National Academies Press: OpenBook

Effective Methods for Environmental Justice Assessment (2004)

Chapter: Chapter 2 - Identifying Protected Populations

« Previous: Chapter 1 - Guidebook Overview
Page 19
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 19
Page 20
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 20
Page 21
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 21
Page 22
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 22
Page 23
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 23
Page 24
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 24
Page 25
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 25
Page 26
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 26
Page 27
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 27
Page 28
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 28
Page 29
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 29
Page 30
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 30
Page 31
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 31
Page 32
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 32
Page 33
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 33
Page 34
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 34
Page 35
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 35
Page 36
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 36
Page 37
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 37
Page 38
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 38
Page 39
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 39
Page 40
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 40
Page 41
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 41
Page 42
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 42
Page 43
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 43
Page 44
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 44
Page 45
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 45
Page 46
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 46
Page 47
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 47
Page 48
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 48
Page 49
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 49
Page 50
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 50
Page 51
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 51
Page 52
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 52
Page 53
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 53
Page 54
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 54
Page 55
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 55
Page 56
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 56
Page 57
Suggested Citation:"Chapter 2 - Identifying Protected Populations." National Academies of Sciences, Engineering, and Medicine. 2004. Effective Methods for Environmental Justice Assessment. Washington, DC: The National Academies Press. doi: 10.17226/13694.
×
Page 57

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

19 CHAPTER 2. IDENTIFYING PROTECTED POPULATIONS OVERVIEW Environmental justice assessment traditionally has focused on identifying distributive effects to minority populations and low-income populations. This focus has evolved out of the language of Executive Order 12898 issued by President Clinton in 1994. From a technical perspective, however, the same analytical process can be used to identify distributive effects on nearly any population group. Although considerable attention has been given to minority and low-income populations in the past decade, federal and state policies and regulations offer some level of protection to many other population groups. A review of federal law and regulations shows that the universe of protected populations includes those defined by age, disability, gender, religion, class, race, low-income, limited English proficiency, and national origin. Assessment of distributive beneficial and adverse effects is an objective, analytical part of the environmental justice assessment process. Common transportation planning practice is to evaluate the effects of transportation system changes to “the public” or “local populations,” in other words to the population at large. By identifying how effects may be differentially distributed among various population groups, the methods provided in this guidebook give you the ability to evaluate transportation system changes with greater precision. This form of analysis is a vital element in performing an environmental justice assessment. Assessment of distributive effects involves combining demographic and spatial analyses with social, economic, and environmental effects analyses. The objective of environmental justice analysis in transportation is to assess the extent to which the benefits and costs of a proposed transportation system change would be experienced differentially by protected populations and other members of society. To make such an assessment, it is essential to have a clear sense of the areas in which minority populations and low-income populations move about most frequently, that is, where they are most likely to experience positive or negative impacts. The most common means of defining areas where impacts are likely to be concentrated is through place of residence. This is a logical approach for many types of effects. For example, noise impacts are generally most significant when they occur near a person’s home, as are community cohesion and aesthetic impacts. Unless a person spends nearly all of his/her time at home, however, many other types of effects are likely to be experienced throughout the day during daily activities. To assess the nature and magnitude of impacts that vary spatially throughout a community, it is first necessary to gain a sense of the geographic space within which protected populations tend to circulate. This geographic space is commonly referred to as “activity space.” To determine the activity space of protected populations, you must examine the social, affective, and physical aspects of these communities. Specifically, we present methods that can assist in estimating: • The location and relative importance of activity spaces. • Accessibility to these locations.

20 • How the proposed changes would affect protected population groups. The methods presented in this chapter include many effective techniques for identifying protected populations using demographic data. This information can be collected either directly or from sources such as the U.S. Census Bureau. In addition, various geographic information system (GIS) and database applications allow you to process 1990 and 2000 census data and apply a number of the methods presented in this chapter. The methods presented generally involve the following steps: • Collect the necessary information. • Verify the accuracy of the information if possible. • Calculate pertinent population statistics. • Assess the presence or absence of protected populations. Many of the methods are quantitative and use census data, survey data, and GIS. Other methods are more qualitative and rely heavily on local knowledge and the public participation process. Which method is best to use depends on various factors including the probable magnitude of a particular impact, data availability and cost, and the capabilities and experience of the person performing the analysis. The most important factor is whether or not the method is appropriate for the type of transportation system change being evaluated. For example, some effects of transportation system projects are distributed geographically, whereas other effects are distributed among system users based on their demographic characteristics. The best methods and data for identifying the demographic characteristics of affected populations may differ from case to case. One purpose of any protected population assessment is to accurately represent the demographic characteristics of the affected populations. Any assessment should include input from members of the public and individuals with comparatively high levels of knowledge of the local area. Many methods rely directly upon these sources of information. For other methods, such as those that rely on census data, local knowledge and public input can be used to validate source data and study results. A number of special considerations must be addressed when evaluating census data to identify protected populations. Some of these considerations include identifying appropriate comparison thresholds for analysis; selecting the appropriate scale of census data; estimating population characteristics for study areas; and comparing historic census data with current (2000) census data. These special topics are addressed in this chapter and in the appendices to this guidebook. STATE OF THE PRACTICE Most studies conducted since the mid-1990s to conform to requirements of the National Environmental Policy Act (NEPA) include some level of demographic review that includes analysis of minority and low-income population information. Inclusion of this information in

21 transportation planning studies is common but not universal. A review of recent NEPA- conformance studies and a survey of practitioners show that local knowledge and analysis of large-scale census data are the most commonly used techniques for identifying protected populations (Sheeley and Forkenbrock 2002). Similar types of environmental justice assessment are also becoming commonplace in planning products such as major investment studies (MISs). Field survey and data collection techniques also are commonly used. Assessment results released to the public often are brief summaries with very little detail. The specific methods used and analytical results obtained generally are not documented in detail. Some techniques are used more rarely because of their technical complexity. These include formalized public participation-based qualitative assessment techniques, detailed assessment of small-scale census data, and complex methods such as historical data analysis and population projection. The state of practice could clearly be enhanced, however, if these rigorous assessment methods were more accessible to practitioners. Making these methods more accessible is one of the goals of this guidebook. The methods presented in this chapter cover the spectrum of potential approaches from simple to complex and qualitative to quantitative. A key consideration in environmental justice analyses related to potential transportation projects is how protected populations move about in time and space. Because few people spend the majority of their time at home, it often is not sufficient to determine only where minority and low-income populations live; it also is important to ascertain prevalent daily activity spaces. Trip diaries and surveys are the most common means of identifying how people move about in time and space. These methods can be very simple in nature or quite involved. The survey data can reveal the activity space within which protected populations frequently, occasionally, or seldom travel. Revealed preference analyses are a practical way to evaluate how people actually respond to the choices available to them, and therefore what their preferences are when trading off attributes. These analyses enable you to assess, for example, how groups of people balance lower prices against convenience when purchasing goods. With sufficient data, this approach can enable you to “make direct inferences about the trade-offs travelers make among site and distance attributes” (Pipkin 1986, p. 183). When using these analyses, you should keep in mind that people will reveal preferences only in terms of the available alternatives. In considering shopping trips, Pipkin pointed out that individuals who do not own a vehicle or are otherwise disadvantaged might be forced to forego certain kinds of trips or to restrict their travel to nearer, smaller, and perhaps more expensive alternatives. Variables that help explain travel patterns include the following: • Personal attributes such as age, ethnicity, income, and level of education; • Site characteristics such as price, quality and convenience for shopping, service, or recreational destinations; and

22 • Tension between locations that are of significant value yet geographically dispersed—i.e., home, work, and school. Kitamura et al. (1997) identified four fundamental elements that influence urban travel patterns. Two of these elements, time budget and activity pattern, involve the individual traveler. The other two elements, land use and transportation, relate to the urban system. These four elements interact within a framework of space and time. Miller (2001, p. 2) attempted to synthesize time geography, activity theory, and GIS into something he called “people-oriented GIS.” Method 12 builds on Miller’s approach by synthesizing several global positioning system (GPS) space-time activity studies. SELECTING AN APPROPRIATE METHOD OF ANALYSIS Each chapter of this guidebook includes a table that summarizes considerations you can apply to select an appropriate method of analysis. You can use the table as a concise list of the methods discussed in each chapter to identify specific methods to read about in more detail. Each method discussion provides further information to help you understand appropriate uses. Table 2-1 summarizes the protected population identification methods presented in this chapter. We present 12 methods below, which generally pertain to identifying the areas in the community where protected populations currently reside and where they may live in coming years. The last three methods apply to analyzing the activity space within which these populations typically move about. These methods vary in complexity—how complex an analysis of activity space you should undertake may be based on the following considerations: • Spatial nature of likely impacts. If a proposed project is expected to affect a sizable portion of the community, it would be important to assess the extent to which the impacts would be experienced in areas where protected populations frequently carry out their activities. • Perceived complexity of potential impacts. If the potential impacts of the proposed project are likely to be substantial or complex in nature, it is essential to understand how these impacts would be distributed among protected populations as opposed to other groups. • Perceived importance of potential impacts. If members of a protected population viewed a particular type of effect as particularly significant, it would be wise to use a relatively powerful method to assess how greatly this effect would impact the common activity space of that population. As these considerations make clear, a basic knowledge of the protected population and its activity space will be necessary before a method can be selected. To determine which method to use, you should have a general idea of the potential problem and the area’s population and characteristics.

