other words, if groups stay in the coun-try for a longer period of time and improve their socioeconomic resources,segregation will be reduced. However, the results only show the segrega-tion patterns among European groups. To fully appreciate the complexity ofrace and ethnic relations in contemporary Canadian cities, we need a morecomprehensive picture that considers the segregation patterns of Europeangroups with visible minority groups.

In general, the level of residential segregation between visible minoritygroups and Europeans is similar to the level of segregation between thecharter and western European groups and the southern European groups(see Table III). The residential segregation levels between visible minorityand European groups are moderately high: all are above 0.5 except threepairs. Some groups, such as Jews and Vietnamese, have averages above 0.7.The results seem to demonstrate the effect of length of stay in the countryon ethnic segregation levels. Most visible minority groups arrived in Canadain recent decades. The development of immigrant communities in majorCanadian cities, such as Toronto and Vancouver, may become an ecologicalforce drawing ethnic members together spatially. However, the high level of

Page 15: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 591

Table III. Uneven Distribution Between European Ethnic Groups and Visible Minority Groupsin Major Canadian Metropolitan Areas, 1991

Asian Indian Chinese Filipino Vietnamese Black Average SD

British 0.513 0.494 0.553 0.704 0.478 0.548 0.103French 0.585 0.557 0.599 0.702 0.502 0.589 0.122German 0.527 0.506 0.566 0.713 0.502 0.563 0.102Dutch 0.567 0.560 0.601 0.749 0.541 0.604 0.101Polish 0.545 0.516 0.545 0.652 0.483 0.548 0.092Hungarian 0.572 0.562 0.576 0.697 0.544 0.587 0.086Ukrainian 0.547 0.527 0.573 0.712 0.523 0.576 0.103Portuguese 0.629 0.600 0.622 0.660 0.585 0.619 0.093Spanish 0.625 0.574 0.625 0.623 0.552 0.600 0.074Greek 0.605 0.584 0.642 0.746 0.569 0.629 0.111Italian 0.563 0.513 0.601 0.695 0.514 0.577 0.103Jewish 0.742 0.681 0.718 0.851 0.759 0.750 0.091Average 0.585 0.556 0.601 0.709 0.546SD 0.089 0.089 0.107 0.096 0.096

Source: 1991 Canadian Census Profile file.

segregation between visible minority groups and European groups may alsobe related to racial discrimination in the housing market, which has beenwell documented in the literature (Henry, 1994).

The results presented in Table III also demonstrate two other uniqueresidential patterns. First, Jews highly segregated not only from other Eu-ropean groups, but also from visible minority groups. The average level ofresidential segregation of Jews from other visible minority groups is 0.75. Sec-ond, Vietnamese experience a higher level of residential segregation fromEuropean groups than do other visible minorities. Their average residentialsegregation level with European groups is 0.71, while the segregation lev-els between other visible minorities and European groups are 0.6 or below.The consistently higher residential segregation of Jews from visible minor-ity groups and of Vietnamese from European groups may be the result ofsocioeconomic differences between Jews and visible minority groups, and ofthe fact that most Vietnamese arrived in Canada under the refugee categoryin the 1970s with little financial capital.

It is also noteworthy that Blacks in Canada do not experience as high alevel of residential segregation as Blacks in the United States. The residentialsegregation patterns of Blacks with European groups are similar to those ofother visible minority groups. These results are consistent with other stud-ies on Canadian residential patterns, which reflect the distinct backgroundof Blacks in Canada (Fong 1996, 1997). About 60% of Blacks in Canadaare immigrants, many of whom entered Canada as independent class immi-grants with reasonable levels of education and skill. The patterns reflect someAmerican findings that the residential attainment of immigrant Blacks is

Page 16: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

592 Fong and Wilkes

Table IV. Uneven Distribution Between Visible Minority Groups in Major CanadianMetropolitan Areas, 1991

