The Social Science Journal 38 (2001) 13–25
Social class, race, and toxic releases in American counties, 1995 David W. Allen* Department of Political Science, Colorado State University, Fort Collins, Colorado 80523-1782, USA
Abstract This article, using data on 2,083 counties in 1995, tests the environmental racism/classism hypotheses and concludes that both have merit, however, specific findings demonstrate that the relationships are more complex than heretofore reported. Class and race relationships are conditional: while high social class reduces the level of toxic releases, it does so by moderating the relationship between fiscal capacity, pollution potential and this environmental harm. Further, while toxic releases increase as a function of the Black population, this relationship is stronger in the Sunbelt. © 2001 Elsevier Science Inc. All rights reserved.
1. Introduction The environmental racism/classism hypotheses ask whether environmental hazards disproportionately affect minorities and the poor. Quantitative research supports both positions by reporting race/class correlates for a variety of environmental harms (Adeola, 1994; Allen, Lester, & Hill, 1995; Anderton et al., 1994, 1994a; Asch & Seneca, 1978; Attah, 1992; Been, 1994; Berry, 1977; Boer et al., 1997; Boerner & Lambert, 1995; Bowen et al., 1995; Bowman & Crews-Meyer, 1997; Bullard, 1990; Burch, 1976; Burke, 1993; Costner & Thornton, 1990; Cutter, 1994; Davies, 1972; Downey, 1998; Freeman, 1972; Gianessi, Peskin, & Wolfe, 1979; Gianessi & Peskin, 1980; Gould, 1986; Greenberg, 1993, 1994; Hamilton, 1993, 1995; Handy, 1977; Harrison, 1975; Hird, 1993, 1994; Hird & Reese, 1998; Kohlhase, 1991; Krieg, 1995, 1998; Kruvant, 1975; Lester & Allen, 1996; Mohai & Bryant, 1992; Perlin et al., 1995; Polloch & Vittas, 1995; Ringquist, 1997; Shaikh & Loomis, 1998; Szasz et al., 1993; United Church of Christ, 1987; US Council on Environmental Quality,
* Tel.: ⫹1-970-491-5751. 0362-3319/01/$ – see front matter © 2001 Elsevier Science Inc. All rights reserved. PII: S 0 3 6 2 - 3 3 1 9 ( 0 0 ) 0 0 1 0 9 - 9
14
D.W. Allen / The Social Science Journal 38 (2001) 13–25
1971; US General Accounting Office, 1983; West, 1992; White, 1992; Zimmerman, 1993, 1994; Zupan, 1973). This article, while studying both hypotheses, does focus on environmental racism in relation to toxic releases in American counties. While research into this subject should avoid the “race versus class” trap in discussing questions of environmental justice, it is important to focus on the environmental racism hypothesis. Unlike social class, individuals cannot move out of a racial classification; therefore, the race/risk nexus forms part of the corpus secundum of racial discrimination. Further, class, unlike race, has been widely used in previous research. Thus, much has been known, and for sometime, about its effect. Conversely, the race/environmental risk nexus allows for continued inquiry wherein “new” findings are possible. The dependent variable for this study is the level of toxic releases in American Counties in 1995 as abstracted from the Toxic Release Inventory (US Environmental Protection Agency, 1995). Using Toxic Release Inventory data are not without problems. The data base includes only manufacturing facilities falling within primary SIC codes 20 –39. Thus, not all toxic release facilities are included. Second, only SIC 20 –39 facilities with ten or more employees which process more than 25,000 pounds of the 350 TRI list chemicals/chemical compounds annually or who use more than 10,000 pounds of these chemicals per year file reports. Therefore, toxic releases from small firms and firms below the processing/use thresholds are not included—indicating that reasonably significant levels of toxic releases are not reflected in the Inventory. Third, TRI data are the result of industry self-reporting and reports only receive spot checks by the Environmental Protection Agency. This may mean that producers are under reporting emission levels (Krieg, 1998: 192–193). Finally, TRI data are a product of currently industrialized areas with active processors/users of a restrictive list of toxins. It does not reflect other toxins at abandoned waste dumps, landfills, or at other storage facilities. As a consequence, the TRI data should not be conceptualized as measuring the overall toxic waste in any community. However, use of the Toxic Release Inventory is still valid. Toxic waste sites of defunct industry, while dangerous, represent a fixed level of an environmental hazard. Toxins moved to storage facilities and active dumps do constitute an expanding supply of environmental hazards, but these locations are fixed and, to some degree, contain the spread of the waste. On the other hand, TRI data reflects continually expanding environmental hazards discharged directly to the land, air, and water. Thus, the Inventory represents annual amounts of released toxins which are not necessarily accounted for in other data bases. Finally, even though the data base does not cover all toxic chemicals, reported amounts are substantial and a cause for concern. Measurement of toxic releases presents another problem. Reporting only the total pounds of toxic releases masks the method of release. Releases occur as stack air, fugitive air, land, and water releases and the amounts of these individual release modes vary dramatically even though total releases can be highly similar for different locales. One way to account for this variation is to create a weighted scale using factor analysis. Principal components factor analysis of pounds of stack air, fugitive air, land, and water releases, abstracted from the 1995 Toxic Release Inventory and transformed to eliminate extreme outliers, produced a single unrotated factor.1 High positive scores reflect high levels of toxic releases and high negative scores reflect the opposite.
