Masters Project Report
On
Analyzing Environmental Justice in the Dallas/Forth Worth Area
Advisor: Dr. Ronald Briggs
Acknowledgement: Dr. Michael Tiefelsdorf
Course: Masters Project (GISC 6389)
Prepared by
Rumana Reaz Arifin
Fall 2007
Table of Contents
1. Introduction / 42. Project Objective / 5
3. Literature Review / 6
3.1 General Studies of Environmental Justice / 6
3.2 Methodological Issues / 8
3.2.1 Variable Selection / 8
3.2.2 Classification of Pollutant Zones / 9
3.2.3 Analysis in Different Geographic Unit / 10
4. Project Hypotheses / 11
5. Data and Data Sources / 12
5.1 Spatial Data / 12
5.2 Attribute Data / 13
6. Analysis and Methodology / 14
6.1 Classifying Zones for Pollutant Exposure / 15
6.1.1 Standard Deviational Elliposoidal Distribution of Pollutant Sites / 16
6.1.2 Classification of Pollutant Zones / 17
6.1.2.1 Spatial Join / 17
6.1.2.2 Hot Spot Analysis / 18
6.1.2.3 Results of Classifying Pollutant Zones / 22
6.2 Analysis of Socio-Economic Variables / 22
6.2.1 Analysis of Variance (ANOVA) / 23
6.2.2 Kruskal-Wallis Chi-Squared Test / 23
6.2.3 Pearson’s Product Moment Correlation Coefficient / 24
6.2.4 Multinomial Regression / 24
7. Results and Discussion / 25
7.1 Results of Graphical Exploration / 25
7.1.1 Population of Racial Group as % of Total Population / 25
7.1.2 Percentage of Geographic Unite where Minority Ratio > 1 / 26
7.1.3 Differences in Income Level / 27
7.1.4 Percentage of Population Below and Above Poverty Level / 28
7.1.5 Percentage of Population at Different Education Level / 29
7.1.6 Summary of Graphical Exploration / 30
7.2 Results of Statistical Analysis / 30
7.2.1 Results of ANOVA and Kruskal-Wallis Chi-Squared Test / 30
7.2.2 Results of Correlation Matrix / 34
7.2.3 Results of Multinomial Regression / 38
8. Conclusion / 40
8.1 Future Work / 41
9. References / 41
List of Figures and Tables
Figures- Standard Deviational Ellipsoidal Distribution of Pollutant Sites at Census Tract Level
- Standard Deviational Ellipsoidal Distribution of Pollutant Sites at Zipcode Level
- Clean Tracts
- Result of Hot Spot Analysis
- Zoomed In View of Hot Spot Analysis Overlaid with Kernel Density
- Kernel Density Analysis
- Classified Pollutant Zones at Census Tract Level
- Classified Pollutant Zones at Zipcode Level
- Racial Comparison at Census Tract
- Racial Comparison at Zipcode
- Minority Ratio Comparison at Census Tract
- Minority Ratio Comparison at Zipcode
- Comparison of Income Level at Census Tract
- Comparison of Income Level at Zipcode
- Comparison of Poverty Level at Census Tract
- Comparison of Poverty Level at Zipcode
- Comparison of Education Level Attained at Census Tract
- Comparison of Education Level Attained at Zipcode
Tables
- Some Selected Studies of Racial and Income Disparities in the distribution of Environmental Hazards (1967-1993).
