The Dynamics of Racial Residential Segregation Author(s): Camille Zubrinsky Charles Source: Annual Review of Sociology, Vol. 29 (2003), pp. 167-207 Published by: Annual Reviews Stable URL: http://www.jstor.org/stable/30036965 Accessed: 17/02/2009 09:36 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=annrevs. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org. Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review of Sociology. http://www.jstor.org Annu.Rev.Sociol.2003.29:167-207 doi:10.1146/annurev.soc.29.010202.100002 © 2003by AnnualReviews.All rightsreserved Copyright onlineas a Reviewin AdvanceonJune4, 2003 Firstpublished THE DYNAMICS OF RACIAL RESIDENTIAL SEGREGATION CamilleZubrinskyCharles Philadelphia, Pennsylvania of Sociology,University Department of Pennsylvania, 19104-6299;email:ccharles@pop.upenn.edu Key Words race,racerelations,urban/racial inequality,stratification U Abstract The publicationof AmericanApartheid(Massey& Denton1993)was influentialin shiftingpublic discourseback towardracialresidentialsegregationas fundamentalto persistingracialinequality.At the end of the twentiethcentury,the majorityof blacksremainedseverelysegregatedfromwhites in majormetropolitan areas.Due to the persistenceof high-volumeimmigration,HispanicandAsiansegregationfromwhites has increased,althoughit is still best characterizedas moderate. This reviewexaminestrendsin the residentialsegregationof blacks,Hispanics,and the causesof persistingsegregaAsiansandrecentresearchfocusedon understanding tion. This discussionis organizedaroundtwo broadtheoreticalperspectives-spatial assimilationandplace stratification. Afterdetailingthe consequencesof segregation andgoals for futureresearch. for affectedgroups,I identifygapsin ourunderstanding INTRODUCTION Sociologists andpolicymakershave long viewed racialresidentialsegregationas a key aspectof racialinequality,implicatedin both intergrouprelationsandin larger processesof individualandgroupsocial mobility.At the dawnof the twentiethcentury,Du Bois (1903) recognizedthe importanceof neighborhoods-the "physical proximityof home and dwelling-places,the way in which neighborhoodsgroup themselves, and [their] contiguity"-as primarylocations for social interaction, lamentingthat the "colorline" separatingblack and white neighborhoodscaused each to see the worst in the other (1990, pp. 120-21). Indeed, studentsof racial inequality,from Myrdal (1944) to Taeuber& Taeuber(1965), believed that segregationwas a major barrierto equality,assertingthat segregation"inhibitsthe developmentof informal,neighborlyrelations,""ensuresthe segregationof a variety of public and privatefacilities"(Taeuber& Taeuber1965, p. 1), and permits prejudice"tobe freely vented on Negroes withouthurtingwhites"(Myrdal1944, p. 618). Moreover,residentialsegregation"underminesthe social and economic well-being"irrespectiveof personalcharacteristics(Massey& Denton 1993, pp. 23). Whethervoluntaryor involuntary,living in racially segregatedneighborhoods has seriousimplicationsfor the presentand futuremobility opportunitiesof those 036030572/03/0811-0167$14.00 167 168 CHARLES who are excluded from desirable areas. Where we live affects our proximity to good job opportunities,educationalquality,and safety fromcrime (both as victim and as perpetrator),as well as the qualityof our social networks(Jargowsky1996, Wilson 1987). By the late 1960s, unrestin urbanghettos across the countrybroughtresidential segregation-and its implicationin racialinequality-to the public's attention, leadingto the now famousconclusionof the KernerCommissionthatAmericawas "movingtowardtwo societies, one black, one white-separate andunequal"(U.S. National Advisory Commission on Civil Disorders 1988) and the passage of the Fair Housing Act in 1968. In additionto ending legal housing marketdiscrimination, passage of the Fair Housing Act markedthe end of public discussion of legislationwas the residentialsegregation,as manybelievedthatantidiscrimination beginningof the end of residentialsegregation.Withlegal barriersto educational, occupational,and residentialopportunitiesremoved,blacks could finally achieve full-fledgedintegration,and social scientists,politicians,andthe generalpublicignoredthis dimensionof the color line for the next two decades(Massey & Denton 1993, Meyer 2000). By the late 1970s, conditions in the nation's urbanareaswhere the majorityof blacks were still concentrated-had declined precipitously. Social scientists scrambledto explainthe emergenceof a disproportionatelyblack urbanunderclass,paying little or no attentionto persistingresidentialsegregation by race. In The TrulyDisadvantaged(1987), Wilson outlinedthe most widely accepted theory of urban poverty: Geographicallyconcentratedpoverty and the subsequent developmentof a ghetto underclassresultedfrom structuralchanges in the economy combined with the exodus of middle- and working-class black families from many inner-city ghetto neighborhoods.The shift from a goods- to a service-producingeconomy saw huge declines in the availabilityof low-skilled manufacturingjobs thatpaid enough to supporta family; owing to past discrimination, blacks were disproportionatelyconcentratedin these jobs and therefore sufferedmassive unemployment.Having benefitedmore substantiallyfrom civil rights gains thatincludedaffirmativeaction policies as well as antidiscrimination legislation, Wilson argued,middle- and working-classblacks were able to take advantageof residentialopportunitiesoutside of the ghetto. The impact of these events was an "exponentialincrease"in the now well-known social dislocations associatedwith suddenand/orlong-termincreasesin joblessness-under- andunemployment,welfare dependence,out-of-wedlockbirths,and a blatantdisregard for the law. The emigrationof nonpoorblacks, Wilson argued,removed an important"social buffer,"leaving poor blacks in socially isolated communitiesthat lacked materialresources, access to jobs andjob networks,exposure to conventional role models, and therefore"generate[d]behavior not conducive to good work histories"(Wilson 1987, pp. 56-60).1 seeWilson(1987)andMassey 'Foranoverviewof otherpopulartheoriesof theunderclass, & Denton(1993);for detaileddiscussionsof the characteristics poverty of concentrated see Wilson(1987),Massey& Denton(1993),andJargowsky neighborhoods, (1996). DYNAMICSOF RESIDENTIALSEGREGATION 169 Massey & Denton (1993) show, however,thatwithoutresidentialsegregation, these "structuralchanges would not have producedthe disastroussocial and economic outcomesobservedin innercities... Althoughrates of black povertywere drivenup by the economic dislocationsWilson identifies, it was segregationthat confined the increased deprivationto a small numberof densely settled, tightly packed, and geographicallyisolated areas."Retooling existing theories of urban poverty,they argue,resolves unansweredquestionsregardingthe disproportionate representationof blacksandPuertoRicansin the ranksof the underclass,as well as the concentrationof underclasscommunitiesin older,largercities of the Northeast and Midwest. In the largesturbanareas,blacks and PuertoRicans were the only groupsto experienceextremeresidentialsegregationand steep rises in povertyat the same time, the latter stemming from the fact that areas of black concentration were also hit especially hardby the economic reversalsof the 1970s (Massey & Denton 1993, pp. 146-47).2 Emphasizingthe interactionof segregationand rising povertyalso furthersour understandingof the inability of nonpoorblacks to escape segregationand its consequences, despite increasingclass segregation withinblackcommunities(Jargowsky1996;Massey & Denton 1993, pp. 146-47). Focusing on a black middle-classexodus, they argue,detractsattentionfrom the devastatingconsequencesof residentialsegregationfor all blacks, irrespectiveof socioeconomic status. The publicationof AmericanApartheid(Massey & Denton 1993) was singularly influentialin shifting public discourse "back to issues of race and racial segregation"as "fundamentalto... the statusof black Americansand the origins of the urbanunderclass."The book arguedpersuasivelythat "themissing link"in each of the underclasstheoriesprevalentat the time was "theirsystematicfailureto considerthe importantrole thatsegregationhas playedin mediating,exacerbating, and ultimatelyamplifyingthe harmfulsocial and economic processes they treat" (Massey & Denton 1993, p. 7). As a result, social scientists have rediscovered racialresidentialsegregationas a constituentfactorin persistentracialinequality in the United States. Recent researchaddressesseveralkey issues, including the following: (a) trendsin the residentialsegregationof racial/ethnicgroups,(b) factorsthatinfluencethe spatialdistributionof groups,and(c) the social andeconomic consequencesof segregation. This review addresseseach of these issues. I begin with a summaryof trends in the residentialsegregationof blacks, Hispanics, and Asians from whites since 1980. Despite declines in black-whitesegregation,blacks remainseverely segregated in the majorityof U.S. cities. As a result of massive immigration,Hispanic and Asian segregationfrom whites is on the rise; but except for a small number of cases among Hispanics, both groups still remain only moderatelysegregated from whites. Following the discussion of trends in segregation,I review recent literaturededicatedto understandingthe causes of residentialsegregation.Two 2Massey&Denton(1993)andothers(see,forexample,Loganet al. 1996)arguethatgroups with obviousAfricanphenotypeare similarlydisadvantaged, explainingthe divergence betweenblackandwhiteHispanics. 170 CHARLES broad theoreticalperspectives shape this discussion and are indicative of ongoing sociological-indeed, societal-debates regardingthe relative importanceof race and class in determiningsocial outcomes. The spatial assimilation model posits that objectivedifferencesin socioeconomic statusand acculturationacross racial/ethnicgroupsareprimarilyresponsiblefor residentialsegregation,squarely addressingthe issue of social mobility in its suggestionthat increasededucation, occupationalprestige,andincome will lead to greaterracialresidentialintegration. This explanationadequatelydescribesthe residentialmobility of both phenotypically white Hispanicsand of Asians. Alternatively,the place stratificationmodel emphasizes the persistence of prejudiceand discrimination-key aspects of intergrouprelations-that act to constrainthe residentialmobility options of disadvantagedgroups, including supraindividual,institutional-levelforces.3 Available evidence suggests that this model bettercharacterizesthe inability of those who arephenotypicallyblack (bothAfricanAmericansandblack Hispanics)to escape segregation.At first glance, these perspectives may appearoppositional. Upon closer inspection, however, these seemingly oppositional explanationscomplement one another.Race still matters;however,its relativeimportance-and that of socioeconomic status-depends on group membership.Finally, I end with a discussion of the consequences of residentialsegregation,followed by a discussion of the currentstateof knowledgeregardingthe dynamicsof racialresidential segregation,including suggestions and/orefforts to alleviate segregationand its consequences. TRENDS IN RESIDENTIAL SEGREGATION, 1980-2000 Blacks in 16 metropolitanareas were hypersegregatedfrom whites in 1980, exhibiting extremeisolation on at least four of five standardmeasuresof residential distribution(Massey & Denton 1989).4By 1990, thatnumberhad nearlydoubled: in the emergenceandmaintenance 3Theroleof the federalgovernment of raciallysegreto discriminatory is fromFHAmortgage zoningpractices, gatedneighborhoods, guarantees well documented (seemostrecentlyMeyer2000).Morethan30 yearssincethesediscriminatorypracticeswerebanned,however,theireffectsareevidencedby massiveblack-white disparities in accumulated wealth(Oliver& Shapiro1995). 4Studiesof residentialsegregationgenerallyrely on one or moreof five measures,each of whichcapturesa differentdimensionof the spatialdistribution of groups.Evennessmeasuredas the indexof dissimilarity-describes the degreeto whicha groupis evenly distributed acrossneighborhoods or tracts.A scoreover60 is interpreted as extremesegregationbetweentwo groups,indicatingthe percentageof eithergroupthatwouldhave to moveto anothertractto achievewithin-tract thatmirrorthat populationdistributions of the metroarea.Isolation-measuredas (P*,)-is interpreted as thepercentthatis the ortract;scoresof 70 andoverare sameracein theaveragegroupmember'sneighborhood extreme(indicating thattheaveragepersonlivesin anareathatis 70%thesame considered as theaverageprobability of race).Theinverseof isolationis Exposure(P*,), interpreted DYNAMICSOF RESIDENTIALSEGREGATION 171 In 29 U.S. metropolitanareas-containing 40% of the total black populationblacks experienced"extreme,multidimensional,andcumulativeresidentialsegregation"(Denton1994, p. 49). Blacks areuniquein this experience,which contrasts sharply with the limited and temporary segregation experienced by other groups (Denton 1994, Massey & Denton 1993). Hispanics and Asians are only moderately segregated from whites, although their levels of segregation and isolation are increasing as a result of continuous, high-volume immigration since 1970.5 Preliminary data from the 2000 Census (Logan 2001a) documents nationwide increases in the relative size of the Hispanic and Asian populations since 1980 and declines in the relative size of the white population; the relative size of the black population changed little (an average of 1.5%).6 With no end to immigration in sight, non-Hispanic whites are projected to become a numerical minority in the United States some time during this century (Edmonston & Passel 1992, Massey 1995), and the trend is well underway: 8 of the 50 largest metro areas are already majority-minority(whites are less than half the total population)and two others will be majority-minority by the 2010 Census.7 Thesecompositionalshiftsinfluenceresidentialsegregationin meaningfulways. Isolation is generally low for small groups but is expected to rise with increasing contact with a person-of-another-racecomparisongroup (usually whites). These are the most commonlyreportedmeasures.Threeothermeasures-concentration, clustering,and centralization-address a group's degree of density,proximityto the centralbusiness district, and the contiguity of their neighborhoods,respectively.A group is hypersegregated if it scores over 60 on at least four of the five measures(Denton 1994; Massey & Denton 1989, 1993). 5Accordingto official estimates, nearly 85% of the 15.5 million immigrantsto the United States between 1971 and 1993 are of Latin American or Asian origin (roughly 50% and 35%, respectively);includingestimatesof undocumentedor illegal immigrantspushes the total up by at least another3 million, the majorityof which are largely Mexican (Massey 1995). 6Data are for the 50 largest metropolitan areas. Due to space limitations, these data are not presented here but are available from the Lewis Mumford Center at The relative share of the Hispanic and Asian populawww.albany.edu/mumford/census. tions increasedan averageof 6.8% and 3.8%, respectively.Eleven regions-Los Angeles, Riverside-SanBemardino,OrangeCounty,and San Diego in California;HoustonandDallas in Texas;Phoenix-Mesain Arizona;Miami, Orlando,and Fort Lauderdalein Florida; and Bergen-Passaicin New Jersey-saw theirHispanicpopulationsincreasebetween 10% and 22%. Increases for Asians topped 10% in four areas in California(OrangeCounty, Oakland,San Francisco, and San Jose), and five other areas increased between 6% and 10% (Los Angeles-Long Beach and Sacramento,Seattle-Bellvue-Everettin Washington, New YorkCity, and Bergen-Passaic).On average,the relativesize of the white population declined by 12.4%. For blacks, the patternis more varied;there is little or no growth in many areasand others show slight declines. 7Regionsthatare currentlymajority-minorityare:Los Angeles, Riverside-SanBemardino, Oakland,and San Jose; Houstonand San Antonio; Miami; and New YorkCity. Two other regions-Orange County and San Francisco-are currentlyjust over 51% white. 