Jump to main content.

Guidelines for Conducting Environmental Justice Analyses

Section 2

2.0 Guidelines for Conducting Environmental Justice Analyses

2.1 Environmental Justice Definitions and Data Sources
2.2 Procedure

2.2.1  Develop geographic boundaries for Community of Concern (COC) and conduct a preliminary burden analysis
2.2.2  Compare COC demographics to a statistical reference area
2.2.3  Determine if demographic criteria are met
2.2.4  Develop a comprehensive environmental load profile
2.2.5  Assess whether the burden is disproportionately high and adverse
2.2.6  Summarize and report results

2.3 Additional Factors to be Considered

Back to Table of Contents  


2.0 Guidelines for Conducting Environmental Justice Analyses

The Guidelines for Conducting Environmental Justice Analyses provide guidance and procedures, and identify sources of data for conducting EJ analyses to evaluate if a community is an EJ community. It is divided into the following major sections: Environmental Justice Definitions and Data Sources; Procedure; and Additional Factors. The EJ Factors and Data Sources section addresses definitions and data availability. The Procedure section discusses the methodology for evaluating EJ community and associated analytical tools. The Additional Factors section highlights some unique circumstances and alternatives for addressing such situations.

Further, these guidelines provide methodologies for developing an environmental load profile (ELP) to represent burden. The approach is to incorporate contributing elements into the load profile where there is a defensible method or data to provide a quantifiable estimate of an element's contribution to burden. The Region intends to add categories to the load profile as analytical methods and consistent data sets become available. Section 2.2.5 presents the criteria for determining whether the Community of Concern (COC) is a disproportionately high and adversely burdened community.

2.1 Environmental Justice Definitions and Data Sources
2.1.1. Definitions

Minority Community or Population

EPA's Office of Environmental Justice has defined the term "minority" for EJ purposes to include Hispanics, Asian-Americans and Pacific Islanders, African-Americans, and American Indians and Alaskan Natives. For EJ purposes, the term 'minority' does NOT address religion or people who might be distinguished by sex, age, culture, sexual orientation, or any type of handicap. Section 2.2.1 presents the statistical criterion for determining whether the COC is a minority community for EJ purposes under this Interim Policy.

Low-income Community or Population

The U.S. Census Bureau does not provide a specific definition for "low-income." Rather, the term is used interchangeably with "poverty." In this regard, the Census Bureau established a set of income cutoffs/thresholds to determine the poverty status of families. Those poverty thresholds are based on family size and the number of family members under 18 years old. Further, these groups were differentiated by age of the family householder. In addition, the thresholds for a one-person family (unrelated individual) and two-person family were further differentiated by the number of family members 65 years of age and older. The Census determines poverty by comparing the total income of each family against its corresponding threshold. If the total family income is less than the corresponding cutoff, the family is classified as "below the poverty level." Section 2.2.2 of this Interim Policy presents the statistical criterion for determining whether the COC is a low income community for EJ purposes under this IP. Until the 2000 Census data becomes available, the Region will utilize the 1990 Census data for purposes of determining whether a COC is a low-income and/or minority community.

Disproportionately High and Adverse Burden

The environmental burden or impact can be related to ambient conditions, a specific source or sources, and/or cumulative or area-wide sources. This burden can affect human health, as well as the ecological health of the natural environment. Identifying the magnitude of environmental burden, however, is not a simple process. Whereas high quality and consistent data are available for the development of the required low-income and minority demographic profiles, there currently exists limited data available for assessing the environmental burden.

Back to Top

2.1.2 Data Sources

This section discusses available data sets that will be used for conducting this segment of the EJ analysis.

Low-income and Minority Environmental Justice Demographic Data
Region 2 has developed a Geographic Information System (GIS) layer of the 1990 Census data. Moreover, Region 2 has developed a GIS application that uses the Census data to conduct the demographic portion of the EJ analysis. In mapping the data, it is important to properly define the boundaries of the COC to run the demographic analysis. It is the Region's intent to develop such boundaries with input from the community. Once the boundaries are established, they can be drawn onto the GIS application. Next, the application calculates the percent minority and low-income based on the Census data. Lastly, the application compares the COC to an appropriate statistical reference area as discussed in Section 2.2.2.

