. nu lmum v. :- . Tim... , 31W? . LEW. 1.; “wantfi‘wgwq 3%.? ,qt 1‘ J! A . .. V.‘ ....v ‘ J . .‘uu..”1....v.snu. V. :3... 11.1%,...1'3. u..§,.L...WSE.W!.txz.icihn Two rural and low income areas, one white and the other minority in racial/ethnic make-up. > Two suburban and moderate income areas, one white and the other minority in racial/ethnic make-up. > Two urban and high income areas, one white and the other minority in racial/ethnic make-up. r Two rural and high income areas, one white and the other minority in racial/ethnic make-up. v Two urban and low income areas, one white and the other minority in racial/ethnic make-up. These study area categories are summarized below in Table 1. 2'3 Fahsbender, op. cit, pp. 123, 135, and 144-145. The researcher notes that during 1997 participation in the City of Detroit’s Community Reinvestment Strategy initiative, participants self-organized the city into zip code areas along which was attached certain status, cultural, ethnic and economic identity. 2" Adeola, op. cit, p. 107 and Cutter, op. cit, p. 117. 72 Table 1 WWW StudyAreas - _Lowlncome *fi Moderatelncome Hrghlncomei Urban One White and One Minority Study Area One White and One Minority Study Area Suburban One White and One Minority Study Area Rural One White and One Minority Study Area One White and One Minority Study Area U.S. Census Data: Specific variables utilized from the 1990 U.S. Census data included: Percent White Population Mean household income. Urban Population Inside Urbanized Areas Percent Black Population Urban Population Outside Urbanized Areas Percent American Indian, Eskimo and Aleut Population Rural Population Percent Asian and Pacific Islander Population Percent Other Race Population (Hispanic Origin) Percent Other Race Population I! ll! g") Definitions This study defines terms and measures the demographics of social equity in Michigan's Part 201 policy as described below. 73 Racial/ethnicity Parameters: While recognizing the above perspectives and potential for the introduction of bias in interpretation of findings”, for the purpose of this study race/ethnicity is defined using the 1990 U.S. Census definitions as follows: D White, comprised of persons reporting as White, Canadian, German, Italian, Lebanese, Near Eastemer, Arab or Polish. Black, consisting of persons reporting as Black, Negro, African American, Afro- American, Black Puerto Rican, Jamaican, Nigerian, West Indian or Haitian. American Indian, Eskimo or Aleut, defined as persons reporting as Indian, Canadian Indian, French-American Indian, Spanish-American Indian, Eskimo, Aleut or by tribal affiliation. This category is based upon self-identification, and is without regard to tribal status or the status of U.S. federal or state governmental recognition of tribes. Asian or Pacific Islander, comprised of persons reporting as Chinese, Filipino, Japanese, Asian Indian, Korean, Vietnamese, Cambodian, Hmong, Laotian, Thai, . Other Asian, Hawaiian, Samoan, Guamian, or Other Pacific Islander. Other of Spanish/Hispanic Origin, comprised of persons reporting as of Spanish/Hispanic origin (such as Mexican, Cuban, or Puerto Rican). Other, a final racial/ethnic category comprised of persons reporting as other than the above racial/ethnic groups such as multiracial, mixed, interracial or Wesort. For the purpose of study area identification, these U.S. Census Bureau categories of race were nominally aggregated into "white" and "minority", the latter including all U.S. Census racial/ethnic categories except "white". These combined racial/ethnic categories were then converted into a ratio of minorities to whites, and converted to a 2'5 Zimmerman, op. cit, pp. 663-669; Pulido, op. cit, p. 152; Goldman, 1996, op. cit, p. 134; and Perlin, et. al., op. cit, p. 72. 74 logarithmic scale to divide race into quartiles along a normal distribution?"5 The variable of race was treated as such and coded to allow for statistical analysis. Specifically, race is operationalized in quartiles for study area selection as follows: Ratio of Minorities to Whites: 1= No Minorities 2= Few Minorities 3= Some Minorities 4= Most Minorities Codes of 1, 2 and 3 were operationalized as "white" study areas, and a code of 4 was designated as a "minority" study area. Population Density: The population density of study areas is defined here according to the U.S. Census definitions as follows: v Rural areas are defined in this study to mean zip code areas with populations at or below 2,499 persons, and within which no metropolitan area217 exists. v Suburban areas are defined as zip code areas containing, not containing, or partially containing or be contiguous to a metropolitan area and the study area population is at or between 2,500 and 9,999 persons. > Urban areas are defined as zip code areas containing a metropolitan area and the study area population is at or above 10,000 persons. 2'6 Recommendation of Fahsbender, op. cit, pp. 140-142. 2'7 Metropolitan areas are defined using the U.S. Census definition of an area of a large population nucleus (minimum of 50,000 or :1 Census Bureau defined urbanized area and a total metropolitan area population of at least 100,000 persons), together with adjacent communities that have a high degree of economic and social integration with that nucleus. 75 Specifically, population density categories have been similarly operationalized as a ratio of persons per square mile, and placed on a logarithmic scale. The use of the logarithmic scale achieves a normal distribution, as with race, and allowed the researcher to code and divide population density into quartiles along this normal distribution for the selection of study areas.218 Specifically, population density is operationalized in quartiles as follows: = Least Dense = Little Density = Some Density 4= Most Dense The population density code of 1 was operationalized as "rur ". Codes of 2 and 3 were considered "suburban", and code 4 was designated as "urban". Income Status: Income parameters of low, moderate and high income used herein are all above the U.S. Census Bureau's poverty threshold, and are based upon the researcher’s examination of median annual household income for the State of Michigan.”9 Specifically income categories are defined as follows: > Low Income is defined as zip code areas with mean 1989 family income of less than $21,391. v Moderate Income is defined as zip code areas with mean 1989 family income of between $21,392 and$35,042. 2’8 Recommendation of Fahsbender, op. cit. 2'9 Recommendation of Dr. Elaine Hockman; Gould, op. cit, pp. 21-22; and Perlin, et. al., pp. 72-73, and 76. 76 > High Income is defined as zip code areas with mean 1989 family income of more than $35,043. Specifically, categories of income have been operationalized as a mean family income, and similarly placed on a logarithmic scale. The use of the logarithmic scale again achieves a normal distribution, as with race and population density, and allowed the researcher to code and income into quartiles along this normal distribution for the selection of study areas. Specifically, income is operationalized in quartiles as followsm: Mean Family Income: = < $21,391 per year = $21,392 - $26,999 per year = $27,000 - $35,042 per year = > $35,043 per year Code 1 has been designated in this study to mean "low income". Income codes 2 and 3 have been defined in this study as "moderate income", and 4 has been designated as "high income". Control of Other Factors in Study Area Selection: This study sought to control for geographic size of study areas and potential bias resulting from the distance of Part 201 sites of environmental contamination from the nearest MDEQ, Environmental Response Division (ERD) district office and the number of ERD staff in 1990 within each district office. In all cases, all Part 201 sites within randomly selected categorized zip code study areas were evaluated to derive and record program performance measures. In the selection of study areas, no zip code area with 22° Ibid 77 less than two Part 201 sites were included in the population for the random selection of sample sites. Of course, several Michigan zip code areas possessed no Part 201 sites, and were therefore discarded. Similarly, Part 201 sites lacking detail concerning compliance status reported on the MDEQ Part 201 list were not included in the study. Data Collection - Dependent Variables Currently, the MDEQ, ERD staff within district offices throughout Michigan are required to track and report compliance activities at sites of environmental contamination on a computerized data base known as the Incident Tracking System and through the Part 201 annual site evaluation and scoring process. Unfortunately, these data bases are not well maintained and/or easily accessed within respective MDEQ District Offices, and are often either not collected or utilized by central management and planning units in Lansing ERD Headquarters, nor designed to possess data fields of relevance to this research. No such computerized data base is known to the researcher to exist for Part 201 enforcement data. Consequently, the researcher created a site data check-list to collect data relative to all Part 201 sites within the ten chosen study areas. Given the nature of data collection and management within the MDEQ, the researcher traveled to MDEQ, ERD District offices in Morrice, Jackson, Plainwell, Livonia, Bay City, Cadillac, Gaylord and Grand Rapids, respectively, to collect these data. Refer to Appendix A for an example of the dependent variable data collection check-list used. Data Coding and Preparation Data collected from MDEQ Part 201 site file exarninaticn at various MDEQ District offices was recorded using site check-lists, encoded and compiled through entry 78 into an SPSS" data base of the researcher’s design. This data set whs analyzed, and findings are reported in the form of tables, histograms and line graphs. Refer to Appendix B for a list and description of encoded variables. Data Analysis Following the entry of values for dependent variables for each Part 201 study site, means were calculated for Part 201 program measures. Specifically, MDEQ compliance effort measures calculated included: the average number of days for initial MDEQ response, the average number of site inspections per year, the average number of months needed to control environmental hazards, the average number of compliance oversight meetings per year, the average number of months to attain site cleanup, and a dummy variable mean indicating whether or not final site cleanup had been achieved. MDEQ enforcement measures calculated included the average number of information requests, the logarithmic mean of enforcement notification letters sent by the MDEQ, the average dollar amount of negotiated penalties, and the average number of MDEQ enforcement referrals. The means for each MDEQ compliance, enforcement and public funding measure were then analyzed in accordance within categories of race, income and population density. To assess the statistical significance of these findings, means comparison was undertaken through independent samples for equality of means using SPSS’. Specifically, t-Test analysis was undertaken to determine if observed variability is the result of "usual". or expected variability within the study sample or indicative of "real" or 79 statistically significant difference in program implementation within minority, poor or urban areas. A two-tailed significance finding of .05, or 5 percent, or less was considered statistically significant. In other words, significance findings of less than .05 place observed means at the extreme margins or tails of a normal distribution, and allowed for the rejection of the null hypothesis that demographically varied areas are not treated differently in the MDEQ implementation of the Part 201 program. Secondly, univariate linear regression analysis was undertaken of all statistically significant and normally distributed Part 201 program measures. Finally, multivariate linear regression was undertaken for Part 201 measures statistically significant in more than one demographic category. The R square value from regression analyses explains how well independent study variables of race, income and population density "explain" the observed means difference of the dependent variables of Part 201 program implementation. The Multiple R value generated through regression analyses represents the correlation coefficient between the means difference observed in minority, poor and urban areas (i. e. dependent variables) and the values predicted by the regression model”‘. For interpretation of findings, a Multiple R value of 1 suggests perfect prediction of dependent variables from independent variables, and a zero value indicates that independent variables are not linearly related to the dependent variables. An examination of overall regression analysis, or F test, is utilized in this study to determine the statistical significance of multivariate regression analysis. An observed F value of less than .05 22' Norusis, Marijsa J., SPSS® 6.] Guide to Data Analysis, Prentice Hall, New Jersey, 1995, pp. 422-423, and Andrew Hurley, op. cit, p. 6. 80 allows for the rejection of the null hypothesis that there is no linear relationship between Part 201 program measures (i. e. dependent variables) and demographic categories (i. e. independent variables). In other words, the F value represents the ratio of the regression mean square to the residual mean square, and statistically significant F values suggest that directional (positive or negative) linear correlation exists between dependent and independent variables.222 Finally, the study design proposes that statistically significant findings may validly be generalized to the population of all Michigan Part 201 sites. Study Assumptions, Scope and Limitations Due to the existence of over 11,000 sites of environmental contamination as identified by the Part 201 program, a subset or sample of the total population of sites was selected to undertake this study. Ten study areas were selected randomly by zip code from within ten categories regarding race, income and population density, as described above. This resulted in the selection of 715 zip code areas, from the approximately 1,169 zip code areas in Michigan at the time of this study, each containing at least one part 201 site from which the study sample of ten zip codes was randomly drawn?” All sites of environmental contamination identified and listed by the MDEQ in 1994 were analyzed for select measures of compliance, enforcement and public funding. The researcher acknowledges that specific sites of environmental contamination identified by the Part 201 program exist at various stages of investigation and/or cleanup, ”2 Ibid., p. 478. 223 Approximately 36,000 zip code areas exist within the U.S. in approximately 3,000 counties. 81 and may or may not have undergone escalated enforcement actions or required public funding. Thereby, it is acknowledged that specific sites may show varying amounts of data. In short, data were collected from all sites for all MDEQ program performance measures. Study Assumptions Assumptions regarding the quality of Part 201 site data collected include the absence of variation in: 1) Part 201 program implementation diligence or documentation within and between MDEQ District offices resulting fi'om varying staffing levels and number of sites, i.e. workload variations; 2) professional initiative of staff or MDEQ District management; 3) past and future changes in relevant MDEQ staff and/or management; 4) in Part 201 program management philosophy and policy over time and within select MDEQ District, Regional, Headquarter Offices, and the Governor's Office; and 5) administrative policy changes, legislative amendment and agency reorganization. It is also assumed that political factors have not resulted in varying amounts of compliance, enforcement or the funding of sites of environmental contamination within the ten select study areas. Specifically, this study assumes no significant overall program changes resulting in deviation fiom the MDEQ's constitutional charge to protect and conserve the air, water and other natural resources as a paramount concern of the State, and to protect the public interest in the environment and the protection of public health, safety and general welfare“. 22" Michigan Constitution, Article IV, Section 52, 1963. 82 It is also assumed that Part 201 site security and safety within select study areas have not resulted in variations in compliance, enforcement and/or funding, if found. This assumption relies upon the researcher's experience and knowledge that the assistance of trained and armed MDEQ Conservation Officers, MDEQ Environmental Conservation Offices, and Michigan State Police are available to assist in MDEQ, ERD staff in site access for initial site investigations, inspections, sample collection, compliance monitoring, etc. , if deemed necessary. It is recognized, however, that such precautions demand MDEQ, ERD staff to undertake additional logistical and organizational efforts in contamination site compliance and enforcement oversight. Further, it is assumed that environmental and public health threats exist from sites of environmental contamination, as defined by Part 201, until such locations are remediated to within state standards as provided by administrative rules pursuant to Part 201. Further, this study excludes federal Superfund sites within Michigan from analysis, whether addressed by a state or federal-lead agency. Study Scope Importantly, this study excludes leaking underground storage tank (LUST) sites from analysis, although compliance, enforcement and funding were overseen by the ERD, MDEQ prior to the reorganization and programmatic split in 1994. LUST sites have been excluded here, in acknowledgment of the expenditure of public funds by private parties and others to address these sites from 1989 to 1996 pursuant to the Michigan Underground Storage Tank Financial Assurance Act, PA. 518 of 1988, as amended. These expenditures, totaling over $500 million since fund inception, resulted in increased 83 private sector involvement in site assessment and cleanup, and, if included, would skew this study's data and veil attempts to assess MDEQ diligence in Part 201 program implementation. Similarly, this study excludes sites of environmental contamination owned or operated by the State of Michigan's various departments. A bond fimd of the amount of approximately $30 million was passed in 1989 to investigate and cleanup state-owned facilities with contamination. The inclusion of state-owned or operate Part 201 site would skew the results this study due to the relatively rapid infirsion of large amounts of public funds in a short time periods for this specific subset of the universe of Part 201 sites. Lastly, this study has included site data for Part 201 sites following the implementation of major amendments to Part 201 passed by the Michigan Legislature in June of 1995. These amendments eased cleanup standards, and changed site cleanup approaches from site remediation to site risk assessment and risk management in accordance with varying land-uses and land-use plans. These amendments greatly altered the definition of a site of environmental contamination pursuant to the Part 201 program, and greatly altered cleanup approaches as to result in an overall decrease in site oversight and compliance and enforcement activity by the MDEQ and shifted increasing responsibility for determinations of risk and the completion of risk management to the private sector and to consultants. Limitations of U.S. Census Data Many have recognized the "problem of definition" of race, ethnicity, gender and 84 economic status within U.S. Census Bureau classifications and data collection methodology, and the geographic location of subpopulations based upon these measuresm. The operationalization of such measures in this study is undertaken with knowledge of the limitations of the U.S. Census Bureau 1990 population data. Specifically, these concepts have been questioned as valid units of measure as demonstrated by historically inconsistent racial/ethnic classification (variation in comparison among subpopulations at different points of time and locations) and misclassification (implying a valid standard of racial/ethnic determination) by institutions and researchers, reflecting dynamic social power relationships in definition”. Methods of racial/ethnic classification resulting in variability include self-identification, blood quantum calculation, third party classification based on visible physical features, social mores regarding the classification of off-spring fi'om intermarriagem, and dominant cultural ignorance. For example, U.S. Census racial/ethnic category "White" and "Black" have been questioned as being socially constructed and unidirectionalm. In other words, "whiteness" becomes a social code for membership to dominant culture, and any amount of "blackness" stigrnatizes the carrier as "non-dominantm9. "Asian" is reported to ”5 Zimmerman, op. cit, p. 634, and Pulido, op. cit, pp. 143-148. 226 Zimmerman, op. cit, pp. 636-643, and Pulido, op. cit, p. 142. 227 Ibid, pp. 639-640. 228 See F rankenburg, Ruth, The Social Construction of Whiteness, White Women, Race Matters, University of Minnesota Press, 1993. 229 See Lcrber, Judith and Susan A. Farrell, The Social Construction of Gender, Sage Publications, 1991; Andersen, Margaret L. and Patricia Hill Collins, Race, Class, and Gender: An Anthology, Wadsworth, Inc. Press, 85 contain ten ma' or subcate cries, and "Other Asian" ma contain nineteen cate oriesm. J g y 8 "Pacific Islander" may contain three major categories and "Other Pacific Islander" may subsume seventeen categoriesm. "Hispanic" or " Spanish" may reflect colonial history more than human variability, and can be divided into approximately thirty-six subcategories based on country of origin. Of course concepts of ethnicity seldom know such political boundaries, and may or may not reflect geo-physical boundaries and historical and current migration pathways. U.S. Census categories are often based on country of origin. Similarly, the "American Indian" racial/ethnic category contains nearly six hundred distinct tribesm. As stated throughout sociological literature, the salient questions may be "Who decides?" and "Why?". Pulido (1996) questions three specific assumptions regarding the operationalization of race/ethnicity common to environmental equity empirical studies: 1) that racism can be isolated from other intersecting and overlapping forms of social difference; 2) the focus on racism as distinct and measurable acts of discrimination, rather than its conceptualization as an ideology; and 3) the treatment of racism as monolithic rather than understanding its fragmented and multifaceted nature.233 Some challenge the 1992; and Rose M. Brewer, "Theorizing Race, Class and Gender: The New Scholarship of Black Feminist Intellectuals and Black Women's Labor", in T heorizing Black Feminists: The Visionary Pragmatism of Black Women, Stanlie M. James and Abena P.A. Busia, Editors, Routledge Press, 1993. 