Neighborhood types in Chicago, 2010

The maps below present a classification of Chicago’s residential census tracts based on multivariate analysis. This approach (sometimes called social area analysis or factorial ecology in geography and sociology) is often used to classify small areas in cities. The ten neighborhood types identified on the maps were derived through a two-step process. First, the TRYSYS program was used to factor 34 important tract-level census variables by the Tryon “key-cluster analysis” method. The data come either from the 2010 census or from the 2008-2012 American Community Survey. Three oblique dimensions were identified. Then each tract was scored on the three dimensions (using a simple sum of standardized scores), and tracts were cluster-analyzed using TRYSYS‘s iterative partitioning method. Robert B. Dean did the statistical analysis.

These maps are comparable to those generated for 1990 and 2000 data when I was working at the University of Chicago Library. Essentially the same variables were used, and nearly the same geographic area was covered. The three dimensions (or clusters) are very similar to the first three dimensions found in the 1990 and 2000 data (although the order of the first two is different). In both 1990 and 2000, a fourth dimension was identified, associated with family type and age. A similar dimension in the 2010 data is not at all significant, and, in conjunction with this, the geographical pattern of neighborhood types has changed in some small ways. For example, the well-off/young-adult/non-family area roughly along North Halsted Street that was identified in the 2000 census analysis has vanished in 2010, mostly subsumed into areas classed as very well-off and very urban (type 1). But, in general, the broad pattern of Chicago’s social geography appears to have changed only in subtle ways in the first decade of the 21st century. The area of gentrification on the North Side has clearly moved west and north from Wicker Park and Bucktown, into Ukrainian Village, Logan Square, and even Humboldt Park. Gentrification has also moved west from downtown. Click here for some maps that demonstrate that there was a considerable amount of ethnic-specific internal migration in the area. Note that, because the classifications have changed, the color schemes of the 1990, 2000, and 2010 neighborhood-type maps, while similar, are not completely comparable.

No claim can be made that this is a definitive analysis of neighborhood types in Chicago. It is in the nature of this kind of analysis that a change in the variables selected or in the parameters set by the analyst can change the results significantly. There is also the issue that the 2008-2012 ACS data are not likely to be as reliable as earlier long-form data. The best that can be said is that the maps may provide one useful way of analyzing the differences in Chicago’s residential areas.

Here is a map showing neighborhood types in Chicago and its inner suburbs. Nominal scale is 1:250,000. (Click here for key.)

Map showing result of cluster analysis of circa 2010 data for Chicago's census tracts that resulted in classification into ten neighborhood types, Chicago and vicinity, Illinois

 

Here is a map showing neighborhood types in the Chicago region. Nominal scale is 1:700,000. (Click here for key.)

Map showing result of cluster analysis of circa 2010 data for Chicago's census tracts that resulted in classification into ten neighborhood types, Chicago region, Illinois and surrounding states

 

 

Posted in Urban | Tagged | Leave a comment

The extreme high density of the central portions of the New York area

Sprawl supporters like Wendell Cox and Robert Bruegmann1 (and many other writers) have gleefully pointed out the Census Bureau’s somewhat counterintuitive claim that the Los Angeles urban area is denser than New York’s. Anyone familiar with Los Angeles and New York would understand that the explanation for this must lie mostly in the different characters of the outlying portions of the two urban areas. New York (like other eastern cities) includes a large number of suburban areas with two-or-more-acre zoning. Such low-density suburbs are almost impossible in western cities, where rough terrain and government ownership of much of the mountainous land have discouraged their formation.

One way of demonstrating the extraordinary density of New York’s central portions that I haven’t seen pointed out requires looking at 2010 population density by census tract for all 73,000-odd U.S. census tracts (data from NHGIS). The densest tract, with 196,409 people per square kilometer (508,697 people per square mile), turns out to be in Chicago: a tract created especially for the 2010 census that contains a single apartment building, at 5415 North Sheridan Road. 163 out of the next densest 166 tracts, however, are in the New York area (all but one in New York, N.Y.; the three non-New-York tracts are in San Francisco). New York’s overwhelming dominance extends down the density hierarchy. Out of the 436 census tracts in the United States with population densities of more than 30,000 people per square kilometer (77,700 per square mile), 409 are in the New York area (401 in New York City). Of the remaining 27, 12 are in the San Francisco Bay Area (11 in San Francisco), 6 in Los Angeles County, 5 in Chicago, 3 in Honolulu, and 1 in Baltimore. Of the 967 tracts with more than 20,000 persons per square kilometer (51,800 people per square mile), 867 are in the New York area (842 in New York City). (The runners-up are: Los Angeles County 29, Cook County (almost all in Chicago) 24, San Francisco Bay Area 22, Honolulu 7, Philadelphia 6, Washington area 5.)

