Abstract
This study examines how patterns of racial and ethnic segregation in U.S. metropolitan areas vary by household structure. Specifically, using tract-level summary files from the 2000 decennial census, we estimated levels of metropolitan segregation for different racial and ethnic groups by household composition and poverty status. We find that when using the dissimilarity index, white households with children, and especially poor ones, are more segregated from black, Hispanic, and Asian households than are white households as a whole. Results from the interaction index provide complimentary information. In large part because nonpoor white married-couple households are more numerous than other groups in most metropolitan areas, such households tend to have relatively less interaction with other racial and ethnic groups, and black and Hispanic households in particular. In contrast, minority group members often live in neighborhoods with a high proportion of non-Hispanic white households. Among all three minority group families with children, nonpoor married householders had the highest levels of interaction with whites. These results show that household structure shapes racial and ethnic residential patterns in U.S. metropolitan areas.
Keywords: racial and ethnic residential segregation, racial and ethnic inequality, household structure
Black-white residential segregation in U.S. metropolitan areas, while still high in absolute terms, has declined substantially over the past few decades. This suggests at least some easing of the stark racial divide between blacks and whites in American life. However, this decline may be less dramatic for some groups within these populations. For example, some research has indicated that black and white children in schools and neighborhoods are more segregated than blacks and whites overall (Logan et al. 2001; Logan 2004). In addition, research on neighborhood change suggests that white households with children are amongst the least likely to live in integrated neighborhoods (Ellen 2007).
Researchers have long noted that life course considerations play an important role in household residential decisions, as residential mobility is often a response to housing needs that accompany changes in household composition (Rossi 1955; Speare, Goldstein, and Frey 1975). For example, married couples may desire more space than single individuals. In addition to space, families with children are also likely concerned about particular kinds of neighborhood amenities, such as good schools, parks, and safe spaces (Rosenbaum and Friedman 2001). The implication is that white households with children may be especially averse to minority neighborhoods because of the perception (whether unfounded or not) that such neighborhoods have fewer amenities, higher crime, or inferior public schools (Harris 1997; Ellen 2007).
Despite these suggestive findings that household structure influences patterns of racial and ethnic residential segregation, there are relatively few studies to date that directly focus on this issue. The goal of our study is to therefore provide a detailed, mainly descriptive, analysis of how racial and ethnic residential segregation varies by household structure in U.S. metropolitan areas. The following questions guide our analysis: To what extent are households with children, and particularly white households with children, more racially and ethnically residentially segregated than other kinds of households, such as non-family households? Does household composition play a different role in the segregation of white households from others than it does in the segregation of black, Hispanic, and Asian households? Do socioeconomic status and marital status of the householder also play a role in shaping these patterns?
To answer these questions we use data from the 2000 decennial census and calculate different measures of residential segregation in all metropolitan areas where various groups are present in sufficient numbers. We examine the segregation of whites, blacks, Asians, and Hispanics from alternative reference groups and calculate levels of segregation by household poverty status and the marital status of the householder. In doing so, we aim to get a better sense of not only the role that race plays in shaping residential patterns, but how race interacts with household composition in producing the levels of racial and ethnic segregation we see today.
Background
Black-white segregation in American metropolitan areas remains quite pervasive, though it has declined significantly over the past few decades. Hispanic-white and Asian-white segregation levels are lower than black-white ones, though they did not decline in the same fashion over the same time period, in large part due to continued immigration that is reinforcing immigrant enclaves (Iceland 2009; Iceland and Scopilliti 2008; Iceland and Nelson 2008).
A number of theories have been offered to explain these patterns of racial and ethnic residential segregation in the U.S. Two broad perspectives that encompass many of these views have been termed spatial assimilation and place stratification. According to the spatial assimilation model, differences in socioeconomic status (SES) and, among immigrants groups, acculturation, shape patterns of segregation. Minority members may simply not be able to afford to live in the same neighborhoods as more affluent whites. Indeed, research has shown that people of different classes are segregated from one another, even if only moderately so (Clark 2007; Fischer et al. 2004; Iceland, Sharpe and Steinmetz 2005; Iceland and Wilkes 2006).
In contrast to the spatial assimilation model, the place stratification perspective emphasizes the role of prejudice and discrimination in shaping residential patterns. Segregation is the tool used by whites to maintain social distance from other minority groups. The effects of structural barriers are thought to be greatest for blacks in the U.S. because blacks have historically been perceived in the most unfavorable terms (Bobo and Zubrinksky 1996; Charles 2000). Despite some declines in discrimination in recent years, many believe that both it and white avoidance of mixed or minority neighborhoods still play central roles in shaping the residential patterns of various ethnic groups in the United States (Charles 2005; Squires and Kurbin 2006; Ross and Turner 2005).
While these two general approaches to racial and ethnic residential segregation address its underlying social causes, they say little about how these larger mechanisms are manifested differently in residential patterns according to household structure.