D.W. Allen / The Social Science Journal 38 (2001) 13–25
15
2. Concepts, measurement, and hypotheses 2.1.1. Race. By focusing on the Black population it is possible to determine whether this group, the object of a high level of discrimination in other forums, is similarly situated with regard to environmental hazards. However, focusing only on this group also means that the ability to generalize findings to all minority groups is sacrificed in order to obtained detailed information about a specific group which has historically suffered visible and long term discrimination in this county. The race variable is measured as the 1990 % Black population in each county as enumerated in the County and City Data Book (US Department of Commerce, Bureau of Census, 1994).2 The hypothesis, drawn from existing literature, is: increasing levels of the Black population, when other explanations of environmental hazards have been held constant, is positively associated with increased levels of toxic releases. An extensive literature review located measures used, not only in the environmental racism/classism literature, but in literature which dealt with environmental pollution in general. The review lead to the development of measures specifically applicable to the “classism” hypotheses (social class), to measures which were applicable to explaining the level of pollution in counties (political mobilization, fiscal capacity, pollution potential, and land area), and a series of state level measures which account for pollution levels (interest groups, political institutions/institutional arrangements, public opinion, political culture, and region). Scores on state level measures were given to each county in the sample. 2.2. County level measures 2.2.1. Social class. “Environmental classism” maintains that poor communities are targeted for environmental hazards. Land is cheap in poor areas, possibly because of existing industry. New industries select these areas as potential locations because of depressed land prices and existing infrastructure. Additionally, wealthier locales, with well-educated individuals, want to protect against decreasing property values, and, thus, polluting industries meet opposition when trying to locate in “upscale” areas. The social class measure results from principal components factor analysis of six traditional class-based indicators.3 High positive factor scores indicate high social class while the opposite end of the scale indicates low social class. The hypothesis reflects the prevailing trend in the literature, that is, a negative relationship between social class and toxic releases. 2.2.2. Political mobilization. Politically mobilized communities capture the attention of decision makers. Increased political mobilization should have the effect of minimizing environmental harms because policy makers are likely to pay attention to problems articulated by this type of community (Hamilton, 1993, 1995; Hird, 1993, 1994; Hird & Reese, 1998). Political mobilization was measured by dividing total votes cast in the 1992 presidential election by 1990 total population for a crude measure of voter turnouts. High scores indicate higher levels of political mobilization. A negative relationship is expected between political mobilization and toxic releases. 2.2.3. County fiscal capacity. A political jurisdiction with more funds can exercise more influence over a policy problem such as environmental hazards (Ringquist, 1993). County
16
D.W. Allen / The Social Science Journal 38 (2001) 13–25
revenues and expenditures, subject to principal components factor analysis, yielded an unidimensional scale which was used to measure this concept.4 High positive scores reflect high fiscal capacity while high negative scores reflect the opposite. A negative relationship is expected between county fiscal capacity and toxic releases. 2.2.4. Pollution potential. Surrogates of pollution potential—total population, population density, and manufacturing capacity (Allen, Lester, & Hill, 1995; Anderton et al., 1994; Been, 1994; Bowman & Crews-Meyer, 1997; Cutter, 1994; Hamilton, 1993, 1995; Hird, 1993, 1994; Hird & Reese, 1998; Holm, 1994; Krieg, 1998; Lester & Allen, 1996; Ringquist, 1997)—were subjected to principal components factor analysis in order to create a weighted unidimensional scale.5 High positive factor scores indicate high pollution potential. A positive relationship is expected between this measure and toxic releases. 2.2.5. Land area. The geographic size of counties varies dramatically. To account for this variation, log transformed county land area in square miles in 1990 was included as a control variable. 