Whites Who Say Theyd Flee 675
Demography, Volume 39-Number 4, November 2002: 675696 675
WHITES WHO SAY THEYD FLEE: WHO ARE THEY, AND
WHY WOULD THEY LEAVE?*
Questions have been raised about whether white flightone factor contributing to U.S. resi-
dential segregationis driven by racial, race-associated, or neutral ethnocentric concerns. I useclosed- and open-ended survey data from the Multi-City Study of Urban Inequality to explore whosays they would leave and their reasons for doing so. Thirty-eight percent of white respondents said
they would leave one of the integrated neighborhoods, with Detroiters and those endorsing negativeracial stereotypes especially likely to do so. When asked why they might leave, whites focused on thenegative features of integrated neighborhoods. Expressions of racial prejudice were also common,
but neutral ethnocentrism rare. The results of an experiment asking about integration with Asiansand Latinos are also discussed.
For a neighborhood, it is not a problem how many blacks come into it. The problem ishow many whites go out.
respondent in the Detroit Area Study
ne of todays central debates about the residential segregation process centers on themotivation behind whites opposition to integrated neighborhoods. The traditional inter-pretation is that opposition is motivated by racial prejudice (e.g., Charles 2000; Farley etal. 1994; Zubrinsky and Bobo 1996). But recently, researchers have raised questions aboutwhether whites objections are driven instead by (1) the desire to avoid neighborhoodswith certain characteristics that whites associate with African American neighborhoods(e.g., Harris 1999, 2001; Taub, Taylor, and Dunham 1984) or (2) the desire to live nearones own kind, out of a sense of neutral ethnocentrism (Clark 1992). Which of theseforces underlies attitudes toward white flightand whites residential preferences moregenerallyhas relevance to our understanding of contemporary racial attitudes and resi-dential preferences, as well as significant policy implications. Although demographicanalyses of neighborhood change and residential segregation have routinely made assump-tions about the motives underlying the white-flight process, relatively little empirical evi-dence has been brought to bear on this issue. That is, little is known about what whitesthink of integrating neighborhoodswhat these neighborhoods are like, what they willbecome, and why whites may leave them. This article reports on individual-level attitudi-nal data related to white flight as one way to answer these questions.
White flight, as described by the Detroit-area resident quoted at the beginning of thearticle, occurs when the arrival of African Americans in a neighborhood promptsthe rapid departure of whites, thus turning a community from all white to all African
*Maria Krysan, Department of Sociology (m/c 312), University of Illinois at Chicago, 1007 W. HarrisonStreet, Chicago, Illinois, 60607; E-mail: email@example.com. This research was supported by grants from the RussellSage Foundation, the National Science Foundation (SES 96-18700 and SES 00-95658), and the Ford Founda-tion. The author gratefully acknowledges the Russell Sage Foundation because much of the work for this articlewas completed while she was a visiting scholar at the foundation. Nakesha Faison and Kelly Harr Shomo pro-vided invaluable research assistance, and Kyle Crowder, William P. Bridges, and Howard Schuman gave manyinsightful comments and suggestions on earlier drafts of this article.
676 Demography, Volume 39-Number 4, November 2002
American. To be sure, this process is not the only cause of racial residential segregation.Indeed, its contributions have been hotly contested, with some maintaining its impor-tance (Galster 1990; Goering 1978; Mayer 1960; Quillian 1999; Schelling 1971; Wolf1963; Wurdock 1981) and others downplaying its significance (Frey 1979; Marshall1979; Marshall and OFlaherty 1987; Molotch 1969). More recently, researchers haveoffered more nuanced conclusions that specify the conditions under which white flightdoes and does not occur and have demonstrated that it cannot be ignored as one of themany contributing causes of persistent segregation (e.g., Crowder 2000; Galster 1990;Lee and Wood 1991).
While neither an analysis of neighborhood change nor a complete explication of theconditions under which white flight occurs, this article presents results that have impor-tant implications for the interpretation of these demographic processes. Specifically, Iexamine the often-assumedbut seldom empirically examinedsocial psychological fea-tures of them. By examining both who said they might leave an integrating neighborhoodand the reasons they gave for this decision, this analysis focuses on individual-level atti-tudinal expressions of white flight, although the insights may also apply to the questionof what shapes whites residential preferences more generally.
What Is the Evidence on White Flight?
White flight is a pithy phrase that conjures up images of an inevitable and inexorableprocess of neighborhood change. As a description of one of the facts of urban life, it hasproved to be as controversial and complex as it is catchy. Over the past four decades,researchers have disagreed about the validity of the white-flight hypothesis, with earlywork tending to come down on one side or the other. More recently, recognizing thatwhite flight is neither universal nor inevitable, the emphasis has shifted toward specify-ing the conditions under which it does and does not occur. For example, Lee and Wood(1991) studied integrated neighborhoods in U.S. metropolitan areas and found that pat-terns consistent with white flight varied considerably across region and city: between1970 and 1980, 90% of integrated census tracts in Atlanta and Detroit but just one-thirdin Boston and Los Angeles experienced racial succession. Galster (1990) also highlightedthe contingent nature of white flight in his study of Cleveland, where he found that cen-sus tracts in which segregationist sentiment was the highest (as measured by aggregatepublic opinion) were most likely to undergo neighborhood racial transition.
Studies such as these, which have used the neighborhood as the unit of analysis andhave found, under certain conditions, patterns of neighborhood change that are consistentwith white flight, have come up short as tests of the white-flight hypothesis because theyhave failed to take into account individual-level factors. Put simply, we need to knowwhy people move, not simply that they do so. In his classic work on this topic, Rossi(1955) emphasized that individual characteristics are important predictors of general resi-dential mobility. People move because they get married, get divorced, have children, orbecome empty nesters. They move because they are renters and want to own. Theymove because of their age, gender, or income. To understand white flight, in particular,both the neighborhood context (in this case, its racial composition) and individual charac-teristics must be considered. Neglecting the latter risks overstating the effects of whiteflight because mobility (apart from racially motivated mobility) may eventually result inneighborhood transition (e.g., Frey 1979).
In a study of Nashville, Lee, Oropesa, and Kanan (1994) included both individualfactors and neighborhood context and found that both predicted mobility. They developeda comprehensive conceptualization of neighborhood context that included objective andsubjective perceptions of neighborhood features, as well as static and dynamic
Whites Who Say Theyd Flee 677
dimensions. Their results showed that neighborhood characteristicsparticularly subjec-tive perceptions of theminfluenced mobility behavior indirectly through their influenceon thoughts about mobility. Temporality was important as well: expected changes in thefeatures of the neighborhood were one of the strongest predictors of mobility.
Although Lee et al. (1994) found that objective racial composition did not predictmobility, Crowders (2000) explicit test of the white-flight hypothesis showed that it didmatterbeyond individual characteristics and in the face of other changes in the socialand economic character of the neighborhood. Using individual-level national data,Crowder demonstrated that recent changes in the proportion of African Americans in atract were a significant predictor of white mobility from a tract. In addition, he pointedout that although the substantive impact of neighborhood racial composition is relativelymodest on an annual basis, when compounded across several years, the modest effect cancontribute to significant neighborhood change because of its constant downward pres-sure on white representation.
Research has begun to identify important caveats and nuances with respect to whenand under what conditions neighborhood change and white flight operate, but little isknown about what motivates whites to leave integrating neighborhoods. As Lee et al.(1994) noted, studies of contextual effects fail to answer the question, Why does aparticular context have an influence on a particular behavior? Lee et al. speculated thatthe answer pertains to the degree of threat that the attributes are felt to pose to ahouseholds investment in the residential setting . . . [and] . . . probing this calculusshould rank high among the tasks to be undertaken in future mobility research (p. 265).In my research, I took on this task and asked, What is the calculus at work when whitesexpress a desire to leave an integrating neighborhood? What are the threats and risks thatwhites perceive in an integrating neighborhood? In short, what are the mechanismsthrough which racial context operates on thoughts of mobility? Although the optimalapproach for answering questions about motivations would be to use longitudinal datathat measure both attitudes and behavior at the individual level, these data do not exist.Nevertheless, much remains to be learned from a close examination of the existing atti-tudinal data to shed light on whites expectations about integrated neighborhoods and thereasons they give when they say they would move from them. In this article, I firstexamine what distinguishes whites who are more likely and less likely to state theirintention to leave integrated neighborhoods. Second, and more important, I examinewhat underlies the expressions of white flight by analyzing whites explanations of whythey say they would try to leave.
