PRESERVATION VALUES FOR VISIBILITY
PROTECTION AT THE NATIONAL PARKS
Draft Final Report
Prepared by:
Lauraine G. Chestnut
Robert D. Rowe
RCG/Hagler, Bailly, Inc.
P.O. Drawer 0
Boulder, Colorado 80306-1906
(303) 449-5515
Prepared for:
Economic Analysis Branch
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
and
Air Quality Management Division
National Park Service
Denver, Colorado
Partially Funded Under:
U.S. EPA Cooperative Agreement No. CR813686
Research Grant Awarded to the Regents of
the University of Colorado Center for Economic Analysis
Boulder, Colorado
February 16, 1990
[PARV5]
-------
DISCLAIMER
Although prepared with EPA funding, this report has
neither been reviewed nor approved by the U.S.
Environmental Protection Agency for publication as an
EPA report. The contents do not necessarily reflect the
views or policies of the U.S. Environmental Protection
Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for
use.
-------
I
ACKNOWLEDGEMENTS
Many individuals greatly enhanced the quality and enjoyment of this research. Most
importantly, we thank the nearly 1,800 anonymous citizens who participated in the varying
stages of the survey research. While not every comment from these citizens could be
included in the report, we have attempted to be faithful in accurately reporting the
information provided to us.
Parts of this work were funded by the National Park Service, Air Quality Management
Division, and by the U.S. Environmental Protection Agency, Office of Air Quality Planning
and Standards, Economics Analysis Branch, through contracts with the University of
Colorado, Mathtech, and directly with RCG/Hagler, Bailly, Inc. We particularly thank
Leland Deck, Allen Basala, and Alan Carlin, of the U.S. EPA; and David Joseph and
Christine Shaver of the NPS who supported the study with research monies, contract
management, and other important assistance when needed. The NPS, working with Air
Resource Specialists of Fort Collins, Colorado, provided the slides and photographs used
in the survey.
We greatly benefitted from thoughtful review and comment of many colleagues at various
stages of the work. These colleagues included William Schulze, Leland Deck, Bruce
Madariaga, Robert Horst, Doug Rae, William Key, Don Dillman, Dave Ross, William
Malm, John Trijonis, John Molenar, Christine Shaver, and David Joseph. In house, we
were supported by the fine efforts of Gloria Starry, Andrea Litt, Dale Willingham, and our
entire word processing staff.
RCG/Hagler, BaiHy, Inc.
-------
List of Tables
Table 2.2-1
Typology of Potential Response Effect Biases in CV Studies*
Table 2.3-1
Comparison of Schulze et al. and Tolley et al. Grand Canyon
Visibility Value Results
Table 2.3-2
On-Site Use Values for Visibility Protection at Recreation
Sites: Selected Studies
Table 3.1-1
Summary of Questionnaire Versions
Table 3.2-1
Photographs Used to Illustrate Visibility Conditions
Table 4.1-1
Sampling Plan and Survey Response Rates
Table 4.1-2
Comparisons of Response Rates for Different Park Regions
Table 4.1-3
Comparisons of Response Rates for Different Questionnaire Versions
Table 4.2-1
Comparison of Respondents and National Population Characteristics
Table 4.3-1
Telephone Follow-up Response Rates
Table 4.3-2
Summary of National Park Visitation Characteristics of
Telephone Respondents
Table 4.3-3
Summary of Importance to Telephone Respondents of Visibility
Protection for National Parks in the Selected Region
Table 4.3-4
Willingness of Telephone Respondents to Pay Something for
Visibility Protection for National Parks
RCG/Hagler. Bailly, Inc.
-------
V
List of Tables, Page 2
Table 4.3-5
Mean Adjusted WTP by Response Group
Table 4.4-1
National Park Visitation History
Table 4.4-2
Probability of Future National Park Visitation in Next Five Years
Table 4.4-3
Reasons for Visiting National Parks
Table 4.4-4
Most Frequent "Other" Reasons for Visiting National Parks
Table 4.4-5
Reasons for Preserving National Parks, Regardless of Own Visitation
Table 4.4-6
Most Frequent "Other" Reasons for Preserving National Parks,
Regardless of Own Visitation
Table 4.4-7
Priorities for Protecting National Park Resources from Human
Activity Outside the Parks
Table 4.4-8
Most Frequent "Other" Responses for Priorities in Protecting
National Park Resources
Table 4.4-9
Effect of Visibility Improvement on Park Visit Enjoyment
Table 4.4-10
Willingness to Pay Something for Visibility Protection
Table 4.4-11
Importance of Visibility Protection at National Parks
Table 4.4-12
Non-Response Rates for Selected Questions
4-13
4-15
4-16
4-18
4-19
4-20
4-21
4-22
4-23
4-24
4-25
4-27
4-28
RCG/Haglcr, BaiHy, Inc.
-------
List of Tables, Page Three
Table 4.5-1
Example of Distribution of WTP Responses Across the Checklist 4-30
Table 4.5-2
Frequency of Refusals and Zeros in WTP Responses 4-33
Table 4.5-3
Visibility Focus Verification: Question 17 4-34
Table 4.5-4
High WTP and Non-Response WTP Evaluation 4-35
Table 4.5-5
Frequency Distribution of Adjusted* WTP Responses For
Versions 1, 2, and 3 4-42
Table 4.5-6
Perceived Accuracy of WTP Responses 4-43
Table 4.5-7
All Equal Versus Different WTP Responses 4-46
Table 4.5-8
WTP Means for All Equal Versus Different Responses 4-48
Table 4.5-9
Logit Analysis on whether Respondent Gave the Same WTP 4-49
Table 4.5-10
Means of Adjusted Willingness to Pay Responses by Version and by State 4-52
Table 4.5-11
Average Sample Characteristics by State 4-58
Table 4.5-12
Combined Means of Adjusted Willingness to Pay for the Southwest
Parks by State (Versions 3, 5, and 6) 4-59
Table 4.5-13
WTP1 Means by Likelihood of Future Visit 4-61
RCG/Hagler, BaiHy, Inc.
-------
List of Tables, Page Four
Table 4.5-14
Medians of Adjusted Willingness to Pay by Region of Focus and
by State of Residence
4-62
Table 4.5-15
Significant Positive Correlations with Willingness to Pay
for Visibility Protection
4-64
Table 4.5-16
Regression Results for Adjusted WTP, Versions 1, 2. and 3
4-67
Table 4.5-17
Regression Coefficients for Different Regions for WTP1
4-69
Table 4.5-18
Regression Results Highlighting Effects of Different Versions
for Southwest Parks
4-71
Table 4.5-19
WTP Regression Results with Change in Percentile 4-77
Table 4.5-20
WTP Regression Results with Change in Visual Range (km) 4-78
Table 4.5-21
WTP Regression Results with Log of Ratio of Visual Range (km) 4-79
Table 4.5-22
WTP Regression Results with Change in Average Atmospheric Transmission 4-80
Table 4.5-23
Means and Standard Deviations of Regression Variables 4-81
Table 4.5-24
Average Percentage of Regional WTP for the Individual Park 4-85
Table 4.5-25
WTP Means for individual Parks 4-87
Table 4.5-26
Regression Coefficients for Different Parks for WTP1 4-88
RCG/Hagler, BaiHy, Inc.
-------
Vlll
List of Tables, Page Five
Table 4.5-27
WTP Regressions for Yosemite with Change in Visibility Percentile 4-90
Table 4.5-28
WTP Regressions for Yosemite with Log of Ratio of Visual Range (km) 4-91
Table 4.5-29
WTP Regressions for Grand Canyon with Change in Visibility Percentile 4-92
Table 4.5-30
WTP Regressions for Grand Canyon with Log of Ratio of Visual Range (km) 4-93
Table 4.5-3 1
WTP Regressions for Shenandoah with Change in Visibility Percentile 4-94
Table 4.5-32
WTP Regressions for Shenandoah with Log of Ratio of Visual Range (km) 4-95
Table 4.5-33
Average Percentage Allocation of WTP by Motive 4-97
Table 4.5-34
Correlations Between Responses to Question 19 and Reasons for
Visiting and Protecting National Parks 4-99
Table 4.5-35
Average Option Price, Bequest and Existence Values for WTP1 4-100
Table 4.5-36
Selected Regression Sensitivity Analyses for the Grand Canyon WTP 4-102
Table 4.6-1
Summary of Written Comments 4-104
Table 4.7-1
Comparing Point Estimates of Visibility Value Studies 4-105
RCG/Hagler, BaiHy, Inc.
-------
ix
List of Figures
Figure 3.2-1
Insert Title Page, Version 1 through Version 5 3-20
Figure 3.2-2
Insert Title Page, Version 6 3-21
Figure 3.2-3
Insert Map 3-22
Figure 3.2-4
Insert Photograph Layout, Versions 1 through 5 3-23
Figure 3.2-5
Insert Photograph Layout, Version 6 3-24
Figure 3.3-1
Responses Received Each Week 3-28
Figure 4.5-1
Frequency of Question 17 Responses: Percentage for Visibility 4-37
Figure 4.5-2
Adjusted WTP1 as a Percent of Household Income 4-40
Figure 4.5-3
Frequency of Question 18 Responses: Percentage for Illustrated Park 4-86
Figure 4.5-4
Frequency of Responses for Option Price, Bequest Value, and
Existence Value 4-98
RCG/Hagler, BaiHy, Inc.
-------
1-1
1.0 INTRODUCTION
This document presents the design and results of a study conducted for the U.S.
Environmental Protection Agency, the National Park Service, and the University of
Colorado concerning the estimation of preservation values held by the general public for
the protection of visibility at national parks from air pollution impacts.
1.1 BACKGROUND
Under the Organic Act of 1916, the National Park Service (NPS) is charged with protecting
the resources of areas under its jurisdiction to assure their continued availability for the
enjoyment of the public. The fact that these areas have been set aside reveals a
Congressional intent to preserve the resources, purposes, and values of these unique
national treasures in perpetuity. These actions reveal a value for preserving the resources
that goes well beyond the use of the resource in the current period and perhaps beyond
future use as well.
Because of the increasing use of cost-benefit analysis in various decision-making processes,
there is a need to develop methodologies that can be used to quantify the economic
benefits associated with preserving park resources to assure that credible and useful
information concerning how the public values these resources can be provided to decision-
makers. The costs and benefits of alternative industrial development scenarios or
regulatory approaches are routinely weighed by federal, state, and local agencies during
certain decision-making processes. For example, the Clean Air Act allows regulatory and
permitting agencies to consider costs and benefits before deciding whether to require
installation of retrofit controls on existing stationary sources that are contributing to
visibility impairment in Class I areas. Costs and benefits may also be weighed by
permitting authorities in permitting decisions for new sources that may adversely affect a
scenic vista, but not cause any adverse impact within park boundaries. Finally, the
Environmental Protection Agency has recently suggested that the Clean Air Act might allow
the Agency to consider costs and benefits when establishing secondary national ambient air
quality standards. Secondary standards are designed to prevent "welfare" effects (e.g.,
adverse effects on visibility, soils, water, vegetation, etc.) and are, therefore, critical to
assure protection of park resources.
In addition to the likelihood that costs and benefits of alternative actions will be considered
in decisions made by others, regulatory actions initiated by the EPA and NPS are subject
to the requirements of Executive Order 12291. The order requires that costs and benefits
to society be weighed before new regulations are issued, that a regulation not be issued
unless the potential benefits to society outweigh the costs, and that of the alternative
RCG/Hagler, BaiHy, Inc.
-------
1-2
approaches to a given regulatory objective, an agency must select the alternative involving
the least net cost to society. Even in cases where a statute explicitly excludes economic
considerations from entering into a decision, a regulatory impact analysis must be
performed for new major regulations. The regulatory impact analysis must include a
discussion of the costs and benefits of alternative regulatory approaches.
Current estimation methods and empirical estimates of social benefits associated with
preservation of natural area park resources, and visual air quality at these sites, other than
for current period use, are quite limited. Analyses conducted for the Integral Vista
Regulatory Impact Analysis (Chestnut and Rowe 1983) exemplified the limited information
currently available for use in quantifying benefits of NPS resource protection, especially
with regard to preservation benefits that extend beyond direct benefits to current park
visitors. One study has extensively considered preservation values for visibility protection
at national parks of the Southwest (Schulze e al. 1981); a few other studies have, in a
limited way, addressed issues in this study (" Iley et al. 1986, Rahamatian 1986); and
another only casually addressed the issue for one national park in the eastern U.S. (Rae
1984). These studies provide only limited evidence for visibility-related preservation values
at a handful of national parks. Moreover, the accuracy of the results has been questioned
as practitioners learn more about the design and application of the contingent valuation
method used to obtain these results. The limited nature of the work done to date makes
general application of findings vulnerable to criticism based on questionable transferability
of results from one setting or issue to another. Because of the general lack of agreement
on, or acceptance of, any preservation value estimate, federal, state, and local agencies may
be forced to give less weight to potential social benefits in regulatory analysis.
For all of the above reasons, this study has been designed with the intention of advancing
the state-of-the-art in estimation of preservation values and to produce additional empirical
results that can be used to provide information to decision-makers who are authorized, if
not required, to consider costs and benefits when making regulatory or permitting decisions
affecting visual air quality in and around national parks. Therefore, the objective of the
study is to attempt to establish a set of defensible benefit estimates for visibility protection
for a variety of national parks with sufficient accuracy, reliability, and variety to be useful
in answering broad national policy questions and in addressing specific issues on a case-
by-case basis.
1.2 ORGANIZATION OF THE REPORT
Chapter 2 provides background information on the concepts of visibility values and issues
concerning the use of the contingent valuation method to obtain preservation values for
visibility protection at national parks. Chapter 2 also reviews key literature covering
previous empirical studies of preservation values for visibility protection at national parks,
and other related preservation value studies.
RCG/Hagler, BaiHy, Inc.
-------
1-3
Chapter 3 describes the design and implementation of a new contingent valuation method
(CVM) mail survey used to obtain preservation value estimates for visibility protection in
and around national parks of the Southwest, California, and the Southeast. Included in this
chapter are discussions of how the various features are designed to minimize, test for, and
correct selected potential biases in the CVM instrument.
Chapter 4 presents a detailed summary of results and their implications. Because the
analysis is quite detailed and extensive, Chapter 5 presents a simple bulleted summary of
the key findings, and discussions of the interpretation and use of the results. Chapter 5 also
discusses potential future directions for the use of the CVM method in estimating
preservation values for national park resources.
Sample mail and telephone survey instruments are found in the Appendices.
RCG/Hagler, BaiHy, Inc.
-------
2-1
2.0 ESTIMATING VALUES FOR PROTECTING VISIBILITY AT
NATIONAL PARKS
This chapter introduces the concepts and measures of value for changes in non-market
natural resources that have been developed in the economics literature, and briefly discusses
key issues in the contingent valuation method (CVM), which is used in this study. The
second section of this chapter summarizes the results of previous studies that have
estimated values for changes in visibility at national parks, and discusses selected related
literature concerning estimation of preservation values for natural resources.
2.1 CONCEPTS OF VALUE
Before talking about the "value of visibility" at national parks, it is important to define what
is meant by this in the economics literature. The accepted economic measure of the dollar
value to an individual for a change in the quantity or quality of any good or service is the
change in income that would cause the same (or offsetting) change in the individual's well-
being (utility) as a specified change in the good or service. This measure is commonly
referred to as "consumer's surplus.'" Thus, the information desired for this benefit analysis
is the change in income, for all affected parties, that would cause the same (or offsetting)
change in utility as the change in visibility that is being considered.
For environmental public goods that are not exchanged in a market, there are two ways
traditionally used to define this change in income:
Willingness to nav (WTP) is the maximum dollar amount the individual is willing
to pay to obtain an increase, or prevent a decrease, in the quantity or quality of
the good.
Willingness to accent compensation (WTA) is the minimum dollar amount the
individual is willing to accept to voluntarily forgo an increase, or to accept a
decrease, in the quantity or quality of the good.
One important distinction that has been drawn in the literature is that some value is related
to one's own use of the resource, while some value may not be related to one's own use.
Use values for visibility at national parks are the values associated with the park's visual air
quality during an individual's own on-site visits to a park and through off-site enjoyment of
park features with films, photographs, paintings, etc. Non-use values for visibility at
national parks are values the individual may have for protecting visibility unrelated to his
or her own use, which may be held even if he or she never visits the parks.
1 See Freeman (1979), Just et al. (1982), Morey (1984), Vartia (1983), and Randall and Stoll (1980) for more
rigorous definitions and discussion of welfare value measures for environmental goods based on standard
economic utility theory, and see Rowe and Chestnut (1982) for a discussion of this theory specifically related to
visibility.
RCG/Hagler, BaiHy, Inc.
-------
2-2
The concept of non-use value for natural resources was first elucidated by Krutilla (1967)
who observed that "there are many persons who obtain satisfaction from mere knowledge
that part of wilderness North America remains even though they would be appalled by the
prospect of being exposed to it." The concepts of use and non-use values for natural
resources have received considerable attention in the economics literature and more
rigorous definitions of use and non-use values have been developed since Krutilla first
introduced the concept in the resource economics literature. Important contributions to this
literature have been made by Krutilla and Fisher (1975), Randall and Stoll (1983),
McConnell (1983) Freeman (1988), Cicchetti and Freeman (1971), Krutilla et al. (1972),
Freeman (1984) and others. Several different categories of, and nomenclature for, use and
non-use values have been developed in this literature, including (but not limited to) the
following. These terms are used in this report as defined here, but may be used somewhat
differently by other authors.
Dm ion price. Option price is a measure of use value that reflects uncertainty
regarding future use of a resource. It equals the expected value of impacts upon
current and future use plus a risk premium, which may be positive, negative, or
zero. The risk premium is related to uncertainty regarding desired future use
and the impact of the resource change upon future use, and its sign (positive or
negative) depends on whether the individual prefers to err toward preserving a
resource that may not be wanted for use in the future, or toward losing a
resource that may be wanted for use in the future.
Bequest, value. This is the component of non-use value that is related to the
use of the resource by others now and in the future. This value is typically
thought of as altruistic in nature.
Fxistence value. This is the component of non-use value that is related to
preservation of the resource, even if there is no human use of the resource in
the traditional sense. In practice, bequest and existence values are difficult to
distinguish and are often together referred to as existence value.
Preservation value. This term refers to the total value of a resource and includes
all use and non-use values.
While much of this literature has focused on whether or not a particular natural resource,
such as an endangered species, is to be preserved, the same value categories apply to
changes in the quality of a resource (Freeman 1988). With visual air quality at national
parks, the issue is typically at what level it is to be maintained, not whether or not it will
exist at all.
Option price differs from on-site use value because it is an ex ante measure based on
expected use rather than an ex post measure of value based on actual use, and includes a
risk aversion premium. As the ex ante measure, option price is the appropriate measure
for analysis of proposed regulatory decisions that may affect the quality or availability of
the future use of a resource (Chavas et al. 1986).
RCG/Hagler, BaiHy, Inc.
-------
2-3
2.2 THE CONTINGENT VALUATION ESTIMATION METHOD
2.2.1 Why the Method is Selected
For private market goods, WTP and WTA measures of value can be derived from market
information on prices and quantities of the goods sold. For non-market goods such as
environmental quality, this direct market information is not available and other methods
must be used to estimate these measures of value.
Contingent valuation methods and travel cost/time allocation methods are potentially useful
for estimating on-site use values related to visibility at national parks, although only the
contingent valuation methods have been applied to date to obtain dollar estimates of these
types of visibility-related on-site use values. The contingent valuation methods are the only
methods available at this time for estimating non-use values, because these values are not
expected to be revealed in observable market behavior, which is used as the basis for most
other value estimation methods to value changes in non-market goods.
The contingent valuation method (CVM) involves the use of survey instruments to obtain
information on the values respondents believe they would place on potential changes
described to them in the survey. Variations of the CVM method include direct WTP or
WTA questions, referendum questions, and contingent ranking questions. In this report we
highlight key issues in applying CVM for estimating preservation values for visibility at
national parks. Thorough reviews of CVM as applied to non-market resource and visibility
valuation are available elsewhere (See Mitchell and Carson 1989; Rowe and Chestnut 1982;
Cummings et al. 1986; Freeman 1988; Fischhoff and Furby 1988).
2.2.2 Accuracy and Reliability of CVM Responses
CVM is a developing empirical method and the credibility and reliability of the results have
been the subject of some controversy in the economics profession. The skepticism results
primarily from the hypothetical nature of the method. It is based on what people say, not
necessarily on what they would actually do. Several potential sources of inaccuracy in
implementing CVM include:
1. Failed Correspondence between Theory and Method While the underlying
theory may correctly define the value measures of interest, the method chosen
to implement the theory may not be designed to obtain the correct measure.
2. Failed Correspondence Between Method and Questionnaire The method may
be appropriate, but the questionnaire itself may fail to correctly implement the
method. The researcher may simply ask the wrong questions.
RCG/Hagler, BaiHy, Inc.
-------
2-4
3. Questionnaire Design Problems. Even if the correct questions are asked from
the perspective of the researcher, they may not correspond with how the
respondent views the issue, or some survey information may mislead the
respondent. The researcher may be asking the right kind of questions, but in
the wrong way. For example, the respondent could reject the property rights
implied in the questionnaire, be influenced by the design of the questions, value
a slightly different good or service due to a different understanding of the
problem, or the respondent may be overtaxed by the difficulty of the questions.
The result is the respondent may give valid responses, but for a different set of
circumstances than the researcher believes he is addressing, or the results may
be biased by the instrument.
4. Response Problems. Even if the questions are designed properly, correspond
to the respondent's view of the issue and are well understood, the respondent
may have difficulty accurately quantifying the responses required. This may be
due to a lack of familiarity with the scenario of the good being valued, questions
that are too difficult or inconsistent with the cognitive processes used in decision-
making or due to a lack of effort in responding to the questions.
5. Implementation and Statistical Problems. Even with valid and accurate
responses for individual respondents, incorrect sampling and small sample sizes
may still lead to invalid results. Miscoding and statistical misinterpretation of
data can add further error or bias.
Most research addressing the validity and accuracy of CVM applications has focused upon
categories 2-4, although significant errors can result from all categories. Cummings at al.
(1986) have identified several conditions they suggest must be met to have a high level of
accuracy in CVM responses which include:
1. Subjects must understand and be familiar with the commodity to be valued.
2. Subjects must have had (or be allowed to obtain) prior valuation and choice
experience with respect to consumption levels of the commodity.
3. There must be little uncertainty.
4. WTP, not WTA, measures should be elicited.
Cummings et al. suggest that when these conditions are not met, the CVM responses may
still be valid, but the accuracy decreases. These conditions are useful to consider in
evaluating a CVM, but they are not comprehensive (they fail to consider many other
potential sources of inaccuracy), nor are they uniformly applicable to all CVM applications.
RCG/Hagler, BaiHy, Inc.
-------
2-5
Mitchell and Carson (1989) have provided a useful and structured presentation of many of
the potential sources of inaccuracy and bias in the application of a CVM valuation exercise.
These are found in Table 2.2-1. Other problems not identified above may also contribute
to inaccuracy in CVM.
To address the accuracy and reliability of CVM results, Cummings et al. review 15
comparisons of CVM results with results using other valuation techniques, such as hedonic
pricing models and travel cost models. In all of the comparisons, the calculated value
estimates were within 60 percent of one another and many are much closer. This does not
prove the accuracy of any one CVM is within 60 percent of the "true" value, but suggests
that each of these methods may often converge to the same range of values and that the
CVM method may be more desired in some instances due to its relative strengths as to
when and how it can be applied. Mitchell and Carson (1989) provide similar evidence for
a selected set of CVM studies.
Conclusions on the Accuracy and Reliability of CVM
Reviewing the list of potential sources of error in CVM studies, one might conclude the
results may often be invalid and inaccurate. However, careful design, pretesting,
implementation, and interpretation of results can minimize biases and inaccuracies and
yield valid and accurate value information. Next, the hypothetical nature of the questions
need not result in invalid or inaccurate answers. Respondents often face WTP decision in
markets, through political decisions and elsewhere that require real behavior similar to the
hypothetical behavior in a CVM. According to Mitchell and Carson, "But can CV surveys
actually measure values that are sufficiently reliable and valid for use in benefit estimation?
Our conclusion is basically affirmative." (1989, p. 295).
CVM is now widely used and is among the methods recommended by the United States
Department of Interior for evaluation of natural resource injuries. Use of the method for
these purposes was challenged recently in the United States Court of Appeals (Ohio vs.
U.S. DOI, No. 86-1529). The court ruling denied the challenge stating, "We find DOI's
promulgation of CV methodology reasonable and consistent with Congressional intent, and
therefore worthy of deference" (p. 94), and "We find no cause to overturn DOI's considered
judgment that CV methodology, when properly applied, can be structured so as to eliminate
undue upward biases" (p. 96).
2.2.3 Specific. Application Tssnes for Measuring Visihilitv Preservation Values
While most all general CVM design and application issues are of concern in the valuation
of visibility preservation values for national parks, a few issues are of specific concern, and
are discussed below.
RCG/Hagler, BaiHy, Inc.
-------
Table 2.2-1
Typology of Potential Response Effect Biases in CV Studies*
1. Incentives to Misrepresent Responses
Biases in this class occur when a respondent misrepresents his or her true willingness to pay (WTP).
A. Strategic Bias: where a respondent gives a WTP amount that differs from his or her true WTP amount (condition on the perceived
information) in an attempt to influence the provision of the good and/or the respondents's level of payment for the good.
B. Compliance Bias
1. Sponsor Bias: where a respondent gives a WTP amount that differs from his or her true WTP amount in an attempt to comply
with the presumed expectations of the sponsor (or assumed sponsor).
2. Interviewer Bias: where a respondent gives a WTP amount that differs from his or her true WTP amount in an attempt to either
please or gain status in the eyes of a particular interviewer,
2. Implied Value Cues
These biases occur when elements of the contingent market are treated by respondents as providing information about the "correct" value for the
good.
A. Starting Point Bias: where the elicitation method or payment vehicle directly or indirectly introduced a potential WTP amount that
influences the WTP amount given by a respondent. This bias may be accentuated by a tendency to yea-saying.
B. Range Bias: where the elicitation met hod presents a range of potential WTP amounts that influences a respondent's WTP amount.
C. Relational Bias: where the description of the good presents information about its relationship to other public or private commodities [hat
influences a respondent's WTP amount.
D. Importance Bias: where the act of being interviewed or some feature of the instrument suggests to the respondent that one or more levels
of the amenity has value.
E. Position Bias: where the position or order in which valuation questions for different levels of a good (or different goods) suggest to
respondents how those levels should be valued.
3. Scenario Misspecification
Biases in this category occur when a respondent does not respond to the correct contingent scenario. Except in A, in the outline that follows it
is presumed that the intended scenario is correct and that the errors occur because the respondent does not understand the scenario as the
researcher intends it to be understood.
A. Theoretical Misspecification Bias: where the scenario specified by the researcher is incorrect in terms of economic theory or the major
policy elements.
B. Amenity Misspecification Bias: where the perceived good being valued differs from the intended good.
1. Symbolic: where a respondent values a symbolic entity instead of the researcher's intended good.
2. Part-Whole: where a respondent values a larger or a smaller entity than the researcher's intended good.
-------
Table 2.2-1
Typology of Potential Response Effect Biases in CV Studies* (cont.)
a. Geographical Part-Whole: where a respondent values a good whose spatial attributes are larger or smaller than the
spatial attributes of the researcher's intended good.
b. Benefit Part-Whole: where a respondent includes a broader or a narrower range of benefits in valuing a good than
intended by the researcher.
c. Policy-package Part-Whole: where a respondent values a broader or a narrower policy package than the one intended
by the researcher.
3. Metric: where a respondent values the amenity on a different (and usually less precise) metric or scale than the one intended by
the researcher.
4. Probability of Provision: where a respondent values a good whose probability of provision differs from that intended by the
researcher.
C. Context Misspecification Bias: where the perceived context of the market differs from the intended context.
1. Payment Vehicle: where the payment vehicle is either misperceived or is itself valued in a way not intended by the researcher.
2. Property Right: where the property right perceived for the good differs from that intended by the researcher.
3. Method of Provision: where the intended method of provision is either misperceived or is itself valued in a way not intended by
the researcher.
4. Budget Constraint: where the perceived budget constraint differs from the budget constraint the researcher intended to invoke.
5. Elicitation Question: where the perceived elicitation question fails to convey a request for a firm commitment to pay the highest
amount the respondent will realistically pay before preferring to do without the amenity. (In the discrete-choice framework, the
commitment is to pay the specified amount.)
6. Instrument Context: where the intended context or reference frame conveyed by the preliminary nonscenario material differs
from that perceived by the respondent.
7. Question Order: where a sequence of questions, which should not have an effect, does have an effect on a respondent's WTP
amount.
* From Mitchell and Carson (1989). [Permission may be needed for inclusion in final draft]
-------
2-8
Familiarity With the Resource Change
Estimating non-use values may pose some additional difficulties for CVM. The primary
concern that has been raised is that the lack of familiarity with the good, or with thinking
about dollar values for the good, may make respondents particularly susceptible to
unintended influences in the survey instrument. Lack of familiarity is a concern with non-
use values because the respondents will necessarily include individuals who have not used,
and may never use, the resource. This may be particularly important when considering
subtle ecological changes or obscure endangered species. However, in the present case,
most people know what national parks are and have experience with various levels of visual
air quality in their daily lives, even if they have never experienced varying visibility levels
at a specific park. Therefore, they may be better able to comprehend and value visibility
changes at national parks than for some other types of resource changes. The design of the
questionnaire will allow this issue to be further addressed.
Context and Information
Recent work by Fischhoff and Furby (1988) has extended the discussion of the context of
simulated market transactions for visibility valuation. They assert, "In general, the more
novel a transaction, the more of its details will need to be explained and the more difficult
it will be to ensure that those details are understood." (page 152) They suggest that paying
for visibility improvements may be a reasonably novel transaction and that seemingly
irrelevant factors may affect responses. One might then infer that paying for preservation
of visibility at national parks, especially by those who are non-users, may be even more
novel. In part, these views are consistent with Shuman and Presser (1981), who have
argued that the more crystallized the values and attitudes are, the less important minor
context impacts are likely to be in survey design.
Fischhoff and Furby continue by identifying an array of potential characteristics of a
visibility valuation transaction that may be of concern. These characteristics are broken
down as defining main features of (1) the good, (2) the payment and (3) the social context
of the transaction. For each of these three features, characteristics can be considered as
helping to convey to the respondent "substantive" and "formal" definitions of the transaction.
Substantive definitions refer to characteristics that help identify how the transaction affects
the respondents directly, such as how their activities would be affected and how they would
pay for the transaction. Formal definitions refer to technical specifications of the
characteristics, such as duration and certainty of impacts. Many of these issues overlap the
scenario development concerns identified by other CVM researchers, and summarized by
Mitchell and Carson (1989).
Fischhoff and Furby suggest that to obtain accurate values for unfamiliar transactions, as
visibility valuation may be, the transaction may need to be extensively defined. When
characteristics of the transaction are left unspecified, respondents must rely upon their
default assumptions, which may differ from one another and from what the researcher
intended. They note, however:
RCG/Hagler, BaiHy, Inc.
-------
2-9
,. simply telling people everything provides no guarantee that they have understood
everything. Such a strategy might even impede understanding if attention to critical
features of the contingent market is diverted by a deluge of details about features
that could have gone without saying because they have little practical effect on
decisions, (page 152)
The researcher must, therefore, balance the need for information against the interests of
the respondents to absorb information. Relatively minor context features will have to be
unstated for a CVM instrument to be manageable. The challenge for the researcher is to
determine which information is critical in terms of the impact upon WTP measures of
interest. In the current research, reported in Chapters 3 through 5 below, the baseline
instruments are designed to incorporate many of the characteristics identified by Fischhoff
and Furby. The guiding principle used in designing the context for the value elicitation is
to keep it realistic and credible, but as simple and straightforward as possible. Several
survey variations then alter the context and information presented to begin to specifically
address some of the issues raised by these authors.
Part-Whole. Sequencing and Aggregation/Disaggregation
Part-whole bias, sequencing and the appropriate level of aggregation in the valuation, are
three interrelated issues of particular concern to the valuation of visibility protection at
national parks. Each is discussed and practical solutions identified below.
Part-Whole Bias. Mitchell and Carson (1989) define part-whole bias as occurring when a
respondent values a larger or smaller entity than the researcher intended. Potential part-
whole bias has been a significant concern in reviews of past urban visibility value studies
where, for example, Fischhoff and Furby (1988) indicate:
... respondents might be told to disregard how a change in air pollution affected their
health risk. However, such selective forgetting may not always be possible. If it is
natural to think of an intervention's impacts as a whole, there may be no way to
segregate mentally its individual effects, (page 155)
Concerns for the separation of health from visibility impacts of urban air pollution control
are the focus of two ongoing CVM studies (Carson et al. 1989 and Irwin 1989), and
preliminary results confirm, to some degree, the difficulty respondents have in separating
these two characteristics of air pollution control in urban environments.
In the valuation of visibility protection for national parks, part-whole bias could enter in
at least three ways:
1. Respondents fail to isolate visibility effects from other effects of air pollution,
such as damage to vegetation and risks to human health.
RCG/Hagler, Bailly, Inc.
-------
2-10
2. Respondents fail to isolate visibility effects from other concerns about national
parks, such as preservation of natural geologic features and prevention of water
pollution. "Symbolic bias," as suggested by Mitchell and Carson, is also
potentially related in that a respondent may more broadly value protection at
national parks rather than be concerned with the specific resource impairment
of concern to the researcher.
3. Respondents fail to isolate visibility effects at the identified park(s) from similar
effects at other park(s) or in other geographic areas.
Therefore, it is of significant importance to minimize the potential of these types of part-
whole bias in the survey design, and to test and correct for any remaining part-whole bias
in the CVM responses.
Sequencing. Sequencing bias occurs when the respondent provides different bids for the
same resource protection depending upon the number and order of environmental
protection issues to be valued. For example, Tolley et al. (1986) found the average stated
WTP value for the visibility protection at the Grand Canyon was lower when respondents
were first asked to give WTP values for visibility protection at other areas and then asked
an incremental WTP for the Grand Canyon, versus when they were asked a WTP for
visibility protection at only the Grand Canyon. These results are discussed further in
Section 2.3.
In part, this sequencing effect may be due to individuals having a "mental account" for a
category of environmental protection items based upon a limited ability to adjust their
budget at any one time. As a result, as more and more resource protection items are to
be simultaneously funded, the available financial resources per item decrease, and the
average amount that can be paid for each item decreases. If this is only revealed to the
respondent in an incremental manner (i.e., respondents are not told they ill also have to
purchase additional visibility protection at other - , as was done by Tolley et al. 1986),
the stated WTP for the later items considered may diminish as compared to if these items
were first in the list, if the items are grouped under the same mental account.
These budget constraint and sequencing issues are not a unique problem for resource
economics. A household's WTP for a bundle of consumer goods may change if it is also,
or first, required to buy a second bundle of goods and services. This may logically be the
case because components of the two different bundles may be perceived as substitutes for
meeting some of the goals (or motives) for which the bundles would be purchased; and
because diminishing marginal utility for bundles of goods and services, and budget
constraints, reduce the WTP for like incremental goods and services.
Several approaches may be taken to address sequencing effects. The simplest is to alter
the sequence for which items are valued and retest the results, but this simply recreates the
problem. The second, which is tied to the aggregation issue discussed next, is to define the
entire policy package to be valued before the valuation commences.
RCG/Hagler, Bailly, Inc.
-------
2-11
Aggregation/Disaggregation. Aggregation refers to the appropriate level of resource
impacts to consider; i.e., what is the total policy package of interest? Disaggregation refers
to how value estimates should be obtained for individual components of the policy package.2
As discussed for sequencing, as more and more items are added to a policy package to be
obtained at any one time, the WTP for any individual item may fall. In a similar vein,
Irwin et al. (1989) and Boyce et al. (1990) found that while WTP estimates given for
individual components of a good, such as visibility effects and health effects of air pollution
control, are less than the WTP amount for the good as a whole, the sum of the values for
individual components, when estimated separately, exceeded the WTP bid for the entire
good. This may be reflecting part-whole bias, budget constraints and/or other economic
and psychological response behavior. Therefore, it might occur that the sum of WTP
values obtained from separate CVM studies for individual parks in a region (or regions in
the country) may equal or exceed the WTP for simultaneous visibility protection throughout
the region (or throughout the country).
Similar problems are observed when WTP questions are asked in terms of monthly
payments versus annual payments: the monthly WTP estimates are less than the annual
WTP estimates, but they sum to more than the single annual estimate. This may be
reflecting failure to accurately consider budget constraints, or to fully comprehend the long
term nature of the payment stream when providing monthly payments.
The solution to these problems is not entirely clear. Available research does suggest that
the conservative direction to take in CVM exercises is to define and value the entire policy
package of interest, then disaggregate to component values through follow-up questions or
through statistical procedures. These component values may be for different parks, for
different value motives or for other components of interest.
This solution does not mean that, for example, all potential natural resource protection
issues must be simultaneously considered and valued in order to obtain a value for any one
resource protection issue. For practicality, the researcher must take a common-sense
approach concerning what the typical respondent will be able to isolate and what must be
treated as a total package. Finally, natural resource policy often occurs sequentially and
using values from independent studies covering independent issues in relative isolation may
be appropriate as, through time, economic agents have time to readjust their budgets and
may have the flexibility to fund additional (subsequent) resource protection.
The Motives Behind the Value Statements
Understanding motives behind preservation values is important to the valuation exercise.
This understanding allows more appropriate interpretation of what these motives are and
why they exist, which may be used to refine the CVM design to measure values for motives
2 Irwin et al. (1989) discuss aspects of this issue using the term "additivity," rather than aggregation.
RCG/Hagler, Bailly, Inc.