2.3. State level measures 2.3.1. Interest groups. Interest groups influence political decisions (Ringquist, 1993). A well-constructed equation therefore has to account for organized interests. Membership in the Sierra Club, National Wildlife Federation, and Friends of the Earth per 1,000 persons in each state in 1987 was used to measure this concept (Hall & Kerr, 1991). 2.3.2. Political institutions and institutional arrangements. The political institution of interest is the state legislature. Existing literature has demonstrated that more professional state legislatures are better able to deal with problems of environmental pollution. Squire’s (1992) index of legislative professionalism is adopted for this article.6 A negative relationship is expected between professionalism and toxic releases. Institutional arrangements also effect the level of environmental hazards in a community. These arrangements include the level of policy commitment and the fiscal capacity dedicated to promoting environmental protection. Lester (1994) developed a four-cell topology which reflects these twin considerations. States with “high” policy commitment and capacity are labeled “progressive” states. Those with a high policy commitment but a limited capacity are labeled “strugglers.” The third classification, “delayers,” consist of states with a low policy commitment but a relatively high capacity. Finally, “regressive” states are low in terms of both policy commitment and capacity. The categories are measured as dummy variables, with “1” equal to the presence of the category and “0” if otherwise. The regressive states constitute the default category. Counties in progressive states, because of the state’s high commitment and capacity, should have lower levels of toxic releases. Counties in delayer and struggler states, because of an absence of commitment or a lack of capacity, should have higher levels of toxic releases. 2.3.3. Public opinion and political culture. In a democratic society, public opinion is supposed to shape government response. This linkage is well understood. Two forms of
D.W. Allen / The Social Science Journal 38 (2001) 13–25
17
public opinion are partisanship and ideology, both of which have been measured by opinion data (Wright, Erickson, & McIver, 1985). Previous literature has demonstrated that Republicans and conservatives are not within the mainstream of the environmental protection movement, and because high scores on the indices reflect both considerations, the hypothesis posits a positive relationship between these concepts and toxic releases. Political culture constrains the modes of political action available to political institutions and resultant institutional arrangements. The predominant political cultures in the United States are the individualistic, moralistic, and traditionalistic cultures. The individualistic and moralistic cultures capture our attention because of their emphasis on the marketplace and private concerns and because substantial segments of the American polity operate within the two cultures (Elazar, 1972: 94 –96). The individualistic culture places a premium on limiting community intervention into private activities to the minimum necessary to keep the marketplace in proper working order (Elazar, 1972: 94) and favors the needs of the business community over those of the general polity. As a consequence, environmental hazards, a byproduct of industry, would be more tolerated in an individualistic culture. The moralistic culture emphasizes politics as activity centered on some notion of the public good properly devoted to the advancement of the public interest (Elazar, 1972: 96). A moralistic culture has less tolerance for environmental pollution because it affects the public good. As a consequence, the moralistic culture requires a cleaner and safer environment. Johnson’s (1976) moralistic and individualistic culture indices are used in this research.7 2.3.4. Region. The Environmental Protection Agency has divided the nation into a series of regions which are collapsed, for purposes of statistical controls, into the northeast, northcentral, western, and, Sunbelt regions. These dummy variables, scored “1” for the presence of a region and “0,” if otherwise, account for the regional distribution of toxic releases. The Sunbelt region, the default category in this measurement system, is used as a control variable in a later section of the analysis. This decision was made because large percentages of the Black population resides within the Sunbelt, and this region has, within the past two decades, developed a large number of polluting industries. A race/Sunbelt interaction is expected to produce a stronger relationship between race and toxic releases in the Sunbelt than is evident elsewhere in the nation.
