Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 260


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 259
places twice as much dollar value on the benefit of cutting travel time to their workplace as on the visual quality of a traffic corridor, that ratio can be factored into the analysis of the overall impact of a project on that population. Using relative valuation in this way obviates the problem caused by the fact that different populations may have different ideas about the value of a dollar, as might be anticipated when comparing, say, a low-income population with a high-income population. It also reduces any possible concern about the absolute accuracy of the estimates. The following sections contain detailed descriptions of the visual preference survey and contingent choice methods. Much of the material describing CC is derived from uncopyrighted material found on the Ecosystem Valuation Web site (King, et al. undated; see www. ecosystemvaluation.org). In addition, a method is described for analyzing distributive effects of visual quality using either objectively or subjectively weighted visual quality information. METHODS Table 11-2 summarizes the methods for analyzing visual effects described in the remainder of this chapter. Table 11-2. Summary of methods for analyzing visual effects Assessment Appropriate Use Data Expertise Method level uses when needs required 1. Visual Screening/ Select among Visual quality impacts are Low Survey methods, preference detailed design choices to be analyzed apart from statistical methods survey or compare other impacts standards and values of populations 2. Contingent Detailed Compare Visual quality impacts are Medium Survey methods, choice standards and to be analyzed vis--vis statistical methods, experiments values of other factors or when a economic analysis populations dollar estimate of visual quality impact is desired 3. Distributive Screening/ Analyze Differences in population Medium/ Statistical methods, effect detailed distributive standards and values are high GIS analysis effects deemed important Environmental justice assessment of visual quality effects consists of four major steps: 1. Identification of protected populations (covered in Chapter 2). 2. Identification of standards and values of the impacted populations. 3. Design and communication of the visual impacts to the affected populations (covered in "Visual quality design and communication techniques," earlier in this chapter). 4. Analysis of distributive effects. 265

OCR for page 259
This sequence is recommended because in most situations the key to environmental justice will be making design and visual aesthetic decisions that meet the approval of the populations that will have to look at the project and of the populations that will use the project. Identifying whether or not these populations include protected population groups (Step 1) and, if so, identifying the unique standards and values of those groups (Step 2) is crucial to making the best design decisions (Step 3). Distributive effects assessment (Step 4) can then be used to evaluate various design alternatives and the distribution of adverse and beneficial visual quality effects. Our discussion of "Selecting an appropriate method for analysis" provides an overview of techniques for identifying standards and values of protected populations and a method for evaluating distributive visual quality effects. Three of these methods are described in more detail below. Method 1. Visual preference survey (VPS) When to use. The VPS method can be used to determine which of a set of possible design choices is most appealing from a visual quality perspective. It may also be used to estimate the value placed on particular aspects of visual quality or on changes in visual quality resulting from a transportation project. By conducting equivalent surveys on groups of respondents representing different protected and nonprotected populations, the standards and values of impacted populations can be characterized and integrated into subsequent analyses of distributive effects and environmental justice. Population standards and values may be expressed in terms of the relative appeal or aversiveness of a particular visual feature or design option. Analysis. This method consists of the following five steps. Step 1 Define the valuation problem. Determine the visual quality impacts to be assessed, and the relevant protected and nonprotected populations. Use interviews or focus groups to broadly identify which aspects of visual quality are of concern to the affected populations. For these preliminary inquiries, images of similar projects or computer simulations of the proposed project may be used to elicit comments. Step 2 Make preliminary survey decisions. Make preliminary decisions about the survey itself, including how images will be presented (slides, computer projector, video), how many respondents will be surveyed in each session, and whether photographs of similar, completed projects or computer simulations will be used for the actual survey. If any of the protected populations to be studied include a significant number of members for whom English is not their first language, conduct the survey in the preferred language of the population. Step 3 Survey design. Select scenes or images that illustrate the visual quality issues of concern to the impacted populations. Furthermore, select a range of levels of impact for each aspect of visual quality. For example, if foliage was identified as an important aspect of visual quality, use multiple images showing different types and amounts of foliage. Generally speaking, no more than 80 images should be used. Step 4 Survey implementation. The first implementation task is to select the survey sample. Ideally, the sample should be a randomly selected group of participants from each relevant population, gathered using standard statistical sampling methods. 266