It is important to recognize that the connection between attitudes and behaviors iscomplex (Schuman and Johnson 1976). By focusing on white-flight attitudes, I do notmean to imply a one-to-one correspondence between, for example, those who tell an in-terviewer they would leave a neighborhood and those who would actually do so. Indeed,some have argued that this connection may be loose with respect to whites intentions toflee (Molotch 1969). Equally inappropriate, however, would be the assumption that atti-tudes and subjective perceptions play no part in decisions about housing and mobilitybehavior. Galsters (1990) study in Cleveland that showed the importance of segrega-tionist sentiment is one such piece of evidence, and Lee et al.s (1994) findings thatsubjective perceptions are important is another.
Finally, white flight is clearly not the onlyor even the mainengine behind racialresidential segregation. It is not simply the individual actions of individual whites withparticular racist attitudes that cause the stubborn patterns of racial residential segregation.As Massey and Denton (1993), among others, persuasively argued, there are considerableinstitutional and structural barriers to racial integration. African Americans are and havebeen routinely and systematically barred from living and purchasing homes in predomi-nantly or overwhelmingly white neighborhoods by practices, such as block busting,
678 Demography, Volume 39-Number 4, November 2002
redlining, racial steering, discrimination in obtaining financing and insurance, and count-less other subtle and not-so-subtle policies and practices (e.g., Massey and Denton 1993;Turner and Wienk 1993; Yinger 1995). At the same time, one such manifestation of thesevarious practices has been the movement of whites out of areas into which African Ameri-cans move and whites decisions not to move into areas that already have more than atoken number of African Americans. In short, white flight is neither the only nor the maincause of residential segregation. And while the evidence points to it as having a contrib-uting roleunder certain circumstancesour understanding of the attitudes underlyingthese behaviors remains incomplete.
Who Says Theyd Flee?
Numerous analyses (e.g., Schuman et al. 1997) have demonstrated the importance of age,education, and gender in shaping whites racial attitudes, especially related to social dis-tance. Thus, white-flight attitudes should also be related to these characteristics. Indeed,analyses of general racial residential preferences have been consistent with this expecta-tion (Bobo and Zubrinsky 1996; Farley, Fielding, and Krysan 1997; Farley et al. 1994).Put simply, women, the young, and the well educated should be less likely to say theywould leave a neighborhood upon the arrival of African Americans than should their lesswell-educated, older, and male counterparts.
The more specific focus of this research suggests four additional factors from theliterature on residential mobility. First, because home owners are more investedat leastfinanciallyin their neighborhoods, they may be more concerned about possible declinesin the quality of their neighborhoods and the potential loss of their investment and there-fore more likely than renters to say they would move in the face of change (Lee et al.1994; Wolf 1963). Second, parents with children under age 18 in their households may bemore concerned about their childrens safety and the quality of their schools and so bemore likely to say they would leave (Harris 1999). Third, those with higher incomes maybe best able to afford to flee integrating neighborhoods (Berry 1979; Crowder 2000)and thus may be more likely than those with lower incomes to indicate they would do so.Fourth, analyses of the same data set used in the present analysis (but that have examineddifferent measures of residential preferences) have shown that whites who hold stereo-typical beliefs about African Americans are less comfortable with the prospect of livingin integrated neighborhoods (Charles 2000, 2001; Farley et al. 1994). Related to this find-ing, however, Bobo and Zubrinsky (1996) and Charles (2001) found that perceived so-cial-class differences do not predict residential preferences.1
Why Do They Say Theyd Leave?
Previous research provided descriptions of the patterns and social and demographic cor-relates of residential preferences (Charles 2000, 2001; Farley et al. 1978, 1994; Zubrinskyand Bobo 1996). But few attitudinal studies have examined white-flight attitudes, in par-ticular, and even fewer have dug beneath expressions of white flight to understand whywhites hold the attitudes they do.2 The importance of doing so is highlighted by the fact
1. The present study differs in at least three ways from other analyses that used the same data set. First,whereas Farley et al.s (1994) study examined data just from Detroit, this study used data from all four cities.Second, Charles (2000, 2001) measured preferences with questions on comfort levels and willingness to moveinto various integrated neighborhoods (as well as a draw your own neighborhood technique). The desire tomove out of an integrating neighborhoodor white-flight attitudesis a different dimension of residentialpreferences that has not been the subject of prior analyses. Third, this study complemented the closed-endedquantitative analysis of residential preferences with a more qualitatively oriented approach in which responsesto open-ended questions about residential preferences were systematically analyzed; this, too, has not been doneto the same degree, if at all, in previous studies.
2. Farley et al. (1978, 1994) conducted some analyses of the reasons whites give to explain their desire tomove from an integrating neighborhood, but the analyses were limited and restricted to the Detroit area.
Whites Who Say Theyd Flee 679
that demographic analyses of white flight have based their conclusions on untested as-sumptions about the basis of whites motivations to leave. At the same time, however, thedebate about whether white flight is an important contributor to residential segregationhas pointed out that it may not be the decision to move out of a neighborhood that is asimportant as what kind of neighborhood whites decide to move into (Crowder 2000; Frey1979; Marshall 1979; Marshall and OFlaherty 1987; South and Crowder 1998). Indeed,both demographic and attitudinal data indicate that whites choose to move into (and pre-fer) communities with no more than a token number of African Americans. But studieshave provided little direct analysis of why whites prefer whiter neighborhoods. So, al-though my study examined attitudes about white flight, the analysis of the reasons whitesgive for why they would leave a particular neighborhood is also likely to be applicable tothe reasons whites would give for why they would choose not to move into such a neigh-borhood in the first place. The research literature suggests three motivations: racial, raceassociated, and neutral ethnocentrism.
Racial reasons. Traditional interpretations of white-flight behavior and attitudes haveidentified racial prejudice as the motivating factor. Generally, racial prejudice is viewedfrom a psychological perspective, and is an individual-level phenomenon involving ex-pressions of antiAfrican American hostility and negative beliefs about other racialgroups. Thus, operationally, racial reasons are those that refer to a dislike or distrust ofAfrican Americans, a simple desire not to live near African Americans, and references tonegative stereotypes about African Americans as a group (see, for example, Allport 1954;Bobo and Zubrinsky 1996; Jackman 1994).
Bobo and Zubrinsky (1996) directed attention to a second, more sociological, con-ception derived from Blumers (1958) theory of prejudice as a sense of group position, inwhich prejudice is not an individuals irrational response, but a reaction of the membersof the dominant group to a threat by the subordinate group. Applied to the case of resi-dential preferences, integration poses a threat to whites dominant position in the neigh-borhood and their sense of what ought to be. Whites want to leave neighborhoods withAfrican Americans because they do not want to be in the minority, because they perceivethat African Americans are taking over, and/or because they view their neighborhoodsas their territories.
Race-associated reasons. Several scholars have argued that white flight is drivennot by racial prejudice, but by concerns about the characteristics of neighborhoods thatcorrelate with its racial composition (Harris 1999, 2001; Taub et al. 1984). That is,whites are motivated to avoid particular neighborhoods because of race-associated char-acteristics, such as crime or neighborhood deteriorationnot the actual race of theirneighbors.
In the context of white-flight attitudes, the distinction between racial and race-associated reasons is problematic because it glosses over the role of subjective percep-tions. Specifically, when a white family is deciding to leave or not, what is important iswhether they believe an integrated neighborhood will become a high-crime, deteriorat-ing, low property-value area. It matters less whether it eventually does become such aplace (Wolf 1963). Thus, from the vantage of attitudes and perceptions, the question of arace-associated explanation for white flight loses the race-neutral veneer that Harris(1999, 2001) and others put on it. Indeed, it becomes difficult to separate race-associatedreasons from what are more explicitly racial reasons, such as stereotypes about AfricanAmericans or antiAfrican American hostility. For example, in a study of Detroit, nearlyhalf the white respondents believed that African Americans are more violent and takeless good care of their properties than do whites (Krysan 1998). These negative stereo-types of African Americans as a group may translate into negative perceptions of inte-grated neighborhoods, such that when whites object to integrated neighborhoods on thebasis of concerns that it isor will becomea bad place to live, they are, to a degree,
680 Demography, Volume 39-Number 4, November 2002
articulating racial stereotypes. Although it is conceptually difficult to separate racial andrace-associated reasons, I maintained this distinction and conducted analyses to deter-mine if they have different features.