Asian Indian Chinese Filipino Vietnamese Black

Asian Indian — 0.489 0.499 0.714 0.500Chinese — 0.528 0.620 0.507Filipino — 0.679 0.528Vietnamese — 0.662BlackAverage 0.551 0.536 0.559 0.635 0.570SD 0.130 0.093 0.119 0.100 0.096

Source: 1991 Canadian Census Profile file.

better than that of American-born Blacks (Alba et al., 2000; Crowder, 1999).The segregation patterns of Blacks can also be related to their small pop-ulation size in Canadian cities. Reduced visibility has meant that the Blackpopulation has not experienced a high level of discrimination (Blalock, 1967;Blauner, 1972; Lieberson, 1980).

In short, the segregation patterns between European and visible minor-ity groups shown in Table III, like the segregation levels among Europeangroups, point to the possible significant effect on segregation patterns ofduration in the country and socioeconomic resources. However, the resultsalso suggest some other explanations, such as housing discrimination againstvisible minority groups.

Table IV compares the residential segregation levels among visible mi-nority groups, which are similar to the segregation levels between visibleminority and European groups. The average segregation level between vis-ible minority groups is between 0.54 and 0.64. The Vietnamese stand out asthe group with the highest level of residential segregation. Half of their lev-els of residential segregation from other visible minority groups are above0.6. These results suggest that Vietnamese not only experience a high levelof residential segregation from European groups but are also segregatedfrom other visible minority groups. Since Vietnamese in general have fewersocioeconomic resources than other visible minority groups, the results onceagain suggest the possible effect of socioeconomic resources on residentialpatterns.

The foregoing discussion provides a picture of residential patternsamong racial and ethnic groups in Canada. The charter and westernEuropean groups who arrived in the country earlier than other Europeangroups are more likely to share neighborhoods. Segregation levels are higherbetween these groups and more recently arrived European groups. Thesegregation of visible minority groups remains moderately high with allEuropean groups and other visible minority groups. To improve our

Page 17: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 593

understanding of residential segregation patterns in Canada, we developedmultivariate models to account for interurban differences.

MULTIVARIATE ANALYSIS

In this section, we present the results of a series of multivariate regres-sion analyses. Our aim is to study the relative effects of city demographiccomposition, city structural factors, and group economic differences on thevariation in residential segregation levels among groups, controlling for eth-nic groups included in the observation and whether the observation is inMontreal. We created a data set with 1471 observations for this analysis.Each observation is composed of the dissimilarity indices of two specificracial and ethnic groups in the 12 selected cities and the metropolitan in-formation associated with each pair of groups.12 For example, we created136 observations for Toronto. Each observation consists of the dissimilarityindex of any pair of groups in Toronto and all the information related toToronto’s demographic composition, structural factors, and variables of dif-ferences between the two groups in question. Therefore, we have included1471 observations comprise dissimilarity indices of each possible pairwisecomparison of the 17 racial and ethnic groups in each metropolitan area.Since the nature of the dissimilarity index is symmetrical, we did not includethe score of the dissimilarity index between A and B if the dissimilarity indexbetween B and A was already included.

Demographic composition includes two variables: the total city pop-ulation and the proportion of visible minority population in the city. Bothvariables are directly obtained from the Canadian census.

The structural context of the city is captured by three variables. First,the economy is measured by the city unemployment rate. Second, the ageof the city is approximated by the proportion of housing built after 1986. Fi-nally, a set of functional specialization variables is included to reflect varioustypes of city functions. In their original approach, Farley and Frey (1994)classified cities into different categories according to the dominant type ofjobs available. Instead, we have used the proportion of retirement, manu-facturing jobs, government jobs, and trade-related jobs in the city. This ap-proach avoids classification based on arbitrary criteria. In addition, in manycities more than one type of employment has a similarly large proportion of

12The setup of the analysis is such that the independent variables have the same value for allcomparisons within one city. The setup is fine as long as the solution of the inverse [X′X] isfound, because the estimated coefficient B is [X′X]−1 [X′Y]. The estimate is not related towhether the value is the same for all cases in one city.