230 Zimmerman, op. cit, pp. 638. 23‘ Ibid 23’ [bid 23’ Pulido, op. cit, pp. 152-155, and Goldman, op. cit, 1996, pp 126 and 137. 86 conception of class as income, property values, and/or educational achievement as Weberian?” Further, it is difficult to separate and operationalize socio-economic variables. For example, U.S. EPA notes that 86.1 percent of African-Americans and 91.2 percent of Latinos in the U.S. live in urban areas?” Fmther, although data does not exist for Part 201 sites in Michigan, many federal Superfirnd sites are located in rural areas?“ As reported by Lavelle and Coyle (1993), 18.4 percent of Superfund sites exist in urban areas, 39.6 percent are in suburban and 42 percent are in rural areas.237 It is recognized that such trends likely exist with Michigan Part 201 sites. This may result in the inflation of Part 201 site score in rural areas with potential human exposure to contaminated groundwater drinking sources. Thereby resulting in disparity in agency diligence and public and private spending to address rural contamination sites. Methodological Limitations from Assumptions Used in Environmental Equity Studies A second area of methodological limitation include the assumptions made when setting geographic boundaries for units of equity analysis, distance from sites of environmental risk, selecting a level of aggregation for demographics information, and 23‘ Pulido, op. cit, p. 146. 235 Swanson, op. cit, p. 587. 236 Swanson, op. cit, and Breslin, op. cit, p. 484. 237 Lavelle and Coyle, op. cit, p. S6. 87 choosing a standard of comparison”. Geographic boundaries for analysis in previous equity studies have utilized the entire U.S., geographic regions of the nation, states, counties, municipalities (and other local political jurisdictions), collections of urban areas, agency regions and service areas, zip codes, census tracts (county subdivision usually with 2,400 and 8,000 persons designed to be homogeneous with respect to population characteristics, economic status and living conditions), census block groups (clusters of blocks generally containing between 250 and 550 homes, with an ideal size of 400 housing units), and census blocks (bounded on all sides by visible features such as streets, roads, streams, and railroad tracks, and invisible boundaries including political jurisdictions, property lines and short, imaginary extensions of streets and roads)239. Little or no agreement exist within the literature as to which geographic unit of analysis to use.240 Some researchers, however, have proved the importance of using units at least as large as census tracts so as not to 23‘ [bid, pp. 645-659; Perlin, et. al., op. cit, p. 70; Cutter, op. cit, p. 4; Goldman, op. cit, 1996, pp. 133-135; Greenberg, op. cit, 1993, p. 235, and Glickman, et al., op. cit, pp. 95-114. 23” U.S. Bureau of the Census, 19990 Census of Population and Housing, Area Classifications, Appendices A, A-3 through A-6 (1991), as cited by Zimmerman, op. cit.,1994, p.652; and Perlin, et. al., op. cit, p. 70. 24° Cutter, op. cit, 1995, pp. 111 and 114, and Cutter, op. cit.,1994, p. 4. Cutter states that disagreement among the literature is largely focused on: 1) the environmental threat chosen; 2) the geographic unit of analysis used; 3) the subpopulaticn selected for study; and 4) the time frame analyzed (pp. 1 11-1 14). Perlin, et. al. op. cit, and Zimmerman, op. cit. , add that substantial disagreement also exists within the literature regarding risk and exposure assumptions implied by distance selected from sites of interest, and the choice of statistical methods used to analyze data (see generally). Been, 1995, op. cit, Glickman, et. al., op. cit, pp. 111-112, and Zimmerman, op. cit, recommend the use of multiple statistical methods to analyze numerous time flames, geographic units, and levels of aggregation. 88 hide measures of social difference within study areasz‘“, and others tend to promote the use of zip code areasm. Jurisdictions defined politically are often too large to "capture a facility's immediate neighborhood", but can serve as a useful indicator of "communities... encompassing a shared sense of place, identity and a set of organizations that meet the area's needs" which may or may not be geographically coterminousz‘”. Federal district appellate courts considering issues of environmental justice have generally accepted or implicitly accepted the use of census tracts as the most appropriate geographic unit to analyze social difference in environmental equity studies“. As informed by these findings and in working with the existing Part 201 data base at Wayne State University's Computer Research Lab, this study utilizes zip code designations for the selection of study areas. As stated previously, agreement does not exist within the literature concerning the selection of an appropriate geographic distance in environmental equity studies to use as a surrogate for known or potential migration pathways of environmental risk from 24’ Zimmennan, op. cit, p. 646; Anderton, et. al., op. cit, p. 232; Zax, op. cit, Been, 1994, op. cit, p. 1403, and Hurley, op. cit, p. 6. 242 Commission for Racial Justice, United Church of Christ, op. cit; Mohai and Bryant, op. cit, 1992, pp. 165- 169; Nabalamba, Alice, The Controversy Between Environmental Disposal Systems and Residents of the City of Romulus, MI, Over the Siting of a Deep Injection Well on Wahrman Road, unpublished pater, University of Michigan, Winter 1996, p. 5; and Bryant and Hockman, op. cit, 1994. 2‘3 Zimmerman, op. cit. 2“ See Bean v.s. Southwestern Waste Management Corp., 482 F. Supp. 773, East Tibbs Neighborhood Association v.s. Macon-Bibb County Planning & Zoning Commission, 706 F. Supp. 880 (MD. Ga.) affd 896 F. 2d 1264 (1] Cir. 1989), and Zimmerman, op. cit, pp. 659-665. 89 varying sources.245 Researchers do agree, however, that determinations of risk should be the driving factor of such empiricismz“. As stated by Zimmerman, "without additional information about the physical extent of impacted interests, it is impossible to determine appropriate distances for analyzing equity. From an analytical perspective, the differences between values for a given socioeconomic characteristic become less F significant as the distance becomes greater within a few miles of the site"z‘". Considerations in levels of aggregation include determining a relevant distance of risk (e. g. a chosen number of miles from sites of environmental contamination) and including the census data from partially captured geographic units, or the use of polygonal boundaries from other geographic units. Assumptions made in either approach include the statement bisected units are homogenous with the study areas, or that excluded areas are demographically heterogenous with the study area(s). This "dampening effect" associated with the use of large geographic units of analysis tends to minimize or eliminate demographic mean differences between areas of smaller units of 245 Perlin, et al., op. cit, p. 70, Anderton, et. al., op. cit, p. 140, and Glickman, et. al., op. cit, pp. 111- 112. All environmental equity studies review do this to some extent. The ASTDR’s The Nature and Extent of Lead Poisoning in Children in the United States: A Report to Congress, ASTDR, Centers for Disease Control, Atlanta, GA, 1988 is the only study encountered during this research that examined actual human exposure data. 2‘6 Zimmerman, op. cit, p. 656; Anderton, et. al., op. cit, p. 236; Colquette and Robertson, op. cit, p. 183; Collin, Robert W., “Review of the Legal Literature on Environmental Racism, Environmental Equity, and Environmental Justice”, Journal of Environmental Law and Litigation, Vol. 9, 1994, pp. 157 and 159; Perlin, et. al., op. cit, pp. 69-70; and March 1997 testimony given by the researcher in Genessee Circuit Court in re NAACP v.s. Engler concerning the construction and operation of a demolition wood incinerator and electric generation facility along the northeastern limit of the City of Flint, Michigan. 2"7 Zimmerman, op. cit, p. 656, and Perlin, et al., op. cit. 90 analysis, such as census block and block groupsm. As stated by Zimmerman the bias that can enter into equity studies using large geographic units of analysis "argues for working with smaller geographic units”2"9. Researchers warn that "this assumption may work well with total population figures, it is not likely to work well with subpopulations, which tend to cluster geographically, and are not typically distributed homogeneously", and that difference between an area circumscribed and surrounding areas tend to increase as larger geographic units of analysis are used”25°. Conversely, others find that "studies which have been national in scope and which have provided both income and race/ethnicity information have found race to be more importantly related to the distribution of environmental hazards that incomem‘. Finally, the selection of a standard of comparison is required in equity studies by which population distribution is evaluated relative to source(s) of environmental risk. This area "continues to be one of the more subjective, discretionary areas of environmental equity research" and can lead to bias”. Previous comparison of demographic values with select geographic unit of analysis have been with state or 2“ Zimmerman, op. cit, p.655. 2‘9 Ibid., p. 656; Perlin, et. al., p. 70; Anderton, et. al., op. cit, p. 140; and Glickman, et. al., op. cit, p. 111-112. 25° Zimmerman, op. cit, pp. 653 and 655, and Glickman, et. al., op. cit. 25 ’ Mohai and Bryant, op. cit, University of Colorado Law Review, 1992, p. 927. 252 Zimmerman, op. cit, p. 659, and Been, 1994, op. cit, p. 1384 (the proportionality argument and empirical method ignores population densities of neighborhoods and fails to provide infcnnaticn as to how far the distribution of the population within LULU host neighborhoods deviate from national or state distributions). 91 national means, and represents a significant methodological consideration. Study has shown that the use of the percentage of Afiican-Americans, Hispanics and population below the census-defined poverty line within communities with federal Superfund sites possessed higher percentages when weighted by population size than national percentages”. In other words, the comparison of study area values to state or national standards can a minimize difference and introduce additional bias. This study, therefore, is conservative in methodology in that it possesses the limitation that stronger association may have been found if findings were compared to State of Michigan or U.S. 1990 Census percentages, rather than comparing means of normally distributed Part 201 measures within demographically defined and randomly selected geographic units of analysis. Further, this study analyzed the entire “life cycle”, to the time of data collection, for each Part 201 site captured herein. Data were collected, therefore, from the administrative record for each 201 site from the date of discovery (as early as the 19305) to the time of data collection (Spring through Winter 1997). 253 Perlin, et. al., op. cit. Chapter 4 RESEARCH RESULTS Selection of Study Sites A computerized list of the total population of Michigan sites of environmental contamination was accessed at the Computer Research Lab at Wayne State University”. A data base "dump" was generated of all sites of environmental contamination listed by zip code and site identification number, environmental compliance status, and sorted by population density, income, and race/ethnicity as described in the Chapter 3. This sorted list contained a total of 2,839 sites of environmental contamination. A working list of potential study sites was created by grouping using demographic characteristics to identify ten study zip code areas. The random selection of a single zip code area within each the ten identified social demographic categories was undertaken using a random number table. The number of listed environmental sites contained within the ten selected zip code areas ranged from 4 to 21 , with a mean of 8.9 sites/zip code area as summarized in Table 2 below. 25" Michigan Sites of Environmental Contamination, Volume 1, November 1994, Fiscal Year 1996 92 93 Table 2 W Income Population Density Low Moderate High Urban 2 1/ 5 --- 7/4 Suburban --- 17/8 --- Rural 6/6 -- 1 1/4 numerator = total number of sites within minority zip code area. Mean = 8.9 sites/zip code area. denominator = total aner of sites within white zip code area. Range = 4 to 21 sites/zip code area. This working sample was refined to form the study sample through the identification and elimination of sites with unknown environmental compliance status as listed by the MDEQ255. The study sample was further reduced by cross-referencing site numbers by site name, county, pollutant, and primary MDEQ oversight agency using the MDEQ 201 list”. Cross-referencing was also undertaken to further sample refinement through the elimination of leaking underground storage tank (LUST) sites. LUST sites were identified and eliminated from the study sample either by a site name listing a gas station or related facility, and/or pollutants listed as chemical indicators of refined petroleum products and additives, i. e. purgeable aromatic hydrocarbons, polynuclear aromatic hydrocarbons, methyl-tertiary butyl ether, and/or lead. Sites with primary compliance and enforcement responsibility assigned to a MDEQ division other than ERD were also eliminated, e. g. Underground Storage Tank Division, Waste Management 2” [bid 25‘ [bid 94 Division, Surface Water Quality Division, etc. Additional sites were eliminated from the sample if the site name referenced state agency ownership and/or operation, e. g. "MDOT garage", "MDNR Field Office", or "University of Michigan", etc. The results of study sample identification and refinement is summarized in Table 3 below: Table 3 MDEQmsmt P4201 , V Officer Sites "I ” " ‘ ' ’ ' Plainwell 10 l l 1 2 24 Cadillac 7 2 0 0 9 SE Michigan 6 1 0 2 9 Grayling 5 3 0 0 8 Grand Rapids 4 3 0 0 7 Saginaw Bay 7 9 0 0 16 Jackson 9 12 0 l ' Total I 48 411 5 Total MEDQ listed Part 201 sites = 2,839, source Michigan Sites of Environmental Contamination, Volume 1, November 1994, FY I 996. Subpopulation grouped by social demographics (population density, race/ethnicity, and income) and randomly selected by zip code. From the above evaluation, the percent of MDEQ listed sites determined not to qualify for inclusion in this study equaled 47 of 95 total sites, or nearly 50%. Consequently, an adjusted total population of Michigan 201 sites, accounting for listed non-Part 201 sites, may equal 1,420 sites. Therefore, the 43 study sample sites represent approximately 3.0% of the total adjusted population of Michigan Part 201 environmental contamination sites. 95 Finally, the zip code referenced sample was cross-referenced with MDEQ District Office geographic assignments, and appointments were requested in writing to examine the selected site files pursuant to Michigan’s Freedom of Information Act (F OIA). Each site file was reviewed, and chronologies of site activities were recorded to derive measures of site compliance, enforcement, and public funding as described in Chapter 3. Tracking forms created to record site file information were then encoded, and these data were entered into a computerized data base created for this study using SPSS" Version 6.1 . The results of study sample refinement and FOIA requests is summarized in Table 4 below: Table 4 WWW andResuIts mes : m sou—owner a Plainwell 7 0 l 2 10 Cadillac 5 0 0 2 7 SE Michigan 6 0 0 0 6 Gaylord 5 0 0 0 5 Grand Rapids 4 0 0 0 4 Saginaw Bay 7 0 0 0 7 Jackson 9 0 0 0 9 Total 43 0 1 4 48 Missing Site Files As shown above, the total site files sought for review MDEQ, ERD District Offices through the Freedom of Information Act was 48. Final sample refinement occurred upon researcher file review and/or the ability of the MDEQ, ERD District Office 96 to locate file information, i. e. a site's administrative record. The four missing site files were all within the MDEQ, ERD Plainwell District office, and consisted of Part 201 sites in Allegan and Kalamazoo Counties. Interestingly, each of the four site files that were reported by the MDEQ as “public record did not exist” were within minority zip code areas. Three existed within a minority, rural and high income area, and one within a minority, urban and low income area. As a result, the total number of Part 201 sites included in this study was 43, or 89.5% of 48 site files requested. Study Sample Table 5 below summarizes the total number of environmental contamination sites included within the ten social demographics cases in this study, including relative percentages of each within select demographic categories. Table 5 Smdxfiamflebflnmmeandficmflafionfiemitx Income Population Density Low Moderate High Total # (% Total) Urban 6 (13.95%) -- 6 (13.95%) 12 (27.9%) Suburban --- 19 (44.2%) -- 19 (44.2%) Rural 9 (20.9%) --- 3 (7.0%) 12 (27.9%) Total # (% Total) 15 (34.9%) 19 (44.2%) 9 (20.9%) 43 (100%) #(%) # = total number of sites/zip code area. Mean = 4.3 sites/zip code area. (%) = percent of total number of sites/zip code area. Range = 3 to 19 sites/zip code area. As shown below, Table 6 incorporates race/ethnicity into the summary of the number of study sites by income and population density. Sample percentages for 97 racial/ethnic categories of white and minority are summarized as follows. Table 6 Income Population Low Moderate High Minority: # Sites (%sites) Density White: # Sites (% sites) Urban 5 (16.7%) -- 4 (13.3%) 9 (30%) 1 (7.7%) 2 (15.4%) 3 (23.1%) Suburban -- 15 (50%) -- 15 (50%) 4 (30.8%) 4 (30.8%) Rural 4 (13.3%) -- 2 (6.7%) 6 (20%) 5 (38.5%) 1 (7.7%) 6 (46.2%) Minority #(%) 9 (30%) 15 (50%) 6 (20%) 30 (100%) White #(%) 6 (46.2%) 4 (30%) 3 (23.1%) 13 (100%) top = total number of sites and percent sites/minority zip code area. Mean = 3 sites/ minority zip code areas. bottom = total number of sites percent sites/white zip code area. Mean = 1.3 sites/white zip code areas. Range = l to 15 sites/zip code area. The distribution of responses for each variable was examined using SPSS®, and plotted to determine if distribution was normal. The variable for MDEQ response time (days) or "cl " was converted to a logarithmic scale to achieve a normal distribution. The other five MDEQ compliance effort variables were normally or approximately normally distributed as listed in Table 7 below. Results of Site Compliance Evaluation To begin to evaluate Part 201 program compliance measures (dependent variables) by demographic (independent) variables, means were calculated using SPSS®. The results of compliance measure means calculations are summarized in Table 7 below. 98 Table 7 QalnuatmnnLMeantEarflflLBmgramLanliance; Means Response Inspections # Months to Meetings # Months *Cleanup Time (days) Per Year Control Per Year to Cleanup Ongoing? Urban 4,624 .425 122 .137 169 .834 Suburban 1,423 .570 155 .195 181 .790 Rural 860 .486 99 .292 156 .834 Low 395 .503 86 .280 154 .800 Income Moderate 1,423 .570 155 .195 182 .790 Income High 6,602 .376 151 .105 176 .890 Income Minority 2,890 .540 149 .164 177 .767 White 684 .426 86 .303 157 .923 "' Binomial variable that defines yes = 0 and no = 1. Therefore, higher decimals indicate higher likelihood that site cleanup within the specified demographic category is not complete. MDEQ Response Time Regarding initial agency response time, it was found that rural areas on average receive 39.5% faster response time than suburban areas, and 81% faster response time than urban areas. Response time as used here is defined as the number of days from MDEQ, ERD discovery or awareness of a potential site of environmental contamination to ERD stafl‘ follow-up investigation. According to study results, white areas received a 76% quicker response time than Part 201 sites in minority areas. Lastly, low income areas received a 73% faster response time than sites in moderate income areas, and a 94% faster response time than sites in high income areas. 99 MDEQ Inspections Regarding MDEQ annual inspections at study sites, it was found that suburban area sites on average received 25% more MDEQ effort to oversee site compliance activities through site inspections than sites in urban areas, and 15% more MDEQ inspection effort than sites in rural areas. Also, rural area sites were subject to 13% more agency inspection effort than sites in urban areas. Annual inspections is used here to denote the total number of recorded MDEQ, ERD staff inspections per year from site discovery until cleanup. According to study results, minority areas received 21% more annual MDEQ compliance inspections than sites in white areas. Moderate income area sites received 12% more annual MDEQ inspection efi‘ort than sites in low income areas, and 34% more than sites in high income areas. Further, low income area sites on average received 25% more MDEQ inspection effort per year than sites in high income areas. Time to Control Site Hazards Table 7 also summarizes the results of means calculations regarding the number of months required to control hazards at sites of environmental contamination included in this study. Findings suggest that Part 201 sites in rural areas were brought under control nearly 19% sooner than sites in urban areas, and 36% sooner than sites in suburban areas. Site hazard control is used here to mean the fencing of a site of environmental contamination, the capping of contaminated soils, replacement of impacted drinking water supply wells, the removal of contamination source(s), and/or the initiation of ground water or soil treatment. Site hazard control results in the interim elimination or control of potential or known human and/or environmental exposure pathways. 