Disproportionate Burden Data
The Guidelines advance the concept of an environmental load profile. The profile provides a representation of the environmental burden in the community. It is based on salient characteristics that serve as indicators of environmental burden and provide a consistent basis for comparison. The profile of the community of concern is compared to that of the statistical reference area and the salient characteristics (e.g., indicators of air quality, drinking water, etc.) are used to assess whether the COC is an EJ community. Information for the environmental load profile analysis may be generated from the following available data sources:

Exposure Data
Information on exposure may be found, at least generally or indirectly, in some commonly available data bases. There is a large degree of facility (source) information available for EJ analysis within the following EPA mainframe databases:

Health Data
In addition, we will consider the insights that available health data sources may provide toward the environmental load profile. Data sources that should prove useful include:

Back to Top

2.2 Procedure

The following six (6) steps comprise the procedure to identify potential EJ communities and, further, actual EJ communities:

  1. delineate the boundaries of the COC and conduct, as appropriate, a preliminary environmental burden analysis6;
  2. compare the demographics of the community to an appropriate statistical reference;
  3. determine whether the community is either minority or low income;
  4. develop a comprehensive environmental load profile (ELP) for any community that is either minority or low income;
  5. assess whether the burden is disproportionately high and adverse; and
  6. summarize and report the results.

The evaluation of potential EJ communities is an iterative process - i.e., the defining characteristics of both the community and the actual analysis of that community are refined as the Region moves toward a more detailed analysis.

The following sections address the methodology for conducting steps 1 through 6.

2.2.1 Step 1: Develop Geographic Boundaries for Community of Concern and Conduct a Preliminary Burden Analysis

There are a number of different ways to identify the geographic boundaries of the COC. They include the use of established political boundaries - i.e., city, county or town limits; physical boundaries - e.g., rivers, main roads, or railroad tracks; and U.S. Census boundaries, such as those for census blocks or block groups. Other recommended sources of information on community boundaries include state and local governments. Nevertheless, final boundaries of the COC may be modified following community input.

In addition to delineating the community boundaries, a preliminary burden analysis may be performed with respect to the environmental burden experienced by that particular COC. Conducting an initial burden analysis can serve as a useful tool for regional managers and staff, especially in the case of reviewing applications for federally-issued permits, as the information generated may allude to a potential environmental justice concern in the vicinity of the facility early on during the permitting process. Through the use of the GIS application, along with mapping data received from the environmental burden indicators (See 2.2.4 for further description), a determination can be made as to whether a potential environmental justice concern exists in the vicinity of the permit applicant's facility.

Back to Top

2.2.2 Step 2: Compare COC Demographics to a Statistical Reference Area

Statistical reference areas are evaluated to determine appropriate cutoffs for demographic factors: minority and low income. This evaluation provides a basis for comparison to determine if the COC meets the demographic EJ criteria. A description of the statistical analysis follows.

Statistical Reference Area
Demographic data were analyzed using the 1990 Census. Moreover, the statistical cluster analysis approach was applied using Census block group data. The block group represents the resolution of least-size where the most important data sets are readily available (i.e., both for population and income). Data were evaluated on a state-specific basis. All of the statistical methods evaluated indicated that minority populations in urban areas were skewing the results for the states of New York and New Jersey. Specifically, state-wide benchmarks were similar to those derived from using only urban areas, while the results for only rural areas were considerably lower. Consequently, minority data were evaluated separately for urban and rural areas within these states. These separate analyses yield one statistical reference area for urban and one for rural for percent minority for New York and New Jersey. The following Census Bureau definitions for urban and rural were utilized:

Urban All territory, population, and housing units located in urbanized areas (UA) and in places of 2,500 or more inhabitants outside of UAs. An urbanized area is a continuously built-up area with a population of 50,000 or more.

Rural Territory, population, and housing units that the Census Bureau does not classify as urban are classified as rural.

Cluster Analysis
Block group data were analyzed using the cluster methodology statistical approach. With the use of a cluster analytical approach, data are divided into two distinct groups (e.g., minority and non-minority; low income and non-low income). Cluster analysis examines the natural break of the data. Data on percent minority and percent poverty were ranked separately in descending order for each State. (Note, as discussed above, for minority data in New York and New Jersey, the data were evaluated based on urban and rural settings). An iterative process was employed in which the data were (1) split into two groups; (2) the means for each of the two groups were calculated; (3) the difference between the means for each group was determined; and (4) Steps 1- 3 were repeated until the greatest difference between the means was found. This method results in dividing the data into two groups that are as different as possible.

GIS Comparison of COC to Statistical Reference Area
Region 2 has developed a GIS application to evaluate the demographics of the COC and compare them to a statistically derived reference area. To facilitate the statistical analysis, first the boundaries of the COC are drawn. The GIS application then calculates the percent minority and low income individuals within those boundaries using Census block group data. Where portions of a block group are inside the boundary of the COC, the total block group population is prorated based on the area included. (For example, if of the block group is inside the boundary of the COC, of the population in the block group would be utilized). The following tables were developed to provide a comparison of those percentages to the statistical reference area thresholds as discussed above.