Page 18: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

594 Fong and Wilkes

available jobs, and it is difficult to classify them into one specific functionalspecialization. Including a set of variables for different jobs available in thecity allows us to delineate the relative effects of the availability of differenttypes of jobs on residential segregation.

The third set of variables included in the model represents the economicdifference between two specific groups, measured by the difference in theaverage household income for each pair of groups within each city.13 Sincethe 2B Profile data file does not provide information on average householdincome by racial and ethnic group, we obtained the information from the1991 Public Use Micro Data file (PUMF). We took the absolute value of thedifferences because we were interested in the magnitude, not the directionof differences.14

In addition, we controlled for the possible effect of specific ethnic groupsand the city of Montreal, which reflect unique racial and ethnic relationsin Quebec, on levels of segregation. Previous studies have shown that theunique ethnic relations and the distinctive cultural and social setting ofQuebec distinguish its residential segregation patterns from those of otherCanadian cities (Fong, 1997).

Table V presents the multivariate analysis of how the demographic com-position of the city, the city’s structural contexts, and its group differencesaffect segregation levels among racial and ethnic groups, controlling for spe-cific race and ethnicity and the city of Montreal. For each model, we re-gressed the segregation index score between two racial and ethnic groupson the sets of variables discussed earlier. In other words, the analysis tells ushow these factors affect the level of spatial interaction among groups. Sincewe are interested in understanding residential segregation patterns as anindicator of the social relations among groups rather than as an indicator ofthe integration of minority groups, we did not just focus on the segregationpatterns between minority and majority groups. We ran separate modelsto delineate the effect of these factors on the segregation patterns amongEuropean groups, between European groups and visible minorities, andamong visible minority groups.

13Household income is the sum of the total incomes of all household members. Total incomeincludes income from various sources, including wages and salaries, government transfers,and investment income.

14One of the reviewers raised a concern about using actual income differences rather thanrelative income differences. The use of actual income differences is consistent with the evi-dence from the literature that residential segregation is related to income differences betweengroups. Relative income differences can have social and economic implications for neighbor-hood qualities, but not necessarily for the level of uneven spatial distribution of two groups.In addition, our use of income in absolute rather than relative terms is consistent with pre-vious studies on residential segregation (e.g., Massey and Denton, 1988; Frey and Farley,1996).

Page 19: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 595

Table V. Effects of City Demographic Composition, City Structural Context, and Group Dif-ference on Uneven Distribution of European and Visible Minority Groups in Selected CMAs

in Canada, 1991

European vs. European vs. European vs. Visible minority vs.European visible minority visible minority visible minority

City demographiccomposition

In (population) 0.069∗∗∗ 0.019 0.022 −0.014Proportion of visible −0.004∗∗ 0.000 0.000 −0.001

minorityCity structural contexts

Unemployment rate −0.008 −0.020∗∗ −0.020∗∗∗ −0.023Proportion of housing −0.007∗∗ −0.011∗∗∗ −0.011∗∗∗ −0.008

built after 1986City functional

specializationProportion of 0.002 0.001 0.001 0.000

manufacturing jobsProportion of 0.003 0.003 0.002 0.000

government jobsProportion of 0.005∗∗ 0.013∗∗∗ 0.015∗∗∗ 0.026∗

retired populationProportion of −0.003 0.003 0.000 −0.002

trade-related jobsGroup economic 0.000 −0.001∗∗∗ −0.001∗∗∗ 0.000

differenceRacial and ethnic

groupsBritish o.c. o.c.French 0.047 −0.002German 0.000 −0.032Italian 0.091∗∗ −0.017Ukrainian 0.027 −0.019Dutch 0.060∗ 0.017Polish 0.036 −0.043∗∗Portuguese 0.209 0.028Jewish 0.303∗∗∗ 0.155∗∗∗Greek 0.185∗∗∗ 0.035∗Hungarian 0.072∗∗ −0.002Spanish 0.202∗∗∗ 0.011Black o.c. o.c.Chinese 0.002 −0.037Indian 0.051∗∗∗ 0.007Filipino 0.043∗∗∗ 0.002Vietnamese 0.158∗∗∗ 0.143∗∗∗

