This is a very comprehensive descriptive piece on the trend of residential differentiation in the U.S., in the years 1940 to 80. Chapter 4 looks at the issue of residential segregation - using diversity of indices from race and ethnicity to income and educational attainment, this chapter reviews just how segregated American neighborhoods are. Chapter 5 looks at the pattern of spatial organization - that is, merely looking at the level of segregation does not tell us what sort of people are living in which part of the metropolitan area. This chapter attempts to describe this issue of what type of people tend to live in which area of the SMSA, and tries to assay the predictive power of three theories aiming to account for the pattern of differential spatial organization in a city - the concentric zone theory, sector theory, and the multiple nuclei theory. Chapter 6 investigates into the issue of neighborhood change over time. Has the pattern of spatial organization changed significantly in the years of 1940 to 80? This chapter is divided into three parts: The first part looks at the change in segregation level, the second part looks at changes in spatial organization of the metropolis as described in chapter 5, and the third part attempts to trace the changing pattern of housing and population organization in an individual neighborhood, whereas the first two parts only looked at changes in the aggregate level, treating a whole SMSA as an unit of analysis.
 

Chapter 4

This is an almost wholly descriptive chapter on the level of segregation in the contemporary American neighborhood. Two indices of segregation level are used to capture this level of segregation. One is the commonly used index of dissimilarity D, and the other is the entropy index H, which is less well known but has the advantage of being able to treat variables with more than two categories at the same time, whereas the D can be calculated only for the segregation level of two dichotomous groups. The level of segregation can be described on the basis of variety of different variables, and the author accordingly includes a very comprehensive list of variables in his description in this study. These include race and ethnicity, which has generally been the variable of greatest concern to those studying residential segregation, physical characteristics of the neighborhood as types of building, housing and ownership, home value rent, socioeconomic characteristics as education, income and occupation, and the demographic characteristics as age, household size and marital status.

The conclusion for this chapter is so well organized that it probably suffices for the purpose of the prelim to simply read this part for this chapter. Here is the excerpt of important points from that conclusion.

First, selected variables used to describe the level of segregation can be placed in the "ladder of segregation", placed in a vertical order in terms of the different level of segregation. Using this scheme, then, the first salient point is the continuing significance of race as the most important characteristic on which segregation is based, as the level of racial segregation is higher than that based on any other variables. The black population is the most segregated of any demographic characteristic. No other ethnic groups are segregated to the same extent; the level of segregation among different white ancestry groups is appreciably lower. Also, the foreign born have surprisingly low level of segregation, in fact much lower than that of blacks - indicating that spatial assimilation theory, that posits all different newly arriving ethnic groups "assimilate" to the main stream pattern at the same rate as time passes, may not be tenable.

On the other hand, segregation level on the basis of all life cycle characteristics (age, sex, marital status, and household size) is very low. Neighborhoods on the whole appear to be quite diverse in terms of these characteristics.

The segregation level on the basis of status system - here, income, education and occupation - is quite modest, albeit higher than that of the life cycle variables. On the other hand, if rent and home value statistic is used as a proxy to represent status, then the level of segregation becomes appreciably higher - apparently people are more segregated on the basis of the type of housing they live in, which probably is more stable than income and occupation, than the more direct measurement of status and class. Further, it is also important to note that on the basis of poverty level, the level of segregation is much higher than that of the ordinary status system. Race definitely has some effects on this trend, and White notes that racial segregation is enough to raise overall poverty segregation and household type segregation by several points.
 

Chapter 5

The last chapter looked at the level of segregation, but that alone is not enough to tell what type of people live in which parts of the metropolis. In theory people could change their residential locations without altering the segregation level. This chapter addresses the issue of the spatial organization of the metropolis, that is, is there systematic pattern as far as the residential differentiation of the metropolis is concerned? Many theorists have answered yes to this question. The author first notes that there are basically three theoretical models that tries to provide for the systematic account of the residential structure, as well as the temporal change in that structure, of the metropolis. The first is the concentric zone model, originally formulated by Ernest Burgess. It posits that the distance from the CBD (central business district) is the key factor in residential differentiation, so that the neighborhoods of different characteristics tend to be arranged in the form of a series of concentric zones or rings. Since those higher in socioeconomic status are looking for newer and more spacious housing, the tendency for the growth pattern of a metropolis is outward, and the general pattern is that the successive zones increase in socioeconomic status. The second is the sector model, usually attributed to Homer Hoyt. This model sees the direction, not the distance, as the key factor in determining the spatial organization of a metropolis. It posits that while the growth is still outward, areas in the same direction tend to maintain the original socioeconomic characteristics of more inner areas. In short, the general picture is that sectors alternate in major socioeconomic status groups. The third model provides for even more complex and unsystematic picture. This is the multiple nuclei model, arguing that metropolitan development cannot be explained solely or fundamentally in terms of orientation to the CBD. Rather, it posits the existence of multiple "nuclei", each often being the alternative center of economic activity. Having reviewed these three theories, then, it is the author's task to not just review the trends of spatial organization of the American metropolis but do so in light of these three theories, assaying whether these theories do indeed have explanatory power or not.