-------
2-12
as individuals see them, rather just as economists define them. This may also help in better
understanding what t ypes of resource quality and quantity changes will be of most value to
society. Examining motives is also important as a means to evaluate the credibility of CVM
responses, and to determine the appropriate application of the results for policy analysis.
Madariaga and McConnell (1987) have raised the question of motives behind bequest
values from another perspective and suggest that certain kinds of motives can confound
CVM responses and their interpretation for use in policy analysis. They argue that there
are two types of altruism that could underlie bequest motives. The first is an "individualistic
altruism" that is based on the utility of others and assumes the bequestor would not want
to leave a bequest that costs the receivers more than it benefits them. The second type is
defined as "paternalistic altruism" that is based on another's consumption of a specific good
because it will be good for them.
Based upon the above definitions, Madariaga and McConnell develop a model that suggests
that the usual efficiency criteria (that the sum of individual benefits exceeds costs, where
those benefits do not include any interdependence of utility) still gives the optimal
allocation of resources even in the presence of utility interdependence. This applies to the
"individualistic altruism" bequest motive -- benefits to future users must exceed costs to
future users or the altruist gains no utility. They conclude that knowing who bears future
benefits and costs, and how much they are, is desirable information to be included in
contingent valuation questions. What is not stated, however, is how to present future costs
and benefits when they are uncertain. They apply this conclusion with some small sample
empirical tests and find that willingness to pay is indeed smaller when subjects are told that
future users will incur costs as well as benefits.
We suspect that the paternalistic altruism, which is given limited attention by Madariaga
and McConnell, may be important with regard to parks. To the extent that a person
believes that others should have the perspective and ethical value associated with contact
with the natural environment, he may derive satisfaction from knowing the parks are
available for those he cares about regardless of, and almost certainly not knowing, whether
they would have been better off (in terms of maximum attainable utility) under a different
allocation of resources. Many of the choices that have been made to establish and protect
national parks were made without knowledge of future expenses, and may well reflect a
societal "paternalistic altruism."
An additional aspect of this paternalistic altruism may be the perception that there is a
broad social externality from the availability of the resource that goes beyond the well-
being of any one individual who actually uses the resource. In this case, it is not that the
bequestor's utility is directly enhanced because of use of the resource by specific individuals,
but rather because society as a whole may be better off in a way that exceeds the sum of
each individual's direct benefits. For example, consider programs taking juvenile
delinquents on wilderness experiences, which are intended to help with self confidence and
perspective, and hopefully reduce future crime.
RCG/Hagler, BaiHy, Inc.
-------
2-13
Another potential motive behind bequest value responses may be the perceived
irreversibility of failure to protect many natural resources. Even though in this case
visibility degradation is a reversible impact (the air clears up when emissions are reduced),
there may be a perceived irreversibility in terms of policy precedence. It may be that some
people perceive that if we allow the degradation now, we give up the right to prevent it in
the future, and open the door to other environmental degradation.
To a considerable extent, these discussions of motives behind bequest values are speculation
without empirical research to determine which motives dominate. A detailed analysis of
bequest values motives is beyond this study's focus and resources. This study does attempt
to obtain attitudinal information to understand and verify bequest values, and attempts to
circumvent the problem of interdependence of utility in the survey design.
2.3 PREVIOUS CVM STUDIES CONCERNING PRESERVATION VALUES
The section provides a brief review of selected CVM studies that have been conducted
concerning visibility protection at national parks and concerning other similar natural
resource protection issues.
2.3.1 Preservation Value Studies for Visihilitv Protection at National Parks
One previous CVM study, called the Southwest Parklands study, has estimated preservation
values for changes in visibility at the Grand Canyon National Park and at all national parks
in the Southwestern United States (Schulze et al. 1981). The Southwest Parklands study
is the only previous CVM study that is directly comparable to the study being presented in
this report. Two small follow-up studies of the Southwest Parklands study have been
conducted to examine specific questions raised about the Southwest Parklands study (Tolley
et al. 1986; and Rahmatian 1986). Rae (1984) also examined a few questions concerning
preservation values for visibility protection at Great Smoky Mountains National Park.
Southwest, Parklands Study
The Southwest Parklands study represents the first attempt to estimate preservation values
for visibility protection at national parks with respondents not physically on-site at the parks
in question. The results reflect estimates of preservation values for non-users as well as for
users. Respondents were interviewed in-person in four major metropolitan areas: Los
Angeles, Albuquerque, Denver, and Chicago. Four hundred and fifty completed (non-
protest) responses were obtained for the preservation value questionnaire. Three basic
questions were addressed: What is the value of controlling haze at the Grand Canyon
National Park? What is the value of controlling haze throughout the remainder of the
Southwest region? and, What is the value of controlling plumes visible from the Grand
Canyon National Park?
RCG/Hagler, BaiHy, Inc.
-------
2-14
In the introduction to the preservation value questionnaire, respondents were told that air
pollution from human sources sometimes impairs visibility at the Grand Canyon, and were
shown a set of photographs of the Grand Canyon. The photos showed three different
scenes, each under five different visibility conditions (fifteen photographs in total). The
respondents were told that the five photographs in each set showed conditions ranging from
good visibility to poor visibility, with the middle photographs reflecting current average
visibility conditions. Respondents were not given any quantitative information such as
frequency percentiles or visual range for the conditions shown in the photographs.
The first set of questions concerned previous and expected future visitation to the Grand
Canyon and other national parks in the Southwest. Respondents were then told that
additional industrial emissions controls might be required to prevent visibility at the Grand
Canyon from deteriorating and that such controls would likely make electricity more
expensive. They were asked to estimate what they would be willing to pay in increases in
their monthly utility bills to prevent an increase in air pollution that would cause average
conditions to deteriorate from the middle photos to the next worse photos. This
represented a change from approximately the current (1979) 50th percentile of visibility
conditions to approximately the then current 25th percentile. The photographs represented
approximate visual range levels of 200 km and 155 km. The respondents were then shown
a similar set of photographs showing scenes from three national parks in the Southwest
(Grand Canyon, Mesa Verde, and Zion) and were asked what additional amount they
would be willing to pay to prevent the same deterioration in visibility throughout the region.
Respondents who gave zero WTP were asked if they thought the change in visibility was
not important, or if they thought someone else should pay. The second response was
interpreted as a protest response and these zeros were dropped from the results. The
authors do not report the total number of zero responses obtained or the number
interpreted as protest zeros for the preservation value responses.
The average monthly WTP responses per household were (after exclusion of protest
responses):
Adjustingg to 1988 dollars and multiplying by twelve to obtain an estimate of annual WTP,
these results imply an average annual 1988 WTP per household of $95 to prevent a
degradation in visibility from the 50th to the 25th percentile at the Grand Canyon due to
haze, and an additional $80 for preventing this amount of degradation at all other parks in
the region.
Mean for Grand Canyon NP
SE of Mean
N
$5.50
0.41
450
Mean for remainder of region
SE of Mean
N
$4.66
0.36
450
RCG/Hagler, BaiHy, Inc.
-------
2-15
The analysis of the WTP responses reported by the authors suggests no relationship
between the WTP amount stated and the distance of residence from the Grand Canyon,
and very little relationship between the WTP amount stated and previous or expected
future Grand Canyon visitation. Older respondents gave significantly lower WTPs and
respondents with higher incomes were associated with higher WTPs. The income elasticity
was approximately 0.3, implying that a 10 percent higher income was associated with a 3
percent higher WTP response.
Questions have been raised about the effect in this study of focusing on such a perceived
"national treasure" as the Grand Canyon and whether the responses would be the same if
other areas were also considered at the same time. The concern is that respondents may
have overstated their true WTP for the Grand Canyon and included some value for changes
in visibility in other areas as well. A related question raised is whether stated WTP for
visibility protection at the Grand Canyon would change if respondents were also requested
to simultaneously spend more to protect visibility in other areas.
Southwest Parklands Follow-up Studies
Two follow-up studies have attempted to address some of the questions concerning the
Southwest Parkland Study. While the followup studies do not entirely resolve issues in the
original study, they do suggest that some problems may exist.
Tolley et al. (1986) addressed the question of whether responses would be different if
respondents were asked to give WTP for changes in visibility where they live as well as for
protection of visibility at the Grand Canyon. A sample of residents in Chicago was asked
a set of three WTP questions: (1) WTP to prevent a degradation in visibility in Chicago,
(2) additional WTP to prevent a similar degradation throughout the remainder of the U.S.
east of the Mississippi, and (3) additional WTP to simultaneously protect against a specified
degradation in visibility at the Grand Canyon. The Grand Canyon photographs and
hypothesized change were the same as those used in the Southwest Parklands study. All
the questions were for monthly increases in utility bills to cover costs of pollution controls.
The results are summarized in Table 2.3-1 (adjusted to annual WTP in 1988 dollars).
Even though the sample sizes are small, the WTP estimates for the Grand Canyon are
substantially smaller when asked as the third in a series of WTP questions. The authors
conclude that the WTP responses are influenced by the order of the questions. In the
Southwest Parklands study, the change in visibility being considered was for the Grand
Canyon only, while in the follow-up study the visibility change is for Chicago, the East and
the Grand Canyon all at once. This appears to be a change in the good being valued as
well as the order of the questions. The results therefore are best interpreted as
demonstrating that when considering a change in one area such as the Grand Canyon, it
is important to consider whether changes in visibility would also occur in other areas as
part of the same policy package. It is a substantially different question for the respondent
to give WTP for changes in, for example, the entire U.S. versus WTP for changes in just
RCG/Hagler, Bailly, Inc.
-------
2-16
Table 2.3-1
Comparison of Schulze et al. and Tolley et al.
Grand Canyon Visibility Value Results
Annual
WTP for
Chicago
Annual Additional
WTP for
Remainder of East
Annual Additional
WTP for
Grand Canyon
Tolley et al.
(N = 59)
$296
$36
$21
(SE = 12)
Schulze et al.
(Chicago sample,
N = 130)
$132
(SE = 20)
one location, such as the Grand Canyon. With income constraints, and assuming visibility
is a normal good, one would expect that incremental WTP to prevent degradation at an
additional site would be smaller than WTP for the same site if it were the only site needing
protection. This is the aggregation issue discussed above.
This problem may have been further exacerbated by the procedure used in the Tolley et
al. study. Respondents were not informed at the beginning of the WTP questions that they
would be asked to give estimates of WTP for visibility protection in more than one location,
which reinforces sequencing problems, as discussed above. As a result, respondents may
have allocated a larger share of this "visibility budget" to the first site considered as if it
were the only item in a policy package, than they would have if they had considered all
three sites simultaneously as part of one policy package.
A second follow-up study was conducted in Denver in 1982 (Rahmatian 1986), also with the
aim of examining the effect of considering changes in visibility at the Grand Canyon alone
versus changes in a larger area as well as the Grand Canyon. Some subjects were asked
to give willingness to pay estimates for the Grand Canyon alone. Others were shown
photographs of the Grand Canyon and other parks in the Southwest and then asked to
give willingness to pay estimates (1) for the Grand Canyon and then (2) for the parks in
the remainder of the southwest region. This second protocol is similar to the questions
asked in the Southwest Parklands study except that the photos for the region as well as for
the Grand Canyon were shown before the willingness to pay questions were asked. There
is no statistically significant difference in the mean WTP estimates given for the Grand
Canyon in these two procedures, and in both cases the mean responses are not statistically
significantly different than the results obtained in the Southwest Parklands study. However,
the Grand Canyon question was asked first in both cases; therefore, the results do not
adequately address the question of whether responses are different if the order of the
questions is changed.
RCG/Hagler, BaiHy, Inc.
-------
2-17
Other concerns with the Southwest Parklands study include the separability of visibility from
other resource protection issues. For example, a major concern regarding measuring
visibility values is whether other effects of air pollution, such as health effects or vegetation
damage, are being separated fully from visibility protection values. There is also concern
that park resource protection motives unrelated to visibility may be reflected in the
responses. This is the potential part-whole bias discussed above.
The focus of the Southwest Parklands study was also geographically limited and only one
level of change in visibility conditions related to haze was valued. The uncertainties
involved with transferring these results to other locations and other visibility change
scenarios limits the applicability of the results for current policy issues without further
verification with new research. Finally, some questions have been raised about specific
design elements of the CVM application in this study, such as the order of the questions
and the use of the monthly utility bill payment vehicle.
(1984)
Rae (1984) conducted a CVM study in Cincinnati, Ohio, which primarily focused on
benefits of reducing urban air pollution, but a few questions were also asked later in the
questionnaire about visibility at the Great Smoky Mountains National Park. Interviews
were conducted with 316 adults. Visibility at Great Smoky Mountains was illustrated with
two photographs of a scene at the park showing 20 km visual range and 100 km visual
range. Subjects were asked the maximum their household would be willing to pay annually
to have visibility conditions at the park like those shown in the 100 km photograph most
of the time rather than the 20 km conditions. No payment vehicle was specified. A list of
values (in intervals) ranging from $0 to $500 was shown to the subjects. Subjects were
asked about past and expected future visits to the park. After excluding one $1000
response, the average response to the first visibility question regarding Great Smokies was
about $60. A significant portion of the subjects had visited the park and/or planned to visit
it in the future, so this willingness to pay can be expected to reflect perceived benefits
associated with actual visits as well as potential non-use values.
Subjects were also asked a second set of questions about their willingness to pay for ten
"good causes," including protecting visual air quality at Great Smokies, to see how their
responses might change when more than one good cause was considered. The average
willingness to pay for visibility protection at the park was about one-third of the previous
average response. This suggests the possibility that subjects may not be considering the full
range of alternative uses of their money when they are asked to estimate their willingness
to pay for a single good cause, and that when competing "good causes" are also included,
the willingness to pay for one of the causes may be smaller. However, in this question the
magnitude of the change in visual air quality and in the other good causes was not specified.
This uncertainty may also play an important role in the change in visibility bids. Cummings
et al. (1986), Mitchell and Carson (1989), and Fischhoff and Furby (1988) all suggest
uncertainty leads to inaccurate, and often reduced, bids.
RCG/Hagler, Bailly, Inc.
-------
2-18
2.3.2 On-Site Use Value Studies for Visibility at National Parks
A number of contingent valuation studies have been conducted that have estimated dollar
values for visibility impacts to on-site visitation at national parks. More extensive reviews
of these studies can be found in Chestnut and Rowe (1983). These studies are briefly
discussed here because they are CVM studies concerning visibility protection at national
parks, but their focus was much different than that of the current study.
Results of on-site use value studies concerning visibility at national parks are summarized
in Table 2.3-2. Two of these studies have obtained estimates of use values for changes in
visibility at the Grand Canyon, one of the parks selected for the focus of this study. In both
of these studies survey respondents were asked what they would be willing to pay in
additional daily entrance fees to have one level of visibility rather than another while
visiting the Grand Canyon National Park.
MacFarland et al. (1983) interviewed about 1000 visitors at the Grand Canyon and at Mesa
Verde in the summer of 1980 for the willingness-to-pay portion of their study. Visitors
interviewed at the Grand Canyon were shown two sets of slides, each showing five different
levels of visibility at a particular viewpoint. The authors report approximate levels of visual
range for the five different conditions, as shown in Table 2.3-2. Respondents were asked
to estimate the most they would be willing to pay in additional daily entrance fees, over the
then current two dollars, to have visibility at level B, C, D, and E rather than level A while
visiting the park. Possible response options were provided to respondents using a checklist
format. The mean responses are shown in Table 2.3-2.
Schulze et al. (1981) interviewed 166 subjects for the use value portion of the Southwest
Parklands study. Subjects were interviewed at their homes in Albuquerque, Denver, Los
Angeles, and Chicago. Respondents were excluded from the use value questions if they
said that they had not visited the Grand Canyon in the last 10 years and if they did not
plan to visit the Grand Canyon in the next 10 years. The willingness to pay questions were
very similar to those asked by MacFarland et al. The primary difference was that the
respondents were interviewed at their homes rather than at the park and some of them
(roughly 40%) had not been to the park in the last 10 years. Schulze et al. do not report
visual range estimates for their photographs, but they report Mg/m3 pollutant loadings and
visibility frequency percentiles the photos represent based on point contrast measurements
taken throughout the summer of 1979. We have calculated approximate visual ranges for
these pollutant loadings and percentiles from historic data on visual range levels at the
Grand Canyon (reported by NPS 1988 and by MacFarland et al. 1983). These estimated
levels of visual range and the mean WTP responses obtained by Schulze et al. are shown
in Table 2.3-2.
Other use value studies for changes in visibility at parks or recreation areas have typically
asked visitors what they would be willing to pay in additional park entrance fees to ensure
one level of visibility during their visit to the park versus another level. In general, the
RCG/Hagler, BaiHy, Inc.
-------
2-19
Table 2.3-2
On-Site Use Values for
Visibility Protection at Recreation Sites: Selected Studies
Initial New $1988 WTP
Visibility Visibility Per Visitor Method
Study Site* Observations (miles) (miles) Party Day Used"
Rowe, et al.
Navajo
26
75
50
$6.98
IB
(1980)
Navajo
26
75
25
$11.14
IB
Navajo
26
50
25
$5.19
IB
Schulze, et al.
GCNP
166
75
95
$2.47
CL
(1981)
GCNP
166
75
125
$4.06
CL
GCNP
166
75
175
$5.73
CL
GCNP
166
75
240
$7.57
CL
McFarland, et al.
GCNP
1000
70
100
$1.40
CL
(1983)
GCNP
1000
70
130
$2.34
CL
GCNP
1000
70
165
$3.07
CL
GCNP
1000
70
215
$4.03
CL
MVNP
800
70
100
$1.24
CL
MVNP
800
70
130
$1.68
CL
MVNP
800
70
165
$2.56
CL
MVNP
800
70
215
$3.73
CL
Rae
MVNP
196
74-95c
160
$4.42
CR
(1983)
MVNP
193
95
160
$11.88
CL
GSMNP
202
6-12c
60
$5.67
CR
GSMNP
202
12
60
$3.76
CR
GSMNP
201
12
60
$2.94
CL
a GCNP = Grand Canyon National Park, MVNP = Mesa Verde National Park, GSMNP =
Great Smoky Mountains Park, Navajo = Navajo Reservoir, NM.
b All studies used entrance fee vehicles
IB = iterative bidding
CL = Check list of value ranges to select from
CR = Contingent ranking
c Two scenarios, each with a different baseline, provided indistinguishable results and were
merged together.
RCG/Hagler, BaiHy, Inc.
-------
2-20
mean WTP estimates obtained in these studies range from $2 to $8 per day per household.
These studies are summarized by Chestnut and Rowe (1983), Freeburn (1987) and Gilbert
(1989). The results obtained for national parks other than the Grand Canyon are not
dramatically different, suggesting that there is not a noticeable premium for the Grand
Canyon at least when it comes to on-site use values.
One thing that is striking about the on-site use value results is how small they are compared
to expenditures, stated response behavior to visibility changes, and other consumer's surplus
estimates, given that viewing is often cited as the single most important activity at national
parks. No one has estimated the total consumer's surplus associated with a visit to the
GCNP, but Haspel and Johnson (1982) and Johnson and Haspel (1983) estimated total
WTP for a visit to Bryce Canyon National Park (BCNP), with average visits lengths of 1
day, at between $93 and $130 ($1988). We have no reason to expect the consumer's surplus
for visits to the GCNP would be substantially less than for BCNP. Finally, the results found
by MacFarland et al. (1983) suggest that 50 percent changes in visual range at the GCNP
would result in a majority of visitors changing their time spent at the park by 13 to .5
hours, a figure that seems potentially inconsistent with WTP estimates of a few dollars per
day per visitor party.
There are several possible reasons why the available estimates of the on-site use value
component of visibility related consumer's surplus estimates for the GCNP are relatively
small.
1. The values may be accurate. Values may be small because while viewing is
important, it is not the only aspect of the experience. Moreover, recreators have
the ability to substitute to other sites, or in some cases to other activities.
Finally, it may be the case that with high trip costs, there is little consumer's
surplus left.
2. The reported values may be biased due to survey design elements. Most of these
surveys represent relatively early CVM exercises. The responses to these early
WTP questions may be biased downward as a result of psychological response
behavior that has been identified in the literature, but cannot be statistically
adjusted for with the current data. Psychological research suggests that
"anchoring and adjustment" problems frequently occur in choice making (Slovic
1969). Individuals may anchor on a well understood situation or value, such as
current entrance fees, and adjust this value to respond to the CVM scenario.
This adjustment process often falls short of being complete. Respondents may
be further anchoring upon what is felt to be a "reasonable" payment, rather than
the maximum payment they would make before they would choose the
alternative.
Of importance for the studies summarized in Table 2.3-2 is that they used entrance fee
vehicles that may have anchored respondents upon the then typical $2 entrance fee.
Respondents may have been psychologically prone to consider that if they pay $2 for an
entrance fee (covering the whole experience), then something on this order may be
"reasonable" to protect the viewing component of the experience. Moreover, respondents
RCG/Hagler, BaiHy, Inc.
-------
2-21
may have objected to the suggestion that national park entrance fees be used to generate
funds to control nearby industrial emissions. Little was done or reported in past studies to
test or adjust for these types of problems.
Some recent research may support the above argument for downward bias in the on-site
use values. Gilbert (1989) estimated changes in consumer's surplus for haze impacts to
recreation in Vermont. This study first had respondents allocate trip costs to viewing
scenery (called scenery costs), and to other purposes, then asked for an incremental WTP
to obtain improved visibility, or to prevent visibility degradation. In contrast to earlier
studies, the scenery cost variable serves as the anchor rather than entrance fees and
estimated values per day are much higher. Another study by Rowe et al. (1989) examined
CVM and travel cost consumer's surplus measures for Atlantic salmon fishing in Maine.
The CVM estimates, using a fishing license vehicle, roughly equaled the current license
fee, while the travel cost estimates were nearly 10 times larger. The most apparent
explanation for this difference was the potential influence of the current license fee as an
anchor as to "reasonable payments," which was reinforced by written comments from
respondents.
2.3.3 Other Similar CVM Preservation Value Studies
Several other studies address similar preservation valuation issues and have study designs
and findings that are instructive for the present study. Two studies (Bishop and Boyle 1985,
and Walsh et al. 1982) have estimated total preservation values for protecting sites similar
to parklands. The purpose in these studies was different than for the present study because
the value being elicited was for preservation of the site (or sites) as a nature preserve or
wilderness, rather than for changes in the quality of the resources at these sites. Another
study of interest (Sutherland and Walsh 1985) examined the relationship between
preservation values for protection of water quality at a recreation site and the distance of
the subject's residence from the site.
A fourth study reviewed here (Carson and Mitchell 1988) has estimated total values for
protection of the quality of freshwater lakes, rivers and streams (exclusive of drinking
water) throughout the country. Although the subject matter is different, there are some
similarities to this study in that the subject is the quality of an environmental resource that
occurs in all parts of the country and varies from place to place. It is also similar in that
the general population was sampled whether or not they were involved in any water related
recreation activities.
Bishori and Bovle (1985s)
Bishop and Boyle used a mail questionnaire to obtain values to Illinois residents for
preserving the Illinois Beach State Nature Preserve, which is located at the southern end
of the Illinois Beach State Park on Lake Michigan. The Nature Preserve is currently
RCG/Hagler, Bailly, Inc.
-------
2-22
threatened by erosion of the sand dunes which if allowed to continue will result in the
flooding of the Nature Preserve. A stratified random sample of 600 Illinois residents was
selected and the overall response rates were 63%.
Information about the Nature Preserve was included in the form of maps and questions and
answers. The willingness to pay question was the close-ended referendum style.
Respondents were asked whether they would pay a given amount for a membership to a
private foundation that would provide the funds necessary to build an off-shore breakwater
and manage the Nature Preserve day to day. Follow-up questions were asked to find out
what respondents thought about their answers to the membership question. Additional
sections of the questionnaire probed respondents about their familiarity with the Nature
Preserve and about their attitudes toward environmental protection and this questionnaire.
Respondents' familiarity with the Nature Preserve was not high even for respondents from
nearby counties, but 77% of the sample said that it was somewhat or very important to
them personally that the Nature Preserve be preserved. This percentage was about the
same for the residents of nearby counties, and for those who live further away and were
much less familiar with the Nature Preserve. This suggests that visitation to, or even name
recognition of the specific site, may not be as important as might be expected.
The weighted average annual willingness to pay to preserve the Nature Preserve in its
current state was $28 per household. The authors took a conservative approach in
calculating total values for the State by presuming that non-respondents placed no value on
preservation of the Nature Preserve, because those who do not return the questionnaire are
more likely to care less (but not necessarily zero) about the resource in question.
Responses to the follow-up questions and the environmental protection attitudes provide
some insight about how the respondents were thinking about the issue in general and how
they reacted to the presentation of the issue in this particular questionnaire. These types
of follow-up and attitude questions are very helpful in interpreting the results of the
valuation questions, especially with a mail questionnaire where there is no chance to gage
the respondents reaction in any other way.
More than half of the respondents who said yes to the membership question, also checked
the follow-up response that said "I don't know what I would pay for a membership, but I
thought the State Nature Preserve should be preserved." About 85% of the respondents
said that it was definitely or probably true that they thought their response was important
because they could participate in the decision of whether to preserve the Nature Preserve.
These responses suggest that a significant percentage of the respondents were uncertain
about what they would have paid for a membership, but that they said yes to the amount
asked because they wanted to "vote for" preservation. However, when presented with the
statement, "I felt that preserving the State Nature Preserve would not really cost me
anything because the membership question was hypothetical," about 62% of the
respondents said this was probably or definitely false. It appears that the respondents were
uncertain about the actual payment they were willing to make, but that they took the
questionnaire seriously and believed that their responses would have an effect on the policy
decisions that were made and ultimately on costs that they would bear.
RCG/Hagler, BaiHy, Inc.
-------
2-23
Responses to questions about general environmental protection attitudes also give an idea
how respondents think about these issues. The majority of respondents indicated their
sentiments that the plants and animals have a right to exist and that human interests should
not take precedence. The respondents seemed to be expressing a kind of environmental
ethic that was discussed above in the context of existence value. This may be related to
why such a large portion of the respondents said that it was somewhat or very important
to them personally that the Nature Preserve be protected even though only a small
percentage of them had ever visited it. The findings on response attitudes, accuracy and
motives suggest these are important issues deserving more attention.
Walsh et, al. (1982s)
This study obtained estimates of willingness to pay to preserve wilderness areas in Colorado.
A mail questionnaire was sent to 600 Colorado residents during summer 1980, with a
response rate of 40%. Four maps of the State illustrated alternative levels of wilderness
protection, ranging from the then-current area (1.2 million acres) to up to 10 million acres.
Respondents were also asked their willingness to pay to preserve alternative amounts of
wilderness throughout the remainder of the country.
Respondents were asked to assume that the only way to preserve wilderness in Colorado
would be by paying into a special fund to be used exclusively for that purpose. They were
then asked the maximum they would be willing to pay each year for each of the four levels
of protection. Following this question they were asked to estimate what portion of their
payment they would allocate for the following reasons, which the authors interpreted as use
value, option value, existence value, and bequest value respectively:
Payment to visit existing or potential Wilderness Areas this year.
Payment for the option to visit existing or potential Wilderness Areas in the
future, should you choose.
The value to you from knowing there exists a natural habitat for plants, fish,
wildlife, etc.
The value to you from knowing that future generations will have Wilderness
Areas.
Approximately 84% of the respondents were willing to pay some positive amount for the
preservation of at least the amount of designated wilderness that existed in Colorado in
1980 and 77% were willing to pay some positive amount for wilderness preservation
throughout the country in addition to the payment for Colorado wilderness. Average
annual willingness to pay per household for 1980 level wilderness in Colorado was $26,
with approximately 45% of this amount being allocated to use value. Average annual
additional willingness to pay per Colorado household to preserve current (1980) wilderness
RCG/Hagler, BaiHy, Inc.
-------
2-24
throughout the remainder of the country was $14, with about 20% being allocated to use
value. Willingness to pay was higher than this for larger amounts of wilderness in Colorado
and throughout the country, although the proportion allocated to use value was somewhat
higher for the higher acreage amounts. The percentages of the payment allocated for
option, existence, and bequest values were roughly similar, although the percentage given
for option value tended to be somewhat smaller than for the other two.
Total values, including both use and preservation values, were significantly related to the
distance from the respondent's residence to the nearest wilderness area and to the amount
of wilderness visitation made. More frequent visitors and those living nearer gave higher
values for both use and non-use motives.
Responses to questions concerning the importance of various reasons for preserving
wilderness indicated that preservation related motives were considered somewhat more
important than user motives. Protection of water quality, air quality, and wildlife habitat,
and knowing that future generations will have wilderness, ranked slightly above recreation
use values and option of future use values.
Sutherland and Walsh (1985s)
Sutherland and Walsh conducted a CVM study concerning the protection of water quality
from degradation due to coal mining in the Flathead River and Flathead Lake area in
Montana. This is a recreation area currently used primarily by local residents. Seventy-
five percent of the visitors are from Montana and most of the remaining visitors are from
neighboring states. The authors were particularly interested in examining the relationship
between the WTP responses and the distance the respondent lives from the site.
A mail survey instrument was sent in the summer of 1981 to a sample of residents from
four Montana cities located various distances from the site. These distances were 10, 115,
227, and 420 miles. Usable responses from 171 residents were obtained. The response rate
was 61 percent. Respondents were asked the total annual amount their household would
be willing to pay into a special fund to protect water quality in the Flathead River and
Lake area. They were then asked to allocate this payment across four reasons they may
want to protect water quality in this area:
1. For their own visits to the area in the current year.
2. For their future visits to the area.
3. To know that good water quality exists in the area.
4. To know that future generations will have good water quality in the area.
RCG/Hagler, BaiHy, Inc.
-------
2-25
The mean total preservation value response was $64, with $7, $11, $20, and $26 being the
average amounts (based on percentage of the total) for each of the above reasons,
respectively. The results of the analysis of the WTP responses indicates a significant
relationship between the WTP response and the distance of residence from the site, and
between the WTP and the frequency of visitation to the site, which was also correlated with
distance from the site. The authors report that the observed relationship suggests that the
value falls to near zero at about 640 miles, although this is based on an extrapolation
outside the range of residence distances included in the sample. The authors stress that
unlike well-known national parks that might be considered national environmental resources
due to wide dissemination of information about them, the Flathead River and Lake area
is more of a regional resource with which most non-visitors are not very familiar.
Carson and Mitchell (1988s)
Carson and Mitchell conducted a nationwide survey in 1984 concerning the value of
protecting the quality of freshwater lakes, rivers and streams throughout the country.
In-person interviews were conducted with 813 individual, with 79% of the eligible
respondents completing interviews.
The potential range of water quality levels was described with the help of a ladder on
which four water quality levels were shown: swimmable, fishable, boatable, and too polluted
for any human, plant, or animal contact. It was explained that with current pollution
control efforts, 99% of the nation's freshwater lakes, rivers, and streams are at least
boatable with most being fishable and perhaps 70-80% being as clean as swimmable.
Subjects were told that they are currently paying for pollution control efforts through higher
prices and taxes, and that if these control efforts were stopped that the water quality in
areas that can now be used only for boating would in many cases fall to less than boatable.
Subjects were asked to give the maximum they would be willing to pay for their household
in taxes and higher prices (including the amount they are currently paying) to ensure that
the boatable level is maintained in virtually all (99%) water bodies; the maximum
additional amount they would pay to have a minimum level of fishable; and the maximum
additional amount to have a minimum of swimmable. In follow-up questions subjects were
shown estimates of typical amounts paid through higher prices and taxes for water quality
protection by households in a similar income category, and were asked if they wanted to
revise their responses. A subset of subjects were also told the typical amount a household
in their income bracket is paying for air pollution control as well as water pollution control.
The idea was to test for the possibility that people forget that water pollution control is
only one aspect of environmental quality and would therefore revise their responses
downward given the information about air pollution control expenditures. Subjects were
also asked if they would still be willing to pay the amount they gave for the fishable level
if 5% of the nation's water bodies remained at the boatable level, and if 50% remained at
boatable. Subjects were also asked to divide their willingness to pay between the state
where they live and the remainder of the country.
RCG/Hagler, BaiHy, Inc.
-------
2-26
Usable willingness to pay responses were obtained from 70% of those who were
interviewed. The breakdown of those who did not give usable responses was as follows: (1)
72 said they didn't know; (2) 133 gave protest zeros with explanations falling into one of
two groups, attitudes of anti-governmental taxes and expenditures or strong environmentalist
attitudes with feelings that putting dollars on such things is immoral; (3) 16 gave responses
of more than 5% of their incomes and were judged to be too high; and (4) 10 gave very low
estimates (e.g. $1) that were judged to be protest responses that were not caught because
a value other than zero was given.
The mean response per household to obtain the swimmable level for all freshwater lakes,
rivers, and streams from the baseline of what would occur without current controls (i.e.,
non-boatable in some areas) was $280 (1984 dollars) per year. The authors used a
statistical procedure to account for potential non-response bias when they aggregated to a
U.S. total. Responses were weighted to make the sample more representative of the
Census population. This adjustment resulted in a 12 percent reduction in the mean WTP
value for obtaining the swimmable level.
The authors use an indirect approach to separate values for different motives by comparing
the values given by users versus non-users. When non-users were defined as those who
reported no in-stream (or in-lake) recreation by household members in the past year, non-
use values were at least 30% of total values. When non-users were defined as those who
reported no direct or indirect use (e.g. picnicking by a lake or stream), non-use values were
at least 19% of total value. These are lower bounds in that no non-use value is attributed
to users, although expectations about potential future use by those who were non-users last
year are unknown.
Responses indicated that subjects considered the 95% option to be essentially equivalent
to the full attainment of the fishable level in all water bodies, and the reduction that was
made in responses if only 50% would obtain the fishable level was significantly less than
half the original amount given. There is the possibility that subjects were influenced by
the wording of the question. Since it asked whether they would still pay the amount they
originally gave, there may have been some reluctance to change their answers. A more
neutral approach would have been to ask what they would pay for the alternatives involving
less than complete attainment of the fishable level.
Subjects reported that about two-thirds of their willingness to pay was for water quality in
their own state and one-third for the remainder of the country. The changes in the
estimates after subjects were shown estimates of the amount they currently pay for water
quality protection, were small. Those whose first response was below their current payment
estimates tended to increase their responses somewhat, and those whose first response was
above their current payment estimates tended to stay the same. Overall, the subjects'
willingness to pay responses were surprisingly close to estimates of their current payment.
RCG/Hagler, BaiHy, Inc.
-------
2-27
Responses concerning willingness to pay for water quality protection did not change
significantly when estimates of current payment for air quality protection were also shown
to the subjects. This is different than the findings in some previous surveys that found
significant changes in responses when more issues were introduced. Since the way these
tests were made differ, it is difficult to draw satisfactory conclusions from these findings.
This finding may have been the result of introductory information concerning a variety of
public issues that may have served to help respondents think about water quality protection
issues without forgetting about other competing demands on their budgets. An alternative
explanation is that respondents may become defensive when new information is presented
that suggests they should reconsider their bids, and therefore refuse to revise their
responses.
Finally, income was found to be significantly related to the responses, showing an income
elasticity of about 1. This is a larger income effect than has been found in many willingness
to pay surveys and may have been influenced by the income adjusted anchors used on the
payment cards.
2.3.4 Lessons from the Related Research
Several key lessons from related research highlight issues and directions to be addressed
in the current effort.
The preservation value studies have addressed issues from local sites to nation-
wide concerns. The WTP values also vary considerably, from tens of dollars to
hundreds of dollars each year, and reflect that respondents do not simply give
the same response for any preservation value WTP question. Overall,
evaluations of these surveys indicate respondents generally take the survey
seriously and attempt to give valid responses to the CVM questions.
Respondents acknowledge uncertainty in the accuracy of their responses.
The researcher must carefully present the policy package of interest, and perhaps
present information on related policy actions.
WTP response tend to generally reflect expressed attitudes and behavior
conveyed in other parts of the survey. The use of this type of information is
critical for evaluating the survey responses.
Values in previous studies are found to be related to expected use and often to
distance from a site, although this impact may be lessened for more prominent
sites, such as national parks. Therefore, these data should be collected.
RCG/Hagler, BaiHy, Inc.
-------
2-28
Follow-up questions that ask respondents to comment on and evaluate their
WTP bids were well received and may be of particular importance in evaluating
a CVM application. Respondents acknowledge uncertainty in the accuracy of
their responses to total preservation value questions, and addressing the level
and impact of this uncertainty upon the analysis is of particular importance for
preservation value studies. On the other hand, follow-up questions that ask for
new bids based upon new information may not be well received.
Larger changes in resource provision, beyond the initial proposed change, may
result in relatively flat WTP response surfaces, although little investigation has
been given to this issue.
Separating, or disaggregating, values to individual value motives may be a
difficult exercise for respondents, just as it is for other characteristics of air
pollution control. (See Section 2.2.3). However, using indirect methods, such
as in Carson and Mitchell 1988 may add even more error to the exercise. The
estimates of value for individual components obtained in these studies are
consistent with other survey evidence, although the average option price
estimated in Walsh (1982) is quite large relative to theoretical expectations
(Freeman 1988). It may be inappropriate and overtaxing to ask respondents to
consider separately current use, option value and future use in favor of just
asking for values related to current and potential future use.
RCG/Hagler, BaiHy, Inc.
-------
3-1
3.0 STUDY DESIGN
This chapter describes the design of the survey instrument and the implementation
procedures.
3.1 DEVELOPMENT OF THE SURVEY INSTRUMENT
This study was designed to address some of the questions that have been raised concerning
the estimation of preservation values for visibility protection at national parks and to extend
the information available, from previous studies. Basic objectives include:
Examining how visibility protection values vary across different regions and
parks. To do so, the analysis considers visibility impacts at national parks
throughout the Southwest, California, and the Southeast; and values are
estimated for one selected park in each region.