Montreal 0.049 0.125∗∗ 0.126∗∗ 0.140Intercept −0.502∗∗ 0.336 0.325 0.633N 717 647 647 107Adjusted R2 0.711 0.325 0.418 0.641

Note. o.c. = omitted category. Source: 1991 Canadian Census Profile file.∗ p < 0.10; ∗∗ p < 0.05; ∗∗∗ p < 0.01.

Page 20: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

596 Fong and Wilkes

The first model includes the segregation scores for only two Europeangroups. Since there are 12 European groups in the analysis, there are 11 com-parisons for each city. Segregation levels between two European groups arerelated moderately to both demographic factors included in the analysis.They are higher when there is a larger population and a smaller visible mi-nority population in the city. The significant effect of visible minority pop-ulation on the residential patterns of European groups is noteworthy. Theresults suggest that the presence of visible minorities may act as a catalystthat draws different European groups together.

None of the city structural contexts except the proportions of new hous-ing and retired people are related to segregation levels among Europeangroups. Even when these proportions are statistically significant, the effectsare minimal. In other words, their residential patterns are relatively stabledespite changes in city structures. In addition, economic differences betweentwo European groups are not related to their residential segregation level.The results indicate that economic disadvantage does not affect residentialsegregation levels among European groups.

Finally, the set of ethnic dummy variables suggests that most recentEuropean immigrant groups, such as the Portuguese, Greek, and Spanish,have higher segregation levels than the British, the reference group in themodel controlling for other factors. The results reinforce the observationreported in Table I. They also confirm that Jews are more likely to be segre-gated from other European groups.

The second and third models show the effects of city variables on seg-regation levels between European and visible minority groups. The resultsreflect how various factors affect the sharing of neighborhoods betweenEuropean and visible minority groups, which in turn may affect their contactopportunity. The second column reflects the model that includes only the setof dummy variables that indicate European groups. The third column showsthe results for the model that includes only the set of dummy variables thatindicate visible minority groups. We ran two separate models of segregationlevels between European and visible minority groups because of the uniquesetup of the analysis. Since we are interested in the segregation between Eu-ropean and visible minority groups, each European group has five possiblecombinations with visible minority groups (e.g., French vs. Chinese, Frenchvs. Blacks, French vs. Indian, French vs. Filipino, and French vs. Vietnamese).As the pattern repeats for each European group, any four dummy variablesof visible minority groups are enough to define the remaining visible minor-ity group. If all five visible minority groups are included, linear combinationof one visible minority will occur. A similar condition also applies to the Eu-ropean groups. We can run a model with one visible minority group and oneEuropean group omitted from the analysis. However, this approach makes

Page 21: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 597

the set of racial and ethnic variables difficult to interpret because the omittedgroup is a combination of the European and visible minority groups.

Therefore, we ran two separate models. One model includes city de-mographic factors, city structural contexts, group difference variables, anda set of dummy variables to indicate various European groups, along with adummy variable to identify the city of Montreal. The other model consists ofthe same set of variables, but replaces a set of dummy variables indicating var-ious European groups with another set of dummy variables to identify visibleminority groups. With these two separate models, we can delineate the effectsof various factors on the segregation between European and visible minoritygroups. Without controlling for visible minority groups, variable estimates inModel 2 absorb part of the effect of visible minority groups. The setup allowsus to see the combined effects of the variable and the visible minority groupson the segregation level between European and visible minority groups. Sim-ilarly, without controlling for European groups, variable estimates in Model3 show us the combined effects of the variables and the European groups.