3. Sample and analysis While a variety of existing research has focused on the smallest possible units of analysis, for example, zip codes, census tracts, and SMSAs, and while this strategy does establish proximity to a hazard it does entail a flaw with regard to policy making. The units capable of effectuating/implementing policy solution are the federal, state, county, and city governments. There are no zipcode, census tract or SMSA policy generating/implementing structures. Thus, it makes sense to study an entity that will have to oversee elimination of environmental racism—and counties constitute one of these entities. Thus, the sample consists of 2,083 counties, or roughly 66% of all counties in the United States.8 Data are analyzed using a two-step process. Eq. (1) incorporates known explanations for
18
D.W. Allen / The Social Science Journal 38 (2001) 13–25
toxic emissions, including class but excluding race. Additionally, exhaustive sets of two-way interactive terms, generated from the pool of independent variables, were also included in the first equation. Interactive terms, entered one at a time, were retained if they were significant at the p ⬍ .05 level and if their inclusion increased the equation’s goodness-of-fit in a statistically significant fashion (p ⬍ .05). The first step equation takes the following form: yToxic Releases ⫽ b0 ⫺ b1Class ⫾ bnxn ⫾ bnIn9
(1)
where: xn ⫽ additional explanations; and, In ⫽ interactive terms. The second equation isolates the relationship between race and toxic releases once all other explanations have been accounted for. This equation’s dependent variable is the residuals from Eq. (1). The residuals are conceptualized as a county’s level of toxic releases once all other explanations are held constant. Because Eq. (1)’s regional variables have a default category that is of interest, the Sunbelt, this regional variable is also included in the second equation: first as a simple linear equation, yres ⫽ b0 ⫹ b1Race ⫹ b2Sunbelt;
(2)
and, then as an interactive equation: yres ⫽ b0 ⫹ b1Race ⫹ b2Sunbelt ⫹ b3Race ⫻ Sunbelt
(2a)
4. Results Table 1 reports the results of Eq. (1) in reduced form. This equation explains 33% of the variance in level of toxic releases in American counties. Social class moderates the relationship between fiscal capacity, pollution potential and toxic releases. The class/capacity interaction leads to an unexpected outcome. The negative slope for the interactive term indicates that social class moderates the relationship between fiscal capacity and toxic releases and minimizes the level of this environmental hazard. Possibly high social class communities require the county to devote more fiscal resources toward minimizing environmental harms. Social class moderates the relationships between pollution potential and toxic releases. Counties with wealthier and more educated citizens either block the type of industry that would produce high levels of toxic releases or, in the alternate, would require industries to lower their toxic outputs. A third conditional relationship between county land area and counties in the northeast indicates that large counties in this region experience significantly higher levels of toxic releases than is the case for other counties in this region. In sum, the northeast is home to old and new industries which have high levels of toxic producing entities, and apparently these industries are located in this region’s larger counties. The north-central region also evidences a positive relationship with toxic releases—indicating that this region also has large numbers of facilities which produce this environmental harm. A fourth interactive term indicates that under conditions of a moralistic culture the ability
D.W. Allen / The Social Science Journal 38 (2001) 13–25
19
Table 1 Explaining Toxic Releases, American Counties, 1995 (Including Class, Excluding Race): Results from Reduced Equation 1 (N ⫽ 2083 Counties). Variable Social Class Political Mobilization County Fiscal Capacity Pollution Potential Land Area Environmental Interests Legislative Professionalism Progressive State Delayer State Struggler State Partisanship Ideology Individualistic Culture Moralistic Culture Northcentral Region Northeastern Region Western Region Class ⫻ County Fiscal Capacity Class ⫻ Pollution Potential Ideology ⫻ Individualistic Culture Environmental Interest ⫻ Moralistic Culture Land Area ⫻ Northeastern Region Constant
ba .003 n/a .163 .579**** .050* ⫺.151*** ⫺.134*** n/a .231*** n/a .087** ⫺.60* .056 ⫺.079* .169* .013 n/a ⫺.083*** ⫺.058** ⫺.102*** ⫺.86*** .323*** ⫺.196***
t-statistic .130 n/a 6.713 22.086 2.387 ⫺4.137 ⫺4.734 n/a 4.705 n/a 2.640 ⫺2.110 1.736 ⫺2.211 2.525 .170 n/a 4.877 ⫺3.082 ⫺3.755 ⫺2.858 3.016 ⫺3.811
Adj. R Sq.: .331 F-ratio Equation: 50.29**; Standard Error of Estimate: .81; ****p ⬍ .0001; ***p .001 to .008; **p ⬍ .01; *p ⬍ .05. a Cell entries are unstandardized regression coefficients computed on the basis of standardized normalized data. “n/a” indicates terms that were excluded during the reduction process.