OCR for page 259
Because it is reasonable to anticipate that there may be wide variation in survey results within each subpopulation, a minimum of 50 respondents from each should be surveyed. It is best to divide the sample of respondents from each population into multiple small groups that are surveyed in separate sessions. The order in which the images are presented should be varied randomly across sessions to minimize order effects in the analysis. Respondents should be asked to rank each image on a 21-point (-10 to +10) scale, where the highest negative score indicates extreme aversion and the highest positive score indicates extreme appeal. Respondents should be given a score sheet on which to enter their responses. The score sheet should include a thumbnail of each image to ensure that the responses are properly matched with the images. If the order in which images are presented to each group is random, arrange the score sheets for each group in the same order as the images presented. Respondents should be instructed not to make comments or otherwise indicate their reaction to each image during the survey so that peer pressure does not bias scores. Step 5 Compile, analyze, and report the results. VPS findings can be used for several different purposes, and different methods of analysis are used depending on the intended goal of the survey. For example, VPS findings could be used to determine the most attractive (or least aversive) design for noise abatement or visual screen walls for a particular population. VPS findings could also be used to establish relative attractiveness of, or degree of aversion to, a visual feature across two or more populations. Finally, VPS could be used to estimate differences in standards and values between protected and nonprotected populations--for example, to establish how highly each population values foliage as a visual quality asset. To determine the most appealing design for a visual feature, simply compute the mean score for each alternative design within each population. The design with the highest mean score is preferred by that population. If the objective is to establish a relative value for attractiveness or aversiveness of a particular visual feature, simply use each population's mean score for that feature. However, you should take note if there is a low level of agreement among respondents within a population. Any major disagreements should be resolved through the use of focus groups or other methods for consensus gathering. Differences in standards and values between the various populations regarding broad aspects of visual quality may be analyzed as follows: For each aspect of visual quality studied, group the images showing different levels of the aspect. Perform the subsequent analysis steps for each grouping of images. Compute the range of each respondent's scores for the group of images. This value reflects that respondent's sensitivity to that aspect of visual quality. Combining all respondents from all populations, rank order the sensitivity scores. Divide the ranked sensitivity scores into three groupings--high, medium, and low, such that each grouping contains approximately 1/3 of the respondents. For each population studied, compute the number of respondents exhibiting high, medium, and low sensitivity. Cast these numbers into a 3 X n table (n is the number of population groups studied). 267

OCR for page 259
Use the chi-square test to determine whether the distribution of high, medium and low sensitivities differs among populations. Chi-square is the preferred statistical method because it does not rely on an assumption of the normality of the underlying distributions (see pages 105-108). Data needs, assumptions, and limitations. VPS has relatively low requirements for external data because the majority of data are generated from the survey. Demographic data are required to determine if there are protected populations within the area of concern. Keep in mind the possible pitfalls associated with using VPS for estimating the relative appeal of a particular visual feature or for comparing the standards and values of populations: There may be differences between populations in terms of the degree to which they express extreme likes or dislikes. If one population is relatively more stoic or more strongly opinionated overall, this will influence the outcome. There may be a confounding of visual quality factors with other aspects of an image. For example, if you present images illustrating various levels of truck traffic, the respondents' scores may reflect their attitudes toward noise or congestion in addition to their attitudes toward visual quality. By using ranges within groupings of images as an indication of respondents' sensitivity to different aspects of visual quality, any differences between individuals or populations in the average response to those groupings are lost. For example, one population may react more negatively to all images of a particular scene regardless of differences among the images in, say, the number of trees. Results and their presentation. Results are presented in tabular form or using bar charts to illustrate population differences. Parametric statistics such as the mean and standard deviation may be used for descriptive purposes, but nonparametric statistics such as chi-square should be used for inferential purposes. Assessment. Visual preference surveys involve asking a group of respondents to rate each of a set of images according to their perceived attractiveness or aversiveness. Ratings consist of scores for each image on a 21-point hedonic scale. Results may be used to select among design choices; to establish an absolute measure of attractiveness or aversiveness of a particular visual feature; or to evaluate population differences in standards and values regarding broad aspects of visual quality. Method 2. Stated preference/contingent choice (SP/CC) The CC method is a hypothetical method--it asks people to make choices based on a hypothetical scenario. It differs from the CV method in that it does not directly ask people to state their values in dollars. Instead, values are inferred from the hypothetical choices or tradeoffs that people make. There are a variety of formats for applying contingent choice methods, including: 268