The results of the Model 2 of segregation levels between European andvisible minority groups, controlling for European groups, suggest that thereare statistically significant effects from the city structural contexts and groupeconomic differences, but not from the city demographic factors. Unlike theprevious model of segregation levels among European groups, in the Model2 neither the size of the city population nor the size of the visible minoritypopulation is related to the segregation levels between European and visibleminority groups when controlling for European groups.

In addition, the results show that a number of city structural contextsare related to the segregation levels between European and visible minoritygroups. The unemployment rate, proportion of new housing, and propor-tion of retired population are all related to the segregation levels betweenEuropean and visible minority groups. As well, differences in income be-tween two groups are related to the segregation levels. However, it is note-worthy that the effect of the difference in income between two groups is weakand is opposite to the expected direction. The results may reveal the conse-quences of the housing policies in Canada in reducing the effect of income onresidential locations. Since the mid-1970s, housing policies have reflected thepolicy of “social mix.” Locations of housing provided for poor householdsin the public and nonprofit sectors were planned to mix with other incomegroups to avoid large-scale concentration of poverty (Van Dyk, 1995).

The set of European ethnic variables in the model suggests that thesegregation levels of most European groups, except those of the Poles, Jews,and Greeks, are not statistically different from those of the British. Amongthe three groups, two are recent European immigrant groups. It may be thatboth recent European immigrant groups and visible minority groups have

Page 22: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

598 Fong and Wilkes

their own ethnic communities because of a large immigrant population, andthat they seldom interact spatially.

Taken together, the results suggest that segregation levels between vis-ible minorities and European groups are affected by city structural contexts,controlling for the possible effects of specific European groups. Since visibleminority groups are not controlled, the segregation levels are related to thecity structural contexts, group differences, and the possible indirect effect ofvisible minority groups through these variables.

The third column shows the segregation levels between European andvisible minority groups, controlling for visible minority groups. The effects ofcity demographic composition, city structural factors, and group economicdifferences are similar to those in the model controlling for European groups.The demographic composition is unrelated to the segregation levels betweenEuropean and visible minority groups even when the visible minority groupsare controlled. In addition, as in the previous model, the unemployment rate,proportion of housing built after 1986, and proportion of retired people allaffect the segregation levels. The group economic differences still main-tain a minimal effect. Since the visible minority groups are controlled in theModel 3, the results further confirm those obtained from the previous model.The final set of visible minority variables shows that, except for the Chinese,the segregation patterns of all groups are statistically different from thosefor Blacks.

The final column shows the regression results of the model among visi-ble minority groups. Blacks are the reference group for the racial and ethniccategories. None of the city demographic composition or city structural con-texts, except for the proportion of the retired population, are significant tothe segregation patterns among visible minority groups. The results suggestthat the segregation patterns of visible minority groups are not related tomost factors suggested in the literature, which are mainly drawn from the ex-periences of European groups. Among all groups, only the Vietnamese standout as having a higher segregation level than other visible minority group.

CONCLUSION

This is a study of how city factors affect racial and ethnic residential seg-regation in contemporary multiethnic cities. To disentangle the complexityof the residential segregation of racial and ethnic groups, we followed theapproach developed by White et al. (2003) in order to perform a multivariateregression analysis for 17 groups in 12 metropolitan areas.

The results suggest that the analysis of racial and ethnic residentialsegregation from literature largely based on European experiences in the

Page 23: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 599

beginning of the last century and uniformly applied to different groupsin Canadian cities today. Our findings from Canadian cities suggest thatthe residential segregation of different racial and ethnic groups is affectedby different sets of city factors. Levels of residential segregation betweenEuropean groups are strongly related to city demographic factors, while res-idential segregation patterns between European groups and visible minoritygroups are affected by city structural factors.