Analyzing how different resource protection attitudes and behaviors are tied to
WTP, and to the motive definitions economists traditionally use (option price,
bequest value, existence value). This examines the validity of the separation
of values into these motive categories, which is also done in the study.
Examining the impact of respondent uncertainty in the CVM exercise upon the
reported values, and examining which respondent characteristics are tied to
valuation uncertainty. This is accomplished through a follow-up question on the
respondents' self-perceived accuracy of their WTP responses.
Testing for the ability to control part-whole bias related to other air pollution
impacts and other national park natural resource protection issues, and
correcting for any such bias. This is done through a survey design attempting
to mitigate such impacts before the WTP question, paired with a follow-up
question addressing the existence and significance of the problem.
Developing direct and realistic CVM scenario context information, and testing
the impact of specific changes in context information.
Addressing the issue of what level of information must be presented about
visibility protection at other potentially competing national park sites, while
bidding on only one park region, by including this information in some survey
versions, but not in others.
Addressing how visibility values change with changes in the number of park
regions to be protected by including a survey version that addresses values for
protection in three regions at once.
Refining select CVM design features and analysis procedures to address issues
raised in the literature.
RCG/Hagler, Bailly, Inc.
-------
3-2
Six versions of the questionnaire were developed to address these questions. All versions
included a photograph insert showing alternative visibility conditions at one or more
national parks. These versions are summarized in Table 3.1-1 and discussed below.
3.1.1 Pilot Tests. Peer Reviews, and Pretests
After two rounds of pilot testing, with approximately ten individuals per round, a revised
instrument was prepared and sent out for peer review. The reviewers included sociologists
familiar with national park visitor issues and survey design issues, economists familiar with
CVM design, and an atmospheric scientist familiar with visibility issues. Based upon these
reviews, the instrument, was revised to a pretest draft.
Twenty in-person pretests were conducted by two professional interviewers from Colorado
Market Research, a Denver survey firm. The interviewers obtained responses from twenty
Denver residents in several different neighborhoods selected to represent a range of
socioeconomic characteristics. The interviewers carried a display showing the same
photographs and map as were used in the final questionnaire insert. Respondents were
shown the display and asked to answer the questionnaire on their own. The interviewer
then asked some specifically prepared follow-up questions and made note of any other
comments offered by the respondents.
Overall, the pretest respondents indicated interest in the topic and in the photographs and
seemed to understand the questions, including the WTP questions. The pretest results
indicated that the average time to complete the questionnaire (33 minutes) was still longer
than desired. A few more questions were therefore either simplified or eliminated to get
the expected average completion time closer to 25 minutes. Other minor changes in the
presentation were also made to further streamline and reduce the print on each page, as
well as to further refine some specific wordings.
One of the pretest follow-up questions was whether the respondent felt his WTP responses
were for the specific changes in visibility only, or the responses also included some value
for the protection of national parks in general. The responses to this question suggested
that some of the respondents were including general values. This question was, therefore,
included in the final questionnaire.
3.1.2 Outline of the Final Baseline Questionnaire
The baseline version (labeled Version 3 below) of the questionnaire consists of six sections,
with WTP questions focusing on the national parks in the Southwest. A copy of this
baseline questionnaire is included in Appendix A and includes the sections discussed below.
RCG/Hagler, BaiHy, Inc.
-------
Table 3.1-1
Summary of Questionnaire Versions
Title
Photo Insert Illustrates
Focus Region for WTP
Focus Park
1. California Parks
2. Southeast Parks
3. Southwest Parks
(Baseline version)
4. Multiple Regions
(1 WTP for each of
three region)
5. Limited Information
(reduced WTP scenario
details)
6. Single Region Focus
(presentation for
only 1 region)
3 regions
3 regions
3 regions
3 regions
3 regions
1 region only
California
Southeast
Southwest
California,
Southeast, and
Southwest
Southwest
Southwest
Yosemite
Shenandoah
Grand Canyon
Yosemite, Shenandoah,
and Grand Canyon
Grand Canyon
Grand Canyon
-------
3-4
Cover
The cover provides the title "MANAGING VISIBILITY AT NATIONAL PARKS: WHAT
IS YOUR OPINION?"; a pictograph with trees, mountains, a dollar sign, fish and a family
to indicate a variety of competing issues facing the family and the natural environment.
The cover also states, "Research conducted for the Center for Economic Analysis at the
University of Colorado." The back cover states that the survey should be returned to
RCG/Hagler, Bailly. These affiliations, rather than a federal agency, are listed to reduce
any perceived incentives related to potential sponsor bias (Mitchell and Carson 1989). The
back cover also allows space for comments, which are used in consistency checking of
individual responses, and to better understand the overall responses.
Section 1: About Your Visits to National Parks
The first four questions in this section concern the respondent's past and expected future
visitation to national parks. These questions are fairly straightforward making it relatively
easy for the respondent to get started (Dillman 1978). They also get the respondent
thinking about national parks in general and their own past and intended future visitation
and experiences as a means to help the respondent begin to establish the substantive
importance to themselves, if any, of the hypothesized visibility changes.
National park visitation is expected to be an important factor related to responses
concerning the value of protecting visibility at national parks for two reasons. The first is
that visiting national parks probably reflects a greater interest and concern for the
protection of this sort of resource, and will therefore be one measure of differences in tastes
and preferences across the sample. The second is that national park visitors will have some
actual experience to draw upon in answering the questions, which may result in some
differences in their responses to hypothetical questions. These hypotheses are tested in
Chapter 4.
These questions also lead the respondent to view the map included in the insert, which
shows most of the national parks in the country at which visibility is considered an
important resource1 (See below for discussion of the insert). This distinction allowed the
focus to be on just those units of concern, versus all NPS units, without using the term "class
I areas." This map is intended to help define the regions (Southwest, Southeast and
California) used in the questions, identify the national parks in the region where visibility
is an important resource, and to provide the subsequent perspective that the WTP questions
are about only a portion of all the national parks that might be of interest to the
respondent.
1 Section 169a of the Clean Air Act gives certain measures of visibility protection to federal class I areas
where visibility is an important resource. Subsequently, regulations were promulgated (40 CFR Part 81.400;
November 30, 1979) identifying these sites following NPS recommendations.
RCG/Hagler, BaiHy, Inc.
-------
3-5
The last three questions in this section let the respondent consider the importance, if any,
of reasons (or motives) people may want to visit and protect national parks. Question 5
asks respondents to rate the importance of potential reasons for visiting national parks.
The list is based, in part, on results of national park visitor surveys (Ross et al. 1985) and
is intended to allow some distinction between reasons related specifically to enjoying the
natural environment versus other reasons people visit national parks, such as to spend time
with family or to simply have a change of surroundings. Questions 6 and 7 are about the
respondents' interest in protecting national parks even if they personally could never visit
a national park. The purpose of these two questions is to identify the relative importance
of bequest value and existence value types of motives for the preservation and management
of national park resources, and to get respondents thinking about the reasons they may want
national parks preserved, and protected.
The information on attitudes about national park use and protection are also useful as
consistency checks on WTP responses. I.e., one would expect that those with strong use
and non-use protection attitudes would be more likely to provide positive WTP. One would
also expect those with higher scores for preservation, even if they could not visit the park,
would assign a larger share of their WTP to bequest value and existence value motives in
subsequent questions.
Section 2: Ahout, Pollution Issues Facing National Parks
Question 8 asks respondents to consider several different types of potential pollution
impacts to national park resources from human activities outside the parks, and whether
they consider the prevention of each a low, medium, or high priority. One of the impacts
listed is visibility degradation. This question is aimed at:
Getting respondents to think about the range of potential threats to park
resources before considering one threat in more detail as a means of attempting
to separate visibility protection from other national resource protection issues.
This again is directed to minimize part-whole bias as well as to acknowledge
competing resource protection issues.
Obtaining information about the perceived relative importance of protecting
visibility versus other types of pollution impacts from human activities outside
the parks.
Introducing the impacts as being due to man-made activity, and originating from
outside the parks.
RCG/Hagler, Bailly, Inc.
-------
3-6
Section 3: About Visibility In and Around National Parks
This section serves to introduce, through the photograph insert and text discussions and
questions, the range of effects of air pollution on visibility conditions in the three study
regions. The photograph insert is discussed in detail below in Section 3.2. The photographs
are presented as representing events that occur with different frequencies "on days without
rain or fog" (underline added). Frequencies are presented to communicate that the average
condition does not occur on all days, but rather there is a distribution of different
conditions. The underlined text above is also stated on the photograph insert. The pretest
results indicate respondents seem to understand this caveat.
Question 9 asks what effect having visibility as in Photo B rather than as in Photo C would
have on the respondent's enjoyment of a national park visit in each of the three regions.
Photo B shows a somewhat less than average amount of haze for each of the park regions,
while Photo C shows the average for each region. This is the visibility change considered
in the first WTP scenario. This question is intended to have respondents consider the
potential significance, if any, that such a change in visibility would have for them personally
(and presumably for others as well) during a park visit. This is expected to be related to
option price and bequest value motives and serves to get respondents thinking about how
they would be affected by such a change before asking them the more difficult WTP
questions. This question also provides non-dollar information about how important a
change in visibility might be for the respondent in terms of his or her own park visitation,
which is useful in evaluating and interpreting the WTP responses, and provides some
information from each respondent about potential attitudes toward visibility protection for
all three park regions as opposed to just the one region considered in the WTP section of
most survey versions.
Question 10 begins to define the pollution control mechanisms, the payment vehicle, and
social context of the visibility transaction to be used in the CVM questions terms of who
will pay and how. After introducing this context, it asks how willing the respondent might
be to pay higher prices or taxes to support visibility protection for national park in each of
the three regions. It also highlights that each region is only one of many that might be
considered for additional visibility protection to again recognize the potential for competing
resource protection and to reduce potential part-whole bias. This question also provides
information from each respondent about attitudes toward paying for visibility protection for
all three regions, unlike the specific WTP questions that focus on just one region.
Section 4: About Visibility at National Parks in the Southwest
Question 11 asks that the subject consider the photographs (for Version 3) for the Grand
Canyon NP, as representative of conditions at national parks throughout the Southwest, and
assess the importance of (1) improving visibility and (2) preventing visibility from getting
worse at national parks in the Southwest. This question brings the respondent's focus to
the single region of interest for the subsequent WTP questions and provides some non-
RCG/Hagler, BaiHy, Inc.
-------
3-7
dollar information about the respondent's attitude toward visibility protection for parks in
that region. It also introduces the idea that additional expenditures might be required to
prevent visibility from deteriorating as well as to obtain improvements over current
conditions. Indirectly, one can also assess the applicability to visibility protection of a
prospect theory (Kahneman and Tversky 1979 and 1982) tenet that preventing losses is seen
as more important than obtaining gains.
Section 5: What is the Value of Protecting Visibility at National Parks in the Southwest
This section establishes the context for the WTP valuation, includes the specific WTP
questions, and includes follow-up questions to help in interpreting the WTP responses. Key
elements in the scenario development include:
"New air pollution laws being considered for the protection of visibility at
national parks in the Southwest could mean higher prices and higher taxes
throughout the country." This reinforces the vehicle and social context of the
payment introduced in the prior section.
"These questions concern only visibility at, national narks in the Southwest and
assume there will be no change in visibility at national parks in other regions.
Other households are being asked about visibility, human health and vegetation
protection in urban areas and at national parks in other regions." These
comments are designed to reduce the tendency to include values for other air
pollution impacts and at other locations into the visibility value responses.
"... assume you could be sure that any change would occur next year and
continue forever, ..." This is included to reduce concerns about the certainty of
provision, which has been identified as a concern by Mitchell and Carson (1989),
Fischhoff and Furby (1988) and others.
"... all households now and in the future would also pay the most it is worth to
them to protect visibility." This again establishes the social context of the
transaction in terms of who will pay. Moreover, it is intended to partially
address concerns raised by Madariaga and McConnell (1987) about bequest
values (See Section 2.2). These authors indicate that, to correctly formulate
WTP values related to bequest values, respondents must know the benefits and
costs to future generations. However, future benefits and costs cannot be known.
This comment, therefore, addresses this problem by stating that future
generations will pay the most it is worth to them, implying that net benefits to
others can be assumed to be unchanged.
"... average conditions will change in and around all national parks in the
Southwest..." Again, to minimize potential aggregation biases (since many
policies could affect many national parks rather than just one), regional
aggregate bids are obtained, then disaggregated to values for the individual parks
shown in the photographs.
RCG/Hagler, BaiHy, Inc.
-------
3-8
Three WTP questions follow this introduction. Each question asks "What is the most your
household would be willing to pay every year in increased prices and taxes" for the specified
change in average visibility conditions "at all national parks in the Southwest." While
monthly payments may better proxy financial decision-making for many households, annual
payments are perceived by the research team to better clarify the annual financial impacts
in relation to budget constraints (in order to minimize potential budget constraint bias
identified by Mitchell and Carson, 1989), and are expected to have a downward effect on
the valuation (see the discussion of aggregation in Section 2.2.3).
The WTP approach is selected, as opposed to a willingness to accept payment (WTA) to
forgo improvements, or to incur degradation, due to the operational practicalities. WTA
measures have some appeal in theory, and may be appropriate if the affected individuals
have a property right to be compensated by the lluters. Where WTA measures are
appropriate, they can be expected to exceed WTP measures. However, the theoretical
difference is uncertain and may range from very small under specific assumptions (Randall
and Stoll 1980) to very large if preferences begin to reflect a lexicographic ordering
phenomenon. Operationally, WTA responses are often plagued by those who do not
respond, or respond with infinity, potentially reflecting emotional or ethical rejection of the
WTA premise. As a result, CVM practitioners often advise against the use of WTA
measures (Mitchell and Carson 1989). Finally, any policy the reduces widespread haze
impacts at national parks will likely call for control measures that are ultimately paid for
by a large number of individuals, or even by all of society. Therefore, a WTP measure has
an appropriate foundation in policy analysis as well.
WTP is obtained for the three hypothesized changes in average visibility conditions.
1. Obtaining improvement in average conditions from Photo C to Photo B. This
is a compensating surplus value measure and is subsequently referred to as
WTP1.
2. Obtaining improvement in average visibility conditions from Photo C to Photo
A, which exceeds (or equals) the improvement from Photo C to Photo B. This
is a compensating surplus value measure and is subsequently referred to as
WTP2.
3. Preventing degradation in average visibility conditions from Photo C to Photo
D. This is an equivalent surplus measure and is referred to as WTP3.
Prospect theory, and neo-classical utility theory, would suggest that for a comparable
visibility change, WTP to prevent a degradation would exceed WTP to obtain an
improvement, but the expected magnitude of such a difference is uncertain.
RCG/Hagler, Bailly, Inc.
-------
3-9
The WTP elicitation procedure in the questionnaire employs a payment card approach with
no benchmarks, where respondents have the option to review a variety of potential
alternative payments and choose the best response. This approach was selected to obtain
greater estimation efficiency for the selected sample size, as opposed to referendum
approaches (Cameron and Huppert 1988). In addition, we have found that payment cards
typically obtain lower item non-response rates in mail surveys than do open ended WTP
questions (see, for example, Rowe et al. 1986, Rowe et al. 1985, and Rowe and Schulze,
1987). The dollar checklist is the same for each WTP question and ranges from $0.00 to
"more than $750." To minimize the potential for researcher induced range bias discussed
by Mitchell and Carson (1989), and Fischhoff and Furby (1988), the range of values
included in the checklist was based on the pretest results.
Question 1.5 asks respondents to give any information that might help explain their answers
to the WTP questions. This approach was selected over a checklist of possible explanations
for zero responses because we have found that comments offered in the respondents' own
words are very helpful for interpreting the refusals, zeros, high bids, and other non-zero
WTP responses, and it allows the researcher the benefit of the respondent's own
clarification on the issue and their WTP response. Respondents are also invited to give any
additional comments on the back page of the questionnaire.
Question 16 recognizes the difficulty in such-WTP exercises and asks for a self assessment
of the accuracy of the WTP responses. It allows respondents to indicate the overall strength
of the value signal provided. Fischhoff and Furby (1988) express concern about "forcing"
more out of respondents than they have to give. They suggest it may be more appropriate
to screen respondents for whether they have answers to give, than for them to be forced to
trust and express nascent feelings (page 169). To address the impact of any forcing induced
by survey design, this question was intended to allow examination of the differences in
responses between those who believe their responses are fairly accurate versus those who
believe their responses are very inaccurate.
Following in the same vein, Question 17 asks respondents to consider their WTP responses
and to say whether they were basically for the stated changes in visibility at national parks
or whether the responses also reflected values for other needs. Based upon the pretest,
there was particular concern that the WTP values may reflect a contribution to support
other needs at national parks as well as visibility protection. The second part of the
question asks respondents who say that other concerns are also reflected in their WTP to
then estimate what percentage of their WTP responses is really for visibility. Because the
early sections in the questionnaire separated visibility from other issues, and clearly
indicated that the WTP responses were to be only for the stated visibility changes at
national parks, this question provides information to address whether extensive scenario
development can, on its own, overcome potential part-whole bias for related resource
protection issues, and provides data to correct for any such bias in the value calculations.
RCG/Hagler, Bailly, Inc.
-------
3-10
Question 18 asks respondents to estimate what percentage of their WTP for the region they
would want to allocate to the specific national park illustrated in the photographs. This
reverses the order of these questions as compared to the Southwest Parklands study
(Schulze et al. 1981) in which subjects were first asked a WTP for the Grand Canyon and
then a WTP for the remainder of the region. The decision was made to start with WTP
for the region because:
The long range transport nature of the emissions related to many of the current
visibility effects at national parks suggests that many pollution control strategies
targeting visibility at national parks are likely to have regional, rather than single
park, effects, and
It is presumed that the appropriate aggregation/disaggregation approach is to
start with large units all simultaneously affected by a policy package and then
disaggregate to small units to obtain individual park values.
Question 19 asks respondents to give the percentage of their WTP responses that they
would attribute to the following motives:
So my household and I could enjoy conditions as natural as possible on visits to
national parks in the Southwest
So others, now and in the future, could enjoy conditions as natural as possible
on visits to national parks in the Southwest
To have conditions as natural as possible at national parks in the Southwest,
even if no one were to ever visit
Other (please specify)
These potential motives are defined to reflect option price, bequest value, and existence
value as defined in Section 2.1. Some previous CVM studies (Greenley et al., 1981) have
asked for separate WTP estimates for the different motives, assuming that these could be
summed to obtain a total preservation value. The approach taken here presumes there is
less potential for upward bias in the total value estimate, which is of most importance, and
in the individual component value estimates, if the total is first obtained and then
disaggregated to value components (see Section 2.2.3).
Section 6: About You and Your Household
The last page of questions concerns socioeconomic characteristics of the respondent and the
household that might be related to attitudes about visibility protection at national parks.
These include age, sex, education level, and employment status of the respondent; and the
number and age of all household members and household income.
RCG/Hagler, BaiHy, Inc.
-------
3-11
3.1.3 Description of the Six Versions
As summarized in Table 3.1-1, questionnaire Versions 1, 2, and 3 are identical except that
they focus on different regions in Sections 4 and 5 (Questions 11-19) of the questionnaire.
Version 1 focuses on national parks in California, as illustrated in the photo insert by
Yosemite National Park. Version 2 focuses on national parks in the Southeast, as
illustrated in the photo insert by Shenandoah National Park. Version 3 focuses on national
parks in the Southwest, as illustrated in the photo insert by Grand Canyon National Park.
The photo insert, discussed below, is identical for Versions 1 through 5. As summarized
in Table 3.1-1, Versions 4, 5, and 6 reflect alternative variation of the instrument to allow
for tests of the effects of specific changes in the design.
Version 4: Multiple Region Focus
There is concern in CVM exercises that values for resource protection for one site (or
issue) will be different if resource protection must simultaneously be purchased at multiple
sites (or for multiple issues). As a result, it may be invalid to add together values for
individual policy package components, if they were estimated independently and
individually, to obtain a total value for the entire policy package. I.e., one may not be able
to add together values derived separately for the Southwest and California to value a policy
package that obtains both. This is the aggregation problem discussed in Chapter 2.
For visibility protection at national parks, it may be the case that some policies may impact
multiple regions simultaneously. To examine the potential magnitude of the aggregation
problem for more than one region, Version 4 includes WTP question about visibility
protection for national parks in all three regions simultaneously.
All the introductory questions and information are the same as in the baseline survey
version, except that Question 11 and the WTP introduction refer to all three regions rather
than just one. The introduction to the WTP questions in Version 4 reads as follows:
New air pollution controls being considered for the protection of visibility at national
parks in California, the Southwest, and the Southeast could mean higher prices and
higher taxes throughout the country. The next questions concern how much obtaining
improvements and preventing worsening in visibility at national parks in each of these
regions would be worth to your household if you had to pay for the improvements in
all three regions each year.
These questions concern only visibility at, national narks in California, the Southwest,
and the Southeast, and assume there will be no change in visibility at national parks
in other regions.... (the remainder of the introduction is unchanged).
RCG/Hagler, BaiHy, Inc.
-------
3-12
There are three separate WTP questions, one for each region for obtaining an improvement
in average visibility conditions from the current 50th to the current 75th percentile at
national parks in that region. The WTP follow-up questions are the same except that the
percentage for a single park is not asked, and the remaining questions refer to all three
regions.
The responses to these three WTP questions can be analyzed in concert with the responses
to the first WTP question in Versions 1 through 3: Question 12 in Version 4 asks for a
WTP for the same change in visibility as that asked in Question 12 in Version 1; Question
13 in Version 4 asks for a WTP for the same change in visibility as that asked in Question
12 in Version 3; and Question 14 in Version 4 asks for a WTP for the same change in
visibility as that asked in Question 12 in Version 2.
Version ft: Limited Information
The continuing investigation into the design of CVM scenarios has tended to lead to
identifying more and more transactions characteristics that mav impact the valuation. For
example, Mitchell and Carson (1989) have defined many potential sources of bias due to
the selection or inclusion of scenario information. Similarly, Fischhoff and Furby (1988)
have identified a long list of definitional attributes that mav be important. These issues
were discussed in Chapter 2. In fact, nearly every CVM author has added information
issues of concern to the scenario design. The problem, however, arises that to meet every
identified potential need may result in a deluge of detail that distracts attention from the
critical features of the CVM scenario. Some attempts must be made to begin to look away
from what information may have some impact, and to look to what information has the
most significant impacts upon the magnitude of the estimates.
Version 5 presents one look at this issue by simply deleting most of the second paragraph
in the baseline WTP scenario development, which includes several potential information
needs recently identified in the literature. Specifically, Version 5 Hplptps the following
information used in Version 3 (with identifiers (1), (2) and (3) added for reference in the
subsequent discussion):
(1) Other households are being asked about visibility, human health and vegetation
protection in urban areas and at national parks in other regions. (2) For these
questions, assume you could be sure that any change would occur next year and
continue forever, and (3) all households now and in the future would also pay the
most it is worth to them to protect visibility.
Comment (1) was originally included to reduce incentives to inflate WTP to cover values
for other air pollution control effects as a means of reducing part-whole bias. The effect
of its deletion, if any, is expected to result in higher WTP values. Comment (2) reduced
uncertainty about the provision of the good. Mitchell and Carson (1989) and Fischhoff and
Furby (1988) suggest that if this is unspecified, the subject may have concerns about the
RCG/Hagler, BaiHy, Inc.
-------
3-13
actual provision of the good. The effect of its deletion, if any, is expected to result in lower
WTP values. However, we suspect the effect may be minimal as our experience is that
most individuals accept the assumption of the stated change for responding to the valuation,
and those who are uncertain, or do not accept the assumption, frequently state $0 and add
written comments clarifying the rejection nature of their response, which leads to their WTP
response being treated as a rejection bid rather than a valid value statement for visibility
changes. Comment (3) was included to better establish the social context of the transaction,
and to address the bequest value formation and interpretation concerns raised by Madariaga
and McConnell (1987). Omitting this information reintroduces the "individualistic altruism
motive" these authors identify, and the effect, if any, is expected to increase bids.
Several of the follow-up questions can also be examined to see if the difference in the
introduction has any effect. Overall, the deletion of this information could have minimal
effect if the individual effects are all minimal, or if the effects are offsetting. Because we
expect minimal impact of deleting the comment on certainty, but cannot be sure of these
expectations, we expect the deleting of this paragraph will have a zero or positive impact
upon the bids.
Version 6: Single Region Focus
As identified in Chapter 2, there is concern that CVM experiments need to identify, or even
include, other similar environmental impacts that may compete for funding. The primary
issue that motivates this concern is potential part-whole bias, where the respondent may
inadvertently include values for related goods. Therefore, it is sometimes argued that stated
values for the resource protection issue in question may be overstated, due to failure to at
least consider the existence of other competing resource protection issues in the survey
instrument. If many competing resource protection issues must all be identified, explained,
and even bid upon in CVM experiments, it adds significant complexity and cost to the
exercise, may distract the respondents from focusing upon the one resource protection issue
of most concern, and may result in lower response rates. For visibility protection at
national parks, an important aspect of this question is the extent to which respondents are
able to isolate national parks in a single region, or a single national park, from other
national parks around the country.
To begin to address this issue, Version 6 focuses on the national parks of the Southwest and
does not include many of the questions, or parts of questions, in the baseline version that
refer to visitation to and visibility protection at national parks in other regions. Also the
photograph insert included with the Version 6 questionnaire shows only the photographs
of the Grand Canyon National Park. Sections 3 and 4 are combined and all of the
questions in this section refer only to national parks in the Southwest. The WTP questions
and the remainder of Sections 5 and 6 are identical to Version 3.
RCG/Hagler, BaiHy, Inc.
-------
3-14
3.2 PRESENTATION OF VISIBILITY CONDITIONS
3.2.1 Visual Air Quality Measures and Human Perception
Because the survey relies on photographs to convey important information about visual air
quality at the national parks, it is important to consider what is known about how human
subjects perceive visual air quality depicted in this way. Several studies have been
conducted that have examined the factors influencing human judgements of visual air
quality. Malm et al. (1980, 1981), for example, asked subjects to rate the visual air quality
in a series of scenes on a 1 to 10 scale. This rating is called the Perceived Visual Air
Quality (PVAQ).
The relationship between PVAQ and the importance (and hence WTP) a subject might
place on obtaining better or preventing worse visual air quality has not been examined.
There is no reason to assume that because a subject can perceive a difference in visual air
quality that they value such a change, but it is probably safe to assume that there is no
value for changes that cannot be perceived. It is probably 'also safe to assume that for any
one individual, a larger change in PVAQ would be valued the same or greater than a
smaller change in PVAQ. Thus, factors that are correlated with PVAQ are likely to be
correlated with WTP. We can, therefore, use the results of the PVAQ studies to help
minimize the introduction of extraneous factors in the presentation of alternative levels of
visibility that are known to influence PVAQ judgements. We can also use the results of the
PVAQ studies to identify physical parameters that may be appropriate for relating WTP
based on specific photographs to objective measures of air quality that can be tied to
pollution emissions.
Four findings from the PVAQ studies are particularly important for consideration in
designing and analyzing WTP studies concerning visibility conditions.
1. PVAQ ratings of actual scenes are correlated with PVAQ ratings of slides taken
at the same time.
This means that using photographic representations to illustrate different levels of visual air
quality is not likely to introduce distortions in subjects' responses relative to how they would
respond to the actual scene in person. There are, however, many remaining questions
concerning the use of photographs in WTP studies for changes in visibility. Several
important factors that are not known include how differences in the features of the scene
used to illustrate different levels of air quality might be expected to affect WTP responses,
and how color differences between on-site viewing and photographic representations might
affect WTP responses.
2. PVAQ ratings are affected by the presence of clouds, snow cover, and sun angle,
as well as by air quality.
RCG/Hagler, BaiHy, Inc.
-------
3-15
These findings underscore the importance of holding factors other than air quality constant
when illustrating different levels of visual air quality for evaluation by subjects.
3. When there is a dominant distant feature in the scene, and factors such as cloud
cover and sun angle are held constant, PVAQ ratings are linearly related to the
atmospheric transmittance between the feature and the observer. Therefore,
PVAQ is, to a first approximation, proportional to the apparent contrast of the
feature against the horizon sky.
Because contrast is an objective measure of visibility conditions, it is reassuring that it has
been found to be so highly correlated with PVAQ judgements. This means that subjects
are responding to measurable changes in air quality in some predictable fashion. Malm et
al. (1981) suggest that in the presence of multiple vista elements, PVAQ will be a function
of the contrast (transmittance) of each element weighted by the fractional area subtended
by that element. They qualify this suggestion with the observation that subjects seem to key
in on the most sensitive features of the scene when judging changes in visual air quality.
For example, a foreground feature with unchanged contrast does not seem to affect PVAQ
judgements when more distant features are changing.
4. PVAQ ratings are inversely, and non-linearly, related to measures of light
extinction between the observer and a distant target, such that the PVAQ ratings
are more sensitive to increases in pollution (i.e., decreases in light extinction)
in cleaner atmospheres.
Light extinction is inversely related to visual range, which is the distance at which a large
black object on the horizon is just perceptible. This means that PVAQ ratings can be
expected to be positively related to visual range levels.
3.2.2 Selection of Photographs
Visibility conditions at national parks in each of the regions are illustrated with a set of
photographs from a park in each region selected from the National Park Service air quality
monitoring network. This network includes automatic cameras at several national parks
around the country that take daily photographs of the same view at set times of day. This
network is managed for the National Park Service by Air Resource Specialists, Inc., in Fort
Collins, Colorado.
The parks selected to represent each of the regions are those with visibility conditions
typical of most parks in the region and for which photographs were available. The selected
parks are well-known parks and among the most frequently visited in each region.
Yosemite National Park was selected for California, Grand Canyon National Park for the
Southwest, and Shenandoah National Park for the Southeast.
RCG/Hagler, BaiHy, Inc.
-------
3-16
The decision was made to use actual photographs rather than computer generated
photographs for two reasons. The first was that the availability of the extensive NPS
photograph network made it possible to obtain an acceptable set of photographs for each
park (showing a range of air quality conditions with minimal variation in other factors) at
relatively low cost. The second reason was that using actual photographs enhances the
credibility of the presentation because it is possible to say that these various conditions
actually occur, as opposed to explaining that the photos presented are artificially generated
representations of conditions that do occur. Using actual photographs, however, makes it
impossible to maintain exact uniformity in all factors other than air quality.
The photographs used in this study were selected to show a range of visibility conditions
associated with different'levels of air quality, with differences such as sun angle, clouds,
snow, and color avoided as much as possible. All of the selected photographs were taken
at 3 o'clock in the afternoon. Slight differences in sun angle therefore occur because the
photographs were all not taken on the same day of the year. Slight differences in color due
to slides being processed in different batches were also difficult to avoid.
The WTP questions in this study are framed in terms of changes in average visibility
conditions, but a range of visibility conditions is shown to communicate that there is a
distribution of conditions, that all days are not "average," and that a change in the average
really means a shift in the distribution. Due to differences in the pattern of meteorological
conditions across the seasons, presentation of an annual distribution of visibility conditions
due to fluctuations in air quality can be confounded by the differences in the meteorology.
We therefore decided to show the typical range of conditions during the summer, when the
majority of national park visitation currently occurs. Because the purpose of the
presentation is to communicate the general day-to-day variability of conditions for a lay
audience, rather than give a precise depiction of a year-round distribution, and because it
is necessary to keep the information presentation brief, no information about visibility
conditions during other times of the year is presented. In fact, median visibility conditions
are typically somewhat better in the winter than in the summer, but show a similar
variability between best and worst conditions at most national parks.
Slides were selected by Air Resource Specialists, Inc., to approximately represent typical
10th, 50th, 75th, and 90th percentiles of summertime visibility conditions at each park.
These slides were then reproduced as printed photographs on the glossy inserts. These
photographs are not exact representations of each of these percentiles, but can be expected
to adequately represent the range of these percentiles in most years. This is because the
film processing and photograph reproduction process always introduces some slight changes
and because actual visibility conditions vary from year to year.
Table 3.2-1 gives information about each of the photographs including the view, the NPS
inventory number, and values for different visibility measures.2
2 For additional technical discussion see Malm et al. (1980, 1981), EPA (1985) and Trijonis et al. (1990).
RCG/Hagler, BaiHy, Inc.
-------
Table 3.2-1
Photographs Used to Illustrate Visibility Conditions
Park
View
NPS Photo Representative
Inventory Percentile
Number
Average Visual
Range for the
Percentile (km)
Single-Target
Contrast
Single-Target
Atmospheric
Transmission
Weighted Average
Atmospheric
Transmission
Yosemite Half Dome
A 65
B 657
C 114
D 105
90
75
50
10
150
125
90
45
-0.81
-0.66
-0.23
-0.05
0.00
0.73
0.26
0.06
0.973
0.924
0.723
0.524
Grand
Canyon
Mt. Trumbull A 738
B 887
C 1087
D 1182
90
75
50
10
250
200
155
115
-0.71
-0.72
-0.64
-0.55
0.89
0.90
0.80
0.69
0.941
0.945
0.892
0.830
Shenandoah Rocky Mt.
A 552
B 1869
C 2297
D 379
90
75
50
10
75
50
25
10
-0.84
-0.39
-0.35
-0.15
0.93
0.43
0.39
0.17
0.861
0.445
0.418
0.256
-------
3-18
Representative percentile is the approximate percent of time visibility is less than
or equal to the represented level during the summer season.
Visual range is the distance at which a large black object just disappears from
view, or can no longer be distinguished from the background.
Single-target, atmospheric transmission is a measure of the light transmitted from
a single point in the scene.
Single-target, contrast is derived from the single-target atmospheric transmission
and is a measure of the difference in brightness between the target and the
background.
Weighted average atmospheric transmission is the atmospheric transmission
between each scenic feature and the observer, weighted by the fractional area
of the scene subtended by each feature.
None of these measures fully accounts for the different content of the scenes across the
three national parks considered, and they therefore have limitations for use in comparisons
across the different parks.
The approximate visual ranges typical of each percentile are given in the table. NPS (1988)
reports 10th, 50th, and 90th percentile visual ranges for each of the national parks in the
network based on teleradiometer, photographic densitometry, or extinction measurements.
We took the averages of the 10th, 50th, and 90th percentile estimates reported for
monitored summers (available data varies by park, but typically cover several years during
the 1980s) at each of the three parks and interpolated to get an approximate visual range
for the 75th percentile at each park.
The last three measures given in Table 3.2-1 are taken directly from the photograph insert
used in the survey. NPS staff calcula:. these measurements using information recorded
with a digitizing camera, which converts light reflected from the image into digital density
values. As expected, these measures are somewhat, but not dramatically, different than the
same measures for the original slides.
The single-target measures used Half Dome in the Yosemite photos, the U-shaped ridge
to the right of Mount Trumbull in the Grand Canyon photos, and the near ridge to the
right of the center of the image in the Shenandoah photos. Due to the target specific
nature of these single-target contrast measurements, it is not appropriate to compare these
across the scenes for the different parks, or to necessarily infer that the contrast at the
point is representative of the entire photograph.
The weighted average atmospheric transmission measurements may be somewhat more
comparable across the scenes for the different parks because they take into account each
different-distance feature in the overall atmospheric transmission.
RCG/Hagler, Bailly, Inc.
-------
3-19
One potential problem in the photoset is that both of the atmospheric transmission
measurements for the 75th and 90th percentile photographs for the Grand Canyon are
virtually identical. The visual air quality in both of these photographs is quite high, but
visual inspection shows that the detail of the canyon walls is more clearly delineated in the
90th percentile photograph than in the 75th percentile photograph. Atmospheric
transmission is just one technical measure that can be used to characterize visual air quality,
but it does not necessarily reflect all the information that the human observer sees and
responds to when viewing a scene.
3.2.3 Layout of the Photograph Insert
Two inserts were developed to accompany the survey instrument: the baseline insert, used
with Versions 1 through 5; and the Versions 6 insert, which focused upon the Southwest and
the Grand Canyon National Park. The baseline insert is a single sheet of glossy stock that
measures about 16 by 17 inches and is folded three times. The Version 6 insert measures
about 11 by 13.5 inches. The title pages, shown in actual size in Figures 3.2-1 and 3.2-2,
state that the visibility impacts are due to air pollution on days without rain or natural fog
to reinforce that the survey concerns man-made pollution impacts upon visibility. The map,
which is about 17 by 8.5 inches in the baseline insert and 12 by 5.5 in the Version 6 insert,
is shown reduced in Figure 3.2-3. The map shows the continental United States divided
into six regions and identifies all of the national parks that are defined as Class I areas
under the Clean Air Act and that NPS has identified as parks where visibility is considered
an important resource (40 CFR Part 81.400).
Inside the baseline insert, four 3 by 5 inch photographs for each of three parks are
positioned in vertical columns as illustrated in Figure 3.2-4. The name of the park is given
at the top of each of the three columns: Yosemite, Grand Canyon, and Shenandoah. The
A photos are in the top row, and these are the 90th percentile photos for each park
showing the best visibility conditions. The visibility descends to the bottom row showing
the 10th percentile for each park. The Version 6 insert includes just the Grand Canyon
photographs as arranged in Figure 3.2-5. The captions are the same for the four photos for
each park and read as follows:
A.
Visibility
on
about
15%
of
days
B.
Visibility
on
about
20%
of
days
C.
Visibility
on
about
40%
of
days
D.
Visibility
on
about
25%
of
days
RCG/Hagler, BaiHy, Inc.
-------
Figure 3.2-1
Insert Title Page
Version 1 through 5
VISIBILITY
IN THE
NATIONAL
PARKS
Visibility Impacts
Due to Air Pollution
on Summer Days
Without Rain or Natural Fog
Photographs and Map Inside
RCG/Hagler, BaiHy, Inc.