Here are choropleth maps of the New York, Chicago, Los Angeles, San Francisco, and (just for fun) Toronto areas, all at the same scale, showing persons per square kilometer in 2010 (2011 in Toronto) with the same class intervals and colors. Note how different New York looks. There are, for example, hardly any signs of “population densities as high as [in] Manhattan” that some scholars2 have claimed for Los Angeles. All of these metropolitan areas are radically underbounded, but outlying areas (even in Los Angeles) never get beyond moderate densities.

new_york_denschicago_densla_denssf_dens_1toronto-dens-clegend2

This observation jibes completely with the Census Bureau’s recent creation of a “population-weighted density” statistic for urban areas, in which the New York area ends up being considerably denser than any other American urban region. New York has 31,251 “population-weighted” persons per square mile, San Francisco 12,145, Los Angeles 12,113, and Chicago 8,613and these figures include the suburbs.

Posted in Urban | Tagged , , , , | Leave a comment

Expedition to Kenosha

Kenosha--Streetcar

I had lived in Chicago for thirty years, and I like trains, but I’d never gone and taken a look at the Chicago area’s only streetcar line, in Kenosha, Wisconsin. So I set out one Sunday afternoon on Chicago’s suburban railroad system, Metra, to visit Kenosha. It was a pretty easy hour-and-forty-minute trip, past familiar cityscape for a while, then along a bicycle trail through well-off suburbs, and finally (and much more speedily) through increasingly open country. Kenosha may in some senses be a suburb of Chicago, but there is plenty of unbuilt-on land close to the state border.

The Kenosha streetcar runs for a mile or so between the Metra station and the Lakefront. Its complete route is a loop, said to be two miles long in all.

Kenosha streetcar, Metra, pedestrian facilities, Kenosha, Wisconsin

Map of central Kenosha showing the routes of the streetcar line and of Metra trains to Chicago. GIS data from Geofabrik version of OpenStreetMap.

Its development was justified on the grounds that the line would serve the inhabitants of Harbor Park, a relatively new townhouse condo development on the Lakefront. The Kenosha streetcar seems to have been an abysmal failure at attracting Harbor Park customers though. The current schedule—7.5 hours a day starting at 10 or 11 for most of the year but on weekends only in winter—would not allow commuters to use it. To get to the Loop by 9 you’d have to take a train at 7:13 or earlier, and the first train from the Loop after 5 wouldn’t get you home until 6:38 or later. Clearly the streetcar line has become essentially a tourist attraction and a modest one at that. The latest figures (2014) from the American Public Transit Association (APTA) suggest that there are 45,054 passengers/year, or maybe 140/day (and you wonder how Kenosha Transit counts people who buy a day pass; if you make several round trips on a day pass, does each one-way ride count as an unlinked trip?). When I was there only one PCC car was running. It was taking about ten minutes to complete a loop, that is, making maybe 45 round trips, or ninety one-way trips/day. That means that there were something like 1.5 passengers per trip. Actually, I was struck by the complete absence of passengers on the Sunday afternoon I visited. There were, however, people out photographing the nicely repainted PCC cars, of which Kenosha Transit has acquired six. To someone who likes trains, it’s quite appealing to see these PCC cars running in a small city. It’s not clear that the cars have much to do with urban transit, however.

The Metra trains, interestingly, were just jammed, especially the late afternoon train back from Kenosha to Chicago. There were standees between Ravinia Park and Ravenswood. But the train wasn’t just serving people attending a Ravinia event. People got on and off at every one of the 27 stops. Perhaps there should be more than nine trains a day on Sunday! It was certainly interesting to contrast a transit service that is filling a real need with one that seems to be serving mostly an aesthetic one. The Kenosha streetcar is, admittedly, probably bringing in some tourist dollars, perhaps enough to justify the $6.37 subsidy per passenger trip (but note that, according to APTA, Metra’s tickets—almost always for a much longer journey—are subsidized at only $4.93/ride).