23 Table 2-1. Summary of methods for identifying protected populations Method Assessment level Appropriate uses Use when Data needs Expertise required 1. Local know- ledge and public input All Recommended in all situations Initial evaluation of potential for distributive effects and to assure quality of findings of other methods Low Local area/ community involvement 2. Threshold analysis Screening/ detailed Regional plans, STIP/TIP, sys- tem assessment Demographic patterns must be evaluated for large areas Low GIS, Census data 3. Spatial interpolation Screening/ detailed Corridor/ project Demographic patterns must be evaluated for small areas or population patterns must be evaluated for finite areas of effect Medium GIS, Census data 4. Field survey Detailed Corridor/ project Detailed residence, business, and public space location information is required Low/ medium GPS & photo interpretation can be useful 5. Customer survey Detailed All System users could experience distributive effects Medium/ high Survey design 6. Population surfaces Detailed Regional plans/ corridor/ project Scenario modeling or integration with grid- based modeling packages is required High GIS, Census data 7. Historic data review Detailed All Past projects or investment plans are at issue, or when population trends are needed Medium/ high GIS, Census data 8. Population projection Detailed Regional plans, STIP/TIP Planning horizon is five years or more High Census data, statistical modeling 9. Environmen- tal justice index Screening/ detailed All Combined analysis of multiple demographic factors is needed Medium/ high Census data, GIS 10. Personal interviews Screening/ detailed Regional plans/ corridor/ project Analysis of a relatively well-defined impact area Low/ medium Interview techniques 11. Abbreviated diary Detailed Corridor/ project Analysis of movement along a corridor is needed Medium Sampling, surveys 12. Space-time activity analyses Detailed Corridor/ project Analysis of movement along a corridor is needed High Sampling, surveys, GIS, GPS

24 METHODS As highlighted in Table 2-1, the following methods can be applied to identify the locations and activity space of protected populations. Method 1. Local knowledge and public input Practitioners and stakeholders involved with the planning process usually are able to provide considerable insight into the population within a designated study area. Public input can be obtained as part of the community involvement process through interviews, surveys, focus groups, and feedback from public meetings. When to use. Using local knowledge and public input is among the most commonly applied methods for identifying protected populations. This method is well suited for identifying effects that are distributed spatially, as well as those that are distributed among transportation system users. It can be used in all environmental justice assessments performed for transportation system changes at the project, corridor, and systems level. Even if other methods are used to identify protected populations, local knowledge and public input should be used to verify results (see box titled “Using local knowledge and public input to validate census data,” p. 25). This method is also effective as an initial screening technique to determine if other, more data-intensive methods for identifying protected populations are needed. Analysis. The following techniques should be considered for identifying protected populations through application of local knowledge and public input. Interviews − In-person one-on-one interviews can be conducted with individuals identified as community leaders (people who work with or represent other people, e.g., neighborhood activists, elected officials, clergy, and representatives of local interest groups). These individuals are likely to have knowledge of, and insight into, local issues that cannot be found elsewhere. Interviews should be conducted early in the process so that the information gained can be taken into consideration as soon as possible during project development. Surveys − Surveys may be carried out using samples that are broad-based (general population- based) or more narrowly focused (neighborhood-based). Some are scientific and produce statistically valid quantitative data; others are more informal and produce a mixture of qualitative and quantitative information. Surveys should be designed on the basis of the information that is needed. For example, if quantitative data are required for a particular district or area of the community, a formal survey should be designed to be statistically representative of individuals in the entire district. On the other hand, if information is needed from a group of people who have a particular interest in an issue, an informal survey should suffice. Focus groups − A focus group is a small group discussion run by a facilitator. The group is carefully selected, either randomly or nonrandomly (to secure representation of particular groups). A random group will ensure representation of all segments of a population, whereas the nonrandom group will be helpful in eliciting a particular viewpoint or position. A focus group generally has the following elements and objectives: a scripted agenda (including five or six

25 major questions); an emphasis on gathering many different points of view rather than on presenting information; identification of major points of agreement and disagreement; and approximately 8 to 12 participants. Feedback from public meetings − A public meeting (or open house or hearing) is a forum for receiving comments from the public. These meetings generally have ground rules regarding listening and speaking (e.g., a time limit for speakers). Participants also have an opportunity to submit written comments at the time of the meeting or afterward. An advertising strategy is always necessary to ensure good participation in the public meeting by as large a segment of the local population as possible. Using local knowledge and public input to validate census data The case of Louisiana Energy Services (LES) is a good example of why local knowledge and public input must be used to identify protected populations and issues that can arise if census data are not augmented by additional information sources. In 1994 the Nuclear Regulatory Commission (NRC) heard a complaint filed by the group Citizens Against Nuclear Trash (CANT) against proposed construction of a uranium enrichment plant. LES was planning to build the plant in Claiborn Parish near Homer, Louisiana. CANT represented the small communities of Center Springs and Forest Grove in the complaint. Among other things, the complaint alleged that the environmental impact statement (EIS) did not sufficiently address all adverse environmental, social, and economic effects, and that there was racial bias in the choice of the plant’s location. LES planned to reroute a road and greatly increase the travel distance between the two small communities. These communities, populated by individuals in protected population groups, were not discussed in the Draft EIS and were not identified on a map that showed local communities. They were overlooked because the Draft EIS analysis relied solely on census data and map sources with a coarse level of resolution that did not show the two communities. The NRC agreed that the EIS did not adequately address impacts to the two communities and, ultimately, LES was required to resubmit sections of the EIS. The map was changed in the final EIS. In addition, LES had to revisit the entire siting process and show that racial bias did not play a part in the choice of the site. At this point, LES withdrew its application. Data needs, assumptions, and limitations. Information should be collected on locations and names of communities on maps, and important community centers should be identified, along with public spaces where protected populations may congregate. Assume that a representative cross-section of the communities in the study area are involved in the process. This process may be time consuming and does not produce estimated population sizes or demographic characteristics. Also, these data cannot be readily incorporated with quantitative assessment. Results and their presentation. Map census data indicating neighborhoods and important community centers identified in a survey or set of interviews. This will produce a tabulation of

26 protected population groups, their residential and activity spaces, and important modes of transportation. Assessment. Techniques for obtaining input on protected populations from individuals with local knowledge should be applied to assessment of most transportation policies, programs, and projects. Techniques such as interviews, surveys, focus groups, and feedback from public meetings are often the best way to identify where protected populations live, their activity spaces, and modes of transportation. Both qualitative and quantitative information can be collected, depending on need. Method 2. Threshold analysis using large-area census data Census data from 1990 and 2000 are reported at many levels of analysis and are available from numerous sources, including the U.S. Census Bureau, the Internet, and federal repository libraries. Large-area census data can be obtained for states, counties, census tracts, and census- designated places. Small-area reporting units available in census data include traffic analysis zones (TAZs), block groups, and blocks. Evaluation of small-area census data is discussed as a separate method. It should be noted, however, that the large-area evaluation methods discussed below can be used with small-area data when needed. When to use. Large-area census data are useful for evaluating the distributive effects of state and regional transportation plans and other systems-level planning efforts. These data also may be applied to initial assessment-of-corridor studies, depending upon the size of the corridor and the nature of the transportation system change that is being evaluated. Additionally, this type of data can be used when spatial demographic patterns must be evaluated or to estimate transportation user demographics at the systems level. In general, protected populations are identified from large-area census data using some form of threshold analysis. Analysis. Threshold analyses are carried out in five steps: Step 1 – Define the study area. Consider the scope of the proposed transportation system change and select a study area that encompasses all affected areas and populations. For a transportation investment plan, the study area may even be statewide. A regional long-range transportation plan may require a multicounty study area. Step 2 – Select analysis units to be used. The selected analysis unit (e.g., county, tract, TAZ) must balance the amount of data to be evaluated with the level of precision required to identify the distribution of effects. For example, county units of analysis may be appropriate to evaluate distributive effects of a statewide multimodal transportation investment plan. County-level data are not sufficient, however, to evaluate whether protected populations have equitable access to a regional transit system. In this case, census tracts or TAZs may be a more appropriate choice to address the distances that members of protected population groups must travel to reach regional transit nodes. Step 3 – Acquire data and compute demographic statistics. Information on where to obtain large-area census data is presented in “Data needs, assumptions, and limitations.” Depending on the data source, it may be necessary to compute the demographic statistics needed to identify

27 protected populations; the Census Bureau does not always precalculate these statistics. Some data sources, however, provide precalculated variables that can be used when appropriate. Guidance on techniques for calculating demographic statistics from standard census data files can be found in Appendix D. Step 4 – Determine threshold levels. When using a data-driven technique, it is often necessary to define “protected populations” on the basis of threshold values for various demographic variables. The thresholds are used as comparison values to determine if protected populations exist in a study area. For example, if an area’s percent minority population is not equal to or greater than the established threshold value, the level of environmental justice concern can be assumed to be lower than in areas where the statistic is greater than the threshold. Two basic approaches are used to define thresholds. The first approach commonly used by agencies is to establish a working group to evaluate and determine appropriate threshold levels. Recently in New York, the state environmental agency’s environmental justice working group set statewide thresholds that were approved after being submitted to the public for comment. A second commonly used approach is to set thresholds to equal either state- or county-level averages, depending upon the size and geography of the study area (see box titled “Limitations of using comparison thresholds in environmental justice assessment,” p. 28). Step 5 – Identify protected populations. When using large-area census data, all evaluation units generally include members of protected populations. It is thus more appropriate to consider this technique as an approach for categorizing evaluation units based on the proportion of protected populations that they contain. Evaluation units with protected population levels greater than the established threshold values are considered to have substantial protected populations and higher potential for distributive effects than other evaluation units. An example is shown in Table 2-2. The results were obtained from a review of county-level 2000 census data for the Texas Department of Transportation, Houston District. Table 2-2. Relative level of environmental justice concern in the Houston district Area Percent minority Percent low income Relative level of concern Threshold value (State of Texas) 47.6 17.0 — Brazoria County 34.6 12.5 Lower Fort Bend County 53.8 7.3 Higher Galveston County 36.9 16.8 Lower Montgomery County 18.6 11.9 Lower Harris County 57.9 14.9 Higher Waller County 50.1 20.0 Higher