Neutral ethnocentric reasons. Clark (1992) and others have argued that whitesobjections to living with African Americans are based on a desire to seek neighborhoodswith their own group, rather than to escape contact with another group. People aredrawn to neighborhoods in which they share a cultural background with other residents,and so it is the pull factor of neutral ethnocentrism that perpetuates segregation, notthe push factor of the presence of another group. By this account, white-flight attitudesare conceived of as benign; inevitable; and devoid of racial prejudice, stereotypes, orconflict.
Multiethnic note. Historically, data on racial residential preferences have been lim-ited to African Americans and whites. Increasingly, however, our nation is better charac-terized as prismatic (Zubrinsky and Bobo 1996) because of the increase in immigrationfrom Asia and Latin America. This data set provides an opportunity also to examine whitesattitudes toward living with Asians and Latinos. Using these data for Los Angeles,Zubrinsky and Bobo (1996) found a racial hierarchy in which African Americans wereconsidered the least desirable neighbors, Asians the most, with Latinos falling in themiddle. More recently, Charles (2000, 2001) used the same data set that I analyzed toexamine the preferences of whites toward living with Asians, Latinos, and African Ameri-cans and showed a similar hierarchy.3 My study extended these studies by asking whetherthe reasons whites give for objecting to neighborhoods with Asians and Latinos differ fromtheir reasons for wanting to leave neighborhoods with African Americans.
DATA AND METHODS
The data came from the Multi-City Study of Urban Inequality (MCSUI), which investi-gated the beliefs, perceptions, and attitudes of large samples of adults in metropolitanAtlanta, Boston, Detroit, and Los Angeles.4 MCSUI used a multistage area-probabilitysample of households in each metropolis, drawing on block-group data from the 1990census as the sampling frame. Face-to-face interviews were conducted in the mid-1990swith a randomly selected adult resident of each sampled household, producing a total of2,810 non-Hispanic white respondents (for details on the study and sample design, seeBobo et al. 2000). Despite efforts to match the races of the interviewers and respondents,some slippage occurred, particularly in Atlanta (where 20% of whites were interviewed byAfrican American interviewers). Because of the evidence that the race of an interviewerinfluences expressions of whites racial attitudes (Finkel, Guterbock, and Borg 1989;Hatchett and Schuman 1976; Schuman et al. 1997) and because whites who were inter-viewed by African Americans were more likely to be living in currently integrated neigh-borhoods, the multivariate analyses controlled for these two characteristics.
To assess the neighborhood preferences of whites, MCSUI used an approach devel-oped by Farley et al. (1978) for the 1976 Detroit Area Study. A series of cards were pre-pared showing neighborhoods ranging from all white to one occupied by eight black andseven white families (see Figure 1). White respondents were first shown the all-whitecard and were asked to imagine that they lived in such a neighborhood. They were thenpresented with cards containing greater densities of blacks. With each card, they wereasked how comfortable they would feel living in such a racially mixed neighborhood
3. Charles (2000, 2001) also examined the preferences of Asians and Latinos toward living with white,African American, Asian and Latino neighbors. Clearly, understanding the reasons underlying the preferencesof these groups is an important topic for future research, but it is beyond the scope of this article. See Krysanand Farley (2002) for an analysis of the reasons underlying African Americans preferences.
4. The Los Angeles questionnaire did not include the would you move out? series of questions, so I wasunable to include it in the full analysis. However, I use a related set of questions in the multiethnic analysis.
Whites Who Say Theyd Flee 681
using a 4-point scale, ranging from very comfortable to very uncomfortable. If therespondents said that they would feel somewhat or very uncomfortable, they were askedif they would try to move out should their own neighborhood come to have a racial com-position similar to the one depicted on the card. If they said they would not try to moveout, they were shown cards with higher densities of blacks until they said they would tryto move out or reached the fifth card showing a majority black neighborhood (with eightblack and seven white families).
The dependent variable was constructed from these questions and identifies the levelof integration at which the respondents indicated they would leave. The cards showneighborhoods with 1, 3, 5, and 8 black families (out of 15), which correspond to neigh-borhoods that are 7%, 20%, 33%, and 53% black, respectively. Although this variablemay be considered ordinal, because of the violation of the parallel regression assumption(Long and Freese 2001), I treated it as nominal and used multinomial logistic regression.The omitted category comprises respondents who indicated they would not be uncom-fortable and/or would not move out of a neighborhood in which they were in the numeri-
Figure 1. Neighborhood Diagrams Used to Measure Whites Residential Preferences in the Multi-City Study of Urban Inequality
Scenario 1 Scenario 2
682 Demography, Volume 39-Number 4, November 2002
cal minority (scenario 5).5 The key independent variables are age, education, income,home ownership, gender, and presence of children, as well as a measure of stereotypes6
and controls for the racial composition of the respondents census-block group and therace of interviewer.
A more textured picture of whites attitudes about integrated neighborhoods is pro-vided through a systematic analysis of the responses to an open-ended follow-up ques-tion. Whites who said they would move out of any one of the neighborhoods were askedto explain, in their own words, why they would move out.7 The verbatim responses to thequestionWhy would you try to move out?Were entered into FoxPro, a relationaldatabase program used to code and analyze the open-ended data. A complex codingscheme with numerous categories was created to capture as fully as possible the range ofresponses, but it was ultimately reduced to the 12 substantive and 3 nonsubstantive cat-egories shown in the tables. A single respondents answer could be coded under an unlim-ited number of themes.
After an initial training period, a research assistant and the author independentlycoded all responses. The initial coder agreement was assessed by calculating the percent-age of cases in which the two coders agreed precisely on how to code the completeresponse. For example, if a particular respondent mentioned three themes in his or heranswer, the two coders each had to identify all three themes for the agreement to beconsidered complete. If the coders deviated on even one code, the case was considered adisagreement. Using this conservative estimate of the coders reliability, the initialagreement rate was 84% in Atlanta, 70% in Boston, and 73% in Detroit. After codingeach response independently, the two coders discussed and resolved any disagreements.
In Los Angeles, whites were asked about living not only with blacks, but withAsians and Latinos. Instead of being asked the would you move questions, however,they were asked about their comfort with different levels of integration and then asked toexplain, in an open-ended question, why they would feel uncomfortable. Despite thisinconsistency, given the paucity of data on whites attitudes toward groups other thanblacks, I briefly discuss these items in a supplemental analysis. To avoid invoking thenorm of evenhandedness, a split-ballot design was used, so that a random one-third ofwhites were asked about integration with either blacks, Asians, or Latinos. Because ofthe small samples, multivariate analyses are not possible.8
5. This measure is clearly truncated because the respondents were not asked about neighborhoods thatwere more than 53% black. Presumably, among the omitted group are those who would leave at some higherpercentage of blacks, as well as others who would state that they would not leave even if they were the onlywhite family left in the neighborhood.
6. The stereotype measure was constructed from four questions asking whites to assess whites and blackson a 7-point scale, where one end is a positive characteristic and the other a negative characteristic (easy versushard to get along with, prefer to live off welfare versus prefer to be self-supporting, intelligent versus unintelli-gent, and speak English well versus speak English poorly). Respondents who gave valid answers on at least twoof the four items were included in the analysis. A respondents score was based on the average difference be-tween the rating a respondent assigned to whites versus blacks on the items for which a respondent provided ananswer. Only 4.7% of the respondents failed to answer all four items: of these, 3.2% did not answer one item,0.9% did not answer two, and just four respondents (0.7%) were excluded because they answered just one ortwo of the items. The stereotype scale ranged from 6 to +6, (+6 means a respondent rated blacks at the mostextreme negative point (7) and whites at the most extreme positive point (1).
7. As with all sensitive questions asked of survey respondents, there is always a concern about the level ofcandor and the extent to which the respondents are able to articulate accurately the reasons behind their statedanswers. The respondents seemed open in answering these questions, but nevertheless the responses should betreated as conservative estimates of the admission of socially sensitive attitudes, such as racial prejudice andstereotyping.