-------
Figure 3.2-2
Insert Title Page
Version 6
VISIBILITY
IN
GRAND CANYON
NATIONAL PARK
Visibility Impacts
Due to Air Pollution
on Summer Days
Without Rain or Natural Fog
Photographs and Map Inside
RCG/Hagler, BaiHy, Inc.
-------
Figure 3.2-3
Insert Map
NATIONAL PARKS WHERE VISIBILITY IS
CONSIDERED AN IMPORTANT RESOURCE
Ot
Vmoic
^TV-northwest
Uoun,
c,ai*' U*e
Uheoqore Qoosevefl
1 cua r,^
^ Grand
Peinfiea Sanoet^
i ^ c forest 0
CALIFORNIA # ^1° / ,
yCh."cahu3 fC3"sb3rl Caverns J
voyageurs
isJe Royafe
NORTHEAST
^,caa»a
CENTRAL U.S.
_Jr ¦
SOUTHWEST
Guaaatuoe Mountains
Ca
h/OOnia'r
-------
Figure 3.2-4
Insert Photograph Layout
Versions 1 through 5
-YOSEMITE
GRAND CANYDN-
SHENANDOAH-
A. visibly on about 15% of days
A. Visibifity on about 15% of days
a. visibifity on about 15% of days
8 visibfity on about 20% of days
B. Visjb
-------
Figure 3.2-5
Insert Photograph Layout
¦¦ Version 6 mmmm
A. Visibility on about 15% of days GRAND CANYON B. Visibility on about 20% of days
C. Visibility on about 40% of days
D. Visibility on about 25% of days
-------
3-25
These captions describe the approximate frequencies that a range of conditions represented
by the photographs occur. For example, Photo A is approximately the 90th percentile
photograph and is representative of conditions from the 85th to the 100th percentiles.
Photo B is approximately the 75th percentile photograph and is representative of conditions
from the 65th to the 85th percentiles. Photo C is approximately the 50th percentile
photograph and is representative of conditions from the 25th to the 65th percentile. The
Photo D is approximately the 10th percentile photograph and is representative of conditions
from the 0th to the 25th percentile.
Additional information about the photographs is given in Section 3 of the questionnaire and
reads as follows:
Throughout the U.S., air pollution from outside the parks causes haze that reduces
how well a person can see in national parks and into scenic vistas outside park
boundaries.
The enclosed photographs show different levels of air pollution at three national parks
on days without rain or fog. The conditions at these parks are typical of summertime
conditions at the national parks throughout the region in which each park is located.
Photogranh A shows almost no haze. This occurs on about 18 summer days each
year (about 15% of the time).
Photogranh B shows a little haze. This occurs on about 24 summer days each
year (about 20% of the time).
Photogranh C shows average visibility conditions. This occurs on about 48
summer days each year (about 40% of the time).
Photogranh D shows a lot of haze. This occurs on about 30 summer days each
year (about 25% of the time).
3.3 SURVEY IMPLEMENTATION PROCEDURES
The full study was implemented using a mailing of the final survey instrument and a
telephone follow-up on non-respondents.
3.3.1 Survey Mailings
Random samples of residents of five states were selected to receive the questionnaire. The
number of households from each state selected to receive each version of the questionnaire,
and the response rates, are shown in Table 4.1-1. Five states were selected for the sample:
Arizona, California, Missouri, New York, and Virginia. These states were selected, subject
to project budget constraints, based upon a variety of considerations:
RCG/Hagler, Bailly, Inc.
-------
3-26
1. To include an adequate number of responses for individuals living both near and
far from the national parks of interest because distance was expected to be a
potentially important factor in the WTP responses (Sutherland and Walsh, 1985).
2. For each survey version, residents were sampled from the same state as the park
that served to illustrate the national parks in the focus region. Home- state
residents were oversampled to improve estimation accuracy for individuals with
higher probability of future use. Residents in the home state are expected a_
priori to have a higher probability of previous and future visits compared to
residents of the other four states in the sample.
3. For each survey version, residents were sampled from a state with one of the
other national parks pictured on the photograph insert. In a sense, this allows
for tests for "competing park" impacts.
4. For each survey version, residents were sampled from states located in regions
not represented on the photograph inserts, and with relatively few prominent
national parks. The two states selected in this category also represented a range
of characteristics, as compared to each other and the rest of the sample, in terms
of urbanization and distance from the focus parks.
The sample was selected from a list of U.S. residents maintained by Ed Burnett
Consultants, Inc., Englewood, New Jersey. This list is based on sources that include the
following:
telephone books
drivers license information
car registration
voters registration records
survey information
warranty card information
mail order buyers information
Households from the master files were selected with an approximate equal probability of
inclusion by first randomly selecting a set of zipcodes from each state (the number selected
equaled the sample size and any zipcode could be sampled repeatedly), then randomly
selecting a household from each zipcode. This approach was selected as the number of
addresses per zipcode are targeted by the postal service to be roughly equal.
The questionnaire mailings followed a modified Dillman (1978) approach with an advance
letter and three follow-up mailings. Copies of the letter or postcard sent with each mailing
are include in Appendix B. The schedule of these mailings was as follows.
RCG/Hagler, BaiHy, Inc.
-------
3-27
August 23,
1988:
Introduction letter
September
6, 1988:
First questionnaire copy
September
13, 1988:
Reminder postcard
September
28, 1988:
Second questionnaire copy
November
3, 1988:
Third questionnaire copy
Figure 3.3-1 shows the responses received each week following the mailing of the first
questionnaire copy. Subjects were removed from the mailing list as their responses were
received.
3.3.2 Telephone Folio w-up
A sample of non-respondents was selected from each of the five states for a telephone
follow-up in January 1989. This sample was taken from households originally sent Versions
2 (Southeast national parks) or 3 (Southwest national parks) of the questionnaire.
Telephone directories were used to obtain telephone numbers for selected non-respondents.
CIC Research of San Diego, California, randomly selected among the non-respondents until
300 names and numbers were obtained, and conducted the telephone interviews. The
purpose of the telephone follow-up was to determine the rate of remaining bad addresses
in the original sample, and to examine the likely sign and significance of any potential non-
response bias.
A copy of the questions for the Grand Canyon version of the telephone follow-up is
included in the Appendix. The first three questions concerned past and potential future
visitation to national parks in the Southwest (or Southeast). All of the telephone
respondents were asked if they thought it was "not at all important," "somewhat important,"
or "extremely important," to prevent visibility from getting worse at national parks in the
Southwest (or the Southeast). They were also asked the same question about obtaining
potential improvements in visibility.
Telephone respondents were asked if they still had the photograph insert that was sent with
the questionnaire and, if so, a few specific WTP questions were asked. However, only four
could quickly locate it. Telephone respondents who no longer had the photograph insert
were asked whether they would be willing to pay anything in additional prices or taxes to
(1) prevent visibility from getting worse at national parks in the Southwest (or Southeast),
and (2) obtain improvement in visibility at national parks in the Southwest (or Southeast).
Six questions about the respondent and the household were then asked to compare the
sample characteristics to the mail respondents.
RCG/Hagler, BaiHy, Inc.
-------
Figure 3.3-1
Responses Received Each Week
Number of Responses
First
Enclosure
Postcard
610
1
256
I
I
I
I
312
1
I
Second
Enclosure
160
100
¦
I
61
Third
Enclosure
17 19
J
45
22
8
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Week Number
-------
4-1
4.0 RESULTS
4.1 RESPONSE TO MAILINGS
Table 4.1-1 shows the sampling plan and the responses obtained for each version of the
mail questionnaire. The overall unadjusted mail survey response rate, net of the returned
bad addresses, was 58.2 percent. After adjusting for estimated remaining bad addresses
identified though the telephone follow-up (discussed in Section 4.3) the overall mail
response rate was computed to be 73 percent. Including the telephone follow-up, responses
were obtained from approximately 76 percent of the valid sample. The response rates for
the California and New York residents were somewhat lower than for the residents of the
other three states.1
One issue of concern is whether respondents may be more or less likely to respond to
different survey versions, which may indirectly reflect differences in their ability or interest
in completing the instrument and, therefore, indirectly introduce a bias into the responses.
For example, respondents from the same state as a national park of interest, who may have
higher values, may also have higher response rates.
Table 4.1-2 shows comparisons of the response rates for the questionnaires concerning
different national park regions (Versions 1, 2, and 3). The first section of the table shows
a small, but statistically significant higher response rate for the questionnaire concerning the
Southeastern national parks than for the other two park regions. This difference is seen
in the five-state averages and for Missouri alone, but not for New York alone. The other
three states were not compared separately in this section due to potential confounding
effects of larger sample sizes in the home state of the park. The second section of the
table shows that there is no statistically significant difference in response rates when the
WTP questions focus upon national parks in the respondent's own state versus focusing
upon national parks in other regions.
Table 4.1-3 shows comparisons of response rates for different versions of the Southwestern
national parks questionnaire (Versions 5 and 6 versus Version 3), and for the multiple
region questionnaire (Version 4). Some statistically significant differences are shown, but
the directions of the differences are not consistent.
1 Elsewhere (Sanghvi et al. 1989, Peterson et al. 1987), we have experienced somewhat lower response rates
in California and we do not attribute the small differences in response rates across the states experienced in the
current effort as reflecting anything unique about the visibility survey, but rather we believe that they reflect
different rates of mobility resulting in different rates of undetected bad addresses, other socio-economic
differences across the states and different attitudes about completing such surveys.
RCG/Hagler, Bailly, Inc.
-------
UNADJUSTED
TOTAL BAD ADD REC'D NET X RECO
VERSION NAME:
CALIFORNIA PARKS (VI)
TOTAL MAILED/BAD ADDS
785
128
375
57. IX
ARIZONA
125
22
61
59.2X
VIRGINIA
125
17
67
62. OX
CALIFORNIA
285
50
128
54.5X
NEW YORK
125
23
55
53.9X
MISSOURI
125
16
64
58.7X
PERCENTS
100.OX
16.3X
47.8X
57.1X
UNADJUSTED
TOTAL
BAD ADD
REC'D NET
X RECD
VERSION NAME:
MULTIPLE REGIONS
-------
4-3
Table 4.1-2
Comparisons of Response Rates for Different Park Regions
Statistically
State Number Mean Standard Significant
of of Valid Percent Error of Differences
Residence Version Addresses Returned Mean (Direction)
I. Different Regions
5-State Total
657
665
661
57.1%
61.1%
57.2%
0.95
0.92
0.95
p < .01
(positive)
New York
102
102
106
53.9%
53.9%
56.6%
2.46
2.46
2.39
Missouri
109
113
111
58.7%
65.5%
54.1%
2.32
2.13
2.35
p < .05
(positive)
II. Home State Park
California 1
Arizona
2 + 3
3
1 + 2
Virginia 2
1 + 3
235
224
236
216
225
204
54.5%
53.6%
61.4%
62.5%
62.2%
60.3%
1.61
1.66
1.94
2.02
1.57
1.68
RCG/Hagler, BaiHy, Inc.
-------
4-4
Table 4.1-3
Comparisons of Response Rates for Different Questionnaire Versions
Statistically
State Number Mean Standard Significant
of of Valid Percent Error of Differences
Residence Version (s) Addresses Returned Mean (Direction)
I. Multiple Region WTP (Version 4)
Arizona 4 91
3 236
1+2 216
California 4 97
1 235
2+3 224
New York 4 95
1+2+3 310
II. Limited Information (Version 5)
Arizona 5 92
3 236
Virginia 5 95
3 96
New York 5 96
3 106
3-State Total 5 283
3 438
III. Single Region Focus (Version 6)
Arizona 6 91
3 236
California 6 96
3 97
New York 6 96
3 106
71.4% 2.14 p < .01
61.4% 1.90 (positive)
62.5% 2.02
42.3% 2.48 p < .01
54.5% 1.61 (negative)
53.6% 1.66
52.6% 2.56
54.3% 1.40
65.2 2.36 p < .10
61.4 1.94 (positive)
61.1 2.45
58.3 2.48
51.1 2.55 p < .05
56.6 2.38 (negative)
59.1 1.44
59.6 1.15
63.7 2.42
61.4 1.94
61.5 2.42 p < .01
50.9 2.53 (positive)
51.0 2.55 P < .05
56.6 2.39 (negative)
RCG/Hagler, Bailly, Inc.
-------
4-5
Overall, there appears to be very little variation in response rates for the different versions
of the questionnaire. Rather, response rates appear to vary across states due to other socio-
economic factors. We therefore conclude that variations in the survey versions do not
appear to introduce potential bias to the results as a result of varying response rates.
4.2 CHARACTERISTICS OF RESPONDENTS
Table 4.2-1 shows selected characteristics of the mail and telephone respondents and values
of comparable characteristics for the United States population. Focusing upon the mail
survey respondents, a greater percentage are male and the median age is somewhat higher
than for the national population. Household size is comparable. Education and income
are both higher for the respondents than for the national population. Because low incomes
are expected to be correlated with less education, it is not surprising that this population
group is under represented in the respondent group. It is not clear, however, how much
these differences are due to differences in characteristics in the sampled states versus the
entire U.S., or due to differences in the sample frame versus the characteristics of the
underlying populations, but this issue is addressed in part through the follow-up telephone
survey. The associations between these characteristics and the responses to the willingness
to pay questions are also examined in subsequent sections of this chapter and allow one to
interpret the impact of these differences upon the sample mean WTP estimates.
4.3 TELEPHONE FOLLOW-UP INTERVIEW OF NON-RESPONDENTS
While the mail survey response rates are high, bias may still enter into the data through the
self selection of respondents to the mail survey. The telephone follow-up survey was
undertaken to examine for the existence, direction and general magnitude of any potential
bias affecting the WTP estimates.2
A sample of non-respondents to the mail questionnaire was selected and telephoned to
obtain some information about the characteristics of non-respondents and to identify
potential additional bad addresses. Telephone respondents were asked a few questions
about their visitation at national parks and about the importance to them of visibility
protection for national parks in the specific region of interest. The telephone sample was
taken from the Version 2 (national parks in the Southeast) and the Version 3 (national
parks in the Southwest) samples. The questions asked in the telephone interview are
presented in the Appendix.
2 The results of the telephone survey should be interpreted cautiously. This is in part due to the much
smaller sample size (so the estimates are less reliable), unlisted numbers causing sampling problems, failure to
contact all targeted names, refusals to participate, and necessary differences in the survey instrument.
RCG/Hagler, Bailly, Inc.
-------
Table 4.2-1
Comparison of Respondents and
National Population Characteristics
4-6
Variable
Mail
Respondents
Telephone
Respondents
National
Completed Responses 1632
Sex:
Male 59%
Female 41%
Median Age (18 +) 44 years
Mean Household Size 2.77
Education:
Mean Years 14.0
1987 Mean Household Income $41,441
Income Distribution:
Under $10,000 7.0%
$10,000 - 49,999 63.0%
$50,000 + 30.0%
72
53%
47%
50 years
2.87
$34,483
49%
51%
40 years
2.78
12.6
$32,144
11.7%
65.5%
22.9%
RCG/Haglcr, Bailly, Inc.
-------
4-7
Table 4.3-1 shows the telephone sample and response rates. A total of 304 numbers were
selected from the non-respondents to Versions 2 and 3. An approximately equal number
was selected for each version. Of these, no contact was made with 56, and 12 said that they
had mailed back the questionnaire, leaving a net telephone sample of 236. Bad addresses
made up 47.5 percent of the net telephone sample. The bad address classification included
numbers not in service, business numbers, wrong numbers ("no one by that name lives
here"), and the respondent is dead. The net number of remaining good numbers was 124,
and of these, completed interviews were obtained with 72 (58.1 percent). When the
response rate to the mailed questionnaire is adjusted for this additional rate of bad
addresses (47.5 percent of the non-responses) the overall response rate to the mailed
questionnaire becomes 72.6 percent. The original and adjusted response rates for the
mailed questionnaire are shown in Table 4.1-1.
Included in the bad addresses identified in the telephone follow-up was a relatively large
number of "wrong numbers" and "that person is dead" responses. These are appropriately
interpreted as bad addresses, but to account for the possibility that some people may give
these responses just to get rid of the interviewer, we also calculated a lower bound overall
response rate to the mailed questionnaire treating all of these as refusals. Under this
assumption, 15.3 percent of the net telephone sample is bad addresses (rather than 47.5
percent) implying an overall response rate to the mailed questionnaire of 62.2 percent.
These lower response rates are also shown in Table 4.1-1. The true response is probably
closer to the 73 percent estimate as these responses were usually given prior to the
interviewer stating who they were or the purpose of the telephone contact. The computed
response rates would be higher if the telephone respondents with a language barrier were
also considered as bad addresses (because there is some likelihood of difficulty reading the
survey instrument).
As reported in Table 4.2-1, the telephone sample is older than the mail sample and national
average and has a household income comparable to the national average, but 17 percent
less than the mail respondents.
Table 4.3-2 shows the national park visitation characteristics of the telephone respondents.
Their average past and probable future visitation to the national parks is somewhat lower
than the average for the mail questionnaire respondents to Versions 2 and 3. The results
reported in Table 4.3-3, however, suggest that this lower visitation does not also reflect less
perceived importance of visibility protection for national parks. The responses concerning
the importance of preventing visibility from getting worse and the importance of improving
visibility in national parks were very similar to those given by the mail respondents to
Versions 2 and 3.3
3 The percentages shown in Table 4.3-3 for the mail responses were converted from the five point importance
scale used in the mail questionnaire to the three point scale used in the telephone questions by splitting the
groups at 2 and 4 on the scale in half and allocating half to each of the adjacent points on the scale (see Table
4.4-11).
RCG/Hagler, Bailly, Inc.
-------
4-8
Table 4.3-1
Telephone Follow-up Response Rates
Version 2 Version 3
Southeast Southwest Total Percent
Numbers Selected 149
No Contact 26
No Answer 21
Phone Busy
Respondent Deaf 2
Language Problem 3
Mailed Back Questionnaire 6
Net Phone Sample 117
Bad Numbers 51
Not in Service 13
Business Number 1
Respondent Died 2
Wrong Number 35
Net Good Numbers 66
Non-Response 31
Refusals 19
Terminated 1
Call-Back 11
Completed Interviews 35
155 304
30 56
25 46
2 2
2
3 6
6 12
119 236
61 112 47.5%
17 30
5 6
3 5
36 71
58 124
21 52
15 34
1 2
5 16
37 72 58.1%
RCG/Hagler, Bailly, Inc.
-------
4-9
Table 4.3-2
Summary of National Park Visitation
Characteristics of Telephone Respondents
Percent of
Phone
Respondents
Percent of
Mail
Respondents
Have visited any national park 64.7%
Have visited specific national park
(Grand Canyon or Shenandoah) 18.1%
Will visit a national park in the region (Southwest
or Southeast) in the next five years 14.7%
May visit a national park in the region
(Southwest or Southeast) in the next five years 44.1%
78.9%
35.8%
25.3%
48.5%
Will not visit a national park in the region
(Southwest or Southeast) in the next five years
41.2%
26.2%
Note: Telephone respondents were asked about either the Southwest parks or the
Southeast parks, not both. Fifty-one percent of completed interviews were for the
Southwest.
RCG/Hagler, BaiHy, Inc.
-------
4-10
Table 4.3-3
Summary of Importance to Telephone Respondents of
Visibility Protection for National Parks in
the Selected Region
Phone Responses Mail Responses
Number Percent Percent
Importance of preventing visibility from getting worse:
Not at all important
2
2.9%
3.6%
Somewhat important
16
23.5%
27.6%
Extremely important
50
73.5%
68.9%
Importance of improving visibility:
Not at all important
2
3.0%
5.0%
Somewhat important
25
37.3%
43.8%
Extremely important
40
59.7%
5 1.4%
Note: The rate of refusal to answer these questions was 5.5% to 7.0% for telephone
respondents and 3.5% to 4.5% for mail respondents.
RCG/Hagler, BaiHy, Inc.
-------
4-11
Table 4.3-4 shows the responses to the questions about whether the telephone respondent
would be willing to pay something for visibility protection for national parks. The responses
show that most respondents were willing to pay for visibility protection. The positive
response rate appears to be slightly lower than that obtained for the most comparable
question in the mail questionnaire for Versions 2 and 3."
Analysis of the mail questionnaire responses indicates that the dollar amount the
respondent gave in response to the WTP question was correlated positively with:
the household income;
the probable future visitation to national parks in the selected region;
the willingness to pay (on the five point scale) for visibility protection at national
parks; and
the importance rating for improving or preventing degradation in visibility at
national parks.
The responses to the telephone follow-up indicate that the telephone respondents were
similar for the fourth characteristic, but were somewhat less likely to visit national parks in
the selected regions and had lower household income than did the mail respondents. This
suggests that the dollar WTP for the telephone respondents might be somewhat, but
probably not dramatically, lower than for the mail respondents and the impact of non-
response bias upon the sample wide WTP estimate would be relatively small.5
To further examine for the potential magnitude of any such response bias, the mail
responses were categorized into five groups, each with about 200 observations, according
to the date of receipt, as shown in Table 4.3-5. The mean adjusted WTP response
(discussed in Section 4.5 below) is highest for the early respondents, then fairly constant and
dropping only slightly for the last 176 responses received (10 percent of all respondents, and
13.4 percent of the mail respondents), although such a drop was also experienced in the
middle of the sample and is not statistically significant. If the responses of this last group
are reflective of the entire 27 percent non-respondents, the sample-wide WTP mean would
drop by only six percent. If one assumed a more extreme assumption that the average WTP
for all non-respondents was as little as 50 percent of the value reported by mail survey
respondents, the sample-wide WTP estimates would drop by about 14 percent.
4 Exact comparison is difficult as the telephone survey used only a yes/no format while the mail survey used
a five point scale of willingness to pay for visibility protection (See Table 4.4-10).
s Based upon Table 4.3-2, and the results reported in Section 4.5, the predicted effect of the reduced
likelihood of future visits and reduced income for telephone respondents would be on the order of a 30 to 40
percent reduction in WTP. Assuming the telephone respondents are representative of all non-respondents, the
corresponding impact upon the computed sample-wide mean WTP would then be 10 percent or less.
RCG/Hagler, Bailly, Inc.
-------
4-12
Table 4.3-4
Willingness of Telephone Respondents to Pay Something
for Visibility Protection for National Parks
Phone Respondents
Response Number Percent
Willing to pay to prevent YES 43 70.5%
degradation NO 18 29.5%
Willing to pay to obtain YES 39 63.9%
improvement NO 22 36.1%
Refused to answer this 11 15%
question
RCG/Hagler, BaiHy, Inc.
-------
4-13
Table 4.3-5
Mean Adjusted WTP by Response Group
Adjusted
Date Received WTP1*
Before 9-19 mean 62.88
SE 9.42
N 191
9-19 mean 36.11
SE 3.74
N 226
9-20 mean 41.08
SE 4.81
N 249
9-21 to 9-27 mean 44.68
SE 5.68
N 266
9-28 to 10-11 mean 44.43
SE 7.54
N 205
After 10-11 mean 34.06
SE 4.38
N 176
* See Section 4.5 for definitions.
RCG/Hagler, BaiHy, Inc.
-------
4-14
4.4 RESPONSES TO GENERAL QUESTIONS
The first ten questions of each survey version concerned national park visitation and general
attitudes about protection of resources at national parks. The purpose of these questions
was twofold. One purpose was to obtain information that might be useful in analyzing WTP
responses from different individuals. Another purpose was to get the respondents thinking
about national parks and about the many different resource protection issues facing the
parks, of which visibility is only one (see Section 3.1).
National Park Visitation Patterns of Respondents
Table 4.4-1 summarizes the national park visitation history reported by the respondents.
More than three quarters of the respondents reported having visited a national park at
some time in their lives. Between one-quarter and one-half of the respondents had visited
each of the national parks used in the photographs set. Overall, visitation rates were
highest for the Grand Canyon and lowest for Shenandoah, but Virginia residents had
visitation experience at Shenandoah comparable to that of Arizona residents at the Grand
Canyon. Out-of-state residents were less likely to have visited Shenandoah than the other
two parks.
Visitation in the last two years showed a similar pattern. About one-half of the Arizona
residents said they had visited the Grand Canyon in the past two years. Almost 60 percent
of the Virginia residents had visited Shenandoah in the past two years. About 30 percent
of the California residents had visited Yosemite in the past two years.
Item non-response to the visitation questions, especially about specific parks, was fairly high.
The responses shown in Table 4.4-1 reflect responses made to previous questions when
appropriate. For example, if a respondent indicated he had never visited a national park
in California and did not answer the Yosemite question, he was coded as never having
visited Yosemite. Still, the higher non-response rate for out-of-state residents suggests that
they are more likely to be non-visitors than visitors. The high item non-response rate
overall may reflect some recall difficulty with these questions.
Table 4.4-2 summarizes responses concerning the probability of visiting any national park
in each of the three selected regions during the next five years. The mean response for
each of the regions is about three ("might visit"), a little higher for the Southwest and a
little lower for the Southeast. As with previous visitation, home-state residents give a
higher probability of visiting parks in their own regions in the near future, with a mean
response close to four ("probably will visit") in each case. The mean responses for out-
of-state residents fall between "probably will not visit" and "might visit."
The responses to the past and probable future national park visitation questions suggest that
questions about resource protection for national parks in these three regions are likely to
be meaningful for the vast majority of respondents in terms of their own personal
RCG/Hagler, BaiHy, Inc.
-------
4-15
Table 4.4-1
National Park Visitation History
Yes No No Response
Ever Visited any National Park?
All Respondents (n=1632) 80.5% 15.5% 4.0%
Ever Visited any NP in California?
California Residents (n=343) 77.6% 15.5% 7.0%
Residents of Other States (n=1289) 37.5% 43.7% 19.0%
All Respondents (n=1632) 50.0% 37.6% 16.5%
Ever Visited Yosemite NP?
California Residents (n=343) 61.2% 30.6% 8.2%
Residents of Other States (n=1289) 25.3% 60.3% 14.4%
All Respondents (n=1632) 32.8% 54.0% 13.1%
Visited Yosemite in Last 2 Years?
California Residents (n= 343) 30.6% 53.9% 15.5%
Residents of Other States (n=1289) 9.4% 71.6% 19.0%
Ever Visited any NP in the Southwest?
Arizona Residents (n=455) 88.1% 9.2% 2.6%
Residents of Other States (n=1177) 39.5% 42.1% 18.4%
All Respondents (n=632) 53.1% 33.0% 14.0%
Ever Visited Grand Canyon NP?
Arizona Residents (n=455) 78.2% 18.2% 3.5%
Residents of Other States (n=1177) 33.1% 55.0% 12.0%
All Respondents (n=1632) 45.6% 44.7% 9.6%
Visited Grand Canyon in Last 2 Years?
Arizona Residents (n=455) 52.1% 35.69c 12.3%
Residents of Other States (n=1177) 12.7% 69.9% 17.3%
Ever Visited Parks in the Southeast?
Virginia Residents (n=317) 76.0% 19.9% 4.1%
Residents of Other States (n=1315) 27.5% 53.1% 19.5%
All Respondents (n=1632) 36.9% 61.2% 12.7%
Ever Visited Shenandoah NP?
Virginia Residents (n=317) 76.7% 18.3% 5.0%
Residents of Other States (n=1315) 13.8% 71.6% 14.6%
All Respondents (n=1632) 26.0% 61.2% 12.7%
Visited Shenandoah in Last 2 Years?
Virginia Residents (n=317) 59.6% 31.2% 9.1%
Residents of Other States (n=1315) 4.5% 77.7% 17.5%
RCG/Hagler, BaiHy, Inc.
-------
4-16
Table 4.4-2
Probability of Future National Park
Visitation in Next Five Years
Definitely
Wffl Not
Visit
(1)
Frequency of Responses
Probably
Will Not
Visit
(2)
Might
Visit
(3)
Probably
Will
Visit
(4)
Definitely
Will Visit
(5)
Mean
Any NP in California
California Residents (n=272) 2.6% 9.2%
Residents of
Other States (n=1028) 13.3% 253%
All Respondents 11.1% 21.9%
18.4%
295%
27.2%
28.3%
22.0%
23.3%
41.5%
9.9%
16.5%
4.0
2.9
3.1
Any NP in
the Southwest*
Arizona Residents (n=439) 2.7% 7.3% 18.0% 33.5% 38.5%
Residents of
Other States (n=1027) 13.8% 225% 31.2% 20.5% 12.1%
All Respondents 10.5% 17.9% 27.2% 24.4% 20.0%
4.0
2.9
3.3
Any NP in
the Southeast
Virginia Residents (n=291) 4.5% 8.3% 25.1% 25.8% 36.4%
Residents of
Other States (n=982) 19.3% 37.3% 26.0% 11.7% 5.8%
All Respondents 15.9% 30.6% 25.8% 14.9% 12.8%
3.8
2.5
2.8
Version 6 respondents (n=165) were asked about probable future visitation to Southwest parks only.
RCG/Hagler, BaiHy, Inc.
-------
4-17
experience. For example, only 7 percent of the respondents said they had never visited any
national park in the Southwest and that they definitely would not visit parks in this region
in the next five years. Comparable percentages for California national parks and Southeast
national parks were 8 percent and 13 percent, respectively. Past and/or potential future
visitation to national parks in the region of interest should aid the respondents in
understanding the visibility protection good and how changes in the good may affect them.
Reasons and Priorities for Protecting National Park Resources
Table 4.4-3 shows the responses to the Question 5 concerning the importance to the
respondent of various possible reasons for visiting national parks. The majority of the
respondents ranked all of the listed reasons as at least somewhat important. The two
highest ranked reasons were related to enjoying nature and unique natural places, reasons
most closely linked to the preservation of the natural environment. Next came doing
something enjoyable with other people and having a change from usual surroundings. The
most frequently given "other" reason was also related to the preservation of the environment
as it involved witnessing or enjoying beauty. The most frequent other reasons that were
cited are listed in Table 4.4-4.
These ratings of reasons for visiting national parks are consistent with findings of recent on-
site research conducted at several national parks. Ross et al. (1985) conducted interviews
with visitors at Grand Canyon, Mesa Verde, Mount Rainier, Great Smoky Mountains, and
Everglades National Parks in which respondents were given a list of park features and
asked how important each feature was to their recreational experience. At all five parks,
features related to the natural environment were rated as having the highest importance to
visitors.
In Question 6 respondents were asked if they would want any of their taxes spent to
preserve and manage national parks even if they personally could never visit. Over 96
percent of the respondents said maybe yes or definitely yes. This supports the notion that
there is a widespread perception that protecting and preserving the national parks is
important for society, beyond the individual's interest in his or her own opportunity to visit
these areas.
Question 7 then asked about the importance of various potential reasons for preserving and
managing national parks independent of the respondent's own visitation. The responses to
these questions are summarized in Table 4.4-5 and 4.4-6. Again, the vast majority of the
respondents rated all of the listed reasons as at least somewhat important. The highest
ranked reasons were all related to preservation motives not necessarily related to use.
These were:
So there will be areas preserved in their natural condition, even if no one ever
goes there.
To preserve our national heritage.
So there is not development everywhere.
RCG/Hagler, BaiHy, Inc.
-------
4-18
Table 4.4-3
Reasons for Visiting National Parks
Not at all Slightly Somewhat Very Extremely
Important Important Important Important Important Mean
0) m LLii (4) (5)
To experience unique
natural places
Responses (n=1554) 2.4% 4.4% 19.6% 42.2% 31.5% 4.0
To experience unique
historic places
Responses (n=1549) 3.0% 9.5% 28.6% 39.6% 19.3% 3.6
To do something
enjoyable with other
people (for example,
family and friends)
Responses (n=1555) 3.7% 7.4% 21.2% 39.9% 27.9% 3.8
To enjoy the
vastness of nature
Responses (n=1559) 2.1% 4.4% 14.6 35.5% 43.4% 4.1
To take part in
outdoor recreation,
such as hiking,
fishing, or camping
Responses (n=1540) 12.1% 12.4% 23.2% 28.6% 23.8% 3.4
To have a change
from my usual
surroundings
Responses (n=1546) 5.3% 8.0% 24.1% 35.4% 27.2% 3.7
Please list any other reasons you like to, or would like to, visit national parks: See Table 4.4-4.
RCG/Hagler, Bailly, Inc.
-------
4-19
Most Frequent
Table 4.4-4
"Other" Reasons for Visiting National Parks
Frequency
Reason
56
To witness beauty
52
To relax, experience quiet
51
To see a particular park, or the
country in general
34
To experience fresh air or a different
climate
37
To see wildlife
29
To undertake a special activity such
as photography
25
To get away from the city
25
To observe nature
18
For spiritual restoration
16
For education
RCG/Hagler, BaiHy, Inc.
-------
4-20
Table 4.4-5
Reasons for Preserving National Parks,
Regardless of Own Visitation
-6 Federal taxes are used to preserve and manage national parks. If you personally could never visit a
national park, would you want any of your taxes spent to preserve and manage national parks? (Circle
number)
Definitely No Maybe No Maybe Yes Definitely Yes Mean
(1) (2) (3) (4) Score
Responses (n=1577)
% responding 2.2% 1.3% 16.7% 79.9% 3.74
Q-7
If you personally could never visit a national park in the future, how important to you
would each of the following reasons be to spend taxes to preserve and manage
national parks? (Circle number of best response for each reason)
Not at All Slightly Somewhat Very Extremely
Important Important Important Important Important
ID (2) (3) (_4) (5)
Mean
So other members of
my family will have
the opportunity to
visit these areas now
and in the future
Responses (n=1543)
1.6%
3.3%
14.2%
41.2%
39.7%
4.1
So people outside my
family will have the
opportunity to visit
these areas now and
in the future
Responses (n=1532) 1.6% 5.1% 20.8% 42.1% 30.5% 3.9
So there will be areas
preserved in their
natural condition, even
if no one ever goes there
Responses (n=1540) 3.1% 4.5% 13.6% 29.0% 49.9% 4.2
To allow scientific
research on nature or
history
Responses (n=1520) 3.2% 8.8% 24.9% 35.8% 27.4% 3.8
To preserve our
national heritage
Responses (n=1529) 1.8% 3.3% 14.8% 33.4% 46.8% 4.2
So there is not
development everywhere
Responses (n=1510) 3.8% 4.9% 11.7% 27.0% 52.7% 4.2
Do you have any other reasons? (Please list) See Table 4.4-6.
RCG/Hagler, BaiHy, Inc.
-------
4-21
Table 4.4-6
Most Frequent "Other" Reasons for Preserving
National Parks, Regardless of Own Visitation
Frequency Reason
84
To preserve wildlife/animal habitat
40
To preserve for the future
27
To preserve nature in general
12
To preserve clean areas
Close behind these were opportunities for others to visit, with visitation of other family
members being ranked somewhat higher than visitation of others outside the respondent's
own family. The most frequently given "other" reason was also tied to preservation of the
natural environment and involved the preservation of wildlife habitat.
Question 8 concerned priorities the respondents place on various potential adverse effects
on national park resources from human activities outside the parks. Degraded visibility due
to manmade air pollution is one of these effects. The responses also give us an idea how
visibility degradation ranks relative to other similar potential problems at parks and helps
set-up visibility protection as separate from other protection issues (to help minimize part-
whole bias) and as only one of many potential concerns (to help minimize importance bias).
The responses to Question 8 are summarized in Table 4.4-7. The majority of the
respondents rated protection of national parks against all of the listed effects at medium
or high priority. Notably lower on the list was preventing seeing or hearing mining or
industrial activities located outside park boundaries, with almost 25 percent giving this low
priority. Protecting visibility was about in the middle, with preventing water pollution injury
to aquatic life and preventing air pollution damage to vegetation rated somewhat higher.
Somewhat below visibility was air pollution damage to historic structures and aesthetic
damage from water pollution. Protecting parks from human pollution was the most
frequently given "other" priority (see Table 4.4-8).
General Attitudes About Visibility Protection for Parks
Question 9 asked about the effect of visibility improvements on visit enjoyment.
Respondents were asked to consider the potential effect of having conditions as shown in
the B photographs rather than the average conditions as shown in the C photographs during
a visit to a national park in each of the three regions. This question served to introduce
the photographs on the insert and the "substantive definitional" issue of whether varying
RCG/Hagler, BaiHy, Inc.
-------
4-22
Table 4.4-7
Priorities for Protecting National Park Resources
from Human Activity Outside the Parks
Q-8 Below are some types of effects that are happening or could happen in national
parks due to people's activities outside park boundaries. What priority do you give
to prevention of the following effects in national parks due to human activities
outside park boundaries? (Circle number of best response for each effect)
Low Medium High
Priority Priority Priority Mean
_0) (2) (3)
2.9% 25.2% 71.9% 2.7
Air pollution decreasing the ability
to see scenic vistas
Responses (n=1569)
Air pollution injury to
vegetation
Responses (n=1570)
Air pollution damage to
historic structures
Responses (n=1570)
Water pollution in streams or
lakes that harms fish or other
aquatic life
Responses (n=1581)
Water pollution that muddies streams
or lakes, but does not harm fish or
aquatic life
Responses (n=1562)
Park visitors being able to see or
hear mining or industrial activities
located outside park boundaries
Responses (n=1548)
1.5% 11.2% 87.3% 2.9
5.0% 29.4% 65.7% 2.6
0.7% 3.5% 95.8% 3.0
8.5% 41.7% 49.9% 2.4
22.6% 36.9% 40.5% 2.2
Are there other types of effects of special concern to you? (Please list) See Table 4.4-8.
RCG/Hagler, Bailly, Inc.