Posted in Urban, Transportation, Rail infrastructure | Tagged | Leave a comment

Delhi Metro

Delhi has added more rail transit in the 21st century than any other city outside of East Asia. Although the first line did not open until 2002, there are now 190 route kilometers and 2.4 million riders a day. Among North American rail transit lines, only the New York subway is longer and carries more passengers. And the Delhi subway is still growing. Phase III, now being constructed, will add 160 route kilometers. In a few years Delhi will have one of the world’s largest rail transit systems. Delhi Metro is generally considered an enormous success, not least because the lines were apparently built efficiently, with a minimum of corruption, in a country where this is not the norm.

I’ve been on the Delhi Metro twice, on an earlier trip in 2007 and just recently. There are all sorts of tremendously impressive things about the system. The underground stations and the trains are air-conditioned—a fairly important accomplishment given that temperatures in Delhi were well over 100 degrees every day I was there. The stations are quite spacious and subtly lit (they remind me of stations on the Washington Metro). The trains are very much of the modern type. You can walk from car to car. It is true that there are plenty of reminders that you’re in India. The trains can get awfully crowded (except for the women’s car). And passengers getting on seem incapable of waiting for passengers to get off, and there is thus lots of pushing and shoving. Also, it’s a bit unnerving that you have to get your luggage x-rayed and undergo a body search to enter a station (on one day I had to take out every object from a crowded backpack for a zealous guard). Furthermore, there are frequent announcements warning passengers against picking up unattended objects since they could be bombs. Still, who could argue with the proposition that Delhi Metro is a major achievement?

I have one pretty major caveat, however. It’s that it’s practically impossible to walk to the Metro stations, at least in central Delhi.

When I first rode the Delhi Metro in 2007, I decided to follow the elevated line that emerges a few blocks west of Connaught Circle, to take pictures and to get a feel for how the line fit into the neighborhoods it passed through. I gave up after a couple of stations. One problem is the near absence of sidewalks along the route (see below). In most cases a sidewalk had once been built, but it’s been encroached on in places and has not been maintained in others, so that there is more trash-filled pothole than sidewalk. Most Indians just walk in the traffic lane. Even more serious, you take your life in your hands every time you want to cross a street. There are few traffic lights, and little attention is paid to those that exist. Some Indian intellectuals say they hate the anthropologist Louis Dumont’s phrase “homo hierarchicus,” an attempt to characterize the Indian people in light of the caste system. But there is no doubt that there is a hierarchy on Indian roads. Third World traffic rules are in force everywhere. The larger vehicle has the right of way, and pedestrians have no rights at all. Even though automobile ownership is not that high, Delhi is one of the most pedestrian-unfriendly cities in the world—especially British-planned New Delhi with its innumerable roundabouts. You end up wondering just how people get to the Metro stations (or anywhere else). A surprising number take auto or bicycle rickshaws short distances just to avoid having to walk. There may be fewer pedestrians in central Delhi than in any other large city anywhere. There are many fewer pedestrians in the blocks around around Connaught Circle than in, say, downtown Los Angeles, which is famous in North America for being a car-oriented place. In downtown Los Angeles paved sidewalks are universal, and you can cross streets safely.

Delhi--MetroOverStreet2

I visited Gurgaon on my last trip to Delhi. Gurgaon is Delhi’s Rosslyn or Schaumburg or La Défense. Many successful firms have built high-rise office buildings in Gurgaon. There’s also a huge amount of middle- and upper-class housing. A new elevated Metro line to one of Gurgaon’s major office complexes just opened a couple of months ago. It connects to a branch of the Metro that opened approximately a year ago. Both lines were jammed despite the fact that service is frequent, and I had to stand all the way back to central Delhi. But even in shiny-new Gurgaon it’s painful to get to the stations. There are numerous pedestrians in Gurgaon’s office district—and almost no provision made for them. You often have to walk in moving traffic or across muddy fields to get where you’re going (see below).

Delhi--Gurgaon pedestrians

Some Indians attribute the pedestrian-unfriendliness of Indian cities to cultural factors other than the caste system. It is certainly true that Indians on the whole are probably less interested in fitness than just about any of the world’s peoples. You almost never see runners or recreational bicyclists in Indian residential districts. Still, there are plenty of middle-class Indians who enjoy walking. Small parks all over urban India have de facto tracks where people walk in circles. I spent a couple of hours on my last trip to Delhi in Lodi Garden, a wonderful park in New Delhi that contains several enormous centuries-old tombs; lots of large trees (and therefore shade); and thousands of noisy birds. It may be the most pleasant place in Delhi. And it’s just big enough to contain an approximately one-mile walking loop. Amazingly, there were a very large number of people there making the circuit, including some women (see below). It’s true that the park also has a substantial population of unowned dogs, but they do seem to let people pretty much alone (although they can bark furiously at owned dogs being walked on leashes). Clearly, some Indians would walk to Metro stations if it were easier to do so.