OCR for page 259
Contingent ranking Contingent ranking surveys ask individuals to compare and rank alternate project impacts with various characteristics, including costs. For instance, people might be asked to compare and rank several mutually exclusive roadway beautification projects under consideration for a travel corridor, each of which has different outcomes and different costs. Respondents are asked to rank the alternatives in order of preference. Discrete choice In the discrete choice approach, respondents are simultaneously shown two or more different alternatives and their characteristics and asked to identify the preferred alternative. Paired rating This is a variation on the discrete choice format, where respondents are asked to compare two alternate situations and to rate them in terms of strength of preference. For instance, people might be asked to compare two roadway beautification projects and their outcomes, to state which is preferred, and to indicate whether it is strongly, moderately, or slightly preferred to the other program. Whatever format is selected, respondents' choices are statistically analyzed using discrete choice statistical techniques to determine the relative values for the different characteristics or attributes. If one of the characteristics is a monetary price, then it is possible to compute the respondent's willingness to pay for the other characteristics. When to use. The contingent choice method asks the respondent to state a preference between one group of environmental services or characteristics (a scenario) and another. A typical CC survey might comprise 50 to 100 such choices. If an estimate of the dollar value of each characteristic is desired, a monetary characteristic is included in the set for each scenario. Because it focuses on tradeoffs among scenarios with different characteristics, contingent choice is especially suited to situations where a project or policy might result in multiple different impacts on a population. For example, a highway project might impact accessibility to the workplace, noise levels, and safety, in addition to visual quality. While contingent choice can be used to estimate dollar values, the results may also be used simply to rank impacts, without focusing on dollar values. Analysis. This method consists of the following five steps. Step 1 Define the valuation problem. Determine the visual quality and other impacts to be assessed and the relevant protected populations. Note that the size and complexity of the CC survey increases at roughly the square of the number of different impacts analyzed. For this reason, it is best to limit the number of impacts analyzed to five or fewer. Step 2 Make preliminary survey decisions. The second step is to make preliminary decisions about the survey itself, including whether it will be conducted by mail, phone or in person, how large the sample size will be, who will be surveyed, and other related questions. The answers will depend, among other things, on the importance of the valuation issue, the complexity of the question(s) being asked, and the size of the budget. In-person interviews are generally the most effective for complex questions, because it is often easier to explain the required background information to respondents in person and people are more likely to complete a long survey when they are interviewed in person. In some cases, visual 269