28 Limitations of using comparison thresholds in environmental justice assessment Background. Thresholds often are a necessary component of data-driven evaluation techniques. They serve as a useful way to define “level of environmental justice concern” (a commonly used term in environmental justice assessments) based on demographic characteristics. It is important, however, to use threshold comparisons wisely, as part of a thorough evaluation process. Although they can serve to categorize and prioritize areas based on level of concern, the problem with thresholds is that they are arbitrary. Example. Consider as an example an environmental justice assessment of a proposed airport expansion. Environmental review indicates that most adverse noise and air quality effects on surrounding populations will occur within 1 mile of the airport boundary, so a 1-mile buffer of the airport is selected as the study area. Because the study area is completely contained within a single county, the county’s average demographics are selected as the comparison threshold. A review of small-area census data (described later in this chapter) shows that the percent of study area population living below the poverty level is 14.5 percent, which is below the county average threshold of 16.5 percent. Similar results were obtained in a review of minority population data. Based on this information, it could be concluded that there is limited concern about adverse distributive effects to protected populations. Discussion. The above approach is common practice. The assessment process is relatively easy to perform, and the findings and conclusions are easily presented. It is often advisable, however, to perform a more thorough analysis in many situations. Figure 2-1 continues with the airport expansion example and presents the results of a thorough demographic evaluation performed at each quarter-mile increment from the airport boundary out to 8 miles. The figure supports the conclusion that the low-income percentage within the 1-mile area is less than the county average threshold. A more interesting observation is that the highest percent of people living below the poverty level is from 1.25 to 2.5 miles from the airport. Conclusion. Instead of establishing an arbitrary threshold for analysis, it may be more appropriate to perform a thorough review of the demographic data; identify areas where substantial minority populations and low-income populations live or work; and then evaluate effects to determine the beneficial and adverse distributive effects that would result in those areas. Persons Living Below Poverty Level 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 Distance from Project Site (Miles) P er ce nt o f P op ul at io n County (16.5%) State (21.8%) Figure 2-1. Percent of population below poverty level

29 Data needs, assumptions, and limitations. The single best sources of data are Summary File 3 of the 2000 Census for geographically distributed effects and the Census Transportation Planning Package (CTPP) for effects experienced by system users. Still, useful demographic data are also available in other census summary files. For comparison with previous census data, use Summary Tape File 3 from the 1990 Census. Online resources for these data are included in the resources at the end of this chapter. These data can be combined with census TIGER (Topologically Integrated Geographic Encoding and Referencing) files in GIS if maps are needed to analyze spatial distributions. This method contains several important assumptions that serve to limit its effectiveness. The most important limitations include the following: • Populations are not distributed uniformly. Most techniques that use census data assume that populations and population characteristics are uniformly distributed within census units. This assumption can be especially problematic when working with large- area census data. Census geography is hierarchical. So, for example, counties are composed of tracts; tracts are composed of block groups; and block groups are composed of blocks. In general, population characteristics will be most uniformly distributed in blocks and variability will increase with the size of the census unit. It has long been known that statistical studies will yield different results depending upon the level of census data that are analyzed. This is in part because the data violate the assumption of uniform distribution to differing degrees (Fotheringham and Wong 1991; Amrhein 1995). An example of the variability in results is provided in the discussion of “GIS-based techniques to estimate demographic characteristics,” where block and block group-level data provide different results for the same study area. • The level of demographic resolution should match the scale of effects. Resolution, or scale, is another limitation of large-area census data and is related to the uniform distribution assumption. Data for large-area census units are totals of the small-area census units contained within them. Large-area census units thus have less resolution than small-area census units. Therefore, it is important that large-area census data not be used in instances where a high degree of demographic resolution is needed. An example would be assessing the effects of noise on nearby residences, which is a highly localized effect. In such instances, it is more appropriate to use small-area census data, even if the study area is very large. In other instances, large-area census data are appropriate. An example would be an analysis of transportation policy changes on regional air quality. • The Census may undercount protected populations. Survey and enumeration techniques used in census data collection rely heavily on address lists. It has long been argued that low-income, minority, and other protected populations are consistently undercounted and that the census is not an entirely accurate representation of these populations. Elaborate statistical techniques are available to correct for population undercounting, but these techniques rely on assumptions that are often difficult to validate. It is probably more appropriate when working with census data to understand that protected populations may be undercounted and to apply conservative threshold levels and use local knowledge as a means of verification.

30 GIS-based techniques to estimate demographic characteristics Study area boundaries often do not coincide with the boundaries of census reporting units. When performing environmental justice assessments, it is important to understand the relationship between study area geography and census unit geography. It is also important to understand that use of different census reporting units, such as block groups or blocks, may yield different results. Three common GIS-based methods for estimating study area population characteristics are known as polygon containment, polygon intersection, and areal interpolation. Each involves a different approach for identifying the census reporting units that overlay a study area. Areal interpolation is the most accurate but is more data intensive than the other methods. The polygon-intersection technique tends to over- predict population, whereas the polygon-containment technique tends to under-predict population. Census blocks offer the most detailed census geography and will yield the most precise estimates, but only limited demographic information is reported at the block level. Figure 2-2 shows the census block and census block group geographies used to derive population estimates for a 1-mile study area surrounding a proposed bridge location. The shaded census units show the area from which the 1-mile area estimates are determined. Note that the same 1-mile buffer is shown in each diagram. The population estimates obtained from the various methods and census units vary by as much as 12,092 persons, or 4.7 percent. Figure 2-2. Census reporting units and study area geography Total population: 10,992 Percent minority: 95.8 Total population: 11,187 Percent minority: 96.2 Total population: 13,278 Percent minority: 95.1 Total population: 18,942 Percent minority: 92.9 Total population: 9,648 Percent minority: 97.2 Total population: 6,850 Percent minority: 97.6

31 Results and their presentation. Results of large-scale census data analysis are often presented in narrative format. Using the data presented in Table 2-2, results could be expressed in the following manner: In the Houston District, Fort Bend, Harris, and Waller Counties have a higher proportion of individuals in protected population groups than other counties in the District. Tables can be used to convey large amounts of information in technical reports, but the amount of information should be kept to a minimum if tables are to be used in public forums. Maps can be used to convey spatial patterns, but maps are often more appropriate for presenting small-scale census data. Assessment. Large-scale census data are suitable for identifying protected populations if the patterns of effects are uniform over large areas. Much useful information is available for large- scale census units, such as tracts and counties. This information can be used to assess both the spatial distribution of protected populations and the demographics of transportation system users. Large-scale census data are best suited for assessing state and regional policies and programs, and for transportation system changes that would generate system-wide effects. Such census data are generally not suited for project-level analysis and should be used cautiously for corridor-level assessment. Method 3. Spatial interpolation using small-area census data Small-area census data at the block, block group, and TAZ level offer the most detailed nationally available demographic information useful for identifying protected populations. Database, spreadsheet, and GIS software are often necessary for this type of analysis because even relatively small study areas commonly encompass a large number of small-area census units. When to use. Small-area census data should be used in situations where the scale of effects to be analyzed requires a high degree of demographic resolution, such as when project effects are limited to relatively small, localized areas. Small-area census data can also be used if results of studies using large-area census data are questioned, making it necessary to obtain the “best available” or most accurate census data. The small-area census data and the techniques described below are most applicable to project- and corridor-level analysis. Blocks and block groups can be used to assess spatial demographic patterns, whereas traffic analysis zones should be used in situations where transportation user demographic characteristics are required. Analysis. Spatial interpolation using small-area census data is conducted in five steps: Step 1 – Define the study area. Because small-area census data are best suited for identifying protected populations in situations where effects are localized, it is often possible to define a study area based on detailed geographic patterns of effects. For example, contours or receptors developed from noise and air quality analyses could be used to define areas of effects. Viewsheds may be selected as the area of effects to address visual quality impacts. A more simplified approach is to select a buffer distance that encompasses the geographic extent of

32 effects. GIS can be a useful tool for defining study areas. Contours generated from modeling programs can be brought into GIS for overlay with census data, and buffering tools can be used to develop study areas from site locations or project corridors. Step 2 – Compute statistics for protected populations. If patterns of effects are distributed spatially, block groups and blocks should be selected as the census units for analysis. A combination of block groups and blocks is recommended because, while blocks offer the highest level of resolution, all potentially necessary data are not reported at the block level. If patterns of effects are distributed among transportation system users, traffic analysis zones should be used where possible. Data for traffic analysis zones are available in the 1990 and 2000 CTPP. Table 2- 3 lists the protected population demographic data that are available at the various census reporting levels. Statistics can be calculated using various spreadsheet, database, or GIS applications. Formulas for computing useful demographic statistics can also be found in Appendix D. Table 2-3. Availability of protected population data by 2000 census reporting unit Census reporting unit Total population Minority Income Age Gender Disability English speaking Block     Traffic analysis zone       Block groups, tracts, and larger        Step 3 – Overlay demographic data with area of effects. The purpose of this step is to identify the census-reporting units that fall within the area of effects. The three overlay methods for selecting census-reporting units are known as polygon intersection, polygon containment, and areal interpolation. Each approach can provide a different estimate because areas of effects do not commonly coincide with census unit boundaries. For more information, see the discussion on “GIS-based techniques to estimate demographic characteristics.” Step 4 – Estimate demographic characteristics of the study area population. The demographic characteristics of the study area population can be tabulated once the census units within the area of effects have been identified. Values can be reported for the individual census units within the study area, which is useful for assessing the population distribution and characteristics. It can also be useful to generate an estimate of the population characteristics for the area as a whole. An areal interpolation technique is best suited for developing this estimate. The most common and easily performed method is called area-weighted interpolation, which assumes that populations within the census units are uniformly distributed (Goodchild and Lam 1980). Many interpolation routines do not rely on the uniform distribution assumption. One of the more promising techniques, especially useful in sparsely populated areas, uses an overlay of

33 road networks with census units under the assumption that most residences are near a roadway (Mrozinski and Cromley 1999). Another method uses ancillary data as an input mask to identify areas within census units where there are no residences (Bloom, Pedlar, and Wragg 1996). In this way, nonresidential areas such as parks and areas zoned for commercial or industrial use are assigned zero population. Step 5 – Compare to thresholds for analysis. The final step is to compare demographic estimates to analysis thresholds in order to identify the presence of protected populations. This step is the same as that for the analysis using large-area census data. Data needs, assumptions, and limitations. The following data sources will be necessary, depending on the census-reporting units selected for analysis: • Summary File 1, to compute block-level statistics; • Summary File 3, to compute statistics for block groups, tracts, and larger reporting units; and • The Census Transportation Planning Package to compute statistics for TAZs. Information about acquiring and using these data sources is provided in Appendix D. Analysis of small-area census information can be quite data intensive and requires some combination of spreadsheet, database, and GIS software. GIS software is needed to perform the interpolation methods discussed above. The assumptions and limitations discussed for large-area census data (see Method 2) generally apply to the evaluation of small-area census data. In addition, the complexity of many analytical processes and the amount of data required often limit the use of small-area census evaluation methods to situations where other less intensive techniques have not provided adequate results or results have been contested. Results and their presentation. Results can be presented as maps, graphs, and tables (as shown in Figure 2-3) depending on the purpose and the intended audience. Maps are the best means of presenting geographic patterns and are often essential for conveying the proximity of protected populations to sources of beneficial and adverse effects. Maps should be relatively simple, showing only the census data theme, such as percent minority by census block, and enough other features to orient the reader. Graphs, on the other hand, can be used very effectively to provide a comparison of study area demographics to comparison areas or threshold levels. Tables should be used primarily in technical reports; only tables with five to seven data values or fewer should be used for communication with the public. Tables do, however, offer a very useful way of organizing and summarizing the results of small-area census data evaluations. Assessment. Compared to other techniques, analysis of small-area census data is more complex and more data intensive. However, this method offers the finest demographic resolution available with census data and should be selected over other methods in situations where the effects from a proposed transportation system change will be localized to specific areas.