100 According to study results, Part 201 sites in white areas were controlled 42% sooner than sites in minority areas. Low income area Part 201 sites were controlled 43% sooner than sites in high income areas, and 44% sooner than sites in moderate income areas. Lastly, environmental and human health hazards at sites in high income areas were controlled 3% sooner than sites in moderate income areas. MDEQ Meetings with Regulated Parties Regarding the frequency of MDEQ meetings with PRPs for cleanup oversight, these results suggest that rural study sites on average received the most diligent agency effort to oversee and foster private party cleanup activities through face to face meetings. Specifically, study sites in rural areas were subject to 33% more MDEQ and PRP site meetings than sites in suburban areas, and 53% more than study sites in urban areas. Also, sites in suburban areas were subject to 30% more MDEQ compliance oversight meetings with PRPs than sites in urban areas. Specifically, oversight meetings is used here as an annual mean of the total number of recorded meetings between the MDEQ and PRPs from MDEQ site discovery until site cleanup. According to study results, sites in white areas were 46% more likely to be subject to MDEQ and PRP meetings than sites in minority areas. Study sites in low income areas were subject to 63% more agency and PRP meetings than sites in high income areas, and 30% more than sites in moderate income areas. Study sites in high income areas received 46% more MDEQ compliance oversight meetings when compared to sites in moderate income areas. 1 01 Site Cleanup Pace Table 7 also summarizes the results of means calculations regarding the pace or average number of months required to cleanup sites of environmental contamination included in this study. Findings suggest that cleanup activities at study sites in rural areas were completed 8% sooner than at sites in urban areas, and 14% sooner than at sites in suburban areas. Cleanups at study sites in suburban areas were found to be completed 7% faster than cleanups at sites in urban areas. Site cleanup is used here as the average number of months required to document the removal or reduction of hazardous substances to with state cleanup standards used at the time of site closure study by the MDEQ or PRPm. According to study results, study sites in white areas were cleaned up 11% sooner than sites in minority areas. Low income area study sites were cleaned up 13% sooner than sites in high income areas, and 15% sooner that sites in moderate income areas. Lastly, site cleanup pace in high income areas was 3% faster than sites in moderate income areas. Site Cleanups Completed Regarding the number of sites at which cleanups had been completed at the time of this study, these results suggest that sites in rural and urban areas on average were 2” From 1982 to 1990, Part 201 cleanup standards required the return of site conditions to natural conditions or the documented removal of site contaminants to within laboratory detection limits. Administrative rules promulgated in 1990 created a tripartite cleanup standard, allowing PRPs the choice of standards including active cleanup to the original zero risk standard, and new standards based upon an acceptable risk level of one additional adult cancer increase in one million persons, or the implementation of risk assessment and management site controls to achieve the one in one million acceptable risk standard. Sweeping amendments in 1995 replaced the acceptable risk level with a one in one-hundred thousand additional adult cancers standard for four specific land uses (residential, recreational, commercial, and industrial), and emphasized the utilization of risk assessment and risk management through engineering and/or administrative site controls to prevent unacceptable human exposure and/or ecological damage. 102 equally likely to have been cleaned up. Sites in suburban areas, however, were 5% less likely to have been finished with cleanup than sites in urban or rural areas at the time of this study. Specifically, cleanups completed is used here to mean the total number of recorded cleanups divided by the total number sites within each social demographic category, i.e. population density, income, and race/ethnicity. According to study results, 17% more study sites in white study areas were cleaned up at the time of this study than sites in minority study areas. Lastly, sites in high income study areas were 11% more likely to be finished with site cleanup than sites moderate income areas, and 10% more likely to be cleaned up to within state standards than sites in low income areas. Sites in low income areas were also found to be 1% more likely to have been completed with cleanup activities that sites in moderate income areas at the time of this study. Statistical Significance of Means Differences within Compliance Measures To assess the statistical significance of the difference in means measured within demographic categories of race/ethnicity, income, and population density, t-Tests of independent-sample means was undertaken for Part 201 compliance measures. Assuming population means are equal within social demographic categories (race/ethnicity, income, and population density), statistical significance from independent t-Tests was used to determine if difference observed within means of dependent variables was the result of "usual" variability within a single population, or indicative of statistically significant difference in sample means. Table 8 below smnmarizes the results of independent sample t-Tests of Part 201 compliance measures by race/ethnicity, income, and population density. 103 Table 8 WW 2-tail Response Inspections #Months Meetings Months to Cleanup S' ificance Time Per Year to Control Per Year Cleanup Ongoing? Urban/Suburban 0.246 0.538 0.513 0.494 0.776 0.773 Urban/Rural 0.513 0.638 0.545 0.691 1.000 Suburban/Rural 0.669 0.721 0.274 0.560 0.773 Low [High 0.174 0.340 0.096 0.515 0.591 Income Moderate/Low 0.845 0.754 0.135 0.484 0.942 Income High/Moderate 0.129 0.471 0.941 0.31 1 0.909 0.539 Income Minority/White 0.442 0.539 0.130 0.062 0.610 0.236 Bold numbers and shaded cells are statistically significant. 95% confidence interval. 2-tail significance of <0.05 is considered statistically significant. 2-tai1 significance results more than 0.05 does not allow the rejection of the null hypothesis that two groups of Part 201 sites come from populations with the same average MDEQ compliance effort. These results suggest that mean differences in MDEQ and PRP meetings at Part 201 sites in low and high income areas is statistically significant. Specifically, these results indicate that low income areas receive significantly more MDEQ compliance effort, as measured by MDEQ effort to initiate or attend face to face meetings with PRPs to foster site investigation and cleanup. Following the recommendations of previous studies, the statistical significance of the difference in means observed was analyzed for binomial variables including whether or not site cleanup had been completed at the time of this study within demographic categories of race/ethnicity, income, and population density. The Kruskal-Wallis nonparametric one-way analysis of variation (AN OVA) among all three demographic 104 categories for appropriate dependent variables was undertaken for this purpose. An advantage of the Kruskal-Wallis test is that it does not assume the normal distribution of dependent variable observations. Similar to the independent t-Test above, the null hypothesis for the Kruskal-Wallis test is that the means of Part 201 program measures and the shape of their distributions are the same for social demographic categories (race/ethnicity, income, and population density). Statistical significance corrected for ties from the from Kruskal-Wallis one-way AN OVA was used to determine if the observed difference suggested "usual" variability within the population of Part 201 site, or indicative of statistically significant difference in sample means from influence of socioeconomic independent variables. Table 9 below summarizes the results of Kruskal- Wallis one-way AN OVA of completed cleanups by race/ethnicity, income, and population density. Table 9 safté-A: ‘OI‘ a: slit“! is?! '-U| . 'HI 2" its inificance Corrected for Ties ll Completed Cleanups Population Density 1.00 Income 0.66 Race/ethnicity 0.23 Bold numbers and shaded cells are statistically significant. 95% confidence interval Significance Corrected for Ties of <0.05 is considered statistically significant. Significance of more than 0.05 does not allow the rejection of the null hypothesis that observations are independent samples from populations with the same non-normal distribution (1'. e. same average MDEQ compliance effort). These results suggest that observed means differences regarding the completion of 105 site cleanup at the time of data collection for this study are not statistically significant. The null hypothesis is not rejected that mean difference observations regarding site cleanup status are independent samples from populations with the same non-normal distribution. Results of Site Enforcement Evaluation To evaluate Part 201 program enforcement measures by demographic variables, means of enforcement measures were calculated using SPSS®. The results of Part 201 enforcement means calculations are summarized in Table 10 below. Table 10 MW Means #Information # Notice # Days to Negotiated Enforcement Requests Letters Notification Penalties Referrals Urban 0.000 1.92 8,983 $0.00 .33 Suburban 0.526 1.21 150 $5,985.71 .53 Rural 0.500 3.08 105 $0.00 .50 Low 0.400 3.00 329 $0.00 .47 Income Moderate 0.0526 1.21 150 $5,985.17 .53 Income High 0.000 1.67 11,135 $0.00 .33 Income Minority 0.233 1.70 1,276 $1,373.33 .43 White 0.000 2.46 6,555 $2,366.67 .54 The distribution of responses for each enforcement variable was examined using SPSS®, and plotted to determine if distribution was normal. The variable for PRP notification time (days) or "NOTDAYS#" was converted to a logarithmic scale to achieve 106 a normal distribution. The other four MDEQ enforcement effort variables were normally or approximately normally distributed. MDEQ Information Requests As supported by the researcher's experience, Table 10 suggests that the MDEQ seldom used the information request enforcement provisions of Part 201 as part of its enforcement effort at study sites. Specifically, the number of information requests is defined here as the total number information requests at a site over the years from discovery until cleanup. These findings further suggest that suburban area sites were 5% more likely to have MDEQ information requests of PRPs regarding the nature and extent of site contamination than rural sites. No MDEQ enforcement information requests were captured by urban study sites. According to study results, minority area sites enjoyed the occasional use of the information request by MDEQ enforcers, i.e. 0.233 requests per year. Study sites within white areas were not subject to any information requests. Lastly, low income area sites were subject to 24% more MDEQ information requests than at sites in moderate income areas. No high income sites captured in this study were subject to MDEQ information requests as a part of enforcement actions, if any. MDEQ Notification of PRPs The aggressiveness of MDEQ identification and notification of PRPs was analyzed, and these results are also summarized in Table 10. Findings suggest that rural areas on average receive the most diligent agency efforts to identify, notify and perhaps re-notify PRPs of violations of Part 201, followed by urban and then suburban areas, respectively. Specifically, MDEQ efforts to notify PRPs at rural sites were 38% greater 'LJ‘DQ -1 107 than at urban area sites, and 61% greater than at suburban sites. According to study results, PRPs at white area sites were 31% more likely to be initially notified of violations of Part 201 than PRPs at minority area sites. PRPs at study sites in low income areas were 60% more likely to be notified than PRPs of sites in moderate income areas, and 44% more likely than PRPs at sites high income areas. Finally, PRPs at sites in high P income areas were 28% more likely to have been notified by the MDEQ of Part 201 violations than PRPs at sites in moderate income areas. Time for MDEQ Identification and Notification of PRPs Table 10 also summarizes the results of means calculations for the number of days required by the MDEQ to locate and notify PRPs of Part 201 responsibility to investigate and remediate sites of environmental contamination. Findings suggest that the MDEQ notified PRPs of sites in rural areas of Part 201 violations 30% sooner than PRPs at sites in suburban areas, and 99% sooner than PRPs at sites in urban areas. According to study results, PRPs of minority area study sites were on average notified 81% sooner than those at sites in white areas. Lastly, PRPs of moderate income area sites were notified 65% sooner than PRPs at sites in low income areas, and 99% sooner than PRPs at sites in high income areas. Negotiated Penalties If utilized by the MDEQ, successful enforcement results in the return of reluctant or recalcitrant PRPs to compliance with the provisions of Part 201. This process often concludes with the negotiation and approval of a written agreement between the MDEQ and PRPs, and respective counsel. These documents, referred to as Consent Decrees or 108 Consent Orders, are contractual agreements filed with a Michigan Circuit Court to insure subsequent adherence to agreement provisions through judicial enforcement. As a matter of MDEQ enforcement policy, the agency frequently requires the payment of penalties to off-set PRP financial benefit from pollution and/or damages to natural resources held within the public trust and administered by the MDEQ. Through this negotiation process, PRPs as such are often required to "stipulate" to penalties for past Part 201 non- compliance. The amount of stipulated penalties is at the discretion of the MDEQ and AG, and may serve as a useful measure of the social equity of Part 201 program enforcement. Regarding negotiated penalties in Part 201 enforcement cases included in this study, these results suggest that negotiated penalties at sites in white areas were on average 42% higher than negotiated penalties at minority area sites. Sites with negotiated penalties did not exist within the sample for all population density or income categories, and therefore these measures were not further analyzed. MDEQ Enforcement Referrals Table 10 also summarizes the means calculations of the number of MDEQ enforcement referrals to the AG for sites within each demographic category. These results suggest that sites in suburban areas were 6% more likely to be referred for enforcement than sites in rural sites, and 38% more likely than sites in urban areas. Enforcement referrals are used here to mean official requests by MDEQ management for AG assistance in undertaking escalated enforcement actions against PRPs at Part 201 sites. According to study results, sites in white areas were 20% more likely to be subject 109 to enforcement referral than sites in minority study areas. Sites in moderate income areas were 21% more likely to be referred for AG enforcement action than sites in low income areas, and 38% more likely than sites in high income areas. Finally, sites in low income areas were 30% more likely to have been referred for AG enforcement than sites in high income areas. I MDEQ Negotiation of Cleanup Completion Although not summarized in Table 10, means calculations were also undertaken to determine the relative frequency of MDEQ negotiation of a requirement for PRP site cleanup in association with or in lieu of penalties. Findings suggest very little variation in whether the MDEQ obtained legally enforceable agreement for PRP site cleanup as a result of enforcement actions in areas of varying population density and income. However, sites in minority areas were found to be 17% more likely to be cleaned up by a PRP as a result of enforcement action in lieu of voluntary compliance, and in association with or instead of penalties and/or MDEQ direct cleanup expenditure. Statistical Significance of Means Differences within Enforcement Measures To assess the statistical significance of the difference in means of MDEQ Part 201 enforcement efforts found within this study, t-Tests of independent-sample means was undertaken for each dependent variable. Assuming that population means are equal within social demographic categories (race/ethnicity, income, and population density), statistical significance from independent t-Tests was used here to determine if mean difference observed is the result of usual variability of sample means from a single population or indicative of statistically significant difference. Table 11 below 110 summarizes the results of independent sample t-Test of Part 201 enforcement measures by race/ethnicity, income, and population density. Table 1 l - ' - 2-tai1 #Information #Notice #Days to Negotiated #Enforcement Significance Requests Letters Notification (log) Penalties Referrals Urban/Suburban 0.436 0.281 0.259 0.133 0.565 Urban/Rural 0.328 0.256 0.143 0.333 0.496 Suburban/Rural 0.271 0.667 .. 0.935 Low [High 0.451 0.378 0.599 Income Moderate/Low 0.340 0.944 0.845 Income _. ' High/Moderate 0.502 0.496 0.385 0.200 I 0.598 Income Minority/White 0.454 0.307 0.916 0.657 I 0.696 Bold numbers and shaded cells are statistically significant. 95% confidence interval. 2-tail significance of <05 is considered statistically significant. 2-tail significance results more than 0.05 does not allow the rejection of the null hypothesis that two groups of Part 201 sites come from populations with the same average MDEQ enforcement effort. These results suggest that mean difference in the number of MDEQ notification letters and negotiated penalties comparing suburban and rural, and low and moderate income areas are statistically significant. Specifically, the results indicate that low income areas receive significantly more MDEQ enforcement effort than moderate income areas as measured by MDEQ notification of PRPs, but statistically significantly less negotiated penalties than enforcement cases in moderate income areas. Similarly, the results indicate that rural areas receive significantly more MDEQ enforcement effort than suburban areas as measured by MDEQ notification of PRPs, but statistically significantly 111 less negotiated penalties than enforcement cases in suburban areas. Following the recommendations of previous studies, the statistical significance of the difference in means observed was additionally analyzed for binomial Part 201 program enforcement variables including: whether or not MDEQ information requests were made; if PRPs were identified and notified of Part 201 violations and obligations; if formal referrals were made by the MDEQ to the AG for escalated enforcement actions; if penalties were levied; and if site cleanup was required through negotiated settlement of enforcement proceedings at the time of this study. The Kruskal-Wallis one-way AN OVA was undertaken for nonparametric Part 201 program enforcement variables for all three demographic categories. As stated above, the Kruskal-Wallis test does not assume the normal distribution of dependent variable observations. The null hypothesis for this test is that the means of Part 201 program enforcement measures and the shape of their distributions are the same for each social demographic category. Statistical significance corrected for ties from the from Kruskal-Wallis one-way AN OVA was used to determine if the observed variability suggested usual variability within the population of Part 201 sites, or indicative of statistically significant difference in sample means from influence of socioeconomic independent variables. Table 12 below summarizes the results of Kruskal-Wallis one-way AN OVA of select Part 201 enforcement measures by race/ethnicity, income, and population density. l 12 Table 12 WW Significance as Information PRP Enforcement Penalties Cleanup Corrected for Requests Notification Referral Levied Negotiated Ties Population 0.34 0.67 0.20 0.48 0.85 Density Income 0.49 0.60 0.68 0.19 0.26 Race/ethnicity 0.34 0.87 0.09 0.87 if»if;{Lg-31;;2.0.0233};fjfgi Bold numbers and shaded cells are statistically significant. 95% confidence interval. Significance of <0.05 is considered statistically significant. Corrected for ties significance of more than 0.05 does not allow the rejection of the null hypothesis that observations must be independent samples fiom populations with the same non-normal distribution (1‘. e. same average MDEQ enforcement effort). These results suggest that means differences for court enforceable negotiated agreements for site cleanup from MDEQ enforcement was statistically significant by race/ethnicity. This allows for the rejection of the null hypothesis that the mean difference observed for negotiated site cleanup was from independent samples from the same non-normal distributed population. Results indicate that minority areas receive significantly more MDEQ enforcement effort, as measured by the negotiation of site cleanup by the PRP as the result of agency enforcement actions, in lieu of PRP cleanup from MDEQ compliance effort, or the MDEQ expenditure of public funds. Results of Public Funding Evaluation To begin to evaluate Part 201 program public funding measures by select demographic variables, means were also calculated using SPSS®. The results of public funding means calculations are summarized in Table 13 below. 113 Table 13 WW0: Means Funds Funds Spent Emergency Funds Percent Requested Funds Spent Recovered Recovered Urban $279,166.67 $215,083.33 $29,041.67 $0.00 0.00 Suburban $34,447.39 $12,536.84 $0.00 $0.00 0.00 Rural $140,208.33 $85,417.13 $2,083.33 $3,500.00 4% Low Income $332,166.67 $237,067.03 $24,900.00 $2,800.00 1% Moderate $34,447.37 $12,536.84 $0.00 $0.00 0.00 Income High Income $5,555.56 $5,555.56 $0.00 $0.00 0.00 Minority $159,000.00 $107,103.48 $12,017.24 $0.00 0.00 White $70,538.46 $48,546.23 $1,923.08 $1,923.08 0.7% The distribution of responses for each public funding variable was examined using SPSS®, and plotted to determine if distribution was normal. The above MDEQ public funding effort variables were normally or approximately normally distributed. Public Funds Requested Table 13 suggests that the MDEQ requests for public firnding was nearly 88% higher for sites in urban areas than sites in suburban sites, and 50% higher than sites in rural areas. According to study results, the MDEQ requested 55% more public funds to address sites in minority areas as compared to white areas. Lastly, low income area sites were subject to the highest level of MDEQ site funding. Sites in low income areas were subject to MDEQ requests for nearly 89% more public funds than sites in moderate income areas, and 98% more than sites in high income areas. 