OCR for page 259
aids such as videos or color photographs may be presented to help respondents understand the conditions of the scenario(s) that they are being asked to rate. While in-person interviews are generally the most expensive type of survey, mail surveys that follow procedures aimed at obtaining high response rates can also be quite expensive. Mail and telephone surveys must be kept fairly short, or response rates are likely to drop dramatically. Telephone surveys are generally not appropriate for CC surveys because of the difficulty of conveying the tradeoff questions to people over the telephone. If any of the protected populations to be studied include a significant number of members for whom English is not their first language, consider communicating (interviews, focus groups, questionnaires) in the preferred language of the population. Step 3 Survey design. This is the most important and difficult part of the process and may take 6 months or more to complete. It is accomplished in several steps. The survey design process usually starts with initial interviews and/or focus groups with the types of people who will be receiving the final survey, in this case members of each subpopulation to be studied. In the initial focus groups, the researchers would ask general questions, including questions about peoples' understanding of the issues related to the project, their familiarity with the area of impact (e.g., the travel corridor), and the value they place on the area and its attributes. In later focus groups, the questions would get more detailed and specific, to help develop specific questions for the survey, as well as to decide what kind of background information is needed and how to present it. For example, people might need information on the location and characteristics of the project and what impacts it will have on various environmental, social, and economic aspects of their lives. At this stage, the researchers would test different approaches to the choice question. Usually a CC survey will ask each respondent a series of choice questions, each presenting different combinations and levels of the relevant impacts, possibly including the cost to the respondent associated with each scenario. Each scenario might be described in terms of impact on travel time between two particular points, visual quality (e.g., billboards, trees, and beautification measures), tax burden, etc. The visual quality impact for each scenario should be illustrated using an image. Images may be presented as hard copies or projections. After a number of focus groups have been conducted and researchers have an idea of how to provide background information, describe the hypothetical scenarios, and ask the choice questions, they will start pretesting the survey. People would be asked to fill out the survey. Then the researchers would ask respondents about how they filled it out and let respondents ask questions about anything they found confusing. The researchers would continue this process until they had developed a survey that people seemed able to understand and answer in a way that made sense and revealed their values for the visual quality features being addressed. Step 4 Survey implementation. The first task here is to select the survey sample. Ideally, the sample should be a randomly selected group of participants from each relevant population, selected using standard statistical sampling methods. Although CC surveys are sometimes conducted via mail or phone interview, CC surveys that include visual quality factors should be conducted in a controlled setting. In-person surveys may be conducted with random samples of respondents or may use "convenience" samples--asking people in public places to fill out the survey. 270

OCR for page 259
Step 5 Compile, analyze, and report results. The final step is to compile, analyze, and report the results. The statistical analysis for CC is often more complicated than that for CV, requiring the use of discrete choice analysis methods to infer valuation or willingness to pay from the tradeoffs made by respondents. A full description of the analysis method is beyond the scope of this document; see Louviere et al. (2000) for a thorough discussion of the method. From the analysis, you can estimate the average value for each of the impacts of the project for an individual or household in the sample. This can be extrapolated to the relevant population to calculate the total benefits from the site under different scenarios. The average value for a specific action and its outcomes can also be estimated, or the different policy options can simply be ranked in terms of peoples' preferences. Data needs, assumptions, and limitations. CC methods have relatively minimal requirements for external data because most of the data are generated from the interviews and surveys that comprise the study itself. To collect useful data and provide meaningful results, the CC survey must be properly designed, pretested, and implemented. However, because responses are focused on tradeoffs, rather than direct expressions of dollar values, contingent choice may minimize some of the problems associated with contingent valuation. Often, relative values are easier and more natural for people to express than absolute values. While CC generally is an excellent method for studying a population's values, the following limitations should be addressed when designing a study: Respondents may find some tradeoffs difficult to evaluate because they are unfamiliar. The respondents' behavior underlying the results of a CC study is not well understood. Respondents may resort to simplified decision rules if the choices are too complicated, which can bias the results of the statistical analysis. If the number of attributes or levels of attributes is increased, the sample size and/or number of comparisons each respondent makes must be increased. When presented with a large number of tradeoff questions, respondents may lose interest or become frustrated. By providing only a limited number of options, respondents may be forced to make choices that they would not voluntarily make. Contingent ranking requires sophisticated statistical techniques to estimate willingness to pay. Results and their presentation. The discrete choice analysis methods used to analyze the results of a CC study yield weights that represent the value placed on each factor studied relative to the other factors. For this reason, it is desirable to select one factor, such as dollar cost, that can be used as a standard for comparisons among the other factors. Results may be presented in tabular form or bar charts that describe the differences among populations in relative valuation of the factors studied. 271