172 CHARLES group size even if the group's level of segregationremainsconstant. Moreover, the largerthe relative size of an out-group'spopulation,the greaterexposure to that group is likely to be. Both exposure and isolation are influenced by group settlementpatterns.Specifically, chain migrationpatternscommon among both HispanicandAsianimmigrantsconcentraterapidlygrowinggroupsin a smallnumber of metropolitanareas-and within a small numberof neighborhoodswithin an area-increasing theirisolation and decreasingexposureto out-groups(Logan 2001a, Massey & Denton 1987). Table 1 reportsblack, Hispanic, and Asian segregationfrom whites (dissimilarity),isolation, andexposureto whites for the 50 largestmetropolitanregions in 2000 (and parenthetically,the change between 1980 and 2000).8 Both Hispanics and Asians show increasingsegregationand isolation, along with declining exposure to whites. These patternsare consistent with their rapidpopulationgrowth, settlementpatterns,and declining white populationshare.Over the same period, blacks show declinesin bothsegregationandisolation;trendsin exposureto whites are mixed, but overall reflect a slight increase.These patternsare consistent with the shifts in populationcomposition outlined above and their anticipatedeffects on spatial distribution.These changes also contributedto declining black isolation: In many instances,Hispanicsettlementpatternsconcentratethem in areasof traditionalblack settlement,increasingblack-Hispaniccontact(Alba et al. 1995).9 Nearlyhalf of themetroregionsexperienceddeclinesin black-whitesegregation of at least 10 percentagepointsoverthe 20-yearperiod.10Still, the degreeof blackwhite segregationremainsextreme(over60) in 28 regions.More thanhalf of these areEasternandMidwesternregions,whereblack-whitesegregationhas been most resistantto change(Farley& Frey 1994, Massey & Denton 1993), andmanyof the most segregatedregions saw little or no change in black-whitesegregationover the two decades.At the same time, segregationdeclinedenoughin some mid-sized regions with sizableblackpopulationsto be characterizedas moderate(under50). Areas with the largestdeclines (15% or more) tend to be multiethnic(an aboveaverage presence of at least one other nonwhite group) and/or have relatively small black populations(between 5% and 10%);these metro areas are located in data,themostcommonlyusedlevelof Censusgeography. 8Thesedatausetract-level Census tractstypicallyhavebetween2500and8000residentsandareclosestin size to whatmost Someresearchers considera neighborhood. (e.g.,Farley& Frey1994)reportsegregation measuresbasedon smaller,block-group data.Thisunitof geography usuallycontainsonly As a result,calculating a few hundredresidentsandis, on average,morehomogeneous. segregation levelyieldshigherresults(Ellen2000,p. 14). indicesattheblock-group groupexposurearenotshownhere,butareavailablefromthe 9Dataforminority-minority LewisMumfordCenter(www.albany.edu/mumford/census). as follows:changesof 10 or morepoints 1oChanges in segregation areusuallyinterpreted in a decaderepresentsignificantchange;changebetween5 and 10 pointsovera decade representmoderatechange;changesof less than5 pointsin a decadeareinterpreted as nearlyhalfof theareas littleorno change(Logan2001a,b,c).Basedon thesebenchmarks, experienced moderate declinesin segregation. OFRESIDENTIAL DYNAMICS SEGREGATION 173 A) (-22) (-13) (-9)(-25) (-27) (-8)(-22) (-20) (-13)(-17) (-12) (-8)(-25) (-8)(-7)(-7)(-5)(-8) (-17) (-17) (-29) (-12) (-12) (-28) (-19)(-19) (-26) (-13) (-13) Exposure (80-00 31 46 46 45 65 41 78 38 37 48 48 63 76 68 80 80 77 82 79 45 60 70 70 61 49 62 76 63 62 (Continued) A) (+7)(+13) (+9)(+9)(+3)(+3)(+6)(+2)(+6)(+4)(+7)(+5) (+14) (+8)(+19) (+7)(+17) (+11) (+27) (+6)(+10) 9 15 11 4 5 8 4 9 6 9 8 (80-00 29 11 26 22 19 29 Isolation 40 38 20 24 Asians (+6)(+6)(+10) (+3)(+3)(+2)(+3)(+2) 15 8 12 5 5 4 4 3 A) (+1)(+7)(+12) (+1)(-5)(+4)(+3)(-2)(+9)(+1)(+3) (80-00 48 38 40 47 35 42 32 49 42 49 42 Dissimilarity (+6)(+6)(+1)(+4)(+5)(+2)(+9)(+5)(+6)(+5) 49 45 28 30 42 32 30 30 41 36 (-3)(+5)(+13) (+1)(+3)(+1)(+2)(0) 44 46 43 43 38 35 42 39 1980-2000 A) (80-00 regions,Exposure (-23) (-13) (-13) (-17) (-7) (-15) (-13) (-17) (-23) (-28) (-16) (-25) (-17) (-23) (-17)(-20) (-19) (-15) (-13) (-20) (-4)(-9)(-13) (-20) (-17) (-2)(-30) (+1)(-14) 17 36 31 38 70 36 74 36 30 52 42 31 37 44 50 46 27 49 70 45 44 38 62 67 77 65 64 81 70 A) (80-00 Isolation metropolitan Hispanics largest 50 (+17) (+6)(+12) (+12) (+12) (+21) (+12) (+9)(+11) (+4)(+12) (+10) (+12) (+21) (+16) (+9)(+7)(+13) (+13) (-1)(+23) (+12) (+6)(+2)(+4)(+7)(+1)(+6) (+13) 8 4 17 17 2 7 63 50 54 44 30 15 34 41 21 36 49 45 46 38 37 66 34 22 40 42 48 19 10 A) (+5)(+5)(0) (+1)(0) (-7)(+20) (+8)(0) (+4) 56 54 53 50 48 51 43 43 47 49 (-2)(+4)(+10) (0) (0) (+5)(-1)(+15) 62 46 47 29 58 46 30 44 (-1)(+9)(+4)(+4)(+10) (+1)(+7)(-1)(+5)(+4) 22 33 51 47 40 34 48 74 40 43 (+5)(-2)(-4)(+8)(+5)(+9)(+4)(+7) 16 17 58 32 25 38 39 41 (-16) (-26) (-5)(-2)(-12) (-15) (-21) (-18) (-3)(-4) (-12) (-19) (-26) (-14) (-19) (-28) (-31) (-15) (-6)(-22)(-20) 3 9 24 35 Blacks (80-00 34 15 3 15 14 35 16 23 4 18 18 Isolation 21 24 20 19 47 42 (-10) (0) (-6)(-9)(-7)(-14) (-6)(-12) 73 79 23 65 71 53 58 53 the in (+11) (+14) (+8)(+6)(+5)(+9) (+6)(+4)(+13) (+9)(+11) (80-00 63 43 56 51 31 47 35 54 52 40 47 Dissimilarity A) whites from (-17) (-13)(-7) (0) (-17) (-6)(+4)(-2)(+7)(-3)(-18) (80-00 16 38 48 38 59 26 67 31 38 44 41 Exposure A) segregation Asian and A) (-21) (-14) (-9)(-9)(-10) (-18) (-11) (-7)(-8)(-3) (-11) (-9)(-19) (-18) (-7)(-18) (-20) (-13)(-15) (-8)(-3)(-10) (-12) (-20) (-9)(-8)(-9)(-4)(-9) (80-00 68 46 37 54 50 63 48 61 41 56 52 Dissimilarity 81 85 58 74 77 69 75 71 68 59 44 62 60 50 43 37 52 53 Hispanic, Beach areas average 1 area TABLE average Bernardino Black, County areas Angeles/Long Diego Francisco Jose area Paul Marcos area City/Ogdenareas City Worth/Arlington Lake Antonio Vegas Louis Orange SanSeattle/Bellevue/Everett Portland/Vancouver SanSanSacramentoHouston Denver SanLasSaltAustin/SanChicago St.Cleveland/Lorain/Elyria LosRiverside/San Oakland Dallas Phoenix/Mesa Detroit Kansas Fort Minneapolis/St. Cincinnati Indianapolis MetroWestern Southwestern Southwestern Western Midwestern 174 CHARLES A) (-29) (-30) (-14) (-25)(-12) (-18) (-27) (-10) (-12) (-26) (-19) (-7)(-19) (-12) (-14)(-18) (-12) (-7)(-11) (-20) (-20)(-13)(-16) (-10) (-7)(-14) (80-0078 65 75 57 59 71 75 29 62 59 63 65 51 67 75 70 62 66 71 75 85 69 65 69 68 62 Exposure 40(-15) A) (+5)(+4)(+4) (80-007 5 7 Isolation Asians (+9)(+7)(+5)(+3)(+2)(+4)(+3)(+2)(+3)(0) (+3)(+3)(+5)(+4) 14 8 7 4 3 5 4 6 4 11 4 4 7 6 (+11) (+7)(+1)(+6)(+4)(+7)(+13) (+5)(+7)(+6) 8 5 9 16 6 12 11 27 10 13 (+7)(+9)(+1)(0) (+3)(+4)(+1)(-4)(-4)(-3)(+3)(-1) (+1) 39 45 39 34 31 36 28 34 43 48 46 42 41(0)39 (+3)(+3) (+1)(+3)(-3)(+5)(+3)(+4)(+2)(+10) 51 44 45 36 49 35 36 43 42 40 A) (-3)(+10) (+3) (80-0042 41 41 Dissimilarity A) (-11) (-12) (-20)(-12) (-25) (-27) (-7)(-9)(-16) (-30) (-29) (-18) (-10) (-16) (-19)(-18) (-10) (-7)(-13) (-22) (-3)(-11) (-13) (-31)(-14)(-16) (-12) (80-0071 51 65 21 43 54 56 84 36 38 48 48 51 Exposure 45 49 66 61 18 55 55 60 55 61 58 68 55 54 A) (+5)(+17) (+7) 6 Isolation (80-00 33 16 Hispanics (+17) (+2)(+12) (+2)(+10) (+24) (+10) (+15) (+18) (+2)(+4)(+13) (+21) (+8) (+10) (+5)(+9)(+13) (0) (+9)(+11) (+10) 1 8 11 9 12(+11) 19 20 20 4 23 71 27 23 5 13 27 21 23 36 39 32 28 27 46(+6) Research. A) (+9)(+4)(+4) (80-0038 60 46 Dissimilarity (-1)(-2)(-3)(+18) (+3)(+6) (+4)(+1)(+18) (+9)(+19) (+10) (+2)(-3)(+3)(+10) (+16) (+21) (+3)(-5)(-9)(+10) (+23) (+19) Regional 67 60 59 47 30 65 58 68 57 48 48 53 36 45 44 41 32 32 50 36 51 46 43 43 and A) (+6)(-2) (+4) (80-0047 25 34 Exposure A) (-9)(-2) (-8) Blacks (80-0048 67 59 Isolation A) (-2) (-7) (-10) (80-0063 82 74 Dissimilarity (-1)(+2)(+4)(+7)(-7)(+5)(+6)(+5)(+2)(-2)(+3)(+5)(+1)(+2) 28 28 29 42 11 41 31 42 44 24 41 48 46 35 (-4)(+2)(+4)(-8)(+6)(-4)(-6)(-5) (-2) (+1) 11 28 40 34 50 17 27 56 33 37 number. Urban whole nearest Comparitive (-10)(-8) (-12) for the (-8)(-10) (-5)(-21) (-9)(-10) (-7)(-14) (-8)(-7)(-3)(-10) (-7)(-16) (-18) (0) (-11) (-10) (-11)(-10) to 59 63 66 43 62 41 53 52 45 71 49 46 43 53 60(-3) 62 39 41 47 67 36 13 46 41 Center rounded are (-13) (-6)(-11) (-7)(-14) (-7)(-17) (-22) (-8)(-2)(-8)(-9)(-6) (-10) (-3)(-6) (-7)(-13)(-6) (-10) Mumford (-7)(-11) 82(0) 73 59 72 62 63 66 68 65 74 57 62 46 55 69 59 57 46 61 72 66 74 67 80(-3) changes Lewis Theand News Point Hill and indices Census the of limitations, River/Warwick Hill Beach/Newport (Continued) 1 Petersburg/Clearwater average area areas York space average Bureau to U.S.Due areaaverage Orleans Lauderdale Philadelphia Boston Pittsburgh Bergen/Passaic Providence/Fall Greensboro/Winston-Salem/High Raleigh/Durham/Chapel New Nassau/Suffolk Newark Orlando Ft. Norfolk/Virginia Charlotte/Gastonia/Rock New Nashville Atlanta Baltimore Tampa/St. Miami Columbus Milwaukee/Waukesha Washington, Overall Source: Southern Eastern Notes: Southern Eastern Metro Midwestern area TABLE average DC area areas DYNAMICSOF RESIDENTIALSEGREGATION 175 the newer cities of the West and Southwest (Farley& Frey 1994, Frey & Farley 1993, Logan 2001a). Finally, approximatelyhalf of the areas show double-digit declines in isolation,althoughin manycases, exposureto whites eitherdeclinedor remainedconstant.Thus,for the 50 regions,blackisolationdeclinedby an average of 12%,but exposure to whites increasedby only 1%, on average.As indicated previously,the majorityof the decline in black isolation is due mainly to their increasingexposure to Hispanics (Alba et al. 1995, Frey & Farley 1993, Logan 2001a). Trendsin Hispanic and Asian segregationare the opposite of those observed for blacks.In most areas,Hispanic-whitesegregationremainsmoderate,isolation low, and exposure to whites meaningful, despite explosive population growth. Overall, increases in segregationrange from small to moderate.Hispanic-white dissimilarityneverexceeds 68 (andonly five areasexceed 60, comparedto 28 for blacks)andaveragesa low of 43 in the Southanda high of 57 in the East.Isolation increasedmore substantially,yet the averageHispanicresides in a neighborhood that is between 16%and 42% same-race(comparedto the averageblack person, whose neighborhoodis between 18% and 59% same-race).Exposureto whites declinedmore substantially.Finally,Asians remainthe least-segregatednonwhite group.Increasesin dissimilarityand isolation (except for the West, where Asians are most concentrated)are generallyless than 10% (the averageincreasefor the 50 regions was 3%), and declines in exposureto whites are comparableto those experiencedby Hispanics, once again reflecting the rapid populationgrowth of these largely immigrantgroups, concentratedsettlementpatterns,and declining white populationshare.In contrastto the residentialpatternsof blacks,Hispanics, and Asians, whites' exposure to minorities increased steadily over the past two decades:In 2000, theminoritypercentagein the averagewhiteperson'scensustract was a nontrivial20%, andresearchby Alba & colleagues (1995) documentssharp declines in the numberof all-white neighborhoodssince 1970. In short,although segregationpersists or increases for minoritygroup members,the averagewhite personexperiencesmodest integration.11 Finally, a brief mention of trendsin suburbansegregationis warranted.12In 2000, nearly 60% of Asians, 50% of Hispanics, and 40% of blacks lived in the suburbs,comparedto 71% of whites. These percentagesrepresentsubstantialincreases in minorityrepresentation;however,they have not been accompaniedby meaningfuldeclines in suburbanresidentialsegregation.Patternsof suburbansegregation mirrorthose of the larger metropolitanarea of which they are a part, indicating that new minority residents are moving to suburbswhere coethnics were alreadypresentin 1990. Wheregroupsare smallestin number,they areleast segregatedand least likely to establishsuburbanenclaves;however,in the regions fornon-Hispanic white-minority 11Data groupexposurearenotshownherebutareavailable Center(www.albany.edu/mumford/census). fromtheLewisMumford areasthatare outsideof 12Suburbs are definedas residentialareaswithinmetropolitan centralcities(Ellen2000,Schnore1963,Timberlake 2002). 176 CHARLES where the majorityof blacks, Hispanics, and Asians live and are, therefore, a largershare of the suburbanpopulation,"segregationis higher,more unyielding over time, andminoritypopulationgrowthis more likely to be associatedwith the creation or intensificationof ethnic enclaves" (Logan 2001b). Increasingminority suburbanizationwithin the context of persisting segregationhelps to explain the rising economic segregationamong both blacks and Hispanics documented by Jargowsky(1996). Minority suburbs-although betteroff than poor minority neighborhoods-tend to be less affluent,have poorerqualitypublic services and schools, and experience more crime and social disorganizationcomparedto the suburbsthat comparablewhites reside in (Alba et al. 1994; Logan et al. 2002; Pattillo-McCoy1999). PERSPECTIVES THEORETICAL ON RESIDENTIAL SEGREGATION A large body of research attempts to explain the persistence of residential segregation-particularly among blacks--despite the passage of antidiscrimination legislation, more favorableracial attitudesamong whites, and the dramatic expansion of the black middle class. This section summarizesthree competing explanationsfor persistingracial residentialsegregationthat garnerthe most research attention-objective differences in socioeconomic status, prejudice, and housing-marketdiscrimination-and reviews majorresearchfindingscirca 1980. Explanationsemphasizinggroup differencesin social class status are consistent with the spatialassimilationmodel, whereasthe place stratificationmodel includes explanationsplacingprimacyon persistingprejudiceand/ordiscrimination.Where appropriate,I consideralternativeexplanationsthatdo not fit neatlyinto eithertheoreticalperspective. SpatialAssimilation Racial group differences in socioeconomic status characteristicsare well documented.On average,blacks andHispanicscomplete fewer years of school andare concentratedin lower-statusoccupations,earn less income, and accumulateless wealth comparedto whites (Farley 1996a, Oliver & Shapiro 1995). The persistence and severityof these differenceslead easily to the conclusionthatresidential segregationby race is simply the logical outcome of these differences in status and the associated differences in lifestyle (Clark 1986, 1988; Galster 1988; see also Jackman& Jackman1983 on class identitiesas involvinglifestyle considerations). This assumptionis the basis of the spatialassimilationmodel, which asserts thatindividualsconvertsocioeconomicgains into higher-qualityhousing,often by leaving ethnic neighborhoodsfor areaswith more whites; for immigrants,it also involves acculturation-the accumulationof time in the United StatesandEnglish languagefluency.It should also be noted thatspatialassimilationis influencedby DYNAMICSOF RESIDENTIALSEGREGATION 177 the metropolitan-area characteristicsdiscussed in the previoussection (i.e., group size, ratesof grouppopulationchange,and suburbanization)(Alba & Logan 1993, Farley & Frey 1994, Massey & Denton 1985).13 STATUSDIFFERENCESTests of this hypothesis dominate segreSOCIOECONOMIC gation researchover the past two decades, and findings consistently show that Asians and Hispanics are always substantiallyless segregatedfrom whites than blacks are.As Asian andHispanicsocioeconomicstatusimprovesandgenerations shift from immigrant-to native-born,segregationfrom whites declines substantially. Conversely,objectivedifferencesin socioeconomic statusexplain only part of blacks' residentialoutcomes (Alba & Logan 1993; Denton & Massey 1988; Logan & Alba 1993, 1995; Logan et al. 1996; Massey & Denton 1987, 1993; Massey & Fischer 1999). Moreover,studies distinguishingamong white, black, and mixed-raceHispanics find that black and mixed-raceHispanics' residential patternsmirrorthose of AfricanAmericans.The exceptionalexperienceof groups with black skin leads Massey & Denton (1989, Denton & Massey 1989) to conclude thatblackspay a "higherconstantpenalty"for theirrace thatis not explained by socioeconomic statusdisadvantage. Until recently,the bases for these conclusions were aggregate-levelanalyses, primarilyfromthe Massey-Dentonsegregationresearchprojectthatculminatedin the publicationof AmericanApartheid.14Modeling aggregate-levelstudies suffer from several potentially importantlimitations, however.In particular,modeling individual-levelprocesses at the aggregatelevel (eithertract-or metropolitan-area 13Dueto space limitations,I limit my discussion of the spatialassimilationmodel to group differencesin socioeconomic statusand acculturation.The majorityof multivariateanalyses of the spatialassimilationmodel include one or more measuresof metropolitancontext (generallygroup size, rate of populationgrowth,region, and/ornew housing supply) and find associationsbetween contextualeffects and segregationconsistentwith those outlined in the previous section (see, for example, Alba & Logan 1993; Massey & Denton 1987; Massey et al. 1994; Logan et al. 1996; South & Crowder1997a,b, 1998; South & Deane 1993). Older,largercities, locatedprimarilyin the Northeastand Midwest are more segregatedthanthenewercities of the WestandSouthwest.Oldercities haveecological structures more conduciveto segregation:densely settled cores, thickly packedworking-classneighborhoods,and older housing stock built priorto the passage of the 1968 FairHousing Act (for a detaileddiscussion, see Farley& Frey 1994). 