5-7, we employ these variables to assess the validity of the environmental
injustice argument. Specifically, Chapter 5 examines the argument within
the context of U.S. states; Chapter 6 is concerned with the extent of environmental injustice at the county level; and Chapter 7 deals with this issue
within the context of cities of more than 50,000 inhabitants. In each of these
chapters, we discuss the nature of the environmental harms that are examined at each level, the sample of jurisdictional units that we utilized in each
analysis, a justification for the use of each jurisdictional unit analyzed, and
any idiosyncratic research problems that we encountered at each level of
analysis. We begin with the state-level analysis before proceeding to the
county and city level analyses.
1. Some research does report null class-risk relationships ( Adeola, 1994; Allen, Lester, and
Hill, 1995; Gianessi,
Wolfe, 1979; Lester and
Allen, 1996; West, 1992). Other research reports counterintuitive findings. Bowman and
( 1995, 1997) report that as per capita income increases, so does the presence
of hazardous waste sites or generators in South Carolina counties. Hird ( 1993, 1994)
also reports similar findings for the distribution of NPL sites in 3,000 U.S. counties. 2.
Measurement of social class at the state level was achieved by a four-item factor
scale that combined traditional measures of education and income. The variables included in the factor scale and their factor loadings were: median family income, 1980,
in $1,000,.89; per capita income, 1985, in $1,000, .89; percent college graduates, 1980,
.81; percent population 18-24 years of age in college, 1980, .65. The eigenvalue for
the factor was 2.68, and the unidimensional factor generated by principal components analysis explained 67.1 percent of the variance in the pool of four variables.
The county social-class scale resulted from a principal components factor analysis of
six traditional indicators of social class. Two measures in the scale, median household
income and percent of the population with a B.A. degree, were subjected to square
root transformations in order to normalize univariate distributions, while median
value of owner occupied housing was subjected to a log transformation for the same
reason. Factor loadings for the single unrotated factor were: median family income,
.85; median household income, .94; median value of owner occupied housing, .87;
percent population B.A. degree, .84; percent population with high school degree, .77;
and per capita income, .55. The factor has an eigenvalue of 4.17 and explains 69.6
percent of the variance in the pool of measures. The city-level social-class measure
was constructed via a principal components factor analysis of 1990 per capita income, 1989 median family income, and 1990 percent of the population with B.A. dev
gree or higher. The factor analysis was performed on data that had high values recoded to eliminate outliers. Once recoded, the measures were subjected to square
root transformations (per capita income and median household income) or a log
transformation (percent population with a bachelor's degree). Factor loadings on the
city social-class scale are: per capita income, 1989, in $1,000, .95; median household
income, 1989, in $1,000, .85; percent population, bachelor's degree or higher, 1990,
Questia, a part of Gale, Cengage Learning. www.questia.com
Book title: Environmental Injustice in the United States:Myths and Realities.
Contributors: James P. Lester - Author, David W. Allen - Author, Kelly M. Hill - Author.
Publisher: Westview Press.
Place of publication: Boulder, CO.
Publication year: 2001.
Page number: 74.
This material is protected by copyright and, with the exception of fair use, may
not be further copied, distributed or transmitted in any form or by any means.