Household Composition
Though research has explored how household composition and life course factors such as age, marital status, and the presence of children affect one’s propensity to move (e.g. Landale and Guest 1985; Rossi 1955; Speare, Goldstein, and Frey 1975), the residential segregation literature is thin in its discussion of how these considerations affect racial and ethnic residential patterns. There are at least two reasons to believe that household composition may affect black and white residential decisions differently. First, white (and perhaps Asian) households may avoid minority neighborhoods out of concern for their children. Second, white and Asian married couple households with children tend to be the most affluent and thus able to avoid neighborhoods that they perceive to be less desirable.
Regarding the first explanation, some literature suggests that white parents may avoid neighborhoods with a greater proportion of minority families, and black families in particular. This could be due not only to prejudice, or “racial reasons” for residential segregation, as Krysan (2002) refers to them, but also to beliefs (real or imagined) about school quality and safety associated with the racial composition of areas—what Krysan (2002) terms “race-associated” reasons. Harris (1997, 1999, 2001) terms these “race-proxy” reasons. In contrast, non-family households may care less about these kinds of neighborhood amenities and may be more open to living in more diverse areas often in and around the urban core.
The second reason why household composition may interact differently with residential decisions for white and black households relates to socioeconomic differences by race and household composition. Generally speaking, socioeconomic status may enable more white and Asian families to act on their preferences for particular types of schools and other neighborhood amenities (Holme 2002). Because white and Asian families with children, particularly those headed by a married couple, are generally of higher socioeconomic status than other types of households, they may be the most likely to be able to afford to live in neighborhoods with the most amenities. These neighborhoods are typically higher-cost, less diverse suburban settings, given the considerable racial and ethnic socioeconomic differentials in the U.S. Conversely, black and Hispanic families, which are more likely to be headed by lower-income single parents, tend to be less able to afford to live in high amenity neighborhoods (Johnson 2006; Ellen 2007).
The two explanations offer alternative hypotheses on residential patterns. According to the first model, White (and perhaps Asian) family households with children may avoid more integrated neighborhoods due to either prejudice (racial factors) or notions of neighborhood quality associated with race (“racial proxy” factors). If this is the case, white and Asian households with children will display higher levels of segregation from black and Hispanic group members than other white and Asian households even after controlling for income.
According to the second model, household structure in and of itself does not influence residential preferences. Rather, it is mainly that more white and Asian family households have higher incomes that enable them to live in more affluent and less diverse areas than black and Hispanic households. According to this perspective, we expect to find that, controlling for income, racial and ethnic segregation patterns will not vary by household type.
Our analysis builds on existing studies in a few important ways. First, unlike most of them, we focus on the interaction between race and household composition. Second, most previous studies focus on segregation or mobility patterns in a white-black context. We show how segregation patterns differ when alternative racial and ethnic reference groups are considered, including Hispanic and Asian households. Third, unlike many previous studies, we pay close attention to the potential mediating role of poverty in shaping residential patterns, given differences in the prevalence of poverty across racial and ethnic groups in the U.S.
Data and Methods
We use Summary File (SF) 3 data from the 2000 Census to examine the residential patterns of households by race and household composition across neighborhoods in U.S. metropolitan areas. Our analyses consist of a series of cross-tabulations showing the association between race/ethnicity, household structure, and patterns of residential segregation. First, we calculate segregation indexes for all households by race and Hispanic origin. Then we calculate indexes for different household types: married-couple households with children, single parent households with children, and non-family households.1 A final key computation is segregation by household composition and poverty status. These comparisons will allow us to examine: 1) net differences in racial and ethnic segregation by household composition and 2) whether differences in poverty status help explain any of these differences.
Households are classified by the race/ethnicity of the householder. Based on these questions and the data available in SF 3 tabulations, we examine the segregation of 4 different racial/ethnic groups: those who marked that they were non-Hispanic white alone, black alone, Asian alone, and Hispanic (who can be of any race). Segregation indexes are computed for groups that have at least 1,000 in a particular metropolitan area—as segregation indexes for small populations are less reliable than those with larger ones (Iceland, Weinberg, and Steinmetz 2002). Our analyses include only counts of households and thus exclude people in group quarters (such as prisons, dorms, or other kinds of group houses). We present segregation estimates averaged across metropolitan statistical areas (MSAs), primary metropolitan statistical areas (PMSAs), and for New England states, New England county metropolitan areas (NECMAs) together referred to hereafter as metropolitan areas (MAs), as defined by the Office of Management and Budget (OMB) on June 30, 1999. Using this definition, there are 318 MAs in the U.S. To examine the distribution of different groups across neighborhoods within metropolitan areas, we use census tracts, which have, on average, 4,000 people.
