John Logan
Department of Sociology
State University of New York

POSITION STATEMENT

Research interests

My principal interest in urban development processes is their implications for stratification: for racial and class segregation, for unequal access to resources (employment, schools, health care), and for disparities in exposure to crime and public health problems. I investigate these partly by research on the political economy of land development � what creates and sustains inequalities between places? � and partly by research on locational processes � what people get concentrated in the worst places, and why?

Originally my attention focused mainly on black-white differences. I now give increasing attention to issues of immigration, particularly the phenomena of assimilation and persistent ethnic boundaries. My work has taken on a strong historical dimension, looking at New York and Chicago as far back as 1880. And I have also added some comparative perspective, primarily through research on contemporary Chinese cities.

Data and techniques

I rely almost entirely on census data, supplemented by diverse sources on employment, crime, health, and public services at the local level. I estimate two kinds of multivariate models. The first type is traditional ecological analyses of the relationships among various characteristics of places, sometimes longitudinally. The second is what I call locational attainment models: models that link characteristics of individuals (such as their income, family composition, or race), as predictors, to characteristics of places where they live, as outcomes. For references to this work, see:

http://www.albany.edu/sociology/jlogan/jlogan.html

I use GIS packages (MapInfo and ArcView) to provide a visual representation of spatial patterns. I also use these as working tools:

1. To deal with data using different spatial units, such as crime data for precincts and population data for census tracts. Such boundary problems are well known.

2. To identify neighborhoods, based on geographic contiguity of areas with similar aggregate characteristics, such as a set of adjacent census tracts that have high proportions of Chinese residents.
3. To create geographic variables, such as when we add transit lines to a city map and ask how close a given city block is to a subway station.

There are many difficulties in these applications. Many source maps are in libraries with limited reproduction facilities, and we don�t have equipment or skills to make high resolution photographs ourselves. The methods for creating GIS maps, even when paper maps are in hand, are cumbersome: scanning and digitizing turn out not to be as straightforward as expected. Boundary discrepancies almost always require forcing incomplete matches. Our most skilled GIS consultants (in geography) are unfamiliar with social science applications, so we usually start by doing things the wrong way. And of course there are great limitations in availability of spatial data.

Best practices

I believe one of the biggest gaps is in basic cartography � how to design maps that will communicate to the viewer. This is a problem for social scientists in general, that we are not good at devising tables or charts or figures for presentation. Maps are more difficult than these other media. When we map social characteristics, we often leave out other features (parks, roads, institutions) that would help make sense of the spatial pattern.

One publication in which we used simple maps to identify Russian, Italian, and Irish neighborhoods is: Richard D. Alba, John R. Logan, and Kyle Crowder. 1997. "White Neighborhoods and Assimilation: The Greater New York Region, 1980-1990" Social Forces 75: 883-909.

An innovative project by Phil Ethington (History, USC) is described on the following web page: http://www.usc.edu/isd/locations/cst/IDA. Ethington is building a digital archive for the LA metropolitan region that includes aerial photographs, current and historical census data, building records, and other sources. These are overlaid on a map grid. The archive is not yet functional, however.