Past research has examined the association between preferences for particular racial and ethnic neighborhood compositions, and the composition of the current neighborhood. In this paper, we draw on novel data that allow us to provide a preliminary look at how an important intermediate step—the search for housing—influences these relationships. Overall, our results draw attention to the value of examining the search process that precedes the final choice when studying the correspondence between residential preferences and behaviors in particular, but also as an important component to the perpetuation of racial residential segregation.
In line with earlier research that focused only on the differences between residential preferences and current neighborhood racial/ethnic composition without information on the search itself (e.g., Adelman 2005; Clark 1992), we found that residents generally live in neighborhoods with a larger percentage of their own racial/ethnic group than they report preferring. Yet, our study goes beyond these earlier ones because we find interesting racial and ethnic differences in the stage at which these disparities arise, thus emphasizing the value of studying the search process in relation to residential attainment. Our results suggest that whites experience a mismatch between preferences and behavior at an early stage of the residential process: they generally do not search in neighborhoods that match their preferences. In contrast, blacks—and to a lesser extent Latinos—are more likely to search in neighborhoods that more closely correspond to their preferences in terms of their own racial and ethnic group, but when it comes to where they actually live, their neighborhoods have larger percentages of blacks and Latinos, respectively. The greater degree of match for blacks and Latinos in their search locations and preferences suggests that the explanation for the mismatch between where they prefer and where they actually live should be sought in constraints that prevent the final choice from being the one that matches residential preferences rather than in the choice of search destinations in the first place.
Although our data limit the extent to which we can test a range of theoretical explanations for mismatches at an individual level, our analyses does point to some hints about the sources of the mismatches we observe. For whites, we find that the racial blind spots that Krysan and Bader (2009) identified—such that whites are less likely to be aware of racially diverse neighborhoods—may play a role in the mismatch. That is, white Chicago residents with bigger racial blind spots were the significantly more likely to search in communities that had more whites than they said they prefer. Although education did not predict mismatches, which would have been consistent with a social desirability interpretation, this is a rather weak test. Thus, we suggest that it is still possible that the mismatch for whites at Stage 1 is due in part to whites not being as open to integrated neighborhoods as they say they are. Influenced by social desirability, it may be that whites report a preference for integrated neighborhoods when in fact they favor predominantly white neighborhoods, contributing to a discrepancy between preferences and residential behaviors. That whites do not search in neighborhoods that match their reported preferences is consistent with this interpretation.
However, there are other possibilities apart from social desirability.Footnote 10 For example, it could be that the relative availability of neighborhoods in the metropolitan area that approximate whites’ racial and ethnic preferences is unfavorable. Whites say they prefer a rather diverse neighborhood as long as whites constitute the largest racial/ethnic group. Although these diverse neighborhoods are overrepresented on the search location map (see Table 1), which would imply we overestimate the chances that there will be a match, we see that whites end up living in predominantly white neighborhoods. So, relatively rare as they are, our maps actually increased the possibility for a match between whites’ ideal diverse neighborhoods and being able to report on our maps that they searched in one. Nevertheless, whites (in contrast to blacks and Latinos) live in the most “available” type of neighborhoods, i.e., predominantly white neighborhoods. This means that the relative availability of neighborhoods could play a role in the mismatch we found.
Furthermore, it may be that people tend to start searching close by their current residence (wherever that might be). If this is the case, and in light of patterns of segregation, for whites, these ‘nearby’ neighborhoods are more likely to be white neighborhoods as well (whereas for blacks and Latinos there is a greater chance that the nearby neighborhoods are more diverse). This spatial dynamic of searches therefore precludes more diversity in whites’ search locations. Additionally, it may be that the ‘daily routines’ of whites may be particularly circumscribed, translating into less exposure to diverse neighborhoods and therefore, a lower likelihood that there is sufficient knowledge to seriously consider diverse neighborhoods as destinations.Footnote 11
With respect to the mismatches of blacks and Latinos, we find some suggestion of the role of spatial assimilation for blacks. That is, blacks with an income of less than $20,000 are more likely to reside in a neighborhood with more blacks than in their search locations. These results also lend indirect support to the place stratification model, since the searches may have been unsuccessful due to discriminatory barriers that discouraged blacks and Latinos from residing in the more integrated neighborhoods that they prefer. Although our self-reported measure of housing discrimination did not reveal significant findings, as we noted, we have no measure of actual discrimination experienced by the searchers.
Our study draws attention to a number of directions for future research that put the spotlight on the search process itself as critical to understanding residential outcomes. For example, all we know from our survey data is that a person searched in a particular neighborhood—we do not know what they learned about the neighborhood or what they experienced in it when they searched. It may be, for example, that neighborhoods that are similar to whites’ preferences for more integrated neighborhoods come with race-related factors that stopped them from even searching in these neighborhoods. Indeed, some scholars have asked whether people really have a preference for a particular racial and ethnic composition, or whether these preferences are driven by socioeconomic concerns instead (Harris 2001). Outcomes of vignette experiments indicate that part of the racial preference patterns found in studies using the show card method can be attributed to neighborhood concerns that are related to race, such as poverty, criminality, and bad school quality (Emerson et al. 2001; Krysan et al. 2009; Lewis et al. 2011; St. John and Bates 1990). Nevertheless, the racial and ethnic composition of neighborhoods remained an important and independent factor in shaping residential preferences. In addition, taking into account socioeconomic factors of the neighborhood, Crowder (2000) and Crowder and South (2008) demonstrate that the racial and ethnic neighborhood composition is still an important determinant of whites’ residential mobility. Understanding the manner by which whites determine where to search—and the role of race/ethnicity versus social class—is of great importance.