Though we computed three measures of residential segregation—dissimilarity, interaction, and information theory (or entropy), we present results for just two: dissimilarity and interaction. Residential patterns are quite similar when using dissimilarity and information theory because they are both measures of evenness. The dissimilarity index has the property of not being sensitive to the relative size of the groups in question. It merely provides information on how evenly members of groups are distributed across neighborhoods. In contrast, the interaction index is sensitive to the relative size of the groups being studied. Other factors being equal, larger groups will have lower interaction scores than smaller ones simply because there are fewer other-group members present with which to share neighborhoods.
Limitations
Several limitations of our analyses are worth noting. First, the analysis is largely descriptive, as it consists of presenting patterns of residential segregation by race/ethnicity, household type, and poverty status. With the neighborhood-level data available, we simply cannot look at the role that additional variables (such as alternative socioeconomic status indicators) may play in shaping different groups’ residential patterns. Poverty is a somewhat blunt measurement tool, as are some of our household indicators. Second, the cross-sectional nature of the data imposes limitations on claims of causality—we are mainly examining general relationships between variables. Third, due to data limitations, we cannot look at neighborhood-level residential patterns of mixed-race households by household composition and poverty status. In addition, while SF 3 data have relevant detailed population counts for non-Hispanic white householders, the black and Asian counts include those who are non-Hispanic and Hispanic. This overlap means that we do not compute levels of segregation between Hispanics and blacks or Asians because it is inappropriate to include duplicate counts of households in the two groups being compared in a segregation index.
Despite these limitations, we believe this analysis will provide an important baseline of descriptive information about the racial and ethnic residential patterns of different kinds of households. The existing literature is extraordinarily thin in this area. We hope that our effort will spur further research in this area.
Results
Table 1 shows residential segregation patterns of non-Hispanic white households from black, Hispanic, and Asian households using the dissimilarity index. For each reference group (i.e., column) calculation, we use a constant set of metropolitan areas where all groups are present in sufficient numbers (at least 1,000 group members). We do this to ensure that differences in mean segregation scores across different household types are not merely a function of different sets of metropolitan areas being captured in the calculations.2
Table 1.
Dissimilarity of non-Hispanic White Households by Household Structure and Poverty Status, 2000
| Composition of the white household | Reference Group
|
|||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Households with a Black householder | Households with a Black householder of the same composition | Households with a Hispanic householder | Households with a Hispanic householder of the same composition | Households with an Asian householder | Households with an Asian householder of the same composition | |||||||
|
| ||||||||||||
| Number of metro areas | D Index | Number of metro areas | D Index | Number of metro areas | D Index | Number of metro areas | D Index | Number of metro areas | D Index | Number of metro areas | D Index | |
| All households | 121 | 0.600 | 37 | 0.634 | 121 | 0.435 | 46 | 0.476 | 106 | 0.438 | 13 | 0.445 |
| All married-couple households with children | 121 | 0.657 | 37 | 0.644 | 121 | 0.505 | 46 | 0.555 | 106 | 0.480 | 13 | 0.520 |
| Poor | 121 | 0.683 | 37 | 0.796 | 121 | 0.567 | 46 | 0.730 | 106 | 0.631 | 13 | 0.792 |
| Not Poor | 121 | 0.660 | 37 | 0.641 | 121 | 0.509 | 46 | 0.539 | 106 | 0.481 | 13 | 0.512 |
| All single-parent households with children | 121 | 0.591 | 37 | 0.680 | 121 | 0.437 | 46 | 0.549 | 106 | 0.488 | 13 | 0.622 |
| Poor | 121 | 0.606 | 37 | 0.730 | 121 | 0.486 | 46 | 0.637 | 106 | 0.587 | 13 | 0.802 |
| Not Poor | 121 | 0.607 | 37 | 0.673 | 121 | 0.451 | 46 | 0.554 | 106 | 0.484 | 13 | 0.644 |
| All non-family households | 121 | 0.566 | 37 | 0.607 | 121 | 0.411 | 46 | 0.420 | 106 | 0.427 | 13 | 0.446 |
Source: Census 2000 Summary File 3. Means are unweighted. Metropolitan areas with at least 1,000 members in the group of interest and reference group by table column are included in the calculations. The reference group for the family-level index calculations includes only family households (rather than all households).
In table 1 we see that the overall unweighted dissimilarity score between black and white households is 0.600 in the 121 metropolitan areas where there were over 1,000 people in each group shown. As a rule of thumb, D scores below 0.3 are low, from 0.3 to 0.6 are moderate, and above 0.6 are high. The table shows considerable variation by household type. The dissimilarity between white married-couple households with children and all blacks is 0.657, and the scores are high regardless of the poverty status of the white married-couple household.3
The second set of columns shows black-white segregation among households of the same composition and poverty status. This column includes a more restricted set of metropolitan areas— 37 in which there were over 1,000 blacks and whites of all the household/poverty status groups shown. Among these, we first see that overall black-white segregation is nominally higher (0.634) than the score in the first column (though the difference is not statistically significant in this case, as the number of metropolitan areas under consideration is fairly small), consistent with previous literature indicating that larger metropolitan areas in the U.S. (those where there are large numbers of black and whites of all household types) generally have higher levels of segregation than smaller ones (Iceland, Weinberg, and Steinmetz 2002). The dissimilarity scores in this column also show that poor white married-couple households with children are especially highly segregated from poor black married-couple households with children (0.796). In contrast, among nonpoor white and black married-couple households with children, the D score, while quite high, is nevertheless substantially lower (0.641). Among single-parent households with children, levels of segregation are quite high, though again poor households are more segregated from each other (0.730) than nonpoor ones (0.673). These numbers all suggest that SES—or at least poverty status—is at least moderately related with residential patterns: nonpoor white and black households are considerably less segregated from each other than poor white and black households.