14Twotypes of aggregate-levelanalysis are common. In the first,a populationis separated into categoriesof a socioeconomic indicator(e.g., education,occupation,or income) and segregationindices arerecalculatedwithincategoriesof the selectedindicator.If segregation within categoriesof the indicatoris similarto the overall level, researchersconclude that socioeconomic status is not influential in residential outcomes for that group (see, for example, Darden 1995, Denton & Massey 1988, Massey & Fischer 1999). In the second type of aggregate-levelanalysis, multivariatemodels predict residential outcomes (e.g., probabilityof contact with whites) using the averagecharacteristicsof blacks, Hispanics, and Asians for a set of metropolitanareas (Massey & Denton 1987, Denton & Massey 1988). 178 CHARLES level) risks problemsof ecological inferenceand introducesmulticollinearitythat limits the numberof explanatorymeasures (Alba & Logan 1993, Massey et al. 1987, see also Massey & Denton 1987). Particularlyproblematic,homeownership is never includedin aggregate-levelstudies, despite its obvious implications for residentialoutcomes (Alba & Logan 1993, Charles2001b, Oliver & Shapiro 1995, Yinger 1995).15Finally, these studies measureand/orpredict segregation or, less frequently,central-cityversussuburbanlocationacrossmetropolitanareas. At least as important,however,is to understandvariationsin the characteristicsof the neighborhoods-both central-cityand suburban--wherevariousracial/ethnic groups actually live. For example, suburbanblacks tend to live in older, inner suburbsthat are less affluent,less white, and experience more crime and social disorganizationcomparedto the suburbswherecomparablewhites live (Alba et al. 1994, Logan & Schneider 1984, Pattillo-McCoy1999); thus, not all suburbsare equal, and aggregate analyses cannot detail these importantexperientialdifferences. Individual-levelanalyses addressthese limitations and substantiallyenhance ourknowledgeof residentialoutcomesby race;locationalattainmentmodels (Alba & Logan 1991, 1992) have been particularlyinfluentialin this regard.16An innovative method introducedby Alba & Logan (1991, 1992) transformsaggregatelevel Census data (mainly STF3 and STF4 files) into the functionalequivalentof individual-levelPublic Use MicrodataSample data with characteristicsof each person's communityof residence appended,eliminatingissues of ecological inference and multicollinearity[Loganet al. 1996, p. 858; for a detailedexplanation of the method,see Alba & Logan (1991, 1992)]. Models employ a broadrangeof social class indicators,most notablyhomeownershipand family status to predict neighborhood-leveloutcomes(e.g., medianincome andexposureto crime, as well 15Infairness,it shouldbe emphasized that,untilrecently,obtainingCensusdatawithboth individual-andaggregate-level datawasextremelydifficult.Masseyet al. (1991),Massey & Denton(1985),andVillemez(1980)areexceptionsto thetypeof analysisnotedhere; analyses(specialeditionsof the 1970and/or1980PublicUse Files all areindividual-level thatthe CensusBureauappendedneighborhood racialcompositionto) thatdo not suffer fromproblemsof ecologicalinference.Resultsareconsistentwiththoseof aggregate-level thatanyerrorsof substantive studies;Massey& Denton(1985,p. 94) conclude,therefore, analyses)areconservative (ofaggregate-level innature.Still,theseexceptions, interpretation counterparts, liketheiraggregate-level relyon a limitednumberof indicators(see Alba& Logan1993). mobilitystudiesaresimilarin thattheyconsidertheinfluenceof individual16Residential levelcharacteristics returns onthelikelihoodof movingaswellasthelocational toindividual characteristics;they differ in that mobility studies limit analysis to movers, and data are fromnationalpanelstudies(e.g., the PanelSurveyof IncomeDynamicsor the Annual Housing Survey;see Massey et al. 1994; South& Crowder1997a,b, 1998; South& Deane analysesof Censusdatabecausetheirresultsare 1993).I focushereon individual-level however,thepatternof resultsis consistentforthetwotypesof morebroadlygeneralized; studies. DYNAMICSOF RESIDENTIALSEGREGATION 179 as percentnon-Hispanicwhite andsuburbanversuscentral-cityresidence).Analyses comparethe characteristicsof suburbsinhabitedby whites, blacks, Hispanics, and Asians ratherthan aggregate-levelsegregation.17 These improvementshave yielded interestingandimportantinformation.Most interesting,perhaps,is that at the individual-level,blacks exhibit a positive association between socioeconomic statusand residentialoutcomes, althoughtheir returnsto educationand income are significantlylower thanfor othergroups.Especially troublingis the negative effect of homeownershipon blacks' residential outcomes.Counterto the benefitstypicallyassociatedwith owning a home (rather thanrenting),blackhomeownersresidein neighborhoodsthataremoresegregated and less affluentthan their rentingcounterparts--theyare the only group that is consistentlypenalized for owning a home (Alba et al. 2000a, Logan et al. 1996). Together,these differences keep blacks from reachingparity with whites at any level of affluence--blacks live in neighborhoodsthatare, on average,15%to 20% less affluentthan other groups with comparablestatus. Additionally,contraryto the assertionthatblack residentialsegregationis unchangedby increasingsocioeconomic status, Alba et al. (2000a) find that middle-class and affluentblacks in the most segregatedU.S. cities live in areas with substantiallymore whites than their poor, inner-citycounterpartsdo. This is counterbalanced,however,by the generally lower status of their white neighbors.Thus, the suburbanareas where middle-classand affluentblacks live are significantlyless white and less affluent thanthose of comparablewhites. Patternsfor Asians and Hispanics,on the otherhand,are more similarto those observedin the aggregate.Both show substantialresidentialgains with improved socioeconomic status,andeffects for homeownershiparemixed (often nonsignificant andoccasionallynegative,althoughless so thanfor blacks);effects of education and income are large enough, however,that averageand affluentnative-born Hispanicsand Asians live in communitiesthat are roughly equivalentto those of comparablewhites (Logan et al. 1996, Alba et al. 1999). A comparisonof 1980 and 1990 data suggests a weakeningof the traditionalspatialassimilationmodel regardingthe importanceof acculturation.Being native-bornand speaking only English still improves Hispanics' locational attainment,but the latteris less importantin 1990 comparedto 1980; by 1990, neithercharacteristicdisadvantaged Asians. The emergence of ethnic suburbanenclaves may account for this apparent weakeningof the traditionalspatialassimilationprocess by makingresidence in high-status,suburbancommunitiesan option for recently arrivednon-English speakerswith at- or above-averagesocial class characteristics(Alba et al. 1999, 2000b; Loganet al. 2002). Furthermore,"perceptibleAfricanancestry"costs black because(a)segregation is lowerinthesuburbs compared focusonsuburban 17They residents tothecentralcityforallgroups(seeMassey&Denton1989),suggesting thattheinfluenceof forthetwolocations,and(b)itrevealstheprocess characteristics is different individual-level thatdetermineslocation within suburbiaand elaboratesdifferencesin the characteristicsof suburbanneighborhoodsacross racial/ethnicgroups(Alba & Logan 1993, p. 1400). 180 CHARLES Hispanicsbetween $3500 and $6000 in locationalreturns,placing them in neighborhoods that are comparableto those of black Americans (Alba et al. 2000b, p. 613). Much of the research discussed to this point focuses heavily on the use of statisticallyconvenient,but homogenizing,racialcategories.Consideringcharacteristics specific to immigrationmay account for some intragroupdiversity,and it is certainly a step in the right direction;however, an importantbody of research documents meaningful differences among national-origingroups within the same broadracialcategory,suggestingthe importanceof analysesthatare sensitive to these differences(see, for example,Portes& Rumbaut1996; Waldinger& Bozorgmehr 1996; Waters 1990, 1999). At the aggregate level, Massey & Bitterman's(1985) comparisonof Mexicans in Los Angeles and Puerto Ricans in New York--demonstratingthatdifferencesin segregationareattributableto the lattergroup'sgenerallylower socioeconomic statusand "blackness"-represents both an importantexception to this generaltendency and evidence of potentially importantintragroupvariation. A final advantageof the individual-levelanalyses detailed above is the serious attentionseveral studies have paid to national-origindifferenceswithin each of the four majorracial categories. Consistentwith assimilationhypotheses, for example, Logan & Alba (1993) find thatthe more recently settled Irish-, Italian-, and Polish-originwhites tend to reside in lower-incomeneighborhoodsthan the earlier-arriving(e.g., British,French,and German)and otherwhite ethnic groups, net of individual-levelcharacteristics.For "blacks,"results are consistent with those detailedabove:Afro-Caribbeanblacks experiencemorefavorableoutcomes and see betterreturnsto theirhumancapitalthanAfricanAmericansdo (Alba et al. 1999; Crowder1999; Logan & Alba 1993, 1995).18 National-origindifferencesaremost pronouncedamongAsians and Hispanics, the two most heterogeneousand rapidly growing groups. Logan & Alba (1993) find that Asian Indians,Filipinos, and Vietnamesetend to reside in less affluent neighborhoodsthanChinese, Japanese,Koreans,and otherAsian groupsdo, suggesting that this effect may be tied to the extremepovertycircumstancesof their home countries comparedto those of other Asian groups. Alternatively,Asian Indians and Filipinos are not at all disadvantagedby poor English skills; in this case, the researcherssuggest that this is because English is widely used in both India and the Philippines.As a result, these groups arrivewith more exposureto English; therefore,the "census self-assessmentof English ability has a different meaningfor them"(Alba et al., 1999 p. 457; see also Jasso & Rosenzweig 1990). 18Ithas also been suggested that importantdifferences in socioeconomic status among fromthe WestIndianimmigrants mayinfluenceresidentialoutcomes.Blackimmigrants BritishWestIndies (mainlyJamaica)have significantlyhigherincomes, educationallevels, employmentand homeownershiprates, and are more concentratedin high-statusoccupations than FrenchWest Indians(mainly Haitians)are;Dutch West Indiansfall in between (Crowder1999, p. 103). DYNAMICSOF RESIDENTIALSEGREGATION 181 The disadvantageassociatedwith poor English skills, moreover,declined considerablybetween 1980 and 1990 for the Chinese and Koreans,but increasedamong the Vietnamese.19None of the Asian groups is meaningfully disadvantagedby recent arrival(Alba et al. 1999). Both national-originand racial classificationmatterfor Hispanics. PuertoRicans, Cubans,and all black Hispanicsreside, on average,in lower socioeconomic statusneighborhoodsrelativeto otherHispanics,net of otherindividualcharacteristics (Alba & Logan 1993, pp. 260-64). Poor English skills negativelyimpactthe likelihood of suburban(versus central-city)residence for Mexicans and Cubans, but not for Dominicans and Salvadorans.Also, in sharp contrast to the Asian groups,recentimmigrationto the United Statesis detrimentalto the likelihoodof suburbanresidencefor all Hispanicgroups(Alba et al. 1999). Althoughlimitedin number,these analyses highlightimportantvariationin residentialoutcomes and in the factorsthat influence those outcomes that are hidden by the use of broad, analyticallyconvenientracialcategories;futureresearchshouldcontinueto expose the complicatedsocial realitieshiddenby this social science convention. On the whole, conclusions of aggregate-levelstudies remainintact.The experiences of Hispanicsand (for the most part)Asians are largely consistentwith the spatial assimilationmodel; blacks (African Americans and black Hispanics) on the otherhand, do not see the same payoff for improvedsocial class status.This is best illustratedby the negative effect of homeownership.Alba and colleagues suggest that this representsthe operationof a dual housing marketthat restricts black homeowners-but not black renters--to black neighborhoods,making it difficultfor them to entersome neighborhoodsand addingto the cost they pay for housing. Mixed effects of homeownershipamong Asians and Hispanics suggest that a dual housing marketmay operateto a lesser extent for them as well (Alba & Logan 1993, Alba et al. 1999, Logan et al. 1996, Massey & Denton 1993). Finally, non-Hispanicwhites live in largely white and generally more affluent neighborhoodsirrespectiveof their social class characteristics.The oppositional experiencesof blacks and whites contradictthe tenets of spatialassimilationand suggest the persistenceof an enduringsystem of racial stratification. Place Stratification The emergenceof raciallyseparateneighborhoodsresultedfrom a combinationof individual-andinstitutional-levelactions.Scholarsgenerallyagreethatall levels of government,as well as the real estate,lending, andconstructionindustries,played criticalrolesin creatingandmaintaininga dualhousingmarketthatconstrainedthe amongtheVietnamese 19Albaet al. (1999)suggestthatthecountertrend is attributable to theirrefugeestatusduringthe1970s.Inthisperiod,housinglocationwasdetermined largely by resettlement agencies;morerecently(1980s),housingdecisionsaremademoreon the maybe enteringa basisof householdneeds.AlbaandcolleaguesconcludetheVietnamese groups(Albaet al. 1999, settlement phasecomparable to theearlierphasesof nonrefugee p. 457). 