- Result of ANOVA at Census Tract
- Result of Kruskal-Wallis Chi-Squared Test at Census Tract
- Result of ANOVA at Zipcode
- Result of Kruskal-Wallis Chi-Squared Test at Zipcode
- Pearson’s Product Moment Correlation Matrix for Census Tract
- Pearson’s Product Moment Correlation Matrix for Zipcode
- Summary of Multinomial Regression Result
1. Introduction
The issue of equal and fair distribution of pollution hazards and their impact on surroundingsis a well exercised topic of debate. Basically, when the disproportionate impact of environmental hazards relates with people of color and economically –disadvantaged groups, it leads to the formation of environmental justice issue. Environmental justice in general can be depicted as the right of having equal justice for any environmental phenomena without discrimination based on race and socioeconomic status. A book definition can be like this, “pursuit of equal justice and equal protection under the law for all environmental statutes and regulations without discrimination based on race, ethnicity, and /or socioeconomic status. This concept applies to governmental actions at all levels—local, state and federal—as well as private industry activities.” (Frokenbrok and Sheeley 2004). But the reality is different from what is expected. It has been historically observed that minority group populations (for example, populations of specific races such as African Americans, Hispanics, etc, and/or economically disadvantaged groups) live in closer proximity to pollutant sites compared to the Anglo population and/or populations with higher income levels. Hence it can be said that environmental justice is violated in the proximity of pollutant sites. There are a number of examples supporting this burning issue. A study based on modeling total air toxic concentration in 44 large metropolitan areas reveals that, in large metropolitan area of USA, blacks are more likely than whites to be living in census tracts with higher total modeled air toxic concentrations (Lopez 2002). Location of pollutant sites plays a key role in assessing the environmental justice issue as it has been seen that the people of color and/or low income group tend to live near the toxic sites or in other words, toxic sites are located where people of specific race and income groups are residing. This is again a issue of debate that whether the fact is true or not but a general assumption is like that, owner of pollutant sources selects such location for the pollutant site where there are more minority groups as they will not go for pushing that site elsewhere. Mohai and Bryant in a study suggest that, the minority population is subjected to disproportionate pollution exposure because of their lesser ability to defend themselves due to poverty and political powerlessness (Mohai and Bryant 1992). So anyhow, there is a connection between the location of pollutant sites and people living near it and in most cases those people are found to be people of color and/or of low income group. Here, GIS plays a dominant role in assessing the pollution exposure in surrounding population. To asses the environmental justice issue it is critical to determine the location of pollutant sites which are hazard prone and also the population group living in the proximity of hazard prone area needs to be identified correctly.So, classifying pollutant site locations is a major issue and where GIS can play a vital role. There are several indexes of determining nearness to location of pollutant sites like zoning and proximity indices (Huang and Batterman 2000). However, this study has taken approach to asses the pollution exposure based on new method called Hot spot analysis which will be discussed in detail later on.
2. Project Objective
Based on the discussion above my research objective is set to analyze the environmental justice issue in the five counties of the Dallas/Forth Worth region to determine if people in specific racial and socio-economic groups are differentially exposed to environmental pollution. This region has a large metropolitan area also with sub-urban and rural area. There are a number of pollutant site locations in this region and also a good number of people of color living in this region. Hence, my research will try to determine if there exists any relation between them. As mentioned above the project will show a new method for classifying pollutant sites and after the literature review more specific hypotheses will have been set to accomplish the project objective.
3. Literature Review
As the environmental justice is a well-exercised issue, extensive literature exists on this topic. Hence, the literature review here will mainly focus on two major aspects:
- General studies of environmental justice
- Methodological issues
The relevant discussions from different articles on the topics stated above are reviewed as follow:
3.1 General studies of Environmental Justice
Historically environmental justice issues are related to location of pollutant zones and pollution exposure. Many researches have found a connection between poverty, minority racial status and exposure to environmental toxicity. Below a chronological review of a number of literatures (1967-1993) addressing pollution in correlation with race and income is shown (White 1998):
Table1:Some Selected Studies of Racial and Income Disparities in the Distribution
of Environmental Hazards(1967—1993).
Year / Author / Type of Hazards / Geographic Focus / DisparityRace / Income
1967 / Hoffman et.al. / Pesticides / Chicago, Ill. / Yes
1972 / Freeman / Air pollution / Kansas City/
St. Louis/D.C. / Yes / Yes
1977 / Berry et al. / Pollution/pesticides, etc. / Urban areas / Yes / Yes
1978 / Asch and Seneca / Air pollution / Urban areas / Yes / Yes
1983 / U.S. GAO / Hazardous waste / Southeast / Yes
1987 / UCC and PDA / Hazardous waste / National / Yes / Yes
1989 / Belliveau et al. / Toxic releases / Richmond, Calif. / Yes / Yes
1991 / Conner and Thornton / Hazardous waste / National / Yes / Yes
1992 / Goldman / Toxic air/waste / National / Yes / No
1992 / Mohai and Bryant / Hazardous waste Toxic / Detroit, Michigan / Yes / Yes
1993 / Bowen et al. / Toxic releases / Ohio / Yes / No
1993 / Hamilton / Hazardous waste siting / National / Yes / Yes
1993 / Zimmerman / Hazardous waste / National / Yes / No
Source: Derived from Benjamin A. Goldman, Not Just Prosperity: Achieving Sustainability with Environmental Justice (Washington, D.C.: National Wildlife Federation, 1993).