114 Public Funds Spent Actual expenditures of public funds by the MDEQ was also analyzed, and these results are summarized in Table 13. Similar to the funds requested, it was found that urban area study sites on average received 94% more site spending than study sites in suburban areas, and 96% more than study sites in rural areas. Results also indicated that the MDEQ spent 55% more public funds to address sites in minority areas than sites in white study areas. Lastly, relative percentages for expenditures were also similar to ftmds requested when compared to all income categories. Specifically, low income area study sites received nearly 95% more firnds than sites in moderate income areas, and nearly 98% more than sites in moderate income areas. MDEQ Emergency Fund Expenditures Table 13 also summarizes the results of means calculations for public funds spent by the MDEQ to undertake to emergency response actions including: drum and/or surface removal of hazardous substances; the prevention of human exposure to toxic materials; or the prevention of hazardous material migration into sensitive habitats and/or public water supplies. It was found that 93% more emergency funds were spent by the MDEQ at Part 201 sites in urban areas as compared to sites in rural areas, and no emergency funds were spent at study sites in suburban areas. 84% more emergency funds were spent by the MDEQ at sites in minority areas as compared to sites in white study areas. This study did not capture sites with MDEQ emergency firnd expenditure in moderate or high income areas. 115 Public Funds Recovered If public funds are spent by the MDEQ to address conditions at a sites of environmental contamination, as a matter of policy the MDEQ is to undertake efforts to collect dollars from PRPs equal to the public expenditure. This process is generally referred to as agency "cost recovery". Experience and knowledge of the researcher suggests that cost recovery by the MDEQ has been somewhat discretionary, and has not been diligently undertaken by the MDEQ at numerous sites statewide. Study results indicate that cost recovery actions were undertaken in association with public fund expenditures within rural, low income, and white study areas only. This was found despite the fact that public funds were spent in each demographic category. Statistical Significance of Means Differences within Public Funding Measures To assess the statistical significance of the difference in means found in public funding measures, t-Tests of independent-sample means was undertaken for each measure of MDEQ Part 201 public funding effort. Assuming population means are equal within social demographic categories (race/ethnicity, income, and population density), statistical significance from independent t-Tests was used here to determine if variability observed was statistically significant. Table 14 below summarizes the results of independent sample t-Test of Part 201 public funding measures by race/ethnicity, income, and population density. Table 14 MW 2-tail Significance n Funds Requested Funds Spent Emergency Funds .. Funds Spent Recovered Urban/Suburban 0.199 0.188 0.227 0.432 Urban/Rural 0.575 0.516 0.365 0.350 ‘ Suburban/Rural 0.124 0.067 0.227 0.241 Low [High Income 0.195 0.256 0.418 0.478 Moderate/Low 0.087 0. 106 0.247 0.295 Income High/Moderate 0.465 0.513 0.000 0.504 Income fifi Minority/White ll 0.561 0.629 0.581 0.128 Bold numbers and shaded cells are statistically significant. 95% confidence interval. 2-tail significance of <0.05 is considered statistically significant. 2-tail significance results more than 0.05 does not allow the rejection of the null hypothesis that two groups of Part 201 sites come from populations with the same average MDEQ compliance effort. These results indicate that none of the mean differences observed for MDEQ public funding measures were statistically significant. Specifically, the small sample size of Part 201 sites with public fund expenditure and relatively little variance in observed difference, may have resulted in the failure to reject the null hypothesis that observed Part 201 public funding mean values come from a population of all Part 201 sites with the same means. 117 -._|l .“H... r-.'..-. 2.2. .1, A... .1“ ..1‘ I H. To reflect the varying level of risk or severity of Part 201 sites, dependent variable observations were weighted”. Independent variables considered for use in weighting included: the staff size of various MDEQ, ERD district offices as compared to the number of sites, i.e. workload; the lineal distance in miles from sites to MDEQ district offices; and Part 201 site score. Little variation was found in the application of workload and distance to the sample, therefore a site score ratio was calculated and multiplied by observations to account for the relative risk posed by each Part 201 site. Specifically, the above means calculations and statistical analyses were replicated following the weighting of dependent variable measures by the relative risk to public health and the environment presented by each site. Dependent variable measures were weighted by multiplying by a ratio of Part 201 site score for each site. The site score ratio was found by dividing the site Part 201 score, ranging from 1 to 48, as reported by the MDEQ for fiscal year 1996 by the maximum potential site score of 48. As such, the calculated ratio existed as the relative risk posed by each site as determined by the MDEQ. As undertaken in Part 1 above, the distribution of responses for each weighted Part 201 compliance variable was examined using SPSS", and plotted to determine if distribution was normal. The variable for MDEQ response time (days) or "cl " was again converted to a logarithmic scale to achieve a normal distribution. The other five MDEQ 258 Recommendation of Greenberg, op. cit., (1993), pp. 248-249; and Gould, op. cit, p. 9. 118 compliance effort variables were normally or approximately normally distributed. Following this weighting process, means calculations were repeated for each dependent variable representing MDEQ Part 201 compliance, enforcement, and funding effort. Results of Weighted Site Compliance Evaluation As described above, weighted Part 201 program compliance measures by demographic variables, means were calculated using SPSS®. The results of weighted compliance measure means calculations are summarized in Table 15 below. Table 15 Weighted Response Inspections # Months Meetings # Months *Cleanup Means Time (days) Per Year toControl Per Year toCIeanup Done ? Urban 1,991 .274 80 .093 l 1 1 .486 Suburban 1,010 .352 97 .126 1 15 .471 Rural 277 .261 51 .157 83 .468 Low Income 270 .309 53 .172 94 .486 Moderate 1,010 .3 52 97 .126 1 15 .471 Income High Income 2,572 .199 85 .049 102 .461 Minority 1,494 .33 1 92 .107 1 10 .447 White 190 .245 50 .170 93 .538 "‘ Binomial variable defining yes = 0 and no = 1. Therefore higher decimals indicate higher likelihood that site cleanup within the specified demographic category is ongoing. Weighted MDEQ Response Time Regarding agency response time, it was found that rural areas on average receive 73% faster response time than similar sites in suburban areas, and 86% faster response 119 time than similar sites in urban areas. Response time as used here is defined as the weighted average of the number of days from MDEQ, ERD documented discovery or awareness of a potential site of environmental contamination and ERD staff follow-up investigation. According to study results, white area sites received a 87% quicker response time than similar sites in minority areas. Lastly, sites in low income areas received a 73% faster response time than similar sites in moderate income areas, and a 90% faster response time than similar sites in high income areas. Weighted MDEQ Inspections Regarding agency annual inspections at study area sites, it was found that suburban area sites on weighted average received 77% more agency inspection effort to oversee site compliance activities than similar sites in urban areas, and 74% more inspection effort than similar sites in rural areas. Urban areas received 4% more MDEQ inspection effort than similar sites in rural areas. Weighted annual inspections is used here to denote the total number of recorded MDEQ inspections from site discovery until cleanup as weighted by site score. According to weighted means results, minority area sites received 24% more MDEQ annual inspections than similar sites in white areas. Sites in moderate income areas received 43% more annual MDEQ inspection effort than similar sites in high income areas, and 11% more than similar sites in low income areas. Lastly, low income area sites received 35% more annual MDEQ inspection effort than similar sites in high income areas. Weighted Time to Control Site Hazards Table 15 also summarizes the results of means calculations regarding the number 120 of months to control hazards at sites of environmental contamination as weighted by site score. These findings suggest that Part 201 sites in rural areas were brought under control 36% sooner than similar sites in urban areas, and 47% sooner than similar sites in suburban areas. Sites in urban areas were brought under control 18% sooner than similar sites in suburban areas. Site hazard control is used here to mean the fencing of a site of environmental contamination, the capping of contaminated soils, replacement of impacted drinking water supply wells, the removal of contamination source(s), and/or the initiation of ground water or soil treatment system operation as weighted by site score. Further, Part 201 sites in white areas weighted by site score were controlled 46% sooner than similar sites in minority areas. Low income area sites were controlled 38% sooner than similar sites in high income areas, and 45% sooner than similar sites in moderate income areas. Finally, sites in high income areas were controlled 12% sooner than similar sites in moderate income areas. MDEQ Meetings with Regulated Parties Regarding MDEQ meetings with PRPs for cleanup oversight as weighted by site score, these results suggest that rural sites on average received the most diligent agency effort to oversee private party cleanup activities through personal meetings. Specifically, sites in rural areas were 19% more likely to be subject to MDEQ and PRP site meetings than similar sites in suburban areas, and 44% more likely than similar sites in urban areas. MDEQ meetings with PRPs in suburban areas were 30% more prevalent than meetings regarding site progress in urban areas when weighted by site score. Specifically, oversight meetings is used here as the total number of recorded meetings between the 121 MDEQ and PRPs from MDEQ site discovery until site cleanup, as weighted by site score. According to study results, sites in white areas were 35% more likely to be subject to MDEQ and PRP meetings than similar sites in minority areas. Sites in high income areas were 62% less likely to be subject to agency and PRP meetings than similar sites in moderate income areas, and 71% less likely to be subject to agency compliance meetings when compared to similar sites in low income areas. Sites in moderate income areas were 24% less likely to be subject to MDEQ and PRP meetings than similar sites in low income areas. Weighted Site Cleanup Pace Table 15 also summarizes the results of weighted means calculations regarding the number of months required to cleanup sites of environmental contamination. Findings suggest that cleanups at sites in rural areas were begun 25% faster than at similar sites in urban areas, and 28% quicker than at similar sites in suburban areas. The pace of cleanup at sites in suburban areas was found to be 3% faster than the cleanup pace at similar sites in urban areas. Site cleanup is used here to mean the average number of months required to document the removal or reduction of hazardous substances to within state cleanup standards used at the time of site closure studyz”. According to these results, the pace of cleanup at study sites in white areas was 15% faster than similar sites in minority areas. Low income area study sites cleanup pace was 8% faster than sites in high income areas, and 18% faster than similar sites in moderate income areas. Finally, 2” As stated above, sweeping amendments to the Part 201 program were enacted by the Michigan Legislature in 1995. Overall, these amendments resulted in the significant lessening of cleanup standards. Refer to footnote 159 supra for a detailed explanation. 122 the cleanup pace at sites in high income areas was 11% quicker than at similar sites in moderate income areas. Site Cleanups Completed as Weighted by Site Score Regarding the number of sites at which cleanups were complete as weighted by site score, these results suggest that sites in urban areas were on average 3% more likely to have been cleaned up at the time of this study than similar sites in suburban areas, and 4% more likely to have been cleaned up than similar sites in rural areas. Sites in suburban areas were 6% more likely to have been finished with cleanup than similar sites in rural areas. Specifically, site cleanups completed is used here to mean the total number of recorded cleanups divided by the total number sites within each social demographic category, i.e. population density, income, and race/ethnicity as weighted by site score. According to study results, 17% more study sites in white study areas were cleaned up at the time of this study than similar sites in minority areas. Lastly, sites in high income areas were 2% more likely to be finished with site cleanup than similar sites in moderate income areas, and 5% more likely to be cleaned up than similar sites in low income areas. Study sites in low income areas were also found to be 3% more likely to have been completed with cleanup activities at the time of this study than similar sites in moderate income areas. Statistical Significance of Weighted Means Differences within Compliance Measures To assess the statistical significance of the difference in weighted compliance measures within demographic categories of race/ethnicity, income, and population density, t-Tests of independent-sample means were undertaken. Assuming population 123 means were equal within social demographic categories (race/ethnicity, income, and population density), statistical significance from independent t-Tests was used to determine if variability observed dependent variable means was the result of typical variability within a single population, or indicative of statistically significant difference in sample means. Table 16 below summarizes the results of independent sample t-Tests of Part 201 compliance measures by race/ethnicity, income, and population density. Table 16 I-I If E H [MI'II 1M -IIDEQC 1' Eli I 2-tai1 Response Inspections #Months to Meetings #Months to Cleanup Si ificance Time 1 Per Year Control Per Year Cleanup Ongoing? Urban/Suburban 0.570 0.652 0.675 0.593 0.930 0.888 Urban/Rural 0.337 0.893 0.320 0.270 0.332 0.877 ‘ Suburban/Rural . 0.489 0.598 0.254 0.572 0.411 0.973 Low /High 0.204 0.245 0.298 0.797 0.839 Income Moderate/Low 0.432 0.779 0.232 0.426 0.486 0.881 Income High/Moderate 0.432 0.440 0.796 0.197 0.791 0.931 Income .. Minority/White II 0.330 0.533 0.205 0.217 0.587 0.330 Bold numbers and shaded cells are statistically significant. 2-tail significance of <0.05 is considered statistically significant. 2-tail significance results more than 0.05 does not allow the rejection of the null hypothesis that two groups of Part 201 sites come from populations with the same average MDEQ compliance effort. 95% confidence interval. These results suggest that means differences in MDEQ and PRP meetings per year at sites in low and high income areas is statistically significant. As found in the results of the statistical analysis of unweighted means in Part 1 above, these findings indicate that 124 low income areas receive significantly more MDEQ compliance effort as measured by their initiation of PRP oversight meetings or willingness to attend PRP initiatives for face to face meetings to foster site cleanup. Following the recommendations of previous studies, the statistical significance of the means difference observed for the weighted, binomial compliance effort variable of whether or not site cleanup had been completed at the time of this study was statistically tested. Specifically, the Kruskal-Wallis nonparametric one-way AN OVA for this dependent variable was undertaken for all three demographic categories. Table 17 below summarizes the results of Kruskal-Wallis one-way ANOVA of the weighted Part 201 compliance measure of completed site cleanup by race/ethnicity, income, and population density. Table 17 sl_‘.Z-AZ‘.I'A2612|“0 zszr'-U|| 0 ruurzn' in Significance Corrected for Ties ll Completed Cleanups Population Density 0.99 Income 0.96 Race/ethnicity II 0.29 Bold numbers and shaded cells are statistically significant. 95% confidence interval. Significance of <0.05 is considered statistically significant. Corrected for ties significance of more than 0.05 does not allow the rejection of the null hypothesis that observations must be independent samples from populations with the same non-normal distribution (i. e. same average MDEQ compliance effort). These results suggest that observed means differences regarding the completion of site cleanup at the time of data collection for this study are not statistically significant. The null hypothesis is thereby not rejected that mean difference observations regarding 125 site cleanup status are from independent samples from within the same non-normally distributed population. Results of MDEQ Enforcement Effort Weighted by Site Score To evaluate weighted Part 201 program enforcement measures by demographic variables, means were again calculated and weighted by site score using SPSS®. The results of enforcement measure means calculations are summarized in Table 18 below. Table 18 Weighted # Information # Notice # Days to Negotiated Enforcement Means Requests Letters Notification Penalties Referrals Urban 0.000 1 .33 7,047 $0.00 .24 Suburban 0.21 1 .78 69 $3,657.43 .39 Rural 0.188 1.59 45 $0.00 .26 Low Income 0.150 1.90 195 $0.00 .30 Moderate 0.021 .78 69 $4,788.32 .39 Income High Income 0.000 .71 8,790 $0.00 .17 Minority ll 0.089 1 .10 1,018 $1,082.00 .31 White ll 0.000 ‘ 1.29 5,017 $1,041.33 .32 The distribution of responses for each enforcement variable was examined using SPSS', and plotted to determine if distribution was normal. The variable for PRP notification time (days) or "NOTDAYS#" was converted to a logarithmic scale to achieve a normal distribution. The other four MDEQ enforcement effort variables used were normally or approximately normally distributed. Weighted MDEQ Information Requests As found in Part 1, Table 18 suggests that the MDEQ relatively seldom used the 126 information request enforcement provisions of Part 201 as part of enforcement efforts at study sites. Specifically, the weighted number of information requests is defined as the total number information requests at a site over the years from discovery until cleanup weighted by site score. These findings suggest that suburban area sites were most likely to have MDEQ information requests of PRPs regarding the nature and extent of site contamination. Suburban sites were 11% more likely than similar rural sites to be subject to MDEQ information requests, but no information requests were made of urban sites subject to enforcement efforts. According to study results, minority area sites were subject to the infrequent use of the information requests by MDEQ enforcers, but similar sites within white areas captured in this study were not subject to any information requests. Lastly, low income area sites were subject 86% more MDEQ use of information requests than similar sites in moderate income areas, but no high income area sites undergoing enforcement and identified in this study were subject to information requests. Weighted MDEQ Notification of PRPs The aggressiveness of MDEQ identification and notification of PRPs was also weighted by site score and statistically analyzed. Results summarized in Table 18 suggest that rural areas on weighted average received the most diligent MDEQ effort to identify, notify and/or re-notify PRPs of violations of Part 201, followed by similar sites within urban and then suburban areas, respectively. Specifically, rural sites were 16% more likely than similar urban sites, and 51% more likely than similar suburban sites to be subject to agency PRP notification and/or re-notification. Urban area sites were 41% 127 more likely than suburban sites of similar relative risk to be subject to agency PRP notification. According to study results, PRPs at white area sites were 15% more likely to be notified of violation of Part 201 than PRPs at similar sites in minority areas. Finally, PRPs at sites in low income areas were 59% more likely to be notified than PRPs at similar sites in moderate income areas, and 63% more likely than PRPs at similar sites in high income areas. Sites in moderate income areas were 9% more likely to receive MDEQ notification efforts than similar sites in high income areas. Weighted Time for MDEQ Identification and Notification of PRPs Table 18 also summarizes the results of weighted means calculations for the number of days required by the MDEQ to locate and notify PRP(s) of potential Part 201 responsibility to investigate and remediate sites of environmental contamination. Findings suggest that PRPs at sites in rural areas were notified by the MDEQ 34% sooner than at similar sites in suburban areas, and 99% sooner than at similar sites in urban areas. PRPs at sites in suburban areas were notified 99% sooner than PRPs at similar sites in urban areas. According to study results, PRPs at minority sites were notified 80% sooner than at similar sites in white areas. PRPs at weighted moderate income area sites were notified 65% sooner than at similar sites in low income areas, and 99% sooner than at similar sites in high income areas. Finally, PRPs at sites in low income areas were notified 98% sooner than at similar sites in high income areas. Negotiated Penalties Weighted by Site Score If utilized by the MDEQ, successful enforcement effort brings recalcitrant PRPs back into Part 201 compliance. As stated above, this process often concludes with the 128 negotiation, agreement and approval of a Consent Decree including PRP stipulation to pay penalties for past Part 201 non—compliance. As stated above, the amount of stipulated penalties are ultimately at the discretion of the MDEQ and AG, and may serve as a useful measure of the social equity of Part 201 program enforcement. Regarding negotiated penalties at Part 201 enforcement sites included in this study as weighted by site score, results suggest that negotiated penalties at sites in moderate income areas were 100 percent higher than penalties at similar sites in low and high income areas. Negotiated penalties at Part 201 sites in white areas were on average 4% higher than negotiated penalties at similar minority area sites. Negotiated penalties at sites in suburban areas were 100 percent higher than negotiated penalties at similar sites in urban and rural areas. Sites with negotiated penalties did not exist within the study sample for all population density or income categories, and therefore could not be completely analyzed. Weighted MDEQ Enforcement Referrals Table 18 also summarizes the means calculations of the number of MDEQ enforcement referrals to the AG for weighted study sites within each demographic category. These results suggest that the sites in suburban areas were 33% more likely to be referred for enforcement than similar sites in rural areas, and 38% more than similar sites in urban areas. Rural sites weighted by site score were 8% more likely to be referred by the MDEQ to the AG for enforcement than similar sites in urban areas. Enforcement referrals are used here to mean official requests by MDEQ management for AG assistance in undertaking escalated enforcement actions against PRPs at Part 201 sites as weighted 129 by site score. According to study results, weighted sites in white areas were equally subject to enforcement referral as similar sites in minority areas. Sites in moderate income study areas were 23% more likely to be referred for enforcement than similar sites in low income areas, and 56% more likely to be referred for enforcement than similar sites in high income areas. Finally, low income area sites were 43% more likely to be referred for AG enforcement than similar sites in high income areas. Statistical Significance of Weighted Means Differences for MDEQ Enforcement Measures To assess the statistical significance of the difference in means of weighted enforcement measures, t-Tests of independent-sample means was undertaken for Part 201 enforcement variables. Assuming population means are equal within social demographic categories (race/ethnicity, income, and population density), statistical significance from independent t-Tests is used here to determine if variability observed is statistically significant. Table 11 below summarizes the results of independent sample t-Test of Part 201 enforcement measures by race/ethnicity, income, and population density. 