34 58.1 81.6 78.2 62.3 47.6 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 1/2 Mile 1 Mile 1.5 Miles County State P er ce n t M in o ri ty P o p u la ti o n Area Distance Total population Total minority Percent minority 1/2 mile 124 72 58.1 1 mile 1,249 1,019 81.6 1-1/2 miles 4,717 3,690 78.2 2 miles 9,195 7,497 81.5 County 313,645 195,467 62.3 State 20,851,820 9,918,507 47.6 Figure 2-3. Techniques for presenting study area demographics Method 4. Field survey A field survey, also known as a dashboard or windshield survey, involves obtaining local knowledge by actually traveling about the area and taking notes. When to use. A field survey generally is recommended as part of project-level environmental justice assessments; they are less practical for corridor and system-wide assessments. Field surveys are especially important in situations where project effects such as noise and air quality will be highly localized and census data do not provide a fine enough level of resolution. A field survey is also a good technique for verifying the accuracy of small-area census data. Field surveys can be comprehensive in small, manageable study areas. If projects cover large areas or field surveys are conducted at the corridor level, it may be more appropriate to identify specific locations for a field study. Field surveys can be used to collect the following information that is not obtainable from a review of census data alone: • Mapping the location of residences in a study area.

35 • Delineating specific neighborhoods by identifying visible changes in neighborhood characteristics. • Identifying nonresidential locations, such as churches and community centers, that are used and valued by members of protected population groups. • Identifying small areas of extremely high population density, such as apartment houses, and commercial and industrial areas with no population. Analysis. The four steps in carrying out a field survey are: Step 1 – Obtain location maps. Location maps are necessary for planning the field survey and for documenting your findings. Topographic maps, maps produced in GIS from readily available data, and maps from Internet service providers such as MapQuest and Yahoo will serve the intended purpose. City maps, county maps, and maps sold by companies such as Delorme and Thomas Brothers will also work well. To verify census information, it is also necessary to prepare small-area census maps before conducting the survey. Step 2 – Plan your route. It is important to plan your route before conducting the survey for safety reasons, to save time, and, most essentially, to ensure that you visit all important locations and inspect a representative cross-section of the study area. This is especially important in situations where the study area is too large to be surveyed in its entirety. Step 3 – Perform field survey, collect field notes. During the field survey, it is important to take notes to ensure that all relevant findings are documented. Be on the lookout for “sensitive receptors” such as schools, hospitals, and nursing homes, as well as locations that visually do not appear to corroborate census information. This could include large areas with no population that are depicted as having high population on the census map or affluent neighborhoods marked as low income. Because the census is only conducted every 10 years, be alert for newly developed areas that do not yet appear in the census data. In sparsely populated areas or when certain types of impacts are being evaluated, it can be worthwhile to map the location of individual residences as part of the field survey. This effort can be supported by the use of GPS technology or aerial photography. Photographs are an especially useful method of documenting detailed information about appearance and relative location for future reference. Step 4 – Conduct follow-up activities. It is often necessary to conduct follow-up activities. Questions may arise after the survey based on information that was collected during the initial trip. When in doubt, conduct a follow-up survey, talk with community members, or speak with someone knowledgeable about the local area to answer any questions that you may have. Data needs, assumptions, and limitations. Field maps and possibly census maps are required to conduct a meaningful field survey. Although this technique is an effective way of obtaining information about a study area, there are notable limitations. The technique relies heavily on visual cues, which are not always accurate. Unlike door-to-door surveys or telephone surveys, the dashboard survey is limited to observations obtained in the field from a visual inspection of the study area. Also, although this technique aids in identifying areas of concern, it is only of limited use in estimating population counts (such as in sparsely populated areas). Other demographic information, such as income level, age, and disability, are virtually impossible to obtain through a field survey.

36 Results and their presentation. Results are often printed in narrative form as references to specific locations or features of the study area. Photographs are often used to depict examples of locations of concern. Another useful method of presentation is to include specific information gathered during field surveys as annotation on project site maps. The map in Figure 2-4 includes information from a field survey. The information has been added to the census map for the area near Harbor Bridge, originally shown in Figure 2-3. Figure 2-4. Techniques for presenting field survey results Assessment. A field survey is an effective technique for gaining local knowledge of areas potentially affected by transportation system changes. The method is well suited to use in combination with techniques for evaluating small-area census data. Information from field surveys can be enhanced by performing door-to-door surveys to collect information from local citizens. Method 5. Customer survey Customer surveys are among the most effective methods for evaluating user demand and user perception of transportation system quality. When to use. Customer surveys should be considered to identify protected populations when evaluating effects experienced by transportation system users. These include issues such as transportation safety and transportation choice. Note that the survey should be constructed to evaluate both protected populations and the effects of the transportation system change being studied.

37 Analysis. In general, a customer survey is conducted in two steps: Step 1 – Construct the survey. The transportation survey should be designed to collect information on both demographic characteristics and demand/perceived quality. In many areas, it will be important to prepare the survey in multiple languages. Figure 2-5 includes a series of questions suitable for identifying important social group characteristics of transportation system users. The questions were designed to coincide with attributes available in the 2000 census. Sex Male ____ Female ____ Age Under 18 ____ 18 to 29 _____ 30 to 49 ____ 50 to 64 ____ 65 and older ____ Race African American ____ Asian _____ Hispanic/Latino _____ Native American _______ Caucasian ______ Other _____ Family income Less that $15,000 ___ $15,000 to $25,000 ___ $26,000 to $50,000 ___ $51,000 to $75,000 ___ $76,000 to $100,000 ___ More than $100,000 ___ English proficiency Can you speak English? Very well _____ Well _____ Somewhat _____ Not at all _____ Disability Do you have a disability that affects your mobility or preferred method of transportation? Yes _____ No _____ Figure 2-5. Survey questions to identify protected populations Step 2 – Administer the survey. User surveys can be administered at the point-of-use for many forms of transportation including pedestrian, bicycling, and bus and light rail transit. It is more difficult to conduct point-of-use surveys for vehicle transportation. However, for any mode of transportation it is possible to distribute the surveys through organizations (such as Area Agencies on Aging), businesses, or through the mail. Another promising technique is to administer surveys online via the Internet as more and more people are obtaining access either at home, through work, or at local libraries. It is important, however, to consider whether the population to be surveyed (e.g., low-income and other disadvantaged persons) has Internet access before using an online survey. Data needs, assumptions, and limitations. Surveys reflect the views and opinions of the people who complete them. It is critically important to ensure that a survey adequately captures the characteristics and views of the user population. Survey design and sample size are the two most important elements to ensure adequate results. A simple rule of thumb is that a larger sample provides more precise results. Larger samples, however, often cost more money, so in practice it is necessary to balance tradeoffs between statistical precision and cost. Similarly, good survey design is necessary to develop questions that are unambiguous, easily understood, and provide

38 informative results. NCHRP Report 456 (Forkenbrock and Weisbrod 2001) provides a concise discussion of survey design and sample size. As with any user survey, results will reflect the demographic characteristics of current and past users. People who have not used or are not willing to use the method of transportation will not be represented. Results and their presentation. Results can be summarized in narrative form to present key findings. Statistical analysis can be used to evaluate the accuracy of survey results. Results can be tabulated and presented numerically to support a statistical analysis. Key findings can also be presented graphically. Results can be presented by social group (such as low-income versus middle-high income, or minority versus nonminority) to identify if preference or levels of impacts vary by social group. Thus, if a proposed transportation change will have a greater effect on transit service than on vehicle commuting, it would be possible to determine whether there is a greater preference for transit ridership among low-income individuals. Assessment. Surveys are an effective method of assessing transportation user demands, preferences, and perceived quality of service. To allow for environmental justice assessment, it is important to include in the survey design questions that identify whether respondents belong to any protected population groups. These questions need not be detailed or personally invasive. With a little planning, it is a simple task to administer a survey that identifies the respondents’ pertinent demographic characteristics. Method 6. Population surfaces When to use. This is a special technique for evaluating small-area census data. It is a method for processing census data in situations where it is either necessary or beneficial to estimate population characteristics for grid-based model cells. Population surfaces can be used to evaluate distributive effects using modeling results commonly generated for air quality and noise. Analysis. The four steps in developing a population surface are as follows: Step 1 – Acquire input data. The input data requirements and set-up steps are the same as for traditional GIS-based analysis of small-area census data. Step 2 – Develop analysis grid. This is commonly done in conjunction with a model-based analysis of transportation system effects. For example, the analysis grid developed for a regional air quality model could be used as the population surface grid in GIS. With most GIS software packages, it is best for the grid to be uniform (i.e., all cells have same length and width, such as 100 feet by 100 feet) although it is possible to establish a population surface using nonuniform grids. Step 3 – Perform zonal-to-surface population conversion. The zonal (census polygon) to surface (grid) conversion can be performed using GIS. Numerous methods exist for converting zonal population surfaces to grids. Martin (1996) describes a commonly used technique. Step 4 – Compute population statistics. It is most appropriate to generate a surface (grid) from population counts such as total persons, total minority, or total low-income. It is therefore necessary to compute population percentages from the various population surfaces using map