-------
4-23
Table 4.4-8
Most Frequent "Other" Responses for Priorities
in Protecting National Park Resources
Frequency Concern
44
Human Pollution
36
Threats to wildlife/animal habitat
23
Forest fires/fire management
21
Litter and defacement
20
Too much development
15
Noise pollution
15
Acid Rain
visibility would affect on-site enjoyment, and provided some information to help interpret
willingness to pay responses. The responses to Question 9 are summarized in Table 4.4-
9. Only about five percent of the respondents said that this would have no effect on their
enjoyment. About two-thirds of the respondents said that this would very much increase
their enjoyment. The responses for each region were very similar. I.e., the effect on on-
site enjoyment for the hypothesized changes is perceived to be roughly comparable across
the sites.
Question 10 asked about willingness of the respondent to pay something for visibility
protection at national parks as an introduction to and confirmation of the specific
willingness to pay questions that followed. The responses are summarized in Table 4.4-
10. As with the enjoyment question, responses for the different regions were very similar.
In-state residents indicated a greater willingness to pay for visibility protection than out-
of-state residents, but the differences were not as great as might have been expected based
on the visitation differences. The mean responses for in-state and out-of-state residents
exceeded "somewhat willing" for all three regions. Six to eight percent of in-state residents
were "not at all willing" to pay for visibility protection and nine to eleven percent of out-
of-state residents were "not at all willing."
RCG/Hagler, BaiHy, Inc.
-------
4-24
Table 4.4-9
Effect of Visibility Improvement on Park Visit Enjoyment
Q-9 If you were to visit a national park in each of these regions, you would probably have
average visibility like Photograph C. How do you think having somewhat less than
average haze due to air pollution, like Photograph B rather than Photograph C,
would affect your enjoyment of the visit? (Circle number of best response for each
region)
Region
Have no
effect on
enjoyment
Somewhat
increase
enjoyment
(2)
Very much
increase
enjoyment
(3)
Mean
California
Response (n=1392) 4.1%
27.2%
68.8%
2.6
The Southwest*
Response (n=1559) 5.1%
32.4%
62.5 %
2.6
The Southeast
Response (n=1389) 5.1%
32.3%
62.6%
2.6
* The 165 respondents to Version 6 were asked about the Southwest only.
RCG/Hagler, BaiHy, Inc.
-------
4-25
Table 4.4-10
Willingness to Pay Something for Visibility Protection
Frequency of Responses
Region
Not at all
Willing
(1)
(2)
Somewhat
Willing
(3)
(4)
Extremely
Willing
(5)
Mean
California Parks
California Residents
(n=274)
5.8%
10.6%
27.7%
23.7%
32.1%
3.7
Other State Residents
(n=1104)
10.6%
11.2%
36.0%
25.4%
16.9%
3.3
Southwest Parks*
Arizona Residents
(n=446)
Other State Residents
(n=1097)
7.0%
8.8%
8.7%
10.7%
30.9%
35.3%
28.7%
26.1%
24.7%
19.2%
3.6
3.4
Southeast Parks
Virginia Residents
(n=303)
Other State Residents
(n=1068)
7.9%
10.3%
7.3%
12.1%
27.7%
35.4%
31.7%
24.5%
25.4%
17.7%
3.6
3.3
Version 6 respondents (n=165) were asked about willingness to pay for visibility protection at southwest
parks only.
RCG/Hagler, BaiHy, Inc.
-------
4-26
Question 11 shifted the focus to one specific region and further helped to introduce the
WTP questions, and provided information for evaluating the WTP responses. Subjects were
asked to rate the importance of obtaining improvements, and of preventing degradations,
in visibility at national parks in the selected region. In most versions of the questionnaire
subjects were asked about only one region in this question. In Version 4 the question
referred to all three regions.
The responses to Question 11 are summarized in Table 4.4-11. As with the previous
questions, the responses for different regions are very similar. The difference between in-
state and out-of-state respondents is less pronounced than for Question 10 suggesting a
perceived difference between asking about "importance" versus "will you pay." Preventing
a degradation in visibility is consistently rated more important than obtaining an
improvement, although the average ratings for both are between "somewhat" and
"extremely" important. More than half of both in-state and out-of-state respondents said
that preventing degradation is "extremely important" for all three regions.
Non-Response Rates for Specific Questions
Table 4.4-12 shows non-response rates for selected specific questions for all versions of the
questionnaire. There are usually some respondents who do not answer any particular
questions, so some non-response is always expected. Of interest is whether a question has
a notably higher non-response rate than other questions, suggesting that respondents may
have had particular problems or concerns with that question. In some cases, the respondent
wrote on the questionnaire that he or she did not know the answer or did not accept a
premise, but in most cases the response was left blank without any explanation given.
The non-response rates for questions Q-5 through Q-ll were all between 4 and 6 percent.
Therefore, approximately 5 percent can be considered as the baseline expected non-
response rate for questions in this questionnaire. The non-response rates to the national
park visitation history questions were shown in Table 4.4-1, and were mostly higher than
5 percent, suggesting some possible problems with recall for these questions. The question
concerning the probability of future national park visitation (Q-4) had a non-response rate
of about 10 percent, perhaps reflecting uncertainty about future visits.
The non-response rates to the willingness to pay questions (Q-12 through Q-13) were 8 to
9 percent. Because we expect that some respondents will find these questions difficult, and
others will use non-response as a form of protest response, it is not surprising that the non-
response rate is higher than the 5 percent baseline level. It is reassuring that the WTP
non-response rate is actually lower than for many of the park visitation questions and for
the household income question (which typically draws a relatively higher non-response
rate).
RCG/Hagler, BaiHy, Inc.
-------
4-27
Table 4.4-11
Importance of Visibility Protection at National Parks
Frequency of Responses
Not at all Somewhat Extremely Mean
Important Important Important
(1) (2) (3) (4) (5)
California Parks Focus Version
California Residents
Obtaining Improvement 0.8% 1.6% 21.1% 31.7% 44.7% 4.2
(n=123)
Preventing Degradation 0% 0.8% 12.2% 22.8% 64.2% 4.5
(n=123)
Other State Residents
Obtaining Improvement 1.7% 2.1% 29.4% 28.2% 38.7% 4.0
(n=238)
Preventing Degradation 2.1% 0.8% 14.8% 25.3% 57.0% 4.3
(n=237)
Southwest Parks Focus Version
Arizona Residents
Obtaining Improvement 1.0% 3.7% 21.4% 34.8% 39.1% 4.1
(n= 299)
Preventing Degradation 1.0% 1.3% 13.6% 22.2% 61.9% 4.4
(n=302)
Other State Residents
Obtaining Improvement 2.7% 4.4% 27.5% 30.2% 35.1% 3.9
(n=367)
Preventing Degradation 2.2% 3.3% 14.2% 24.3% 55.9% 4.3
(n=367)
Southeast Parks Focus Version
Virginia Residents
Obtaining Improvement 3.1% 5.4% 27.1% 33.3% 31.0% 3.8
(n=129)
Preventing Degradation 1.5% 3.8% 10.6% 26.5% 57.6% 4.4
(n=132)
Other State Residents
Obtaining Improvement 2.4% 4.1% 28.2% 29.8% 35.5% 3.9
(n=245)
Preventing Degradation 2.0% 2.9% 16.4% 25.4% 53.3% 4.3
(n=244)
All Three Regions (Version 4s)
Obtaining improvement 0.7% 4.0% 27.5% 36.2% 31.3% 3.9
(n=149)
Preventing Degradation 0.7% 2.7% 16.0% 22.7% 58.0% 3.4
(n=150)
RCG/Hagler, Bailly, Inc.
-------
4-28
Table 4.4-12
Non-Response Rates for Selected Questions
Don't Left Total
Know (%) Blank (%) Non-Response (%)
A. For All Respondents Who Answered the Questionnaire (n=1632)
Q-4 Future Park Visitation
California
0.2%
10.0%
10.2%
Southwest
0.3%
9.9%
10.2%
Southeast
0.3%
11.6%
11.9%
Q-12 WTP1
1.8%
6.8%
8.6%
Q-13 WTP2
1.6%
6.4%
8.0%
Q-14 WTP3
1.4%
7.7%
9.1%
Q-27 Education
0.6%
3.1%
3.7%
Q-28 Income
0.1%
13.7%
13.8%
B. For All Respondents who Answered Non-zero to at Least One WTP (n=1403)
Q-16 Accuracy
0.2%
1.7%
1.9%
Q-17 % for Visibility
0.4%
4.8%
5.2%
Q-18 % for One Park
0.9%
6.1%
7.0%
Q-19 % for Option Price
0.3%
7.7%
8.0%
% for Request
0.2%
7.8%
8.0%
% for Existence
0.2%
7.7%
7.9%
RCG/Hagler, BaiHy, Inc.
-------
4-29
Questions Q-16 through Q-19 were relevant for only those respondents who gave a non-
zero dollar response to at least one of the WTP questions. Non-response rates for this
subsample for these questions are shown in the second part of Table 4.4-12.
4.5 WILLINGNESS TO PAY RESPONSES FOR CHANGES IN VISIBILITY
4.5.1 Evaluation of WTP Responses
CVM responses are considered problematic when they reflect a rejection of the scenario,
rather than revealing the consumer's surplus value of the change in visibility being valued.
For example, some respondents may list zero, while stating the problem is very important
to them, because they reject the payment scenario. These types of problems are common
in CVM surveys, and they need not invalidate the use of a large number of valid responses.
To address problems of potentially invalid observations, and to address the overall validity
of the CVM data, we use a procedure of "consistency checks" proposed by Rowe and
Chestnut (1985), which extends the zero bid evaluation procedures used in early CVM
studies (Cummings et al. 1986). Consistency checks identify CVM responses that are
inconsistent with other responses given by that respondent, including written comments.
Comments and responses to other questions may suggest that the CVM responses do not
reveal the value estimate sought by the researcher. Consistency checks can also be used
to help evaluate the apparent validity of the overall CVM responses. One would question
the validity of the CVM responses if a substantial share of respondents failed rudimentary
tests of internal consistency (Fischhoff and Furby, 1988, page 161-162).
Distribution of WTP Responses Across the Checklist
The responses to all of the WTP questions for each version of the questionnaire covered
the whole range of values listed in the checklist. The pattern of responses in terms of the
frequency with which each value was selected is similar for each of the WTP questions.
Table 4.5-1 gives an example of this pattern based on the responses to Question 12 for
Version 1, focusing on national parks in California. The first part of Table 4.5-1 shows the
percentage of respondents who selected values in each of the columns of the checklist
(there were four values in each column). Column 4 showed the greatest frequency of
responses at about 25 percent, with columns 3 and 5 both at about 20 percent. Values in
the first two columns (the lowest value) were selected somewhat more frequently than those
in the last two columns (the highest value).
The second part of Table 4.5-1 sheds more light on this observed pattern of responses. For
this question, approximately 53 percent of the respondents selected one of five values out
of the checklist of 28 values. These "key" values appear to be round numbers and are
located at various portions in the checklist (i.e., sometimes at the top of
RCG/Hagler, Bailly, Inc.
-------
4-30
Table 4.5-1
Example of Distribution of WTP Responses Across the Checklist
A. Distribution of Q-12 Responses for Version 1 Across the Checklist Columns
Checklist
Column 1 2 3 4 5 6 7
Dollar $0 to $2 to $8 to $25 to $60 to $150 to $400 to
Range $1.50 $5 $20 $50 $125 $300 more
than $750
Percent
of
Respondents* 11.1% 12.2% 18.1% 25.1% 21.1% 8.5% 3.8%
B. Most Frequent Values Selected for Q-12, Version 1
Percent of
Value
Respondents
$0
7.3%
$10
8.8%
$25
13.2%
$50
9.9%
$100
13.7%
Total
52.9%
Percentage based on the total number of respondents who gave any answer to this
question (n=342).
RCG/Hagler, BaiHy, Inc.
-------
4-31
columns, sometimes in the middle, etc.). Two of these key values are located in the
fourth column, which may account for the higher response rate for that column. The
tendency to select round numbers, such as $25 or $100 rather than $8 or $125, is suggestive
of the level of precision of the WTP responses.
For all three park regions, the percentage of respondents who selected one of the five "key"
values for the second and third WTP questions dropped to under 50 percent. This suggests
that many respondents may have selected a key value for the first response and then
adjusted their responses for the second and third WTP questions.
Evaluation of Zero WTP
In most contingent valuation studies an effort is made to evaluate zero WTP responses to
determine whether the respondent really means he does not value the hypothetical change
being considered, or whether the response reflects some objection to the question and
should not be interpreted as a true zero value for the change in the good in question. Irwin
et al. (1989) suggest that zero bids may also reflect that respondents do not know their
value, or do not want to expend the effort required in the exercise and therefore opt not
to engage in the transaction.
A common zero bid evaluation approach has been to list several alternative explanations
for a zero WTP response and ask the respondent to select the one that best explains his
response. Examples might be "I don't care about visibility" and "Polluters should pay for
pollution control." Our experience with this procedure suggests that some true zeros may
be incorrectly interpreted as protest responses as respondents may further justify true zero
value responses by also checking a protest comment (Rowe and Chestnut 1985, 1986).
The consistency check approach used to evaluate zero responses was to look for
confirmation in the comments or in responses to other questions that the respondent really
cares very little about the hypothesized changes in visibility at national parks in the selected
region. The evaluation was designed to err on the side of keeping invalid zeros rather than
eliminating valid zeros. Responses to Question 9 and Question 11, and written comments,
were used in this evaluation. Zero responses were retained as valid if any one of the
following conditions was met.
Response to Question 9 indicated that an improvement in visibility from C to
B in the selected region (in the WTP question) would not affect the respondent's
enjoyment of a park visit in that region.
Response to Question 11 indicated that improving or preventing deterioration
in (comparable to the direction of change hypothesized in the WTP question)
visibility conditions at parks in the selected region is not at all important.
Respondent offered a comment that he could not afford to pay anything.
Respondent offered a comment that he does not care about visibility conditions
either at the selected location or at all.
RCG/Hagler, Bailly, Inc.
-------
4-32
Both of the following conditions had to be met for the zero response to be interpreted as
a rejection of the question rather than a valid zero:
1. Respondent said that enjoyment would be enhanced at the better visibility level
(Q9), or that improving visibility (or preventing degradation) is at least somewhat
important (Qll).
2. A rejection comment was given, indicating some objection to the premises of the
WTP questions. Rejection comments include:
Polluters should pay. Taxes shouldn't be raised. The government should
use current tax revenues more efficiently. I don't believe the problem you
describe really exists (e.g., it's humidity, not air pollution). Park users
should pay. Visibility is an urban problem, not a park problem (i.e., your
assumptions are wrong).
Most of the zero WTP responses fell unambiguously into either the accept or reject group,
but for a few respondents neither the acceptance nor rejection conditions were met
(primarily due to non-response to Questions 9 and 11 and no comments being offered).
These zero WTP responses were presumed to be valid. The number of zero responses
given to each of the three WTP questions is shown in Table 4.5-2. Zero responses were
about eight percent of all responses to the questionnaire. After this evaluation process, just
under 70 percent of the zero responses were kept as valid zero value responses. The
rejected zero responses were recoded to missing values. The accepted zeros shown in Table
4.5-2 (the table rows labeled "adjusted") also reflect the addition of some valid zeros as a
result of the adjustments based on Question 17. The rejection/missing response rate is
similar to or lower than for many previous CVM exercises (Mitchell and Carson 1989,
Cummings et al. 1986). Table 4.5-4 also shows that protest and non-response bidders were
less likely to visit the parks and placed somewhat lower importance on visibility protection,
factors associated with lower WTP responses. If one made the extmm£-assumption that all
the protests and non-responses actually reflect true zero value, the sample-wide WTP means
would decrease approximately ten percent.
Adjustments Rased on Question 17- WTP Specifically for Visibility
One of the concerns with previous contingent valuation studies for estimating preservation
values for visibility at national parks has been that respondents may be giving WTP
responses that reflect their general desire to support the preservation and protection of
the parks and may therefore overstate their WTP for the specific incremental changes in
visibility conditions hypothesized in the question. The questionnaire was designed to
examine the magnitude of this potential part-whole bias by identifying as a separate issue
and requesting that respondents focus specifically on visibility values prior to the WTP
questions. After the WTP questions, recognizing the potential cognitive difficulty for
respondents in performing the separation ex ante, Question 17 asked respondents to
consider whether their WTP responses were specifically for visibility in and around national
parks or whether they also reflected some value placed on national park protection in
general. If they indicated that the entire value may not be for visibility, they were asked
to estimate the percentage of their WTP responses that was specifically for visibility.
RCG/Hagler, BaiHy, Inc.
-------
4-33
Table 4.5-2
Frequency of Refusals and Zeros in WTP Responses
Refusals and Zeros
Non-Response
WTP1 (Question 12)
Raw 140 125
(8.6%) (7.7%)
Adjusted* 185 107
(11.3%) (6.6%)
WTP2 (Question 13)
Raw 131 132
(8.0%) (8.1%)
Adjusted* 176 111
(10.8%) (6.8%)
WTP3 (Question 14)
Raw 149 129
(9.1%) (7.9%)
Adjusted* 195 110
(11.9%) (6.7%)
Note: Percentages are of all 1,632 respondents.
*
Adjusted removing protest zeros, questionable high WTP and corrected for Question
17 responses.
RCG/Hagler, BaiHy, Inc.
-------
4-34
Table 4.5-3
Visibility Focus Verification: Question 17
Question: Would you say the dollar amounts you gave in answer to Question 12, 13, and
14 are:
Mean Percentage
Percentage of WTP
of Specifically for
Response Responses* Visibility**
1. Basically for the stated changes 32% 87%
in visibility at the national (n=449) (n=449)
parks
2. Somewhat for the stated changes 45% 51%
in visibility and somewhat to (n=624) (n=607)
help with other needs at the
national parks
3. Basically to help the national 7% 40%
parks and are not related to the (n=103) (n=97)
stated changes in visibility
4. Other 6% 42%
(n-81) (n=74)
No response to first part of question 10% 50%
(n=145) (n=102)
Overall mean % for visibility 62%
(n=1329)
Percent of all respondents who gave a non-zero response to at least one WTP question.
** Respondents who answered 2, 3, or 4 were directed to a box asking "About what
percent of your dollar answers is for visibility at national parks?" Some respondents
who answered 1 to the first part of the question also marked something other than
100% in response to the second part of the question (115 out of 449).
RCG/Hagler, BaiHy, Inc.
-------
4-35
Table 4.5-4
High WTP and Non-Response WTP Evaluation
All Identified Absolute High Protests and Other WTP
High WTP1 WTP Respondents2 Non-Respondents Respondents3
n=47 n=14 n=166 n=1419
Importance of Improving 4.60* 4.86* 3.88 3.98
Visibility at National (SE=0.08) (SE=0.10) (SE=0.10) (SE=0.03)
Parks in the Region
Importance of Preventing 4.89* 5.00* 4.17** 4.35
Visibility Degradation (SE=0.08) (SE=0) (SE=0.09) (SE=0.02)
at National Parks in
the Region
Effect of Visibility Improvement 2.83* 2.79 2.53 2.58
on Visit Enjoyment (SE=0.06) (SE=0.11) (SE=0.05) (SE=0.02)
Willingness to Pay to Protect 4.47* 4.64* 2.88** 3.44
Visibility at National (SE=0.10) (SE=0.17) (SE=0.12) (SE=0.03)
Parks in the Region
Ever Visited a National Park 0.85* 0.91* 0.54 0.59
in the Region (SE=0.06) (SE=0.09) (SE=0.05) (SE=0.01)
Probability of Future Visit 3.72* 3.86 2.80** 3.28
to a National Park in the (SE=0.19) (SE=0.31) (SE=0.11) (SE=0.03)
Region
1 Includes all respondents who gave identified absolute and relative to income high WTP responses accepted
as valid.
3
2 Includes all respondents who gave accepted as valid WTP of "more than $750" for all three WTP questions.
These are also included in the first column.
Includes all respondents who gave accepted WTP responses, including zeros and excluding protests and non-
responses.
* Statistically significantly higher than for other respondents at 95% confidence.
** '¦ Statistically significantly lower than for other respondents at 95% confidence.
RCG/Hagler, BaiJly, Inc.
-------
4-36
Table 4.5-3 summarizes the responses to Question 17. The average percentage of the WTP
response specifically for visibility was 62 percent across everyone who answered this
question.6 The second column in Table 4.5-3 shows that the mean percentage for visibility
decreased, as expected, with responses indicating less and less connection between the
previous WTP answers and the specific changes in visibility described. Respondents who
selected the first response to the first part of Question 17 and did not answer the second
part were assumed to be indicating that 100 percent of their WTP for the specific changes
in visibility.
Figure 4.5-1 shows the frequency of the Question 17 responses regarding the percentage
of the WTP response that the respondent said was specifically for visibility in and around
national parks. About one-fourth of the respondents indicated 100 percent, and about one-
fourth indicated 50 percent. Twenty-six respondents indicated that none of their WTP value
was really for changes in visibility. This was about two percent of those who gave non-
zero responses to the WTP question.
The responses to Question 17 were used to adjust the WTP responses to more accurately
reflect WTP specifically for the hypothesized changes in visibility conditions. Mean
percentages for all respondents were used for subjects who answered the WTP questions
but did not answer Question 17. If they answered the first part of Question 17, then the
mean percentage given by others who selected the same first part response was used (unless
they selected the first response for which 100 percent was presumed unless otherwise
indicated by the respondent). If they did not answer any part of Question 17, then 62
percent was used.
To examine the factors that may be related to the percentages given specifically for
visibility, and to help in understanding the significance of the responses to this question,
correlations between the percentages for visibility and responses to other questions were
examined. This analysis found that the percentages for visibility were not related to the
WTP responses, but there was a statistically significant negative correlation between
responses to Question 17 and responses to Question 16 on perceived accuracy of the WTP
responses. This is consistent with expectations: those who believed their responses were
very accurate said that a higher percentage of their WTP response was specifically for
visibility and those who felt their responses were less accurate were more likely to reveal
a tendency toward part-whole bias. Other statistically significant (p < .01) positive
correlations emerged between the percentage for visibility and:
6 Many of the 449 respondents who responded
indicating what share of their bid was for visibility,
About one fourth of the 449 individuals in this group
"basically for the stated changes ..." completed the box
even though they were not explicitly directed to the box
indicated values less than 100 percent.
RCG/Hagler, Bailly, Inc.
-------
Figure 4.5-1
Frequency of Question 17 Responses:
Percentage for Visibility
Number of Responses
400
300
200
100
0
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percentage for Visibility
Note: includes respondents to all
versions who gave at least one
non-zero WTP response
-------
4-38
national park visitation and home-state residence
percentages of WTP for own household use and use of others now and in the
future (Question 19), which may reflect more concern about how visibility would
substantively affect on-site enjoyment by themselves or others leading to more
attention to the specific visibility concern.
the importance rating for obtaining improved visibility at national parks in the
selected region and for preventing degradation of visibility (Question 11)
the priority rating for protection of scenic vistas from air pollution (Question 8
showing an increased concern for visibility vis-a-vis other issues)
household income (perhaps reflecting education)
These correlations suggest that the WTP responses are likely to be more closely aligned
with the specific visibility changes being hypothesized when, as Fischhoff and Furby suggest,
the substantive definition is relevant to the individual. Such relevance may be greater when
the respondent is more likely to actually visit the park, is more concerned about actual
visitation by his own household and by others, and places a higher importance on protection
of visibility at national parks. The responses to this question also indicate that the
overstatement of WTP for changes in visibility in previous studies may have been
substantial, and that even when visibility is partitioned by the researcher in the WTP set-
up and valuation questions, some respondents have difficulty with the chore. However, the
follow-up question appeared to work well in allowing respondents a chance to reconsider
(in a non-threatening way) the intent of their WTP response and to adjust the response,
thereby minimizing the impact of any potential part-whole bias on the final WTP results.
Evaluation of High WTP Responses
Another way that respondents can object to the WTP question, or misrepresent values, is
by giving very high responses that misstate their true values. Most contingent valuation
studies have found that some respondents give apparently unrealistically large estimates of
WTP. This may occur for strategic reasons (although little evidence of such strategic
response has been found), due to inaccuracy in response combined with lack of
consideration to budget constraints, or due to other reasons. Responses were selected for
detailed review if either of the following conditions was met:
The adjusted (based on Question 17) WTP estimate was greater than one
percent of income for any one of the three WTP responses.
All three WTP responses were the maximum listed on the checklist (greater than
$750, which was coded as $1000).
RCG/Hagler, BaiHy, Inc.
-------
4-39
A total of 51 respondents were identified as meeting at least one of these two criterion.
Forty-two met the first criterion, and fifteen met the second (six met both). Figure 4.5-2
shows the distribution of the adjusted WTP responses to the first WTP question as a
percentage of reported income (truncating observations between 1 percent and 2.7 percent
for presentation purposes). Approximately 95 percent of the WTP responses were less than
or equal to 0.5 percent of annual income, two-thirds were less than .25 percent of annual
income, and all the responses were under 3 percent. Responses to the second and third
WTP questions showed similar distributions.
The evaluation of the identified high responses relied primarily on a review of the
comments offered by the respondent. Several types of comments were classified as protest
comments, reflecting an objection to the premises of the WTP questions, or as support
comments, indicating that the respondent thinks that protecting visibility at national parks
is important. Protest comments included:
The questions are too narrow or vague. Anti-developer comments. Anti-industry or
anti-polluter comments. The government needs to do better.
Support comments included:
Visibility at national parks is important. Protecting national parks is important.
Everyone should help and that would pay for it. Environmental protection is
important. I am willing to spend even more to improve air quality everywhere. If
we don't clean up now, we will suffer later. Concern about pollution in general.
Can't afford more (combined with a high response).
Four of the identified high responses were associated with a protest comment and no
support comment. We interpreted these four responses as rejections of the WTP questions
and recoded them to missing values. The remaining high responses each met at least two
of the following conditions and were retained as valid WTP responses. These conditions
were:
One or more support comments were written by the respondent
Respondent had visited national parks in the selected region
The mean of the three adjusted WTP estimates was less than or equal to one
percent of the reported income
No protest comments were given
Additional analysis, reported in Table 4.5-4, illustrates that, overall, the high WTP responses
apparently reflect attitudes, behavior and respondent characteristics consistent with a priori
expectations. In Table 4.5-4, four data columns are listed. The first presents statistics for
RCG/Hagler, Bailly, Inc.
-------
Figure 4.5-2
Adjusted WTP1 as a
Percent of Household Income
Number of Responses
400
300
200
100
0
0.1 0.2
0.3 0.4 0.5 0.6 0.7
Percent of Income
0.8 0.9
Note: Figure Includes non-zero WTP1
responses, adjusted for Question 17
responses, for which Income was given.
16 additional respondents (1%)
gave values between 1%
and 2.7% of Income.
-------
4-41
the 47 high responders identified by the two criteria above after deleting four apparent
protests. The second column presents statistics for only the 14 respondents who answered
" > $750" for each of the three WTP questions. The fourth column gives statistics for all
other respondents. The analysis indicates high responders assign statistically significantly
higher importance to visibility protection, to willingness to pay for that protection, and to
its effect on enjoyment; are more likely to have visited or to plan to visit; and, for the
absolute high responders (column 2), have substantially larger incomes.
Some analysts have suggested one approach to remove protest zero, and high bids, and to
reduce the effect of inaccuracy on the analysis is to arbitrarily trim a fixed percent off each
end of the bid distribution (alpha trimming, see Mitchell and Carson, 1989). Our analysis
finds this procedure would apparently incorrectly remove valid small and large value
statements.
The distributions of the adjusted WTP responses are shown in Table 4.5-5 for the results
from survey Versions 1, 2 and 3. The adjusted responses reflect the evaluations of zero and
high responses and the responses to Question 17. The net effect of these three adjustments
is an overall decrease in the WTP estimates, with the median values being decreased an
average of 33 percent, as compared to the raw data. The primary source of change is due
to the Q17 adjustment. These adjusted WTP estimates are used throughout the remainder
of the analysis, unless otherwise noted.
Self Evaluated Accuracy of WTP Responses
After answering the WTP questions, respondents were asked in Question 16 to give some
qualitative judgments about how accurate they felt their WTP answers were. The responses
to this question are shown in Table 4.5-6.7 About 80 percent of the respondents felt their
answers were "very accurate" or "within the ballpark". Just under 20 percent said that their
responses were somewhat or very inaccurate. We find these responses reassuring in that
the vast majority of respondents appear to indicate their responses give a meaningful value
signal.
The responses to this accuracy self evaluation were used to evaluate two concerns that have
been raised about contingent valuation results: (1) that difficulty, or uncertainty, in
answering the questions is related to overstatement of WTP, and (2) that unfamiliarity with
the good contributes to inaccuracy and upward bias. Although this question allows only a
crude evaluation of these concerns, the results do not support either of these contentions.
The second column in Table 4.5-6 shows the mean (unadjusted) WTP responses associated
with each accuracy response. Contrary to the first concern, WTP declines in statistically
significant increments as self-perceived accuracy declines. A similar pattern holds with
the adjusted WTP responses, shown in the third column. This pattern of decreasing mean
' Responses to Version 4 of the questionnaire are not included in this table because the structure of the
WTP questions was different. The results for Version 4 with respect to the accuracy question were similar.
RCG/Hagler, BaiHy, Inc.
-------
Table 4.5-5
Frequency Distribution of Adjusted* WTP Responses For Versions 1, 2, and 3
Version 1 (California Parks) Version 2 (Southeast Parks) Version 3 (Southwest Parks)
WTP1 WTP2 WTP3 WTP1 WTP 2 WTP3 WPT1 WPT2 WPT3
10th Percentile
$0.50
$0.60
$0.50
$0.50
$0.80
$0.60
$0.40
$0.30
$0.60
20th Percentile
$2.50
$2.50
$2.40
$2.00
$2.00
$2.00
$2.00
$2.50
$2.50
30th Percentile
$5.00
$6.00
$5.00
$5.00
$5.00
$5.00
$5.00
$5.00
$5.00
40th Percentile
$10.00
$12.00
$10.00
$10.00
$10.00
$10.00
$10.00
$10.00
$10.00
50th Percentile
$17.50
$22.50
$20.00
$12.50
$17.50
$15.00
$14.00
$20.00
$15.00
60th Percentile
$25.00
$30.00
$25.00
$25.00
$25.00
$25.00
$25.00
$30.00
$25.00
70th Percentile
$40.00
$50.00
$50.00
$37.50
$50.00
$40.00
$30.00
$50.00
$40.00
80th Percentile
$70.00
$90.00
$75.00
$50.00
$75.00
$62.50
$50.00
$75.00
$50.00
90th Percentile
$100.00
$135.00
$125.00
$100.00
$126.00
$100.00
$100.00
$125.00
$120.00
95th Percentile
$200.00
$200.00
$200.00
$150.00
$225.00
$200.00
$175.00
$250.00
$250.00
99th Percentile**
$400.00
$525.00
$400.00
$360.00
$1,000.00
$600.00
$500.00
$500.00
$500.00
Total Number of Resp'
onses 331
334
330
346
350
344
337
338
332
* Adjusted removing protest zeros and questionable high WTP responses, and corrected for the percentage of WTP solely for visibility from
Question 17.
** The maximum response was "MORE THAN $750." For analysis, this was coded as $1000.
-------
Table 4.5-6
Perceived Accuracy of WTP Responses
Accuracy
Responses*
Mean Raw WTP1 Mean Adjusted
Response WTP1 Response
(SE of Mean) (SE of Mean)
Responses for
those who
gave accepted
zeros to WTP1
Responses
for those who have
ever visited a
national park
shown on the map
Responses
for those who have
ever visited the park
shown in the
photographs**
Very
14.7%
$90.29
$60.97
63%
13.4%
14.5%
Accurate
(n=192)
(15.61)
(11.46)
(n=63)
(n=141)
(n=78)
(n=190)
(n=190)
Within the
66.3%
$67.43
$43.27
23%
69.2%
68.3%
Ballpark
(n=864)
( 3.72)
(2.40)
(n=23)
(n= 729)
(n=367)
(n=858)
(n=858)
Somewhat
14.1%
$53.01
$32.33
7%
13.0%
12.1%
Inaccurate
(n=184)
( 7.34)
(4.14)
(n=7)
(n=137)
(n=65)
(n=179)
(n=179)
Probably
4.8%
$40.15
$26.62
7%
4.5%
5.0%
Very Inaccurate
(n=63)
( 6.96)
(6.23)
(n=7)
(n=47)
(n= 27)
(n=62)
(n=62)
Total #
Respondents 1303 1289 1289 100 1054 537
These are the accuracy responses for respondents to all versions, except version 4, who gave a dollar response (including accepted zeros) to any of
the WTP questions.
This is the number of respondents who said they had ever visited the specific park that was the focus of the WTP questions they were asked. This
is a subset of those who have ever visited any national park.
-------
4-44
WTP with decreasing self-perceived accuracy is accentuated when zero WTP values are
removed, because a significant share of respondents who gave zero WTP said that their
responses were "very accurate." The accuracy responses for individuals who gave accepted
zero responses to the first WTP question are shown in the fourth column in Table 4.5-6.
The fifth and sixth columns show the accuracy responses for respondents with some
visitation experience at the national parks. The distribution of accuracy responses is
essentially the same for these visitor segments of the sample. This suggests that previous
visitation is not related to the perceived accuracy of the WTP responses, contrary to
expectations.
An examination of potential relationships between self-perceived accuracy responses and
responses to other questions revealed two statistically significant relationships, independent
of the relationship with WTP already discussed. Self-reported accuracy increased as the
respondent's education level increased, and a higher percentage of WTP specifically for
visibility (Q-17) was associated with a higher self-reported accuracy for the WTP responses.
Combined with results of Question 17, the Question 16 results indicate that individuals with
low self-perceived accuracy tended to give lower values for the visibility change and were
more likely to give responses that reflected part-whole bias. This has important
implications:
The less well understood the resource change is for the respondent, the more
uncertain and inaccurate responses are likely to be and values will decrease.
Lowered values with lowered accuracy may reflect a tendency to provide a value
lower than the maximum WTP to reflect uncertainty about the valuation (i.e., a
hedging strategy that downward biases WTP response and similarly may upward
bias WTA responses -- see Schulze et al. 1990 forthcoming). This implies that
visibility studies with poor stimuli may result in reduced WTP values.
Forcing respondents to provide answers to questions about which they are
uncertain, which Fishhoff and Furby (1988, page 169) argue against doing, may lead
to downward bias in the sample-wide mean WTP estimates.
To be conservative, responses with low self-reported accuracy are retained in subsequent
analyses, although this may introduce downward biases into the estimates. To examine this
potential impact on the overall results, selected sample means were calculated weighting
the observation by the inverse of the self-perceived evaluation score. These weighted
means were only slightly larger, on the order of five percent, than the unweighted means
presented in the report. Due to the small amount of difference, these weighted means are
not presented.
WTP Responses All Equal
In general, one might expect that WTP responses would be different for the different
changes in visibility conditions. WTP2 for an improvement in average conditions from the
RCG/Hagler, BaiHy, Inc.
-------
4-45
current 50th percentile to the current 90th percentile would be expected to equal or exceed
WTP1 for an improvement from the 50th to the 75th percentile. Based upon the responses
to Question 11 (Table 4.4-11) and prospect theory (Kahneman and Tversky 1979 and 1982)
one might also expect WTP3 to prevent a degradation to the 10th percent conditions to
exceed the WTP1 but not to be as large as WTP2. This is because the percent change in
visual range in the WTP3 scenarios is generally equal to or larger than in the WTP1
scenarios, but less than in the WTP2 scenarios (Table 3.3-2), and because preventing
degradations is ranked as slightly more important than obtaining improvements.8
The overall mean WTP responses are consistent with these expectations. But, looking at
the data on an individual observation basis, it is found that a significant portion of the
respondents gave the same non-zero bid for all three WTP scenarios.
Table 4.5-7 shows the number and percentage of respondents in each of three categories
of WTP responses: (1) accepted zero WTP for all three, (2) the same non-zero dollar
amount for all three, and (3) a different dollar amount to one or more of the three. The
numbers in Table 4.5-7 are based on the adjusted WTP responses (but for any individual
the same adjustment percentage was used for all three WTP responses). Results for each
national park region are reported. Version 4 is treated separately because the three WTP
questions in this version are for the same percentile change in three separate regions, not
for three different levels of visibility in the same region. The three WTP responses for
Version 4 being all the same raises different issues than for the other versions of the
questionnaire.
Table 4.5-7 shows that the percentage of respondents who gave the same non-zero dollar
response to all three WTP questions is almost identical for the three regions, ranging from
40 to 42 percent. The percentage of all-equal non-zero WTP responses for Version 4,
which may occur for different' reasons than in the other survey versions, is only slightly
higher at 45 percent.
The high percentage of all-equal non-zero WTP responses raises some questions about how
closely the responses are tied to the specific changes in visibility presented for each
scenario. It may be the case that the changes in visibility are perceived as being quite
similar. I.e, the WTP1 and WTP2 visibility measures (Table 3.2-1) are nearly identical for
the Grand Canyon, and the change in these measures is also nearly equal for the WTP1 and
WTP3 scenarios. In other cases, very flat indifference curves for visibility improvements
beyond some level would also result in WTP1 approximately equal WTP2. In cases such
as these, respondents may simply state the same value, rather than make small adjustments
in the value. While these arguments may hold in some cases, the differences in visibility
conditions in some scenarios for some locations, combined with the high (and similar)
percent of all equal response for each survey version, are sufficient to suggest this alone
does not adequately explain the "all-equal WTP" responses. Sanghvi et al. (1989) also
reported an unexpectedly high percentage of equal WTP responses for different changes in
8 If responses reflect the weighted average atmospheric transmission measure, similar implications would
result. If respondents are focusing upon the change in the percentile, then one would expect WTP3 = WTP2
> WTP1. This is because the WTP1 and WTP3 scenarios both have a change of 40 percentile points, while
WTP1 has a change of 25 percentile points.
RCG/Hagler, BaiHy, Inc.