The next set of columns show Hispanic-white segregation scores. Overall white-Hispanic segregation (0.435) is lower than white-black segregation (0.600). The segregation of white married-couple households with children from Hispanics (0.505) is higher than the overall white-Hispanic score. Poor white married-couple households with children show among the highest levels of segregation from Hispanics (0.567), and indeed of Hispanics of the same household composition and poverty status (0.730). Whites living in non-family households show quite moderate levels of segregation from Hispanics in general (0.411) and Hispanic non-family households (0.420). Here again poverty status makes a difference, with nonpoor white households being less segregated from Hispanics than poor white households.
The final set of columns with Asians as the reference group shows similar patterns to those which use Hispanics as the reference group. Asian-white segregation is significantly higher among married-couple households with children (0.480) than all households (0.438), though this difference is moderate. Differences by poverty status are more striking, where the D between poor white married-couple households with children and all Asian households (0.631) is higher than D among nonpoor white married-couple households with children (0.481). The segregation of poor white and Asian single-parent households in the 13 metropolitan areas where they are both present in sufficient numbers is very high (0.802).
We conducted further analyses examining the segregation of black households from Asian households (available upon request). Overall levels of dissimilarity between these groups are moderate to high (0.545) in the 15 metropolitan areas where blacks and all Asian subgroups are present in sufficient numbers. Patterns by household composition and poverty status are fairly similar to what we see among whites and blacks, in that Asian married-couple households with children (0.592), and especially poor ones (0.716), are highly segregated from blacks. Poor Asian and poor black households are extremely segregated from each other (0.827 and 0.823 among married and single-parent households, respectively).
Table 2 shows interaction scores for non-Hispanic white households of various types and by poverty status. It has the same structure as Table 1, but highlights a different dimension of segregation: exposure. As described earlier, interaction scores indicate the average proportion of reference group members in the neighborhood where the typical individual of the group of interest resides. Other factors being equal, large demographic groups will tend to have lower interaction scores, because of the smaller risk set of other groups in a metropolitan area.
Table 2.
Interaction Index for non-Hispanic White Households by Household Structure and Poverty Status, 2000
| Composition of the white household | Reference Group
|
|||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Households with a Black householder | Households with a Black householder of the same composition | Households with a Hispanic householder | Households with a Hispanic householder of the same composition | Households with an Asian householder | Households with an Asian householder of the same composition | |||||||
|
| ||||||||||||
| Number of metro areas | Pxy | Number of metro areas | Pxy | Number of metro areas | Pxy | Number of metro areas | Pxy | Number of metro areas | Pxy | Number of metro areas | Pxy | |
| All households | 121 | 0.068 | 37 | 0.105 | 121 | 0.057 | 46 | 0.108 | 106 | 0.025 | 13 | 0.072 |
| All married-couple households with children | 121 | 0.057 | 37 | 0.032 | 121 | 0.059 | 46 | 0.057 | 106 | 0.027 | 13 | 0.041 |
| Poor | 121 | 0.081 | 37 | 0.004 | 121 | 0.082 | 46 | 0.012 | 106 | 0.026 | 13 | 0.004 |
| Not Poor | 121 | 0.056 | 37 | 0.030 | 121 | 0.059 | 46 | 0.051 | 106 | 0.027 | 13 | 0.038 |
| All single-parent households with children | 121 | 0.079 | 37 | 0.045 | 121 | 0.076 | 46 | 0.033 | 106 | 0.027 | 13 | 0.007 |
| Poor | 121 | 0.099 | 37 | 0.025 | 121 | 0.092 | 46 | 0.017 | 106 | 0.026 | 13 | 0.002 |
| Not Poor | 121 | 0.073 | 37 | 0.027 | 121 | 0.073 | 46 | 0.020 | 106 | 0.028 | 13 | 0.006 |
| All non-family households | 121 | 0.081 | 37 | 0.041 | 121 | 0.063 | 46 | 0.028 | 106 | 0.027 | 13 | 0.023 |
Source: Census 2000 Summary File 3. Means are unweighted. Metropolitan areas with at least 1,000 members in the group of interest and reference group by table column are included in the calculations. The reference group for the family-level index calculations includes only family households (rather than all households).