8. Whites in Boston were also asked about integration with Asians and Latinos. Because of the smallsamples, however, I report only briefly on these data.
Whites Who Say Theyd Flee 683
How Many Whites Said They Would Flee?
Overall, 38% of non-Hispanic whites in Atlanta, Boston, and Detroit said they wouldconsider moving out of one of the four neighborhoods with black residents. The percent-age increased as the percentage of blacks in the hypothetical neighborhood increased:from 3% of whites who said they would move out of a neighborhood with a single blackfamily to 19% who said they would move if their neighborhood was majority black. Fur-thermore, 6% and 10% of whites, respectively, said they would move from a neighbor-hood with five black or three black families.
Who Said They Would Flee?
Table 1 reports the multinomial logistic regression results that identify the difference be-tween those who said they would not move from any of the neighborhoods and those whosaid they would. Beginning with those who would leave a neighborhood with a singleblack family, compared with those who would stay, the early leavers were significantlyless educated and more likely to be male. Although none of the other demographic vari-ables was significant, racial attitudes clearly matter: those who said they would leave aneighborhood at the arrival of the first black family were two times more likely to en-dorse negative stereotypes about blacks than were those who would remain in the neigh-borhood at least to the eight blackseven white composition. However, rating blacks aspoorer than whites does not predict departure at this (or any) level.
Turning to the second column in Table 1, education was nonsignificant, but city ofresidence, stereotypes, and the presence of children in the home all influenced the desireto flee at the 20% black level. Detroit residents were nearly four times more likely thanBoston residents to say they would leave this neighborhood than to say they would stayup to the 53% black neighborhood. Detroiters were also more likely than Atlantans to saythey would leave. It is interesting that whites who had children in their homes were halfas likely as those with no children in their homes to say they would move out at this level,which is opposite to the expectation. Finally, the respondents who lived in neighborhoodswith more African American families were significantly less likely to say they wouldleave a 3 black12 white neighborhood. Conclusions about causal direction cannot bedisentangled, however, because it is unclear whether those who live in integrated neigh-borhoods are selective of those who are the most tolerant of integrated living or whetherthe experience of living in a relatively integrated neighborhood decreases hostility anddiscomfort with such an arrangement.
In the third column of Table 1, Detroiters were again significantly more likely to saythey would leave than were Bostoners. And although the coefficients for the middle-income categories are significant, the addition of the series of dummy variables measur-ing income does not significantly improve the overall fit of the model (detailed resultsnot shown).
Not much distinguishes those who said they would leave the 53% black neighbor-hood from those who said they would not leave, as shown in column 4. Atlantans weresignificantly less likely than Bostoners (and Detroiters) to say so. The social context ofthe interview setting was also influential: whites who were interviewed by an AfricanAmerican interviewer were significantly less likely than those who were interviewed bya white interviewer to say they would leave this neighborhood. Finally, the respondentswho more strongly endorsed negative stereotypical beliefs about blacks were, again, sig-nificantly more likely to say they would move out of this neighborhood in which theywould be in the minority.
In analyses of the MCSUI datausing different dependent variablesCharles(2000, 2001) found a similarly consistent effect of stereotypes on residential preferences.
684 Demography, Volume 39-Number 4, November 2002
Thus, the present results demonstrate that the effect of the endorsement of stereotypes isrobust across a variety of dimensions of residential preferences. At the same time, theresults highlight some nuances with respect to which levels of integration prompt aflee response and how the patterns vary for different types of respondents.
Why Would They Leave?
With this understanding of who said they would leave various neighborhoods, I now turnto the words of the respondents themselves to examine whether whites explanations
Table 1. Multinomial Logit Regression (Log Odds) of Whites Attitudes TowardMoving Away From an Integrating Neighborhood: Atlanta, Boston, andDetroit Residents
Move Out at Move Out at Move Out at Move Out at1 Black14 3 Black12 5 Black10 8 Black7
White White White WhiteFamilies Families Families Families
(1) (2) (3) (4)
CityBoston (omitted) 1.00 1.00 1.00 1.00
Atlanta 0.79 0.89 1.61 0.63*
Detroit 1.85 4.40*** 2.75*** 1.43
Background VariablesEducation (in years) 0.87* 0.94 1.00 0.99
Age (in years) 1.02 1.00 1.01* 1.00
Female 0.42** 0.57* 0.80 0.94
Homeowner 1.77 1.61 0.81 1.12
Children in home 1.45 0.52* 1.07 0.83
Family income< 20,000 (omitted) 1.00 1.00 1.00 1.00
$20,00039,999 0.44 0.89 1.67 1.04
$40,00059,999 0.92 1.05 1.35** 1.07
$60,00079,999 0.93 1.14 1.28* 1.13
> $80,000 0.80 0.93 1.15 1.01
Income missing 0.59 1.33 2.01* 0.54*
Control VariablesBlack interviewer 0.17 0.23 0.43 0.45*
Percentage black intract/block group 1.18 0.04* 0.27* 0.44
AttitudesStereotype index 2.08*** 2.36*** 2.03*** 1.62***
Blacks poorer than whites 1.25 1.03 0.94 0.98
Cox and Snell R2 0.245
Nagelkerke R2 0.274
Note: The omitted category is those who said they would stay in the 8 black7 white neighborhood.
*p < .05; **p < .01; ***p < .001
Whites Who Say Theyd Flee 685
reflect racial, race-associated, or neutral ethnocentric concerns. As such, the sample is nowrestricted to only those who said they would move out of one of the four neighborhoods.The three main themes are categorized into several subgroups as shown in Table 2.9 Illus-trative responses are presented in the text, and a detailed coding scheme and descriptionare available from the author. Each respondent could mention more than one reason, so thetotal percentages sum to more than 100%. The number of reasons mentioned by the re-spondents varied across the three cities: Atlanta (1.17), Boston (1.50), and Detroit (1.46).
Racial reasons. Table 2 shows the breakdown of reasons overall and separately bythe neighborhood that prompted the flee response. Traditional prejudice, revealed bothby expressions of racial hostility and by negative stereotypes about African Americans,constitutes a small but not insignificant percentage of the reasons the respondents gave.Overall, about 6% of the respondents said they would leave a neighborhood because of anopposition to integration in general and/or living next to African Americans in particularor made an explicit statement of hostility toward or a lack of trust or dislike of AfricanAmericans. For example, respondents said something like Blacks and whites shouldntmix, I dont trust them, Because Id be uncomfortable. [What do you mean?] Whatcan I say, I just dont like blacks. Another 10% expressed a negative perception aboutthe characteristics of African Americans as a group, as in Theyre too loud; They arelazy and dirty; Because when they move in, they absolutely destroy things; and Theywould beat my kids to deathblack kids would.
Of the three racial reasons, prejudice as a sense of group position was the mostfrequently mentioned: about one in five respondents who would leave one of the integrat-ing neighborhoods pointed to concerns about the relative proportion of African Ameri-cans, that African Americans would take over the neighborhood, that the races would notget along, or that African Americans would discriminate against them in an integratedneighborhood.10 Some such statements included, Because I would be surrounded by ablack majority, Because I would be in the minority. Nothing personal against blacks,Id feel I was being moved out of my neighborhood, and They are taking over myterritory. The prevalence of responses like these suggests that numbers do count forwhites. These respondents viewed their neighborhoods as places where the presence ofanother racial group reflects a sense of a loss of control over their living situations andintegrated neighborhoods threaten their dominant status. Being a minority was clearly anunacceptable and uncomfortable prospect for them.
Columns 25 in Table 2 show that some racial reasons are more common amongthose who object to certain neighborhoods. For example, those who would leave theneighborhood with a token black family were the most likely to report racial hostility:17% of whites, compared with just 4% of those who said they would leave an eightblackseven white family neighborhood. Those who objected to neighborhoods withhigher proportions of black families were more likely to voice concerns about numbers,but concerns about a takeover were most prevalent among those who objected to thebarely integrated neighborhoods.
Race-associated reasons. The race-associated reasons were divided into two cat-egories: descriptions of integrated neighborhoods as having various undesirable charac-teristics and concerns about property values. Clearly, a substantial percentage of whiteswho said they would try to flee had negative images of what an integrated neighborhoodwould be like: it would be crime ridden and have graffiti, drug problems, and poor
9. There are also three nonsubstantive categories: it depends (those who said their decision depended onthe type of person who moved in or what happened upon their arrival), qualify the response (those who noted,for example, that they were not prejudiced themselves or indicated ambivalence in some other way), and noneof the above (those whose answers did not fit into any of the categories shown on the table).