39 algebra after the surfaces have been created. For example, dividing a grid of total minority population by a grid of total persons can generate a grid showing percent minority. Map algebra can be performed in any GIS software package that supports grids. Data needs, assumptions, and limitations. Census data and census geography are used as inputs in constructing population surfaces. GIS is required to convert the polygon-based population map to a grid-based population map. This function can be performed in most commercial GIS applications. Because this technique uses small-area census data, it is subject to the same assumptions and limitations discussed Methods 2 and 3. Also, it is virtually impossible to perform the polygon-to- surface data conversion without introducing some error into the data. Population surfaces therefore include errors contained in the original census data as well as errors introduced during the conversion. There are two conversion error components: data omission and spatial shifting. Data omission occurs when the population for a census unit is not included in the grid-based population surface. Spatial shifting occurs because grid cells cannot represent census unit boundaries as precisely as zonal polygons. If done properly, however, the conversion can be performed with negligible error. Results and their presentation. Results and presentation of protected population information can rely on maps, tables, charts, or graphs, similar to any other GIS-based census data technique. The difference in results and presentation with this method is that the demographic information can be combined with model-derived effects information. Further examples of this technique are given in guidebook chapters that follow. Assessment. Use of population surfaces is a promising and powerful environmental justice assessment technique that has seen little use in the transportation field. One notable system, the System for Planning and Research in Towns and Cities for Urban Sustainability (SPARTACUS) (see pages 331-332) constructs population surfaces for integration with a raster transportation and land use model. This application is described in an NCHRP report (Cambridge Systematics, Inc. 2002). Defining population surfaces is both data intensive and computationally intensive, and it requires a considerable amount of GIS expertise. The method is best suited for analyzing distributive effects of phenomena that are most effectively modeled as a surface, such as air quality or noise. Method 7. Analysis of historical data When to use. Historical information is useful when issues related to environmental justice have occurred over a long period of time or have a historical context. Historical data are also useful to establish population baselines for comparisons to current or more recent data, and they can be useful for identifying population trends. Review of historical data can be especially important as part of various methods for performing population projections, which are discussed later. Analysis. There are generally two basic steps in completing an analysis of historical data:

40 Step 1 – Acquire data. There are many sources of historical data. Data from U.S. censuses beginning in 1990 are readily available nationwide. Many electronic and paper products from the 1980 and previous censuses are available for most of the U.S. In addition to census information, numerous regional and local data sources are available, such as property ownership records, maps, zoning and planning records, aerial photographs, old newspapers, knowledgeable citizens, government reports, and the like. Step 2 – Perform analysis. Many of the techniques discussed in this chapter can be performed using historical data. In cases where historical data will be used as baseline information, it will be important to format the data to overlay or match with the current data sets that are being used. Data needs, assumptions, and limitations. Historical data often are used for a purpose that was not originally planned for or anticipated. As with any data source, it is important to understand why the data were collected and what are their limitations. Ask questions such as “What was the sample size? Were the data collection methods susceptible to either under counting or over counting protected populations? What were the quality control procedures? Are there other data sources available to corroborate this information?” In addition to these issues related to use of historical data, the assumptions and limitations of the specific analytical method also apply. Results and their presentation. One use of historical data is to analyze population trends within a study area. The example in Figure 2-6 is based on a comparison of 1990 and 2000 census data for the Baton Rouge metropolitan area. Minority population percentages for census tracts were compared for the two census years. These data were overlaid with the interstate highway system to identify segments where there was a greater than 10 percentage point increase in minority population from 1990 to 2000. This statistic indicates areas where it may be especially important to address how transportation system changes affect protected populations. Figure 2-6. Interstate highway corridors with a significant increase in minority population from 1990 to 2000 Assessment. Historical data can be used to evaluate long-term population trends and distributive effects of transportation system changes that have occurred in the past. There are numerous data

41 sources available regionally and locally. At a national level, data from the 1980, 1990, and 2000 censuses are readily available sources that can be used to establish population trends. In addition, trend data can be used to project future population characteristics, our next topic. Method 8. Population projections Population projections are as much art as they are science, and the field of demography has for decades endeavored to develop accurate population projection techniques. Worldwide population projections are relatively accurate, because the future population to be estimated is large and population change at the global scale is driven by the relative size of age cohorts and trends in birth rates and death rates that have been predictable for at least the past couple decades. At smaller scales of states and counties, population projections further involve finer-scale cultural issues and emigration/immigration as confounding factors. Of course, the further into the future you project a population, the less accurate the projection is likely to be. Most transportation system changes, especially projects, require use of small-scale demographic data such as tracts, TAZs, and block groups. At this scale, general population growth trends are much more difficult to evaluate and projections are more likely to be thrown off by cultural changes. Examples such as gentrification, local economic changes, and rapid growth in various industry sectors can render useless even the most scientifically rigorous projections. Add to the above issues the fact that for environmental justice assessment the problem is not merely one of identifying the future population in a small area. Rather, you must add to that the need to predict population change for numerous subgroups of the population. In summary, effective population projection for environmental justice assessment of transportation system changes must utilize techniques for (a) estimating small-area populations and (b) predicting population change for multiple population groups. Such techniques are inherently imprecise, so projections of specific populations in small areas can be highly inaccurate, and they should be reevaluated regularly to determine if they remain valid. When to use. Population projections are best used to evaluate transportation planning projects with a time horizon greater than 5 years in circumstances where the size and composition of affected populations is expected to change quite substantially. Examples would be transportation investment plans, long-range plans for cities and counties, and alternatives studies where project development is expected to begin more than 5 years from the time of analysis. Transportation systems are long-term infrastructure investments that will impact the surrounding environment for years or even decades. As such, it is advisable to evaluate all transportation policies, programs, and projects using both current population data and population data projected for some reasonable, informative future-year scenario. However, the complexity and questionable validity of population projections must be weighed against the additional insight they would provide into identifying future effects to protected populations. Because of their limited precision and the considerable effort involved, small-area population projections of specific groups are not likely to be widely applied.

42 Analysis. In its simplest formulation, population projection for any area can be computed using the following equation: Pf = Pc + (B – D) + (I – E), where Pf = future population Pc = current population B = births D = deaths I = immigration E = emigration B, D, I, and E are computed for the future time period minus the current time period (i.e., T2 – T1). This equation can be cast in terms of the total population or for subpopulations such that the sum of Pf over all subpopulations is the total future population. In application, however, accurate measures of births, deaths, immigration, and emigration are difficult to predict and are data intensive. Most methods rely on a combination of census data and symptomatic variables. Symptomatic variables change over time according to a predictable and logical relationship with population. These variables are often collected from administrative records systems. Some examples of symptomatic variables include housing permits, new utility hookups, birth and death records, vehicle registrations, and school enrollment figures. There are various techniques available for developing population projections, and many metropolitan planning organizations (MPOs) develop projections using standard methodologies. These projections are commonly adopted and used for transportation master planning purposes. Growth forecasts are commonly developed every 2 to 3 years for housing, population, and employment. The time horizon for these projections is generally 20 to 25 years. In general terms, there are four projection approaches commonly in use. These approaches are summarized below. See NCRHP (1999) for further discussion. • Demographic models. These models are based on characteristics of the current population and net migration. Required population characteristics include fertility rates, mortality rates, age cohorts, and gender cohorts. Race and ethnicity are often evaluated for large-scale areas such as states and counties. It can be difficult to generate accurate small-area estimates using this technique, as fertility and mortality information are not readily available, so values for larger areas such as counties must be used. • Trend-based models. As mentioned above, these models work by extrapolating historical trends. They are problematic for small-area projection because they do not account for land use change, such as housing developments, that occur during the historical period. • Land use models. Automated models such as MEPLAN, TRANUS, and UPLAN are based on information characterizing vacant land that can be developed and therefore has the potential for greater population capacity, developed land that has a fixed or only slightly variable population capacity, and service information indicating housing unit

43 density and relative attractiveness of different areas. The Projective Land Use Model (PLUM) utilizes census data, place of work, trip lengths, and population capacity information to derive population projection estimates. For more information on the PLUM model, see Tayman (1996). • General plan models. General plan models are based on information available in community master plans and specific development plans, such as for housing subdivisions and planned annexations. They use information from comprehensive plans, zoning codes, and other land use regulations to develop a future picture of population patterns. One benefit of general plan models is that population projections are built from the ground up, meaning that the land use characteristics and planned changes for specific locations are used as the basis of projection. This approach is beneficial for environmental justice assessment because it can be used to derive estimates for small areas, such as tracts and TAZs, as well as information for larger municipalities, counties, and regions. A drawback, however, is that the technique is extremely data intensive, requiring information on changes in land use policies, zoning, general plan updates, residential densities, city limit boundaries, the status of current development proposals, economic trends, job inventories, estimates of population per housing unit, and housing unit vacancy rates. The basic steps of the general plan method are discussed below. The Sacramento Area Council of Governments (SACOG) approach is used as a model for descriptive purposes (SACOG 2001). This method involves two stages. The first stage is to develop total population projections for counties, cities, tracts, and traffic analysis zones. The second stage is to decompose the total population estimate into the subpopulations of interest, such as minority and nonminority. Stage 1 – Develop county and subarea population projections Step 1a. Define analysis zones. It is best to develop the projections for relatively small analysis units, census tracts at a minimum. For transportation planning purposes, it is also advisable to develop projection for TAZs. Larger reporting zones should be aggregations of the base analysis unit. Reporting zones should include counties, cities, and regional analysis districts (RADs). RADs are smaller than cities or counties and can be developed to mirror local community planning areas or census county divisions. Step 1b. Establish base numbers. Develop estimates of the population, housing stock, job inventory, and school facilities and enrollment for the base year. The housing inventory should provide an annual count of residential housing units. These figures can be obtained from permit completions maintained by the building departments of each jurisdiction in the study area. Data must be collected, categorized by building unit type, and allocated to analysis units based on address. Information on demolished housing units and annexed housing units must also be acquired. Group quarters (e.g., military barracks, penal institutions, or college dormitories) should be included in this information. The current population count can be developed by combining information from the most recent census, the housing inventory, and any current population estimates developed by state or federal agencies. In some states it is mandated that locally developed population estimates must correspond to estimates produced by the state demographic agency. In California, for example,