130 Table 19 WW 2-tai1 # Information # Notice # Days to Negotiated #Enforcement Significance Requests Letters Notification Penalties Referrals — (log) =1 Urban/Suburban 0.436 0.276 0.138 0.090 0.571 Urban/Rural 0.328 0.666 0.107 mm 0.914 Suburban/Rural 0.276 0.102 0.870 0.602 Low /High 0.451 0.316 mm 0.456 Income Moderate/Low 0.346 0.565 0.694 Income ,. _ .. - High/Moderate 0.502 0.881 0.238 0.170 0.448 Income Minority/White Jl 0.450 0.677 0.741 0.975 0.982 Bold and shaded cells are statistically significant. 95% confidence interval. ”---" denotes that the standard deviation in both comparison groups was zero, precluding the performance of t-Test. 2-tail significance of <0.05 is considered statistically significant. 2-tail significance results more than 0.05 does not allow the rejection of the null hypothesis that two groups of Part 201 sites come from populations with the same average MDEQ enforcement effort. These results suggest that weighted means differences in the number of MDEQ notification letters and negotiated penalties between low and moderate income areas are statistically significant. Further, the number of MDEQ notice letters comparing suburban to rural areas is also statistically significant. Specifically, the results indicate that low income areas receive significantly more MDEQ enforcement effort than similar sites in moderate income areas as measured by weighted MDEQ notification of PRPs, but statistically significantly less negotiated penalties than enforcement cases at similar sites in moderate income areas. These results also indicate that rural areas receive significantly less MDEQ enforcement effort than similar sites in suburban areas, as measured by 131 MDEQ negotiated penalties at sites in rural areas as compared to similar sites in suburban areas. The statistical significance of observed means difference within weighted, binomial Part 201 program enforcement variables were also analyzed. Specifically, weighted enforcement variables analyzed included: whether or not MDEQ information requests were made; if PRPs were identified and notified of Part 201 violations and obligations; if formal referrals were made by the MDEQ to the AG for escalated enforcement actions; if penalties were levied; and if site cleanup was required through negotiated settlement of enforcement proceedings at the time of this study within demographic categories of race/ethnicity, income, and population density. The Kruskal- Wallis nonparametric one-way AN OVA was undertaken for Part 201 enforcement measures within all three demographic categories. As stated above, the Kruskal-Wallis test does not assume the normal distribution of dependent variable observations. The null hypothesis for this test is that the means of Part 201 program enforcement measures and the shape of their distributions are the same for social demographic categories (race/ethnicity, income, and population density). Statistical significance corrected forties from the fiom Kruskal-Wallis one-way AN OVA was used to determine if the observed variability suggested statistically significant difference in sample means fi'om influence of independent variables. Table 20 below summarizes the results of Kruskal-Wallis one-way ANOVA of the weighted Part 201 enforcement measures by race/ethnicity, income, and population density. 132 Table 20 Significance as Information Enforcement Penalties Cleanup Population Density 0.61 . 0.78 0.55 Income 0.74 0.88 0.69 0.35 0.65 0.21 0.88 Bold numbers and shaded cells are statistically significant. 95% confidence interval. Significance of <05 is considered statistically significant. Corrected for ties significance of more than 0.05 does not allow the rejection of the null hypothesis that observations must be independent samples from populations with the same non-normal distribution (1'. e. same average MDEQ enforcement effort). These results suggest that weighted means differences regarding PRP notification effort and Part 201 penalties by income; penalties levied by population density; and the MDEQ negotiation of an agreement for site cleanup by race/ethnicity are statistically significant. This allows for the rejection of the null hypothesis that the mean difference for these Part 201 enforcement measures are from independent samples from with the same non-normally distributed population. Specifically, these results indicate that sites in low income areas receive significantly more MDEQ effort regarding PRP notification; PRPs at sites in moderate income areas are fined significantly more than those at similar sites in low or high income areas; PRPs at sites in suburban areas were fined significantly more than those at similar sites in rural or urban areas; and that sites in minority areas receive significantly more MDEQ enforcement effort than similar sites in white areas, as measured by the negotiation of site cleanup as the result of agency enforcement actions. Results of Weighted Public Funding Evaluation To further evaluate the Part 201 program, public funding measures were also 133 weighted by site score and means were calculated by demographic variables using SPSS®. The results of weighted public funding measure means calculations are summarized in Table 21 below. Table 21 WNW Weighted Means Funds Requested Funds Spent Emergency Funds ‘ Funds Spent Recovered Urban ll $2239,083.33 $175,561.67 $24,685.12 $0.00 Suburban ll $26,037.89 $7,937.95 $0.00 $0.00 Rural ll $77,000.00 $46,898.88 $1,000.00 $3,500.00 Low Income ll $238,266.67 $176,168.44 $20,548.33 $2,800.00 Moderate Income " $26,037 .89 $7,937.94 $0.00 $0.00 High Income $3,000.00 $3,000.00 $0.00 $0.00 Minority u $122,028.33 $83,900.62 $10,214.66 $0.00 White $33,451.54 $23,332.98 $923.08 $ 1 ,923 .08 The distribution of responses for each weighted public funding variable was examined using SPSS®, and plotted to determine if distribution was normal. The above MDEQ public funding variables were normally or approximately normally distributed. Public Funds Requested Weighted by Site Score Table 21 suggests that weighted MDEQ requests for public funding is 88% higher for sites in urban areas than at similar sites in suburban areas, and 65% higher than at similar sites in rural areas. Further, sites in rural areas received 83% more requested funding for site investigation and cleanup than similar sites in suburban areas. According to study results, the MDEQ requested 73% more public funds to address sites in minority areas as compared to similar sites in white areas. Low income area sites were subject to 134 the highest level of MDEQ site fimding. Low income area sites were subject to MDEQ requests for 89% more public funds than similar sites in moderate income areas, and 98% more than similar sites in high income areas. Finally, sites in moderate income areas were subject to 88% more requested funds than similar sites in high income areas. Weighted Public Funds Spent Actual mean expenditures of public funds by the MDEQ were also weighted by sites score and statistically analyzed. These results are also summarized in Table 21 above. Similar to the results of analysis of frmds requested, it was found that sites in urban areas on average received 95% more public spending than at similar sites in suburban areas, and 73% more than at similar sites in rural areas. Further, sites in rural areas received 83% more public funding than similar sites in suburban areas. These results also indicated that the MDEQ spent 72% more public funds to address sites in minority areas than similar sites in white areas. Lastly, relative percentages of public expenditures at weighted sites were similar to funds requested when compared to income categories. Specifically, low income area study sites received 95% public funds than similar sites in moderate income areas, and 98% more than similar sites in high income areas. Sites in moderate income areas received 62% more public funds than similar sites in high income areas. MDEQ Emergency Fund Expenditures as Weighted by Site Score Table 21 also summarizes the results of weighted means calculations for public funds spent by the MDEQ to undertake to emergency response actions including: drum and/or surface removal of hazardous substances; the prevention of human exposure to 135 toxic materials; the prevention of hazardous material migration into public water supplies and/or sensitive habitats. It was found that 96% more emergency funds were spent at sites in urban study areas as compared to similar sites in rural areas. No emergency funds were spent at sites in suburban areas captured by this study. 91 % more emergency funds were spent by the MDEQ at sites in minority areas as compared to similar sites in white areas. This study also did not capture sites with MDEQ emergency fund expenditures in moderate or high income areas. Public Funds Recovered as Weighted by Site Score As stated above, results indicated that cost recovery actions were only undertaken in association with public fund expenditures within rural, low income, and white study areas precluding further statistical analyses. This was found despite the fact that public funds were spent in each demographic category. Statistical Significance of Means Differences within Weighted Public Funding Measures To assess the statistical significance of the difference in means found, t-Tests of independent-sample means was undertaken for each weighted MDEQ Part 201 public funding measure. Assuming population means are equal within social demographic categories (race/ethnicity, income, and population density), statistical significance from independent t-Tests is used here to determine if variability observed was the result of usual variability of sample means from a single population or indicative of statistically significant difference. Table 22 below summarizes the results of independent sample t- Tests of Part 201 public funding measures by race/ethnicity, income, and population 136 density as weighted by site score. Table 22 WW 2-tail Significance Funds Requested Funds Spent Emergency Funds Funds S nt Recovered l w Urban/Suburban 0.221 0.199 0.227 0.432 Urban/Rural 0.476 0.441 0.348 0.350 Suburban/Rural "P 0.232 0.070 0.227 0.241 r Low [High Income H 0.262 0.313 0.432 0.478 Moderate/Low 0.142 0.150 0.261 0.295 Income High/Moderate 0.468 0.466 0.000 0.504 Income Minority/White 0.480 0.549 0.549 0.128 Bold numbers and shaded cells are statistically significant. 95% confidence interval. 2-tai1 significance of <0.05 is considered statistically significant. 2-tail significance results more than 0.05 does not allow the rejection of the null hypothesis that two groups of Part 201 sites come from populations with the same average MDEQ compliance effort. These results indicate that none of the means differences observed for weighted MDEQ public funding measures were statistically significant. Specifically, the small sample size of Part 201 sites with public fund expenditures, and relatively little variance in observed difference likely resulted in the failure to reject the null hypothesis that observed Part 201 public funding mean values come from a population of all Part 201 sites with the same means. '.--._ ”a m.“ a” H . :91: . _ 2:5"; 2 ‘ .H14' H UT ,1; Based on the recommendations of previous studies, statistically significant Part 201 compliance, enforcement, and public funding measures, as weighted by site score, were further analyzed using univariate linear regression to determine the direction and 137 strength of correlation to demographic categories of race/ethnicity, income, and population density. The weighted number of meetings per year was the only Part 201 compliance measure found to be statistically significant by one-way AN OVA. Specifically, the weighted means difference of the number of meetings per year was found to be significant between low and high income areas. Regarding Part 201 enforcement measures, statistical significance was found in the comparison of the weighted means of: MDEQ notification effort between low and high income areas, and between moderate and low income areas; negotiated penalties between moderate and low income areas; and negotiated penalties between suburban and rural areas. For binomial or nonparametric weighted enforcement measures, statistical significance was found for penalties levied by population density; PRP notification and penalties levied by income; and MDEQ cleanup negotiation by race/ethnicity. No weighted Part 201 public funding measures were found with statistical significance in any demographic category. Consequently, statistically significant measures of ordinal, ratio, or interval scales were further analyzed using univariate linear regression to determine the direction and strength of correlation between demographic (independent) and Part 201 MDEQ evaluation (dependent) variables. Statistically significant binomial variables of nominal scale reported as above, were not further analyzed. For the purposes of linear regression analyses, "R" represents the Pearson correlation coefficient ranging from -1 to +1. A strongly positively correlated outcome 138 approaches +1, and a strongly negative correlation outcome nears -1. The "R square" represents the percent of observed variability within the dependent variable that is "explained" by the independent variable. The "Multiple of R" indicates how well the regression "fits", as it represents the correlation coefficient between the observed values of the dependent variable and the value predicted by the regression model. The F statistic is the ratio of mean squares that is used to test the null hypothesis that all coefficients are equal to zero, or in other words that the dependent and independent variables are not correlated. If the univariate linear regression coefficient is not zero, the Significance of F, or "overall regression F test", will be less than 0.05 at the 95% confidence interval. The finding of an overall regression P value of <0.05 allows for the rejection of the null, and indicates statistically significant positive or negative linear correlation. Findings of univariate linear regression are summarized in Table 23 below. Table 23 r {'9' H .zr.‘r=.t tr ':e| Measures Meetings/Year PRP Penalties by Penalties by by Income Notifications by Income Population Income Density Multiple R H 0.291 0.354 0.182 0.094 R Square ll 0.085 0.126 0.033 0.009 F It 3.780 5.884 0.755 0.194 Significance of F H 0.0582 0.394 0.664 Bold numbers and shaded cells are statistically significant 95% confidence interval. Significance of F of <0. 05 rs considered statistically significant. Significance of F results more than 0.05 does not allow the rejection of the null hypothesis that there is no linear relationship between the independent and dependent variables. 139 These findings indicate that each of the above Part 201 measures are positively correlated with relevant independent demographic variables. In other words, as the income of the area decreased, MDEQ compliance efforts increased to oversee and foster site cleanup through in-person meetings with PRPs, as did MDEQ enforcement effort to identify and notify PRPs to undertake site investigation and cleanup. But, as area income increased, so did MDEQ negotiated penalties levying against PRPs for violations of Part 201. Similarly, the less rural the area within which a Part 201 site existed, the higher MDEQ levied penalties for Part 201 non-compliance. Further, statistically significant positive correlation exists between the number of notifications of PRPs made by the MDEQ and the income level of the areas within which the site exists. Refer to Figures 2 through 5, pp. 161 through 165 for scatterplots of the linear relationships of these variables. Refer to Appendix D for copies of univariate linear regression analyses for each pair of dependent and independent variables. Note that the Multiple of R for meetings per year by income and PRP negotiations by income indicate that relatively strong positive linear relationships exist for both Part 201 measures. However, the R square results above indicate that only 8.5% of the variability observed in meetings per year, and 12.6% of PRP notifications, was respectively "explained" by the independent variable of income. Also refer to Figures 2 and 3, pp. 161 and 162 for further description. Further, the positive correlation between penalties by income and penalties by population density are weakly correlated, as indicated in Table 23 above and in Figures 4 and 5, pp. 163 and 164. 140 EII'MII' 'II' R . [ SI I'I' 11 5' 'fi Ilil'll ll! IZIIIM Also based on the recommendations of previous studies, weighted Part 201 compliance, enforcement, and public funding measures statistically significant in more than one demographic category, were further analyzed using multivariate linear regression to determine the strength of correlation and predictability of demographic categories of race/ethnicity, income, and population density. Specifically, the Part 201 enforcement measure of negotiated penalties was found to be statistically significant by income and population density. No other enforcement, compliance, or public funding measures were found to be statistically significant in more than one demographic category. Consequently, the enforcement measure of negotiated penalties was further analyzed using multivariate linear regression between income and population density to determine their comparative correlative strength and predictability for this dependent variable. The findings of this multivariate linear regression analysis are summarized in Table 24 below. 141 Table 24 U {'1202' 1‘; tween 1 . 3.1.‘12\_HHI :I A‘f'h' :ul Multiple R R Square F 0.484 I Significance of F 0.623 Bold numbers and shaded cells are statistically significant. 95% confidence interval. Significance of F of <0.05 is considered statistically significant. Significance of F results more than 0.05 does not allow the rejection of the null hypothesis that there is no linear relationship between the independent and dependent variables. These findings indicate that penalties levied by the MDEQ is weakly positively correlated with income and population density. In other words, as the income increased and rural character of the area decreased, MDEQ enforcement increased as reflected by the levying of penalties for violation of Part 201. However, these correlations are not statistically significant. Refer to Figure 6, p. 165 for a scatterplot of the linear relationship of these variables, and Appendix D for a copy of multivariate linear regression analyses for weighted negotiated penalties by income and population density. Note that the R square results for penalties by income and population density indicates that only 4.4% of the observed variability is "'explained" by these two independent variables, and that income was a stronger predictor of penalties than the population density of the area. Chapter 5 CONCLUSIONS AND RECOMMENDATIONS This study examined performance measures to determine the presence and degree of social discrimination, if any, in the implementation and enforcement of Michigan’s Part 201 program. As defined in this study, social equity is "the distribution of amenities and disadvantages across individuals and groups'Q‘E’0 that may result in "the disparate treatment of a group or community based upon race, class .. or some other distinguishing characteristic"26‘. Findings of Compliance Equity Analysis Regarding the analysis of measures of compliance equity within the Part 201 program as weighted by site score it was found that: 1) Part 201 sites in rural areas received 73 percent faster MDEQ response time than similar sites in suburban areas, and 86 percent faster response time than similar sites in urban areas. White area sites received a 73 percent faster response time 26° Zimmerman, Rae, "Issues of Classification in Environmental Equity: How We Manage is How We Measure", F ordham Urban Law Review, Vol. 21, 1994, p. 633. 