10. This group included only those who expressed concern about proportions, numbers, and takeovers butdid not also express some hostility or negative perceptions of African Americans.
686 Demography, Volume 39-Number 4, November 2002
property upkeep. Examples of statements included The houses will go downhill. Whenone moves in, a lot of things come along with it and Crime seems to increase whenblack families move in. Nearly 4 out of 10 white respondents perceived a number ofrisks to the quality of life in an integrated neighborhood. Of the neighborhood features,
Table 2. Reasons Why Whites Would Move Away From an Integrated Neighborhood: Total Sampleand by Neighborhood They Would Move Away From: Atlanta, Boston, and Detroit Resi-dents (Percentages)
Move Move Move MoveTotal 1/14 3/12 5/10 8/7 p Value
Reason (1) (2) (3) (4) (5) (6)
Racial ReasonsTraditional prejudice: racial
hostility 5.5 16.7 10.6 2.3 3.9 .000
Traditional prejudice: negativecharacteristics of AfricanAmericans 10.4 12.5 13.5 12.6 7.8 .216
Prejudice as sense of groupposition 20.5 16.7 26.9 19.5 19.5 .339
Too many (without negativeperceptions) 4.3 0.0 1.9 5.7 4.9 .191
Dont want to be minority(without negativeperceptions) 6.0 0.0 1.9 6.9 7.8 .043
Takeover 7.6 14.6 17.3 5.2 4.5 .000
Wont get along 3.9 2.1 5.8 4.0 3.6 .689
Race-Associated ReasonsCharacteristics of integrated
neighborhoodsa 39.0 37.5 42.3 38.5 38.3 .896
Crime 23.5 20.8 18.3 23.0 26.0 .416
Other qualities of theneighborhood 21.5 18.8 27.9 23.0 18.8 .235
Property values 33.8 27.1 42.3 39.7 28.6 .013
Neutral EthnocentrismWant to be with own kind 6.2 8.3 6.7 3.4 7.1 .364
Cultural differences 4.4 8.3 1.9 4.0 4.9 .322
Nonsubstantive ThemesNone of the above 11.7 14.6 6.7 9.8 14.0 .169
It depends 19.9 12.5 15.4 19.5 22.7 .209
Qualify the response 7.9 8.3 6.7 6.3 9.1 .704
n 634 48 104 174 308
Note: These results are for only those who said they would move out of one of the neighborhoods. More than one reasonmay be coded for a respondent, so the totals sum to over 100%. In the sixth column, p values are based on a chi-square test ofthe statistical significance of the relationship between mentioning a particular theme and the level of integration from which arespondent indicated he or she would leave.
aThis category includes respondents who were coded under either crime or other qualities of the neighborhood, butrespondents could be double counted, so the individual percentages do not sum to the total.
Whites Who Say Theyd Flee 687
crime was the most prominent. In addition, concerns about property values droppingwere widespread: one-third of the respondents who said they would leave said somethinglike this: I wouldnt want to lose house value or The market value of our propertywould drop quick.
Across neighborhood types, there was no difference in the percentage of respon-dents who mentioned negative characteristics of neighborhoods. This finding suggeststhat although thresholds differ (the respondents said they would leave at different lev-els), once individuals reach their thresholds, there is consistency about what people thinkwill happen to the neighborhood. Concerns about property values, however, variedacross neighborhoods: those who objected to the three and five black-family neighbor-hoods were more likely to give this reason than were those who objected at both thehigher and lower levels.
Neutral ethnocentrism. Responses in this category refer to a stated desire not toavoid African Americans, but to live with ones own kind 11 or to live in a neighbor-hood where the respondents would share a cultural background with the other residents.As is clear from Table 2, responses of this type were a relatively small percentage. Just6% and 4%, respectively, said something like, Nothing in common with the neighbors,Feel more drawn to people of my own race, and Not many of my own peoplearound, There was no difference by type of neighborhood for the mention of these twothemes.
Stereotypes of people versus neighborhoods: Are they different? Stereotypes fig-ured prominently in these results. Not only was the use of stereotypes one of the strongestpredictors in the quantitative model, but at least three different types of stereotypes wereprevalent in the open-ended responses. Some stereotypes were explicitly about AfricanAmericans:
I hate to tell you this and talk about people. I just wouldnt want to live with that many
black people. Some dont keep up their houses. They have a tendency not to care for what
belongs to them. Go two blocks and youll see a huge eyesore. And you know its them
that live there. I like my [black] neighbor. I like him, but he doesnt keep up the house.
When they say there goes the neighborhood, they mean it. They steal in the stores.
They litter. They destroy.
Some referred not explicitly to African Americans, but to the neighborhoods in whichAfrican Americans live:
When a neighborhood becomes majority black, then conditions develop, such as prob-
lems of crime, quality of education declining, quality and condition of housing declining.
I wouldnt want to live in a rundown neighborhood with drugs, violence, crime, and thats
how the neighborhood would change.
Finally, others mentioned stereotypes about property values:
My perception would be that the property values would decrease based on historical pat-
terns in the Detroit area. I dont know if thats fact, but thats my perception.
Property values would go down. Economics is everything. Its the way society is, not my
personal feelings for individuals.
11. Any respondent who expressed both a desire to live among his or her own kind and a negative percep-tion of African Americans was not included in this category because the respondents in this instance explicitlyconnected their desire to be around their own group with beliefs about particular negative characteristics aboutanother group.
688 Demography, Volume 39-Number 4, November 2002
Although all three types of responses can be construed as stereotypes, the first wascoded as a racial reason, while the other two were treated as race associated. I nowturn to a series of analyses that examined whether these three types of stereotypes havedistinct features.
Because the respondents could give more than one reason, some respondents mayhave mentioned one, two, or all three stereotypes; thus it is useful to identify the degreeto which there is overlap. Such a comparison reveals a hierarchy: whereas 82% of whiteswho mentioned a negative characteristic of African Americans also mentioned a negativeneighborhood characteristic, just 22% of those who mentioned negative neighborhoodcharacteristics also gave a more explicit negative description of African Americans as agroup. And fully 70% of those who mentioned property values did not provide any othernegative characterization of integrated neighborhoods. Thus, those who gave the mostblatant racial stereotypes (about African Americans themselves) were also likely to char-acterize the neighborhoods negatively, but those who gave the property-values stereo-type were unlikely to point to other negative qualities of either African Americans or theneighborhoods.
In light of this overlap and the conceptual distinction between racial and race-associated responses, it is useful to construct a hierarchical variable in which respon-dents are assigned to just one category on the basis of a system that gives priority to theracial responses, then race-associated responses, and finally responses reflecting neutralethnocentrism. The overall distribution of this new hierarchical variable is shown in col-umn 1 of Table 3, with the priority matching the order in which the reasons are presentedin the table.
Using this hierarchical variable, I first examine the relationship between the reasonsin the open-ended question and the closed-ended stereotype scale. The stereotype scale
Table 3. Reasons for Leaving an Integrating Neighborhood by Level of Stereotyping: Hierarchi-cal Coding for Atlanta, Boston, and Detroit Residents (Percentages)
Negative 1.25 2.25Through Through Through
Total 1.00 2.00 6.00 p ValueReason (1) (2) (3) (4) (5)
Traditional Prejudice: Racial Hostility 5.6 5.0 5.6 6.2 .877
Traditional Prejudice: NegativeCharacteristics of African Americans 9.4 4.5 9.8 14.4 .003
Prejudice as Sense of Group Position 19.0 20.5 15.8 21.0 .327
Characteristics of IntegratedNeighborhoods 27.5 25.0 27.4 30.3 .488
Property Values 21.4 22.7 25.1 15.9 .064
Want to Be With Own Kind 4.3 5.0 4.7 3.1 .595
Cultural Differences 1.1 2.3 0.0 1.0 .077
None of the Above 11.7 15.0 11.6 8.2 .100
Total 100.0 100.0 100.0 100.1a
n 630 220 215 195
Note: In the fifth column, p values are based on a chi-square test of the statistical significance of the relationship betweenmentioning a particular theme and a respondents score on the closed-ended stereotype scale.
aColumn does not sum to 100.0 because of rounding.