Delhi--Lodi Garden3

Delhi of course is hardly the only Third World city where pedestrians are treated with contempt. But, unlike its close competitors in this area (Jakarta and Hanoi, for example), it has made a substantial investment in a form of transport that appears to function fully only when pedestrian access is possible. It’s true that, even as it is, Delhi’s subway is working for millions of people. But a sympathetic foreigner can’t help but wonder whether it wouldn’t be even more of a success if something were done about making it less of an ordeal to walk to the stations. I acknowledge that it would not be a trivial matter to change India’s driving culture—or its long-standing indifference to keeping sidewalks clear.

–May 2014

Posted in Pedestrian infrastructure, Rail infrastructure, Transportation, Urban | Tagged | Leave a comment

Trysys

TRYSYS is a descendent of the BCTRY program, which was developed by Robert C. Tryon at Berkeley during the 1960s. See Cluster Analysis by Robert C. Tryon and Daniel E. Bailey (New York : McGraw-Hill, 1970). There are many other techniques available for performing dimension reduction, most of which lean heavily on orthogonal factor analysis. Tryon key-cluster factoring has the advantage of not forcing dimensions to be orthogonal, that is, it allows a significant degree of correlation among dimensions (for example, in the case of the present analysis, between the “suburban” dimension and the “wealth” dimension). It is arguable that this capability allows its dimensions to approximate those of everyday experience. For additional information on TRYSYS, click here.

Posted in Urban | Comments Off on Trysys

Social area analysis

Social area analysis essentially involves a statistical procedure to identify from a sometimes quite large database of socio-economic data the most salient underlying variables. Very often most of the variance in a group of dozens of variables can be accounted for by three or four dimensions.

It is probably fair to say that the heyday of this approach occurred in the 1970s. It has cropped up frequently in scholarly literature in the years since, but there seems to have been only a modest amount of new theoretical work in this area. Social area analysis and factorial ecology have also been quite important in marketing research.

See, for example, Carl-Gunnar Janson, “Factorial social ecology: an attempt at summary and evaluation,” Annual Review of Sociology, 1980, vol. 6, pp. 433-456 (JSTOR link).

Posted in Urban | Comments Off on Social area analysis

34 variables

Here is a list of variables used in the analysis of neighborhood types in Chicago, 2010.

Variable name Description
DENS_PSQKM population per square kilometer
PERSPHOUS persons per household
PCTAFAM percent non-Hispanic African-American/black
PCTASPAC percent non-Hispanic Asian/Pacific
PCTHISP percent Hispanic/Latino
PCT_LT18 percent less than 18
PCT18_29 percent 18-29
PCT30_49 percent 30-49
PCT50_64 percent 50-64
PCT_65UP percent 65+
Z0PLUSHOUS percent occupied housing units in buildings with 10 or more housing units
SINGFAMH percent occupied housing units that are single-family houses
FEMHEADHH percent households female-headed
MARCOUPHH percent households including married couple
PUBTRANS percent workers 16+ using public transit to travel to work
COLLDEGR percent population with college degree
UNEMPL percent workers 16+ unemployed
OWNOCC percent households owner-occupied
NOCAR percent households with no vehicle
POLISH percent Polish ancestry
ITALIAN percent Italian ancestry
RENT median rent of rented housing units
HOUSVAL median value of owner-occupied housing units
HHINC median household income
PCAPINC per capita income
YEARBILT mean year housing built
PCTNONFM percent households “non-family”
LINGISOL percent “linguistically isolated”
FORBORN percent foreign born
PROFMANAG percent workers 16+ professional/managerial
PUBASST percent population on public assistance
RUSSIAN percent Russian ancestry
WHITE percent non-Hispanic white
SAMESEXP percent households with same-sex partner
Posted in Urban | Comments Off on 34 variables

T-scores

T-scores are standardized scores on each dimension for each type. A score of 50 represents the mean. A difference of 10 from the mean indicates a difference of one standard deviation. Thus, a score of 60 is one standard deviation above the mean, while a score of 30 is two standard deviations below the mean.