According to results in Table 2, the typical white household lives in a neighborhood where 6.8 percent of the households have a black householder. This proportion is a little higher in the more restricted set of 37 metropolitan areas (10.5 percent) which tend to have larger black populations. In the first set of columns in table 2 we see that poor married-couple white households actually tend to live in neighborhoods with more blacks (0.081) than nonpoor married-couple households (0.056). This pattern by poverty status holds among white single-parent households as well. Thus, despite lower levels of dissimilarity (shown in table 1), nonpoor white households, in part because they are a demographically large group, live in neighborhoods with relatively few households that include blacks.
The middle columns in table 2 indicate that different types of non-Hispanic white households show somewhat similar levels and patterns of interaction with Hispanic households. Nonpoor married non-Hispanic white households, for example, have moderately lower levels of interaction (0.059) with all Hispanic households than poor non-Hispanic white households of the same composition (0.082), but again poor white households tend to have particularly low levels of interaction with poor Hispanic households of the same composition.
Levels of interaction with Asian households are on the whole considerably lower (0.025), though a little higher in the more restricted set of metropolitan areas (0.072) that tend to have a higher proportion of Asians than the broader set of metropolitan areas. When all Asians are the reference group, there is very little variation in interaction by household type and poverty status of the white householder. When comparing households of the same composition and poverty status, the highest levels of interaction occur among nonpoor married-couple households.
Table 3 shows the interaction index of Black households with non-Hispanic white and Asian households. Among other things, the table, when compared with results in table 2, illustrates the asymmetrical properties of the interaction index. We see the black interaction is considerably higher with non-Hispanic whites (0.358) than vice-versa (the 0.068 figure from table 2). That is, in large part due to the demographic composition of metropolitan areas, the typical African American lives in a neighborhood with a considerably higher proportion of whites than the typical white individual with African Americans. The table further shows that black married-couple households (0.387) have higher levels of interaction with white households than black single-parent households (0.288). When looking at interaction scores for households of the same composition and poverty status, nonpoor married-couple (0.135) and non-family households (0.161) have relatively high levels of interactions with whites. Levels of interaction with Asians are on the whole low, and the variation by household type and poverty status are rather small. Levels of interaction are again highest among nonpoor married-couple households.
Table 3.
Interaction Index for Black Households by Household Structure and Poverty Status, 2000
| Composition of the black household | Reference Group
|
|||||||
|---|---|---|---|---|---|---|---|---|
| Households with a non-Hispanic White householder | Households with a non-Hispanic White householder of the same composition | Households with an Asian householder | Households with an Asian householder of the same composition | |||||
|
| ||||||||
| Number of metro areas | Interaction | Number of metro areas | Interaction | Number of metro areas | Interaction | Number of metro areas | Interaction | |
|
|
||||||||
| All households | 39 | 0.358 | 37 | 0.358 | 36 | 0.023 | 8 | 0.049 |
| All married-couple households with children | 39 | 0.387 | 37 | 0.136 | 36 | 0.029 | 8 | 0.031 |
| Poor | 39 | 0.292 | 37 | 0.005 | 36 | 0.024 | 8 | 0.006 |
| Not Poor | 39 | 0.397 | 37 | 0.135 | 36 | 0.029 | 8 | 0.027 |
| All single-parent households with children | 39 | 0.288 | 37 | 0.039 | 36 | 0.024 | 8 | 0.008 |
| Poor | 39 | 0.245 | 37 | 0.010 | 36 | 0.021 | 8 | 0.003 |
| Not Poor | 39 | 0.315 | 37 | 0.032 | 36 | 0.026 | 8 | 0.005 |
| All non-family households | 39 | 0.358 | 37 | 0.161 | 36 | 0.025 | 8 | 0.017 |
Source: Census 2000 Summary File 3. Means are unweighted. Metropolitan areas with at least 1,000 members in the group of interest and reference group by table column are included in the calculations. The reference group for the family-level index calculations includes only family households (rather than all households).
Table 4 shows the interaction index for Asian households, where non-Hispanic whites and blacks are the reference groups. Here we see very high levels of interaction with non-Hispanic white households (0.554). Interaction is relatively high among nonpoor married-couple Asian households (0.525) and non-family households (0.570), and lower for poor married-couple (0.384) and poor single-parent (0.387) households. These general patterns hold when looking at households of the same composition and poverty status, though overall interaction levels are lower (in large part due to smaller demographic groups in the risk sets).
Table 4.