182 CHARLES mobility options of blacks (for detaileddiscussions, see Massey & Denton 1993, Meyer 2000, Yinger 1995). It was assumed by many, however, that passage of the 1968 FairHousing Act markedthe beginningof the end of segregation.This, however, has not been the case: residentialsegregationpersists, and substantial evidence points to continued resistance to more than token numbers of black (and, to a lesser extent, Hispanic) neighbors (Bobo & Zubrinsky1996; Charles 2000, 2001b; Farley et al. 1993, 1994; Meyer 2000; Zubrinsky& Bobo 1996) and discriminatorypractices in the real estate and lending markets (Massey & Lundy2001, Galster1990, Yinger1995). The place stratificationmodel punctuates the centralityof these issues, arguingthat "[r]acial/ethnicminorities are sorted by place according to their group's relative standing in society, [limiting] the ability of even the socially mobile members to reside in the same communities as comparablewhites" (Alba & Logan 1993, p. 1391). Whites use segregation to maintainsocial distance, and therefore,present-dayresidentialsegregationparticularlyblacks' segregationfrom whites-is best understoodas emanating from structuralforces tied to racialprejudiceand discriminationthatpreservethe relative statusadvantagesof whites (Bobo & Zubrinsky1996, Logan et al. 1996, Massey & Denton 1993, Meyer 2000). Despite general agreementregardingthe role of prejudiceand discrimination (both individualand institutional)in the emergenceof racially segregatedneighborhoods, the extent to which these factors are implicatedin the persistence of segregationremains contested. Alternativeexplanationsdownplay the continuing salience of prejudiceand/ordiscriminationin favor of otherrace-relatedattitudes and perceptions.The in-grouppreferencehypothesis arguesthat all groups have "strongdesires" for neighborhoodswith substantialnumbersof coethnics (Clark 1992, p. 451) that reflect a simple, naturalethnocentrismratherthan outgroup hostility or an effort to preserverelativestatusadvantages.A strongerversion contends blacks' own preferencefor self-segregationexplains currentlevels of black-white segregation(see, for example, Patterson1997, Thernstrom& Thernstrom1997). Accordingto the racialproxy (Clark1986, 1988; Harris1999, 2001) and the race-basedneighborhoodstereotypinghypotheses (Ellen 2000), it is the collection of undesirablesocial class characteristicsassociatedwith blacks or the neighborhoodswhere they are concentrated-joblessness, welfare dependence, proclivityto criminalbehavior-not race per se, thatmotivatesaversionto black neighbors,not only among out-groups,but among blacks themselves. Still, race is centralto each of these alternatives,and direct assessments of the role of prejudice often include one or more of these alternativeexplanations.As such, I addressthese alternativeexplanationswithin the context of stratification-based explanations. NEIGHBORHOOD RACIAL PREFERENCES COMPOSITION A well-established litera- ture details black-whitedifferencesin preferencesfor integration.In theirclassic article, "Chocolate City, Vanilla Suburbs,"Farley and colleagues (1978) introducedaninnovativeandhighlyregardedmethodfor measuringviews on residential DYNAMICSOF RESIDENTIALSEGREGATION 183 White respondent scenarios Card 1 Card2 Card3 Card4 Card5 Card4 Card5 Blackrespondent scenarios Card1 Card2 Card3 neighborhoodcardsfor blackandwhiterespondentsfromthe Figure 1 Farley-Schuman 1992-1994 Multi-CityStudyof UrbanInequality. segregation.In the experiment,white respondentsare asked about their comfort with and willingness to enter neighborhoodswith varying degrees of integration with blacks;blackrespondentsreceive a similarexperiment:ratingneighborhoods of variousracial compositions from most to least attractiveand indicatingtheir willingness to enter each of the areas. In both cases, scenarios representrealistic assumptionsregardingthe residentialexperiencesand options of both groups; neighborhoodcardssimilarto those used by Farleyand colleages arepresentedin Figure 1 (for details, see Farleyet al. 1978, 1993).20 Resultsrevealedsubstantialresistanceby Detroit-areawhites to even minimal levels of integration:25%said the presenceof a single black neighborwould make them uncomfortable,40% said they would try to leave an areathatwas one-third black, and nearly twice as many would leave the majorityblack neighborhood (Farleyet al. 1978). Blacks, on the otherhand,showed a clearpreferencefor integration.Eighty-fivepercentchose the 50-50 neighborhoodas theirfirstor second choice; when asked to explain their selection, two thirdsstressedthe importance of racialharmony(Farleyet al. 1978, p. 328). Virtuallyall blacks were willing to enterall threeintegratedneighborhoods,and 38%of Detroit-areablacks said they would move into an otherwiseall-white neighborhood. As partof the 1992-1994 Multi-CityStudy of UrbanInequality(MCSUI),the Farley-Schumanshowcardmethodology was replicatedin Atlanta,Boston, Detroit,andLos Angeles; to enhanceourunderstandingof preferencesin multiethnic 20The1976Detroitareastudyintroduced viewson racial aninnovativewayof measuring residentialsegregation thathas influencedimportant generalassessmentsof the statusof AfricanAmericans(e.g.,Jaynes& Williams1989,pp. 141-44;Bok 1996,p. 182)as well (Massey& Denton1993,Yinger treatisesonracialresidential as twoimportant segregation 1995). 184 CHARLES contexts, the experimentwas modified to include Hispanicsand Asians.21Analyses of neighborhoodracial composition preferencesbased on the MCSUI data highlight the influenceof both respondent-and target-grouprace on attitudestowardresidentialintegration(Charles2001b; Clark2002; Farleyet al. 1993, 1997; Zubrinsky& Bobo 1996). Relative to the 1970s, whites express greatercomfort with higher levels of integrationand fewer said they would be unwilling to enter racially mixed areas.Although a sizeable majorityof whites express comfort with a one-thirdout-groupneighborhood,a rankorderingof out-groupsis evident: whites feel most comfortablewith Asians and least so with blacks (Hispanicsfall in between), andcomfortdeclines as the numberof out-groupmembersincreases. The patternof responses regardingwhites' willingness to enter racially mixed neighborhoodsis similar, except that the decline in willingness to enter begins earlier and is never as high as comfort with neighborhoodtransition;thus, 60% of whites are comfortablewith a neighborhoodthat is one-thirdblack, but only 45% of whites are willing to move into that same neighborhood(Charles2001b). Although reflecting meaningful improvementsin whites' attitudes,Detroit-area whites standout as more resistantto integrationcomparedto whites in the other cities (Farleyet al. 1997). Blacks, Hispanics, and Asians all appearto want both meaningfulintegration and a substantialcoethnic presence. The relative importanceof these competing desires,however,dependson boththe respondent-andtarget-grouprace.The overwhelming majorityof blacks selected one of the two most integratedalternatives irrespectiveof out-grouprace, althoughthe one with 10 black and 5 out-group households is slightly more attractivethan the one that best approximatesa 5050 neighborhood.For Hispanicsand Asians, on the otherhand, target-grouprace is especially important:when potentialneighborsare white, their most attractive neighborhoodsare the same as those of blacks (Cards2 and 3), althoughthe order is reversed.When potentialneighborsare black, however,between 60% and 80%of both HispanicsandAsians find one of the two least-integratedalternatives most attractive(Cards 1 and 2). Across respondentracial categories, the all-thesame-racealternativeis least attractivewhen potentialneighborsare white; however, Hispanics and Asians generallyfind this neighborhoodmore attractivethan blacks do.22Both groups are also twice as likely as blacks to select the all-white technique,onethirdof eachrespondent racialcategoryin LosAnge21Usinga split-ballot les (whites,blacks,Hispanics,andAsians)andBoston(whites,blacks,andHispanics)was (e.g., one thirdof Hispanicscompletedthe randomlyassignedto one of threeout-groups onethirdcompleteda Hispanic-black experiment, Hispanic-white andthereexperiment, integration withAsians).Exceptfordifferences mainingonethirdof Hispanicsconsidered in targetgroups,black,Hispanic,andAsianrespondents all completedthesameversionof & Bobo(1996). theexperiment. Fordetails,see Charles(2001b)orZubrinsky 22Research suggestsimportantdifferencesby immigrantstatusand acculturation. The particularly foreign-born, thosewith5 yearsorless in theUnitedStatesand/orthosewith limitedEnglishproficiency, prefersubstantially moresame-raceneighborscomparedto andthosewho communicate counterparts, theirnative-born efandlong-term-immigrant fectivelyin English(Charles2000,2001b,2002). DYNAMICSOF RESIDENTIALSEGREGATION 185 neighborhoodas theirfirstor secondchoice (approximately10%for Hispanicsand Asians, comparedto 5%of blacks),althoughfor all threegroups,the all-out-group alternativeis the least attractive.Patternsof willingness to enter neighborhoods mirrorthosefor attractiveness.Forblacks,thesepatternssuggest a slight shift away from a preferencefor 50-50 neighborhoodsand a significantdecline in willingness to be the only black family in an otherwiseall-whiteareasince 1976 (Charles 2001b; Farleyet al. 1993, 1997). Othermultiethnicstudies of preferencesyield similarresults.Bobo & Zubrinsky (1996) analyze multiracialdata but measure attitudestoward one group at a time using a single forced-choice item. Considerationof a single targetgroup is a limitation of the Farley-Schumanmethodology in multiethnic contexts as well, as aredifferencesbetween white and nonwhiteexperimentsthatmake direct comparisonsdifficult. Charles(2000) presentsa majorinnovationon the FarleySchumanexperimentthatallows the simultaneousconsiderationof whites, blacks, Hispanics,and Asians as potentialneighbors,using a single item in which all respondentsare asked to draw their ideal multiethnicneighborhood.Regardlessof the measure,the patternof resultsis the same. All groupsexhibit preferencesfor both meaningfulintegrationanda substantialpresenceof same-raceneighbors,althoughpreferencesfor same-raceneighborsarenot uniformacrossgroups:whites exhibit the strongest preferencefor same-race neighbors and blacks the weakest. Moreover,preferencesvary by the race of the targetgroup and demonstrate a racial rank ordering of out-groupsin which whites are always the most desirable out-groupand blacks are always the least desirable.Finally, preferences for integrationdecline as the numberof out-groupmembersincreases.These bivariatepatternsmake it clear that race is influentialin the residentialdecisionmaking process (Bobo & Zubrinsky 1996; Charles 2001b; Farley et al. 1993, 1997; Zubrinsky& Bobo 1996). Therefore,the next logical questionis how does race matter? THEALTERNATIVES Althoughsuggestive,thepatternsdetailed PREJUDICE VERSUS above arenot conclusiveevidenceof the primacyof racialprejudice;however,several multivariateanalyses detail whetherand how race mattersat the individual level. Using the MCSUIdatafor Detroit,Farleyandcolleagues (1994) showedthat antiblackstereotypesare stronglyassociatedwith whites' discomfortwith black neighbors,their likelihood of fleeing an integratingarea, and their willingness to entermixedneighborhoods.Similarly,Timberlake(2000) concludesthatnegative racial stereotypesand perceptionsof group threatfrom blacks are the strongest predictorsof whites' resistanceto integration,basedon analysisof MCSUIdatafor Atlanta.ForAtlantablacks, negativeracial stereotypesand (to a lesser extent)the perceptionof whites as tendingto discriminateagainstothergroupscontributeto theirintegrationattitudes.Bobo & Zubrinsky(1996) and Charles(2000) analyzed multiracialdatafromLos Angeles, concludingthatnegativeout-groupstereotypes reduce openness to integrationacross racial categories and influencepreferences for both out-group(Bobo & Zubrinsky1996, Charles2000) and same-raceneighbors (Charles2000). 186 CHARLES TABLE 2 Summarystatistics,multiethnicneighborhood showcardexperiment,2000 GeneralSocial Survey Respondent race Target group race Whites Blacks Hispanics Whites Mean % No whites All whites 57.11% 0 20.28 30.40% 9.21 0 31.50% 1.28 0 Blacks Mean % No blacks All blacks 16.80% 24.71 0 42.01% 0.66 6.58 18.77% 17.95 0 Hispanics Mean % No Hispanics All Hispanics 12.82% 32.17 0 14.47% 27.63 0 33.61% 2.56 1.28 Asians Mean % No Asians All Asians 13.27% 32.75 0 13.11% 32.24 0 16.11% 19.23 0 Numberof cases 858 152 78 Notes: The percentageof each racialgroupin a respondent'sideal neighborhood is the sum of each groupincludedin the experiment,dividedby the total number of houses (14), excluding the respondent's. p < .001 except: all Hispanics (p < .01), no Asians (p < .05), and mean percent Asian (p < .10). Each of these analyses includes a measureof respondents'perceptionsof the social class positions of out-groupsrelative to their own group as a test of the racial-proxyargument;Bobo & Zubrinsky(1996), Charles(2000), andTimberlake (2000) also include measuresof in-groupattachmentto assess the relativeimportance of ethnocentrism.In all instances,racial stereotypesare the most powerful predictorsof preferences.Effects for both perceivedsocial class disadvantageand in-groupattachmentare always smallerand often nonsignificant;indeed, this patternpersists acrossrespondentracialcategoriesfor both out-groupand same-race neighbors (Charles2000). Nationally representativedata from the 2000 General Social Surveybothconfirmandstrengthenthe resultsof these single-city analyses. Table2 summarizespreferencesfor white, black, andHispanicrespondentsand reveals a patternof preferencessimilarto those found by both Bobo & Zubrinsky (1996) and Charles(2000).23 Comparedto the data from Los Angeles, however, 23Themeasureof preferences usedin the 2000 GeneralSocialSurveyis identicalto that usedby Charles(2000).Respondents areshowna neighborhood cardsimilarto thosein Figure1, exceptthatthehousesareblank.Theyaretheninstructed as follows:"NowI'd likeyouto imaginea neighborhood would thathadanethnicandracialmixyoupersonally DYNAMICSOF RESIDENTIALSEGREGATION 187 bothwhitesandblacksprefermoresame-raceneighbors(8%and5%,respecNationally,whitesaremuchmore tively),buttheoppositeis trueforHispanics.24 likelytoexcludeanout-group entirely:25%of whiteswantnoblacksin theirideal neighborhood (comparedto one fifthin Los Angeles)andas manyas one third Hispanics exclude to 16%-17%in LosAngeles).25Blacks andAsians(compared are also substantiallymore exclusionaryat the nationallevel, exhibitingrates of Hispanic and Asian exclusion between three and five times higher than for Los Angeles. Finally, Table 3 presents correlationsbetween neighborhoodracial composition preferencesand each of the race-relatedattitudesoutlinedabove-perceived social class difference,racialstereotyping,andin-groupattachment.In additionto preferencesfor variousout-groupneighbors,the bottom panel of Table3 reports correlationsbetween the selected racial attitudesand preferencesfor same-race neighbors.The perceivedsocial class differenceand racial stereotypingmeasures are scaled from -6 to +6. High (positive) scores indicate unfavorableratingsof out-groupsrelative to one's own group, low (negative) scores indicate favorable ratingsof out-groups,and a score of zero indicatesno perceiveddifference.26Ingroupattachmentcapturesthe extentto whichrespondents"feel close" to members of theirracial group;scores range from 0 (not at all close) to 8 (very close). This item was only asked of white and black respondents. Consistentwith Bobo & Zubrinsky(1996), racialstereotypingis therace-related attitudeor perceptionmost correlatedwith preferences,irrespectiveof respondent or target-grouprace. As stereotypesof out-groupsbecome increasinglyunfavorable,preferencesfor those groups as neighborsdecline and preferencesfor feel most comfortablein. Here is a blank neighborhoodcard, which depicts some houses that surroundyour own. Using the lettersA for Asian, B for Black, H for Hispanicor Latin Americanand W for White, please put a letter in each of these houses to representyour preferredneighborhoodwhere you would most like to live. Please be sure to fill in all of the houses." 24In2000, respondentswere askedwhetherthey were Hispanicin additionto being racially identifiedas white, black, or other.ForTables2 and3, anyoneprovidinga Hispanicidentity is treatedas Hispanic; those unidentified"others"who remainedwere droppedfrom the analysis.Due to the small numberof Hispanicrespondents(this groupis not intentionally sampled),resultsfor this group shouldbe treatedas exploratory. 251suspect that differences in rates of inclusion/exclusion-particularly as they relate to Hispanic and Asian neighbors-are a consequence of the populationcomposition of Los Angeles comparedto the countryas a whole. Unlike the countryas a whole, Los Angeles is heavily populatedby both HispanicsandAsians;these groupsmay not be on the "cognitive maps"of residentsin cities thatdo not attractlarge numbersof immigrants. consistency(Cronbach'salpha)amongthe five traitsincludedin the racialstereo26Internal typing measure(intelligence,laziness, violence-prone,committedto strongfamilies, committedto fairnessandequalityfor all) varyby targetgroupas follows: for whites a = .62, for blacks a = .67, for Hispanicsa = .57, and for Asians a = .64. In the same-racepreference block, measuresof perceived social class difference and racial stereotypingare combined for all out-groups(e.g., forblackrespondents,these measuresreflectperceptionsof/attitudes towardwhites, Hispanics, and Asians). 