-------
4-46
Table 4.5-7
All Equal Versus Different WTP Responses
California Parks Southwest Parks Southeast Parks Multiple Regions
(Version 1) (Version 3, 5, 6) (Version 2) (Version 4)
number percent number percent number percent number percent
All Zero (Accepted)
17
5%
47
7%
23
7%
7
5%
Non-Zero All Equal
138
41%
259
40%
149
42%
61
45%
Different
181
54%
337
52%
180
51%
67
50%
TOTAL*
336
643
352
135
* Net of missing and identified protests.
RCG/Hagler, BaiHy, Inc.
-------
4-47
the frequency of power outages, and we suspect that this is a common occurrence in CVM
studies that has only recently begun to be reported. Explanations for this phenomenon will
remain speculative until further research is conducted to specifically examine this issue.
We have two hypotheses regarding these all-equal responses and have examined the
information available from this study to see if either of these hypotheses is supported.
These hypotheses are:
1. Respondents may be anchoring on the first scenario and failing to expend the
effort necessary to adjust the response to changes in the subsequent scenarios.
2. Respondents may be expressing a value for the good or service in question that
is only loosely tied to the exact change described in the scenario.
The evidence available in this study regarding these hypotheses is mixed and inconclusive.
Both hypotheses are supported to some extent, and they are not necessarily mutually
exclusive.
Table 4.5-8 shows the mean WTP responses for the all-equal and the different WTP groups,
combined across Versions 1, 2, 3, 5, and 6. The raw and adjusted responses to the first
WTP question are not statistically significantly different for the two groups, while for the
second and third WTP questions the mean responses of the all-equal group are statistically
significantly lower. This suggests that there is an anchoring of the first WTP bid on the
first visibility control scenario rather than just being a contribution unrelated to visibility
changes. The anchoring hypothesis is also consistent with the tendency of responses to
select relatively round numbers such as $25 or $100 (as discussed earlier in this section),
having decided upon such a number for the first WTP question, some respondents may not
be inclined to put the effort into fine tuning their responses for what they may see as
relatively small changes in the scenario, or if there is a sense that the accuracy of the
response is less than the magnitude of the refinement.
Table 4.5-8 shows that the all-equal WTP group gave an average percentage specifically for
visibility in Question 17 that is statistically significantly lower for the different WTP group.
This suggests that the raw responses for this group are less closely tied to the specific
visibility changes presented, which is consistent with the second hypothesis. The adjusted
responses account for this difference to the extent that respondents are able to estimate
what percentage really is for the specific changes in visibility.
The results of a logit analysis further examining the differences between the all-equal WTP
group and the different WTP group are shown in Table 4.5-9. Age and education are
statistically significantly related to the likelihood of giving all-equal WTP responses, while
the gender of the respondent is not. The signs of the age and education coefficients are
consistent with the anchoring and less effort hypothesis. Older respondents are more likely
to give all-equal responses and those with more education are less likely to give all-equal
responses.
RCG/Hagler, BaiHy, Inc.
-------
4-48
Table 4.5-8
WTP Means for All Equal
Versus Different Responses
Mean
Mean
Raw WTP Responses
Adjusted WTP Responses
Non-Zero
Non-Zero
All Equal Different
All Equal Different
All Versions
Except 4
WTP1
$73
(6.82)
n=556
$69
(3.92)
n=691
$45
(4.59)
n=546
$48
(3.04)
n=680
WTP2
$73
(6.82)
n=556
$106**
(6.10)
n=700
$45
(4.59)
n=546
$74'
(4.83)
n=689
WTP3
$73
(6.82)
n=556
$88*
(5.46)
n=687
$45
(4.59)
n=546
$62'
(4.28)
n=676
Percent for
Visibility
58%
(1.4)
n=532
65%*
(1.1)
n666
Statistically signicantly different versus non-zero all equal means at 90% confidence.
Statistically different versus non-zero all equal means at 95% confidence.
RCG/Hagler, BaiHy, Inc.
-------
4-49
Table 4.5-9
Logit Analysis on Whether Respondent Gave the Same WTP
Independent Variable Model 1 Model 2
One 1.27** 0.60
(3.03) (1.25)
Male -0.041 -0.060
(-0.31) (-0.45)
Importance (Q-ll)
Age 0.013** 0.013
(2.90) (2.89)
Education -0.14** -0.15**
(-3.9 1) (-4.40)
Allpay Comment 0.62** 0.58**
(3.03) (2.87)
Future Visit (Q-4) -0.10* -0.13
(-1.86) (-2.44)
Percent for Visibility (Q-17) -0.005** -0.005'1
(-2.33) (-2.37)
Support WTP (Q-10) -0.15**
(-2.32)
0.094
(1.24)
N 1059 1056
% Correctly Predicted 60.9 60.7
P < 10
p < .05
NOTE: Dependent variable = 1 if respondent gave same dollar response to all three WTP
question, 0 otherwise. Sample excludes Version 4 and excludes respondents who
gave zero or no response to all three WTP questions, t-ratios in parentheses.
RCG/Hagler, BaiHy, Inc
-------
4-50
Model 2 results show that responses to Question 11 asking respondents to rate the
importance of obtaining improvements in visibility at national parks in the specific region
of focus in their questionnaire are not related to the likelihood of giving all-equal WTP
responses. However, Model 1 shows that respondents who say they are less willing to pay
(qualitatively) for protection of visibility at national parks are more likely to give all-equal
WTP responses. Combined with Model 2 results, this suggests that respondents giving all-
equal WTP responses do not necessarily care less about visibility at national parks, but may
have more trouble with the idea of being asked to pay for such protection. This is also
consistent with the anchoring hypothesis in that these respondents appear to be having more
trouble with the WTP questions.
The results of the logit analysis also indicate that respondents who say they are less likely
to visit national parks in the specific region of the focus in the subsequent WTP questions
are more likely to give all-equal WTP responses. This is consistent with the possibility that
these respondents are less concerned about the exact visibility changes and therefore tend
to see the different scenarios as roughly equivalent.
Some respondents volunteered a comment along the lines of "if everyone paid some amount
then we could take care of this problem." This type of comment suggests a contribution
type of attitude toward the WTP questions, which might be expected to be associated with
less concern about the specific scenarios. The logit analysis results are consistent with this
possibility, showing that respondents giving this type of comment (ALLPAY=1) are more
likely to have given the same WTP response to all three questions.
It is important to note that the effect of the all-equal WTP responses in this study is to
lower the estimated coefficient for changes in visibility, because the first scenario generally
presents the smallest change in visibility and the respondents giving all-equal responses
appear to be anchoring on the first WTP question. The analysis presented here suggests
that the available evidence points to several possible explanations for this phenomenon, but
this needs to be further examined in future research designed to specifically address this
issue.
4.5.2 Means. Medians, and Simple Correlation of Willingness to Pav Responses
This section presents the WTP means and medians by state and by park region. Simple
correlations between the WTP responses and responses to other questions are also
discussed.
Means of Willingness to Pav Responses
Mean willingness to pay responses by questionnaire version for respondents in the sample
from each state are given in Table 4.5-10. These are means of the adjusted responses, as
discussed in the previous section. All of the willingness to pay figures are annual values for
the household.
RCG/Hagler, BaiHy, Inc.
-------
4-51
For all questionnaire versions except Version 4, WTP1 is willingness to pay for an
improvement in average conditions from the 50th to the 75th percentile. WTP2 is for an
improvement in average conditions from the 50th to the 90th percentile. WTP3 is to
prevent a degradation in average conditions from the 50th to the 10th percentile. For
Version 4: WTP1 is for an improvement in average conditions from the 50th to the 75th
percentile at national parks in California; WTP2 is for an improvement in average
conditions from the 50th to the 75th percentile at national parks in the Southwest; and
WTP3 is for an improvement in average conditions from the 50th to the 75th percentile at
national parks in the Southeast.
Several patterns emerge from the results in Table 4.5-10. For all questionnaire versions
except Version 4, WTP2 is greater than WTP1. For the most part, WTP3 falls between
WTP1 and WTP2. In one case, WTP3 exceeds WTP2. Responses for residents of the state
located in the region of interest are always the highest or second highest state means for
the states sampled for each questionnaire version.
The home-state effect is present in the Version 4 results, but appears to be smaller than
for the other versions. The order of the means by state of residence is the same for all
three park regions: Arizona residents give the lowest WTP and New York residents give
the highest WTP. However, Arizona residents have a higher mean WTP for southwestern
parks than for the other two regions and California residents have a higher mean WTP for
California parks than for the other two regions, although these differences are not
statistically significant.
Comparisons of the mean WTP responses for the different states may be confounded by
differences in the average characteristics of the residents of each of the sampled states.
Table 4.5-11 shows selected characteristics of the sample by state. The most notable
difference between the states is in average household income. The lower average income
combined with the distance from the selected park regions may explain why the mean WTP
responses for Missouri residents were among the lowest for all three regions. In Table
4.5-10, whenever the home state WTP mean is exceeded by the mean for another state, the
other state also has a higher average income. For example, the mean WTP1 for Arizona
is exceeded by the mean WTP1 for Virginia. The Virginia respondents have an average
household income of $45,000, while the Arizona respondents have an average household
income of $36,000. It is interesting that there do not appear to be significant differences
across the states in terms of general perceived importance of preserving natural resources
at national parks. The relationships between various characteristics of the respondents and
the WTP responses are further explored in subsequent sections of this chapter.
Responses for different questionnaire versions, including those for different regions of
interest, do not appear to be statistically significantly different. In particular, Versions 3,
5, and 6 obtained very similar WTP responses for visibility at parks in the southwestern U.S.
Table 4.5-12 shows the WTP means by state for the southwest region, combined across all
three of these questionnaire versions. The question of the effect of different versions of the
questionnaire is explored further in section 4.5.4 below.
RCG/Hagler, BaiHy, Inc.
-------
4-52
Table 4.5-10
Means of Adjusted Willingness to Pay
Responses by Version and by State
Version 1: California Parks
State of Residence WTP1 WTP2 WTP3
Arizona
$27.54
$34.29
$3/
(5.05)
(5.70)
(5.
n=55
n=56
n = 57
Virginia
$41.94
$46.37
$49.73
(9.62)
(11.48)
(12.76)
n=59
n=60
n=57
California
$60.19
$71.92
$64.13
(10.88)
(12.25)
(11.12)
n=113
n=113
n=113
New York
$57.20
$77.86
$73.01
(14.90)
(23.36)
(23.31)
n=47
n=47
n=47
Missouri
$32.74
$40.09
$35.77
(6.68)
(7.56)
(7.21)
n=57
n=58
n=56
All States
$46.36
$56.33
$52.79
(4.85)
(5.95)
(5.75)
n=331
n=334
n=330
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
4-53
Table 4.5-10 (continued)
Means of Adjusted Willingness to Pay
Responses by Version and by State
Version 2: Southeast Parks
State of Residence WTP1 WTP2 WTP3
Arizona
$24.70
$56.21
S53.56
(5.69)
(22.80)
(22.88)
n=61
n=61
n = 61
Virginia
$59.77
$74.89
$67.60
(11.78)
(13.48)
(11.72)
n=120
n=124
n=124
California
$38.24
$54.65
$47.68
(9.93)
(13.62)
(11.43)
n=55
n=55
n=55
New York
$39.55
$53.96
S45.69
(8.20)
(10.19)
(9.90)
CD
II
CD
II
n = 45
Missouri
$25.62
$30.52
$26.40
(4.04)
(4.84)
(4.08)
ti-
ll
CD
II
n=62
All States
$41.16
$57.59
$5 1.77
(4.73)
(6.78)
(6.28)
CD
CO
II
n=350
n=347
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
4-54
Table 4.5-10 (continued)
Means of Adjusted Willingness to Pay
Responses by Version and by State
Version 3: Southwest Parks
State of Residence WTP1 WTP2 WTP3
Arizona
$45.35
$65.91
$54.87
(7.55)
(10.29)
(9.05)
n= 130
n= 130
n=130
Virginia
$66.51
$69.94
$72.96
(20.24)
(17.47)
(18.05)
n=48
n=49
n=48
California
$34.80
$53.86
$36.80
(8.56)
(14.71)
(10.35)
n=49
n=49
n=47
New York
$44.15
$57.41
$50.72
(11.76)
(15.23)
(13.15)
n=55
n=55
n=53
Missouri
$17.86
$21.32
$19.63
(2.80)
(3.29)
(3.18)
P
II
n=55
n=54
All States
$42.14
$56.11
$48.53
(4.75)
(5.79)
(5.17)
n=337
n=338
n=332
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
Table 4.5-10 (continued)
Means of Adjusted Willingness to Pay
Responses by Version and by State
Version 4: Three Region
State of Residence WTP1 WTP2 WTP3
(California) (Southwest) (Southeast)
Arizona
$28.96
$35.36
$26.55
(6.81)
(8.08)
(6.90)
n=58
n=58
n=54
California
$40.50
$37.93
$33.70
(8.47)
(7.59)
(7.92)
CO
II
CO
CO
II
n=31
New York
$58.59
$56.38
$60.88
(24.23)
(23.53)
(24.36)
n=42
CO
II
CO
II
All States
$41.17
$42.74
$39.81
(8.43)
(8.50)
(8.93)
n= 134
n= 134
n= 128
Standard error of mean in parentheses.
RCG/Hagler, Bailly, Inc.
-------
4-56
Table 4.5-10 (continued)
Means of Adjusted Willingness to Pay
Responses by Version and by State
Version 5: Limited Information (SW)
State of Residence WTP1 WTP2 WTP3
Arizona
$63.22
$73.28
$65.50
(18.24)
(18.88)
(18.35)
n=56
n=56
n=57
Virginia
$39.12
$42.47
$37.79
(16.57)
(16.37)
(16.02)
n=47
n=48
n=49
New York
$26.55
$39.10
$30.79
(5.79)
(9.90)
(6.89)
n=46
n=46
n=46
All States
$44.30
$52.94
$46.06
(8.84)
(9.34)
(8.89)
n=149
n=150
n=152
Standard error
of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
4-57
Table 4.5-10 (continued)
Means of Adjusted Willingness to Pay
Responses by Version and by State
Version 6: One Region (SW)
State of Residence WTP1 WTP2 WTP3
Arizona
$51.14
$66.94
$52.72
(13.94)
(16.19)
(15.93)
n=55
n=54
n=54
California
$55.07
$85.01
$70.22
(13.54)
(24.78)
(19.91)
n=51
n=52
n=51
New York
$31.32
$37.20
$29.67
(7.84)
(8.53)
(7.62)
n=44
n=44
n=43
All States
$46.66
$64.48
$52.05
(7.25)
(10.72)
(9.30)
n=150
n=150
n=148
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
4-58
Table 4.5-11
Average Sample Characteristics by State
Characteristic Arizona
(n=446)
Virginia
(n=305)
California
(n=333)
New York
(n=306)
Missouri
(n=188)
1987
Household Income $36,000
$45,000
$44,000
$47,000
$36,000
% Male Respondent
59%
66%
61%
57%
63%
Age of Respondent
48
46
46
47
46
% Have Visited a
National Park
94%
86%
88%
67%
75%
Number of Children
in Household
.7
.7
.6
.7
.7
Importance of Natural
Conditions at National
Parks (Q-5a)
4.0
3.9
4.0
3.9
3.9
Importance of
Preservation
for Its Own
Sake (Q-7c)
4.1
4.2
4.2
4.3
4.2
RCG/Hagler, BaiHy, Inc.
-------
4-59
Table 4.5-12
Combined Means of Adjusted Willingness to Pay
for the Southwest Parks by State
(Versions 3, 5, and 6)
State of Residence
WTP1
WTP2
WTP3
Arizona
$50.83
(6.67)
n=241
$67.86
(7.95)
n=240
$56.90
(7.42)
n=241
Virginia
$52.96
(13.12)
n=95
$56.35
(12.00)
n=97
$55.19
(12.12)
n=97
California
$45.14
(8.11)
n=100
$69.89
(14.63)
n=101
S54.19
(11.56)
n = 98
New York
$34.67
(5.38)
n=145
$45.47
(7.06)
n=145
$37.89
(5.89)
n=142
Missouri
$17.86
(2.80)
n=55
$21.32
(3.29)
n=55
$19.63
(3.18)
n=54
All States
$43.72
(3.67)
n=636
$57.33
(4.53)
n=638
$48.76
(4.08)
n=632
Standard error of mean in parentheses.
RCG/Hagler, Bailly, Inc.
-------
4-60
Table 4.5-13 shows mean WTP responses for WTP1 for each of the three regions, split
according future visitation expectations. Visitation and residence location were related, as
was shown in Section 4.4, and visitation may be the underlying cause of the home-state
effect seen in the responses. Those who responded to Question 4 saying that they probably
or definitely would visit a national park in the region of interest in the next five years gave
WTP responses about twice those given by respondents who indicated less likelihood of
visiting. For all three regions, these means are statistically significantly different. A similar
pattern is seen when the sample is split according to previous visitation, although the
difference in the means is greatest when the sample is split according the future visitation
expectations.
Medians of Willingness to Pav Responses
Table 4.5-14 shows the medians of the adjusted WTP responses for each park region by
state. Fifty percent of the sample gave responses this high or higher, and fifty percent gave
responses this low or lower. The Version 4 medians are reported separately. The
responses for the southwest region are combined from Versions 3, 5, and 6. The home-
state effect is reflected in the medians as it was in the means. In general, the medians are
about one-third to one-half the magnitude of the means, due to the effect of the skewed
distribution of responses. This pattern is typical of all WTP estimates because they are
necessarily truncated at zero and there are typically a few relatively high responses.
Simnle Correlations with Willingness to Pav Resnonses
One of the questions that this study was designed to address is whether WTP responses for
protection of visibility at national parks can be substantiated by responses to non-
quantitative questions about the general importance of the visibility and natural resource
protection for national parks. Even though questions will remain about the accuracy of the
dollar responses to the WTP questions, it increases our confidence that the response are
meaningful if they are at least correlated with responses to other questions concerning the
importance of protecting the visibility at national parks.
Simple correlations were estimated between the responses to Questions 5 through 11 and
the response to the first WTP question (Question 12) for all versions of the questionnaire.
Table 4.5-15 lists the questions the responses to which were significantly positively
correlated with the WTP response. The group showing the strongest correlations (p < .001)
had correlation coefficients typically in the range of .10 to .30. The second group (.001 <
p < .01) had correlation coefficients typically in the range of .05 to .15.
These correlations are consistent with expectations. For Question 5 concerning reasons for
visiting national parks, the items most closely related to WTP for visibility are those
concerning enjoyment of the natural environment. The importance ratings for all of the
items listed for Question 7, reasons for protecting national parks even if you yourself never
visit, are related to the WTP response. The Question 8 response concerning the priority
rating for visibility was significantly related to the WTP given for visibility.
RCG/Hagler, BaiHy, Inc.
-------
4-61
Table 4.5-13
WTP1 Means by Likelihood of Future Visit
Probably or
Definitely Less Likely
Will Visit to Visit
(Q-4 = 4 or 5) (Q-4 = I 2, or 3)
California Parks $66.4 $27.9
(version 1) (9 1) (2.9)
n=191 n=240
Southwest Parks $58.3 $31.2
(version 3, 5, 6) (6 1) (3.6)
n=357 n=360
Southeast Parks $67.1 $32.6
(version 2) (12.5) (3.9)
n=126 n=298
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
4-62
Table 4.5-14
Medians of Adjusted Willingness to Pay
by Region of Focus and by State of Residence
California Parks (Version 1)
State of Residence
WTP1
WTP2
WTP3
Arizona
$12.50
$20.00
$15.00
Virginia
$10.00
$12.50
$12.50
California
$25.00
$35.00
$30.00
New York
$15.00
$20.00
$17.50
Missouri
$10.00
$15.00
$15.00
All States
$17.50
$22.50
$20.00
Southwest
Parks (Versions 3,
5, and 6)
State of Residence
WTP1
WTP2
WTP3
Arizona
$24.00
$30.00
$25.00
Virginia
$10.00
$10.00
$10.00
California
$14.00
$20.00
$15.00
New York
$10.00
$15.00
$10.00
Missouri
$10.00
$12.00
$10.00
All States
$15.00
$20.00
$15.00
RCG/Hagler, Bailly, Inc.
-------
4-63
Table 4.5-14 (continued)
Medians of Adjusted Willingness to Pay
by Region of Focus and by State of Residence
Southeast Parks (Version 2)
State of Residence
WTP1
WTP2
WTP3
Arizona
$10.00
$10.50
$7.50
Virginia
$20.00
$25.00
$20.00
California
$10.00
$15.00
$15.00
New York
$15.00
$30.00
$25.00
Missouri
$10.00
$10.00
$12.00
All States
$12.50
$17.50
$15.00
Version 4
State of Residence
WTP1
WTP2
WTP3
California
Southwest
Southeast
Parks
Parks
Parks
Arizona
$5.00
$10.00
$4.80
California
$25.00
$25.00
$20.00
New York
$20.00
$20.00
$20.00
RCG/Hagler, BaiHy, Inc.
-------
4-64
Table 4.5-15
Significant Positive Correlations with
Willingness to Pay for Visibility Protection
p < .001
Question 5: Importance of following reasons to visit national parks:
To experience unique natural places
To enjoy the vastness of nature
Question 7: Importance of following reasons for protecting national parks even if you
never visit:
So other members of my family can visit
So other people outside of my family can visit
So there will be areas in natural condition even if no one visits
Scientific research opportunities
Preserve national heritage
So there is not development every where
Question 8: Priority of preventing following potential effects on national parks:
Air pollution decreasing ability to see scenic vistas
Question 10: Willingness to pay higher prices and taxes to support visibility protection in
selected region
Question 11: Importance of improving or preventing degradation in visibility at parks in
selected region
RCG/Hagler, BaiHy, Inc.
-------
4-65
Table 4.5-15 (continued)
Significant Positive Correlations with
Willingness to Pay for Visibility Protection
.001 < p < .01
Question 5: Importance of following reasons to visit national parks:
Outdoor recreation opportunities
Change from usual surroundings
Question 8: Priority of preventing the following potential effects on national parks:
Air pollution injury to vegetation
Water pollution - aesthetic effects only
Hearing or seeing mining/industrial activities outside boundaries
Question 9: Effect of an improvement in visibility on enjoyment of a visit to selected park
RCG/Hagler, BaiHy, Inc.
-------
4-66
Questions 10 and 11 were even more closely tied to the WTP questions and concerned
willingness to pay for visibility protection at parks in the selected regions in general (rather
than a specific dollar amount for a specific level of protection) and the importance of
improving or preventing degradation in visibility at parks in the selected region. The
response to Question 9 concerning the perceived effect of an improvement in visibility on
a potential visit to the selected park was also significantly correlated with the WTP
response.
These types of analyses provide an overall consistency check on the data quality and reflect
that, on the whole, the WTP responses are meaningfully related to underlying attitudes and
behaviors.
4.5.3 Variations in WTP for Different Regions
The means of the adjusted WTP responses for the different regions were reported in Tables
4.5-10 and 4.5-12. Comparing overall means for the basic questionnaire Versions 1, 2, and
3 suggests that the WTP responses for the different regions are nearly identical. Regression
analysis was used to further examine this question.
Exploratory regression results for the adjusted WTP responses from Versions 1, 2, and 3
are shown in Tables 4.5-16 and 4.5-17. The purpose of these regressions is to determine
if there is a statistically significant difference in the WTP responses for the different regions
after controlling for other factors that might influence the responses, such as income or park
visitation. The levels of visual range and the features of the parks, as illustrated in the
selected photographs, were different for each region, and it was expected that this might
cause differences in the WTP responses. The WTP1 equations are for obtaining
improvements in average conditions from the current 50th percentile to the current 75th
percentile. The WTP2 equation is for obtaining an improvement in average conditions
from the current 50th percentile to the current 90th percentile. The WTP3 equation is for
preventing degradation in average conditions from the current 50th percentile to the current
10th percentile. Dummy variables (0,1) were included if the focus of the WTP questions
was the Southeast parks or the California parks.
The results shown in Table 4.5-16 confirm what was seen in the WTP means, that there is
no statistically significant difference in responses for the different park regions. The
coefficients for the Southeast and California variables are small and not statistically
significant in any of the equations. Explanatory variables that are statistically significant
and fairly stable across the different WTP questions are household income and the
respondent residing in the state in which the illustrated park is located. The probability of
a future visit to a park in the region is also statistically significant as shown in Equation 1
for the first WTP question. Residence in the park state and probability of future visitation
are highly correlated and were not therefore included in the same equation.
RCG/Hagler, BaiHy, Inc.
-------
4-67
Table 4.5-16
Regression Results for Adjusted WTP,
Versions 1, 2, and 3
Variable
WTP1
Equation 1
Coefficient
(t-statistic)
WTP1
Equation 2
Coefficient
(t-statistic)
WTP2
Equation
Coefficient
(t-statistic)
WTP3
Equation
Coefficient
(t-statistic)
Intercept
Income
Age
Male
F-Visit
P-State
Southeast
California
R
n
F
1.85
(0.12)
0.73 **
(5.81)
-0.07
(-0.33)
-6.95
(-1.06)
5.67 **
(2.17)
-0.85
(-0.11)
3.06
(0.40)
15.54
(1.19)
0.77 **
(6.19)
-0.14
(-0.69)
-6.24
(-0.95)
.05
831
7.69 **
16.98 :
(2.65)
-2.36
(-0.31)
2.95
(0.39)
.06
831
8.10 **
26.89
(1.58)
1.01 **
(6.28)
-0.31
(-1.16)
-7.71
(-0.90)
20.40 **
(2.44)
1.84
(0.18)
-1.00
(-0.10)
.06
836
8.39 **
26.93 *
(1.69)
0.82 **
(5.42)
-0.32
(-1.24)
-5.27
(-0.65)
15.08 *
(1.92)
3.64
(0.39)
5.07
(0.54)
.04
828
6.14 **
P < 10
p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-68
Table 4.5-16 (continued)
Regression Results for Adjusted WTP1,
Versions 1, 2, and 3
Dependent, variable definitions:
WTP1 - Annual household WTP to obtain an improvement in average conditions from
the current 50th percentile to the current 75th percentile.
WTP2 - Annual household WTP to obtain an improvement in average conditions from
the current 50th percentile to the current 90th percentile.
WTP3 - Annual household WTP to prevent a degradation in average conditions from
the current 50th percentile to the current 10th percentile.
Indenendent, variable definitions:
Income - Annual 1987 household income before taxes in thousands (mid-point of the
range).
Age - In years.
Male - Equals 1 if respondent is male.
F-Visit - Probability of a future visit to a park in the region of focus in the next five
years. A five-point scale: 1 = definitely will not visit and 5 = definitely will
visit.
P-State - Park illustrated is in the state in which respondent resides.
Southeast - Equals 1 if the region of focus is the Southeast.
California - Equals 1 if the region of focus is California.
RCG/Hagler, BaiHy, Inc.
-------
4-69
Table 4.5-17
Regression Coefficients for Different
Regions for WTP1
California Southwest Southeast
Equation 1 Equation 2 Equation 1 Equation 2 Equation 1 Equation 2
Income 0.82 ** 0.64 ** 1.04 ** 0.88 ** 0.72 ** 0.65 1
(4.56) (3.12) (6.99) (5.40) (3.60) (3.07)
Age 0.14 -0.06 -0.11 -0.23 0.11 -0.03
(0.69) (-0.27) (-0.63) (-1.21) (0.47) (-0.13)
Male -8.00 -9.09 -2.06 -5.48 -6.42 -8.82
(-0.71) (-0.80) (-0.24) (-0.63) (-0.57) (-0.78)
F-Visit 8.95 ** 6.87 ** 6.64 *
(2.48) (2.56) (1.81)
P-State 25.40 ** 19.65 ** 22.77 **
(2.30) (2.41) (2.01)
R2 .27 .27 .24 .24 .21 .21
F 24.82 * 25.10 * 41.51 * 41.74 * 18.20 * 17.96
n 274 274 534 534 275 275
p < .10
p < .05
RCG/Hagler, BaiJly, Inc.
-------
4-70
Other explanatory variables that were included in earlier estimations were education and
whether there were children in the household. Education was found to be highly correlated
with income and therefore dropped from subsequent estimations, as income showed a
stronger explanatory power. Having children in the household was not related to the WTP
responses.
Table 4.5-17 shows the estimated regression coefficients for each region for WTP1 (the
change from 50th to 75th percentile). The purpose of these regressions is to assess the
stability of the coefficients for the socio-economic variables for the different regions. The
results in Table 4.5-17 show that the statistically significant coefficients are reasonably
stable across the regions. These are income, residence in the region (P-STATE), and the
probability of visiting a park in the region in the next five years (F-VISIT).
4.5.4 Effects of Questionnaire Design Variations
Three different versions of the questionnaire were developed to test for effects of certain
variations in the instrument design. All three of these versions focused on the Southwest
and provide comparison to the baseline Version 3 for the Southwest, and to one another.
Version 5 presented a different amount of information in the introduction to the WTP
questions. Version 6 excluded photographs and information about the other two regions.
Version 4 asked a WTP question for each of the regions.
Mean and median WTP responses can be compared using the information in Tables 4.5-
10, 4.5-12 and 4.5-14. However, pairwise (across two survey versions) mean tests are
relatively weak due to changing sample characteristics across the small samples for survey
Versions 4, 5 and 6. Therefore, to be able to control for visitation, income and other
important respondent characteristics across the versions and to use the information in all
four versions (Versions 3 through 6), regression analysis was undertaken with dummy
variables for survey Versions 4, 5 and 6. Results are reported in Table 4.5-18. For Version
4, only responses to the Southwest WTP question are included in the results reported in
Table 4.5-18.
Effect of Information
Version 5 was the same as Version 3 except that the introduction to Question 12 (the first
WTP question) was shortened from two paragraphs to one paragraph. Information was
deleted stressing that the WTP questions refer to visibility only and specifying additional
assumptions about when payments begin and who would make payments. This change was
expected to cause WTP responses to increase or to change only minimally (see discussion
in Section 3.1.3).
While the mean WTP responses for Arizona residents who received Version 5 are higher
than for Arizona residents who received Version 3, the opposite is true for residents of
Virginia and New York, and only a few of the differences in means are statistically different
at very low confidence levels. The regression analysis confirms that the change in
information had no apparent consistent effect on the results.
RCG/Hagler, BaiHy, Inc.
-------
4-71
Table 4.5-18
Regression Results Highlighting Effects of Different Versions
for Southwest Parks
Variable
WTP1
Equation 1
Coefficient
(t-statistic)
WTP1
Equation 2
Coefficient
(t-statistic)
WTP1
Equation 3
Coefficient
(t-statistic)
Intercept
Income
Age
Male
F-Visit
P-State
Vers4 * P-State
Version 4
Version 5
Version 6
Vers6 * CARes
-19.68
(-1.00)
0.90 **
(5.82)
-0.09
(-0.36)
-8.27
(- 1.03)
10.33 **
(3.12)
2.97
(0.27)
6.84
(0.65)
0.04
(0.003)
11.83
(0.76)
1.00 **
(6.35)
-0.25
(-1.02)
-6.02
(-0.75)
14.66 **
(1.86)
3.50
(0.32)
6.76
(0.64)
2.65
(0.26)
13.90
(0.91)
1.00 **
(6.31)
-0.26
(-1.05)
-6.16
(-0.77)
16.28 **
(1.95)
-9.05
(-0.57)
4.70
(0.45)
0.63
(0.06)
R2
F
n
.07
7.22
645
.06
6.27
645
.06
6.31
645
RCG/Hagler, Bailly, Inc.
-------
4-72
Table 4.5-18 (cont.)
Regression Results Highlighting Effects of Different Versions
for Southwest Parks
WTP 1
WTP2
WTP 3
Eauation 1
Eauation 2
Ecmation 3
Variable
Coefficient
Coefficient
Coefficient
(t-statistic)
(t-statistic)
(t-statistic)
Intercept
11.30
24.96
25.65
(0.72)
(1.31)
(1.46)
Income
0.99 **
1.21 "
0.93 **
(6.32)
(6 ~:6)
(5.16)
Age
-0.25
-C i *
-0.43
(- 1.02)
(-1.35)
(-1.50)
Male
-6.35
-0.75
-3.95
(-0.79)
(-0.08)
(-0.43)
F-Visit
P-State
16.69 **
22.83 **
19.42 **
(2.06)
(2.35)
(2.16)
Vers4 * P-State
Version 4
3.43
(0.32)
Version 5
6.81
-0.39
-0.31
(0.64)
(-0.03)
(-0.03)
Version 6
-3.75
8.94
5.27
(-0.31)
(0.76)
(0.49)
Vers 6 * CARes
18.66
(1.02)
R2
.07
.08
.06
F
5.62
8.56
5.75
n
645
573
568
WTP values arc for Southwestern parks. Questionnaire Versions 3, 4 (WTP2), 5, and 6 arc included
and WTP3 equations include Versions 3, 5. and 6. See variable definitions in Table 4.5-13.
*p < .10, **p < .05
in the WTP1 equations. The WTP2
RCG/Hagler, Bailly, Inc.
-------
4-73
Responses to Question 16, concerning perceived accuracy of the WTP responses, and
responses to Question 17, concerning allocation of the WTP to visibility only, were very
similar for Versions 3 (baseline) and 5. Response rates for the two versions were also very
similar (Table 4.1-1). In general, it is expected that the shorter the questionnaire, the
higher the response rate. However, this one paragraph in this 12-page questionnaire
apparently made little difference.
These results suggest that the additional information included in Version 3 had minimal,
if any, effect on the WTP responses. Fischhoff and Furby (1988) have addressed the issue
of balancing the need for a very complete scenario development versus information
overload. These results suggest that omitting some features, which in this case are more
formal than substantive using the Fishhoff and Furby nomenclature, may have minimal
impact, or as a group may have offsetting impact. This does not mean that WTP questions
do not need to be carefully introduced, but it suggests that the introduction included in
Version 5 was sufficient (or at least that the additional explanation included in Version 3
did not alter things in any appreciable way).
Effect of Limited Regional Focus
Version 6 presented information on the southwestern region only. The same national map
was included in the insert, but the only photographs included were those for the Grand
Canyon. In addition, Questions 3, 4, 9, and 10 referred to parks in the southwestern region
only. The WTP questions themselves were exactly the same, but it was expected that the
responses might be higher for Version 6 as there was less emphasis placed on visibility in
regions other than the Southwest in the earlier questions.
The differences in the means for Versions 3 and 6 are generally statistically insignificant,
although for California residents the Version 6 results are substantially larger. However,
the means decrease for New York residents and are virtually unchanged for Arizona
residents. The California residents may have been affected differently than the New York
residents because California was one of the three regions included in the set of photographs
and some of the preliminary questions in Version 3 of the questionnaire. Seeing some
information about their own region may have had more of a downward influence on their
WTP responses for another region than for the New York residents who lived in none of
the three focus regions. It is also possible that since California is a neighboring state to the
region of focus in Version 6, the California respondents may have responded more similarly
to Arizona residents than they did when information about California parks was also
included. For these reasons, a dummy variable for both Version 6 and California residents
(Ver6*CARes) was included in the analysis reported in Table 4.5-18.
The regression results shown in Table 4.5-18 confirm that the WTP responses to Version
6 are not statistically significantly different. The coefficient for the Version 6 dummy
variable is not statistically significant in any of the regressions for the three WTP responses.
Equation 4 for WTP1 finds that the coefficient for the dummy variable for both Version
6 and California residence is positive and similar in magnitude to the home-state coefficient.
The coefficient is not, however, statistically significant.
RCG/Hagler, BaiHy, Inc.
-------
4-74
It appears that showing photographs from only one region does not necessarily cause a
significant upward bias in the WTP responses; or at least such a bias is not changed by
including photographs and preliminary questions for other regions. The exception may be
if photographs from the respondent's own region are included before asking questions about
another region. The results for the California respondents are statistically inconclusive on
this question, but suggest that it should be further explored in future research. The results
could have important implications for CVM design and expense. They suggest that it may
not be necessary to spend substantial effort (including researcher resources and survey
pages) to address at length many weakly related resource changes.
Effect of Multiple Region Bidding
Version 4 respondents were asked what they would be willing to pay to improve avera
visibility individually in each of the three regions from the 50th percentile level to the 75th
percentile level. The WTP introduction indicated "... would be worth to your household if
you had to pay for the improvements in all three regions each year." It was expected that
the per region WTP responses would be lower than for the versions that asked for WTP
for one region only (See Section 3.1.3). Of particular interest is whether residents from the
same state as the park in a WTP question respond differently from other respondents. To
examine this, an additional dummy variable (Vers4 * P-state) is equal to one when the
respondent is answering Version 4 and lives in the same state as the park pictured for the
focus region in question.
Regression results shown in Table 4.5-18 illustrate the lack of a statistically significant effect
of Version 4 on the WTP responses for the southwestern region. The Version 4 dummy
variable appears only in the WTP1 regressions because only the change from the 50th
percentile to the 75 percentile was considered in Version 4. The Version 4 coefficient was
not statistically significant in either of the first two WTP1 regressions. The estimated
coefficient for the V4*P-state variable suggests that the responses to Version 4 showed a
negative impact on the "home state" effect, but the difference is not statistically significant.
Similar results (not shown) were obtained with the WTP responses for the California parks.
This suggests that it may not be incorrect to add together WTP responses given for one
region at a time, if more than one region were to be affected by a particular pollution
control option being considered. However, these results are inconsistent with the findings
of some previous contingent valuation studies (discussed in Chapter 2). It is possible that
these results are an artifact of having broken the WTP question into three parts, asking
about each of the regions separately. It may be the case that had one WTP been asked to
protect all three regions at once, that the value would be less than the sum of the three
individual WTPs, even though the introduction indicated all three regions were to be paid
each year. For example, in recent research, Irwin et al. (1989) and Boyce et al. (1990)
found that the sum of WTP values for individual components of a resource change, when
asked separately, exceeded a single WTP for the :al change, which the authors refer to
as a problem of additivity. The conclusion Irv. " et al. draw is that it may be more
appropriate to value the entire policy package in one WTP and then disaggregate to
component values rather than value individual resource components and aggregate.