Interaction Index for Asian Households by Household Structure and Poverty Status, 2000
| Composition of the Asian household | Reference Group
|
|||||||
|---|---|---|---|---|---|---|---|---|
| Households with a non-Hispanic White householder | Households with a non-Hispanic White householder of the same composition | All households with a black householder | Households with a black householder of the same composition | |||||
|
| ||||||||
| Number of metro areas | Interaction | Number of metro areas | Interaction | Number of metro areas | Interaction | Number of metro areas | Interaction | |
|
|
||||||||
| All households | 15 | 0.554 | 13 | 0.584 | 15 | 0.080 | 8 | 0.102 |
| All married-couple households with children | 15 | 0.510 | 13 | 0.204 | 15 | 0.076 | 8 | 0.029 |
| Poor | 15 | 0.384 | 13 | 0.006 | 15 | 0.114 | 8 | 0.005 |
| Not Poor | 15 | 0.525 | 13 | 0.207 | 15 | 0.072 | 8 | 0.026 |
| All single-parent households with children | 15 | 0.445 | 13 | 0.058 | 15 | 0.101 | 8 | 0.059 |
| Poor | 15 | 0.387 | 13 | 0.013 | 15 | 0.121 | 8 | 0.032 |
| Not Poor | 15 | 0.467 | 13 | 0.048 | 15 | 0.094 | 8 | 0.035 |
| All non-family households | 15 | 0.570 | 13 | 0.306 | 15 | 0.082 | 8 | 0.053 |
Source: Census 2000 Summary File 3. Means are unweighted. Metropolitan areas with at least 1,000 members in the group of interest and reference group by table column are included in the calculations. The reference group for the family-level index calculations includes only family households (rather than all households).
Patterns in Asian-black interaction tend to be similar to those observed between whites and blacks. The overall interaction score between Asian and black households is low (0.080). As among whites, we also find nominally higher levels of Asian-black interaction among poor Asian households than non-poor ones, though these differences are not statistically significant, in part due to the relatively small number of metropolitan areas in the sample (15).
Finally, Table 5 shows the interaction between Hispanics and non-Hispanic whites. Overall levels are fairly high (0.494), suggesting that about half the residents are white in the neighborhood where the typical Hispanic lives. Interaction with whites is relatively high among nonpoor married-couple households and non-family households, and lower among poor households (both married-couple and single-parent). Comparing those of the same household composition/poverty status (the final column), interaction is relatively high among nonpoor married-couple Hispanic and white households, as well as non-family ones.
Table 5.
Interaction Index for Hispanic Households by Household Structure and Poverty Status, 2000
| Composition of the Hispanic household | Reference Group
|
|||
|---|---|---|---|---|
| Households with a non-Hispanic White householder | Households with a non-Hispanic White householder of the same composition | |||
|
| ||||
| Number of metro areas | Interaction | Number of metro areas | Interaction | |
| All households | 65 | 0.494 | 46 | 0.537 |
| All married-couple households with children | 65 | 0.444 | 46 | 0.169 |
| Poor | 65 | 0.356 | 46 | 0.007 |
| Not Poor | 65 | 0.461 | 46 | 0.170 |
| All single-parent households with children | 65 | 0.406 | 46 | 0.066 |
| Poor | 65 | 0.364 | 46 | 0.017 |
| Not Poor | 65 | 0.431 | 46 | 0.052 |
| All non-family households | 65 | 0.528 | 46 | 0.253 |
Source: Census 2000 Summary File 3. Means are unweighted. Metropolitan areas with at least 1,000 members in the group of interest and reference group by table column are included in the calculations. The reference group for the family-level index calculations includes only family households (rather than all households).
Conclusion
The goal of this analysis has been to examine the relationship between household composition and racial and ethnic residential segregation. Using tract-level summary files from the 2000 decennial census, we estimated levels of metropolitan segregation, as represented by the dissimilarity and interaction indexes, for different groups. Dissimilarity and interaction provide complementary pieces of information about the residential patterns of these groups because they tap into different dimensions of segregation: evenness and exposure. We find that when using the dissimilarity index, white households with children are indeed more segregated from black, Hispanic, and Asian households than white households as a whole. Poor white households tend to display the highest levels of dissimilarity, particularly with corresponding poor households of other racial and ethnic groups.
The interaction index tells a different, complementary, story. In large part reflecting the relative demographic dominance of nonpoor white married-couple households in many metropolitan areas, such households tended to have somewhat limited “interaction” with other groups, and blacks and Hispanics in particular. For example, the typical nonpoor white married-couple householder in 2000 lived in a neighborhood that was 6 percent black, 6 percent Hispanic, and another 3 percent Asian. The typical poor white married-couple household actually lived in slightly more diverse neighborhoods, where, on average, 8 percent of the households were black, 8 percent were Hispanic, and another 3 percent were Asian.
Also reflecting the demographic dominance of whites in U.S. metropolitan areas, we found that minority group members often lived in neighborhoods with a relatively high proportion of non-Hispanic white households. The typical black, Asian, and Hispanic householder lived in neighborhoods that were, on average, 36 percent, 55 percent, and 49 percent non-Hispanic white, respectively. Among all three minority group households with children, nonpoor married householders tended to have the highest levels of interaction with whites.