OCR for page 259
Assessment. The CC method can be used to establish how populations value a project as a whole, as well as the various visual quality attributes or visual quality effects of the project. The method allows respondents to think in terms of tradeoffs, which may be easier than directly expressing dollar values. In addition, respondents may be able to give more meaningful answers to questions about their behavior (i.e., they prefer one alternative over another), than to questions that ask them directly about the dollar value of a social or economic impact or the value of changes in environmental quality. Thus, an advantage of this method over the CV method is that it does not ask the respondent to make a tradeoff directly between environmental quality and money. Method 3. Distributive effects analysis When to use. Use distributive effects analysis when screening data indicate that there may be a disproportionate distribution of visual effects with respect to protected populations in the project impact area. Analysis. This method consists of the following three steps. Step 1 Quantify the distribution of protected populations. Before proceeding with any more detailed analysis, determine whether any protected populations are disproportionately represented in the project area as a whole or whether they are unevenly distributed within the project area. Because visual quality impacts people in their homes as well as when they travel, both population distribution analysis and use analysis should be performed. Population distribution analysis may be performed using the Environmental Justice Index or one of the other methods described in Chapter 2. Use analysis may be performed using one of the transportation demand methods described in Chapter 7. If protected populations are proportionally represented in the project impact area as a whole and are uniformly distributed within the project area, no further analysis need be performed. If protected populations are not proportionally represented or are unevenly distributed within the project area, the maldistribution should be quantified. The project impact area should be divided into analysis areas that can be characterized with respect to the number of members of each protected and nonprotected population group that live in or use each area. Step 2 Quantify the level of visual quality impact on each affected population. The level of visual quality impact on each affected population can be quantified using objective or subjective measures of impact. Objective measures can be based on counts of visual features, such as number of trees planted or number of billboards in each analysis area. An economic measure might also be used, such as the amount spent on beautification or visual screens in each analysis area. In either case, the estimate of visual quality impact is based simply on objective information and does not differ between protected and nonprotected populations. Subjective measures make use of information gathered using VPS, CC, or some other measure of the value placed on particular visual quality features by the various affected populations. The level of subjective impact may not be the same for all populations. For example, suppose that a VPS revealed that a protected population rated a particular visual feature (such as a screening 272

OCR for page 259
wall) as 3 on a 21-point hedonic scale, whereas the nonprotected population rated the same feature as +2. Those values could be used as direct measures of visual impact. It is also possible to combine subjective measures with objective measures. For example, if VPS or CC revealed that a protected population was 25 percent more sensitive than the nonprotected population to a particular type of visual quality effect (e.g., in their aversion to billboards), that could be used as a factor to estimate total aversiveness for each population. In this example, the impact of billboards on visual quality for the nonprotected population could be quantified simply as the number of billboards anticipated, whereas the impact on the protected population could be quantified as 1.25 times the anticipated billboard count. Step 3 Combine the visual quality impact metrics with the population distribution data. This step involves computing the visual quality impacts--either objectively or subjectively measured--for each protected and nonprotected population within each analysis area. That is, for each area, the following calculations are performed: Tpa = I p N pa where Tpa = total visual quality impact on population p in analysis area a Ip = level of visual quality impact on population p as calculated in Step 2 N pa = number of members of population p in analysis area a Values of Tpa for each population and area are cast into an n X m table, where n is the number of populations and m is the number of areas. A Friedman two-way analysis of variance or similar nonparametric test may then be used to determine whether there is a significant interaction effect between populations and areas in total visual quality impact. A significant interaction effect would indicate that the visual quality impacts were not proportionately distributed among protected and nonprotected populations. Data needs, assumptions, and limitations. These considerations depend on the methods selected to estimate the impact of visual quality factors on different populations. One limitation of this method is that it does not account for any additive effects, such as disproportionate distribution of counterbalancing benefits. 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. Many examples of results and their presentation are included in Chapter 2. Assessment. This method allows you to combine objective or subjective measures of visual quality impact with information about the demographics of protected populations. 273