of environmental interests to minimize pollution is enhanced. Environmental interests seek to minimize pollution. A moralistic culture emphasizes politics as centered on the notion of public good and advancement of public’s interests. Under conditions of a moralistic culture, efforts of environmental groups are enhanced because the predominant cultural effect encourages interest group activity which is devoted to the public interest. The relationship between legislative professionalism and toxic releases is, as hypothesized, negative. Professional legislatures are apparently better qualified to navigate their way though the technical issues which surround environmental hazards and therefore create policies designed to minimize resultant harms. Counties within these states apparently benefit from the statewide policies developed by professionalist legislatures because toxic releases are lower in these counties. The capacity/commitment topology of states adopted for this research produced results which are contrary to and in accordance with the various hypotheses. Results indicate the absence of relationships between progressive and struggler states and toxic releases and a positive relationship between delayer states and the level of environmental hazards. The relationship between partisanship and toxic releases, although small, is in the hypothesized direction. Counties within highly Republican states have higher levels of toxic
20
D.W. Allen / The Social Science Journal 38 (2001) 13–25
releases. This follows the general trend in the literature which maintains that the Democratic party is more concerned with environmental protection. While the relationship between partisanship and toxic releases is as hypothesized, the second public opinion variable, ideology, evidences a conditional relationship with toxic releases based on political culture. The ideology/individualistic culture conditional relationship can be explained in post hoc fashion. A conservative ideology is concerned, in part, with property values and toxinproducing facilities diminish the value of this commodity. Counties in highly conservative states which also operate within the confines of an individualistic culture, a culture which is concerned with the marketplace value of property, would want to safeguard property values. As such, in inadvertent fashion, counties in highly conservative individualistic states wind up with marginally better environmental conditions. Using the residuals from Eq. (1), the relationship between race and toxic releases can now be assessed. Eq. (2) reveals a small but significant relationship between race and toxic releases—a finding which replicates the general trend in the literature. y Toxic Releases ⫽ .029 ⫹ .05关 percent black兴* ⫺ .071关region兴
(2)
However, Eq. (2a), which includes the race/region interactive term, indicates that the relationship between race and toxic releases is stronger in the Sunbelt region. yToxic Releases ⫽ .011 ⫹ .011关 percent black兴 ⫺ .101关Sunbelt兴* ⫹ .113关black ⫻ Sunbelt兴** **p ⬍ .01 *p ⬍ .05
(2a)
Indeed, the manipulation of the intercept and slope terms from Eq. (2a), noted in Eq. (2a.1), indicated that the coefficient between race and toxic releases, when the Sunbelt is taken into account, increases to b ⫽ 0.124. Sunbelt Region: yToxic Releases ⫽ ⫺ .091 ⫹ .124关 percent black兴 Non-Sunbelt: yToxic Releases ⫽ .001 ⫹ .001关 percent black兴
(2a.1)
This effect can be explained: large amounts of toxic releases occur in the Sunbelt which contains 858 of the counties included in the analysis. Counties within this region are also home to large percentages of Black citizens. The combination of these two terms results in Black citizens being subjected to higher levels of toxic releases in this region than is the case elsewhere in the nation.
5. Conclusions The major findings in this study support the environmental racism and classism hypotheses. Minorities and the poor are disproportionately affected by environmental harms. However, current results also demonstrate that the race/class/risk nexus, as regard toxic releases, is more complex than anticipated. As the percentage of the Black population in American counties in 1995 increased so did the level of toxic releases; however, this relationship was stronger in the Sunbelt, indicating that larger proportions of Black Amer-
D.W. Allen / The Social Science Journal 38 (2001) 13–25
21
icans are exposed to higher levels of toxic releases in Sunbelt counties that are the case elsewhere in the nation. The current research, using counties as the unit of analysis, has replicated results from studies which used units of analysis both more finely graduated (urban neighborhoods and large cities) and more grossly graduated (American counties). Also, current results replicate findings generated on the basis of more restricted samples. This commonality would tend to indicate that existing findings do not appear to be an artifact of either the unit of analysis employed or sample size. Replicating results under these conditions increases confidence in the race/toxic release nexus. Second, while class-based findings support the environmental classism hypothesis, the current classism results also indicate that the relationship is more complex than originally anticipated, that is social class interacts with both fiscal capacity and pollution potential such that the intersection of high class, high fiscal capacity, and high pollution potential leads to lower levels of toxic releases. Two sets of ancillary findings, separate and apart from the environmental racism/classism hypotheses, provide additional information about the level of toxic releases in American counties. First, conditional relationships, involving ideology, interest groups and culture, and one simple relationship, involving legislative professionalism, provides information about lower levels of toxic releases in American counties. A conservative ideology is concerned, in part, with the maintenance of property values—and property values decrease in areas where toxic releases are high. An individualistic culture is also concerned with the marketplace value of property and seeks to preserve property values. As such, in inadvertent fashion, counties in highly conservative and highly individualistic states do wind up with marginally better environmental conditions as regards toxic releases. Additionally, as expected, increases in interest group size decrease the level of toxic releases. However, environmental group efforts were magnified if they operated within the confines of a moralistic culture, that is toxic releases decreased out of proportion to the size of the interest group because the group’s purpose is augmented by a cultural ethos that seeks to protect the public good. Finally, legislative professionalism also decreases the level of toxic releases in American counties. States with professionalized legislatures are probably better able to deal with the technical issue of toxic releases—and the state-level ability to manage the problem apparently filters down to substate regions, that is, counties. The second set of ancillary findings provides information about increases in toxic releases in American counties. Counties in delayer states evidence increasing levels of toxic releases. There was no relationship between the progressive and struggler categories and level of toxic releases. Possibly these findings are a consequence of measure construction. Further, the relationship between partisanship and toxic releases indicates that counties in states with a Republican bias have higher levels of toxic releases. Finally, region is related to the level of toxic releases in two out of three instances: the north-central region and large counties in the northeast both evidence high levels of toxic releases. Both regions have high concentrations of toxic producers, however, in the northeast these producers seem to be concentrated in larger counties. The west evidenced no relationship with level of toxic releases. Two explanations may account for this null finding. First, the distribution of toxic release may be such that they are not associated with the west. This is a weak explanation, since California fits into the western region. Secondly, and probably a more reasonable explanation for the reported outcome, the west contained the highest number of missing counties of all regions.