RCG/Hagler, BaiHy, Inc.
-------
4-75
4.5.5 Estimated WTP Functions for Changes in Visibility
If WTP estimates are to be used to analyze policy alternatives that might cause changes in
visibility other than the exact changes presented in the WTP questions, it is necessary to
know how WTP can be expected to change as a function of the change in visibility. This
section presents the results of regression analysis on the WTP responses using several
different measures of the changes in visibility reflected in the WTP questions. It is not
certain what measure best reflects respondents' perceptions of the visibility conditions
presented in the photographs. The four quantitative measures used are:
Change in Percentile (DPTTLEV The difference between the percentile score of
current visibility conditions for each photo. For example, C to A = 90 - 50 = 40, C
to B = 75 - 50 = 25, etc. The change in percentile is the same for each of the regions
for WTP1, then again for WTP2 and for WTP3 (except in Version 4). These
measures are examined because they were the only quantitative information presented
in the survey instrument and photo handout.
Change in Average Visual Range (PVR): The difference between the current average
visual range associated with each of the percentile conditions illustrated in the
photographs. These visual range estimates are not derived directly from the
photographs, but are based on the historic record of visibility conditions at the sites
at which the photos were taken.
Log of the Ratio of Visual Range (LNVR): The natural logarithm of the ratio of
the hypothetical new average visual range to the current average visual range. Using
this measure in the WTP function implies that WTP is constant for a given percentage
change in visual range.
Change in Average Atmospheric Transmission (DATRAN): These estimates are
based on measurements taken directly from the photographs in the insert and reflect
an average measure of contrast across the whole scene. The atmospheric transmission
is higher when the visibility is better. Previous perceptions studies have found a
correlation between observer ratings of perceived visual air quality and atmospheric
transmission (See Section 3.2).
The values of these four measures for each of the scenes shown in the questionnaire insert
were given in Table 3.2-1. The simple correlation coefficients among these measures are
quite high (from .60 to .94), suggesting that all four measures are reflecting similar
information about the differences shown in the photographs. However, as can be seen in
Table 3.2-1, there are some noticeable differences between the regions in how these
measures change across the set of four photographs. The percentiles are the same for all
RCG/Hagler, Bailly, Inc.
-------
4-76
three regions. The changes in visual range (and the levels of visual range) are greatest for
the Southwest and smallest for the Southeast. The spread in the atmospheric transmission
estimates is greatest for the Southeast photographs and smallest for the Southwest
photographs.
These four different measures of the change in visibility conditions shown in the
photographs are used to estimate WTP functions in this section of the analysis. The WTP
function attempts to explain variations in the WTP responses by considering the change in
visibility conditions presented in the WTP questions, various characteristics of the scenario,
and various socio-economic characteristics of the respondent. The WTP function has a
direct relationship with a utility function, therefore the form of the WTP function implies
a related form of the utility function. The WTP function is derived from the utility function
in that it is a representation of the change in income that would keep utility constant in the
event of the change in visibility conditions. A general form of the utility function with
respect to visibility might be:
U = a * f(V) + b * g(X) * f(V) + d * Y (4-1)
utility
some function of visibility conditions V
some function of the individual's socio-economic characteristics X
that influence the effect of V on the individual's utility
income
coefficients
The WTP function for a change in visibility conditions from Vt to V2 derived from this
utility function is:
WTP = a, [f(V2) . f(V0] + b, * g(X) * [f(V2) . f(V2)] (4-2)
Where a! and bj are new coefficients.
Tables 4.5-19 through 4.5-22 show estimation results for several different specifications of
a WTP function using the four measures of the change in visibility conditions defined above.
For three of these measures (DPTILE, DVR, and DATRAN), f(V) is a simple linear
function. For LNVR, f(V) is the natural log of visual range. All of the adjusted WTP
responses for each region and each different change in visibility are included in these four
tables. For the most part, this means three WTP estimates per respondent. The means and
standard deviations for all of the regression variables in these tables are given in Table 4.5-
23.
Where:
U
f(V)
g(X)
Y
a,b,d
RCG/Hagler, BaiHy, Inc.
-------
4-77
Table 4.5-19
WTP Regression Results with Change in Percentile
Equation Equation Equation Equation Equation Equation
(1) (2) (3) (4) (5) (6)
Full Full Excludes Full Full Full
Sample Sample WTP All Equal Sample Sample Sample
> $0
DPTILE
1.42 **
(31.24)
1.25 **
(22.09)
i 70 **
(27 50)
0.07
(0.27)
1.41 **
(21.29)
0.0016
(0.006)
DP* CA
0.05
(0.45)
0.04
(0.35)
DP * SE
0.02
(0.15)
0.10
(0.78)
DP * AGE
-0.007 **
(-2.01)
-0.007
(-1.93)
DP* MALE
-0.24 **
(-2.21)
-0.24
(-2.21)
DP * INC
.024 **
(11.50)
0.024
(11.42)
DP * FVIS
0.27 **
(6.16)
0.27
(6.20)
DP * PSTATE
0.48 "
(5.09)
R2
.18
.19
.24
.23
.18
.23
F
975.90
503.68
755.97
215.45
325.23
153.92
n
4340
4340
2441
3573
4340
3573
t-statistics in parentheses.
* p < .10
** p < .05
RCG/Hagler, Bailly, Inc.
-------
4-78
Table 4.5-20
WTP Regression Results with
Change in Visual Range (km)
Equation
Sample
Equation
Full
Sample
Equation
(3)
Excludes
WTP All Equal
> $0
Equation
Sample
Equation
Pull
Sample
Equation
full
Sample
DVR
0.88 **
(28.87)
0.79 **
(20.54)
1.08 **
(25.45)
¦0.02
(-0.10)
0.75 **
(20.45)
-0.30 *
(-1.74)
DVR*
CA
0.34 **
(4.38)
0.35 **
(4.08)
DVR *
SE
**
0.66
(6.55)
0.79 **
(6.70)
DVR *
AGE
-0.004*
(-1.94)
-0.003
(-1.45)
DVR *
MALE
-0.15 **
(-2.06)
-0.15 **
(-1.99)
DVR *
INC
0.016 **
(11.36)
0.015 **
(11.03)
DVR *
FVIS
0.17 **
(5.68)
0.20 **
(6.68)
DVR *
PSTATE
0.26 **
(4.15)
R2
.16
.16
.21
.21
.17
.22
F
833.31
426.81
647.80
186.78
299.03
142.99
n
4340
4340
2441
3573
4340
3573
t-statistics in parentheses.
* p < .10
** p < .05
RCG/Hagler, Bailly, Inc.
-------
4-79
Table 4.5-21
WTP Regression Results with
Log of Ratio of Visual Range (km)
Equation Equation Equation Equation Equation Equation
Full* Full Excludes Full Kill Si
Sample Sample WTP All Equal Sample Sample Sample
> $0
LNVR
i
77.73 **
(28.16)
66.69 **
(19.69)
95.57 **
(24.97)
4.09
(0.28)
139.73 **
(21.16)
88.45 **
(5.28)
LNV * CA
-43.28 **
(-4.82)
-47.45 **
(-4.62)
LNV * SE
-84.49 **
(-11.36)
-87.92 **
(-10.19)
LNV * AGE
¦0.35
(-1.61)
-0.55 **
(-2.57)
LNV * MALE
-14.20 **
(-2.13)
-14.17 **
(-2.16)
LNV * INC
1.21 **
(9.49)
1.28 **
(10.18)
LNV* FVIS
16.29 **
(6.26)
11.79 **
(4.54)
LNV * PSTATE
32.44 **
(5.59)
R:
.15
.16
.20
.19
.18
.22
F
792.81
414.77
623.68
172.13
319.68
142.48
n
4340
4340
2441
3573
4340
3573
t-statistics in parentheses.
* p<.10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-80
Table 4.5-22
WTP Regression Results with Change in
Average Atmospheric Transmission
Equation
AJ)
Sample
Equation
%
Sample
Equation
Excludes
WTP All Equal
> $0
Equation
Sample
Equation
full
Sample
Equation
full
Sample
DATRAN
220.49 **
(23.67)
190.12 **
(16.53)
272.27 **
(20.84)
17.25
(0.34)
894.84 **
(20.94)
782.55
(11.18)
DAT * CA
-660.38 **
(-14.63)
-708.25
(-13.69)
DAT * SE
-736.64 **
(-16.60)
-780.58
(-15.28)
DAT * AGE
¦0.78
(-1.06)
-1.30
(-1.81)
DAT * MALE
-40.44 *
(-1.78)
-40.11:
(-1.83)
DAT * INC
3.32 **
(7.72)
3.53'
(8.50)
DAT * FVIS
42.32 **
(4.75)
31.06 :
(3.57)
DAT * PSTATE
87.58 **
(4.48)
R2
.11
.12
.15
.14
.17
.19
F
560.12
29135
434.45
117.91
290.92
123.07
n
4340
4340
2441
3573
4340
3573
t-statistics in parentheses.
* p < .10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-81
Table 4.5-23
Means and Standard Deviations of Regression Variables
Standard
Variable N Mean Deviation Minimum Maximum
WTP
4340
19.14
114.27
¦1000.00
1000.00
LNVR
4896
0.18
0.56
-0.92
1.10
DVR
4896
26.38
45.26
-45.00
95.00
DATRAN
4896
0.06
0.17
-0.20
0.44
DPTILE
4896
9.90
33.41
-40.00
40.00
AGE
4698
47.00
15.83
18.00
94.00
MALE
4896
0.59
0.49
0.00
1.00
INC
4224
41.44
25.52
5.00
105.00
FVIS
4417
3.22
1.25
1.00
5.00
PSTATE
4896
0.34
0.48
0.00
1.00
RCG/Hagler, Bailly, Inc.
-------
4-82
The first equation in each of the four tables shows the results of the WTP responses
regressed on the visibility measure alone. In each case the estimated coefficient is
statistically significant at greater than the 99 percent confidence level. The R-squared
statistics indicate that these measures alone explain between 11 and 18 percent of the
variation in the WTP responses. The highest R-squared was obtained with the change in
percentile measure, and the lowest was obtained with the weighted average atmospheric
transmission measure.
The second equation in each of the four tables shows the results when a dummy variable
(PSTATE) is added that equals one when the respondent is a resident of the state in which
the illustrated park is located, multiplied by the visibility measure. In each case this "home-
state" variable is statistically significant at greater than the 95 percent confidence level.
When the home-state variable is added, the coefficients for the visibility measures alone
(now representing the WTP of the out-of-state residents) are reduced, but remain
statistically significant. The results of these second equations suggest WTP is 33 percent
to 50 percent higher for home state residents.
The third equation in each of the four tables shows the results of regressing the WTP
response on the visibility measure when respondents who gave the same non-zero WTP
response for different changes in visibility are excluded. It was expected that this might
increase the statistical significance and the magnitude of the estimated coefficient as
compared to the estimated coefficients in the first equations. The question is to what
degree the coefficients might be affected due to the inclusion of the all equal responses.
In all cases, the statistical significance increased, as expected, and the coefficients increased
by about 23 percent. It is not entirely clear whether deleting the all-equal non-zero
responses might be more appropriate for evaluating a policy alternative involving a change
in visibility. This depends on the reasons for the all-equal responses, which remain
somewhat speculative without further research (See Section 4.5.1 above). To remain
conservative, the all-equal responses have been retained throughout the remaining analysis.
The fourth equation in the four tables includes four socio-economic variables: (1) age of
respondent (AGE), (2) sex of respondent (MALE), (3) 1987 annual household income in
thousands (INC), and (4) probability of a visit to a park in the region in the next five years
(FVIS). Following the form of the WTP function shown above, the socio-economic
variables are multiplied by the visibility measure.' The results show a statistically significant
positive relationship between WTP and both income and future park visit expectations.
This is consistent with findings reported in previous sections of the analysis. The future
park visitation probability and the home-state variable are significantly correlated
(correlation coefficient is .44, p < .001) and show fairly similar explanatory power in
9 Strictly speaking, income would not enter interactively with visibility under specification 4-1, although it
might enter in this way under alternative specifications. It is modeled this way in the regression analysis for
consistency with other socio-economic variables. Alternatively, it could be considered a proxy for other socio-
economic variables, such as education, that would enter the specification in this manner.
RCG/Hagler, Bailly, Inc.
-------
4-83
separate regressions. The coefficient for MALE is negative and significant at least at the
90 percent confidence level in ail four tables. The coefficient for AGE is negative and
significant in some cases. In all of these fourth equations, the coefficient for the visibility
measure alone is no longer statistically significant. The effect of the visibility measure is
now reflected in the coefficients for the other independent variables.
Dummy variables for the different focus regions have been added in the fifth equations in
the four tables. This was done to see if there were significant differences in the WTP
responses for the same change in each visibility measure in the different regions. This
could occur for many different reasons. One may be that the differences in the scenery
provoke different responses. Another may be that there are non-linearities in the values
for changes in visibility that are not fully reflected in the visibility measures used in these
estimations. The results show significant differences across the regions for some of the
visibility measures. If the coefficients in the fifth equations are labeled in order as A, B,
and C, they can be interpreted as follows:
WTP for Southwest parks = A * change in visibility
WTP for California parks = (A + B) * change in visibility
WTP for Southeast parks = (A + C) * change in visibility
The results shown in Table 4.5-19 indicate that the coefficient for DPTILE is essentially the
same in the three regions. The DVR coefficient, shown in Table 4.5-20, is statistically
significantly different for the each of the regions, being lowest for the Southwest and highest
for the Southeast. The coefficients for LNVR and DATRAN, shown in Tables 4.5-21 and
4.5-22, are also statistically significantly different for all three regions. For both of these
measures the coefficients are smallest for the Southeast and largest for the Southwest. The
results for Equations 4 through 6 in Tables 4.5-19 through 4.5-22 show that the coefficients
for the regional visibility shift variables (i.e., DP*CA) are robust to the inclusion or
exclusion of the socio-economic variables, and vise versa.
The choice of visibility measure in Tables 4.5-19 through 4.5-22 has little impact on the
overall explanatory power of the variance in the WTP responses, as based upon the F and
R2 statistics, although the percentile and ln(V2/Vl) measures are slightly preferred in the
selected functional form. Other results concerning the relative merit of the alternative
visibility measures might be found using alternative functional forms.
The consistent sign and statistical significance of the percentile visibility measure for each
of the three study regions is of some interest as this measure has not previously been used
in visibility value or visual air quality rating studies. One reason for this finding may be
that the percentile of current visibility conditions was the only visibility measure tied to the
photographs explicitly presented to the respondents. Respondents may have anchored upon
this information as a basis for their WTP response. Similar WTP responses across regions
RCG/Hagler, Bailly, Inc.
-------
4-84
might occur if respondents generally feel that visibility protection is equally important for
national parks in each region, and use the percentile figures to determine the range of
possible improvements to be obtained, or degradations to be prevented. In essence,
respondents may be anchoring upon the status quo conditions at each site (Samuelson and
Zeckhauser 1986) and bidding for modest and large changes relative to the maximum
obtainable changes. The results are also consistent with the hypothesis that the visibility
changes at parks in each region are of comparable value because the differences in visibility
condition changes across the regions are offset by differences in the scenic quality impacts
due to the visibility impairment.
4.5.6 Allocations of WTP to Specific Parks and Motives
Two questions that followed the WTP questions asked respondents to estimate the
percentage of their WTP responses that they would want to allocate in various ways.
Question 18 asked them what percentage they would want to go to the specific park shown
in the photograph. Question 19 asked them what percentage of their WTP responses they
would attribute to various specified motives.
Values for Specific Parks
Table 4.5-24 shows the average percentages given for each of the specific parks. The results
are given by home-state residents and residents of states outside the region. Overall, the
average percentage given for the illustrated park ranged from 41 percent to 45 percent.
Figure 4.5-3 shows the frequency of the responses to Question 18 for all respondents who
gave at least one non-zero WTP response. All the regions are combined in this figure, but
Version 4 is excluded because this question was not asked. The pattern was very similar
for each of the regions. For each region, the share to the park included in the photographs
exceeds 1/n where n is the number of parks shown on the map for each region. This may
be because the pictured parks are more prominent than some of the other parks in the
same region.
One factor that appears to affect the responses to Question 18 is whether the respondent
lives in the state in which the specific park is located. These home-state residents have a
significantly higher average percentage for that specific park (roughly 50 percent versus
about 40 percent for residents of other states). Home-state residence is correlated with
previous and future visitation to the parks in the region, including the specific park of
interest. As each of the illustrated parks is among the most frequently visited in each of
the states, it is likely that the home-state effect is related to familiarity with the park due
to previous visitation and to higher expectations regarding future visits to that park.
Question 18 allows some comparison to the results of the Southwest Parklands study
(Schulze et al. 1981), in which respondents were asked to give WTP for the Grand Canyon
and then to give WTP for the same change in visibility throughout the remainder of the
RCG/Hagler, BaiHy, Inc.
-------
4-85
Table 4.5-24
Average Percentage of Regional WTP
for the Individual Park
Yosemite Grand Canyon Shenandoah
(Version 1) (Versions 3, 5, 6) (Version 2)
Home-State Residents 46% 52% 54%
Residents of Other States 38% 41% 38%
All Respondents 41% 45% 44%
region. It was expected that the percentage of total WTP for the region allocated to the
Grand Canyon might be higher in the Southwest Parklands study because of the order in
which the questions were asked. The results are consistent with this expectation, although
the difference is not large. The percentage of total WTP attributed to the Grand Canyon
was about 54 percent in the Southwest Parklands study versus about 45 percent in the
current study.
Table 4.5-25 shows the average WTP estimates for each of the potential changes in visibility
for each of the specific parks. These were calculated by multiplying the percentage given
by each respondent for the specific park by the WTP responses given for the region. If the
respondent gave WTP responses, but did not answer Question 18, then the average
percentage response for that park given by other residents of the same state was used. The
results in Table 4.5-25 show an even greater spread in WTP means between home-state
residents and residents of other states than was the case with the regional WTP estimates.
This is the result of the combined effect of the higher regional WTP and higher percentages
for the specific park. As with the percentages, the overall average WTP estimates for each
park are similar.
Table 4.5-26 shows the regression coefficients for WTP1 (for the change from the 50th to
the 75th percentile) for the different parks. The dependent variable is the product of the
WTP1 response and the percentage given for the illustrated park. The income coefficients
are reasonably stable, as they were for the regions (Table 4.5-17), but the visitation and
home-state coefficients are somewhat less stable than for the regions. The magnitude of
the visitation coefficient (F-VISIT) is fairly consistent, but the statistical significance is not.
The reduced stability relative to that observed for the regions presumably reflects the fact
that the original WI'P responses were tied to visibility conditions at all parks (on the map)
in the region. Therefore, it is appropriate that regional park visitation is more highly
correlated to the regional WTP responses.
RCG/Hagler, Bailly, Inc.
-------
Figure 4.5-3
Frequency of Question 18 Responses:
Percentage for Illustrated Park
Number of Responses
350
300
250
200
150
100
50
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percentage for Illustrated Park
Note: Includes respondents to all
versions, except version 4, who gave at
least one non-zero WTP response.
-------
4-87
Table 4.5-25
WTP Means for Individual Parks
Home State
Residents of
Park
Residents
Other States
All Respondents
Yosemite
WTP1
$29.1
$20.5
$23.4
(5.2)
n=113
(3.0)
n=218
(2.7)
n=331
WTP2
$35.1
$24.4
$28.0
(5.9)
n=113
(3.6)
n=221
(3.1)
n=334
WTP3
$30.7
$23.9
$26.2
(5.3)
n=113
(3.6)
n=217
(3.0)
n=330
Grand Canyon
WTP1
$31.0
$19.5
$23.8
(5.2)
n=241
(2.6)
n=395
(2.6)
n636
WTP2
$40.8
$23.8
$30.2
(6.0)
n=240
(2.7)
n=398
(2.8)
n=638
WTP3
$35.4
$21.5
$26.8
(5.8)
n=241
(2.6)
n=391
(2.7)
n=632
Shenandoah
WTP1
$39.6
$14.6
$23.3
(9.8)
n=120
(1.8)
n=226
(3.7)
n=346
WTP2
$48.8
$22.3
$31.7
(H.l)
n=124
(3.4)
n=226
(4.5)
n=350
WTP3
$42.4
$19.3
$27.6
(8.5)
n=124
(3.1)
n=223
(3.7)
n=347
Note: Missing values for the percentages for specific parks were replaced by means for the
corresponding questionnaire versions and states.
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
4-88
Table 4.5-26
Regression Coefficients for Different Parks for WTP1
Yosemite Grand Canyon Shenandoah
Equation Equation Equation Equation Equation Equation
(i) (2) (1) (2) (1) (2)
Income 0.37 ** 0.26 ** 0.55 ** 0.47 ** 0.42 ** 0.41 "
(3.69) (2.31) (5.18) (4.05) (2.71) (2.49)
Age 0.13 0.002 -0.05 -0.07 0.06 0.02
(1.10) (0.01) (0.45) (-0.51) (0.33) (0.11)
Male -4.34 -5.27 -1.99 -3.53 -8.60 -9.79
(-0.69) (-0.84) (.0.32) (-0.56) (-0.99) (-1.11)
F-Visit 4.85 ** 3.33 * 3.44
(2.44) (1-73) (1.21)
P-State 10.63 * 15.38 ** 20.27 **
(1.74) (2.64) (2.32)
R2 .22 .23 .16 .16 .14 .13
F 18.73 19.66 25.63 24.45 10.84 9.73
n 274 274 534 534 275 275
t-statistics in parentheses
* p < .10
** p < .05
RCG/Hagler, BaiJly, Inc.
-------
4-89
Tables 4.5-27 and 4.5-28 show regression results for WTP for Yosemite National Park.
Table 4.5-27 shows the results when the change in percentile (DPTILE) is used as the
visibility measure. Table 4.5-28 shows the results when the log of the ratio of visual range
(LNVR) is used. Tables 4.5-29 and 4.5-30 show the same regressions for Grand Canyon
National Park, and Tables 4.5-31 and 4.5-32 show the same regressions for Shenandoah
National Park.
As with the region-wide WTP responses reported in Table 4.5-19, the coefficients for the
change in the visibility percentile (Equation 1 in Tables 4.5-27, 4.5-29, and 4.5-31) are very
similar for the different parks. Equation 2 in Tables 4.5-27, 4.5-29, and 4.5-31 suggests that
the home-state effect is more variable than the responses for out-of-state respondents. The
home-state coefficient for Shenandoah is almost three times the home-state coefficient for
Yosemite. The home-state coefficient for Grand Canyon lies between these two. This
reflects more variability in the home-state coefficient than was seen for the regions. Adding
the age, sex, and income variables in Equations 3 and 4 increases the explanatory power
of the regressions somewhat, primarily due to the statistical significance of the income
coefficient. The sign and magnitude of the coefficients for socio-economic variables are
generally consistent in the regressions for each of the three regions. Age and sex were not
significant except for Shenandoah, where male respondent gave significantly lower WTP.
This is apparently related to the percentage for Shenandoah rather than to the WTP for the
region because this difference was not seen in Table 4.5-17.
The coefficients for the log of the ratio of the second to the first level of visual range are
different for the three parks, as they were for the regions. The highest is for Grand Canyon
and the lowest is for Shenandoah. The explanatory power of this measure of visibility is
very similar to that obtained with the change in percentile. The results for the other
variables are also very similar.
Values for Specific Motives
In Question 19, all the respondents were asked to estimate what percentage of their WTP
responses they would attribute to three specific motives. Respondents were given the
opportunity to give another motive if the three listed were not sufficient to account for 100
percent of their WTP response. The three motives were intended to reflect (1) option
price, (2) bequest value, and (3) existence value (See Section 3.1.4).
"Other" responses were given by 173 respondents (about 13 percent of those who answered
the question). Review of the other motives listed found they all fell into two categories:
(1) they were for things other than visibility at national parks, or (2) they could be
interpreted as one of the three motives already listed. Therefore, no other motives were
added to the original list of three for further analysis. Even when we accounted for the
"other" responses, we found that 109 respondents (8 percent) gave percentages that did not
sum to 99 or 100 percent. About two-thirds of these summed to less than 99 and about
one-third summed to more than 100. In order to exclude the "other" responses and adjust
for the summation problems, the responses given for the three motives were adjusted to
sum to 100 percent. Each response was adjusted proportionately. This affected the mean
percentage responses for each motive in the following way:
RCG/Hagler, Bailly, Inc.
-------
4-90
Table 4.5-27
WTP Regressions for Yosemite with
Change in Visibility Percentile
Equation Equation Equation Equation
(1) (2) (3) (4)
DPTILE 0.72 ** 0.64 ** 0.39 * -0.17
(15.23) (10.99) (1.88) (-0.66)
DP * AGE -0.001 0.001
(-0.40) (0.37)
DP * MALE -0.08 -0.09
(-0.70) (-0.80)
DP * INC 0.009 ** 0.008 **
(4.32) (3.60)
DP * FVIS 0.18 **
(3.95)
DP * PSTATE 0.24 ** 0.24 **
(2.41) (2.17)
R2 .19 .19 .22 .23
F 231.88 119.41 46.08 48.85
n 995 995 821 821
t-statistics in parentheses
* p < .10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-91
Table 4.5-28
WTP Regressions for Yosemite with
Log of Ratio of Visual Range (km)
Equation Equation Equation Equation
(1) (2) (3) (4)
LNVR 47.33 ** 42.11 ** 26.35 * -10.06
(14.83) (10.74) (1.90) (-0.58)
LNV * AGE -0.10 0.08
(-0.43) (0.31)
LNV * MALE -5.01 -5.74
(-0.66) (-0.76)
LNV * INC 0.58 ** 0.49 **
(4.17) (3.47)
LNV * FVIS 11.75 **
(3.80)
LNV * PSTATE 15.31 ** 15.03 **
(2.28) (2.00)
R2 .18 .19 .21 .22
F 219.99 113.05 43.60 46.23
n 995 995 821 821
t-statistics in parentheses
* p < .10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-92
Table 4.5-29
WTP Regressions for Grand Canyon with Change
in Visibility Percentile
Equation Equation Equation Equation
(1) (2) (3) (4)
DPTILE 0.75 ** 0.60 ** 0.09 -0.36
(17.17) (10.88) (0.46) (-1.40)
DP * AGE -0.003 -0.0004
(-0.99) (-0.11)
DP * MALE -0.06 -0.11
(-0.53) (-1.02)
DP * INC 0.017 ** 0.014 **
(8.08) (6.98)
DP * FVIS 0.19 **
(4.26)
DP * PSTATE 0.40 ** 0.52 **
(4.42) (5.01)
R2 .13 .14 .18 .17
F 294.87 158.66 70.17 68.50
n 1906 1906 1598 1598
t-statistics in parentheses
* p < .10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-93
Table 4.5-30
WTP Regressions for Grand Canyon with Log
of Ratio of Visual Range (km)
Equation Equation Equation Equation
(1) (2) (3) (4)
LNVR 74.45 ** 59.43 ** 8.34 -36.50
(17.00) (10.76) (0.42) (-1.43)
LNV * AGE -0.32 -0.03
(-0.97) (-0.10)
LNV * MALE -5.35 -10.50
(-0.50) (-0.98)
LNV * INC 1.65 ** 1.42 **
(8.04) (6.94)
LNV * FVIS 18.77 **
(4.26)
LNV * PSTATE 39.77 ** 51.25 **
(4.43) (4.98)
R2 .13 .14 .18 .17
F 289.05 155.72 68.90 67.31
n 1906 1906 1598 1598
t-statistics in parentheses
* p < .10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-94
Table 4.5-31
WTP Regressions for Shenandoah with Change
in Visibility Percentile
Equation Equation Equation Equation
(1) (2) (3) (4)
DPTILE 0.77 ** 0.53 ** 0.57 " 0.65**
(12.00) (6.72) (1.98) (1.90)
DP * AGE -0.006 -0.006
(-1.24) (-1.22)
DP * MALE -0.33 ** -0.35 **
(-2.12) (-2.22)
DP * INC 0.012 ** 0.013 **
(4.18) (4.35)
DP * FVIS 0.028
(0.46)
DP * PSTATE 0.68 ** 0.50 **
(5.12) (3.19)
R2 .12 .14 .15 .14
F 144.07 86.86 31.15 28.81
n 1043 1043 832 832
t-statistics in parentheses
* p < .10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-95
Table 4.5-32
WTP Regressions for Shenandoah with Log
of Ratio of Visual Range (km)
Equation Equation Equation Equation
(1) (2) (3) (4)
LNVR 30.16 ** 20.72 ** 22.30 ** 25.43 **
(12.06) (6.74) (2.00) (1.92)
LNV * AGE -0.25 -0.25
(-1.25) (-1.23)
LNV * MALE -13.32 ** -13.94 **
(-2.19) (-2.28)
LNV * INC 0.49 ** 0.51 **
(4.21) (4.38)
LNV * FVIS 1.12
(0.48)
LNV * PSTATE 26.69 ** 19.51 **
(5.17) (3.23)
R2 .12 .14 .16 .15
F 145.50 87.90 31.61 29.20
n 1043 1043 832 832
t-statistics in
* p < .10
** p < .05
parentheses
RCG/Hagler, Bailly, Inc.
-------
4-96
Motive
Unadjusted
Adjusted
Option Price
29%
31%
Bequest Value
34%
37%
Existence Value
30%
32%
Other
5%
0%
The relative magnitudes of the mean percentages did not change with this adjustment.
They were all increased by a similar amount.
Table 4.5-33 shows the mean adjusted percentages for each region, for home state and
other residents. The responses for the different regions appear very similar. Home-state
residents gave higher percentages for option price, as might be expected, and lower
percentages for bequest and existence values. The overall mean percentages for each
motive are each close to one-third, although they are a bit higher for bequest than for the
other two motives. Figure 4.5-4 illustrates, however, that the responses varied across the
sample, with as many respondents giving a very low percentage for one of the motives as
those who gave 33 percent for each of the three.
Table 4.5-34 shows how the WTP motives are correlated with prior ratings of reasons for
visiting and protecting national parks (Questions 5 and 7 discussed in Section 4.4). This
table indicates whether a statistically significant correlation exists between the respondent's
rating of a particular reason and the percentage allocated to each motive. These
correlations are essentially consistent with a priori expectations and, on the whole, support
the credibility of the responses to Question 19. The existence value percentage is correlated
with the importance placed on the natural environment and on the concept of preservation
in general. The option price percentage is correlated with the importance the respondent
placed on things related more to use, such as outdoor recreation activities, doing things with
others, and having a change from usual surroundings. The percentage for option price is
also significantly correlated with the following factors:
the number of children in the household
previous visitation to national parks in the region
probable future visitation to national parks in the region
Table 4.5-35 shows the adjusted mean WTP values for option price, bequest value, and
existence value for the first visibility change scenario (WTP1). These motive values are
computed for each individual as their WTP * percent for visibility * percent to each motive,
then averaged across all respondents. For the total WTP and for each of the motives, and
each of the regions, the means for the home-state residents are higher than the means for
residents of other states. In general, bequest value and existence value for each park region
show less variability across state of residence than does option price.
RCG/Hagler, BaiHy, Inc.
-------
4-97
Table 4.5-33
Average Percentage Allocation of WTP by Motive
Option
Price
Percentage
Bequest
Value
Percentage
Existence
Value
Percentage
California Parks (version 1)
California Residents
36%
33%
31%
(n=110)
(2.2)
(1.9)
(2.5)
Residents of Other States
27%
39%
34%
(n=195)
(1.5)
(1.9)
(2.1)
All Respondents
30%
37%
33%
(n=305)
(1.3)
(1.4)
(1.6)
Southwest Parks (versions 3, 5, 6)
Arizona Residents
34%
39%
27%
(n=224)
(1.4)
(1.6)
(1.7)
Residents of Other States
29%
38%
33%
(n=353)
(1.3)
(1.5)
(1.6)
All Respondents
31%
38%
31%
(n=577)
(1.0)
(1.1)
(1.2)
Southeast Parks (version 2)
Virginia Residents
35%
34%
31%
(n=109)
(2.1)
(2.1)
(2.7)
Residents of Other States
28%
38%
34%
(n=202)
(1.7)
(1.9)
(2.2)
All Respondents
30%
36%
33%
(n=311)
(1.3)
(1.4)
(1.7)
All Three Regions (version 4)
All Respondents
32%
37%
31%
(n=153)
(1.7)
(1.9)
(2.0)
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
Figure 4.5-4
Frequency of Responses for Option Price
Bequest Value, and Existence Value
Frequency of Responses
~i i ' i ' i " i ' i 1 1 r
5% 15% 25% 35% 45% 55% 65% 75% 85% 95%
Percentage of Willingness to Pay
Hi Option Price
Bequest Value
Existence Value
-------
Table 4.5-34
Correlations Between Responses to Question 19 and
Reasons for Visiting and Protecting National Parks
Significant Positive Correlation Found (p < .05)
Percent for Percent for Percent for
Use by Self Use by Others Existence
and Household Now and Future Regardless of
Use
Reasons for Visiting (Q-5)
1. Experience unique natural places x
2. Experience unique historic places x
3. Do something enjoyable with others x
4. Enjoy vastness of nature x
5. Outdoor recreation activity x
6. Change from usual surroundings x
Reasons for Protecting (Q-7)
1. Others in family can visit x
2. Others outside family can visit x
3. Natural preservation, even if no one visits x
4. Scientific research x
5. National heritage x
6. Not development everywhere x
-------
4-100
Table 4.5-35
Average Option Price, Bequest and Existence Values for WTP1
Option Bequest Existence
Price Value Value
California Parks (version 1)
California Residents $18.08 $17.89 $24.22
(n=113) (2.65) (2.74) (9.13)
Residents of Other States $10.81 $15.53 $12.85
(n=218) (1.91) (2.48) (2.27)
All Respondents $13.29 $16.34 $16.73
(n=331) (1.56) (1.88) (3.46)
Southwest Parks (version 3,5,6)
Arizona Residents $15.52 $17.42 $17.89
(n=241) (2.42) (2.13) (3.38)
Residents of Other States $10.87 $14.14 $14.37
(n=395) (1.31) (1.70) (2.63)
All Respondents $12.63 $15.38 $15.70
(n=636) (1.23) (1.330) (2.08)
Southeast Parks (version 2)
Virginia Residents $23.43 $19.15 $17.18
(n=120) (4.99) (4.14) (3.93)
Residents of Other States $7.64 $13.48 $10.15
(n=226) (1.09) (2.17) (1.20)
All Respondents $13.12 $15.45 $12.59
(n = 346) (1.91) (2.02) (1.58)
All Three Regions
Range of values for
ratio of (other state/
home state) 33-70% 70-87% 53-80%
Note: Missing values for the percentages by motives were replaced by the means for the
corresponding questionnaire versions and states. Adjusted WTP values are used.
Standard error of mean in parentheses.
RCG/Hagler, BaiHy, Inc.
-------
4-101
4.5.7 Additional Sensitivity Tests
Tables 4.5-19 through 4.5-22 and 4.5-27 through 4.5-32 pool observations to examine the
effect of changes in visibility upon the WTP bids. However, in these regressions it is often
the case that when socioeconomic variables are entered into the equation, interactively with
visibility, that the individual visibility variable (not interacted with other variables) losses
statistical significance. One issue, then, is whether the inclusion of the visibility variable is
in fact explaining a significant share of the variation in the data, or whether socio-economic
variables alone would be sufficient. Table 4.5-36 verifies that visibility is an important
variable in the analysis using the Grand Canyon WTP data as an example. The first
equation in the table replicates Equation (3) results earlier reported in Table 4.5-30. The
second equation eliminates the visibility variable. It is quickly apparent that the inclusion
of the visibility variable is statistically significant.
Because the WTP distribution is truncated at zero, a tobit model (Tobin 1958) is a more
correct analytic procedure than ordinary least squares (OLS). Tobit models were run for
several of the specifications estimated earlier with OLS. The magnitude of the coefficients
for each of the previously significant variables changed by no more than ten percent, and
usually by much less. Previously significant variables remained statistically significant. The
minimal change in results may be due to the relatively small share (7%) of zero values in
the data.
Two additional modified regression analyses were undertaken to further examine the
sensitivity of the data to specific data characteristics. Recognizing the nature of the data
as pooled data with multiple observations (up to three) per respondent, fixed effects models
and random effects models (also referred to as error components models) were run (see
Judge et al. 1982, pages 497-498; or Pindyck and Rubinfeld 1981, pages 252-261, for
additional discussion). Selected specifications from Tables 4.5-19, 4.5-21, 4-5.29 and 4.5-
30 were rerun using these statistical techniques. In nearly all cases, neither the fixed effects
models nor random effects models significantly altered the analysis. While some individual
coefficients changed, very few changes were statistically significant, and the predicted WTP
values never changed by more than 10 percent (usually increased) for any of the visibility
scenarios considered, and were never statistically significantly different from what would be
predicted with the ordinary least squares models already presented. The second analysis
used a weighted regression approach that weighted observations by the inverse of the self-
reported accuracy score. Again, the results were not statistically significantly different from
the ordinarily least squares models.
4.6 SUMMARY OF WRITTEN COMMENTS
Respondents were asked, following the WTP questions, to provide any additional
information that might help explain their responses to these and other questions. They
were also asked to provide any other comments they might have at the end of the
questionnaire. Sixty-two percent of the respondents gave one or more comments in
response to these questions.
RCG/Hagler, Bailly, Inc.