Whites Who Say Theyd Flee 689
is divided into three groups in Table 3: (1) those who perceived almost no differencebetween blacks and whites, (2) those who endorsed the negative stereotypes to a smalldegree, and (3) those who strongly endorsed the stereotypes. The respondents who men-tioned characteristics of African Americans in the open-ended question scored higher onthe stereotype scale than did those who did not. The respondents who mentioned prop-erty values (and not either of the other two stereotypes) showed a relationship in theopposite direction: those who scored low on the stereotype scale were more likely tomention property values. Mentions of other negative characteristics of the neighborhoodwere unrelated to the quantitative measure of stereotypes.12
Table 4 examines the relationship between whites reasons for saying they would leaveand their level of education. In addition to shedding light on the distinctiveness of the threestereotypes, it also adds information about whether the respondents with more and lesseducation (who did not show large differences in whether they said they would leave)nevertheless thought about this issue differently. The results show that the respondentswith less education were significantly more likely to express hostility and dislike of Afri-can Americans and to report negative characteristics of African Americans themselves.Concerns about declining property values, in contrast, were more popular among the re-spondents with more education, even when home ownership and income were controlledfor (detailed results not shown). In addition to a substantive interpretation of these varying
Table 4. Reasons for Leaving an Integrating Neighborhood by Level of Education: HierarchicalCoding for Atlanta, Boston, and Detroit Residents (Percentages)
012 1315 16 orTotal Years Years More Years p Value
Reason (1) (2) (3) (4) (5)
Traditional Prejudice: Racial Hostility 5.5 8.3 3.7 2.1 .013
Traditional Prejudice: NegativeCharacteristics of African Americans 9.5 13.5 8.5 2.1 .001
Prejudice as Sense of Group Position 18.9 18.8 21.3 16.1 .489
Characteristics of IntegratedNeighborhoods 27.6 26.1 29.3 28.7 .707
Property Values 21.3 13.2 23.4 35.7 .000
Want to Be With Own Kind 4.4 6.3 2.7 2.8 .094
Cultural Differences 1.1 1.3 0.5 1.4 .668
None of the Above 11.7 12.5 10.6 11.2 .799
Total 100.0 100.0 100.0 100.1a
n 634 303 188 143
Note: In the fifth column, p values are based on a chi-square test of the statistical significance of the relationship betweenmentioning a particular theme and a respondents level of education.
aColumn does not sum to 100.0 because of rounding.
12. At the same time, whites who rated blacks as poorer than whites were more likely to mention propertyvalues as their reason for wanting to move out than were those who perceived no differences in social class, butnone of the other themes showed such a relationship. However, after I controlled for the respondents education(it was those who were least educated who were most likely to deny the existence of social-class differences),the effect was reduced to nonsignificance.
690 Demography, Volume 39-Number 4, November 2002
relationships with education, these patterns may be due to differences by education in thesusceptibility to social desirability pressures.
A multiethnic note. Table 5 compares the responses of Los Angeles whites to integra-tion with three groups: blacks, Asians, and Latinos. Recall that the Los Angeles surveyused a slightly different question, asking not about moving out, but about how comfortablethe respondents would be in various types of neighborhoods. As Zubrinsky and Bobo(1996) reported using the same data, whites considered blacks the least desirable neigh-bors, Asians the most, and Latinos in the middle.13
But do whites object to integration with blacks, Asians, and Latinos for differentreasons? For example, if whites are driven by neutral ethnocentrism, then the reasonsthey give when objecting to the three groups should be similar. But, as Table 6 indicates,Los Angeles whites showed marked differences across groups. They were about twice aslikely to make disparaging stereotypical comments about the characteristics of Latinosthan about Asians or African Americansa difference that is statistically significant.The nature of the stereotypes about the three groups also differed. Among those whonegatively characterized Latinos, about equal percentages referred to beliefs aboutgangs, crime, and violence and poor property upkeep. Also mentioned to a lesser degreewere concerns that Latinos are noisy and have too many children. The Los Angeleswhites stereotypes about African Americans were less variablemainly related to Afri-can Americans not keeping up their property or having an attitude problem (a chip ontheir shoulders). Finally, negative stereotypes about Asians differed from those of bothof the other groups: the modal response was that Asians are not friendly, stick to them-selves, or are uninterested in integration.
Los Angeles whites also had different perceptions of what a neighborhood would belike if it were integrated with blacks or Latinos versus Asians. Over twice as many per-ceived that neighborhoods with Latinos and blacks would suffer from high rates of crime,lower property values, and other negative qualities as those who felt this way about neigh-borhoods integrated with Asians. The problem with Asian neighborhoods, according tothese whites, are cultural differencesparticularly expressed as language concerns.
Table 5. Percentage of Los Angeles Whites Who Said They Would Be Uncomfortable in a Neigh-borhood With Various Racial Compositions, by Race/Ethnicity of the Target Group
Black Hispanic AsianNeighborhood Racial Composition Neighbors Neighbors Neighbors
Not Uncomfortable in Any Neighborhood 44.2 64.6 73.0
Uncomfortable in an 87 Neighborhood 25.3 17.7 15.0
Uncomfortable in a 510 Neighborhood 13.7 9.7 8.0
Uncomfortable in a 312 Neighborhood 10.0 4.0 1.0
Uncomfortable in a 114 Neighborhood 6.8 4.0 3.0
Total 100.0 100.0 100.0
n 249 277 300
13. Boston also included a split-sample experiment comparing responses to integration with Asians,Latinos, and blacks using the move out question. When contemplating integration with blacks, 33% of thewhite respondents in Boston said they would leave; this figure dropped to 23% when the group arriving in theirneighborhood was either Latino or Asian. The one distinction that the Boston whites made between Asians andLatinos was that they said they would be likely to move when they faced a smaller proportion of Latino neigh-bors compared with Asian neighbors.
Whites Who Say Theyd Flee 691
DISCUSSION AND CONCLUSION
During the late 1960s and 1970s, demographic analyses of residential segregation centeredon the debate about whether it was white flight or some other natural mobility processthat drove neighborhood racial turnover. Recently, Crowder (2000:242) highlighted thatan either-or perspective is unsupported by the data and that although mobility decisionsare shaped by a variety of individual and contextual level factors independent of size of theblack population . . . local racial conditions remain a salient influence on the mobility ofwhite householders after these other factors are taken into account.
The first section of this article examined the characteristics of those who were mostlikely to state that they would participate in white flightthat is, that they would leave aneighborhood that was undergoing integration. To begin with, it is clear that locationmatters: Detroiters stood out as much more likely to say they would leave.14 This is animportant finding, given that much of the research on this topic, until recently, has relied
Table 6. Reasons Why Los Angeles Whites Would Be Uncomfortable in an Integrated Neighbor-hood With Black, Hispanic, or Asian Neighbors (Percentages)
Black Hispanic AsianNeighborhood Racial Composition Neighbors Neighbors Neighbors p Value
Racial ReasonsTraditional prejudice: racial hostility 4.3 1.0 2.5 .313Traditional prejudice: negative characteristics
of African Americans/Hispanics/Asians 12.2 26.5 13.6 .010Prejudice as sense of group position 30.9 32.7 39.5 .418
Too Many (without negative perceptions) 6.5 5.1 8.6 .635Dont want to be minority (without
negative perceptions) 12.2 10.2 13.6 .779Takeover 6.5 12.2 13.6 .165Wont get along 8.6 7.1 12.3 .468
Race-Associated ReasonsCharacteristics of integrated neighborhoods 43.9 41.8 17.3 .000
Crime 30.9 23.5 9.9 .002Other qualities of the neighborhood 25.9 28.6 11.1 .012
Property values 15.1 12.2 2.5 .014
Neutral EthnocentrismWant to be with own kind 6.5 5.1 9.9 .439Cultural differences 12.9 22.4 38.3 .000
Nonsubstantive ThemesNone of the above 11.5 5.1 6.2 .155It depends 25.2 12.2 7.4 .001Qualify the response 7.2 3.1 7.4 .342
Total n 139 98 81
Note: These results are for only those who said they would be uncomfortable in one of the neighborhoods. More than onereason could be coded for a respondent, so the totals sum to over 100%. In the final column, p values are based on a chi-squaretest of the relationship between the target group (African Americans/Hispanics/Asians) and mentioning a particular theme.