Table 1.  State-Wide Urban & Rural Percentage Thresholds for Minority Populations

State Urban Rural
New York 51.51 34.73
New Jersey 48.52 29.39
Puerto Rico na7 na
Virgin Islands na8 na
Indian Nations9 na na

Table 2.  State-Wide Percentage Thresholds for Low-Income Populations

State Percentage
New York 23.59
New Jersey 18.58
Puerto Rico 52.0
Virgin Islands 45.2
Indian Nations 41.2

Back to Top

2.2.3 Step 3: Determine if Demographic Criteria are Met

In accordance with the Executive Order, a community is a potential EJ community if it is either minority or low income. The GIS application described above indicates whether either of the demographic criteria is met, based on a comparison of the COC demographics to statistical reference area cutoffs. If the COC demographics are equal to or above either cutoff then the COC is considered a potential EJ area that should be more fully evaluated.

2.2.4 Step 4: Develop A Comprehensive Environmental Load Profile

The environmental burden of a community can be represented by the concept of an environmental load profile (ELP). This profile is based on salient characteristics that serve as indicators of environmental burden and provide a consistent basis for comparison. The profile of the COC is compared to that of the statistical reference area and the salient characteristics (i.e., indicators of air quality, drinking water, etc.) are used to assess whether the COC is experiencing a disproportionately high and adverse burden.

Environmental Burden Indicators
Region 2 has developed a GIS application to assist in the development of the environmental load profile for a COC. While this application currently includes the following components:

i.  TRI Air Emissions,
ii. Facility Density/Population Density,
iii. Land Use Index,
iv. Ambient Air Quality Mapping (Attainment/Non-Attainment Designation),

additional indicators are planned for future development.

TRI Air Emissions Indicator
This indicator was developed using the EPA's Office of Pollution Prevention and Toxics (OPPT) model which evaluates TRI emissions. Documentation for the model can be found at the Risk Screening Indicators Website.  The model takes into account the quantity of chemicals emitted from each facility, the toxicity weight of each chemical, the potential exposure and the size of population at the receptor. It uses this information to generate an indicator value for each TRI facility. The indicator value is an unit-less value that reflects the magnitude of the relative risk impact of the facility on chronic human health. In developing the application, the Region first ran the OPPT model and generated the indicator values for all the facilities in the region and then mapped the indicator values to the TRI facilities. An inverse distance formula is used to determine the facilities' impact on census block groups within a 10 kilometer (or 6.2 miles) radius. An indicator value is calculated for each census block group by adding the cumulative impacts from facilities affecting that block group. As with the demographic analysis, the boundaries of the COC can be drawn using this tool. The TRI emission indicator is calculated by summing the indicators for the block groups included within the COC boundaries.

Facility Density/Population Density Indicator
This indicator was developed from a geographic coverage of permitted facilities and the 1990 Census block group data for total population. The unique facility data layer was created to ensure that double counting of facilities would not occur. The unique data layer was created by screening out permitted entities that were shown as separate businesses in TRI, AIRS, PCS, RCRIS, or CERCLIS, but had the same Facility Identification Number across all different program systems. Furthermore, facilities that are tracked by RCRIS that were unlikely to produce significant environmental loadings (small quantity generators and certain other subclasses) were excluded from these data.

Facility density is calculated by adding all the facilities in the community and dividing by the area of the community to derive the number of facilities per square mile in the COC. This number is then multiplied by the population density of the delineated COC to produce an indicator that gives greater weight to facilities that are concentrated in high population density areas. The index is calculated as follows:

index = (Total fac.)      x (population)
(area of COC)    (10000)

Land Use Index Indicator
The land use index was derived from the National Land Cover Data (LCD) system, which is a product of the joint effort of the Multi-Resolution Land Characteristics (MRLC) Consortium formed by six federal environmental monitoring programs. The LCD was produced from 1992(+1/-1 year) Landsat TM coverage with a 30 meter resolution. The LCD has 15 land use types in a grid format. For the purposes of this application, however, the LCD was summarized into 5 land use types:

  1. industrial/commercial;
  2. residential (combines two LCD classes - low density residential, and high density residential);
  3. open space (combines eight LCD classes - deciduous forest, emergent wetland, evergreen forest, mixed forest, other grasses, pasture/hay, row crop and woody wetland);
  4. water; and
  5. other (combines three LCD classes - bare rock/sand, quarry/strip mine, transitional barrier).

As with the other GIS tools, the COC boundaries can be drawn or imported if the COC boundaries are already defined from the EJ Demographic Screening Tool, and the percentages associated with each land use category within the COC, are calculated. An index is developed by comparing these percentages to a statistical reference area.