188 CHARLES TABLE 3 Correlationsof race-relatedattitudesand perceptionsand neighborhood racial compositionpreferences,2000 GeneralSocial Survey Respondent race Target group race Whites Blacks Hispanics Total Percentageof white neighbors Perceived social class difference Racial stereotyping In-groupattachmenta - -.108 -.148+ .119 -.077 -.192+ - -.095 -.167' Percentageof black neighbors Perceivedsocial class difference Racial stereotyping In-groupattachmenta -.056 -.390'" -.091* - -.302" -.454'" - -.082* -.390'" -.037 Percentageof Hispanicneighbors Perceived social class difference Racial stereotyping In-groupattachmenta -.034 -.319*"' -.150' -.019 -.104 -.095 Percentageof Asian neighbors Perceivedsocial class difference Racial stereotyping In-groupattachmenta -.051 -.285*' -.105' .003 -.204* .040 Perceivedsocialclassdifferenceb .041 .074 Racial stereotypingb In-groupattachmenta .429**' .142"* .048 -.202+ - .109"** -.065' -.305"' -.141" -.051 -.269'" -.082* Percentageof same-raceneighbors .162+ -.064 .285* .208'" .500*** - .421" .104** Notes: Figures are Pearsoncorrelationswith preferencesfor the correspondinggroup as neighbors.The perceived social class difference and racial stereotypingmeasuresare scaled from -6 to +6. High (positive) scores indicate unfavorableratings of out-groupsrelative to one's own group;low (negative) scores indicate favorableratings of out-groups;0 indicatesno perceiveddifference.Internalconsistency (Cronbach'salpha)amongthe traitsincludedin the stereotypedifference score (intelligence, laziness, violent, committedto strongfamilies, committed to fairness and equalityfor all) varyby targetgroupas follows: for whites, a = .6217; for blacks, a = .6736; for Hispanics,a = .5719; and for Asians, a = .6384. aIn-groupattachmentquestionsask aboutfeelings of closeness to whites and blacks only. bInthe panel of correlationsbetween racialattitudesandpreferencesfor same-raceneighbors,measuresof perceived social class difference and racial stereotypesare combined for all out-groups(e.g., for Hispanic respondentsthese measuresreflect perceptionsof/attitudesaboutwhites, blacks, and Asians). +p < .10;'p < .05; *p < .01; ^p < .001. same-raceneighborsincrease. This is especially true among whites, where correlations between racial stereotypingand preferencesare between two and four times largerthanfor in-groupattachmentand those for perceivedsocial class difference are both weak and nonsignificant.Correlationsamongthese variablesare weakestfor blacks.Of 12 figures,3 barelymeet the most-liberallimits of statistical significanceacknowledgedin the social sciences. Hispanicsare in between these extremes,exhibitingthehighestcorrelationbetweenbothperceivedsocial class difference andracialstereotyping,and preferencesfor black neighbors;stereotyping DYNAMICSOF RESIDENTIALSEGREGATION 189 is only marginallycorrelated(p < .10) with preferences for white and Asian neighbors.Despite some differences,preliminaryevidence from the 2000 General Social Surveysubstantiatespriorfindingsfrom single-city analysesandhighlights the primaryimportanceof racialprejudicerelativeto both concerns about social class disadvantageand/or ethnocentrismin understandingneighborhoodracial compositionpreferences. More recently,Krysan& Farley(2002) supplementquantitativeanalyses with an examinationof black MCSUI respondents'open-endedexplanationsof their integrationattitudes.Contraryto proponentsof both the ethnocentrismand racial proxy hypotheses,they find thatbelief in the principleof integrationand/ora desire to improverace relationsdrivesblacks' preferencesfor integration:This was the most common explanationfor the attractivenessof the two most popular(and most integrated)neighborhoods(see Figure 1, Cards2 and 3). Moreover,strong desiresfor a substantialcoethnicpresenceare "inextricablylinked"to fearsof discriminationand white hostility (Krysan& Farley2002, pp. 968-69); the latteris consistentwith otherdescriptiveanalysesdetailingan inverseassociationbetween perceived white hostility and overall neighborhooddesirability(Charles2001b, Farley et al. 1993, Zubrinsky& Bobo 1996). They find virtuallyno supportfor eitherthe ethnocentrismor the racial-proxyhypothesis.Few blacks invokeethnocentric attitudes,even when favoringthe all-blackover the 50-50 neighborhood; contraryto the assertion that blacks (and whites) use race as a proxy for negative neighborhoodcharacteristics,only 10% of black respondentscite negative neighborhoodcharacteristicsas the primaryreasonto avoid all-blackareas. A similar analysis examines open-endedelaborationsfrom "whites who say they'd flee" (Krysan2002). Once again, evidence of ethnocentrismis sparewhen consideringintegrationwithblacks.Concernsaboutculturaldifferencesweremore salient for whites in Los Angeles contemplatingintegrationwith Asians and Hispanics, andareexpressedmainlyin termsof languagedifferences.Consistentwith the racial-proxyandrace-basedneighborhoodstereotypinghypotheses,whites offer race-associatedreasonsmost often (e.g., concernsaboutcrime and/orproperty values); however, meticulous analysis of the characteristicsof whites offering clearly racial versus race-associatedresponses finds that education makes the difference. More educated whites are both less willing to negatively stereotype out-groupsandmore likely to offer race-associatedexplanationsfor theirdecision to flee an integratedarea. Krysan (1998) suggests that because better-educated respondentsare both more susceptible to social desirabilitypressuresand more adept at articulatingtheirracialgroup'sinterestin more subtle ways (Jackman& Muha 1984), the differencebetweenexplicitlyracialand so-calledrace-associated explanationsis semantic:"[I]nthe end, each of the reasons is an articulationof a racial stereotype"(Krysan2002). The analyses discussed to this point representimportantmethodological,empirical,andsubstantivecontributionsto ourunderstandingof integrationattitudes. Especially insightful are those analyses that elaboratepreferencesin multiethnic contexts and those that employ multiple methods to gain leverage on the 190 CHARLES complexitiesof racialattitudes.Nonetheless, critics aptlypoint to importantlimitations associatedwith studies thatrely on measuresof expressedpreferences.In particular,because it is clear to respondentsthattheirracialattitudesare at issue, responses are susceptibleto social desirabilitypressuresand the difficultyof distinguishingbetweenthe directeffect of the racialcompositionof the neighborhood and the indirecteffect of the neighborhoodcharacteristicsthat respondentsmay associate with the racial composition of the neighborhood.Although preference studies include a measureof perceived social class difference,other unmeasured aspects of the proxy argument(e.g., crime, school quality) are left uncontrolled. Each of these limitationscan bias results(Ellen 2000, Emersonet al. 2001, Harris 1999). Alternatively,tests of the racial-proxyhypothesis that use respondents' actualresidentiallocation and the value of theirhomes as indicatorsof neighborhood desirabilityare confoundedby the fact that "[e]ven if people prefer to live in racially mixed neighborhoods,they may not end up in such neighborhoods" because of discriminationor a shortageof housing (Emersonet al. 2001, p. 924). A recent analysisby Emersonet al. (2001) standsout for creativelyand effectively addressingthese limitationsusing a factorialexperimentto assess whites' attitudestowardintegrationwith blacks, Hispanics, and Asians. Respondentsare asked to imagine that they have two school-aged childrenand are looking for a house;they havefounda house they like betterthananyother(it has everythingthe respondentis looking for) that is both close to work and within their price range. Before askingwhitesif theywouldbuy the home, they areoffereda set of randomly generatedneighborhoodcharacteristics-public school quality,crime level, direction of propertyvaluechange,home valuecomparedto othersin the neighborhood, and racial composition(between 5% and 100%Asian, black, or Hispanic). They findthatthe presenceof HispanicsandAsians does not matterto whites, butblack neighborhoodcomposition matterssignificantlyeven after controllingfor proxy variables. Whites are neutral about buying a home in a neighborhoodbetween 10%and 15%black and areunlikelyto buy the home in a neighborhoodover 15% black. This patternis especially pronouncedamong families with children.27 27Ellen(2000)arguesthattheheightenedsensitivityof whitehomeowners and/orfamilies it is actualormeasured aschangeovertime)is withchildrento racialcomposition (whether thanrace-based evidencethatpureprejudiceis less important stereotyping; neighborhood shouldnotvary,shearguesbythesestatusesamongwhites.Thepresenceand pureprejudice amongwhites,sheargues,indicatesthatwhitesareexpressing strengthof thesedifferences concernsaboutproperty valuesandschoolquality.It is alsopossible,however,thatthese simplyreflecttheincreasedsalienceof particular differences aspectsof blackstereotypesand/orparents.To welfaredependence andintellect,forexample-forwhitehomeowners of blacksby housingtenureandparenting testthis,I comparedwhites'racialstereotypes statususingdatafromthe 1993-1994Los AngelesSurveyof UrbanInequality(results morenegaarenot shown,butareavailableuponrequest).Ownersexpresssignificantly tive stereotypesof blacksrelativeto whites(p < .001);a similarpatternemergeswhen whiteparentsto nonparents, comparing signifialthoughthe differenceis notstatistically cant.Differencesin perceptions of blacks'socialclassstatusby tenureandparenting status DYNAMICSOF RESIDENTIALSEGREGATION 191 The overall conclusion to be drawnis that active racial prejudiceis a critical componentof preferencesfor integration,andtherefore,the persistenceof racially segregatedcommunities.Whites' racialprejudiceis a doublewhammy:influential not only for its effect on theirown integrationattitudes,butalso for its implications for minoritygroup preferencesand residentialsearch behavior.Areas perceived as hostile towardparticularminoritygroups are also perceived as less attractive, even when otheraspects of the communitiesshouldbe desirable(Charles2001b). Indeed,blacksopenly admitthatfears of white hostility motivatedesiresfor more thana handfulof coethnic neighbors(Krysan& Farley2002). Althoughthe influence of racial stereotypingis the same for all groups, all three nonwhite groups want substantiallymore integrationthanwhites do. Contraryto the popularadage that "birdsof a feather flock together,"ethnocentrismplays a minimal role at best; moreover,with respectto blacks, the most thoroughand detailedanalysesto date suggest thatwhites move out becauseblacks move in-black densitymatters because the presence of too many blacks (and, to a lesser extent, Hispanics and Asians) suggests a change in "traditionalstatus relations of relative dominance and privilege"(Bobo & Zubrinsky1996, p. 904; see also Charles2000, Krysan 2002).28 In sum, the attitudes,preferences,and (potential) behaviors of whites alone cannotfully accountfor actualresidentialpatterns;all groupsexpress some desire for coethnic neighbors,and these preferencesplay a role in shapingactual outcomes (Clark2002). A much more sizable share of availableevidence, however, points to the influentialrole of racial prejudice,both as a motivatingfactor in the avoidance of particularout-groupsand as a motivatorof minority group preferencesfor same-raceneighbors. DISCRIMINATIONThe institutional practices that created and HOUSINGMARKET maintainedresidentialsegregationrepresentthe translationof white prejudiceinto "systematic,institutionalizedracialdiscrimination"in the housingmarket(Massey & Denton 1993, p. 51; Meyer2000; Yinger1995). A growingbody of empiricalevidence pointsto the persistenceof discriminationin the housing market,although the formthatpresent-daydiscriminationtakes is markedlydifferentthanin previous eras (Bobo 1989; Cutleret al. 1999). Although formalbarriersto integration have been eliminated, discriminatorywhite tastes remain; segregationpersists, it is argued, through a process of decentralizedracism, in which "whites pay more"to live in predominantlywhite areas(Cutleret al. 1999, pp. 445). As such, Theseresultsareconsistentwitha pureprejudiceinterpretation: werenonsignificant. for of blacks whites,thesestatusesincreasethe salienceof widelyheldnegativestereotypes and(potentially) increasethemotivation and/orlikelihoodof actingon theseattitudes. 28Negative simpleout-group racialstereotypes capturetwovariantsof prejudice: hostility group(s); the and"asenseof groupposition" thatmembers of onegrouphaveabout(an)other latteris a collectiveprocesswheregroupsdefinetheirsocialpositionsvis-h-viseachother groupstatusorrelativestatusposition. andthesociallylearnedcommitments to maintaining Whatmattersis themagnitude of differencethatin-groupmembersperceivebetweentheir own groupand particularout-groups(Blumer 1958, pp. 3-4; Bobo 1999). 192 CHARLES discriminatorybehaviorhas become more subtle and thereforemore difficult to detecteven by its victims (Galster1990, 1992;Yinger 1995). Since the mid-1950s, auditstudieshave provenuseful in detectingthese subtleforms of discrimination. In an audit study,pairsof trainedtesters--one white and the othereitherblack or Hispanic-with similareconomic andfamily characteristicssuccessively inquired about housing, carefully detailing their experiences with the real estate agent or landlord.After a visit, each auditorcompletesa detailedreportof her experiences; discriminationis defined as systematicallyless favorabletreatmentof the black or Hispanictester and is documentedby directobservationduringthe interaction (Ondrichet al. 1998). Housingunitsaresampledrandomlyfrommetropolitan-area newspapers;examples of experiencesdetailedby auditorsrange from seemingly race-neutralaspectsof interactionsuch as the promptnessof returningphone calls or volunteeringto show an auditpair additionalunits, to the obviously racial act of steeringminorityauditorstowardmixed or segregatedareas. Despite its advantages,the auditmethodology is not without critics, most notably Heckman& Siegelman(1993) (see also Heckman1998). By samplinghousing units only from majornewspapers,for example,auditstudieslikely underestimate the incidence of discrimination.Otheraspects of the methodrun the risk of overstatingthe frequencyof discrimination.As partof the trainingprocess,auditors arefully informedof the purposeof the studyandas a resultmaybe unintentionally motivatedto findit; similarly,it has been suggestedthatothercharacteristicsof the individualauditorsmay influenceagent behavior(i.e., the presenceor absence of facial hairand/oran accent).Thereis also concernaboutthe use of gross measures of discriminationthatcount "all errorsmade"by agents/landlordsas unfavorable or discriminatorytreatment,arguingthatthis inaccuratelyassumesthatfirmsnever makerace-neutralerrors,andconfoundsrandomand systematiceffects. Heckman & Siegelman (1993, p. 272) suggest beginningwith a net measureof discrimination experiencedby minoritytestersrelativeto theirwhite teammatesbecause (a) it takesrace-neutralerrorsinto account,and(b) if the net measurerevealsevidence of discrimination,the gross measurewill as well. In response, Yinger (1993, 1995, 1998) agrees that audit studies measurediscriminationin a majorsegmentof the housing market-units advertisedin major newspapers-that is accessibleto all homeseekers,irrespectiveof raceor ethnicity; althoughresultscannotbe generalizedto all housing transactions,they do account for a large shareof the action. Concedingthe potentialbenefitsof blind auditsfor avoidingexperimentereffects, proponentsof full disclosurearguefor deliberately informingauditorsof the natureand purposeof the study while at the same time emphasizingthe importanceof accurate,completereportingavoids otherkinds of "experimentereffects." Specifically,some minorityauditorsmay