The review shows that most of the research have a found a correlation between pollution exposure and race. So, comprehensively race is dominant factor in analyzing environmental justice issue. Berry et al.(1977) and Asch and Senneca (1978) conducted study for urban areas all over USA and found both race and income as dominant factor. The first national level study conducted by UCC and PDA in 1987 found disparity for both race and income. This study reveals that 60% of African-Americans live in communities with one or more abandoned toxic waste sites (Hockman and Morris 1998). However another national level study conducted by Goldman in 1992 found the correlation for race but not for income. Similarly Hamilton (!993) found adverse effect for both race and income whereas in the same year Zimmerman (1993) found evidence only for race. Results are subject to variation as different methodology in assessing pollution exposure has been used.
However, some recent studies continue to show the evidence for the violation of environmental justice in hazard prone areas. A study conducted in Michigan in 1998 found that, minority group in terms of race and low income are more exposed to environmental pollution. African American population is more exposed compared to White population. Racial discrimination shows stronger association compared to income (Hockman and Morris 1998). In Southern California Morello-Frosch et al. in 2002found that census tracts with a pollutant site or within 1 mile distance of a pollutant site have significantly higher percentage of residents of color (particularly Hispanic), lower per capita and household income. The results remained consistent even when the percentages of African American and Hispanic residents were evaluated as separate groupings (Morello-Frosch et al. 2002). A recent national level study by Mohai and Saha reveals that over 40% of population living within 1 mile of hazardous waste TSDFs (TRI facilities and treatment, storage and disposal facilities) were people of color (using distance-based method) (Mohai and Saha 2007). Hence, the evidence for racial discrimination continues up to present time and sometimes income disparity is also evident. Hence, this research will do the analysis to determine the racial and socio-economical discrimination in D/FW area.
3.2. Methodological Issues
Several methodological issues have been reviewed from previous articles to get the proper guidance about the analysis. The prime issues are discussed below:
- Variable selection
- Classification of Pollutant Zones
- Analysis in different Geographic units.
3.2.1.Variable Selection
To do the analysis selection of variable is an important criterion. The result depends much on which and how much relevant variables are analyzed. From the review of general studies on environmental justice it is evident demographic variables of race are major variables for analysis. Along with that socio-economic variables for determining income disparity and other social status are also analyzed. Variables for measure of education is also an important issue as it sometimes assumed that people who are living near the pollutant sites are less solvent and being less solvent not reaching towards higher education level and so becoming less conscious about the adverse affect of pollution exposure. The studies conducted upon environmental justice usually analyze different socio-economic variables. Examples from the literature review are as follow: (Bowen et al. 1995) and (Clutter et al. 1995).
Population density
Minority (Black and/or Hispanic) proportion of the total population
Minority Ratio (Black/White and/or Hispanic/White)
Median household income
Poverty Status
Median value of owner-occupied housing
Education level attained
- % of at least High School level education
- % of College Degree
% of Employment and Unemployment
Based on this review the socio-economic variables for my project have been selected.
3.2.2Classification of Pollutant Zones
As mentioned earlier, assessing pollution exposure is a major concern and results vary depending on the methodology adopted for classifying pollutant sites for different exposure level. Several methods are used for classifying the pollutant sites based on their location. However two common indicators are Zonal and Proximity indicators (Huang and Batterman 2002).
Zonal indicators: which classify locations using ‘zones’ with the assumption that individuals within each zone will have the same exposure level.
Proximity indicators: which determine the distance from the pollutant source with the assumption that exposure to pollution increases as the distance decreases.