2“ Gelobter, op. cit, p. 9, and Bullard, op. cit. 142 2) 3) 143 than similar sites in minority areas. And low income areas received 73 percent faster response time than similar sites in moderate income areas, and 90 percent faster response time than similar sites in high income areas. Using an equality means (t-Test) analysis, none of these findings was found to be statistically significant within a 95 percent confidence interval. These findings, while not statistically significant, are generally consistent with a study hypothesis, except that response time at low income area exceeded moderate and high income areas. Part 201 sites in suburban areas received 77 percent more MDEQ compliance inspections per year than similar sites in urban areas, and 74 percent more inspections than similar sites in rural areas. Minority area sites received 24 percent more MDEQ inspection effort than similar sites in white areas. And moderate income area sites received 43 percent more MDEQ annual inspection effort than similar sites in high, and 11 percent more than similar sites in low income areas. Using an equality means (t-Test) analysis, none of these findings was found to be statistically significant within a 95 percent confidence interval. These findings are generally inconsistent with a study hypothesis, except that MDEQ inspections at moderate income area sites slightly exceeded those at similar sites in low income areas. Part 201 sites in rural area sites were brought under control 36 percent sooner than similar sites in urban areas, and 47 percent sooner than similar sites in suburban areas. White area sites were controlled 46 percent sooner than similar sites in minority areas. And low income area sites were brought under control 38 percent 4) 5) 144 sooner than similar sites in high income areas, and 45 percent sooner than similar sites in low income areas. Using an equality means (t-Test) analysis, none of these findings was found to be statistically significant within a 95 percent confidence interval. These findings are generally consistent with a study hypothesis, except that cleanup pace at low income area sites exceeded those at moderate and high income areas. Part 201 sites in rural areas were subject to 19 percent more meetings between the MDEQ and polluters than at similar sites in suburban, and 44 percent more than at similar sites in urban areas. Sites in white areas were subject to 35 percent more MDEQ oversight meetings than similar sites in minority areas. High income area sites were subject to 62 percent less MDEQ meetings than similar sites in moderate, and 71 percent less meetings than similar sites in low income areas. Using an equality means (t-Test) analysis, the difference in prevalence of oversight meetings between the MDEQ and polluters between low and high income areas was found to be statistically significant within a 95 percent confidence interval. Univariate linear regression indicated a statistically significant correlation between income and meetings per year. As income decreased, MDEQ effort to arrange and attend oversight meetings significantly increased. These findings are generally consistent with a study hypothesis, except that meetings in low income areas exceeded both moderate and high income areas. The cleanup pace at Part 201 sites in rural area sites was 25 percent faster than at 6) 145 similar sites in urban, and 28 percent faster than at similar sites in suburban areas. The cleanup pace at white area sites was 15 percent faster than at similar sites in minority areas. And the cleanup pace at sites in low income areas was 8 percent faster than similar sites in high, and 18 percent faster than similar sites in moderate income areas. Using an equality means (t-Test) analysis, none of these results was found to be statistically significant within a 95 percent confidence interval. These findings are generally consistent with this study’s hypothesis, except that cleanup pace in low income areas slightly exceeded that of both moderate and high income areas. At the time of this study, three percent more Part 201 sites were completed with cleanup in urban areas than at similar sites in suburban areas, and 4 percent more than at similar sites in rural areas. Seventeen percent more white area sites were completed with cleanups than similar sites in minority areas. Two percent more high income sites had finished cleanups than similar sites in moderate income areas, and 5 percent more than similar sites in low income areas. Using an equality means (t-Test) analysis, none of these results was found to be statistically significant within a 95 percent confidence interval. These results may be a function of the frequency of ground water cleanups (expensive and time consuming) that were required rural sites, and not as likely to be required by the MDEQ at more urbanized sites are more likely to be supplied by municipal water supplies. However, the comparatively few completed cleanups in minority and low income areas, given the success of cleanup in urban 146 areas, may be a cause for concern by policy makers and the MDEQ. Findings of Enforcement Equity Analysis Regarding the analysis of measures of enforcement equity within the Part 201 program as weighted by site score it was found that: 1) 2) Part 201 sites in suburban areas were subject to 11 percent more MDEQ information requests for enforcement than similar sites in rural areas. Low income area sites were subject to 86 percent more MDEQ information requests than similar sites in moderate income areas. Information was not available for urban, minority, white and high income area sites for this measure of MDEQ Part 201 enforcement equity. Using an equality means (t-Test) analysis, none of these findings was found to be statistically significant within a 95 percent confidence interval. These findings were not statistically significant or complete, and are generally inconsistent with a study hypothesis. Part 201 sites in rural areas were subject to 16 percent more MDEQ enforcement effort, as measured by the number of official NDEQ enforcement notifications of polluters, than similar sites in urban areas, and 51 percent more than similar sites in suburban areas. White area sites received 15 percent more MDEQ enforcement notification effort than similar sites in minority areas. And low income area sites received 59 percent more MDEQ enforcement notification effort than similar sites in moderate, and 63 percent more than similar sites in high income areas. Using an equality means (t-Test) and non-parametric one-way analysis of variance, the differences found between low and high, and between low and 3) 4) 147 moderate income area sites, were found to be statistically significant within a 95 percent confidence interval. Univariate linear regression analysis of these findings evidenced a positive, statistically significant correlation between enforcement notification and income. Specifically, as income decreased, MDEQ enforcement notification effort significantly increased. These findings are generally consistent with a study hypothesis, except the statistically significant finding of strong MDEQ enforcement notification effort in low income over high and moderate income areas. Parties deemed responsible by the MDEQ for creating Part 201 sites in rural area sites were notified of enforcement actions 34 percent sooner than those responsible for similar sites in suburban areas, and 99 percent sooner than at similar sites in urban areas. Enforcement notifications for polluters at minority area sites were 80 percent quicker than at similar sites in white areas. And polluters at moderate income area sites were notified of MDEQ enforcement actions 65 percent sooner than at similar sites in low, and 99 percent sooner than at similar sites in high income areas. Using an equality means (t-Test) analysis, none of these findings were found to be statistically significant within a 95 percent confidence interval. These findings are, in part, consistent with a study hypothesis. Results indicating more MDEQ enforcement effort at sites in minority over white, and low over high income areas are not consistent with the hypothesis of this study. Part 201 sites in white areas were subject to 4 percent higher enforcement 5) 148 penalties than similar site in minority areas. Enforcement penalties were 100 percent higher at sites in moderate income areas as compared with low and high income areas. And penalties levied by the MDEQ for sites in suburban areas were 100 percent higher that at sites in urban and rural areas. Insufficient data for income and population density categories precluded complete analysis. Using an equality means (t-Test) and non-parametric one-way analysis of variance, the difference in penalties found between moderate and low income, and suburban and rural income areas were found to be statistically significant within a 95 percent confidence interval. Univariate linear regression analysis of these findings evidenced a positive, though statistically insignificant, correlation between penalties and income, and penalties and population density. Specifically, as income and population density increased, so did MDEQ negotiated penalties. Multivariate linear regression of penalties by income and population density evidenced a weak, positive correlation. Specifically, as income increased and rural character decreased, penalties levied by the MDEQ increased, though not at a statistically significant level. Finally, income was determined to be a stronger indicator of Part 201 penalties than the population density of an area. These findings are partially consistent with this study’s hypothesis. The referral of Part 201 sites by the MDEQ to the AG for enforcement proceedings in suburban areas exceeded that of similar sites in rural areas by 33 percent, and exceeded that at similar sites in urban areas by 38 percent. No difference was found in enforcement referral between white and minority area 6) 149 sites. Enforcement referral for sites in moderate income areas exceeded those for similar sites in low income areas by 23 percent, and exceeded those at similar sites high income areas by 56 percent. Using an equality means (t-Test) analysis, none of these results was found to be statistically significant within a 95 percent confidence interval. This findings are primarily inconsistent with a study hypothesis. Finally, whether or not site cleanup was required as a part of a court-enforceable settlement agreement between the MDEQ and a polluter was statistically determined to be significantly more likely for Part 201 sites in minority over white areas. Through the use of non-parametric one-way analysis of variance, difference observed within income and population density categories were not determined to be significant. This finding refutes the study hypothesis regarding race/ethnicity and Part 201 enforcement, and neither refutes nor confirms hypotheses regarding income and population density. Findings of Public Funding Equity Analysis Regarding the analysis of measures of public funding equity within the Part 201 program as weighted by site score it was found that: 1) Public funds requested by the MDEQ to address orphaned Part 201 sites in urban areas were 88 percent higher than requests for similar sites in suburban, and 65 percent higher than similar sites in rural areas. Seventy-three percent more public funds were requested by the MDEQ to address similar orphaned sites in minority areas over similar sites in white areas. Low income area sites were subject to 89 2) 3) 150 percent more MDEQ frmding requests than similar sites in moderate income areas, and 98 more than similar sites in high income areas. Using an equality means (t-Test) analysis, none of these findings was found to be statistically significant within a 95 percent confidence interval. These findings were not statistically significant and are contrary to a study hypothesis. Public funds spent by the MDEQ to address orphaned Part 201 sites in urban areas were 95 percent higher than funds spent by the MDEQ at similar sites in suburban areas, and 73 percent higher than similar sites in rural areas. Seventy- two percent more public funds were requested to address orphaned sites in minority areas over similar sites in white areas. Low income area sites were subject to 95 percent more MDEQ funding than similar sites in moderate income areas, and 98 more than similar sites in high income areas. Using an equality means (t-Test) analysis, none of these findings was found to be statistically significant within a 95 percent confidence interval. These findings were not statistically significant and are inconsistent with a study hypothesis. Emergency funds spent by the MDEQ to address Part 201 sites posing imminent hazards to human health or the environment in urban areas were 96 percent higher than funds spent at similar sites in rural areas. No emergency funds were spent at suburban sites captured in this study. Ninety-one percent more emergency funds were spent at in minority areas over similar sites in white areas. This study also did not capture sites with emergency expenditures in moderate or high income areas. Using an equality means (t-Test) analysis, none of these findings was 4) 151 found to be statistically significant within a 95 percent confidence interval. These findings were not statistically significant and are inconsistent with this study’s hypothesis. MEDQ efforts to recover costs of public expenditures at orphaned Part 201 sites occurred only at sites within rural, white and low income areas. Using an equality means (t-Test) analysis, these findings were not found to be statistically significant within a 95 percent confidence interval. These findings, while not statistically significant, are partially consistent with a study hypothesis. CONCLUSIONS This study hypothesized that measurable and statistically significant difference exists within minority, urban, and/or low income communities in Michigan in the implementation and enforcement of the Part 201 program. Based on these findings, the null hypothesis was not rejected. Specifically, it was found that: 1) 2) 3) 4) 5) Low income is strongly correlated with increased MDEQ compliance effort to foster and oversee site cleanup through frequent face to face meetings. Low income is strongly correlated with increased MDEQ Part 201 enforcement effort to locate and notify PRPs. Higher income was found to be strongly correlated with increased MDEQ penalties levied against PRPs in Part 201 enforcement cases. The more suburbanized an area, the higher penalties imposed upon PRPs by the MDEQ. Sites in minority areas are more likely to be compelled by the MDEQ to be 152 cleaned up in enforcement negotiations and settlement agreements. 6) Income was a stronger predictor than population density of higher Part 201 penalties in enforcement cases. INTERPRETATIONS AND IMPRESSIONS This study hypothesizes that significant bias against poor, minority, and urban areas exists in Michigan’s Part 201 program. Due to the relatively few sites undergoing enforcement or subject to public funding in Michigan, this hypothesis is supported only in part by these findings. However, several implications regarding the implementation of the Part 201 program extend from these results. Major program and/or policy implications include: 1) Missing site files, over 8 percent of those requested, only existed for sites within minority areas. This finding could be interpreted critically by those seeking to monitor and/or participate in the cleanup of sites of environmental contamination in these areas. The MDEQ should take special care to maintain the public record for sites of environmental contamination, and forgo any perception of bias in doing so. 2) The MDEQ should collect, evaluate and report the social impact of their implementation of “socially-blind” environmental and human health protection programs, such as the Part 201 Environmental Response Program. If inequity is found, that policy reform should be undertaken in the light of day to address it. To date, the MDEQ does not track, let alone evaluate, program performance measures regarding “Where” and “For whom” it is spending its compliance, 3) 4) 5) 153 enforcement, and public funding effort under this program. Very little research has been previously undertaken regarding the social equity of environmental and human health protection programs, state or federal. Previous research has been openly or allegedly politically motivated“? Some past research has been undertaken by opposing sides in “NIMBY” debates or commissioned for use in site-specific litigation. It is important to begin to develop and refine this area of study to consider potential social inequity in program implementation. Such research if done by implementing agencies, should be guided by and/or open to public evaluation and participation. Although not required currently by Michigan’s Part 201, enhanced opportunities for public participation in overall site cleanup is an important first step to enhancing sustainable community development and the reuse of contaminated or potentially contaminated facilities. Research and public debate regarding environmental justice issues generally should be widened to consider social equity in terms of income, gender, age, inter- generational equity, and other forms of social difference. Current debate and research is dominated by issues of race, and fails to consider the interbedded nature of social oppression. A good start may be made by shifting the focus public environmental policy to the most vulnerable segments of the population in determining public policy issues of “acceptable risk” or “safe levels”. As suggested by these findings, it is important to insure MDEQ compliance effort 262 See Zax, op. cit. and Goldman. Op. Cit. 1992- 1 54 is effective and equitable in minority areas. Concern is raised by the finding that cleanup in minority areas was significantly associated with MDEQ enforcement effort, as relatively little enforcement is undertaken within the universe of Part 201 sites. These findings suggest potential bias against the effective and efiicient cleanup at sites in minority areas. RECOMMENDATIONS FOR FURTHER RESEARCH State and federal environmental and human health agencies have been criticized for disassociating themselves from the structural and social contexts within which they exist, and for approaching environmental protection in a socially-neutral manner. Agencies have also been criticized for implementing compliance and enforcement programs without any "appreciation, or acknowledgment, of the social context and structural dynamics that influence choices, mobility, and employment of people of color"2‘3. To most accurately assess the social equity of the implementation of governmental programs, it is critical that a wide variety of performance measures be operationalized and analyzed using multiple statistical methods. Further, based upon the findings of relevant performance measures, an overall assessment of an environmental program of interest should be undertaken. The presence of one or more measures of inequity, while perhaps not representing institutionalized discrimination, may be sufficient for the public to request and guide governmental reform. However, the degree 253 Foster, op. cit, pp. 729, and 736-737. Also see Hurley, op. cit. 155 of discrimination, if found, that is "acceptable" is as much a public policy decision as the degree of risk deemed acceptable from the public's enviromnental exposure to hazardous materials. Based on this and previous research, it is recommended that future research consider the following: 1) 2) 3) Sample size of similar studies must be sufficiently large to capture meaningful variation within the population of cases subject to a governmental environmental protection program. This study may have failed to support its hypothesis, especially concerning NHDEQ enforcement and public funding measures, due the relatively small sample size. On the other hand, relatively few Part 201 sites are subject to MDEQ enforcement or the recipient of public funding to address hazardous waste contamination. Based upon the lack of agreement within the literature, it is important to undertake analyses using both zip code, census track and other geographic units of analysis to represent the "affected community" or neighborhood. Once meaningful comparative study has been undertaken, this methodological problem thus far vexing environment equity studies may finally be resolved. It is important to search for more informed and meaningful demographic measures of race/ethnicity, income, and population. In an attempt to do so, it is recommended that previous national or statewide study be replicated using several geographic units of analysis and additional study be undertaken within specific communities. This local focus may better capture significant social difference 4) 5) 6) 156 potentially masked at the state or national level, and lend more meaning and relevance to findings regarding environmental program implementation by various levels of government within the U.S. More qualitative, community-led research should be undertaken regarding environmental justice. Community perceptions of the location and meaning of environmental risk are not well understood. Further research should attempt to move to a more nuanced understanding of environmental discrimination, its many forms, and how it is produced and reproduced. Such research should include more radical social analyses of how to enhance democratic participation and self-determination in public policy decisions with deleterious environmental consequences. Further research should include consideration of how to dramatically limit and eventually eliminate exposure to toxins for all, rather than the equitable distribution of risk and/or harm?“ A paradigmatic shifi is needed from spatial, locational equity empiricism to the processes that create inequity in which unjust outcomes are imbedded?" As stated by Lake (1996), “focus (should be) on the structure of production and ..the ways in which communities are linked (or not linked) to ...decisions (through which negative environmental consequences are created) and to the process of 26‘ Pulido, op. cit., p. 143; Swanson, op. cit., p. 602; Michael K. Heiman, “Race, Waste And Class: New Perspectives On Environmental Justice”, Antipode, Vol. 28, No. 2, 1996, pp. 113-114; and Robert Lake, “Volunteers, NIMBYs, And Environmental Justice: Dilemmas Of Democratic Practice, Antipode, Vol. 28, No.2, 1996, p. 162. 