The results also demonstrate that the explanatory power of the factorssuggested in the literature is reduced substantially with regard to the segre-gation patterns of visible minority groups. Residential segregation patternsamong visible minority groups are not related to most of the factors suggestedin the literature. Since most theories of racial and ethnic residential patternshave been developed from the earlier experience of European ethnic groupsor Black–White relationships, they may not help to explain the experience ofvisible minority groups who have distinctive cultural backgrounds and whoare attempting to integrate into cities that have economic and ecologicalstructures different from those faced by European immigrants in previousdecades.

The results lead to two possible view of residential segregation amongracial and ethnic groups in Canada. First, the results portray an optimisticpicture of residential segregation among European groups. As the visi-ble minority population increases, their residential segregation levels de-cline. Since the effects are actually a recent dominant demographic trendin Canadian society, the results imply that spatial integration among allEuropean groups is possible.

Second, the segregation levels of European and visible minority groupsare strongly related to city structural contexts. City unemployment rates,the proportion of new housing built, and city functional specialization arelarger social forces that will not be easily altered in the near future. Resi-dential segregation of European and visible minority groups is less likely tobe reduced in the foreseeable future. In other words, the sharing of neigh-borhoods between European and visible minority groups, which may fosterinterracial relations, is less likely to improve. This conclusion is further re-inforced by the similar results obtained when controlling for the indirecteffects of European or visible minority groups.

In short, this first systematic, multivariate analysis of racial and ethnicsegregation in Canadian cities reveals three distinctive patterns of how cityfactors affect segregation. It confirms that racial and ethnic segregation pat-terns, which reflect racial and ethnic relations, are complicated in Canadiancities as the country has undergone tremendous changes. The study suggestsa need for a wider perspective, perhaps including variables to capture othergroup differences (e.g., degree of cultural differences) that have rarely been

Page 24: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

600 Fong and Wilkes

discussed. Finally, we hope that the new approach undertaken in this studycan allow us to understand residential segregation at the metropolitan levelwith multivariate perspective and open new avenues for us to further disen-tangle racial and ethnic relations in Canadian cities.

ACKNOWLEDGMENTS

The research was supported by a research grant from the Social Scienceand Humanities Research Council of Canada.

REFERENCES

Alba, Richard, John R. Logan, and Brian J.Stults2000 “The changing neighborhood contexts

of the immigrant metropolis.” SocialForces 79:587–621.

Anderson, Alan, and James Frideres1981 Ethnicity in Canada: Theoretical Per-

spectives. Toronto: Butterworths.Avery, Donald H.1995 Reluctant Host: Canada’s Response

to Immigrant Workers: 1896–1994.Toronto: McClelland & Stewart.

Avery, Donald H., and J. K. Fedorowitz1982 The Poles in Canada. Ottawa:

Canadian Historical Association.Badets, Jane, and Tina W. L. Chui1994 Canada’s Changing Immigrant Popu-

lation. Toronto: Prentice Hall Canada.Balakrishnan, T. R.1976 “Ethnic residential segregation in the

metropolitan areas of Canada.” Cana-dian Journal of Sociology 1:481–498.

1982 “Changing patterns of ethnic residen-tial segregation in the metropolitan ar-eas of Canada.” Canadian Review ofSociology and Anthropology 19(1):92–110.

Balakrishnan, T. R., and John Kralt1987 “Segregation of visible minorities

in Montreal, Toronto and Vancou-ver.” In Leo Driedger (ed.), EthnicCanada: Identities and Inequalities:138–157.Toronto: Copp Clark Pitman.

Balakrishnan, T. R., and K. Selvanathan1990 “Ethnic residential segregation in

metropolitan Canada.” In Shiva S.Halli, Frank Trovato, and Leo Drei-

dger (eds.), Ethnic Demography: 393–413. Ottawa: Carleton UniversityPress.

Blalock, Hubert M., Jr.1967 Toward a Theory of Minority-Group

Relations. New York: Wiley.Blauner, Bob1972 Racial Oppression in America. New

York: Harper & Row.Bolaria, B. Singh, and Peter S. Li1985 Racial Oppression in Canada.

