Teaching Research CV/Pubs Contact Links
 
 
 

Environmental Justice

 

Case Study: Air Toxic Releases in New Jersey

(from Mennis, J. and Jordan, L., 2005. The distribution of environmental equity: exploring spatial nonstationarity in multivariate models of air toxic releases. Annals of the Association of American Geographers, 95(2): 249-268)

 

Introduction

Geographic information systems (GIS) and multivariate regression are used to analyze socioeconomic inequity in the spatial distribution of New Jersey air toxic release facilities listed in the Environmental Protection Agency's (EPA) Toxic Release Inventory (TRI).  The map below shows New Jersey and the locations of TRI facilities with air releases.

 

 

 

Data and Methods

TRI facility data for 2000 were acquired from the EPA.  Socioeconomic data (population density, percent black, percent Hispanic, percent living below the poverty line, and percent employed in manufacturing) were acquired from the U.S. Bureau of the Census for the year 2000 at the tract level.  Land cover data were acquired from the U.S. Geological Survey's National Land Cover Data (NLCD) program.  Land cover data were used to calculate the percentage of each tract classified as industrial/commercial/transportation.  A TRI point density surface was calculated using a bandwidth of 5 km.  The mean point density was then summarized for each tract.  Maps of these variables are given below.

 

 

Kendall's tau-b correlation and multivariate regression were used to assess the relationship of the socioeconomic and land cover variables with TRI density.

 

 

Results

The table below reports the correlation of TRI density with the other variables.

 

 
 

Kendall Tau-b Correlation Coefficients of TRI Density

 

Variable

TRI Density

0.221**

0.354**

0.440**

0.255**

0.231**

0.265**

Percent Black

Percent Hispanic

Population Density

Percent Industrial

Percent Poverty

Percent Manufacturing

** p < 0.01

 

The table below reports four regression models.

 

 
 

Standardized Coefficients for Regression of TRI Density (Square Root)

 

Independent Variables

Model 1

Model 2

Model 3

Model 4

Constant

-0.259***

-0.344***

-0.250***

-0.386***

 

Percent Black

0.066***

0.123***

 

 

 

Percent Hispanic

 

 

0.064***

-0.103***

 

Population Density (natural log)

0.494***

0.469***

0.485***

0.518***

 

Percent Industrial

0.310***

0.273***

0.303***

0.278***

 

Percent Poverty

 

-0.047*

 

0.053**

 

Percent Manufacturing

 

0.273***

 

0.295***

 

 

 

 

 

N

1,933

1,933

1,933

1,933

Adjusted R2

0.430

0.500

0.429

0.496

* p < 0.05,** p < 0.01,*** p < 0.005

 
 

 

 

Conclusion

Concentrations of TRI facilities in New Jersey are associated with high concentrations of African Americans and Hispanics.  Population density, industrial/commercial land use, and employment in manufacturing  explain much, though not all, of the variation in these statistical relationships