44 population projections for jurisdictions can be no greater than the estimates produced by the state’s Demographic Research Unit. The housing inventory is used to estimate population by analysis unit. This is done by combining housing unit information with base-year census population data and correlations of population by dwelling type, average household size, and housing unit vacancy rates. These factors are combined to generate a base-year household population estimate. Populations for group quarters (available from local agencies) and homeless individuals (available in the census) are added to household population estimates to generate the base-year total population estimate for each analysis zone. For purposes of environmental justice assessment, it is important to identify the protected population characteristics of the base-year population. The employment inventory is established through surveys of the study area. Employment is considered because it is a significant factor in drawing new residents to an area and in retaining current residents. School facilities and enrollment information is collected in order to (a) better plan for educational infrastructure needs and (b) refine age-cohort information. Step 1c. Evaluate holding capacities. Determine the maximum number of jobs and housing units that can be accommodated by each analysis zone. This is the zone’s holding capacity. For housing, compute densities for each type of residential land use from general plan and zoning information. The average housing density can then be applied to the total acreage of each land use type in the analysis zone. These holding capacities are not fixed; they vary as land use changes and must be updated on a regular basis, at least every few years. For jobs, develop an employment yield matrix. The SACOG approach is to develop estimates of number of employees per acre for five employment types: retail, office, medical, manufacturing, and other. Each employment type can be correlated with land use maps and acreages computed for each analysis zone. Multiplying acreage by number of employees per acre for each employment type yields the employment holding capacity for each analysis zone. Estimates for education-related employment can be gathered from surveys of local educational institutions, and the number of education employees in each analysis zone assigned by address. Step 1d. Determine phasing. Phasing is the process of developing growth curves, using current population and employment as the starting point for the base year and holding capacity as the end point for the horizon year. Individual growth profiles should be developed for each unique type of development pattern that exists in the study area. SACOG uses the following four development patterns: Limited room for growth – applied to areas already at or near their holding capacities; Growth occurring currently – steady growth beginning in the base year and continuing through a future build-out date determined from general plan information; Current development static; growth occurring later – areas expected to begin growing at a later date and to continue growing through a build-out date based on general plan information; and Redevelopment – when land use changes alter the housing and employment composition of an analysis zone. In this case, you should base projections on the expected year of redevelopment and the extent of redevelopment.

45 Step 1e. Compute jobs-housing ratio. The jobs-housing ratio is calculated by dividing an analysis zone’s total number of jobs by its total housing units. The ratio is a measure of the mix of employment and housing in a given area. Step 1f. Derive population. Compute the number of households by household type for each analysis zone and for each year and then multiply by the estimated number of persons per household by each household type to calculate the total household population. Add the population living in group quarters. Adjust the population per household and vacancy rates based on the growth profile of the area. Adjustments are related to changes in housing type from single family to multifamily and changes in age of development. Step 1g. Conduct jurisdictional review. As described earlier, plan-based population projection is an iterative process that relies heavily on the knowledge and understanding of planners and decision-makers as to how proposed and expected land use changes will affect population over time. Jurisdictional review is therefore an opportunity for experts to validate results and, if necessary, provide the further information necessary to alter projections. Stage 2 – Estimate populations of protected groups The commonly used small-area projection techniques do not consider protected population characteristics. This stage of the population projection method is a technique for estimating future protected population characteristics based on the results of Step 1f above and trend information. Step 2a. Collect and prepare data. Use census data for the base year to generate predictor variables for the response variables percent low-income population and percent minority population. The following variables utilized in Stage 1 should be considered as candidate predictor variables for both percent low-income and percent minority: housing costs (housing unit values and rental costs), housing unit density, housing unit vacancy rates, single family to multi-family housing ratio, average household size, jobs-housing ratio, unemployment rate, and housing density to housing capacity ratio. For minority population, also consider low-income as a predictor variable and vice versa. Housing cost and unemployment information are both potential predictors of protected populations. These variables must be added to the Stage 1 data collection efforts and future-year estimation efforts in order to be used as predictor variables. Step 2b. Conduct exploratory study. The result of a multivariate analysis is a best-fit curve allowing estimates of the response variable to be derived from known values of predictor variables. A benefit of this approach is that prediction error estimates can be reported. Note that prediction error rates can be high for exploratory observational studies such as that described here. The purpose of the exploratory study is to reduce the set of candidate predictor variables listed in Step 2a to the set to be used in the regression model. Highly intercorrelated predictor variables should be eliminated, as should other variables found to have low correlation with the response variable. A variable reduction procedure should be used to develop correlation matrices and identify candidate predictors that should be retained in the final model. Although it is beyond the scope of this guidebook to provide an in-depth discussion of variable reduction, there are numerous books on applied regression analysis that can be read for further information, such as Neter, Kutner, Nachtsheim, and Wasserman (1996).

46 Step 2c. Refine and select model. The result of Step 2b is a small subset of candidate regression models with a limited number of explanatory variables that provide good predictive ability. Step 2c results in selection of the final model, based on review of residual plots and analyses to identify lack of fit, outliers, and influential observations. Step 2d. Validate model. Model validity is determined based on the ability to make generalized inferences. Validation should be performed in two stages. First, evaluate the ability of the model to predict known values for percent minority and percent low-income from base year analysis zones. Ascertain if the model provides a suitable fit across all ranges of land use types. Next, evaluate the predictions made for the projection years. Alternative step – generate trends. It may be necessary to compute protected population projections from generalized trends if an acceptable regression model cannot be developed. This alternative step is technically less complex than the regression model steps, but it is otherwise less desirable because it does not provide the ability to report prediction errors. Estimates for changes in percent minority and percent low-income are categorical, more representative of trends than of actual population numbers. This may not be a severe drawback, however, because effects being evaluated over a long time horizon tend to be nonspecific in the case of policies and programs. In addition, for projects it is valuable insight to know, for example, that there is no present day concern about distributive effects but that the project area is expected to experience an increase in low-income population during the ensuing 15 years. The approach is similar to Step 1d (Determine phasing). The trends are developed based on general plan information and can be used to estimate percent low-income and percent minority population for each year in the projection. The following trends are provided as examples, but the actual trends must be developed to reflect the characteristics of the study area. Stable community – the area is near housing capacity; and jobs-housing ratio and housing costs are stable. In this situation it could be expected that percent minority and low income would be stable through time. Growing community – new housing is being developed and new jobs are being created in the area. Percent low-income would be expected to remain stable or even fall. Growth in the low-income population would be related to the availability of affordable housing. Changes in percent minority could be based on the racial composition of the expanding workforce. Declining community – upper and middle-income residents are moving out of the area, housing vacancies are increasing, housing costs are stable or declining, jobs-housing ratio is declining. In this situation, percent low income could be expected to increase because affordable housing is important to low-income individuals. Changes in percent minority could be tied to the correlation between race and income in the study area. Redeveloping community – land use changes alter the housing and employment composition of an analysis zone. These changes in turn affect the relative desirability of housing and could affect percent minority and percent low income. In this case you should base projections on the expected year of redevelopment and the nature of redevelopment. Step 2e. Compute protected population statistics. Using the regression model approach, percent minority and percent low income for each analysis zone can be calculated directly from the model. Reporting the confidence intervals allows for further assessment of the certainty as to

47 whether the minority and income composition of the population will change as predicted. Using the trend-based approach, percent minority and percent low income can be adjusted for each projection year based on knowledge of general demographic trends (i.e., general trend of increasing minority population throughout the study area), knowledge of housing unit turnover rates (to understand phasing), and findings from other areas that have experienced similar land use changes. Calculate number of low-income persons and number of minority persons by multiplying by the total population estimate. Step 2f. Conduct jurisdictional review. As with Stage 1, jurisdictional review is necessary to validate findings and to refine the assessment if the findings are deemed to be invalid. Data needs, assumptions, and limitations. Many of the data needs, assumptions, and limitations have been described above. In general, the technique for projecting changes in minority and low-income population groups is data intensive and time consuming. Trend-based or regression-model techniques can be used to derive protected population projections from base- year and future-year data. General population projection techniques such as trending and extrapolation assume that current population patterns will continue through time and are not well suited to small-area assessment. Land use and plan-based techniques work to overcome this limitation by incorporating information on how the study area is planned to change through time. Common projection methods used in practice estimate future year changes in total population but do not predict changes to subpopulations such as protected groups. The method described above is exploratory at best, and considerable effort must be given to fit the method to the characteristics of the study area it is being applied to. Results and their presentation. Results of this method are estimates of future year populations, including percentages and derived counts for minority populations and low-income populations. Present results as you would results for current year estimates. Assessment. Population projections are a difficult, complex, and time-consuming effort, but in certain instances they are a useful means for assessing the environmental justice of transportation policies, programs, and projects with long time horizons. Population projection techniques useful for transportation environmental justice purposes must provide estimates for small areas such as tracts and TAZs, and they should include predictions for protected population groups. The planning-based method presented in this guidebook must be carefully calibrated to the study area in which it is being applied. As with any small-area projection approach, it is important to refine projections so that they are current with planned land use changes, population trends, and economic fluctuations. Method 9. Environmental justice index The environmental justice index (EJI) is a method of scoring the relative level of environmental justice concern for census-reporting units based on population density, minority population, and low-income population factors. Because the EJI uses multiple factors, it is a good method for combining information to show the distribution of protected populations on a single map.