be upsetby blatantmistreatmentandunableto accuratelycompletetheirevaluations,invalidating the audit.Moreover,both membersof an auditteam must receive identicaltraining to minimizebehavioraldifferences.Bringingteammatestogetherwithoutfull disclosure opens the door for (inaccurate)speculationamong the auditorsabout the purposeof the study and/orappropriatebehavior.With respect to aspects of DYNAMICSOF RESIDENTIALSEGREGATION 193 auditors'appearancesor behavior-aside fromrace-that could influenceagents' behavior,more recent audit studies are more careful in the selection of testers, particularlywith respect to the presenceor absence of an accent (Yinger 1995). Finally,althoughsimplegrossmeasuresof discriminationalmostcertainlyoverestimate the frequencyof systematicdiscriminationand should be interpretedas upper-boundestimates of discrimination,net measuressubtractboth randomand systematicdifferencesin treatmentand probablyunderestimatethe frequencyof discrimination.As a consequence,net measuresunderestimatethe gross incidence of discrimination(Yinger 1995, pp. 45-46). Analyses of audit studies generally presentnet measuresfollowed by gross measures.Yingerpointsout thatthe "story told by the simple net measureis bleak enough... in some ways the storymay be even worse"(1995, p. 46). In light of such intense scrutiny,researchon housing marketdiscriminationbasedon auditstudiesis highlyregardedin boththe research and legal communitiesand is now widely accepted for use both as enforcement tools andas evidenceof discriminationin U.S. courts(Metcalf 1988, Yinger1998). Bothnational-andlocal-level studiesfindevidenceof substantialdiscrimination thathas not changed meaningfullyover time (Yinger 1995, 1998). In a review of 50 local audit studies completed throughoutthe United States duringthe 1980s, Galster (1990) concluded that racial discriminationis a dominantfeatureof the housing market,conservativelyestimatingthat(a) housing discriminationagainst black and Hispanic home and apartmentseekers occurs in roughly half of their interactionswith agents or landlords,(b) the discriminationis subtle and difficult for the individualto detect,and(c) the frequencyof discriminationhadnot changed over time (Galster1992, p. 647). These figuresareconfirmedby evidencefromthe 1989 Housing DiscriminationStudy (HDS), the most recent nationalauditstudy. In ClosedDoors, OpportunitiesLost, Yinger(1995) deliversa comprehensiveand influentialdiscussion of housing marketdiscrimination,using HDS datato detail the incidence and severityof discrimination.At the beginningof a transaction,an individualinquires about an advertisedunit and then asks about the availability of other,similarunits, at which time an agent may withhold informationor limit the numberof units shown to the client. In the second stage, the actions taken by agents to facilitate the transactionare considered. These would include the discussion of terms and conditions, the agent's sales effort, and/orassistance in securing financing;at this point, an agent may offer less assistance to minority clients. The third aspect of the interactioninvolves the geographic location of units otherthanthe advertisedunit thatopened the interaction.Access to housing is constrainedif a client is only shown housing in neighborhoodswith particular racial/ethnicmake-ups(Yinger 1995, pp. 31-33). Duringthe firststage of the interaction,Yingerfoundthatblacks andHispanics are denied access to housing between 5% and 10% of the time-information is completely withheld.More often, minorityaccess to housing was constrained: Black andHispanictesterslearnedabout25%fewerunitsthancomparablewhites. Whites were also significantly more likely to receive other forms of favorable treatment,including follow-up calls, positive comments about an availableunit, 194 CHARLES and special rental incentives (e.g., one month's free rent or a reduced security deposit).Minorityauditorssufferedmanyminorinconveniences,includingwaiting longer to be served, inattentionto their housing needs but overemphasison their incomes, and less assistance with obtaining financing.Racial/ethnic steering is also common. Yingerestimatesthatblack andHispanichomeseekersvisiting four real estate agents will encountersteering40% and 28% of the time, respectively, whereaswhites aremore likely to hearnegativecommentsaboutintegratedareas. Racial/ethnicsteeringof this sortis prohibitedby fairhousinglegislation;however, steering throughmarketingpractices is completely legal, and evidence suggests thatrealestateagenciesdo muchof theirsteeringthroughtheirmarketingpractices. Units in black neighborhoodsarenot advertisedas often, have fewer open houses, and are more likely to be representedby firms that are not part of a Multiple Listing Service. This may also be an issue for units in predominantlyHispanic areas.These practicesare the exact opposite of those employed for units in white neighborhoods(Yinger 1995, pp. 55-59). A growingbody of evidence documentsracialdiscriminationin lending as well (Dedman 1988, 1989; Jackson 1994). The Boston Fed Study compares conventionalloan denialratesfor whites, blacks,andHispanicsin Boston using 1990 data fromthe HomeMortgageDisclosureAct supplementedwithothervariablesknown to influence credit decisions. Together,these data offer "themost comprehensive set of credit characteristicsever assembled"(Yinger 1995, p. 71, for details, see Munnell et al. 1996). Results from the Boston Fed Study indicatethatcontrolling for "therisk and cost of defaultand for loan and personalcharacteristics,"blacks and Hispanicsare 56% more likely thanwhites to be denied a conventionalmortgage loan, which amountsto a minoritydenial rate of 17%,comparedto a white rateof 11%.Analysis of the Boston Fed databy Carr& Megbolugbe(1993) found thatminoritiesreceive systematicallylower creditratings.This meansthata "slow paying" white applicant,for example, would be consideredcreditworthy,but a similarblack applicantwould not. Thereis evidence of racialbias in nearlyevery other aspect of the lending process, includingprivatemortgageinsurance,redlining by home insurancecompanies, methods of advertisingand outreach(Yinger 1995, p. 83-85), and bank branchlocations and closing patterns(Caskey 1992), in additionto evidence of an associationbetween the likelihood of blacks' loan approvalandthe racialcompositionof the financialinstitutionworkforce(Squires & Kim 1995). The latterconfirmsevidence thatprejudiceand economic interests motivatebiased behavior(Yinger 1995, Ondrichet al. 1998). More than a decade has passed since the collection of the 1989 HDS. To remedy this, researchersat the Urban Instituteare back in the field for HDS 2000. This updatedstudy promises to be the most ambitiousand thoroughanalysis to date. In additionto black/whiteand Hispanic/whitetests replicatedfrom the 1989 HDS for comparativepurposes,HDS 2000 will ultimatelyincludetests of housing marketdiscriminationagainst both Asians and Native Americans (Turneret al. 2002). Newly releasedPhase I resultsoffer mixed messages aboutchanges in the incidenceof housingmarketdiscriminationagainstblacks andHispanicssince the DYNAMICSOF RESIDENTIALSEGREGATION 195 1989 study,revealingboth improvementand persistentdiscrimination.Improvementsin the sales marketareencouraging;in 2000, bothblacksandHispanicswere significantlyless likely thanthey were in 1989 to receive consistentlyunfavorable treatmentrelative to whites. For blacks, the overall incidence of white-favored treatmentdroppedto 17%in 2000, down 12 percentagepoints over the 10-year period.Despite this overallimprovement,blacks arenow more likely to be steered away frompredominantlywhite neighborhoodsthanthey were 10 years ago. The overall incidence of discriminationagainstHispanicsdeclined by 7.1 percentage points over the decade (to 19.7%)and saw no significantchange in the likelihood of geographicsteering(Turneret al. 2002). The experienceof black and Hispanic renters,however,offers little optimism.Blacks are significantlyless likely to receive unfavorabletreatmentthan in the previous decade; however,the decline is much smaller(9%)thanin the sales market.Moretroublingis thatHispanicrenters show no significantchange in theirlikelihood of receiving unfavorabletreatment relativeto whites, and they now experience a higher incidence of discrimination (26%) thantheirblack counterparts(Turneret al. 2002). In preparationfor the Asian and Native Americanaudits, Phase I of the 2000 HDS includespilot studiesfor bothgroups-Chinese andKoreansin Los Angeles, SoutheastAsians in Minneapolis,and Native Americansin Pheonix. In Los Angeles, resultssuggest thatChineseandKoreanrenters"mayface differentpatterns of adversetreatment"thantheirblack and Hispaniccounterpartsdo (Turneret al. 2002, p. 4.18). Results indicate that both groups are told about and shown more units than their non-Asian minority counterparts;on the other hand, black and Hispanictestersreceivedbetterservice from agents thaneitherAsian group.DiscriminationagainstChineseandKoreansin the sales market,however,is similarto thatof blacksandHispanics-indeed, Koreanhomebuyershavethe highestoverall net estimateof discrimination(22.2%) for any of the minoritygroups studiedin Los Angeles (Turneret al. 2002). Turnerand colleagues (2002) reportedthat local organizationshad difficulty recruitingand retaining Southeast Asian and Native American testers because (a) little testing has been conducted with these groups in the past and (b) few of these testers had any experiencewith homeownershipand found it especially difficult to complete sales audits. These challenges help to inform subsequent tests. As a result, pilot-testresults are only availablefor the rentalmarket,where both groups appearto experience significantdiscrimination(Turneret al. 2002, p. 2.6). SoutheastAsian rentersin Minneapolisface more adversetreatmentthan either of the Asian groups in Los Angeles, and their experiences more closely mirrornationalresults for black and Hispanicrenters,particularlyin the areasof housing availabilityand inspections.Native Americanrentersexperienceadverse treatmentat levels slightly above those found at the nationallevel for blacks and Hispanics(Turneret al. 2002). These exploratoryresultsillustratethe value of fullanalysesof residentialprocesses;we look forward fledgedmultiethnic/multiracial to the completionand disseminationof results from Phases II and III of the 2000 HDS. 196 CHARLES The importantinformationprovidedby national audit studies comparableto both the 1989 and the 2000 HDSs is juxtaposed by significant challenges involved in utilizingthe method(e.g., training,recruiting,andmaintainingminority testers);chief amongthese challengesis, no doubt,the substantialexpense of these studies. Recent researchby Massey & Lundy (2001) offers a lower-cost alternative to the in-personaudit method used in the 1989 and 2000 HDSs: conducting a telephone-basedaudit study of racial discriminationin the Philadelphiarental housing market.Citing evidence that individuals "are capable of making fairly accurateracial attributionson the basis of linguistic cues," the authorsarguethat a good deal of discriminationis likely to occur before a personalencountercan takeplace (Massey & Lundy2001, p. 454). They findthatthis is, in fact, the case: Comparedto whites, blacks were significantlyless likely to speak to the rental agent and, if they spoke to a landlord,significantlyless likely to be told of a unit's availability.Alternatively,blacks were more likely thanwhites to have theircredit worthinessmentionedas a potentialobstacle in qualifyingfor a lease (Massey & Lundy 2001, p. 466).29 Thus, in one way or another,and to a greateror lesser degree, discrimination in the housing marketconstrainsthe ability of nonwhitesto rent and/orpurchase housing. Access to housing is constrained,the searchprocess is more unpleasant (i.e., more visits, more waiting,etc.), homeseekersreceive farless assistancefrom lenders in the mortgageapplicationprocess and are more likely to have their applications denied, and their moving costs are higher.Yinger estimates that every time that black and Hispanic households search for housing-whether they encounterdiscriminationor not-they pay a "discriminationtax" of approximately $3,000. Cumulatively,he estimates that blacks and Hispanics pay $4.1 billion per year in higher search costs and lost housing opportunities.Included in this the 29Thisapproach-usingthe telephoneratherthanface-to-faceinteraction-addresses betweenpairsof testersresultfromunmeasured intreatment difpossibilitythatdifferences conducted byPurnelland andis similarto research ferencesintheirpersonalcharacteristics employedandtheoverallresults. colleagues(1999)withrespectto boththemethodology Similarly,a recentstudyof racialdiscrimination in theChicagoandBostonlabormarkets to respondto advertised positionsby mail.Matchedpairs employedtheauditmethodology of resumesmanipulated theperception of raceusingeitheranobviouslyAfricanAmerican (e.g.,Tamika,Jamal)ora white-sounding possible name(e.g.,Kristen,Brad),eliminating &Mullainathan fromotherexperimenter biasintreatment characteristics. resulting Bertrand thatthismanipulation producesa significant gapin therateof callbacksfor (2002)reported interviews.Specifically,whitenamesreceivedroughly50%morecallbacksthanAfrican American names. Moreover,an additionalmanipulationof qualificationsfound that, for whites, higher-qualityresumes increase callbacksby 30%;for blacks, resume qualityhas no effect on the likelihoodof callback.Resultsof this study,andof Massey & Lundy(2001), supportthe reliabilityof face-to-face audit studies as evidence of discriminationbecause both find ample evidence withoutface-to-face contact. DYNAMICSOF RESIDENTIALSEGREGATION 197 estimate is the decision of 10% of blacks and 15% of Hispanics not to look for housingbecausethey anticipatediscrimination(Yinger1995, pp. 95-103; for more on the impactof anticipateddiscriminationon searchbehavior,see Farley 1996b). By makingit moredifficultfor minoritiesto purchasehousing,discriminationcontributesto racialdisparitiesin homeownershipandwealth accumulation,which in turnfosterpersistingsuburbanresidentialsegregation. THE CONSEQUENCESOF RESIDENTIALSEGREGATION A voluminous body of research documents the powerful influence of place on individuallife chances. Concentratedpoverty neighborhoodsexhibit high rates of long-termjoblessness, out-of-wedlockbirths, school drop-out,crime and social disorder,and lower average wages for those who work (Cutler & Glaeser 1997, Jargowsky1996, Krivo & Peterson 1996, Massey & Denton 1993, Wilson 1987). Withoutsufficientresources,publicservices-particularly publicschoolsdeteriorateas well. Residential segregationis, as detailed at the outset, deeply implicatedin the concentrationof poverty in black communities.Yet, even after accountingfor the social and economic disadvantagesassociatedwith residential segregation,Cutler& Glaeser(1997, p. 865) found that"aone-standard-deviation reductionin segregation(13 percent)would eliminate one-third"of white-black differencesin ratesof high school completion,single parenthood,andemployment as well as earnings. The neighborhoodswhere poor blacks are concentratedare characterizedby extremelevels of disadvantage,and middle-classand affluentblacks are exposed to higherlevels of neighborhooddisadvantagethantheirstatuswould imply (Alba et al. 1994, Massey & Fischer 1999, Pattillo-McCoy1999, Wilson 1987). Suburban blacks are as segregatedas their central-citycounterparts;their suburbs are part of a contiguous set of black neighborhoodsthat are, collectively, the ghetto, differentiatedonly by their status as the "best, mixed, and worst areas" (Galster 1991, Jargowsky & Bane 1991, Logan 2001a, Morenoff & Sampson 1997, Pattillo-McCoy1999). Indeed,the well-knownperils associatedwith ghetto life documentedby quantitativeresearchers(Jargowsky1996, Cutler & Glaeser 1997, Wilson 1987) andethnographerslike Anderson(1990, 1999) andVenkatesh (2000) are found, albeit to a lesser degree, in the neighborhoodsof middle-class blacks (see, for example, Alba et al. 1994, Pattillo-McCoy 1999, Timberlake 2002). In a study of Philadelphia,Massey