Our data are limited in that the descriptive information they provide cannot allow us to judge which among several competing explanations best accounts for the differences in residential segregation patterns between households with and without children. However, we can draw some tentative conclusions. First, differences in socioeconomic status do not explain differences in segregation patterns by household composition among whites. Both nonpoor and poor white family households were more segregated from blacks than non-family households.
Nevertheless, socioeconomic status still influences patterns of segregation. Perhaps contrary to common perception, poor white and Asian households with children are more segregated from blacks than similar nonpoor white and Asian households when using the dissimilarity index, though poor white households do live in neighborhoods with a slightly higher proportion of minorities. This suggests that nonpoor white households, because of their demographic dominance and higher incomes are, so to speak, less at “risk” of living in neighborhoods with many minority group members (poor African Americans and Hispanics in particular) than poor white households. It could be that there is greater “effort” among poor whites to avoid minority group members to achieve a desired neighborhood racial/ethnic mix.
Conversely, nonpoor black, Hispanic, and Asian households with children live in neighborhoods with a higher proportion of whites than poor black, Hispanic, and Asian family households. These minority families with children may seek the better neighborhood schools and amenities of neighborhoods in which more whites live. Nonpoor families are more likely to be able to afford to live in these neighborhoods, and thus use their resources to “buy into” the schools and services in neighborhoods with more whites. A third tentative conclusion, then, is that socioeconomic status matters somewhat differently for the choices of white families with children compared to Hispanic and black families with children.
The higher racial and ethnic segregation of white married-couple families with children from other groups than non-family households suggests that having children matters. However, analyses that looked at white married couple families without children, and others that compared households by age of the children (not shown), found that these factors played, at most, small roles in segregation patterns. This suggests that life course factors associated with the presence of young children do not account for why children influence patterns of residential racial and ethnic segregation. It could be that white families with and without children are relatively similar to each other (at least as compared with non-family households) in a number of ways, such as in their incomes and neighborhood preferences. Many of the married couples without children may also either anticipate having children shortly or may have had their children move out before the survey. With the data available, we cannot tell.
Finally, we note that our analyses do not allow us to definitively judge whether or not families respond to the services and amenities of neighborhoods, like their parks or schools, or to the race or ethnicity of their neighbors. White families may seek to avoid schools that they perceive to be of low quality that are located in integrated neighborhoods, as the “racial proxy” or “race-associated reasons” perspectives suggest. Or, it could be that white families want to prevent their children from interacting with children of other races and ethnicities, according to the “racial reasons” perspective. Minority families with children may seek neighborhoods with more whites to gain access to better services like schools, to expose their children to white social and peer networks, and/or to avoid the perceived social problems of high minority neighborhoods. Future analyses with other kinds of data should focus on the reasons why children influence households’ residential decisions.
Footnotes
We also examined whether the ages of children in the household were associated with residential patterns by comparing households that had children less than 5 years only vs. children 5 to 17 years old only. However, there were only minimal differences by age of children.
We chose not to have a constant set of metropolitan areas for the entire table (or all of the tables in the paper) because we are most interested in making comparisons in segregation scores within columns (i.e., the role of household structure and poverty by reference group). In addition, restricting the number of metropolitan areas to the smallest cell in an entire table ends up confining the analysis to a very small number of metropolitan areas that are hardly representative of the contexts in which all groups reside.
One set of findings not shown in the table indicates that the dissimilarity between white married-couples without children and blacks is only nominally (not statistically significantly) lower than the figure for white married couples with children. In a constant set of metropolitan areas where all of these groups are present in sufficient numbers, the corresponding D scores are 0.664 for married couples with children and 0.647 for couples without. Why are married couples without children more similar to married couples with children than to non-family households? With the data available it is difficult to say, though the literature reviewed in the background section suggests at least three explanations. Married couples without children may resemble married couples with children in: 1) either their anticipation of having children shortly or having had children in the past that are no long present; 2) incomes; 3) residential preferences for neighbors of different races and incomes. We do not include results for married-couple households without children in any of the tables because such families tend to fairly closely resemble married-couple households with children, and because their inclusion considerably reduces the sample size of metropolitan areas in the tables. We include some discussion on this general issue in the paper’s conclusion.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
John Iceland, Penn State University.
Kimberly A. Goyette, Temple University
Kyle Anne Nelson, University of Maryland.
Chaowen Chan, University of Maryland.