The empirical data analysis for the spatial organization analysis utilizes several different methodological technique. One is the traditional form of mapping a metropolis with the census tract line, and different shadings given to different tracts indicate the differing proportion of people with certain characteristics according to the chosen variable. Another technique is the mapping of "metropolitan contours", utilizing the continuous lines somewhat similar to those used in the map showing levels of altitudes. These lines are supposed to connect areas with similar characteristics according to the chosen variable. It has the added merit of being able to represent fairly complex pattern of distribution and quantify the ability of location to explain statistically the variation in characteristics of neighborhoods. The third method is the statistical analysis to test whether a particular pattern is significant or not in the formal sense. That is, is the pattern of distribution of people according to certain variables deviate from the simple "linear" model based on the distance from the CBD in a statistically significant way? If so, then we need a more complex model, which would be the "contour" model just mentioned, but the question then is to what extent does this contour model have the added predictive power over the linear model? These are the questions asked in this statistical analysis part.

On the whole, for the results again there are way too much of rich details in the main body to summarize effectively. The conclusion for this chapter is also well written, fortunately. So...

In general, White concludes that it is still possible to say that the pattern of urban spatial organization is systematic. It is a highly complex picture, but on the other hand it is certainly not the world of totally patternless chaos. For many variables, the concentric zone model fits best, while for some others sectorial or the multiple nucleus model fares off better. But on the whole, the majority of variables seem to follow the zonal pattern, while few, if any, are strictly based on the sectorial model pattern.

Status distribution is multifaceted. In terms of income and housing expenditures, the pattern is principally zonal. Occupation and education are much less consistent, usually showing a variation of a multiple nuclei pattern. Life cycle differentiation is very weak (just as it was so for the simple segregation level), but to the extent it is present the pattern is zonal. In terms of race, perhaps not surprisingly, blacks are usually concentrated in the inner most region of the metropolis. On the other hand, sometimes the area of this concentration is not the central region of the metropolis, but either way the minority population, particularly that of blacks, is highly concentrated - rendering support to the multiple nucleus model for ethnicity. In fact, in the statistical analysis the "contour" model fared off very well in predicting the pattern of distribution based on race and ethnicity. Finally, in comparing the newer and older, larger cities, the tendency is that the former shows much less systematically organized pattern of spatial arrangement. In the new SMSA it is quite common that the spatial organization seems to follow no clearly discernible pattern, and consequently the statistical models fare off poorly. It remains to be seen whether this less systematic form of organization is the precursor of the future pattern, or those newer cities would eventually adopt the pattern of the older cities.
 

Chapter 6

Previous two chapters have looked at the pattern of residential differentiation at a fixed point of time, around 1980. This chapter looks at the temporal change of the residential differentiation, for the years 1940-80. As mentioned in the introduction, this task is divided into three parts. Again, the findings are extraordinarily rich in detail; the excerpts may be:

First, the good news is that segregation by race and ethnicity declined appreciably, and this is not just an illusory finding due to changes in the pattern of census tract or other mechanisms. On the other hand, race does remain to be the most segregated characteristic, and on the whole the ordering in the "ladder of segregation" has remained remarkably stable over the years. Among the other variables, occupations seem to have become less segregated and oriented to the CBD, perhaps more nucleated. On the other hand, segregation on the basis of poverty increased considerably during the 70s. From the analysis of neighborhood level information, it is noted that the vast majority of inner city tracts are experiencing decline in relative status and are depopulating. The periphery tracts of the metropolis are gaining population rapidly and gaining in relative status as well. The inner and outer most tracts show least gains in population black. The outer regions are becoming relatively younger and are oriented toward childrearing, while the middle two tracts are aging.