22
D.W. Allen / The Social Science Journal 38 (2001) 13–25
Thus, the null relationship in the west could be a function of missing data, and therefore, this finding should be taken merely as advisory for the present.
Notes 1. High values on each release component, interpreted as outliers, were sequentially recoded at one integer higher than the highest break in the distribution. A log transformations was used to normalize univariate distributions after a constant of “1” was added to each case to eliminate zeros. Factor loadings: stack air, 0.812; fugitive air, 0.812; water releases, 0.764; land releases, 0.654. Eigenvalue: 2.33. Explained variance: 58.3%. 2. A log transformation was used to correct a skewed distribution. 3. Unless otherwise noted, the data source for independent variables is the County and City Data Book, 1994. To normalized univariate distributions, three measures included in the social class factor scale were given various transformations. Social class factor loadings: median family income, 1989, 0.954; square root median household income, 1989, 0.940; log median value owner occupied housing, 1989, 0.879; square root percentage population B.A. degree, 1990, 0.840; percent population high school degree, 1990, 0.773; per capita income, 1989, 0.550. Eigenvalue: 4.17. Explained variance: 69.6%. 4. Square root transformations, prior to factor analysis, were used to normalize univariate distributions. County capacity factor loadings: local government finances, per capita taxes, 1986 –1987, 0.748; federal funds and grants, total expenditures, 1992, 0.665; general revenues per $1,000 personal income, 1986 –1987, 0.558; per capita expenditures, federal funds, 1992, 0.537. Eigenvalue: 1.62. Explained variance: 40.0%. 5. The manufacturing variables had a constant of 0.001 added to each case to eliminate zeros so that the log transformation could be used. Pollution potential factor loadings: log number of manufacturing establishments, 1987, 0.954; log total population, 1990, 0.950; log population density, 1990, 0.902; log number of manufacturing employees, 1987, 0.761. Eigenvalue: 3.20. Explained variance: 80.0%. 6. Squire’s index contains a series of outliers recoded as follows: New York, 0.659 to 0.309; Michigan, 0.636 to 0.308; California, 0.626 to 0.307; Massachusetts, 0.614 to 0.306; Pennsylvania, 0.336 to 0.305; Ohio, 0.329 to 0.304. 7. Johnson’s (1976) indices contain outliers which were recoded. Moralistic culture: South Dakota, 0.37 to 0.34; Wyoming, 0.42 to 0.35; Kansas, 0.44 to 0.36; Idaho, 0.68 to 0.37; Utah, 0.93 to 0.38. Individualistic culture: Rhode Island, 0.87 to 0.83; New Mexico, 0.84 to 0.82. 8. The loss of cases was produced by a variety of problems. First, the deletion of Alaska and Hawaii, because of missing data on the public opinion and political culture measures, reduced the sample to 3,106 counties. Second, a large number of counties had no facilities reporting to the Toxic Release Inventory. This absence of reports for the dependent variable further reduced the N of counties to 2,097 counties. Finally, preliminary analysis revealed statistically significant residual outliers, causing deletion
D.W. Allen / The Social Science Journal 38 (2001) 13–25
23
of an additional 14 cases. A complete analysis of missing data on county level TRI releases and its relationship to both the regional and overall state environmental capacity measures is available from the author on request. 9. Because interactive terms are employed, all variables which can be standardized are standardized prior to construction of multiplicative terms. Reported results are unstandardized regression coefficients generated on the basis of standardized data (Aiken & West, 1991: 35). Achen’s (1982: 65) procedure was employed to reduce Eq. (1).