Overall, in examining the racial and ethnic composition of neighborhoods in which people searched for housing, we seek to highlight the potential role that housing searches play in contributing to the discrepancy between residential preferences and current neighborhood racial and ethnic composition, in particular. More broadly, however, our goal is to bring systematic attention to the role of housing searches to the residential attainment processes and their stratified outcomes. Our data have several limitations, but these suggestive results draw attention to the need to better understand the housing search process and call for the identification of data collection and research design strategies that can address them and provide a more decisive answer to the broader theoretical frameworks that underlie our analyses.
In interpreting our results, several of these shortcomings should be kept in mind. First, we used a non-random sample of 41 communities on our map of the Chicago metropolitan area. For those who prefer white neighborhoods, the proportion of neighborhoods on the map that match their preferences is smaller than they in fact are. Similarly, for those who prefer racially and ethnically mixed neighborhoods, the proportion of neighborhoods on the map is larger than they would encounter in Chicago as a whole. It is hard to determine what consequences of imperfectly measuring the relative racial and ethnic distribution of neighborhoods has for our observed mismatches. Yet, given that a majority of respondents say they prefer very mixed areas, we believe that it most likely reduced mismatches because respondents had more opportunities among the 41 to choose their ideal (which many said was racially mixed).
Second, the use of cross-sectional data means that we cannot draw causal inferences. With respect to the correlate of mismatches, for example, we have measured income at the time of the survey, not at the time of the search; it is very possible that economic constraints were very different at the two times. In addition, with respect to our key measures of interest, information on preferences and current neighborhood were measured at the same time, and information on respondents’ searches (though not their reporting of them) predated statements of preferences. Because of this, we cannot rule out the possibility that residents adjusted their preferences based on their search experiences. Similarly, not only searches, but also people’s current neighborhood might have informed racial residential preferences as they were measured simultaneously in time. This might have affected our results in two ways: it might have either underestimated mismatches in the case of people who could not realize their initial preference but adjusted their attitudes to the situation in the current neighborhood (or based on their search experiences) or overestimated mismatches for those residents who search or reside in a neighborhood that matched their original preferences but—for whatever reason—change their preferences.
Thus, we clearly do not have the appropriate temporal order of preference, search, and move data. We ask respondents to indicate their current preferences, fully aware that their past searches and current residences might influence their current racial preferences. Future research should prospectively investigate residential preferences and where people move. Such a study is obviously complicated by the fact that the very nature of moving makes following respondents difficult; the payoff, however, would be clear insight into an area potentially ripe for intervention—encouraging residents to search in neighborhoods with their stated level of diversity even if it is not on their initial list of search destinations.
This focus on housing searches not only has value in terms of social science debates and understanding of the causes of residential segregation as well as neighborhood selection processes that relate to a range of ‘neighborhood effects’ research, but also there are significant policy implications as well. Indeed, the U.S. Department of Housing and Urban Development recently issued a 5-year Research Roadmap (Office of Policy Development and Research 2013), citing one of HUD’s four programmatic goals to “Build inclusive and sustainable communities free from discrimination.” The Research Roadmap identified a number of top priority research projects to accomplish this goal, including one that examined the housing search process of racial and ethnic minorities. Understanding this process was identified as foundational for a number of core HUD programs and policies, including the Housing Choice Voucher program, housing integration strategies, and discrimination testing and enforcement. As the Roadmap notes,
…HUD does not know how households search for housing and what their preferences are when searching for housing. This research will shed light on how housing search and preference affect HUD’s fair housing/Affirmative Furthering Fair Housing goals and how they promote or deter goals to build inclusive communities (2013, p. 98).
Our research contributes to this policy debate, suggests next steps, and reinforces its importance.
The present study is, therefore, a first step in pointing out the mismatch between stated preferences, search locations, and outcomes. The results compel future research to gather new large-scale longitudinal data in which residential preferences and complete housing search and neighborhood careers can be followed over time to test our causal implications and to take into account the full set of possible residential destinations. It is also important for future studies to link these housing careers to information about individual socioeconomic resources, housing market knowledge and local discrimination to understand more thoroughly the relevance of the housing search for interpreting spatial assimilation, place stratification, and information models. Using an innovative dataset that measured for the first time the search locations of residents of a major metropolitan area, we have drawn attention to the possibility of the mismatch between stated preferences, search locations, and outcomes in general, and the importance of examining more carefully the search process in particular if we hope to better understand the way in which racial residential segregation is perpetuated—or the possibilities for helping to break it down.