These two approaches may be combined for classifying the pollutant zones. For example, the population in a polluted area may be identified by proximity to sources and subsequently divided in to subgroups using zones. Alternatively high pollution areas, may be identified by the location of pollution sources and subgroups within this areas demarcated by distance to the source. Considering the location of a pollutant site, zoning can be done based on the count and adjacency of site locations.
So, considering the location of a pollutant site, zoning can be done based on the count and proximity of site locations. A study conducted by Bowen et al. had taken an approach to classify the census tracts of Cleveland, Ohio combining zonal and proximity indicators (Bowen et al. 1995)
- Clean Tract: No TRI site.
- Potentially Exposed Tract: In proximity of a TRI site.
- Dirty Tract: Have one or more sites in the tract
In my study, I adopted similar approach and identified pollutant zones based on site density and distance to pollutant sites using Hot Spot analysis.
3.2.3Analysis in Different Geographic Unit
Another major reason for varying result of different studies of environmental justice is selected geographic unit for analysis. In general it is assumed that the smaller the area coverage, the more variability is induced and results are more dependable. For larger geographic area, variability decreases due to averaging of variables values. But after reviewing the literature, it is found that there is no anonymous choice about geographic unit. Different researchers have used different geographic units and had spoken in favor of it. Arguments have been advanced in favor of both Zipcodes and Census tracts. Clutter, Holm and Clarke suggest Census tracts and Block groups may provide better approximation of neighborhood because of their smaller aerial coverage, compared to Zipcodes (Clutter et al. 1995). However, Hockman and Morris argue that, neighboring Census tracts are too similar in socio-economic characteristics and suggest Zip codes as a good compromise between Counties and Census tracts (Hockman and Morris 1998).
Hence, my approach is to do the analysis for both geographic units and compare results.
- Project Hypotheses
Based on the literature review, the following hypotheses are made, to specify the project objective more clearly:
Hispanic and African American populations will be positively correlated with high pollution exposure.
Low income groups will be positively correlated with high pollution exposure.
Lower education levels will be positively correlated with high pollution exposure.
Minority (racial) population will have a positive correlation with lower income and lower education level.
Results for Census tract should produce stronger association than those of Zipcode because, as the geographic unit gets larger, less variability is induced due to averaging of variables.
Hence, the aim of the analysis will be to determine how far the assumed hypotheses can be proved.
- Data and Data Sources
5.1. Spatial Data
Counties: The project is carried out at five counties in Dallas/Forth Worth region. The counties are:
- Dallas
- Denton
- Collin
- Rockwall
- Tarrant
Total Census tract count within this region: 945
Total Zipcode tabulation area count within this region: 225
As mentioned earlier, the project is carried out at both Census Tract and Zipcode level.
Pollutant Sites: Five types of Pollutant sites are included for analysis. They are as follow:
- TRI Sites
Tri sites are all Toxic Release Inventory sites in Texas. TRI contains information about more than 650 toxic chemicals that are being used, manufactured, treated, transported, or released into the environment. The locations of such activities are marked as TRI sites.
- Solid Waste Sites
Registered landfills and associated municipal solid waste (MSW) facilities for Texas. The dataset contains both closed and open landfills.
- Water Pollution Sites
The waprt pollution sites include all permitted municipal and industrial wastewater quality outfalls.
- Hazardous Sites
These are all permitted industrial and hazardous waste locations in Texas. Facilities which store, process, or dispose of hazardous waste are listed as hazardous sites
- Superfund Sites
These are all superfund clean-up sites in Texas.
Total 999 sites in Census tract area
Total 1078 sites in Zipcode tabulation area
The reason for varying number of sites for Census tract and Zipcode as Zipcode tabulation areas in some cases extend slightly beyond County boundaries.
Data Source
- NCTCOG:
- Texas Commission on Environmental Quality:
- Environmental Protection Agency:
5.2Attribute Data
Socio-Economic Variables: Based on the literature review the following socio-economic variables are selected for analysis of environmental justice issues for race, income and education level.
Measures for Minority Population groups:
- White
- African American
- Hispanic
- Others (Asian, Alaskan Native, Pacific Islander)
Percentage of total population