Toronto: Garamond Press.Breton, Raymond1964 “Institutional completeness of ethnic

communities and personal relationsof immigrants.” American Journal ofSociology 70:193–205.

Coleman, William1984 The Independence Movement in Que-

bec, 1945–1984. Toronto: University ofToronto Press.

Crowder, Kyle D.1999 “Residential segregation of west In-

dians in the New York/New Jerseymetropolitan area: The roles of raceand ethnicity.” International MigrationReview 33:79–113.

Darroch, A. Gordon, and Wilfred G. Marston1971 “The social class basis of ethnic res-

idential segregation: The Canadiancase.” American Journal of Sociology77:491–510.

Denis, Wilfrid1999 “Language policy in Canada.” In

Peter Li (ed.), Race and Ethnic Re-lations in Canada: 178–216. Toronto:Oxford University Press.

Page 25: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

Racial and Ethnic Residential Patterns in Canada 601

Driedger, Leo1996 Multi-Ethnic Canada: Identities and

Inequalities. Toronto: Oxford Univer-sity Press.

Farley, Reynolds, and William H. Frey1994 “Changes in the segregation of Whites

from Blacks during the 1980s: Smallsteps toward a more racially integratedsociety.” American Sociological Re-view 59:23–45.

Fischer, Claude S.1975 “Toward a subcultural theory of urban-

ism.” American Journal of Sociology80:1319–1341.

Fong, Eric1996 “A comparative perspective of racial

residential segregation: American andCanadian experiences.” SociologicalQuarterly 37:501–528.

1997 “Residential proximity with the char-ter groups in Canada.” Canadian Stud-ies in Population 24:103–124.

Fong, Eric, and Kumiko Shibuya2000 “Spatial separation of the poor in

canadian Cities.” Demography 37:449–459

Fong, Eric, and Rima Wilkes1999 “The spatial assimilation model reex-

amined: An assessment by Canadiandata.” International Migration Review33:594–620.

Fong, Eric, and Wsevolod W. Isajiw2000 “Determinants of friendship choices in

multiethnic society.” Sociological Fo-rum 15:249–271.

Frey, William H., and Reynolds Farley1996 “Latinos, Asian, and Black segregation

in U.S. metropolitan areas: Are mul-tiethnic metros different?” Demogra-phy 33:35–50.

Gerus, Oleh W., and Rea, James Edgar1985 The Ukrainians in Canada. Ottawa:

Canadian Historical Association.Henry, Frances1994 The Caribbean Diaspora in Toronto:

Learning to Live with Racism.Toronto: University of TorontoPress.

Isajiw, Wsevolod W.1990 “Ethnic-identity retention.” In

Raymond Breton, Wsevolod W. Isajiw,Warren E. Kalbach, and Jeffrey G.Reitz (eds.), Ethnic Identity andEquality: 34–91. Toronto: Universityof Toronto Press.

Jackman, Mary R., and Marie Crane1986 “Some of my best friends are Black’:

Interracial friendship and Whites’racial attitudes.” Public Opinion Quar-terly 50:459–486.

Johnston, Hugh1984 The East Indians in Canada. Ottawa:

Canadian Historical Association.Kalbach, Warren E.1987 “Growth and distribution of Canada’s

ethnic populations, 1871–1981.” In LeoDriedger (ed.), Ethnic Canada: Identi-ties and Inequalities: 82–110. Toronto:Copp Clark Pitman.

1990 “Ethnic residential segregation andits significance for the individualin the urban setting.” In RaymondBreton, Wsevolod Isajiw, Warren E.Kalbach, and Jeffrey G. Reitz (eds.),Ethnic Identity and Equality: 92–129. Toronto: University of TorontoPress.

Krysan, Maria, and Reynolds Farley2002 “The residential preferences of Blacks:

Do they explain persistent segrega-tion.” Social Forces 80: 937–980.