14. The addition of the two dummy variables representing city improved the overall fit of the model (de-tailed results not shown).
692 Demography, Volume 39-Number 4, November 2002
on data from Detroit (Farley et al. 1978, 1994). The current results suggest that racialsuccession is more likely and will happen more quickly in Detroita finding consistentwith Lee et al.s (1994) cross-city comparisons on actual patterns of racial succession.Indeed, when whites in Detroit were asked to explain their reasons, there were hints ofthe citys history in this regard. Additional analyses of the open-ended data (not shown)revealed that 1 in 10 Detroiters who said they would flee pointed to either personalexperience with racial turnover in their own or relatives neighborhoods or to direct ob-servations of declining property values, racial turnover, and/or neighborhood decline.This finding is significantly (p < .001) different from the findings for residents in Atlantaand Boston, where just 1% referred to local experiences or observations.
Consistent with what others have shown using the MCSUI data, but drawing on dif-ferent dimensions of residential preferences (Charles 2000, 2001; Farley et al. 1994), thestrongest predictor of white-flight attitudes is whether or not a person holds negative ste-reotypical beliefs: whites who view blacks as having negative characteristics are muchmore likely to indicate they will leave an integrating neighborhood. This finding goesagainst assertions that race, per se, is not a factor in residential preferences (e.g., Harris1999, 2001) and is consistent with Emerson, Yancey, and Chais (2001) findings from asurvey experiment that show that the proportion of African Americans in a neighbor-hoodeven after race-associated characteristics are controlled forhas an independenteffect on whites assessments of neighborhoods.
Finally, the respondents educational attainment influences only the decision to leaveor stay at the first sign of blacks; beyond that, it is not significant. This finding contra-dicts much research on racial attitudes, which has usually shown that education plays astrong and consistent role in predicting social distance from African Americans. Thesedata suggest that education does not influence the desire to flee an integrating neighbor-hood but does make respondents more tolerant of token integration. Respondents withdifferent levels of education, however, differ in how they explain why they would leave,an issue I turn to now.
Recently, the debate about white flight and residential preferences in general hasraised questions about the underlying motivations. The evidence is complex, and the re-sults in this article highlight that complexity. To be sure, white-flight attitudes are shaped,in part, by explicit racial prejudice. When directly asked to explain why they thought theywould flee, in addition to the small percentage of whites who explained that they didnot like or trust African Americans and held negative stereotypes about blacks, a substan-tial percentage of whites expressed prejudice in the sense of group position by stating thatthey preferred not to be in a minority or preferred not to be around so many of anotherrace. This finding is consistent with the findings of other studies that used differentmethodsalthough, in some cases, the same data as in this studythat have highlightedthe importance of the group-position interpretation of prejudice (Bobo and Zubrinsky1996; Charles 2000, 2001; Zubrinsky and Bobo 1996).
An entirely neutral ethnocentric interpretation of residential preferences is also un-supported by the data, given that only a small proportion of whites explained their atti-tudes in terms that can be construed as a cultural explanation (Clark 1992), at least inrelation to integration with blacks.15 When whites in Los Angeles were asked about living
15. Even some of the neutral ethnocentric reasoning can be construed as racial because the respondentswho indicated a desire to be with their own peoplepeople with whom they have something in common, notwith others who are so differentmay at least have implicitly expressed negative perceptions of the othergroup and attributed more desirable features to their own group. Indeed, as was noted in the text, many respon-dents who used such phrases as I want to be with my own kind quickly followed that remark with a negativecharacterization of African Americans or integrated neighborhoods. Some respondents who did not spontane-ously explain their attraction to their own group in this way may have revealed such notions, if probed.
Whites Who Say Theyd Flee 693
with Latinos and Asians, however, concerns about cultural differenceswhich wereexpressed mainly in terms of language differenceswere salient.
The most common explanations whites gave were race associated. Although thisfinding seems at first glance to be consistent with Harriss (1999, 2001) arguments, theanalyses reported in this article highlight the problems with interpreting race-associatedexplanations as having little to do with the race of ones neighbors and more to do withwhat the neighborhoods are actually like. First, one racial and two race-associatedthemes can be construed as different versionsalbeit more and less blatant onesofnegative racial stereotypes. The respondents stereotyped African Americans, the qualityof their neighborhoods, and the prospects for dropping property values in integratedneighborhoods. Not only did these responses blur into one anothera single respondentmight express all threebut it can be difficult to distinguish between, for example, astatement that African American neighborhoods have certain features and a statementthat African Americans do certain things. Indeed, it is difficult to imagine if and how arespondent who stereotyped an integrated (African American) neighborhood couldexplain it, apart from assuming that it is the group (African Americans) that bringsthese features to the neighborhoods. In short, the distinction between negative stereo-types of African Americans and of African American neighborhoods may be entirelysemantic.
There are, however, clear differences among the respondents who used these differ-ent reasons. Negative characteristics of African Americans as a group were more likely tobe mentioned by those with less education and those who also endorsed negative stereo-types on the closed-ended questions. Negative characteristics of neighborhoods were un-related both to education and to the use of stereotypes. Finally, concerns about propertyvalues were expressed most often by those with more education and those who denied theexistence of group differences on the stereotype scale.
One interpretation of the pattern of results is that the three types of stereotypes repre-sent more and less blatant expressions of what are, nevertheless, fundamentally negativeracial attitudes. Both the questions used to construct the stereotype scale and the sponta-neous mentions of negative traits of African Americans in the open-ended question areblatantly racial responses. That the two are statistically related, then, is no surprise. Bycontrast, to say that integrated neighborhoods have certain undesirable features only im-plies that it is blacks fault. And a reference to falling property values is even more subtle.Indeed, it was easy for the respondents who gave this answer to distance themselves fromholding negative racial attitudes because the forces of economics and the ways of soci-ety could be blamed. Indeed, as one respondent put it, Because it means the propertyvalues are going down. And Im not prejudiced. In short, the respondents could expressa concern about property values without feeling that they were racist. This finding is con-sistent with the finding that the respondents who scored the lowest on the stereotype scalewere also more likely to give this answer.
The relationship between education and the various stereotypical responses is alsoconsistent with this interpretation. Previous research has shown that respondents withhigher levels of education are both more susceptible to social desirability pressuresagainst admitting to negative racial attitudes (Krysan 1998) and more skilled at articulat-ing their racial groups interests in subtler ways (Jackman and Muha 1984). Thus,although the explicitly racial reasons (negative stereotypes of African Americans them-selves) are more likely to be given by those with less education, it is the most subtle (andleast obviously racial) stereotype about property values that is most likely to be given bythose with more education. Somewhere in the middle are responses that mention otherstereotypical features of neighborhoods, which show no relationship to education. Inshort, although education serves to generate more subtle rational racial responses, inthe end, each of the reasons is an articulation of a racial stereotype. To be sure, these
694 Demography, Volume 39-Number 4, November 2002
stereotypes are embedded in a system of residential segregation that both emanates fromand perpetuates these beliefs. As Yinger (1995:122) noted:
[R]esidential segregation is one outcome of a complex system in which prejudice, seg-
regation, discrimination, and racial or ethnic economic disparities are simultaneously de-
termined. Each of these phenomena influences the others. Because of their complexity,
these relationships are difficult to study, but most scholars now recognize that racial and
ethnic prejudice and discrimination are both causes and consequences of residential seg-
From a theoretical perspective, the interpretation of the race-associated reasons as amotivation for white-flight attitudes is further complicated by the fact that, for whiteflight, it is the perception, not the reality, that is crucial. Indeed, the more subtle nature ofthe race-associated reasons makes them more insidious because they appear to be ratio-nal and not susceptible to the charge of racism. But the implications for holding thesekinds of perceptionsor stereotypesabout the quality of integrating neighborhoods andtheir property values are neither nonracial nor purely a function of the correlates of theobjective characteristics of these neighborhoods. Thus, Harriss (1999, 2001) assertionsthat certain residential preferences are not racialized because they are linked to neighbor-hood characteristics is problematic because what matters for white-flight attitudes is notthe reality of these neighborhood characteristics but, rather, whites expectations aboutwhat will happen to such neighborhoods.