and colleagues (1987) documented rates of welfare dependence, out-of-wedlock births, and below-averageeducational outcomes several times higher than those of comparablewhites, and the experienceof middle-class blacks is only marginallyimprovedover that of poor blacks. The clusteringof blacks within an expansive ghetto underminesblack homeownership,eitherbecause housing in these areasis unattractiveto residentsor because of difficultiesassociatedwith securinglending;segregationalso undermines 198 CHARLES Hispanic home ownership(Flippen 2001, pp. 354-5).30 The dwellings occupied by blacks (and black Hispanics) are older, of poorerquality,and depreciatedin value relativeto similarwhites (Massey & Denton 1993;Rosenbaum1994, 1996), and differencesin home values andratesof ownershipareimplicatedin persisting black andHispanicwealthdisparities-$414 billion and$186 billion, respectively, relativeto whites (Oliver& Shapiro1995, Yinger 1995). Exposureto crime is also a persistentconcernbecause "eventhe most affluentblacks are not able to escape from crime, for they reside in communitiesas crime-proneas those housing the poorest whites (Alba et al. 1994, p. 427). Thus, whites' worst urbancontexts are betterthan"theaveragecontextof blackcommunities"(Sampson& Wilson 1995, p. 42). Thus, as a consequenceof residentialsegregation,the vast majorityof blacks experience residential circumstancesthat are-to a greater or lesser degreedetrimentalto theirfuturesocial mobility because "anyprocess that concentrates poverty within racially isolated neighborhoodswill simultaneouslyincrease the odds of socioeconomic failure within the segregatedgroup"(Massey & Denton 1993, p. 179). Indeed,in-depthinterviewswith employersrevealthatspace is used as a mechanismfor discriminatingagainstminorityjob applicants(Kirschenman & Neckerman1991, Wilson 1996). A recentstudyof participantsin the Gautreaux program-one of the first scattered-site,low-income housing programs-details substantialimprovementsin the educationaland employmentoutcomes of those who movedwith theirfamiliesfromsegregatedurbanhousingprojectsto predominantlywhite suburbancommunities.Comparedto theircity-dwellingcounterparts, Gautreauxparticipantswere significantlymore likely to be in high school, in a college-prep track,enrolled in a four-yearcollege, employed with benefits, and not outside eitherthe educationalor employmentsystems. Many of the suburban participantssaid that safety contributeda great deal to their success (Rubinowitz & Rosenbaum2000). Ongoingresearchby Massey,Charles,andcolleagues (Massey& Fischer2002, Charleset al. 2002) finds a similarrelationshipbetween neighborhoodviolence and the educationaloutcomes of middle-class and affluentstudents at selective colleges and universities.For both blacks and Hispanics,growingup undersegregatedcircumstancessignificantlylowerslateracademicperformance.The negative effect holds after controllingfor socioeconomic status and is not attributableto differences in school quality or variationsin intellectual, social, or psychological preparationamong students from integratedand segregatedneighborhoods. Apparently,segregationmattersbecause it results in exposureto unusuallyhigh levels of violence while growing up. These studentsare also more likely to experience stressfullife events that lead to greaterfamily stress, poorerhealth,and 30ForHispanics,however,the centralissue is a shortageof housing due to largepopulation growthunderconditionsof segregation.Unlike blacks,however,clusteringin barriosfacilitates ownershipbecause thereis greateraccess to informationand coethnic real estate and lending agents thatbufferhomeseekersfrom discriminationin the largermarket. DYNAMICSOF RESIDENTIALSEGREGATION 199 greaterfamily involvementwhile in college-all of which negativelyimpact academic performance(Charleset al. 2002). Although substantiallybetter off than theirpoorcounterparts,residentialsegregationlimits black andHispanicstudents' ability to reachtheirfull potential. CONCLUSION The past decade has seen a remarkableincreasein our understandingof the processes that maintainracially segregatedneighborhoodsin the United States. Increasing attentionto the multiethniccharacterof many metropolitanareas-and of our nation-has improvedourunderstandingof groupdifferencesin locational returnsto humancapitalandhow the racialattitudesof the fourmajorracialgroups contributeto residentialpatterns.Indeed, with so little informationregardingthe racialattitudesof nonwhitesor of whites towardHispanicsand Asians, this alone is a boon to the study of race relations. The use of audit methodology has forever alteredthe landscapeof discussion regardingdiscriminationin the housing and lendingmarkets,detailingwidespread,currentdiscriminationagainstblacks, Hispanics,andpreliminarilyAsians andNative Americansthatoccursat virtually every point in the searchprocess. The expansionof much of this work to include the four majorracialcategoriesis an advancethatcannotbe underscoredenough. As our nation becomes increasingly"prismatic,"understandingthe dynamics of race relationsand processes of social mobility becomes both more complicated and more important. Logan and colleagues (2002, p. 320) recently lamentedthat "we are near the limit of what can be accomplishedthroughthe analysis of publicly availablecensus data."Theirresearchis particularlyillustrativeof the benefitsof highlighting the nation'sincreasingmultiethnicity;still, locationalattainmentmodels thatrely solely on census datacannotadequatelyassess the effect of massive disparitiesin accumulatedwealth, nor can they accountfor the mannerin which respondents' attitudesinfluencetheirresidentialoutcomes. The majorlimitationto researchof this sort,then, is its inabilityto capturethe dynamicnatureof residentialsegregation. This is a monumentaltaskat the nationallevel, but seems possible in selected metropolitanareas. Futureresearchshould continue to actively engage this complexity. To date, much of what we know regardingthe racial attitudesof Hispanics and Asians is limited to analyses of Los Angeles (and, to a lesser extent, Boston). Regional differencesin settlementpatternsand the vast heterogeneitywithin these broad racial categories make it imperativethat we continue to pursue informationon these groupsandpush past the convenienceof broadracialclassificationschemes. In the area of individual-levelracial attitudes,the factorialexperimentaldesign introducedrecently by Emerson and colleagues (2001) presents a substantial improvementon prior methods and should be pursued.Futureresearch should vary factorsthat they did not (e.g., the presence of children),consider the importance of the characteristicsof surroundingneighborhoods,move beyond biracial 200 CHARLES neighborhoodsand offer the full, multiethniccomplementof neighbors, and incorporatenonwhiterespondents(Emerson2001, p. 932). Qualitativeanalyses--whether from elaborationsto closed-endedquestionsor in-depthinterviews-represent anotherdirectionfor futureresearch.The next logical step in this case is to explore the attitudes,perceptions,andjustificationsof Hispanicsand Asians for theirneighborhoodracialcompositionpreferences;this charis particularlyimportantfor capturingthe importanceof immigration-related acteristics.Ongoingresearchby Charles(unpublishedobservations),for example, details importantdifferencesin the motivationsbehindpreferencesfor same-race neighborsbased on immigrantstatus and English language ability. Results from Phases II and Ill of the 2000 HDS, moreover,will providenew and much-needed informationon the housing marketexperiencesof Asians and Native Americans, and for the first time, the ability to make comparisonsamong nonwhite groups. Whatwe have alreadylearnedfrom the Phase I resultsprovidesvaluableinformation aboutchangesin the experiencesof blacks andHispanicssince the 1989 study. Periodicfollow-upsof nationwidehousingmarketauditstudiessimilarto the 2000 HDS would continueto keep us abreastof the extentto which discriminationin the housing marketpersists. Similarly,updatedanalysis of nationallyrepresentative, multiraciallending marketdata, preferablyat regularintervals, would provide crucial and complementaryinformationon this aspect of the residential sorting process. Finally, as is characteristicof social science research,the tendencyhas been to focus on the problem-segregation. Withoutdoubt,this has been justified, given the deleteriouseffects of segregationon intergrouprelations,social mobility,personal safety,andultimately,effortsto reduceracialinequalityin America.Relative to the body of researchon segregation,however,far too little attentionis paid to understandingthe processes that produceand maintainthe small but meaningful numberof stably integratedneighborhoods.Although not discussed at length in this review,effortsto understandracialresidentialpatternsthatfocus on the comparativelysmall but criticallyimportantnumberof success storiesratherthan our well-known failures are a much-neededbreathof fresh air,remindingus that, althoughfragileandfew andfarbetween,"raciallyintegratedneighborhoodsarenot, as once thought,inevitablydoomed to rapidresegregation"(Ellen 2000, p. 152). To wit, a study of 14 stably integratedurbancommunitiesestimates that as many as 10 million Americansreside in racially/ethnicallydiverse communities, areas defined as having racial/ethniccompositions closest to city racial/ethnic averagesin both 1980 and 1990, althoughmost have been integratedfor longer (Nydenet al. 1998, p. 6). More recently,Ellen (2000) examinesthe characteristics of stablyintegratedneighborhoodsand theirresidents,analyzingdatafor 34 U.S. metropolitanareas.She estimatesthatnearly20% of all U.S. neighborhoodswere racially mixed in 1990; these neighborhoodswere home to 15% of whites and roughlyone thirdof blacks.Moreover,morethan75%of neighborhoodsthatwere integrated(between 10%and50%black)in 1980 remainedso a decadelater(Ellen 2000, p. 1). DYNAMICSOF RESIDENTIALSEGREGATION 201 Both studies find that stably integratedcommunitiestend to be economically diverse, includingmiddle-class,college-educatedhomeownerswith professional occupations, as well as low-income families in entry-level, service-sectorjobs. This economic diversity tends to reflect the presence of varied housing opportunities, includingrentalhousing constitutingat least 25% of housing units. Integratedcommunities also tend to have attractivephysical characteristics(e.g., good location,architecturallyinterestinghomes, and a secureset of neighborhood amenities),places where cross-racialinteractiontakesplace as partof day-to-day life (e.g., grocery stores, schools, parks, or neighborhoodfestivals), and strong community-basedorganizationsand social institutionscommittedto maintaining diversity--either directly or indirectlyby addressingcommunitywide,nonracial service issues (largely schools and safety, but also neighborhoodpreservation) and/orpromotingcross-groupdialogue (Ellen 2000, Nyden et al. 1998). Stable integrationis also more likely in communitiesthataremore distantfrom an area's centralminorityconcentrationandin areaswith smalleroverallblack populations without an intense racial competitionfor housing and widespreadneighborhood change (Ellen2000, pp. 153-54). Whetherby design or by circumstance,residents of these communitiesareboth awareof and value the diversityof theircommunities and work to promoteand maintainit (Nyden et al. 1998), and they are likely to have moretolerantracialattitudes.Clearlythen,futureresearchshouldincrease andexpandanalysesof the "substantialandincreasingminorityof neighborhoods that are currentlyintegratedand likely to stay that way for many years" (Ellen 2000, p. 176). The centralconsiderationof Hispanicswill be crucialhere. Given boththe size andheterogeneityof this group,theirresidentialtrajectorywill have a large effect on overall residentialpatterns.Theirability to integrate-particularly those who are phenotypicallyblack-may introducenew options for increasing the residentialintegrationof non-Hispanicblacks. In the dawnof the new millennium,a color line more complex thanthe one Du Bois describedcontinuesto separateourneighborhoods,maintainingourtendency "to see commonlythe worst of each other"(Du Bois 1990, p. 121) and thwarting the upward social mobility of a substantialportion of our population. Recent efforts to understandthe causes of persistingresidentialsegregationhighlightthe complexity of our emergingmultiethnicworld at the same time that they remind us, matter-of-factly,that race still matters.As the dominantgroup, whites have the luxury of living in relatively affluent,safe neighborhoodswith high-quality schools andservices,even whentheirown financialresourcesarelimited.Although recentimmigrantsmay be initiallydisadvantagedby low socioeconomicstatusand limitedEnglishproficiency,theycan be assuredof graduallymakingtheirway into neighborhoodscomparableto those of whites. As has been the case for much of our history,however,groups racially defined as black continue to face profound barriersto theirquest for the Americandream. The agendafor both social science and public policy should also include the articulationof policy responsesthat are both economically feasible and likely to have wide public appeal.Yinger (1995) offers a set of responses of this type that 202 CHARLES involve attackingracial disparitiesthroughpolicies that address social and economic outcomes "for which the minority-whitedisparitiesare greatest"but are availableto all qualifiedapplicantsregardlessof race. Programsthat supportand encouragelow-income homeownershipand/orassist public schools in poor communitiesaregood examplesof this. Moreover,given the largenumberof minorities and the increasingnumberof whites who arewilling to enterintegratedcommunities, programsthatsupportstableintegrationat all levels of social class shouldalso be pursued.This could include(a) the expansionof Gautreaux-typeprogramsas an alternativeto traditionalpublic housing programs,(b) aggressivepublic relations campaignsandcommunitybettermentprojectsthatpromotethe generalattractiveness of integratedneighborhoods,and(c) affirmativemarketingandpro-integrative mortgageincentivesthatencourageblacksto enterpredominantlywhite areasand whites to enterraciallymixed neighborhoods(Ellen 2000, Yinger 1995). The past three decades have witnessed meaningful improvementin whites' racialattitudesandunparalleledexpansionof the blackmiddleclass. Nonetheless, black-white segregationremains so extreme and its consequences so severe that Denton (1994, p. 74) forcefully concluded"[w]hateverwe are now doing to combat residentialsegregationis not nearlyenough and in manycases is not working at all" (see also Glazer 1980, Massey & Denton 1993, Yinger 1995). In places where Hispanics are heavily concentrated,they may soon confront similar circumstances.We have learneda greatdeal aboutthe dynamicsof racialresidential segregationduringthis period and documentedsignificantdeclines in the degree of residentialseparation.Continuedimprovementis crucialif we areto realizeour full nationalpotential,and futureresearchshouldadvancethe achievementof this most importantgoal. ACKNOWLEDGMENTS The authorwishes to thankRichardD. Alba, LawrenceD. Bobo, ReynoldsFarley, and Douglas S. Massey for their insightfulcomments.The authoralso wishes to thankKimberlyTorresandJessicaWelburnfor theirinvaluableresearchassistance. The Annual Reviewof Sociologyis online at http://soc.annualreviews.org CITED LITERATURE on two Alba RD, DentonNA, LeungSJ, LoganJR. AlbaRD, LoganJR. 1991.Variations 1995.Neighborhood themes:racialandethnicpatternsin the atchangeunderthe conditionsof massimmigration: the New York residence.Demogratainmentof suburban City region, 1970-1990. Int. Migr Rev. phy28:431-53 31(3):625-56 AlbaRD,LoganJR.1992.Analyzinglocational constructing reAlba RD, Logan JR. 1993. Minorityproxattainments: individual-level imity to whitesin suburbs:an individual- gressionmodelsusingaggregate data.Sociol. level analysisof segregation. MethodsRes.20:367-97 Am.J. Sociol. AlbaRD, LoganJR,BellairPE. 