Langberg, Mark, and Reynolds Farley1985 “Residential segregation of Asian

Americans in 1980.” Sociology and So-cial Research 70:71–75.

Li, Peter1998 Chinese in Canada. Toronto: Oxford

University Press.Lieberson, Stanley1970 Language and Ethnic Relations in

Canada. Toronto: Wiley.1980 A Piece of the Pie. Berkeley: Univer-

sity of California Press.Logan, John R., and Harvey L. Molotch1987 Urban Fortunes: The Political Econ-

omy of Place. Berkeley: University ofCalifornia Press.

Massey, Douglas S.1985 “Ethnic residential segregation: A the-

oretical synthesis and empirical re-view.” Sociology and Social Research69:315–350.

Massey, Douglas S., and Nancy A. Denton1987 “Trends in the residential segregation

of Blacks, Hispanics, and Asians: 1970–1980.” American Sociological Review52:802–825.

1988 “Suburbanization and segregation inU.S. metropolitan areas.” AmericanJournal of Sociology 94:592–626.

Page 26: Racial and Ethnic Residential Patterns in Canada

P1: GXB

Sociological Forum [sofo] pp1028-sofo-475163 October 22, 2003 18:27 Style file version June 4th, 2002

602 Fong and Wilkes

1992 “Residential segregation of Asian-origin groups in U.S. metropolitan ar-eas.” Sociology and Social Research76:170–177.

Min, Pyong Gap1995 “An overview of Asian Americans.”

In Pyong Gap Min (ed.), Ameri-cans: Contemporary Trends and Is-sues: 10–37. Thousand Oaks, CA:Sage.

Prymak, Thomas M.1991 “Recent scholarship on polyethnic em-

igration from the Republic of Poland toCanada between the wars.” CanadianEthnic Studies 23:58–70.

Ray, D. Michael1996 “Employment creation by small firms.”

In John N. H. Britton (ed.), Canadaand the Global Economy: The Geog-raphy of Structural and Technologi-cal Change: 175–194. Kingston, ON:McGill-Queen’s University Press.

Rutherford, Tod1996 “Socio-spatial restructuring of Cana-

dian labour markets.” In John N.H. Britton (ed.), Canada and theGlobal Economy: The Geography ofStructural and Technological Change:407–432. Kingston, ON: McGill-Queen’s University Press.

Salee, Daniel, and William Coleman1997 “The challenges of the Quebec

question: Paradigm and counter-paradigm.” In Wallace Clement (ed.),Understanding Canada: Building onthe New Canadian Political Econ-omy: 262–285. Kingston, ON: McGill-Queen’s University Press.

Sassen, Saskia1991 The Global City: New York: Princeton

University Press.Sigeleman, Lee, Timothy Bledsoe, SusanWelch, and Michael W. Combs1996 “Making contact? Black–White so-

cial interaction in an urban setting.”American Journal of Sociology101:1306–1332.

Statistics Canada1993a Census of Canada 1991: Profile

Files—Part A. Ottawa: StatisticsCanada.

1993b Census of Canada 1991: ProfileFiles—Part B. Ottawa: StatisticsCanada.

1993c Census of Canada 1991: Public UseMicrodata File. Ottawa: StatisticsCanada.

Taylor, K. W.1991 “Racism in Canadian immigration pol-

icy.” Canadian Ethnic Studies 23:1–20.Van Dyk, Nick1995 “Financing social housing in Canada.”

Housing Policy Debate 6:815–848.White, Michael J., Eric Fong, and Qian Cai2003 “The segregation of Asian-origin

groups in the United States andCanada.” Social Science Research32:148–167.

Wilson, William Julius1987 The Truly Disadvantaged: Essays on

Inner City, the Underclass, and PublicPolicy. Chicago: University of ChicagoPress.

1996 When Work Disappears: The Workof the New Urban Poor. New York:Knopf.