As Lee et al. (1994) noted, it is the fears, risks, and threats that whites hold aboutneighborhoods that are integrating (or already integrated) that is important. Thus, al-though the perceptions may be grounded in subtle racial biases and are articulated in aseemingly benign race-associated fashion, they have potentially concrete and blatant im-plications: the perpetuation of racial residential segregation. To be sure, future studiesare needed to investigate more directly the connection between attitudes about whiteflight and subsequent mobility decisions and actual white flight (Crowder 2000). Fornow, this study has moved along efforts to gain a more nuanced and complex under-standing of the motivations underlying white-flight attitudes, in particular, and racialresidential preferences, in general.
Allport, G.W. 1954. The Nature of Prejudice. New York: Addison-Wesley.
Berry, B.J.L. 1979. The Open Housing Question: Race and Housing in Chicago, 19661976. Cam-
bridge, MA: Ballinger.
Blumer H. 1958. Race Prejudice as a Sense of Group Position. Pacific Sociological Review 1:37.
Bobo, L., J. Johnson, M. Oliver, R. Farley, B. Bluestone, I. Browne, S. Danziger, G. Green, H.
Holzer, M. Krysan, M. Massagli, and C. Zubrinsky Charles. 2000. Multi-City Study of Urban
Inequality, 19921994: [Atlanta, Boston, Detroit, and Los Angeles] [Household Survey Data]
(3rd ICPSR version) [Computer file]. Mathematica, Atlanta GA; University of Massachussets,
Survey Research Laboratory, Boston, MA; University of Michigan, Detroit Area Study and In-
stitute for Social Research, Ann Arbor, MI; University of California, Survey Research Program,
Los Angeles [Producers]. Inter-university Consortium for Political and Social Research [Dis-
Bobo, L.D, and C. Zubrinsky. 1996. Attitudes on Residential Integration: Perceived Status Differ-
ences, Mere In-Group Preference, or Racial Prejudice? Social Forces 74:883909.
Charles, C.Z. 2000. Neighborhood Racial-Composition Preferences: Evidence From a Multieth-
nic Metropolis. Social Problems 47:379407.
. 2001. Processes of Racial Residential Segregation. Pp. 21771 in Urban Inequality:
Evidence From Four Cities, edited by A. OConnor, C. Tilly, and L.D. Bobo. New York: Russell
Whites Who Say Theyd Flee 695
Clark, W.A.V. 1992. Residential Preferences and Residential Choices in a Multiethnic Context.
Crowder, K. 2000. The Racial Context of White Mobility: An Individual-Level Assessment of the
White Flight Hypothesis. Social Science Research 29:22357.
Emerson, M.O., G. Yancey, and K.J. Chai. 2001. Does Race Matter in Residential Segregation?
Exploring the Preferences of White Americans. American Sociological Review 66:92235.
Farley, R., E. Fielding, and M. Krysan. 1997. The Residential Preferences of Blacks and Whites:
A Four Metropolis Analysis. Housing Policy Debate 8:763800.
Farley, R., H. Schuman, S. Bianchi, D. Colasanto, and S. Hatchett. 1978. Chocolate Cities, Va-
nilla Suburbs: Will the Trend Toward Racially Separate Communities Continue? Social Sci-
ence Research 7:31944.
Farley, R., C.G. Steeh, M. Krysan, K. Reeves, and T. Jackson. 1994. Stereotypes and Segregation:
Neighborhoods in the Detroit Area. American Journal of Sociology 100:75078.
Finkel, S.E., T.M. Guterbock, and M.J. Borg. 1989. Race of Interviewer Effects in a Pre-Election
Poll: Virginia 1989. Public Opinion Quarterly 55:31330.
Frey, W.H. 1979. Central City White Flight: Racial and Nonracial Causes. American Sociologi-
cal Review 44:42548.
Galster, G.C. 1990. White Flight From Racially Integrated Neighborhoods in the 1970s: The
Cleveland Experience. Urban Studies 27:38599.
Goering, J.M. 1978. Neighborhood Tipping and Racial Transition. Journal of the American In-
stitute of Planners 44:6878.
Harris, D.R. 1999. Property Values Drop When Blacks Move In, Because . . .: Racial and Socio-
economic Determinants of Neighborhood Desirability. American Sociological Review 64:461
. 2001. Why Are Whites and Blacks Averse to Black Neighbors? Social Science Research
Hatchett, S. and H. Schuman. 1976. White Respondents and Race of Interviewer Effects. Public
Opinion Quarterly 39:52328.
Jackman, M.R. 1994. The Velvet Glove: Paternalism and Conflict in Gender, Class and Race. Ber-
keley: University of California Press.
Jackman, M. and M. Muha. 1984. Education and Intergroup Attitudes: Moral Enlightenment, Su-
perficial Democratic Commitment, or Ideological Refinement? American Sociological Review
Krysan, M. 1998. Privacy and the Expression of White Racial Attitudes: A Comparison Across
Three Contexts. Public Opinion Quarterly 62:50644.
Krysan, M. and R. Farley. 2002. The Residential Preferences of Blacks: Do They Explain Persis-
tent Segregation? Social Forces 80:93780.
Lee, B.A., R.S. Oropesa, and J.W. Kanan. 1994. Neighborhood Context and Residential Mobil-
ity. Demography 31:24970.
Lee, B.A. and P.B. Wood. 1991. Is Neighborhood Racial Succession Place-Specific? Demogra-
Long, J. S. and J. Freese. 2001. Regression Models for Categorical Dependent Variables Using
Stata. College Station, TX: Stata Press Publication.
Marshall, H. 1979. White Movement to the Suburbs: A Comparison of Explanations. American
Sociological Review 44:97594.
Marshall, H. and K. OFlaherty. 1987. Suburbanization in the Seventies: The Push-Pull Hypoth-
esis Revisited. Journal of Urban Affairs 9:24962.
Massey, D.S. and N.A. Denton. 1993. American Apartheid: Segregation and the Making of the
Underclass. Cambridge, MA: Harvard University Press.
Mayer, A. 1960. Russell Woods: Change Without Conflict. Pp. 198220 in Studies in Housing
and Minority Groups, edited by N. Glazer and D. McEntire. Berkeley: University of California
696 Demography, Volume 39-Number 4, November 2002
Molotch, H.L. 1969. Racial Change in a Stable Community. American Journal of Sociology
Quillian, L. 1999. Migration Patterns and the Growth of High-Poverty Neighborhoods, 1970
1990. American Journal of Sociology 105:137.
Rossi, P.H. 1955. Why Families Move. Glencoe, IL: Free Press.
Schelling, T.C. 1971. Dynamic Models of Segregation. Journal of Mathematical Sociology
Schuman, H. and M.P. Johnson. 1976. Attitudes and Behavior. Annual Review of Sociology
Schuman, H., C.G. Steeh, L.D. Bobo, and M. Krysan. 1997. Racial Attitudes in America: Trends
and Interpretations, revised edition. Cambridge, MA: Harvard University Press.
South, S.J. and K. Crowder. 1998. Leaving the Hood: Residential Mobility Between Black,
White, and Integrated Neighborhoods. American Sociological Review 63:1726.
Taub, R.P., D.G. Taylor, and J.D. Dunham. 1984. Paths of Neighborhood Change: Race and Crime
in Urban America. Chicago: University of Chicago Press.
Turner, M.A. and R. Wienk. 1993. The Persistence of Segregation in Urban Areas: Contributing
Causes. Pp. 193219 in Housing Markets and Residential Mobility, edited by G.T. Kingsley
and M.A. Turner. Washington, DC: Urban Institute Press.
Wolf, E. 1963. The Tipping-Point in Racially Changing Neighborhoods. Journal of the Ameri-
can Institute of Planners 29:21722.
Wurdock, C. 1981. Neighborhood Racial Transition: A Study of the Role of White Flight. Urban
Affairs Quarterly 17:7589.
Yinger, J. 1995. Closed Doors, Opportunities Lost: The Continuing Costs of Housing Discrimina-
tion. New York: Russell Sage Foundation.
Zubrinsky, C.L. and L.D. Bobo. 1996. Prismatic Metropolis: Race and Residential Segregation in
the City of the Angels. Social Science Research 25:33574.
/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 149 /GrayImageMinResolutionPolicy /Warning /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 150 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 599 /MonoImageMinResolutionPolicy /Warning /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False
/CreateJDFFile false /Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ]>> setdistillerparams> setpagedevice