265 Lake, op. cit, p. 170. 7) 8) 9) 157 uneven development”.266 More study should be done such the National Law Journal 1992 report and this research, to focus on equity within the implementation of environmental protection programs.267 The literature should seek to mature from the plethora of hazard or risk proximity studies that exist vexed by methodological shortcomings to the analyses of program implementation and their consequences. Future research should seek to reflect a broader conceptualization of environmental justice beyond the “chicken and egg”debate of which came first, social or environmental decay in urban cores. The focus should shift to exploring and testing ways in which to improve the overall situation. As common in all other developed economies, the MDEQ and other U.S. environmental regulatory agencies, should discard the “socially-blin ” approach to policy and program implementation, and begin to collect and analyze demographic information to augment the monitoring of the efficiency and social efficacy of programs. 2“ [bid 2‘57 Lavelle and Coyle, op. cit, and Goldman, op. cit, 1996, pp. 132, 138. FIGURES Figure 1 Part 201 Process PRP Violation - Reportable Release of Hazardous Substance 1 PRP Identification by MDEQ l MDEQ Notification of PRP L 1 PRP Site Investigation Plan MDEQ Review of Investigation Plan 1 l 1 PRP Cleanup Plan 1 PRP Compliance ‘— v PRP Continued Noncompliance MDEQ Enforcement Action MDEQ Enforcement Notification of PRP PRP Compliance MDEQ Review of Cleanup Plan l PRP Site Cleanup! MDEQ Review of Closure Report Consent Decree & Penalty Negotiations MDEQ Suit Against PRP H MDEQ Unilateral C leanup Order HL___9 MDEQ Site Closure Letter & MDEQ Site "Delisting" 1 59 Funding l L__/ See Fig. 1(contlnued) Figure 1 (continued) Part 201 Process From Fig. 1 __p MDEQ Declaration of Emergency - .9. Direct Contact Hazard Affected Water Supply Fire/Explosion Hazard PRP Noncompliance at High Priority Site I iscretionary MDEQ Finding of PRP Recalcitrance MDEQ Staff Request for Public Funding Cost Recovery Litigation Cost Recovery 8 Penalty Negotiations l ” < MDEQ Management Decision to Fund so 1 yes MDEQ Hires Consultant! Contractor i MDEQ Site Plan (Consultant) DEQ Seeks Cost Recovery MDEQ Site MDEQ Site Investigation Closure & and/or Cleanup ———> Delistin (Consultant) 9 Expected Cumulative Probability 8 Normal P-P Plot of Regression Standardized Residual Meetings/Year by Income Figure 2 1.00 .757 in o 1 i\) or I ° Normal Line ° Observed Weighted Means (zresid) Total Population I I I 0.00 Observed Cumulative Probability .25 .50 .75 161 1.00 Expected Cumulative Probability Normal P-P Plot of Regression Standardized Residual PRP Notifications by Income Figure 3 1.00 .75‘ .50“ .25 . / I / I 0.00 0.00 I .25 .50 I I .75 Observed Cumulative Probability 162 ° Normal Line ° Observed Weighted Means (zresid) Total Population 1.00 Expected Cumulative Probability Normal P-P Plot of Regression Standardized Residual Enforcement Penalties by Income Figure 4 1.00 .75 ‘ '01 O l \ \ id or .0 8 ° Normal Line ° Observed Weighted Means (zresid) Total Population I I I 0.00 .25 .50 .75 Observed Cumulative Probability 163 1.00 APPENDICES APPENDIX A Data Collection Sheet Site Name: MDEQ District: Site Number: Date: Compliance: C1 - number of days (discovery to response) C2 - inspections per year (after discovery) C3 - meetings per year (MDEQ & PRP) C4 - number of months to control C5 - number of months to cleanup Enforcement: Funding: E1 - number of information requests Eld - date of information request (earliest) E1d2- date of information request (latest) E2 - number of notice letters (Part 201) E2d - date of notice letter (earliest) E2d2- date of notice letter (latest) E3 - number of enforcement referral(s) E3d - date of enforcement referral(s) E3d2- date of enforcement referral(s) E4 - number of months (referral to start to negotiate) E5 - number of months (conclude settlement) E6 - amount of financial settlement E7 - date of lawsuit (earliest) E8 - date of lawsuit (finish) E9 - financial penalties (lawsuit) F1 - funds ($) requested F2 - funds ($) spent F3 - funds (8) cost recovered F4 - emergency firnds (S) spent F5 - number of private dollars spent 167 APPENDIX B I . | [If . l I Variable Name Label Type Width (decimal) SITEID# MERA# String 8 SITENAME NAME String 20 SCORE Site List Score Numeric 2(0) SCORERAT Site Score Ratio Numeric 4(2) COUNTY County String l8 ZIP Zip Code Numeric 5 INCOME Income String 12 INCODE Income Code Numeric 1(0) Value Label 1 Low 2 Moderate 3 High RACE Race String 12 RCODE Race Code Numeric 1(0) Value Label 1 Minority 2 White DENSITY Density String 8 DCODE Density Code Numeric 1(0) Value Label 1 Urban 2 Suburban 168 3Rura1 DISCOVER Discovery Date Date 8 C1 # Days Response Numeric 8(0) LOGCl Log # Days Rep Numeric 8(2) C2 #Inspections/Year Numeric 8(4) C3 #Meetings/Y ear Numeric 8(4) F IRSTMTG First Mtg Date 8 C4 #Months to Control Numeric 8(0) C5 #Months to Cleanup Numeric 8(0) C6 Ongoing Numeric 1(0) Value Label 0 Cleanup done 1 Cleanup ongoing E1 Information requests Numeric 1(0) E 1 D Info Request Date Date 8 E1D2 2nd Info Request Date 8 E2 #Enf Notices Numeric 2(0) NOTDAYS# #Days Until Notice Numeric 6(0) LOGNOT#D Log #Days to Notify Ntuneric 8(2) E2D Enforce Notice Date Date 8 E2D2 2nd Notice Date Date 8 E3 #Enforce Referral Numeric 1(0) E3D Enf Referral Date Date 8 169 E3D2 Last Enf Notice E4 Enf Start Date Date E5 Enf Complete Date E6 Enf Settlement INJUN C Injunctive Relief Value Label 0 No 1 Yes E7 Litigation Start E8 Done Litigation E9 Penalties Fl Funds Requested F2 Funds Spent F3 Cost Recovery F4 Emergency Funds DISTRICT MDEQ District STAFF# 1990 #Staff DISTANCE Distance (Miles) Date Date Dollar Numeric Date Date Dollars Dollars Dollars Dollars Dollars String Numeric Numeric 170 15(2) 1(0) 15(2) 15(2) 15(2) 15(2) 15(2) 20 2(0) 3(0) BIBLIOGRAPHY BIBLIOGRAPHY Adeola, Francis 0., “Environmental Hazards, Health, and Racial Inequality in Hazardous Waste Distribution”, Environment and Behavior, Vol. 26, No. l, 1994, pp. 99- 126. Agency for Toxic Substances Disease Registry, The Nature and Extent of Lead Poisoning in Children in the United States: A Report to Congress, Centers for Disease Control, Atlanta, Georgia, 1988. Asch, Peter and Joseph L. Seneca, "Some Evidence on the Distribution of Air Quality", Land Economics, Vol. 54, No. 3, 1978, pp. 278-297. Anderson, Andy B., Douglas L. Anderton, and Michael Oakes, "Environmental Equity: Evaluating TSDF Siting Over the Past Two Decades", Waste Age, Vol. 25, No. 7, July 1994, pp. 83-84, 86, 88, 90, 92, 94, 96, 98 and 100. Anderton, Douglas L., Andy B. Anderson, John Michael Oakes, and Michael R. Fraser, "Environmental Equity: The Demographics of Dumping", Demography, Vol. 31, No. 2, 1994, pp. 229-248. Anderton, Douglas L., Andy B. Anderson, Peter H. Rossi, John Michael Oakes, Michael R. Fraser, Eleanor W. Weber, and Edward J. Calabrese, "Hazardous Waste Facilitiesz’Environmental Equity’ Issues in Metropolitan Areas", Evaluation Review, Vol. 18, No. 2, April 1994, pp. 123-140. Baumol, W. J. and W. E. Owens, Economics, Environmental Policy, and the Quality of Life, Prentice -Hall, 1979. Been, Vicki, "What's Fairness Got To Do With It? Environmental Justice and the Siting of Locally Undesirable Land Uses", Cornell Law Review, Vol. 78, 1993, pp. 1001-1085. Been, Vicki, "Locally Undesirable Land Uses in Minority Neighborhoods: Disproportionate Siting or Market Dynamics?", Yale Law Journal, Vol. 103, 1994, pp. 1383-1422. 172 Been, Vicki, "Analyzing Evidence of Environmental Justice", Journal of Land Use and Environmental Law, Vol. 11, No. 1, Fall 1995, pp. 1- 36. Been, Vicki, "Environmental Justice", Chapter 25 in P. Rohan, Zoning and Land Use Controls, 1995, D02. Belliveau, Michael, Michael Kent, Brant Rosenblum, "Richmond At Risk: Community Demographics and Toxic Hazards from Industrial Polluters", San Francisco, CA: Citizens for a Better Environment, 1989. Berry, Brian, J. L., Susan Caris, Daniel Gaskill, Charles P. Kaplan, John Piccinni, Neil Planert, James H. Rendell Ill, and Alexandra de Ste. Phalle, The Social Burdens of Environmental Pollution: A Comparative Metropolitan Data Source, Ballinger Publishing Company, Cambridge, MA, 1977. Boemer, Christopher and Thomas Lambert, Environmental Justice In the City of St Louis: The Economics of Siting Industrial and Waste Facilities, Working Pater 156 for the Center for the Study of American Business, Washington University, St. Louis, Missouri, April 1995. Bowen, William, M., Mark J. Salling, Ellen J. Cyran, Heather A. Moody, The Spatial Association Between Race, Income and Industrial Toxic Emissions in Cuyahoga County, Ohio, Atlanta, GA: Presentation for the Annual Meeting of the Association of American Geographers, May 1993. Breslin, Karen, “In Our Own Backyards: The Continuing Threat of Hazardous Waste”, Environmental Health Perspectives, Vol. 101, No. 6, November 1993, pp. 484- 489. Brion, Adam "An Essay on LULU, NIMBY and the Problem of Distributive Justice", B. C.J. Environmental Aflairs, Vol. 15, p. 437. Brown, Kevin, L. "Environmental Discrimination: Myth or Reality", St. Louis, MO: Supervised Research Paper, Washington State University School of Law, March 29, 1991. Browner, Carol, "Carol Browner on EPA Priorities", EPA Journal, April-June 1993, p. 4. Brueggemann, Martin, R., "Environmental Racism in Our Own Backyard: Solid Waste Disposal in Holly Springs, North Carolina", Chapel Hill, NC. Master's Thesis for the University of North Carolina Graduate School of Journalism, 1993. 173 Bryant, Bunyan and Paul Mohai, Editors, Environmental Racism: Issues and Dilemmas, University of Michigan Press, 1991. Bryant, Bunyan and Paul Mohai, Race and the Incidence of Environmental Hazards: A Time for Discourse, Westview Press, Boulder, 1992. Bryant, Bunyan and Elaine Hockman, Hazardous Waste and Spatial Relations According to Race and Income in the State of Michigan, June, 1994. Bullard, Robert D., Editor, Confionting Environmental Racism: Voices from the Grassroots, Boston, Massachusetts, South End Press, 1993. Bullard, Robert D., Dumping in Dixie: Race, Class, and Environmental Quality, Westview Press, Boulder, San Francisco and Oxford, 1990. Bullard, Robert D., "Ecological Inequities and the New Black South: Black Communities Under Siege", Journal of Ethnic Studies, Volume 17, Number 4, Winter 1990, pp. 101-1 15. Bullard, Robert D., "Overcoming Racism in Environmental Dccisionmaking", Environment, Vol. 36, No. 4, May 1994, p. 10. Bullard, Robert D., "Solid Waste Sites and the Houston Black Community", Sociological Inquiry, Volume 53, Winter 1983, pp. 273-288. Bullard, Robert D., "Politics of Race and Pollution", Multinational Monitor, June 1992, pp. 22-25. Burch, Jr., William R., "The Peregrine Falcon and the Urban Poor: Some Sociological Interrelations", Human Ecology: An Environmental Approach, Peter J. Richerson and James McEvoy, Editors, Belmont California, Duxbury Press, 1976, pp. 308- 3 16. Burke, Lauretta M., “Race and Environmental Equity: A Geographical Analysis in Los Angeles”, Geo Info Systems, October 1993, pp. 44-50. Calabrese, Edward J ., Pollutants and High Risk Groups: The Biological Bias of Increased Human Susceptibility to Environmental and Occupational Pollutants, 1978. Clean Sites, Hazardous Waste Sites and the Rural Poor: A Preliminary Assessment, March, 1990. 174 Clean Water Action Fund, If It's Broke, Fix It: Why Michigan’s Environmental Health Agencies Must Make Changes to Help Detroiters and Others Fish Without Fear, Lawton K. Jackson, Detroit Coordinator, January 1993. Cole, Luke W., "Remedies for Environmental Racism: A View from the Field", Michigan Law Review, June 1992, Vol. 90, No. 7, pp. 1991-1997. Collin, Robert W., Environmental Equity: A Law and Planning Approach To Environmental Racism", Virginia Environmental Law Journal, 1992, Vol. 11, No. 4, pp. 495-546. Collin, Robert W., "Review of Legal Literature on Environmental Racism, Environmental Equity, and Environmental Justice", Journal of Environmental Law and Litigation, Vol. 9, 1994, pp. 121-171. Colquette, Kelly M. and Elizabeth A. Henry Robertson, "Environmental Racism: The Causes, Consequences, and Condemnations", Tulane Environmental Law Journal, 1992, Vol. 5, Issue 1, pp. 153-207. Costner, Pat and Joe Thornton, Playing with Fire: Hazardous Waste Incineration, Greenpeace, USA, Washington, D.C., 1990. Cutter, Susan L., “Race, Class and Environmental Justice”, Progress in Human Geography, Vol. 19, No. 1, 1995, pp. 111-122. Cutter, Susan, L., “The Burdens of Toxic Risks: Are They F air?”, Business and Economic Review, Vol. 41, October-December, 1994, pp. 3-7. Davis, C., "Group Environmental Preference Aggregation - The Principle of Environmental Justice", Journal of Environmental Management, Vol. 22, No. l, 1986, pp. 55-65. Epstein, Samuel 8., “Environmental And Occupational Pollutants Are Avoidable Causes Of Breast Cancer”, International Journal of Health Services, Vol. 24, No. 1, 1994, pp. 145-150. F ahsbender, John J., "Student Article: An Analytical Approach To Defining the Affected Neighborhood In The Environmental Justice Context", New York University Environmental Lanournal, Vol. 5, 1996, pp. 120-180. F eagin, Joe R. and Clairece Booher Feagin, Discrimination American Style: Institutional Racism and Sexism, Malabar, F 1a.: Robert E. Krueger Publishing, 2nd Edition, 1986. 175 Federal Register, 1.59 No. 32, February 16, 1994, Executive Order 12898 of February 11, 1994, pp. 7629 - 7633. Fitton, Laura J., " A Study of the Correlation Between the Siting of Hazardous Waste Facilities and Racial and Socioeconomic Characteristics", Ithaca, New York, A Master's Thesis for Cornell University, December 1992. Fitton, Laura J ., John Choe and Richard F egan, Environmental Justice: Annotated Bibliography, The Center For Policy Alternatives; Washington D.C., September 1994. Fisher, Michael, "Environmental Racism Claims Brought Under Title VI of the Civil Rights Act", Northwest School of Law Lewis and Clark College Law Review, Vol. 25, No. 2, Spring 1995, PP. 285 - 333. Foster, Sheila, "Review Essay: Race(ial) Matters: The Quest for Environmental Justice", Ecology Law Review, Vol. 20, 1993, pp. 721-753. Freeman, A. Myrick, "The Distribution of Environmental Quality", in Environmental Quality Analysis, A. V. Kneese and B. T. Bower, Editors, Baltimore: Johns Hopkins University Press for Resources for the Future, 1972. Friedman-Jimenez, George, "Occupational Disease Among Minority Workers: A Common and Preventable Occupational Health Problem", American Association of Occupational Health Nurses Journal, Volume 37, 1989, pp. 64-65. Gelobter, Michel, The Distribution of Outdoor Air Pollution by Income and Race: 1970- 1986, unpublished Master's Thesis, Energy and Resource Group, University of California, Berkeley, California, 1988. Gelobter, Michel, "Toward a Model of 'Environmental Discrimination” in The Proceedings of the Michigan Conference on Race, Bryant and Mohai, Editors, University of Michigan School of Natural Resources, 1990, pp. 87-107. Gelobter, Michel, "The Meaning of Urban Environmental Justice", F ordham Urban Law Journal, Vol. 21,1994, pp. 841- 856. Geschwind, Sandra A., Jan A. J. Stolwijk, Michael Bracken, Edward Fitzgerald, Alice Stark, Carolyn Olsen, and James Meilus, “Risk of Congenial Malformations Associated with Proximity to Hazardous Waste Sites”, American Journal of Epidemiology, Vol. 135, No. 11, 1992, pp. 1197 - 1207. 176 Geschwind, Sandra A., “Should Pregnant Women Move? Linking Risks for Birth Defects with Proximity to Toxic Waste Sites”, Chance: New Directions for Statistics and Computing, Vol. 5, No. 3-4, 1992, pp. 40-45, and 86. Gianessi, Leonard P., Henry M. Peskin and Edward Wolff, "The Distributional Effects of Uniform Air Pollution Policy in the U.S.", Quarterly Journal of Economics, May 1979, pp. 281-301. Gianessi, Leonard P. and Henry M. Peskin, "The Distribution of Federal Water Pollution Control Policy", Land Economics, Vol. 56, No. 1, February 1980, pp. 85-102. Glickman, Theodore 8., Dominic Golding, Robert Hersh, “GIS-Based Environmental Equity Analyses; A Case Study of TRI Facilities in the Pittsburgh Area”, in Computer Supported Risk Management; Wallace, WA., Beroggi, E. G., Editor, Kluwer Academic Publishers, Netherlands, 1995, pp. 95-114. Godsil, Rachel D., "Remedying Environmental Racism", Michigan Law Review, Vol 90, No. 4, pp. 394-427. Goldman, Benjamin A., The Truth About Where You Live: An Atlas for Action on Toxins and Mortality, New York, New York, Times Books/Random House, 1991. Goldman, Benjamin A., Not Just Prosperity: Achieving Sustainability with Environmental Justice, commissioned by the National Wildlife Federation and the Corporate Conservation Council, February, 1992 Goldman, Benjamin A., “What Is The Future Of Environmental Justice”, Antipode, Vol. 28, No. 2, 1996, pp. 122-141. Goldman, Benjamin A. and Laura F itton, Toxic Wastes and Race Revisited, 1994. Goldstien, et. al., "Indoor Air Pollution: Exposure of Low-Income Inner City Residents", Environmental International, 1986, Vol. 12, pp. 211-219. Gould, Jay M., Quality Of Life In American Neighborhoods: Levels Of Afiluence, Toxic Waste, And Cancer Mortality In Residential Zip Code Areas, Westview Press, Boulder and London, 1986. Greenburg, Michael R., "Proving Environmental Inequity in Siting Locally Unwanted Land Uses", Risk Issues in Health & Safety, Vol. 4, Summer 1993, pp. 235- 252. 177 Greenberg, Michael R., “Separate and Not Equal: Health, Environmental Risk and Economic-social Impacts in Remediating Hazardous Waste Sites”, in Shyamal K. Majumdar, F. J. Brenner, E. W. Miller, and L. M. Rosenfeld, Editors, Environmental Contaminants and Health, Philadelphia, PA: Pennsylvania Academy of Sciences, 1994. Handy, Femida, "Income and Air Quality in Hamilton, Ontario", Alternatives, Vol. 6, No. 3, 1977, pp. 18 - 24. Hamilton, James T., “Politics and Social Costs: Estimating the Impact of Collective Action on Hazardous Waste Facilities”, RAND Journal of Economics, Vol. 24, No. 1, Spring 1993, pp. 101-125. Hamilton, James T., “Testing For Environmental Racism: Prejudice, Profits, Political Power?”, Journal of Public Policy Analysis and Management”, Vol. 14, 1995, pp. 1 07-1 3 2. Harding, Sandra, "Sustainability, Experience and Knowledge: An Epistemology from/for Rainbow Coalition Politics", Development and Change, Vol. 23, No. 3, 1992, pp. 1 75-193. Harrison, David, Jr. and Daniel L. Rubinfeld, "The Distributional Benefits from Improvements in Urban Air Quality", Journal of Environmental Economics and Management, Vol. 5, 1978, pp. 313 - 332. Harrison, David, Jr., Who Pays for Clean Air: The Cost and Benefit Distribution of Automobile Emission Controls, Ballinger Publishing Company, Cambridge, MA, 1975. Hazardous Waste Sites and the Rural Poor: A Preliminary Assessment, Clean Sites Consortium, Kate Probst, Project Director, Alexandria, Va., March 1990. Heiman, Michael K., “Race, Waste, And Class: New Perspectives On Environmental Justice”, Antipode, Vol. 28, No. 2, 1996, pp. 111-121. Hird, John A., "Environmental Policy and Equity: The Case of Superfund", Journal of Policy Analysis and Management, Vol. 12, No. 2, 1993, pp. 323-343. Hurley, Andrew, “The Social Biases of Environmental Change in Gary, Indiana, 1945- 1980", Environmental Review, Vol. 12, No. 4, Winter 1988, pp. 1-19. Jones, S. C., “Inequities of Industrial Siting Addressed”, National Law Journal, Vol. 15, 1993,p.20. 178 Ketlar, Kusum, "Hazardous Waste Sites and Property Values in the State of New Jersey", Applied Economics, Volume 24, 1992, pp. 647 - 653. Kneese, Allen V. and Blair T. Bower, Environmental Quality Analysis, 1972. Knoepfel, P., "Distributional Issues in Regulatory Implementation: The Case of Air Quality Control Policies", Distributional Conflicts in Environmental-Resource Policy, Allan Schnaiberg, et. al., Editors, New York, New York, St. Martin's Press, 1986. Krieger, Martin H., "Six Propositions on the Poor and Pollution", Policy Sciences, Volume 1, 1970, pp. 311-324. Kruvant, William J ., "People, Energy and Pollution", American Energy Consumer, Dorothy K. Newman and Dawn Day, Editors, Ballinger Publishing Company, Cambridge, Massachusetts, 1975, pp. 125-167. Kutz, F. W., A. R. Yobs and S. C. Strassman, "Racial Stratification of Organochlorine Insecticide Residues in Human Adipose Tissue", Journal of Occupational Medicine, Volume 19, 1977, pp. 619-622. Lake, Robert W., “Volunteers, NIMBYs, And Environmental Justice: Dilemmas Of Democratic Practice”, Antipode, Vol. 28, No. 2, 1996, pp. 160-174. Landolt, Marsha L., Hafer R. Fritz, Ahmad Nevissi, Gerald van Belle, Kirk Van Ness, and Christopher Rockwell, "Potential Toxicant Exposure Among Consumers of Recreationally Caught Fish fiom Urban Embayments of Puget Sound", National Oceanic and Atmospheric Administration, National Ocean Service, Rockville, MD, NOAA Technical Memorandum NOS OMA 23, November 1985. Lavelle, Marrianne, and Marcia Coyle, "Unequal Protection: The Racial Divide in Environmental Law", The National Law Journal, September 21 , 1993. Lazarus, Richard J ., "Pursuing 'Environmental Justice: The Distributional Effects of Environmental Protection", Northwestern University Law Review, Volume 87, No. 3, 1993, pp. 787-857. Lee, Charles, "Evidence of Environmental Racism", Sojourners, February/March 1990. Lee, Charles, "The Integration of Justice", Sojourners, F ebruary/March 1990, pp. 22-25 179 McAllum, M., Recreational and Subsistence Catch and Consumption of Seafood fi'om Three Urban Industrial Bays of Puget Sound: Port Gardner, Elliot Bay, and Sinclair Inlet, Washington State Department of Health, Olympia, Washington, 1985. McCall, Julian, “Discriminatory Air Pollution: If Poor, Don’t’ Breathe”, Environment, Vol. 18, No. 2, March 1976, pp. 26 - 31. McMichael, A. J ., R. Spirtas, J. F. Gamble and RM. Tousey, "Mortality Among Rubber Workers: Relationship to Specific Jobs", Journal of Occupational Medicine, Volume 18, 1976, pp. 178-184. McWilliams, Douglas A., "Environmental Justice and Industrial Redevelopment: Economics and Equity in Urban Revitalization", Ecology Law Quarterly, Vol. 21 , No. 3, 1994, pp. 705 - 783. Michigan Natural Resources and Environmental Protection Act, 1994 PA. 451, as amended, Part 201, 299.601 et seq. and its Administrative Rules, R 299.5101, et seq. Mitchell, Carolyn M., “Environmental Racism: Race As A Primary Factor In The Selection Of Hazardous Waste Sites”, National Black Law Journal, Vol. 12, 1993, pp. 176-1 88. Mohai, Paul and Bunyan Bryant, "Environmental Racism: Reviewing the Evidence", in Race and the Incidence of Environmental Hazards, Bunyan Bryant and Paul Mohai, Editors; West View Press, Boulder Colorado, 1992. Mohai, Paul and Bunyan Bryant, "Environmental Injustice: Weighing Race And Class As Factors In The Distribution of Environmental Hazards", University of Colorado Law Review, Vol. 63, 1992, pp. 921-932. Mushak, Betty, "Focus: Environmental Equity: A New Coalition for Justice", Environmental Health Perspectives, Vol. 101, No. 6, November 1993. National Law Journal, "Unequal Protection: The Racial Divide in Environmental Law: A Special Investigation", September 21, 1992. Nieves, Leslie A. And Alvaro L. Nieves, Regional Dififerences in the Potential Exposures of U.S. Minority Populations to Hazardous Facilities, presented to the Annual Meeting of the Regional Science Association, Chicago, IL, November 15, 1992. Norusis, Marijsa, SPSS® 6.] Guide to Data Analysis, Prentice Hall, New Jersey; 1995. 