1994.Living 98(6):1388-427 DYNAMICSOF RESIDENTIALSEGREGATION with crime: the implicationsof racial/ethnic differencesin suburbanlocation.Soc. Forces 73(2):395-434 Alba RD, Logan JR, Stults BJ. 2000a. The changing neighborhoodcontexts of the immigrantmetropolis.Soc. Forces 79(2):587621 Alba RD, LoganJR,StultsBJ. 2000b. How segregatedaremiddle-classAfricanAmericans? Soc. Probl. 47(4):543-58 Alba RD, Logan JR, Stults BJ, Marzan G, ZhangW. 1999.Immigrantgroupsin the suband urbs:a reexaminationof suburbanization spatialassimilation.Am.Sociol. Rev.64:44660 AndersonE. 1990.Streetwise:Race, Class, and Change in an Urban Community.Chicago: Univ. Chicago Press AndersonE. 1999. Codeof the Street:Decency, Violence,andtheMoralLifeof theInnerCity. New York:Norton BertrandM, MullainathanS. 2002. Are Emily and BrendanMoreEmployablethanLakisha and Jamal? A Field Experimenton Labor MarketDiscrimination Blumer H. 1958. Race prejudiceas a sense of groupposition.Pacific Soc. Rev. 1(1):1-7 Bobo L. 1989. Keeping the linchpin in place: testing the multiple sources of oppositionto residentialintegration.Int.Rev.Soc. Psychol. 2(3):305-23 Bobo L. 1999. Prejudice as group position: micro-foundations of a sociological approach to racism and race relations.J. Soc. Issues 55(3):445-72 Bobo L, ZubrinskyC. 1996. Attitudestoward residentialintegration:perceived status differences, merein-grouppreference,or racial prejudice?Soc. Forces74(3):883-909 Bok D. 1996. The State of the Nation: Governmentand the Questfor a Better Society. Cambridge,MA: HarvardUniv. Press Carr JH, Megbolugbe IF. 1993. The Federal Reserve Bank of Boston study on mortgage lending revisited.J. Housing Res. 4(2):277313 Caskey J. 1992. Bank representationin lowincome and minority urban communities. 203 Work. Pap., Res. Div. Fed. Reserve Bank KansasCity Charles CZ. 2000. Neighborhood racialcomposition preferences: evidence from a multiethnic metropolis. Soc. Probl. 47(3): 379-407 CharlesCZ. 2001a. Socioeconomic status and segregation:African Americans,Hispanics, andAsiansin Los Angeles. In Problemof the Century:Racial Stratificationin the United States, ed. E Andersen,DS Massey,pp. 27189. New York:Russell Sage Charles CZ. 2001b. Processes of residential segregation.In Urban Inequality:Evidence FromFour Cities, ed. A O'Connor,C Tilly, L Bobo, pp. 217-71. New York: Russell Sage Charles CZ. 2002. Comfort zones: immigration, assimilation, and the neighborhood racial-compositionpreferences of Latinos and Asians. Paperpresentedat Annu. Meet. Amer. Soc. Assoc., Chicago Clark WAV. 1986. Residential segregationin Americancities: a review andinterpretation. Popul.Res. Policy Rev. 5:95-127 Clark WAV. 1988. Understandingresidential segregationin American cities: interpreting the evidence, a reply to Galster.Popul. Res. Policy Rev. 8:193-97 ClarkWAV.1992. Residentialpreferencesand residentialchoices in a multiethniccontext. Demography29(3):451-66 ClarkWAV.2002. Ethnic preferencesand ethnic perceptionsin multi-ethnicsettings. Urban Geogr.23(3):237-56 CrowderKD. 1999. Residentialsegregationof West Indians in the New York/NewJersey metropolitanareas:the roles of race andethnicity.Int. Migr.Rev. 33:79-113 CutlerDM, GlaeserEL. 1997. Are ghettosgood or bad? Q. J. Econ. 112(3):827-72 CutlerDM, GlaeserEL, VigdorJL. 1999. The rise and decline of the American ghetto. J. Polit. Econ. 107(3):455-506 DardenJT. 1995. Black residentialsegregation since the 1948 Shelley v. Kraemerdecision. J. Black Stud.25(6):680-91 DedmanB. 1988. The color of money.Atlanta 204 CHARLES Journal,Atlanta Journal Constitution,May 1-4 DedmanB. 1989. Blacks turneddownfor home loans fromS&Lstwice as often as whites.Atlanta Journal,AtlantaJournal Constitution, Jan. 22 Denton NA. 1994. Are AfricanAmericansstill hypersegregated?In ResidentialApartheid: TheAmericanLegacy, ed. R Bullard,C Lee, JE Grigsby,pp. 49-79. Los Angeles: UCLA Cent. Afr. Am. Stud. DentonNA, Massey DS. 1988. Residentialsegregationof blacks, Hispanics,and Asians by socioeconomic status and generation. Soc. Sci. Q. 69:797-817 Denton NA, Massey DS. 1989. Racial identity among CaribbeanHispanics: the effect of double minoritystatuson residentialsegregation.Am. Sociol. Rev. 54:790-808 Du Bois WEB. 1990 (1903). TheSouls of Black Folk: Essays and Sketches.New York:Vintage Books Edmonston B, Passel JS. 1992. Immigration andimmigrantgenerationsin populationprojections. Int. J. Forecast.8(3):459-76 Ellen IG. 2000. SharingAmerica's Neighborhoods: The Prospectsfor Stable Racial Integration. Cambridge, MA: HarvardUniv. Press EmersonMO, YanceyG, Chai KJ. 2001. Does race matter in residential segregation?Exploring the preferencesof white Americans. Am. Sociol. Rev. 66(6):922-35 Farley R. 1996a. The New American Reality: How WeAre, How WeGot There,WhereWe Are Going. New York:Russell Sage FarleyR. 1996b.Racialdifferencesin the search for housing: Do whites and blacks use the same techniquesto find housing?Hous. Policy Debate 7:36786 Farley R, Fielding EL, Krysan M. 1997. The residentialpreferencesof whites andblacks: a four-metropolisanalysis.Hous. Policy Debate 8(4):763-800 FarleyR, Frey WH. 1994. Changesin the segregation of whites from blacks during the 1980s: small steps towarda more integrated society.Am. Sociol. Rev. 59:23-45 FarleyR, SchumanH, Bianchi S, ColasantoD, HatchettS. 1978. Chocolatecity, vanillasuburbs:Will the trendtowardracially separate communitiescontinue?Soc. Sci. Res. 7:31944 FarleyR, SteehC, JacksonT, KrysanM, Reeves K. 1993. Continuedracial segregationin detroit:chocolatecity, vanillasuburbs.J. Hous. Res. 4(1):1-38 FarleyR, SteehC, KrysanM, JacksonT,Reeves K. 1994. Stereotypesandsegregation:neighborhoodsin the Detroit area.Am. J. Sociol. 100(3):750-80 FlippenCA. 2001. Residentialsegregationand minority home ownership. Soc. Sci. Res. 30(3):337-62 Frey WH, Farley R. 1993. Latino, Asian, and black segregationin multi-ethnic metro areas:findingsfromthe 1990 Census.Res. Rep. Popul. Stud. Cent., Univ. Mich. Galster GC. 1988. Residential segregation in American cities: a contraryreview. Popul. Res. Policy Rev.7:93-112 Galster GC. 1990. Racial steering in housing marketsduringthe 1980s: a review of the audit evidence. J. Plan. Educ. Res. 9:165-75 GalsterGC. 1991. Black suburbanization:Has it changedthe relativelocation of races? Urban Aff. Q. 26:621-28 GalsterGC. 1992. Researchon discrimination in housing and mortgage markets: assessment andfuturedirections.Hous. Policy Debate 3(2):639-83 Glazer N. 1980. Race and the suburbs.In The Workof CharlesAbrams, ed. OH Koenigsberger, S Groak, B Bernstein, pp. 175-80. Oxford:PergamonPress Harris DR. 1999. Propertyvalues drop when blacks move in, because...: racial and socioeconomic determinantsof neighborhood desirability.Am. Sociol. Rev. 64:461-79 HarrisDR. 2001. Why are whites and blacks averse to black neighbors? Soc. Sci. Res. 30(1):100-16 HeckmanJJ. 1998. Detectingdiscrimination.J. Econ. Perspect. 12(2):101-16 Heckman JJ, Siegelman P. 1993. The Urban Institute Audit Studies: their methods and DYNAMICSOF RESIDENTIALSEGREGATION findings.In Clearand ConvincingEvidence: Measurementof Discriminationin America, ed. M Fix, RJ Struyk,pp. 187-258, 271-76. Washington,DC: UrbanInst. Press Jackman MR, Jackman RW. 1983. Class Awareness in the United States. Berkeley, CA: Univ. Calif. Press JackmanMR, Muha MJ. 1984. Prejudice,tolerance, and attitudestoward ethnic groups. Soc. Sci. Res. 6:145-69 JacksonWE II. 1994. Discriminationin mortgage lending marketsas rationaleconomic behavior: theory, evidence, and public policy. In AfricanAmericansand the New Policy Consensus:Retreatof the Liberal State, ed. ME Lashley, MN Jackson, pp. 157-78. Westport,CT:GreenwoodPress. JargowskyPA. 1996. Povertyand Place: Ghettos, Barrios, and the American City. New York:Russell Sage JargowskyPA, Bane MJ. 1991. Ghettopoverty in the United States, 1970-1980. In The Urban Underclass, ed. C Jencks, PE Peterson, pp. 235-73. Washington,DC: Brookings Inst. Jasso G, Rosenzweig M. 1990. The New Chosen People: Immigrantsin the UnitedStates. New York:Russell Sage Jaynes GD, WilliamsRM Jr. 1989. A Common Destiny: Blacks in AmericanSociety. Washington, DC: NationalAcademy Press KirschenmanJ, NeckermanKM. 1991. 'We'd love to hirethem,but.. .': the meaningof race for employers.In The UrbanUnderclass,ed. C Jencks, PE Peterson, pp. 203-32. Washington, DC: BrookingsInst. Krivo LJ, PetersonRD. 1996. Extremely disadvantagedneighborhoodsand urbancrime. Soc. Forces 75:619-48 KrysanM. 1998. Privacyandthe expressionof white racial attitudes:a comparison across three contexts. Public Opin. Q. 62:506-44 Krysan M. 2002. Whites who say they'd flee: Who are they, and why would they leave? Demography39(4):675-96 KrysanM, FarleyR. 2002. The residentialpreferences of blacks:Do they explainpersistent segregation?Soc. Forces 80:937-80 205 Logan JR. 2001a. Ethnic diversity grows, neighborhoodintegrationlags behind. Rep. Lewis Mumford Cent. Comp. Urban Reg. Res. Logan JR. 2001b. The new ethnic enclaves in America's suburbs.Rep. Lewis Mumford Cent. Comp. UrbanReg. Res. Logan JR. 2001c. Living separately:segregation rises for children.Rep. Lewis Mumford Cent. Comp. UrbanReg. Res. Logan JR, Alba RD. 1993. Locationalreturns to human capital: minority access to suburban community resources. Demography 30(2):243-68 Logan JR, Alba RD. 1995. Who lives in affluent suburbs?Racial differencesin eleven metropolitanregions. Sociol. Focus 28(4): 353-64 Logan JR, Alba RD, Leung S-Y. 1996. Minority access to white suburbs:a multiregional comparison.Soc. Forces 74(3):851-81 Logan JR, Alba RD, Zhang W. 2002. Immigrant enclaves and ethnic communities in New YorkandLos Angeles. Am.Sociol. Rev. 67:299-322 Logan JR, Schneider M. 1984. Racial segregation and racial change in American suburbs, 1970-1980. Am. J. Sociol. 89(4):87488 Massey DS. 1995. The new immigrationand ethnicity in the United States. Popul. Dev. Rev.21(3):631-52 Massey DS, BittermanB. 1985. Explainingthe paradox of Puerto Rican segregation. Soc. Forces 64(2):306-31 Massey DS, CondranGA, Denton NA. 1987. The effect of residentialsegregationon black social and economic well-being. Soc. Forces 66:29-57 Massey DS, Denton NA. 1985. Spatialassimilation as a socioeconomic outcome.Am. Sociol. Rev.50:94-106 Massey DS, Denton NA. 1987. Trendsin the residentialsegregationof blacks, Hispanics, and Asians: 1970-1980. Am. Sociol. Rev. 52(6):802-25 Massey DS, Denton NA. 1989. Hypersegregation in U.S. metropolitanareas: black and 206 CHARLES Hispanic segregationalong five dimensions. Demography26(3):373-91 Massey DS, Denton NA. 1993. American Apartheid:Segregationand theMakingof the Underclass.Cambridge,MA: HarvardUniv. Press Massey DS, Fischer MJ. 1999. Does rising income bring integration?New results for blacks, Hispanics, and Asians in 1990. Soc. Sci. Res. 28:316-26 Massey DS, FischerMJ. 2002. TheLong Term Consequences of Segregation. Paper presented at Ann. Meet. Popul. Assoc. Am., Atlanta Massey DS, Gross AB, Eggers M. 1991. Segregation, the concentrationof poverty, and the life chances of individuals.Soc. Sci. Res. 20:397-420 Massey DS, Gross AB, Shibuya K. 1994. Migration, segregation and the geographic concentrationof poverty. Am. Sociol. Rev. 59(4):425-45 Massey DS, Lundy G. 2001. Use of black English and racial discriminationin urban housing markets:new methodsand findings. UrbanAff Rev. 36(4):452-69 Metcalf GR. 1988. FairHousing ComesofAge. New York:Greenwood Meyer SG. 2000. As Long as TheyDon't Live Next Door: Segregation and Racial Conflict in AmericanNeighborhoods.New York: Rowman & Littlefield MorenoffJD, SampsonRJ 1997. Violent crime and the spatial dynamics of neighborhood transition:Chicago, 1970-1990. Soc. Forces 76:31-64 MunnellAH, Tootell GMB, BrowneLE, McEneaneyJ. 1996. Mortgagelendingin Boston: interpretingHMDA data. Am. Econ. Rev. 86(1):25-53 MyrdalG. 1972 (1944). AnAmericanDilemma: TheNegroProblemandModernDemocracy. New York:RandomHouse Nyden P, LukehartJ, Maly MT, PetermanW, eds. 1998. Racially and ethnically diverse urbanneighborhoods.In Cityscape 4(2):128, 261-69. Washington,DC: US Dep. Hous. UrbanDev. Oliver ML, Shapiro TM. 1995. Black Wealth/WhiteWealth:A New Perspectiveon Racial Inequality.New York:Routledge OndrichJ, StrickerA, Yinger J. 1998. Do real estate brokerschoose to discriminate?Evidence from the 1989 housing discrimination study.South.Econ. J. 64(4):880-902 Patterson 0. 1997. The Ordeal of Integration: Progressand Resentmentin America's Racial Crisis. Washington,DC: Civitas Pattillo-McCoyM. 1999. Black Picket Fences: Privilege and PerilAmong the Black Middle Class. Chicago:Univ. Chicago Press PortesA, RumbautRG. 1996. ImmigrantAmerica: A Portrait. Berkeley, CA: Univ. Calif. Press Purnell T, IdsardiW, Baugh J. 1999. Perceptual and phonetic experimentson American English dialect identification.J. Lang. Soc. Psychol. 18(1):10-30 RosenbaumE. 1994. The constraintson minority housing choices, New York City 19781987. Soc. Forces 72(3):725-47 RosenbaumE. 1996. The influence of race on Hispanic housing choices, New York City, 1978-1987. UrbanAff.Rev. 32(2):217-43 RubinowitzLS, RosenbaumJE. 2000. Crossing the Class and Color Lines: FromPublic Housing to WhiteSuburbia.Chicago: Univ. Chicago Press SampsonRJ, Wilson WJ. 1995. Towarda Theory of Race, Crimeand UrbanInequality.In Crimeand Inequality,ed. J Hagen,RD Peterson, pp. 37-54. Stanford,CA: StanfordUniv. Press Schnore LF. 1963. Some correlates of urban size: a replication.Am. J. Sociol. 69(2):18593 SouthSJ, CrowderKD. 1997a. Residentialmobility betweencities andsuburbs:race,suburbanization,and back-to-the-citymoves. Demography34(4):525-38 South SJ, CrowderKD. 1997b. Escaping distressed neighborhoods:individual, community, and metropolitanareas. Am. J. Sociol. 103(4):1040-84 South SJ, Crowder KD. 1998. Leaving the 'hood: residential mobility between black, DYNAMICSOF RESIDENTIALSEGREGATION white, and integratedneighborhoods.Am. Sociol. Rev.63:17-26 SouthSJ,DeaneGD. 1993. Race andresidential mobility: individualdeterminantsand structural constraints. Soc. Forces 72(1):14767 Squires GD, Kim S. 1995. Does anybodywho works here look like me?: mortgagelending, race, and lender employment. Soc. Sci. Q. 76(4):823-38 Taeuber KE, TaeuberAF. 1965. Negroes in Cities: Residential Segregationand Neighborhood Change. West Hanover, MA: Atheneum ThernstromS, ThernstromA. 1997. America in Black and White:One Nation, Indivisible. New York:Simon & Schuster TimberlakeJM. 2000. Still life in black and white: effects of racial and class attitudes on prospectsforresidentialintegrationin Atlanta. Sociol. Inq. 70(4):420-45 TimberlakeJM. 2002. Separate,but how unequal? Ethnicresidentialstratification,1980 to 1990. City Community1(3):251-66 TurnerMA, Ross SL, Galster GC, Yinger J. 2002. Discriminationin metropolitanhousing markets:national results from Phase I HDS 2000. US Dep. Hous. Urban Dev., Washington,DC. US Natl. Advis. Comm.Civil Disord. 1988. The KernerReport.New York:PantheonBooks Venkatesh SA. 2000. American Project: The Rise and Fall of a Modern Ghetto. Cambridge, MA: HarvardUniv. Press Villemez WJ. 1980. Race, class, and neighbor- 207 hood: differencesin the residentialreturnon individualresources.Soc. Forces59(2):41430 WaldingerR, BozorgmehrM. 1996. The making of a multiculturalmetropolis. In Ethnic Los Angeles, ed. R Waldinger,M Bozorgmehr,pp. 3-37. New York:Russell Sage Found. WatersMC. 1990. Ethnic Options: Choosing Identities in America. Berkeley, CA: Univ. Calif. Press WatersMC. 1999. BlackIdentities:WestIndian ImmigrantDreams and AmericanRealities. New York:Russell Sage Found. Wilson WJ. 1987. The TrulyDisadvantaged: The Inner City, the Underclass,and Public Policy. Chicago:Univ. Chicago Press Wilson WJ. 1996. WhenWorkDisappears: The Worldof the New Urban Poor. New York: Knopf YingerJ. 1993. Auditmethodology:comments. In Clear and Convincing Evidence: Measurementof Discriminationin America, ed. M Fix, RJ Struyk,pp. 259-70. Washington, DC: UrbanInst. Press Yinger J. 1995. Closed Doors, Opportunities Lost: The ContinuingCosts of Housing Discrimination.New York:Russell Sage Found. Yinger J. 1998. Evidence on discrimination in consumer markets. J. Econ. Perspect. 12(2):23-40 Zubrinsky CL, Bobo L. 1996. Prismatic metropolis:race and residentialsegregation in the city of the angels. Soc. Sci. Res. 25:335-74