References Achen, C. H. (1982). Interpreting and using regression. Beverly Hills: Sage Publications. Adeola, F. O. (1994). Environmental hazards, health, and racial inequity in hazardous waste distribution. Environment and Behavior, 26, 99 –126. Aiken, L. S., & West, S. G. (1991). Multiple regression: testing and interpreting interactions. Newbury Park, NJ: Sage Publications. Allen, D. W., Lester, J. P., & Hill, K. M. (1995). Prejudice, profits and power: assessing the eco-racism thesis at the county level. Paper presented at the meeting of the Western Political Science Association, Portland, Oregon. Anderton, D., Anderson, A., Oates, J. M., & Fraser, M. (1994). Hazardous waste facilities: environmental equity issues in metropolitan areas. Evaluation Review, 18, 123–140. Anderton, D., Anderson, A., Oates, J. M., & Fraser, M. (1994a). Environmental equity: the demographics of dumping. Demography, 31, 229 –248. Asch, P., & Seneca, J. J. (1978). Some evidence on the distribution of air quality. Land Economics, 54, 278 –297. Attah, E. B. (1992). Demographics and siting issues in region IV. Proceedings of the Clark Atlanta University and EPA Region IV Conference on Environmental Equity. Been, V. (1994). Locally undesirable land uses in minority neighborhoods: disproportionate siting or market dynamics? Yale Law Journal, 103, 1383–1422. Berry, B. J. (1977). The social burdens of environmental pollution. Cambridge, MA: Ballinger Publishing Company. Boer, J. T., Pastor Jr., M., Sadd, J. L., & Synder, L. D. (1997). Is there environmental racism? The demographics of hazardous waste in Los Angeles County. Social Science Quarterly, 78, 793– 810. Boerner, C., & Lambert, T. (1995). Environmental justice in the City of St. Louis: the economics of siting industrial and waste site facilities. Center for the Study of American Business. Washington University, Campus Box 1208, One Brookings Drive, St. Louis, MO. 63130. Bowen, W. M., Salling, M. J., Hayes, K. E., & Cryan, E. J. (1995). Toward environmental justice: spatial equality in Ohio and Cleveland. Annuals of the Association of American Geographers, 85, 641– 663. Bowman, A., & Crews-Meyer, K. (1997). Locating southern LULUs: race, class, and environmental justice. State and Local Government, 29, 110 –119. Bullard, R. D. (1990). Dumping in Dixie: race, class, and environmental quality. Boulder, CO: Westview Press. Burch, W. R. (1976). The peregrine falcon and the urban poor: some sociological interrelations. In P. Richerson & J. McEvoy (Eds.), Human ecology: an environmental approach. Belmont, CA: Duxbury Press. Burke, L. M. (1993). Race and environmental equity: a geographic analysis in Los Angeles. Geo Info Systems. (October): 44 –50. Costner, P., & Thornton, J. (1990). Playing with fire: hazardous waste incineration. Washington, DC: Greenpeace. Cutter, S. (1994). The burdens of toxic risks: are they fair? Business and Economic Review, 40, 1–25. Davies, J. (1972). The role of social class in human pesticide pollution. American Journal of Epidemiology, 96, 223–238.
24
D.W. Allen / The Social Science Journal 38 (2001) 13–25
Downey, L. (1998). Race and income as predictors of environmental injustice. Social Science Quarterly, 79, 766 –778. Elazar, D. (1972). American federalism: a view from the states (2nd ed.). New York, NY: Thomas Crowell Co. Freeman, M. A. (1972). The distribution of environmental quality. In A. Kneese & B. T. Bower (Eds.), Environmental quality analysis. Baltimore, MD: Johns Hopkins Press. Gianessi, L., & Peskin, H. (1980). The distribution of federal water pollution control policy in the United States. Land Economics, 56, 25–102. Gianessi, L., Peskin, H., & Wolfe, E. (1979). The distributional effects of uniform air pollution policy in the US. Quarterly Journal of Economics, (May), 281–301. Gould, J. M. (1986). Quality of life in American neighborhoods: levels of affluence, toxic waste and cancer mortality in residential zip code areas. Boulder, CO: Westview Press. Greenberg, M. R. (1993). Proving environmental equity in siting locally unwanted land uses. Risk—Issues in Health and Safety, Summer, 235–252. Greenberg, M. R. (1994). Separate and not equal: health-environmental risk and economic-social impacts in remediating hazardous waste sites. In S. K. Majumdar, et al. (Eds.), Environmental contaminants and health. Philadelphia, PA: Pennsylvania Academy of Sciences. Hall, B., & Kerr, M. L. (1991). 