180 Pfaff, Dennis, "Pollution and the Poor", Detroit News, November 26, 1989, pp. 1A, 14A and 15A. Perfecto, Ivette and Baldemar Valasquez, “Farm Workers Among the Least Protected”, EPA Journal, March / April 1992, p. 13. Perlin, Susan A., R. Woodrow Setzer, John Creason and Ken Sexton, “Distribution of Industrial Air Emissions By Income and Race in The United States: An Approach Using The Toxic Release Inventory”, Environmental Science & Technology, Vol. 29, No. 1, 1995, pp. 69-80. Peskin, Henry M., "Environmental Policy and the Distribution of Costs and Benefits", Current Issues in U.S. Environmental Policy, Paul R. Portney, Editor, Johns Hopkins University Press, Baltimore, Maryland, 1978. Puffer, Harold W., Stanley P. Azen, Marusia J. Duda, and David R. Young, Consumption Rates of Potentially Hazardous Marine Fish Caught in the Metropolitan Los Angeles Area, U.S. EPA, Washington, D.C., January 1983, U.S. Governmental Printing Office: 1983/659-095/573. Pulido, Laura, “A Critical Review Of The Methodology Of Environmental Racism Research”, Antipode, Vol. 28, No. 2, 1996, pp. 142-159. Rejeski, David, "Framework for Comprehensive Equity Analysis", U.S. EPA Environmental Equity Project, Washington, D.C., 1990. Rios, Richard, Gerald Poje, and Roger Detels, "Susceptibility to Environmental Pollutants Among Minorities", Toxicology and Industrial Health, Vol. 9, 1993, pp. 797-812. SR1, Seafood Consumption Data Analysis, U.S. EPA, Washington, D.C., 1980. Swanson, Samara F., "Race, Gender, Age, And Disproportionate Impact: What Can We Do About The Failure To Protect the Most Vulnerable?", F ordham Urban Law Journal, Vol. 21, 1994, pp. 577-604. Templet, Paul H., "The Emissions-J obs Ratio", Environmental Science and Technology, Volume 27, Number 5, 1993, pp. 810-812. Templet, Paul H. and Stephen F arber, "The Complirnentarity Between Environmental and Economic Risk: An Empirical Analysis", Ecological Economics, forthcoming 1993. 181 Tomboulian, Alice, Paul Tomboulian, Kurt Metzger, Doug Towns and Lawrence Hinds, T ri-County Detroit Area Environmental Equality Study, United Way Community Services, Detroit, December 1995. Torres, Gerald, "Race, Class, and Environmental Regulation: Introduction: Understanding Environmental Racism", University of Colorado Law Review, Vol. 63, No. 4, 1992, pp. 839 - 848. Toxic Waste And Race In The United States: A National Report on the Racial and Socio - Economic Characteristics of Communities with Hazardous Waste Sites, Commission for Racial Justice, United Church of Christ, New York: N. Y., Dr. Benjamin F. Chavis Jr., Executive Director and Charles Lee, Director Special Project on Toxic Injustice, Public Data Access, Inc., 1987. U.S. Bureau of Census, 1990 Census of Population, General Population Characteristics, Michigan, U.S. Department of Commerce, Economics and Statistics Administration, 1990. CP-1-24. U.S. Bureau of Census, 1990 Census of Population, General Population and Housing, Summary of Social, Economic and Housing Characteristics, Michigan, U.S. Department of Commerce, Economics and Statistics Administration, 1990, CPH- 5-24. U.S. Department of Health and Human Services, Health Status of Minorities and Low Income Groups (3d ed.), 1991. U.S. Department of Health and Human Services, Trends in Indian Health, 1991. U.S. Environmental Protection Agency, Environmental Equity: Reducing Risk For All Communities, 1992, Washington, D.C., Government Printing Office. U.S. Environmental Protection Agency, "Environmental Equity: Reducing Risk for All Communities", Vol. 1 and 2, Office of Policy, Planning and Evaluation, Washington, D.C., 1992. Urban Enviromnent Conference, Inc., Taking Back Our Health: An Institute on Surviving the Toxic Threat to Minority Communities, Washington D.C., General Accounting Office, 1983. U.S. General Accounting Office, Siting of Hazardous waste Landfills and Their Correlation with Racial and Economic Status of Surrounding Communities, Washington D.C., General Accounting Office, 1983. 182 Wemette, D. R. and Nieves, Leslie A., "Breathing Polluted Air; Minorities are Disproportionately Exposed", EPA Journal, No. 18, March/April, 1992, pp. 16- 1 7. West, Patrick C., J. Mark Fly, Frances Larkin, and Robert W. Marans, “Minority Anglers and Toxic Fish Consumption: Evidence from a Statewide Survey of Michigan”, Chapter 8 in Race and the Incidence of Environmental Hazards, Paul Mohai and Bunyan Bryant, Editors, Westview Press, 1992, pp. 100-113. Wood, G. Craig, “Risk of Hazardous Waste Exposure for Different Levels of Percent Minority Populations in Detroit and Vicinity”, unpublished paper, University of Michigan, School of Public Health, Department of Biostatistics, 1997. Zax, Jeffiey 8., Public Policy, Social Welfare, and the Incidence of Airborne Pollution in Genessee County, unpublished paper, February 3, 1997. Zimmerman, Rae, "Issues of Classification in Environmental Equity: How We Manage is How We Measure", Fordham Urban Law Journal, Vol. 21 , 1994, pp. 633-669. Zupan, Jeffrey M., "The Distribution of Air Quality in the New York Region", Johns Hopkins University Press, Baltimore, Maryland, 1973, pp. 2-5. 183 GENERAL REFERENCES GENERAL REFERENCES Alston, Dana, Editor, We Speak For Ourselves: Social Justice, Race & Environment, Washington, D.C., The Panos Institute, 1990. Andersen, Margaret L. and Patricia Hill Collins, Race, Class and Gender: An Anthology, Wadsworth, Inc. Press, 1992. Andrews, John, "Minorities & Toxics", Environmental Action, September/October 1987, p. 9. Askari, Emilia, "The Politics of Pollution", Detroit Free Press, June 5, 1994, pp. 1F and SF. Austin, Regina and Michael Schill, "Black, Brown, Poor & Poisoned: Minority Grassroots and Environmentalism and the Quest for Eco-Justice", The Kansas Journal of Law & Public Policy, Summer 1991, pp. 69-82. Basquet, C. R, et al. , "Socioeconomic Factors and Cancer Incidence Among Black and Whites", Journal of the National Cancer Institute, No. 83, pp. 551-557. Been, Vicki, “Compensated Siting Proposals: Is It Time To Pay Attention?”, Fordham Urban Law Journal, Vol. 21, 1994, pp. 787-826. Been, Vicki, “What’s Fairness Got To Do With It? Environmental Justice And The Siting Of Locally Undesirable Land Uses”, Cornell Law Review, Vol. 17, 1993, pp. 1001-1085. Benett, Graham, Dilemmas: Coping With Environmental Problems, Earthscan Publishers, 1992. Bernstein, Joan Z., "The Siting of Commercial Waste Facilities: An Evolution of Community Land Use Decisions”, The Kansas Journal of Law & Public Policy, Summer 1991, pp. 83-88. Boemer, Christopher and Thomas Lambert, Environmental Justice ?, Center for the Study of American Business, Policy Study Number 121, Washington University, St. Louis, April 1994, pp. 1-30. 185 Brewer, Rose M., “Theorizing Race, Class and Gender: The New Scholarship of Black Feminist Intellectuals and Black Women’s Labor”, in Theorizing Black Feminists: The Visionary Pragmatism of Black Women, Stanlie M. James and Abena P. A. Busia, Editors, Routledge Press, 1993. Brown, A. L., “Environmental Justice: New Civil Rights Frontier”, Trial, July, 1993, pp. 48-53. Brown, Michael H., Laying Waste: The Poisoning of America by Toxic Chemicals, Pantheon Books, New York, 1979. Brown, P. and S. Masterson-Allen, “The Toxic Waste Movement: A New Type of Activism”, Society and Natural Resources, Vol. 7, 1994, pp. 269-287. Bullard, Robert D., “Environmental Justice For All: It’s The Right Thing To Do”, Journal of Environmental Law and Litigation, Vol. 9, 1994, pp. 281-308. Bullard, Robert D. and Beverly H. Wright, "The Quest for Environmental Equity: Mobilizing the Afiican-American Community for Social Change", Society and Natural Resources, Vol. 3, October/December 1990, pp. 301-311. Bullard, Robert D. and Beverly H. Wright, "Toxic Waste and the Afiican American Community", Urban League Review, Summer/Winter 1989, pp. 67-75. Bullard, Robert D. and Beverly H. Wright, "Environmentalism and the Politics of Equity: Emerging Trends in the Black Community", Mid-American Review of Sociology, Winter 1987, Volume 12, No. 2, pp. 21-37. Bullard, Robert D. and Beverly H. Wright, "Blacks and the Environment", Humboldt Journal of Social Relations, 1987, Vol. 14, pp. 165 - 184. Bullard, Robert D. and Beverly H. Wright, "The Politics of Pollution: Implications for the Black Community", Phylon, 1986, Vol. 41, No. 1, pp. 71-78. Capek, S. M., "Environmental Justice, Regulation and the Local Community", International Journal of Health Services, Vol. 22, pp. 729-746. Carter-Pokras, Olivia, James Pirkle, Gilberto Chaves, and Elaine Gunter, "Blood Levels in 4 to 11 Year-Old Mexican American, Puerto Rican and Cuban Children", Public Health Reports, Volume 105, 1990, pp. 388-391. 186 Chase, A. R., “Assessing and Addressing Problems Posed By Environmental Racism, Rutgers Law Review, Vol. 45, 1993, pp. 335-369. Chavis, Benjamin F ., Jr., "Essay: Race, Justice and the Environment", Nature Conservancy, September/October 1992, p. 38. Cole, Luke W., "Empowerment as the Key to Environmental Protection: The Need for Environmental Poverty Law, Ecology Law Quarterly, Vol. 19, No. 4, 1992, pp. 619 - 683. Coplan, Karl S., "Protecting Minority Communities with Environmental Civil Rights Claims", New York Law Journal, Vol. 206, No. 36, August 20, 1991, pp. 1-2, 5. Council on Environmental Quality, The Second Annual Report of the Council on Environmental Quality, Washington D.C., U.S. Governmental Printing Office, 1971. Cross, Frank B., "When Environmental Regulations Kill: The Role of Health / Health Analysis, Ecology Law Quarterly, Vol. 22, No. 3, 1995, pp. 729-784. Davies, John E., et al., "The Role of Social Class in Human Pesticide Pollution", American Journal of Epidemiology, Volume 96, 1972, pp. 334-338. Davis, Morris E., "The Impact of Workplace Health and Safety on Black Workers: Assessment and Prognosis", Labor Law Journal, Volume 31, 1980, pp. 723-724. Dorfrnan, Nancy S. And Arthur Snow, “Who Will Pay For Pollution Control? - The Distribution by Income Of The Burden Of The National Environmental Protection Program, 1972-1980, National Tax Journal, Vol. 28, No. 1, 1983, pp. 101-115. Environmental Forum, "ELI Report: Panel Outlines Strategies to Fight Racial Injustice", Jan/Feb. 1994, pp. 53 - 55. Environment Reporter, The Bureau of National Affairs, Inc., Washington, D.C., 1993. Ferris, Deeohn, “A Challenge To EPA: An Environmental Justice Office Is Needed”, EPA Journal, March/April 1992, pp. 18-27. Ferris, Deeohn, “Communities of Color And Hazardous Waste Cleanup: Expanding Public Participation In The Federal Superfund Program”, F ordham Urban Law Journal, Vol. 21, 1994, pp. 671-687. 187 Frankenberg, Ruth, The Social Construction of Whiteness, White Women, Race Matters, University of Minnesota Press, 1993. Freeman, James S. and Rachel D. Godsil, “The Question of Risk: Incorporating Community Perception Into Environmental Risk Assessments”, F ordham Urban Law Journal, Vol. 21, 1994, pp. 547-575. " From Toxic Racism to Environmental Justice", E, The Environmental Magazine, May/June 1992. Gauna, Eileen, "Federal Environmental Citizen Provisions: Obstacles and Incentives on the Road to Environmental Justice”, Ecology Law Quarterly, Vol. 22, No. 1, 1995, pp. 1-87. Gould, J. M., Quality of Life in American Neighborhoods: Levels of Afiluence, Toxic Waste, and Cancer Mortality in Residential Zip Code Areas, Boulder, Westview Press, 1 986, Greenburg, Michael R. and Richard F. Anderson, Hazardous Waste Sites: The Credibility Gap, Center for Urban Policy Research, New Brunswick, New Jersey, 1984. Greenburg, Michael R., "Social and Economic Effects of Hazardous Waste Management Sites", Hazardous Waste, Vol. 1, 1984, pp. 287-396. Greenburg, Michael R. and Dona Schneider, Environmentally Devastated Neighborhoods: Perceptions, Policies and Realities, Rutgers University Press, New Brunswick, New Jersey, 1996. Greer, Marsha L. and Anna K. Harding, "The Health Impact of Hazardous Waste Sites on Minority Communities: Implications for Public Health and Environmental Professionals", Journal of Environmental Health, Vol. 55, No. 7, 1993, pp. 6-9. Grossman, Karl, "From Toxic Racism to Environmental Justice", E Magazine, 1992, Vol. 3, No. 8, pp. 28-35. Harris, Cynthia H. and Robert C. Williams, "Research Directions: The Public Health Service Looks to Minorities", EPA Journal, Vol. 18, March/April 1992, pp. 40- 41. Hinds, C., et. al., "From the Front Lines of the Movement for Environmental Justice", Social Policy, Vol. 22, No. 4, 1992, pp. 12-22. 188 Holtzrnan, Elizabeth, Smokescreen: How the Department 0 Sanitation's Solid Waste Plan and Environmental Impact Statement Cover Up the Poisonous Health Eflects of Burning Garbage, Office of the Comptroller, New York, New York, June 1992. Hoover, Kenneth R., The Elements of Social Scientific Thinking, Fifth Edition, St. Martin's Press, New York, 1992. Israel, Brian D., “An Environmental Justice Critique of Risk Assessment”, New York University Environmental Law Journal, Vol. 3, 1994, pp. 469-522. Jarris, Cynthia A. And Robert C. Williams, “Research Directions: The Public Health Service Looks at Hazards to Minorities”, EPA Journal, March/April 1992, pp. 40- a 41. E Johnson, James H. and Melvin L. Oliver, "Blacks and the Toxic Crisis", Western Journal 1 of Black Studies, Vol. 13, No. 2, 1989, pp. 72-78. ' 1 Kay, Jane, "Minorities Bear the Brunt of Pollution, San Francisco Examiner, April 7, 1991. Keeva, Steven, "A Breath of Justice", ABA Journal, February 1994. Kish, Leslie, Survey Sampling, Wiley Publishing, New York, 1965. LaBalme, J ., A Road To Walk: A Struggle for Environmental Justice, Durham: Regulator Press, 1989. Lapan, Nora and Karen Hoffman, Editors, Occupational Disease Among Minority Workers: An Annotated Bibliography, National Institute for Occupational Safety and Health, Cleveland, Ohio, 1981. Lavelle, Manianne, "EPA Enforcement to be Probed", The National Law Journal, April 5, 1993, pp. 3 and 34. Lloyd, J. William, Frank E. Lundin, Jr., Carol K. Redomond, and Patricia B. Geiser, "Long-term Mortality Study of Steelworkers, Journal of Occupational Medicine, Volume 12, 1970, pp. 151-157. Lorber, Judith and Susan A. Farrell, The Social Construction of Gender, Sage Publications, 1991. Love, Sam, "Ecology and Social Justice: Is There A Conflict?", Environmental Action, No. 4, August 5, 1972, pp. 3-6. 189 Lucas, Robert E. B., "The Distribution of Job Characteristics", Review of Economics and Statistics, Volume 56, 1974, pp. 530-533. Mann, Eric, L. A. 's Lethal Air: New Strategies for Policy, Organizing, and Action, Labor /Community Watchdog, Los Angeles, California, 1991. McDermott, Charles, "Balancing The Scales of Environmental Justice", F ordham Urban Law Journal, No. 21, No. 3, Part 2, Spring 1994, pp. 689-705. Mitchell, Robert Cameron and Richard T. Carson, “Property Rights, Protest, and the Siting of Hazardous Waste Facilities”, American Economic Review, Vol. 76, No. 2, May 1986, pp. 285-290. Mohai, Paul, "Black Environmentalism", Social Science Quarterly, Vol. 71, No. 4, 1990, pp. 745-765. Mohai, Paul, "Public Concern and Elite Involvement in Environmental - Conservation Issues", Social Science Quarterly, Vol. 66, 1985, pp. 820 - 838. Mohai, Paul and Bunyan Bryant, "Race, Poverty, and Environment: The Disadvantage Face Greater Ris ", EPA Journal, March/April, 1992. Moore, Marjorie W., "Environmental Health and Community Action", New State Bar Association Environmental Law Journal, F ebruary/May 1991., p. 12. Morell, David, "Siting and the Politics of Equity", Hazardous Waste, No. 1, 1980, pp. 55 -59. Morell, David and Christopher Magorian, "Risk, Fear and Local Opposition: 'Not in My Back Yard'", in Siting Hazardous Waste Facilities: Local Opposition and the Myth of Pre-emption, 21 , 1982. Morrison, Denton E. and Riley E. Dunlap, "Environmentalism and Elitism: A Conceptual and Empirical Analysis", Environmental Management, Vol. 10, No. 5, 1986, pp. 581 - 589. National Public Radio, Morning Edition, transcripts from the "Series on Environmentalism in Minority Communities", transcripts of broadcasts fiom September 9, 10 and 11, 1993, Phillip Davis reporting. Navarro, V., "Race or Class v. Race and Class: Mortality Differentials in the U.S.", The Lancet, Vol. 336, 1989/90, pp. 1238 - 1240. 190 Newton, Kathy Cone with Francis Ortega, foreword by Robert D. Bullard, "Beyond Ankle-Biting: Fighting Environmental Discrimination Locally, Nationally, and Globally", The Workbook, Vol. 16, No. 3, Fall 199. Nelson, A. C., et. al., "Price Effects of Landfills on Residential Land Values", Journal of Urban Planning and Development, Vol. 118,, No. 9, 1992, pp 128 - 137. O’Connor, John, “The Promise of Environmental Democracy”, Chapter 4 in Toxic Struggles: The Theory and Practice of Environmental Justice, Richard Nofrichter, ed., New Society, Philadelphia, PA, pp. 47-57. Pfaff, Dennis, "Minorities Speak Out More About Pollution", Detroit News, November 26, 1989, pp. 15A. Pointdexter, Georgette C., "Addressing Morality in Urban Brownfield Redevelopment: Using Stakeholder Theory to Craft Legal Process", Virginia Environmental Law Journal, Vol. 15, No. 1, Fall 1995, pp. 37-76. Pollack Sue and Joann Grozuczak, Reagan, T oxics and Minorities, 1984. Portney, Kent E., “Environmental Justice And Sustainability: Is There A Critical Nexus In The Case Of Waste Disposal Or treatment Facility Siting?”, F ordham Urban Law Journal, Vol. 21, 1994, pp. 827-839. Portney, Kent E., Siting Hazardous Waste Treatment Facilities: The Nimby Syndrome, New York: Auburn House, 1991. Principles of Environmental Justice, The First People of Color Environmental Leadership Summit, October 24-27, 1991, Washington D.C., pp. 1-2. Reich, Peter L., “Greening the Ghetto: A Theory of Environmental Race Discrimination”, The University of Kansas Law Review, Vol. 41, 1992, pp. 271-314. Reilly, William K., “Environmental Equity: EPA’s Position: Protection Should Be Applied Fairly”, EPA Journal, March/April 1992, pp. 18-22. Roberts, Sam, "In My Backyard? Where New York City Puts its Problems", The New York Times, December 6, 1993, p. 54. Robinson, James C., "Exposure to Occupational Hazards Among Hispanics, Blacks and Non-Hispanic Whites in California", American Journal of Public Health, Volume 79, 1989, pp. 629-630. 191 Robinson, James C., "Racial Inequity and the Probability of Occupation-related Injury or Illness", Milbank Quarterly, Volume 62, 1984, pp. 567-588. Robinson, James C., "Trends in Racial Inequity and Exposure to Work-related Hazards 1968-1986", Milbank Quarterly, Volume 65, 1987, pp. 404-412. Russell, Dick, "Environmental Racism: Minority Communities and Their Battle Against Toxics", Amicus Journal, No. 11, Spring 1989, pp. 22-32. Singleton, Royce A., Jr., Bruce C. Straits and Margaret Miller Straits, Approaches to Social Research, Second Edition, Oxford University Press, New York and Oxford, 1993. Smith, James N., ed, Environmental Quality and Social Justice in Urban America, Washington: Conservation Foundation, 1974. Stafford, H. A., "Environmental Protection and Industrial Location", Annals of the Association of American Geographers, Vol. 75, No. 22, 1985, pp. 227 - 240. State Bar of Michigan, Environmental Law Section, Environmental Justice, 1994 Fall Update, October 29, 1994 (unpublished). Taylor, Dorceta E., "Blacks and the Environment: Toward an Explanation of the Concern and Action Gap Between Blacks and Whites", Environment and Behavior, Voltune 21, March 1989, pp. 175-204. Taylor, Dorceta E., “The Environmental Justice Movement: Predominately White Memberships Are Not Defensible”, EPA Journal, March/Ade 1992, pp. 25-27. Tomain, Joseph P., “The Distributional Consequences of Environmental Regulation: Economics, Politics, and Environmental Policymaking”, The Kansas Journal of Law & Public Policy, Summer 1991, pp. 102-112. Torres, Gerald, "Environmental Burdens and Democratic Justice", F ordham Urban Law Journal, Vol. 21, 1994, pp. 431-460. Unger, Donald G., Abraham Wandersman and William Hallrnan, "Living Near a Hazardous Waste Facility: Coping with Individual and Family Distress", American Journal of Orthopsychiatry, 55:62, 1992. Vaughan, Eliane and Brenda Nordenstam, "The Perception of Environmental Risk Among Etlmically Diverse Groups", Journal of Cross - Cultural Psychology, Vol. 22, No. 1,1991, pp. 29 - 60. 192 Vaias, Emily, "The Environmental Justice Movement: Discriminatory Intent of NIMBYism and LULUs at Work?", American Planning Association and the Law Division Newsletter, 1994, p. 3. Wenz, Peter S., Environmental Justice, Albany: SUNY Press, 1988. White, Harvey L., “Hazardous Waste Incinerators and Minority Communities”, Chapter 10 in Race and the Incidence of Environmental Hazards, Bryant and Mohai, eds., Westview Press, 1992, pp. 126-139. Willard, W. "Environmental Racism: The Merging of Civil Rights and Environmental Activism", Southern University Law Review, Vol 31, Fall 1992 / Spring 1993. Wright, R. G., "Hazardous Waste Disposal and the Problems of Stigrnatic and Racial Injury", Arizona State Law Journal, Vol. 23, 1991, pp. 777 - 800. Zimmerman, Klaus, "Distributional Considerations and the Environmental Policy Process", Distributional Conflicts in Environmental-Resource Policy, Allan Schnaiberg, et. al., Editors, New York, New York, St. Martin's Press, 1986. Zwerdling, Daniel, "Poverty and Pollution", Progressive, Volume 37, January 1973, pp. 25-29. 193 MICHIGnN smrE UNIV. LIBRARIES 111111111111111111111111111111111111 31293017141353