References
- Bobo Lawrence, Zubrinsky Camille. Attitudes on Residential Integration: Perceived Status Differences, Mere In-Group Preference, or Racial Prejudice? Social Forces. 1996;74(3):883–909. [Google Scholar]
- Charles Camille Zubrinsky. Can We Live Together? In: de Souza Briggs Xavier., editor. The Geography of Opportunity. Washington, D.C: Brookings Institute Press; 2005. [Google Scholar]
- Charles Camille Zubrinsky. Neighborhood racial-composition preferences: evidence from a multiethnic metropolis. Social Problems. 2000;47(3):370–407. [Google Scholar]
- Clark William AV. Race, Class, and Place. Urban Affairs Review. 2007;42(3):295–314. [Google Scholar]
- Ellen Ingrid Gould. How Integrated Did We Become During the 1990s? In: Goering John., editor. Fragile Rights within Cities: Government, Housing, and Fairness. Lanham, Maryland: Rowman and Littlefield; 2007. [Google Scholar]
- Fischer Claude S, Stockmayer Gretchen, Stiles Jon, Hout Michael. Geographic Levels and Social Dimensions of Metropolitan Segregation. Demography. 2004;41:37–60. doi: 10.1353/dem.2004.0002. [DOI] [PubMed] [Google Scholar]
- Harris David. Population Studies Center Research Report 97–386. University of Michigan; 1997. The Flight of Whites: A Multilevel Analysis of Why Whites Move. [Google Scholar]
- Harris David. ’Properties Values Drop When Blacks Move in Because…’ Racial and Socioeconomic Determinants of Neighborhood Desirability. American Sociological Review. 1999;64(3):461–479. [Google Scholar]
- Harris David. Why Are Whites and Blacks Averse to Black Neighbors? Social Science Research. 2001;30:100–116. [Google Scholar]
- Holme Jennifer J. Buying Homes, Buying Schools: School Choice and the Social Construction of School Quality. Harvard Educational Review. 2002;72(2):177–205. [Google Scholar]
- Iceland John. Where We Live Now: Immigration and Race in the United States. Berkeley, CA: University of California Press; 2009. [Google Scholar]
- Iceland John, Sharpe Cicely, Steinmetz Erika. Class Differences in African American Residential Patterns in U.S. Metropolitan Areas: 1990–2000. Social Science Research. 2005;34(1):252–266. [Google Scholar]
- Iceland John, Wilkes Rima. Does Socioeconomic Status Matter? Race, Class, and Residential Segregation. Social Problems. 2006;52(2):248–73. [Google Scholar]
- Iceland John, Weinberg Daniel H, Steinmetz Erika. US Census Bureau, Census Special Report, CENSR-3. Washington, DC: U.S. Government Printing Office; 2002. Racial and Ethnic Residential Segregation in the United States: 1980–2000. [Google Scholar]
- Iceland John, Nelson Kyle Anne. Hispanic Segregation in Metropolitan America: Exploring the Multiple Forms of Spatial Assimilation. American Sociological Review. 2008;73(5):741–765. [Google Scholar]
- Iceland John, Scopilliti Melissa. Immigrant Residential Segregation in U.S. Metropolitan Areas, 1990–2000. Demography. 2008;45(1):79–94. doi: 10.1353/dem.2008.0009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Johnson Heather Beth. The American Dream and the Power of Wealth: Choosing Schools and Inheriting Inequality in the Land of Opportunity. New York: Routledge; 2006. [Google Scholar]
- Krysan Maria. Whites Who Say They’d Flee: Who are They and Why Would They Leave? Demography. 2002;39(4):675–696. doi: 10.1353/dem.2002.0037. [DOI] [PubMed] [Google Scholar]
- Landale Nancy S, Guest Avery. Constraints, Satisfaction, and Residential Mobility: Speare’s Model Reconsidered. Demography. 1985;22:199–222. [PubMed] [Google Scholar]
- Logan John R. Report by the Lewis Mumford Center, State University of New York at Albany. 2004. Apr 26, Resegregation in American Public Schools? Not in the 1990s. [Google Scholar]
- Logan John R, Oakley Deirdre, Smith Polly, Stowell Jacob, Stults Brian. Report by the Lewis Mumford Center, State University of New York at Albany, May 4, 2001. 2001. Separating the Children. Updated December 28, 2001. [Google Scholar]
- Rosenbaum Emily, Friedman Samantha R. Differences in the Locational Attainment of Immigrant and Native-Born Households with Children in New York City. Demography. 2001;38(3):337–348. doi: 10.1353/dem.2001.0030. [DOI] [PubMed] [Google Scholar]
- Ross Stephen L, Turner Margery Austin. Housing Discrimination in Metropolitan America: Explaining Changes between 1989 and 2000. Social Problems. 2005;52(2):152–80. [Google Scholar]
- Rossi Peter H. Why Families Move: A Study in the Social Psychology of Urban Residential Mobility. New York: The Free Press; 1955. [Google Scholar]
- Speare Alden, Jr, Goldstein Sidney, Frey William H. Residential Mobility, Migration, and Metropolitan Change. Cambridge, MA: Ballinger Publishing Company; 1975. [Google Scholar]
- Squires Gregory D, Kurbin Charis E. Privileged Places: Race, Residence, and the Structure of Opportunity. Boulder, CO: Lynne Rienner Publishers, Inc; 2006. [Google Scholar]