48 When to use. The EJI is particularly effective for showing relative concentrations of minority or low-income populations. It also is suitable as a screening technique to identify areas where detailed assessment and outreach should be conducted. Finally, the EJI can be used as a demographic variable in combination with indicators of the level or presence/absence of environmental effects in statistical evaluations to determine if effects are disproportionately high and adverse (U.S. DOT 2000). Analysis. There are three steps in constructing an environmental justice index: Step 1 – Gather census data and compute demographic variables. Census data should be collected for the study area, and the level of analysis (tracts or block groups) must be selected. Demographic variables should be computed for population density, percent minority population, and percent low-income population. These variables need to be computed for the selected unit of analysis and at the state level for comparison purposes. Step 2 – Calculate EJI factors. The standard EJI is represented by the following formula: EJI = DVPOP x DVMAV x DVECO where DVPOP = degree of vulnerability based on population density DVMAV = degree of vulnerability based on presence of minority population DVECO = degree of vulnerability based on presence of low-income populations These factors are computed as follows. DVPOP Population per square mile Score 0 0 > 0 and < 200 1 > 200 and < 1,000 2 > 1,000 and < 5,000 3 > 5,000 4 DVMAV and DVECO Percent minority or percent low income Score < State average 1 > State average and < 1.33 times the state average 2 > 1.33 times and < 1.66 times the state average 3 > 1.66 times and < 2.0 times the state average 4 > 2.0 times state average 5 In the standard formulation, the EJI thus ranges from 0 to 100. High EJI values indicate that a large population density is present and that a large proportion of that population is minority

49 and/or low-income. It is also possible to modify the EJI to include additional protected population factors for disability, age, and other characteristics. Step 3 – Review maps and identify presence of protected populations. Once the EJI factors and the EJI are computed for each census unit, GIS can be used to produce maps of the study area. The maps will depict areas of high population density and large proportions of minority or low-income population. In many cases, specific neighborhoods can be identified and labeled on the maps. Data needs, assumptions, and limitations. The EJI is best suited for relatively large study areas and in situations where evaluation of small-area census data is desired. Block-group-level data are the most commonly used evaluation unit. Census data from 1990 or 2000 are required for the evaluation units (block groups, tracts, or TAZ), as are state data. For mapping in GIS, the 1990 or 2000 TIGER files for the study area also are needed. Appendix D contains instructions as to how these data can be obtained. The census data can be formatted and variables calculated using spreadsheet software such as Microsoft Excel or database software such as Microsoft Access. The EJI has the limitations endemic to any mathematical index. Indexes are useful for depicting combinations of variables as a single value, which makes them valuable for screening assessments and for mapping. The underlying factors must be used, however, if more detailed analysis is required. For example, an area with an EJI score of 40 to 45 indicates that there is potential for effects to protected populations. The EJI score by itself cannot be used, though, to determine if the area is densely populated (high DVPOP), has a high proportion of minority individuals (high DVMAV), or has a high proportion of low-income individuals (high DVECO). Furthermore, the EJI does not provide meaningful results in areas with relatively uniform population density and protected population characteristics. Results and their presentation. Figure 2-7 is a map of the EJI for block groups in a 1-mile study area surrounding a project corridor. The map clearly depicts areas of minority population and low-income population concentrations, based on EJI score. These areas are often well- defined neighborhoods and can be labeled for illustrative purposes. Figure 2-7. Example presentation of EJI results

50 Assessment. Identifying protected populations through use of the EJI relies upon census data and is best suited to block, block-group, and TAZ levels of analysis. The EJI can be used effectively to identify areas where further environmental justice assessment should be targeted based on demographic characteristics. Results of EJI evaluation are readily presented in simple maps that are easy to interpret. The EJI can also be used in statistical evaluations to identify disproportionate adverse and beneficial effects. Examples of this use of the EJI are given in other sections of the guidebook. As with any census-based evaluation method, local knowledge and public participation efforts should be used to verify that protected populations have been accurately identified within the area of concern. Method 10. Activity space analysis using personal interviews This simple approach can be an effective tool to provide a general idea of the areas that comprise the individual and communal activity spaces of the populations of interest. Such information supplements (it does not replace) insights derived from the foregoing methods that are largely based on the place of residence. When to use. This method is most appropriate for projects that are likely to be low impact and whose effects are expected to occur within a reasonably well-defined communal activity space. In such cases, a sample of residents is likely to provide useful insights on local activity spaces of the residents represented by the sample. Analysis. Space-time activity survey data should be able to help answer the following questions: • What are the important daily activity centers (i.e., relatively frequent destinations) and what routes do individuals take to and from them? • How frequently do members of the community typically access these centers? Are there numerous centers for different subgroups or do a few dominant? • How do the important activity spaces relate to the impact area of the proposed project? Data needs, assumptions, and limitations. The key to any survey-based analysis, of course, is to draw a representative sample that is large enough to provide a clear picture of the population. In this case, the sample should be of members of protected populations in the general area of the community likely to be affected by a proposed transportation project. You should have a general idea of where protected populations move about before carrying out a survey to learn more about this activity space. The most important assumption is that the activity space defined by the survey sample will enable you to assess, with a fair degree of accuracy, the nature and magnitude of probable impacts. A limitation of survey-based analyses of activity spaces is that they afford only a general indication of each respondent’s daily activity space. Thus, in the aggregate, a survey can be expected to provide a general sense of the activity space of a population of interest. Results and their presentation. The survey will enable you to derive a distribution of origin- destination pairs that depict the relative concentration of activity on the part of protected populations. This distribution can be presented graphically using GIS methods. Cells or zones

51 that have a relatively high concentration of activity should be given special attention when analyzing the several types of effects addressed in the chapters to follow. Assessment. It is widely understood that surveys are only as good as the sampling strategy used to select respondents and the questionnaire developed to address the issues. A statistically significant sample size and a well-conceived questionnaire can provide considerable insight into the areas of the community frequently visited by minority and low-income populations. Method 11. Activity space analysis using an abbreviated diary Measuring habitual travel behavior—determining what places a given population consistently travels to and the routes that they use—can be an effective way of measuring communal activity space. Because time and space constraints limit the number of activities that a person can perform on any given day, it is logical to assume that people try to maximize their utility in their choice of daily activities. A variety of research indicates a great degree of stability in weekday daily activities. Schonfelder and Axhausen (2003) found that two to four locations could account for up to 70 percent of all destinations visited in a 6-week period. Their study also indicated that eight locations represent fully 90 percent of the total trips made. Kitamura and van der Horn (1987) also characterized daily participation in different activities—work, leisure, shopping—as very stable. From an environmental justice perspective, the questions of interest are: • What locations do protected populations access on a relatively frequent basis? • How and by what mode do they access these locations? • How much travel variability is there within the population of interest? Habitual behavior studies typically involve a simplified version of a travel diary that asks the time and place of locations visited throughout a specified period. When to use. This option is most appropriate for use when the understanding of a community’s activity space is less certain than is needed for Method 10 but where potential impacts of the project or the data collection needs are not great enough to warrant a more time-consuming and expensive process. Because of the sensitive and personal nature of the data collected, anonymity is a prerequisite to gain the confidence of subjects. This approach requires both sufficient sample size and adequate administration of the equipment. Analysis. While the overall approach to designing the database and extracting the necessary data will vary, space-time activity data for a given study area should at the very least enable you to answer the following questions: • What are the principal daily activity centers? • What are the most common routes to and from these activity centers? • How frequently do members of the community typically access these centers?

52 Data needs, assumptions, and limitations. The habitual travel survey should have the following qualities: 1) An adequate sample size and representation. As noted earlier, a carefully selected sample of sufficient size to mirror the population being represented is critically important. Recruiting the sample and adequately briefing participants is a substantial undertaking. 2) Clear and brief design. The user should be able to complete the diary in a relatively short amount of time. 3) Prompts that do not lead. Questions should strike a balance between stimulating participants’ thoughts and pointing to a certain response. Sample questions could include the following: • Where are the locations to which you travel daily or almost daily (i.e., work, school, day care, and any others you can think of)? How do you typically travel to these locations? Are these locations commonly visited by your neighbors? • What other locations do you regularly (at least once a week) travel to (i.e., stores, restaurants, church, public park, and any others you can think of)? How do you typically travel to these locations? Are these locations also visited by your neighbors? • What other locations do you travel to less than once a week and yet consider an important part of your life outside the home? How do you typically travel to these locations? 4) An adequate time frame. Participants will often vary their activities over a period of days (e.g., one may grocery shop every Thursday). It is thus helpful to give them a few days to complete their diaries to gain an accurate picture of these activities. Results and their presentation. Habitual travel behavior surveys can provide a fairly accurate picture of where protected populations travel within a community. They allow you to distinguish between frequently versus occasionally or rarely visited locations. Once you have an understanding of the general areas where respondents live and the destinations they travel to, it becomes possible to gain considerable insight into activity spaces. As with Method 10, GIS can be used to develop maps and other graphic depictions of relatively common activity spaces and the routes connecting people’s homes to them. Assessment. This method relies on a diary format to define representative destinations and routes to and from them. As with Method 10, it is critical that the selected sample is sufficiently large and representative of the protected populations whose activity space you are interested in defining. It also is essential that the respondents understand exactly what information you are seeking. Because the data needed are quite simple, the questions should be simple and very clear. Method 12. Space-time activity analysis using GIS A more advanced method for exploring how the activity space of protected populations is configured within a given community is a form of daily trip diary. Miller (2001, p. 11) identifies four traditional forms for collecting spatial activity data using diaries:

53 1) Recall methods require subjects to recall and report activities during some previous time period. 2) Stylized recall methods require subjects to report “normal” activities that occur during some typical time period. 3) Diary methods require subjects to record activities in a diary, either in a free-format manner or at predetermined time periods. “Beeper studies” complement this approach by prompting subjects via a pager at selected time intervals to record their current activities. 4) Prospective methods, typically game-based, are employed in conjunction with other methods to investigate the effect of potential changes in the activity environment. Miller has sought ways to combine these traditional forms with GIS to create a robust, sophisticated model for analyzing activity space. Although he acknowledges some inherent flaws, the method combines space-time activity data with new GIS technology to analyze people’s movements. When to use. With a sufficient sample size and appropriate sample selection design, this method can provide a fairly accurate picture of the activity space of the sample frame (i.e., the protected population within a particular area of a community or across a community). As this is a relatively advanced method, it should typically be used under two conditions: (1) when the nature of the project is large enough to warrant the time and expense and (2) when the data collected will help meet the data needs for multiple objectives. Robust data such as this may be necessary to reveal how people—especially low-income and/or minority populations—move and interact in their community differently than the broader population. Analysis. Typical travel diary analysis utilizes a point-based approach to mark the characteristics of points A and B. A path-based approach, however, is likely to better represent the type of disaggregate activity data required for a full environmental justice analysis. Although a more complex data-gathering task, path-based assignment meets the dually important objectives of identifying communal activity spaces and providing a reasonable estimate of the paths linking homes with these spaces. Path-based approaches can encompass a variety of methods for gathering data on personal activity space. A modern approach is to provide the respondent with a GPS receiver combined with recording devices [e.g., personal digital assistants (PDAs) or cellular transmitters tied to location-based services (LBS)]. The data stored by a recording device typically is entered into a GIS file to first calculate the shortest path between two points and then assign the trip along that path. Shaw and Wang’s study (2000, pp. 167-168) separated the data for each trip into four components: • Spatial—trip ends and path; • Temporal—date, start and end times; • Individual—person(s) making the trip; and • Event attributes—trip purpose, travel mode.