1991–1992 green index. Washington, DC: Island Press. Hamilton, J. T. (1993). Policies and social cost: estimating the impact of collective action on hazardous waste facilities. Rand Journal of Economics, 24, 101–125. Hamilton, J. T. (1995). Testing for environmental racism: prejudice, profits, political power. Journal of Policy Analysis and Management, 14, 107–132. Handy, F. (1977). Income and air quality in Hamilton, Ontario. Alternatives, 6, 18 –24. Harrison, D. (1975). Who pays for clean air. Cambridge, MA: Ballinger. Hird, J. A. (1993). Environmental policy and equity: the case of superfund. Journal of Policy Analysis and Management, 12, 323–343. Hird, J. A. (1994). Superfund: the political economy of environmental risk. Baltimore, MD: Johns Hopkins. Hird, J. A., & Reese, M. (1998). The distribution of environmental quality: an empirical analysis. Social Science Quarterly, 79, 693–716. Johnson, C. A. (1976). Political culture in the American states: Elazar’s formulation examined. American Journal of Political Science, 20, 491–509. Kohlhase, J. (1991). The impact of toxic waste on housing values. Journal of Urban Economics, 30, 1–26. Krieg, E. J. (1995). A socio-historical interpretation of toxic waste sites. American Journal of Economics and Sociology, 54, 1–14. Krieg, E. J. (1998). Methodological considerations in the study of toxic waste hazards. Social Science Journal, 35, 191–201. Kruvant, W. J. (1975). People, energy and pollution. In D. K. Newman & D. Day (Eds.), The American energy consumer. Cambridge, MA: Ballinger Publishing Company. Lester, J. P. (1994). The states and environmental policy: a new federalism. In N. Vig, & Kraft, M. (Eds.) Environmental policy in the 1990s. Washington, DC: Congressional Quarterly Press. Lester, J. P., & Allen, D. W. (1996). Prejudice, profits and power: reassessing the eco-racism thesis at the city level. Paper presented at the Western Political Science Association, San Francisco, California. Mohai, P., & Bryant, B. (1992). Environmental racism: reviewing the evidence. In B. Bryant & P. Mohai (Eds.), Race and the incidence of environmental hazards: a time for discourse. Boulder, CO: Westview Press. Perlin, S. A., Setzes, R. W., Creason, J., & Sexton, K. (1995). Distribution of industrial air emissions by income and race in the United States: an approach using the toxic release inventory. Environmental Science and Technology, 29, 69 – 80. Polloch, P., & Vittas, M. (1995). Who bears the burden of environmental pollution? Race, ethnicity, and environmental equity in Florida. Social Science Quarterly, 76, 294 –310. Ringquist, E. J. (1993). Environmental protection at the state level: politics and progress in controlling pollution. New York, NY: M.E. Sharp.
D.W. Allen / The Social Science Journal 38 (2001) 13–25
25
Ringquist, E. J. (1997). Equity and the distribution of environmental risks: the case of TRI facilities. Social Science Quarterly, 78, 811– 829. Shaikh, S., & Loomis, J. B. (1998). An investigation into the presence and causes of environmental inequity in Denver, Colorado. Social Science Journal, 36, 77–92. Squire, P. (1992). Legislative professionalism and membership diversity in state legislatures. Legislative Studies Quarterly, 17, 69 –79. Szasz, A. M., Meiser, H., Aronson, H., & Fukura, H. (1993). The demographics of proximity to toxic releases: the case of Los Angeles County. Paper presented at the meeting of the American Sociological Association, Miami, Florida. United Church of Christ, Commission for Racial Justice. (1987). Toxic wastes and race: a national report on the racial and socioeconomic characteristics of communities with hazardous waste sites. New York, NY: United Church of Christ. US Council on Environmental Quality. (1971). Annual report to the president. Washington, DC: Government Printing Office. US Department of Commerce, Bureau of the Census. (1994). County and city data book 1994. Washington, DC: Government Printing Office. US Environmental Protection Agency. (1995). Toxic release inventory database. http://rtk.net/data/TRIgen.html. US General Accounting Office. (1983). Siting of hazardous waste landfills and their correlations with racial and economic status of surrounding communities. Washington, DC: Government Printing Office. West, P. C. (1992). Minority anglers and toxic fish consumption: evidence from a state-wide survey of Michigan. In B. Bryant & P. Mohai (Eds.), Race and the incidence of environmental hazards: a time for discourse. Boulder, CO: Westview Press. White, H. L. (1992). Hazardous waste incinerators and minority communities. In B. Bryant & P. Mohai. (Eds.), Race and incidence of environmental hazards. Boulder, CO: Westview Press. Wright, G. C., Erickson, R. S., & McIver, J. P. (1985). Measuring state partisanship and ideology with survey data. Journal of Politics, 47, 469 – 489. Zimmerman, R. (1993). Social equity and environmental risk. Risk Analysis, 13, 649 – 666. Zimmerman, R. (1994). Issues of classification in environmental equity: how we manage is how we measure. Fordham Urban Law Journal, 21, 633– 669. Zupan, J. M. (1973). The distribution of air quality in the New York region. Baltimore, MD: Johns Hopkins University Press.