Posted in Urban | Leave a comment

Neighborhood types, 2010

In the statistical study of neighborhoods in Chicago, 2010, cluster analysis of the three dimensions underlying the 34 variables yielded ten neighborhood types. For each neighborhood type, the following list includes:

[a] the neighborhood-type number;
[b] the T-scores on the three dimensions; and
[c] a short name derived mechanically from the scores on the three dimensions. A simple descriptive phrase (like “Welloff” or “NonEng”) indicates a standard deviation from the mean of .5 to 1 ; the same phrase preceded by V (e.g., “VWelloff” for very well-off) indicates a standard deviation from the mean of between 1 and 1.5; repeated “V”‘s are used for additional half-standard deviations from the mean.

The ten neighborhood types are:

1. (64.68 67.87 45.98). VWelloffVVUrban. Well-off, very urban neighborhoods. Prosperous inner-city areas with “professional” populations. The North Side Lakefront, with outliers in Hyde Park, Evanston, Oak Park, and a very few suburban apartment districts. Relatively few non-English speakers.

2. (52.24 67.90 52.26). VVUrban. Somewhat less well-off very urban areas, mostly found on the edges of the more thoroughly gentrified tracts of type 1. Generally more non-English speakers than in type 1.

3. (66.01 37.83 44.54). VVWelloffVSuburbanNative. Very well-off suburban neighborhoods with few non-English speakers. Old-line wealthy suburbs, largely in northern Cook County and in parts of DuPage County.

4.
(62.30 46.44 45.88). VWelloff. Prosperous neighborhoods, mostly in suburban areas, with few non-English speakers. Mostly on the edges of tracts of type 3.

5. (54.46 40.28 44.32). SuburbanNative. Suburban neighborhoods with few non-English speakers. Very common. Forms a rough ring around the more distinctive inner suburbs.

6. (50.20 46.60 45.88). Neutral. None of the dimension scores is even half a standard deviation from the mean, but these areas tend to be suburban with few non-English speakers. Generally inner-suburban areas on all sides of the city.

7. (33.38 63.08 41.63). VVPoorVUrbanNative. The core, generally impoverished African-American neighborhoods of the South and West Sides and Gary.

8. (37.91 51.75 44.11). VPoorNative. Generally African-American neighborhoods in the outer parts of Chicago as well as in Gary, Maywood, and a few older central places like Joliet and Kankakee.

9. (41.03 54.44 76.75). PoorVVVVNonEng. The core Hispanic areas: Little Village and parts of Humboldt Park. More than two standard deviations above the mean on dimension 3.

10.
(46.21 51.01 64.10). NonEng. Mostly urban or inner-suburban areas that have many (but not an overwhelming number of) non-English speakers. Includes many tracts on the Northwest Side of Chicago where there are substantial numbers of relatively recent immigrants from Asia, Eastern Europe, and Latin America, as well as similar areas in the inner northwestern suburbs.

Posted in Urban | Comments Off on Neighborhood types, 2010

Three dimensions, 2010

The analysis of neighborhood types in Chicago, 2010, suggested that three oblique key-cluster dimensions underlaid the 34 variables used in the computations. The dimensions are listed below with their correlations to the most salient dimension definers (note that the names of the dimensions are a little arbitrary; some of these dimensions are quite complex in character):

  • Wealthy/High Professional Status vs. Impoverished/Low Professional Status [Welloff vs. Poor]. The eleven definers are:

[+.876] Mean per capita income

[+.872] Median household income

[+.863] Percent with college degree

[-.846] Percent of households with female head of household

[+.807] Percent with managerial/professional occupation

[+.797] Percent non-Hispanic white

[+.745] Median value of owner-occupied housing units

[-.716] Percent unemployed

[-.678] Percent African-American

[-.600] Percent on public assistance

[+.545] Median monthly rent of rented housing units

  • Urban vs. Suburban. The ten definers are:

[-.894] Percent in owner-occupied dwelling

[-.888] Percent in married-couple household

[-.867] Percent in single-family housing unit

[+.850] Percent of households with no car

[+.788] Percent of households that are “non-family”

[+.761] Percent using public transportation to get to work

[+.621] Percent of population 19-29

[+.616] Percent of housing units in buildings with ten or more housing units

[+.539] Population density

[-.474] Year housing unit built

  • Linguistically Isolated/Hispanic vs. English-Speaking/ Native-Born [NonEng vs. Native]. The three definers are:

[+.936] Percent linguistically isolated

[+.869] Percent Hispanic

[+.861] Percent foreign-born

 

The three dimensions are intercorrelated as follows.

1 2 3
1 1 -0.31 -0.18
2 -0.31 1 0.14
3 -0.18 0.14 1

 

Posted in Urban | Comments Off on Three dimensions, 2010