54 There are numerous potential benefits of this approach, including the ability to support space- time queries. A complicating factor is that the respondent must enter the trip purpose into the recording device as there is no way to identify this information from the trip geography. Data needs, assumptions, and limitations. A critically important component of this analysis is sample selection. It is essential that you select, recruit, and fully brief a representative sample of the low-income population and minority population that potentially would be affected by the proposed project. Special attention may be required to prepare illiterate or non-English speaking participants. There are several steps involved in applying this method. Using a path-based approach, the GIS database should include addresses (or dummy equivalents) and an interactive address-matching process to identify common locations and expedite the coding process. A GPS receiver should be tied into the database, and a data-recording device is needed to develop a record of respondents’ activity space. Using this basic approach, Shaw and Wang (2000) successfully matched 85 percent of more than 6,000 trip ends to the GIS streets database. An alternative approach is to use a cellular data transmitter to forward data obtained using a GPS receiver to a central data management facility. With this approach, it is possible to gather path- based data that give a very clear indication of respondents’ travel patterns in time and space. As with the preceding methods, the assumption must be made that participants’ activity patterns are representative of the population of interest. Another assumption is that participants will not alter their activity patterns as a result of taking part in the study. A limitation of this type of analysis is that it is static in nature. The travel patterns identified using this approach may not reflect the activity spaces that would develop if, over time, new employment centers, shopping centers, and other facilities were to emerge. Thus, it may not accurately predict the future state of things if the proposed transportation project were to move forward. Results and their presentation. While the database design may vary, space-time activity data should at the very least be able to answer three questions for a given study area: • What are the vital individual daily activity centers? • Where are frequently visited communal centers? • How and when do members of the community access these centers? Because this method utilizes such sophisticated data collection and processing needs, it has considerable potential for providing you with a rich data repository able to satisfy a variety of analysis needs across a range of environmental justice-related issues. Assessment. Due to the sensitive and personal nature of the data collected, assurance of absolute anonymity is a prerequisite in order to gain the confidence of subjects. To ensure this, sample size must be sufficient and administration of the equipment adequate. If the data are collected heeding the points raised in this discussion, a very good sense of the activity space of the

55 sampled population can be obtained. The insights provided can be extremely useful in assessing the extent to which positive and negative impacts of a proposed project would be experienced disproportionately by low-income and minority populations. RESOURCES 1) Carlstein, T. (Editor). 1978. Timing, Space and Spacing Time Volume 2: Human Activity and Time Geography. London: Edward Arnold. Of particular interest are Chapter One, “Human Time Allocation in the City” by F. Stuart Chapin, and Chapter Four, “Rhythms of Urban Activity” by Mary Shapcott and Phillip Steadman. Chapin uses a conditional response model of human behavior, noting the relationship between the necessity of activities and how they are prioritized in terms of choice and timing. Shapcott and Steadman discuss the interconnectivity, coordination, and routine of daily activity patterns. 2) Census 1990 and Census 2000 TIGER files for use in GIS-based analysis are available free for download at http://www.esri.com/data/download/census2000_tigerline/index.html. These files include useful demographic information for some census reporting units. 3) Census 2000 Summary File 1 and Summary File 3 data sets are available from the census bureau at http://www2.census.gov/census_2000/datasets/Summary_File_1/ and http://www2.census.gov/census_2000/datasets/Summary_File_3/. These files can be downloaded and used with spreadsheet, database, and GIS programs to calculate demographic variables useful in identifying protected populations. 4) Census 1990 Summary File 3 data sets are available from the census bureau at http://www2.census.gov/census_1990. The files are online copies of the Census Bureau’s Summary Tape File 3 CDs. 5) Kitamura, Ryuichi, Patricia L. Mokhtarian, and Laura Laidet. 1997. “A Micro-analysis of Land Use and Travel in Five Neighborhoods in the San Francisco Bay Area.” Transportation, Vol. 24, No. 2 (May), pp. 125-158. This article examines the effects of land uses, socio-demographic characteristics, and attitudinal characteristics on travel behavior for five diverse San Francisco Bay Area neighborhoods. It employs models for numerous measures of travel behavior and confirms that neighborhood characteristics add significant explanatory power when socio-economic differences are controlled for. 6) Kitamura, Ryuichi, Takamasa Akiyama, Toshiyuki Yamamoto, and Thomas F. Golob. 2001. “Accessibility in a Metropolis - Toward a Better Understanding of Land Use and Travel.” Transportation Research Record 1780. Washington, DC: Transportation Research Board, National Research Council, pp. 64-75. This article uses several accessibility indices to determine how accessibility affects aspects of long-term and short-term travel behavior in an urban area. It uses data from the Kyoto-

56 Osaka-Kobe area and southern California to examine a variety of conjectures regarding time availability, accessibility, and engagement in activities. 7) Kitamura, Ryuichi, and Toon van der Hoorn. 1987. “Regularity and Irreversibility of Weekly Travel Behaviour.” Transportation, Vol. 14, No. 2 (May), pp. 227-251. Kitamura and van der Hoorn use weekly travel diaries to analyze the regularity and persistence of daily activities (work, shopping, and recreation). Their study examines the “behavioral lag” between routine in activity participation and changes in socioeconomic and other factors. 8) Kuhn, Walter. 2001. “Ontologies in Support of Activities in Geographical Space.” International Journal of Geographic Information Science, Vol. 15, No. 7, pp. 613-631. This paper seeks to represent human activities and the objects to which activities are directed as the basic units of analysis. 9) Schlich, Rober, and Kay Axhausen. 2003. “Habitual Travel Behaviour: Evidence From a Six- week Travel Diary.” Transportation, Vol. 30, No. 1 (February), pp. 13-36. Schlich and Axhausen have synthesized several methods that measure habitual travel behavior. They discuss data types and methods of calculating similarity. 10) Shaw, Shih-Lung, and Dongmei Wang. 2000. “Handling Disaggregate Spatiotemporal Travel Data in GIS.” GeoInformatica, Vol. 4, No. 2 (June), pp. 161-178. Frequently referred to throughout Method 3, Shaw and Wang’s article provides more detail regarding data representation issues, reducing data redundancy, query types, and other relevant information. REFERENCES Amrhein, C.G. 1995. “Searching for the Elusive Aggregation Effect: Evidence from Statistical Simulations.” Environment and Planning A, Vol. 27, No. 1 (January), pp. 105-119. Bloom, L.M., P.J. Pedlar, and G.E. Wragg. 1996. “Implementation of Enhanced Areal Interpolation using Map Info.” Computers and Geosciences, Vol. 22, No. 5, pp. 459-466. Cambridge Systematics, Inc. 2002. Technical Methods to Support Analysis of Environmental Justice Issues. Transportation Research Board, National Research Council. Washington, DC: National Academy Press. Forkenbrock, David J., and Glen E. Weisbrod. 2001. Guidebook for Assessing the Social and Economic Effects of Transportation Projects. NCHRP Report 456. Transportation Research Board, National Research Council. Washington, DC: National Academy Press. Also available at http://trb.org/trb/publications/nchrp/nchrp_rpt_456-a.pdf. Fotheringham, A.S., and D.W.S. Wong. 1991. “The Modifiable Areal Unit Problem in Multivariate Statistical Analysis.” Environment and Planning A, Vol. 23, No. 7 (July), pp. 1025-1044.

57 Goodchild, M.F., and N.S. Lam. 1980. “Areal Interpolation: A Variant of the Traditional Spatial Problem.” Geo-Processing, Vol. 1, pp. 297-312. Martin, D. 1996. “An Assessment of Surface and Zonal Models of Population.” International Journal of Geographical Information Systems, Vol. 10, No. 8 (December), pp. 973-989. Miller, Harvey J. 2001. “What About People in Geographic Information Science?” Salt Lake City: University of Utah. Available at http://www.geog.utah.edu/~hmiller/papers/ what_about_people.pdf. Mrozinski, R.D., and R.G. Cromley. 1999. “Singly- and Doubly-Constrained Methods of Areal Interpolation for Vector-Based GIS.” Transactions in GIS, Vol. 3, No. 3 (June), pp. 285-301. NCRHP. 1999. Land Use Impacts of Transportation: A Guidebook. NCHRP Report 423a. Transportation Research Board, National Research Council. Washington, DC: National Academy Press. Neter, John, Michael H. Kutner, Christopher Nachtsheim, and William Wasserman. 1996. Applied Linear Regression Models. Third edition. Chicago, Illinois: McGraw-Hill College. Pipkin, John S. 1986. “Disaggregate Travel Models.” In The Geography of Urban Transportation, Susan Hanson (editor). New York: The Guilford Press. Robinson, W. 1950. “Ecological Correlations and the Behavior of Individuals.” American Sociological Review, Vol. 15, pp. 351-357. Sacramento Area Council of Governments (SACOG). 2001. Documentation: Projections of Population, Housing, and Primary and Secondary Students. Available at http://www.sacog.org/demographics/proj2001/overview.htm. Schonfelder, S., and K.W. Axhausen. 2003. “Activity Spaces: Measures of Social Exclusion.” Transport Policy, Vol. 10, No. 4 (October), pp. 273-286. Sheeley, Jason, and David J. Forkenbrock. 2002. Washington, DC: National Cooperative Highway Research Program. Interim Report: Effective Methods for Environmental Justice Assessment. Austin, TX: URS Corporation. Tayman, J. 1996. “The Accuracy of Small-Area Population Forecasts Base on a Spatial Interaction Land Use Modeling System.” Journal of the American Planning Association. Vol. 62, No. 1 (Spring), pp. 85-99. U.S. DOT. 2000. Environmental Assessment of the Proposed Longhorn Pipeline System (November). Washington, DC.

Next: Chapter 3 - Air Quality »
Effective Methods for Environmental Justice Assessment Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 532: Effective Methods for Environmental Justice Assessment is designed to enhance understanding and to facilitate consideration and incorporation of environmental justice into all elements of the transportation planning process, from long-range transportation systems planning through priority programming, project development, and policy decisions. The report offers practitioners an analytical framework to facilitate comprehensive assessments of a proposed transportation project’s impacts on affected populations and communities.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!