Ambient Air Quality Mapping (Attainment/Non-Attainment Designation)
EPA's air quality data - i.e., as characterized from the collection of data from the Ozone (O3), Particulate Matter (PM10), Carbon Dioxide (CO2), and Carbon Monoxide (CO) monitoring stations throughout the Region, are mapped using the GIS tool. Spatial interpolation was used to estimate the concentration values in areas between ambient monitoring stations. The ambient air quality is then translated to air quality index (a.i.) to provide general information to the public about air quality and associated health effects. More information on air quality index can be found on the AirNow Website.  As with the other GIS tools, the boundaries of a COC can be drawn to determine if the COC is within a non-attainment area for any of these parameters.

Back to Top

2.2.5 Step 5: Assess whether the burden is disproportionately high and adverse.

Evaluating Disproportionately High Burden
The Region intends to use statistical methods for evaluating whether the burden in a COC is disproportionately high. The analysis would be performed by evaluating data on a Census block group level. The Indicators described above would provide the basis for comparison to a statistical reference area. The Region expects to use the same statistical reference areas as were used with the demographic evaluation. Those demographic statistical reference areas were established on a state-by-state basis and by evaluating data for urban and rural settings. The statistical methodology would yield cutoff values for each of the environmental load profile indicators. The load profile indicators would be combined to evaluate the overall profile for the COC compared to the statistical reference. Initially, predetermined weights would be assigned to each load profile element. As appropriate, the Region may conduct ground truthing10 to adjust the weights to reflect actual conditions.

Evaluating Adverse Burden
There is no established methodology for evaluating cumulative risk and there are uncertainties associated with assessing environmental burden. In any event, when an acknowledged health standard for the burden in question is exceeded, the Region will consider the burden to be adverse unless otherwise indicated by supportive data.

Back to Top

2.2.6 Step 6: Summarize and Report the Results

In general, the results of the analyses should be presented in a decision document11 which may include the following:

In addition, any information collected during the EJ analysis used in support of rendering a decision on the EJ characterization of the COC (e.g., permit application, community correspondence, maps) should be incorporated into the decision document.

2.3 Additional Factors to be Considered

Notwithstanding the Region's effort to develop consistent and comprehensive methodologies for EJ analyses, there will arise exceptions and situations that are not easily adaptable to a prescribed methodology and, therefore, flexibility has been built into the Guidelines. Examples of such exceptions may include:

2.3.1 Additional Demographic Considerations

In certain circumstances, a COC may be virtually indistinguishable from any of its neighbors for a given EJ demographic factor. The examples in Region 2 are in Puerto Rico and the U.S. Virgin Islands (USVI), where every community is classified as Hispanic, in the case of Puerto Rico, and as communities of color in the case of the USVI, even though additional racial differences may exist. When the population in the larger area incorporating the COC is relatively homogeneous for a given EJ demographic factor, it is usually not useful to compute a difference in that factor between the COC and the reference area.

2.3.2 Additional Population Considerations

There will be some limited circumstances in which the Census data are not sufficient or appropriate for a specific EJ analysis. For example, certain areas in the Region may have large numbers of undocumented or transient residents who are not recorded in any official Census data bases. When it is clear that a specific community may have demographics that are significantly different from the official census figures, the Region may need to consider additional measures to develop more meaningful data. In such cases, the Region will need to work with the local government and community to develop a more representative demographic profile.

Back to Top


6.  Depending on the particular regional action involved (e.g., processing a permit application), a preliminary burden analysis may provide managers and staff with advance notification of a potential environmental justice concern.
7.  See Additional Factors.
8.  Ibid.
9.  There are seven federally-recognized Indian nations located within the external boundaries of New York State. They include: Tuscarora Nation, Tonawanda Band of Senecas, Cayuga Nation, Onondaga Nation, St. Regis Mohawk Tribe, Seneca Nation of Indians, and Oneida Indian Nation. The U.S. Government has a unique relationship with the federally-recognized Indian nations. In particular, these nations do not come under New York State jurisdiction and are not included in the above tables. These seven federally-recognized Indian nations are also provided special consideration as discussed elsewhere in this document.
10.  "Ground truthing" refers to the collection of reference data materials. Data used in the ELP would be verified by a variety of methods, which may include: collection of field data, "site checking" for land use, etc.
11.  It is recognized that not all issues will require a full EJ analysis to address the concerns of a community. In those instances, a letter or memorandum may suffice to document the Region's actions.

Back to Top

Return to Table of Contents

Go on to Section 3



Return to Table of Contents


Jump to main content.