-------
4-102
Table 4.5-36
Selected Regression Sensitivity Analyses
for the Grand Canyon WTP
Equation
(1)
Equation
(2)
Method
LNVR
LNVR * AGE
AGE
LNVR * MALE
MALE
LNVR * INC
INC
LNVR * PSTATE
PSTATE
R2
F
N
OLS
8.34
(0.42)
-0.32
(-0.97)
OLS
-5.35
(-0.50)
1.65 *
(8.04)
51.25 **
(4.98)
.18
68.90
1598
-0.04
(-0.53)
-0.00
(-0.00)
.000216
(3.25) **
5.99*
(1.655)
.02
9.486
1716
t-statistics in parentheses.
* p < .10
** p < .05
RCG/Hagler, BaiHy, Inc.
-------
4-103
Table 4.6-1 provides a summary of the written comments offered by respondents.
Comments have been grouped into fourteen categories. These categories cover the most
frequent types of comments received. The percentages in Table 4.6-1 sum to more than
100 because some respondents gave more than one comment. The most frequently given
comment indicated that visibility protection and/or protection of national parks in general
is seen by the respondent as an important issue. A comment along these lines was given
by about 20 percent of those who gave any comment. This contrasts with the 5 percent who
wrote that the issue is not very important and the 1.8 percent who wrote that they saw little
value in visibility protection for national parks.
About 17 percent of those who gave written comments wrote that they thought everyone
should pay to clean up air pollution, while about 9 percent said that polluters and park
users should pay to protect the park resources. About 13 percent said that they could not
afford to pay more than they had given in response to the WTP questions. Many of these
comments included suggestions that the respondent would have liked to have given a higher
dollar amount, but felt he or she could not. This comment was given with the whole range
of dollar responses, including several of the zeros.
About 10 percent indicated that they had other concerns related to national parks that they
would also like to see addressed, such as water pollution, littering, congestion, fire
management, acid rain and wildlife protection. About 14 percent also complained about
taxes being too high or about government inefficiency of some sort. Other comments are
summarized in the table.
4.7 COMPARISONS OF RESULTS TO PRIOR NATIONAL PARK VISIBILITY
PRESERVATION VALUE STUDIES
Table 4.7-1 summarizes the WTP responses, adjusted to 1988 dollars, for the most
comparable scenario for visibility protection at the Grand Canyon National Park for this
study, the Southwest Parklands study (Schulze et al. 1981), and Tolley et al. (1986). (See
Section 2.3 for a review of the latter two studies). Each of these scenarios presents the
WTP to prevent degradation in average conditions, and the WTP and visual range levels
in the study scenarios used for comparison are listed in the first three columns. Because
the WTP values are for different changes in visibility, the last two columns normalize the
results in terms of WTP per percent change in visual range, and WTP per km change in
visual range. Using these normalizations, mean WTP for visibility protection at the Grand
Canyon National Park in this study is about 25 to 32 percent of the WTP mean obtained
for the Southwest Parklands study and is about 13 to 45 percent larger than the mean
obtained by Tolley et al.
The Southwest Parklands study is most comparable to the current effort. Among the
reasons results of the current study are so much lower than those obtained in the Southwest
Parklands study, we believe, are:
RCG/Hagler, BaiHy, Inc.
-------
4-104
Table 4.6-1
Summary of Written Comments
Number Percent of all Respondents
Respondents giving one or more comments
1012
62.8%
Comment Categories
Number
Percent of Respondents
Giving Comments
Visibility/national parks important
207
20.5%
General concerns about pollution
175
17.3%
Everyone should help pay
169
16.7%
Complaints about government or taxes
144
14.2%
Can't afford to pay more (or anything)
133
13.1%
Other issues also important
103
10.2%
Others (polluters, users) should pay
93
9.2%
Problems answering questions
70
6.9%
Complaints about industry/developers
61
6.0%
Issue isn't very important
51
5.0%
Complaints about study/questions
34
3.4%
Low value for visibility protection
18
1.8%
Concerned about parks only in own region
16
1.6%
Complaints about National Park Service
4
0.4%
RCG/Hagler, BaiHy, Inc.
-------
4-105
Table 4.7-1
Comparing Point Estimates of Visibility Value Studies
Study Mean Annual WTP per WTP per km
Household VI V2 Percent Change Change in
WTP ($1988) (km) (km) in Visual Range Visual Range
Southwest
Parklands $95 200 155 $4.22 $2.11
Tolley et al. $21 200 155 $0.93 $0.47
NP Visibility
Values* $27 155 115 $1.05 $0.68
Using the WTP to prevent a decrease in visibility scenario
RCG/Hagler, BaiHy, Inc.
-------
4-106
1. The adjustment for visibility alone correcting for part-whole bias. This
adjustment alone reduces the Grand Canyon values in the current study by
about 35 percent.
2. Asking for the percentage of WTP for the Grand Canyon after asking WTP for
the region, rather than asking for WTP for the Grand Canyon first, to minimize
aggregation impacts. I.e., starting with the larger policy package then
disaggregating to individual components. As a result, the portion of total region-
wide WTP assigned to the Grand Canyon is about 45 percent versus about 55
percent in the earlier effort, or an 18 percent decrease in the reported Grand
Canyon value.
3. Using annual versus monthly WTP questions, to minimize aggregation impacts,
and other CVM design refinements. The magnitude of these effects is unknown,
but they are expected to have further decreased the estimated values in the
current effort.
4. Lead questions about other resource protection issues, and information and
photos about visibility in other regions to place the issue in perspective. These
again are expected to decrease the values in the current effort.
5. Different procedures for identifying potential protest zero and large bid
responses. Unfortunately, the number of accepted and rejected zero and high
bids, and missing responses, were not reported in the Southwest Parklands study
making it impossible to determine the significance of the different approaches
used.
6. Differences in sample socioeconomic characteristics and sampling procedures
and response rates. The Southwest Parklands study used on-site interviewers,
versus the current mail survey approach. The refusal to participate rates were
not reported for the on-site interview solicitations, and the effect of any potential
interview bias, or importance bias created by on-site interviewers, is unknown.
In total, items 1 and 2 account for a 50 percent reduction in the current study values as
compared to the Southwest Parklands study (.62 * .82 = .50). Items 3 through 6 may
account for much of the remaining differences in the values received.
The Tolley et al. WTP values are much different. They are incremental values for visibility
protection at the Grand Canyon NP as part of a policy package that also protects visibility
in Chicago and throughout the eastern U.S. This is a significantly different (larger) policy
package than considered in the current study. In other regards the study is similar to the
Southwest Parklands study. Due to the differences in the policy package alone, it is not
surprising that the WTP estimates for visibility protection at the Grand Canyon are lower
than those obtained in the Southwest Parklands study and in the current study. Moreover,
RCG/Hagler, Bailly, Inc.
-------
4-107
in Tolley et al, the Grand Canyon presentation and value elicitation occurred after the
Chicago and eastern U.S. bids were elicited. The Grand Canyon values might have been
larger had information been initially presented that this was to be included in the package.
The limited documentation on survey implementation and data handling procedures
provided by Tolley et al. make more detailed comparisons difficult.
RCG/Hagler, Bailly, Inc.
-------
5-1
5.0 SUMMARY CONCLUSIONS AND DIRECTIONS FOR FUTURE
RESEARCH
5.1 SUMMARY OF RESULTS
There are many conclusions of the research. Because the results presentation is extensive,
a bulleted summary of key findings is presented in this section grouped by topic area. For
each conclusion, the corresponding tables and figures are cited for easy follow-up reference.
The text corresponding to each table should be consulted as well and can be found in the
section with the two initial digits of the table. For example, Table 4.4-3 is discussed in
Section 4.4.
5.1.1 About the Sample and Response Rates
Over 70 percent of the valid sample was contacted. Response rates in individual
states are not consistently or statistically significantly different whether the survey
version focuses upon parks in the respondent's home state, or on parks in
another region of the country. Any potential response bias due to interest in
parks in the home region, versus elsewhere, appears to be minimal. (Tables 4.1-
1 and 4.1-2).
Variations examined in the survey design dealing with the amount of
information, whether one or more parks were pictured, and whether WTP
questions were asked for parks in one region or in three regions had no
consistent statistical effect upon response rates. It is concluded that these
features, in and of themselves, exerted little influence on the results through
different response rates. (Table 4.1-3)
Non-response bias was examined through a small telephone follow-up survey.
Non-respondents have lower probability of visiting national parks and slightly
lower income, which would decrease WTP responses. However, just as in the
returned mail surveys, the large majority of telephone respondents felt protecting
visibility at national parks is important, and would be willing to pay for visibility
protection. Some non-response bias may exist, but the effect on the sample-
wide WTP mean estimate appears to be relatively small. (Tables 4.3-1 through
4.3-4 and Section 4.4)
The sample varies somewhat from national characteristics. It has a higher
percentage of males and is somewhat older, which are characteristics that are
related to lower WTP values. The sample also has somewhat higher income and
education than the national average, which is related to higher WTP values.
(Table 4.2-1) The effect of these sample characteristics can be adjusted for
when applying the results in policy scenarios (Tables 4.5-16, 4.5-17, 4.5-18
through 4.5-22, 4.5-26 through 4.5-32).
RCG/Hagler, BaiHy, Inc.
-------
5-2
5.1.2 Importance of Protecting Visibility at National Parks
Protecting against air pollution decreasing the ability to see scenic vistas was a
high priority for 72 percent of respondents and a medium priority for nearly all
other respondents. (Table 4.4-7 and Table 4.6-1 where comments are
summarized)
Non-use related motives were as, or more, important, than use-related motives
for protecting visibility at national parks. (Table 4.4-5)
Improving visibility from current average conditions to somewhat above average
conditions would enhance on-site enjoyment for about 95 percent of respondents.
(Table 4.4-9)
Over 90 percent of respondents would be willing to pay something for visibility
protection at national parks. This sentiment was nearly equally strong for all
three regions. Sentiment was strongest for protection in a region by residents
of the same region. (Table 4.4-10)
Preventing visibility degradation was felt to be slightly more important than
obtaining improvements. (Table 4.4-11)
5.1.3 WTP for Specific Visibility Scenarios
On the order of 83 percent stated non-zero WTP for the visibility scenarios
presented. After deleting non-respondents and protest responses, the percent
increases to 93 percent. Some respondents held very high values for visibility
protection. Both the valid zero and high bids were highly correlated with income
and with past and expected future national park visitation. (Tables 4.5-2, 4.5-
4, Figure 4.5-2)
The mean WTP values are quite similar for each of the three national park
regions investigated. The mean WTP values (adjusted to the percent for
visibility) are around $40 to $60 per year per household for the three scenarios
considered. Values by residents of the same state as the illustrated park are
typically higher than values for the same region by individuals who live outside
the region. (Table, 4.5-10 and 4.5-12)
Median bids are 25 to 50 percent of the means. This is to be expected as the
distribution is necessarily truncated at zero and includes responses from
individuals who have very large values, based upon income, visitation
expectations and other reasons, for this visibility protection. (Table 4.5-14)
RCG/Hagler, BaiHy, Inc.
-------
5-3
The mean percent of the regional bid that is attributable to the specific park
depicted in the photograph varies depending upon the park and whether
respondents are from the same state or from other states. Residents from the
same state assigned between 46 and 54 percent for pictured park while residents
from other states assigned between 38 and 41 percent for the pictured park.
These shares exceed 1/n, where n is the number of parks in each region. This
may be because the pictured parks are more prominent than most of the other
parks in the same region. (Tables 4.5-24 and 4.5-25, and Figure 4.5-3)
Differences in characteristics of respondents appear to explain differences in
mean WTP for residents of different states for protection at the three national
park regions. These characteristics include income and the probability of visiting
parks in the region of focus. The influence of these characteristics is consistent
across different focus regions. (Tables 4.5-11, 4.5-13, Tables 4.5-16 through 4.5-
23 and 4.5-26 through 4.5-32)
WTP bids are found to follow ex ante expectations in terms of their correlations
to attitudes and opinions about protecting national parks, protecting visibility at
national parks and WTP for this protection. (Table 4.5-15)
Visibility value functions were examined that relate WTP to changes in visibility
conditions and socio-economic characteristics of the respondent. The use of four
different visual air quality measures, in separate estimations of the selected
functional forms, does not appreciably alter the explanatory power of the
regressions. The four measures were: visual range, natural log of the ratio of
new to old visual range, average atmospheric transmission coefficient and the
percentile of annual visual range conditions. (Tables 4.5-19 through 4.5-23 for
regional WTP regressions, and 4.5-27 through 4.5-32 for individual park WTP
regressions)
Values were partitioned, sample-wide, to option price (32%), bequest values
(37%) and existence values (31%). Residents of the same state as the focus park
assigned higher percentages to option price than did non-residents. As a result,
the share of the WTP that is assigned to bequest values and existence values for
each park region shows smaller variability across state of residence than does
option price. This reflects the higher probability of one's own visitation to parks
in the same region as one is currently residing in. (Tables 4.5-32 through 4.5-
34, and Figure 4.5-4)
RCG/Hagler, BaiHy, Inc.
-------
5-4
5.1.4 CVM Method Findings
About 11 percent of the respondents did not answer the WTP questions or gave
apparent protest responses. About 7 percent gave apparently valid zero value
responses. These are within normal ranges for CVM studies. (Table 4.5-2)
Ex ante attempts to lead respondents to value only visibility changes were
insufficient. Over two-thirds stated, after answering the WTP questions, that only
a portion of their WTP estimate was for visibility at national parks in the
specified region. The average percent for visibility was 62 percent. The follow-
up question readdressing the part-whole issue appears to have worked well in
helping to separate values for visibility from values for other concerns. This
adjustment was used to compute visibility WTP results used throughout the
report. (Table 4.5-3, Figure 4.5-1)
Over 80 percent stated a self perceived accuracy of their WTP answers as either
"very accurate" or "within the ballpark." Only 19 percent felt their answers were
"somewhat inaccurate" or "probably very inaccurate." Self reported accuracy was
not related to past visitation. Some have argued that inaccuracy may lead to
overstated WTP responses. However, mean WTP decreased significantly as self
reported accuracy decreased. Deletion of those individuals with low self
reported accuracy would significantly increase, not decrease, the mean bids. This
deletion was not done to maintain conservative results. (Table 4.5-6)
About 40 percent of the respondents providing WTP responses provided the
same WTP response for all three scenarios presented to them (including 7
percent all zero bidders). This finding is only recently being reported in CVM
studies. In this study it may be due to:
1. Anchoring and lack of attention. Uncertainty and lack of attention (or effort
to quantify values) may lead to all equal responses. If values are uncertain
and have some reasonable error, the difference in values across scenarios may
not merit small adjustments in bids. Some respondents may see improving
conditions from photo C to photo A as having a small additional value as
compared to just obtaining photo B. Similarly, preventing photo D may have
similar value to obtaining photo B. Given edit behavior, or the lack of desire
to deal with small differences, the respondent may fail to adjust to the second
and third scenarios and the same value is reported for all three scenarios.
2. Making a contribution. The all equal bids may reflect a tendency to make
a contribution to visibility protection, regardless of the level of protection.
Some individuals also state that "if everyone paid then the problem would be
solved." Because this all equal tendency has been observed in other surveys
where the making of contributions would be much less likely, we discount this
explanation as being dominant.
RCG/Hagler, Bailly, Inc.
-------
5-5
The mean of the all equal responses tend to be anchored upon the first WTP
response. Because the means of WTP2 and WTP3 for respondents who do not
provide all equal bids are larger than WTP1, this anchoring may therefore
downward bias the WTP2 and WTP3 results, and downward bias the estimated
slope of the underlying visibility value functions. (Tables 4.5-7 through 4.5-9)
Differences in the level of information provided did not consistently influence
the WTP results. Relatively minor characteristics of the scenario were deleted
in one survey version with no impact upon response rates or mean WTP results.
Fischhoff and Furby (1988) have indicated a need for a very detailed scenario
development. It may be the case that omitting some minor features has minimal
or offsetting impacts. (Table 4.5-18) Potential explanations range from (1) the
omitted information may be relatively less significant than other design features,
(2) the impacts are offsetting, to (3) the information may not have been well
understood (which contradicts the pretest findings) and therefore has no effect.
It has been argued that respondents must be presented with information on many
other goods simultaneous to the presentation of the good for which WTP values
will be elicited. This, in part, is to address the potential part-whole bias.
Presenting information on only one region (Version 6), versus on three regions
(Version 3), in the form of maps, photographs and questions had some minor
impacts upon the results, and did not appear to lead to any consistent increase
in the WTP results. Further, when a bid was elicited separately, but in the
same survey instrument, for each of the three regions (Version 4), the bids again
did not show any consistent or statistical difference. Individuals appeared to
have addressed each national park region independently. (Table 4.5-18)
If individuals had to simultaneously pay their stated WTP for all three regions,
rather than for just one region, it is uncertain whether the total WTP would
decrease as compared to the sum of the bids for all three regions. Survey
Version 4 elicited a separate WTP for each of the three regions stating that the
WTP for each region would be paid each year. The results of Version 4 were
not statistically different from the other versions. (Table 4.5-18) However, other
recent evidence on these types of aggregation issues suggests that a single WTP
for all three regions at once may obtain a lower value than the sum of values
for three separate WTP questions. This alternative was not examined.
The sample-wide mean percentages of the WTP bids allocated to different
motives (option price, bequest value and existence value) are all generally
between 30 and 40 percent. From this it might be inferred that many or most
individuals pay little attention to (or do not know) their allocation and simply
allocate equal shares to each of three motives. However, the individual data
reveal that this does not appear to be the case. Rather, a wide distribution of
allocations is reflected and correlation analysis reveals statistically significant
relationships to previously stated behavior and attitudes related to each motive.
(Tables 4.5-33 through 4.5-35 and Figure 4.5-4)
RCG/Hagler, BaiHy, Inc.
-------
5-6
5.1.5 Comparisons to Prior Related Studies
The results here are 25 to 32 percent as large as the comparable scenario values
obtained in the Southwest Parklands study (Schulze et al. 1981). These
differences are primarily attributed to the visibility adjustment questions (Q-
17), asking for region-wide bids, then disaggregating to a Grand Canyon NP
value component, using annual versus monthly bids, sample differences, and
other CVM design characteristics.
The results are 13 to 45 percent larger than those for the Grand Canyon NP in
Tolley et al. (1986). However, the later study is for a substantially different
policy package. Tolley et al. obtain values for visibility protection at the Grand
Canyon NP as the 3rd identified component of a policy package that first request
WTP estimates for simultaneously protecting visibility in Chicago and throughout
the eastern U.S. As a result, the valuation scenarios are quite different.
5.1.6 Conclusions
Some conclusions can be based upon the statistical analyses presented in Chapter 4, while
others are impressions gained through review of surveys responses and written comments,
other statistical analysis not presented, and evidence from the literature.
In general, respondents seemed to perceive the survey issues as valid and to make a sincere
effort to answer the WTP questions accurately. The WTP responses are generally quite
consistent with expressed attitudes and behavior. Nonetheless, valuation of such changes
is not an easy, or precise, task. While 80 percent of respondents felt their WTP responses
were "within the ballpark" or better, only 15 percent felt their responses were "very
accurate" and a third of those were zero bidders. Inaccuracy is also potentially reflected
in that a large percent of individuals gave the same response to all WTP questions. I.e.,
responses may reflect a rough estimate of value for these types of visibility changes, but
their WTP responses may not be so accurate as to merit refinement for the exact
differences across the scenarios.
In formulating WTP responses, respondents appear to have focused upon key information
about the visibility scenario and visibility conditions. Among these may have been the
percent of time conditions exist at each of the different regions. When answering the WTP
questions, respondents also had difficulty separating the visibility impacts at national parks
from other resource protection concerns facing national parks, but the follow-up question
served to help clarify this difficulty and its magnitude upon the visibility value estimates.
It goes without saying that the estimates must be used with care. The statistical precision
of the estimates, as reported in the tables in Chapter 4, may overstate the accuracy of the
valuation. For example, the statistical precision of predicted WTP values in the visibility
RCG/Hagler, BaiHy, Inc.
-------
5-7
value regression analyses is on the order of +_ 10 percent. However, other uncertainties in
the respondents' formation of the WTP responses, and in the researchers' interpretation of
the data suggest the uncertainty may be much larger. We would find it reasonable to
suggest the values may be indicative of an accuracy no better than ±_ 50 percent. But this
should not be taken as an opportunity to argue for "true values" half, or double, what is
reported here based upon selective identification of potential biases. For each potential
upward bias, there is an offsetting potential downward bias, and vice versa. Therefore, the
estimates must be taken as indicative of the range respondents feel best reflects their
monetary values for the visibility scenarios presented.
Use of the estimates in policy analysis must also account for the sample characteristics
versus the characteristics of the population to which the estimates would be applied. For
example, at a minimum, coefficients in the visibility value functions should be used to adjust
predicted values for the population of interest.
5.2 DIRECTIONS
Many research efforts could be undertaken, some with minor modifications to the existing
instruments, that would further the understanding of societal values for visibility protection
at national parks. These might include a version that asks a WTP for all three regions at
once, which might be compared to the sum of WTP's obtained in Version 4 to test the
impact of different levels of aggregation when asking for component values.
The survey results indicate a diminishing marginal utility (as reflected by diminishing
marginal WTP) for visibility improvements, which is consistent with the other preservation
value literature reviewed in Chapter 2. This may be an accurate reflection of underlying
values, or may be an artifact of the structure of the WTP elicitation. If some respondents
think of their first response more as a general contribution than a payment for a specific
quantity of a good, then asking them a second WTP for a larger change may suggest they
should give the same or a slightly larger response. This may look like diminishing marginal
utility, but may not really be. Additional survey versions could test these issues by excluding
Photo B and asking for WTP for visibility changes from Photo C to Photo A. The results
of this comparison may also aid in interpreting the "all equal" responses and their impact
upon the valuation by examining whether these responses remain anchored upon the first
scenario.
The survey focused upon fairly substantial changes in average visibility conditions. It would
also be of interest to consider impacts to only a limited number of days, which would result
in smaller average visibility changes, to more fully examine the shape of a visibility value
function. It would also be of interest to examine the effect of variations in the season of
impact on the value responses.
RCG/Hagler, Bailly, Inc.
-------
5-8
The current survey could also be easily modified to more fully examine the impact of
alternative information upon response rates and WTP responses. One important and useful
test would be to dramatically reduce the background motives questions and the CVM
scenario development, which would reduce the complexity of the instrument development
for the researcher and reduce the effort to complete the instrument for the respondent.
More substantive extensions could also be considered. This might include the examination
of substantially different visibility impacts such as plumes, different colors of haze, etc., and
different policy packages, such as visibility impacts at other parks, combined haze and
plume control, etc.
In general, several design and evaluation issues have been initially addressed and merit
additional research. This includes more investigations into: the "all equal" responses; the
accuracy of responses, including how respondents perceive the accuracy of their responses;
and the use of follow-up questions to partition natural resource protection values to obtain
values for the visibility component. More research is also merited on the issue of what
information is most important to CVM scenario development, and when too much scenario
detail overloads respondents to the point of reducing the quality of responses.
RCG/Hagler, BaiHy, Inc.
-------
6-1
6.0 BIBLIOGRAPHY
Air Pollution Control Association. 1987. Visibility Protection: Research and Policy
Aspects. P.S. Bhardwaja (Ed.). Transactions of APCA International Specialty Conference,
Grand Teton National Park, Wyoming, September 1986.
Bell, P.A., W. Malm, R.J. Loomis, and G.E. McGlothin. 1985. "Impact of Impaired
Visibility on Visitor Enjoyment of the Grand Canyon: A Test of the Ordered Logit Utility
Model." Environment and Behavior 17:459-74.
Bishop, R.C., and T.A. Heberlein. 1979. "Measuring Values of Extramarket Goods: Are
Indirect Measures Biased?" American Journal of Agricultural Economics 61(5):926-930.
Bishop, R.C., and K.J. Boyle. 1985. The Economic Value of Illinois Beach State Nature
Preserve. Final report to the Illinois Department of Conservation, December.
Boyce, R., T. Brown, G. McClelland, G. Peterson, and W. Schulze. 1990. "An
Experimental Examination of Intrinsic Environmental Values." University of Colorado,
Department of Economics, manuscript. Boulder, Colorado.
Boyle, K.J., and R.C. Bishop. 1985. 'The Total Value of Wildlife Resources: Conceptual
and Empirical Issues." Paper presented at the Association of Environmental and Resource
Economists Workshop on Recreation Demand Modeling, Boulder, Colorado, May 17-18.
Brookshire, D.D., L.S. Eubanks, and Alan Randall. 1983. "Estimating Option Prices and
Existence Values for Wildlife Resources." Land Economics 59: 1-15.
Brown, T.C. 1984. "The Concept of Value in Resource Allocation." Land Economics 60:
231-46.
Cameron, T.A., and D.D. Huppert. 1988. "The Relative Efficiency of Payment Card vs.
Referendum Data in Non-Market Resource Valuation," Department of Economics, UCLA.
Carson, R.T., and R.C. Mitchell. 1988. "The Value of Clean Water: The Public's
Willingness to Pay for Boatable, Fishable, and Swimmable Quality Water." UCSD
Discussion Paper 88-13, San Diego, California.
Carson, R.T., R.C. Mitchell, and P.A. Ruud. 1989. "Valuing Air Quality Improvements:
Simulating a Hedonic Pricing Equation in the Context of a Contingent Valuation Scenario."
Paper presented at the Air and Waste Management Association Specialty Conference on
Visibility and Fine Particles, Estes Park, Colorado, October 15-19.
Chavas, J.P., R.C. Bishop, and K. Segerson. 1986. "Ex-Ante Consumer Welfare Evaluation
in Cost-Benefit Analysis," Journal of Environmental Economics arid Management, vol.13,
no. 3, pp. 255-268.
RCG/Hagler, BaiHy, Inc.
-------
6-2
Chestnut, L.G., and R.D. Rowe. 1983. "Integral Vista Benefits Analysis." Report prepared
for the National Park Service, Air Quality Division, Washington, D.C.
Chestnut, L.G., R.D. Rowe, and J. Murdoch. 1986. "Review of Establishing and Valuing
the Effects of Improved Visibility in Eastern United States." Report prepared for the U.S.
EPA, Office of Policy and Planning Evaluation, Washington, D.C. Contract No. 68-01-
7033, October.
Chestnut, L.G., and R.D. Rowe. 1989. "New National Park Visibility Value Estimates."
Paper presented at the Air and Waste Management Association Specialty Conference on
Visibility and Fine Particles, Estes Park, Colorado, October 15-19,
Cicchetti, C.J., and A.M. Freeman, III. 1971. "Option Demand and Consumer Surplus:
Further Comment." Quarterly Journal of Economics 85: 528-539.
Council on Environmental Quality. 1980. Public Opinion on Environmental Issues:
Results of a National Public Opinion Survey. Washington, D.C.: U.S. Government Printing
Office.
Crocker, T.D. 1985. "The Economic Benefits of Altering Forest Slash Burning Induced
Atmospheric Visual Impairment in the State of Oregon." University of Wyoming,
Department of Economics.
Cummings, R.G., D.S. Brookshire, W.D. Schulze. 1986. Valuing Environmental Goods:
An Assessment of the Contingent, Valuation Method. Totowa, New Jersey: Rowman and
Allenheld.
Dillman, D.A. 1978. Mail and Telephone Surveys: The Total Design Method. New York:
John Wiley & Sons, New York.
Fischhoff, B., and L Furby. 1988. "Measuring Values: A Conceptual Framework for
Interpreting Transactions with Special Reference to Contingent Valuation of Visibility."
Journal of Risk and Uncertainty l(June): 147-184.
Fisher, A., J. Krutilla and C. Cicchetti. 1972. "The Economics of Environmental
Preservation: A Theoretical and Empirical Analysis," American Economic Review. 62
(September): 605-619.
Fisher, A., and R. Raucher. 1984. "Intrinsic Benefits of Improved Water Quality:
Conceptual and Empirical Perspectives." In Advances in Applied Micro-Economics.
Volume 3, pages 37-66. JAI Press Inc.
Freeburn, S.A. 1987. "Rating Selected Methods for Reducing Prescribed Burning Impacts
on Class I Areas in Western Oregon Using Benefit and Cost Estimates." in P. S.
Bhardwaja, ed., Visibility Protection: Research and Policy Aspects. Transactions of the
APCA Specialty Conference, Grand Teton National Park, Wyoming, September 1986.
RCG/Hagler, BaiHy, Inc.
-------
6-3
Freeman, A.M., III. 1979. The Benefits of Environmental Improvement. Baltimore: Johns
Hopkins University Press for Resources for the Future.
Freeman, A.M., III. 1984. "The Sign and Size of Option Value." Land Emnomins 60: 1-
13.
Freeman, A.M., III. 1988. "Non-use Values in Natural Resource Damage Assessment."
Paper prepared for the Resources for the Future Conference on Assessing Natural
Resource Damages, Washington, D.C., June 16-17.
Gilbert, A.H. 1989. "Factors Affecting Visibility Values in Vermont." Paper presented
at the Air and Waste Management Association Specialty Conference on Visibility and Fine
Particles, Estes Park, Colorado, October 15-19.
Greenley, D.A., R.G. Walsh, and R.A. Young. 1981. "Option Value: Empirical Evidence
from a Case Study of Recreation and Water Quality." Quarterly Journal of Economics
96(4):657-683.
Haspel. A.E., and F.R. Johnson 1982. "Multiple Destination Trip Bias in Recreation
Benefit Estimation." Land Economics 58 (August).
Irwin, J.; D. Schenk, G. McClelland, W. Schulze, T. Stewart, and M. Thayer. 1989. "Urban
Visibility Values: Some Experiments in the Use of the Contingent Valuation Method."
Paper presented at the Air and Waste Management Association Specialty Conference on
Visibility and Fine Particles, Estes Park, Colorado, October 15-19.
Johnson, F.R., and A.E. Haspel. 1983. "Economic Valuation of Potential Scenic
Degradation at Bryan Canyon National Park" in R. Rowe and L. Chestnut (eds.), Managing
Air Quality and Scenic Resources at National Parks and Wilderness Areas. Westview Press,
Boulder, CO.
Judge, G.G., R.C. Hill, W. Griffith, H. Lutkepohl, and T. Lee. 1982. Introduction to the
Theory and Practice of Econometrics. New York:John Wiley & Sons.
Just, R.E., D.L. Hueth, and A. Schmitz. 1982. Applied Welfare Economics and Public
Policy. Prentice Hall.
Kahneman, D., and A Tversky. 1979. "Prospect Theory: An Analysis of Decisions Under
Risk," Econometrica. Vol. 47, No. 2, pp. 263-91.
Kahneman, D., and A. Tversky. 1982. "The Psychology of Preferences," Scientific
American. Vol. 246, No. 1, pp. 160-73.
RCG,/Hagler, BaiHy, Inc.
-------
6-4
Krutilla, J.V. 1967. "Conservation Reconsidered." American Economic Review 57: 777-
786.
Krutilla, J.V., C.J. Cicchetti, AM. Freeman, III, and C.S. Russell. 1972. "Observations
on the Economics of Irreplaceable Assets." in A.V. Kneese and B.T. Bower, eds.,
Environmental Quality Analysis: Theory and Method in the Social Sciences. Baltimore:
Johns Hopkins University Press for Resources for the Future.
Krutilla, J.V., and A.C. Fisher. 1975. The Economics of Natural Environments: Studies
in the Valuation of Commodity and Amenity Resources. Baltimore: Johns Hopkins
University Press for Resources for the Future.
MacFarland, K, W. Malm, and J. Molenar. 1983. "An Examination of Methodologies
and Social Indicators for Assessing the Value of Visibility." in Rowe and Chestnut (Eds.),
Managing Air Quality and Scenic Resources at National Parks and Wilderness Areas.
Boulder, Colorado: Westview Press.
Madariaga, B., and K.F. McConnell. 1987. "Exploring Existence Value." Water Resources
Research 23(5):936-942.
Malm et al. 1989. National Park Service Report on the Winter Haze Intensive Tracer
Experiment. National Park Service, Draft Final Report, Fort Collins, Colorado, September
26.
Malm, W. 1985. "An Examination of the Ability of Various Physical Indicators to Predict
Judgement of Visual Air Quality." Paper presented at the 78th Annual Meeting of the Air
Pollution Control Association, Detroit, Michigan, June 16-21.
Malm, W., K. Kelley, and J. Molenar. 1981. "Human Perception of Visual Air Quality
(Regional Haze)." Atmospheric Environment 15(10).
Malm, W.C., K.K. Leiker, and J.V. Molenar. 1980. "Human Perception of Visual Air
Quality." Journal of the Air Pollution Control Association 30:122-131.
McConnell, K.E. 1983. "Existence Value and Bequest Value." in R.D. Rowe and L.G.
Chestnut, eds., Managing Air Quality and Scenic Resources at National Parks and
Wilderness Areas. Boulder, Colorado: Westview Press.
Mitchell, R.C. 1984. "Public Opinion and Environmental Politics in the 1970s and 1980s,"
in Vig and Kraft (Eds.), Environmental Policy in the 1980's: Reagan's New Agenda.
Congressional Quarterly Press, Washington, D.C.
Mitchell, R.C., and R.T. Carson. 1981. An Experiment in Determining Willingness to Pav
for National Water Quality Improvements. Report to the U.S. Environmental Protection
Agency, Washington, D.C.
RCG/Hagler, BaiHy, Inc.
-------
6-5
Mitchell, R.C., and R.T. Carson. 1989. Using Surveys to Value Public Goods: The
Contingent Valuation Method. Washington, D.C.: Resources for the Future.
Morey, E.R.. 1984. "Confuser Surplus." American Economic Review. Vol. 74, No. 1, pp.
163-173.
National Park Service, Air Quality Division. 1988. "Air Quality in the National Parks."
Natural Resources Report 88-1. Denver, Colorado.
Peterson, D.C., R.D. Rowe, W.D. Schulze, G.W. Russell, R.B. Boyce, S.R. Elliot, and B.
Hurd. 1987. Valuation of Visual Forest, Damages from Ozone. University of Colorado
and RCG/Hagler, Bailly, Inc. report to the U.S. EPA Office of Policy, Planning and
Evaluation, Washington, D.C.
Pindyck, R.S. and D.L. Rubinfeld. 1981. Econometric Models and Economic Forecasts.
2nd edition. New York:McGraw-Hill Book Company.
Rae, D.A. 1983. "The Value to Visitors of Improving Visibility at Mesa Verde and Great
Smoky National Parks." in R.D. Rowe and L.G. Chestnut, eds., Managing Air Quality arid
Scenic Resources at National Parks and Wilderness Areas. Boulder, Colorado: Westview
Press.
Rae, D.A. 1984. Benefits of Visual Air Quality in CincinnatiResults of a Contingent
Ranking Survey Draft report prepared for Electric Power Research Institute by Charles
River Associates, Palo Alto, California, August 1984.
Rahmatian, M. 1986. "Extensions, of the Disaggregate Bid Experiment: Variations in
Framing." Journal of Environmental Management 22: 191-202.
Randall, A. and J. Stoll. 1980. "Consumers Surplus in Commodity Space." American
Economic Review. 70:449-455.
Randall, A, and J.R. Stoll. 1983. "Existence Value and a Total Valuation Framework."
in R.D. Rowe and L.G. Chestnut, eds., Managing Air Quality and Scenic Resources at
National Parks and Wilderness Areas. Boulder, Colorado: Westview Press.
Ross, D., W. Maim, and R. Loomis. 1985. "The Psychological Valuation of Good Visual
Air Quality by National Park Visits." In Proceedings of the 78th Annual Meeting of the
Air Pollution Control Association, Detroit, Michigan, June 16-21.
Ross, D. 1989, in progress. "Importance of Visibility to National Park Visitors," National
Park Service, Fort Collins, Colorado.
RCG/Hagler, BaiHy, Inc.
-------
6-6
Ross, D. and W. Malm. 1985. "An Assessment of Visitor Enjoyment and Values at Grand
Canyon and Mesa Verde National Parks," in Assessment of Visibility Impairment on Visitor
Enjoyment and Utilization of Park Resources. Cooperative Institute for Research in the
Atmosphere, Colorado State University. Fort Collins, Colorado.
Ross, D., G. Haas, R. Loomis, and W. Malm. 1985. "Visibility Impairment and Visitor
Enjoyment," in Assessment of Visibility Impairment on Visitor Enjoyment and Utilization
of Park Resources. Cooperative Institute for Research in the Atmosphere, Colorado State
University. Fort Collins, Colorado.
Rowe, R.D, R. d'Arge, and D. Brookshire. 1980. "An Experiment on the Economic Value
of Visibility." Journal of Environmental Economics and Management 7: 1-19.
Rowe, R.D., and L.G. Chestnut. 1982. The Value of Visibility: Economic Theory and
Applications for Pollution Control. Cambridge, Mass.: Abt Books.
Rowe, R.D., and L.G. Chestnut. 1983. "Valuing Environmental Commodities: Revisited."
Land Economics 59: 404-410.
Rowe, R.D., and L.G. Chestnut. 1985. Oxidants and Asthmatics in Los Angeles: A
Benefits Analysis. Washington, D.C., U.S. Environmental Protection Agency, EPA-230-07-
85-010, March (NTIS #PB85-228997).
Rowe, R.D., and L.G. Chestnut. 1986. "Addendum to Oxidants and Asthmatics in Los
Angeles: A Benefits Analysis." EPA-230-09-86-017. Report prepared for the U.S.
Environmental Protection Agency, Washington, D.C.
Rowe, R.D., W.D. Schulze, B. Hurd, and D. Orr. 1985. Economic Assessment of Damage
Related to the Eagle Mine Facility. Energy and Resource Consultants, Inc., report to the
State of Colorado Attorney General's Office, Denver, Colorado.
Rowe, R.D., W.D. Schulze, and B. Hurd. 1986. A Survey of Colorado Residents' Attitudes
About Cleaning Up Hazardous Waste-Site Problems in Colorado. Energy and Resource
Consultants, Inc. report to the State of Colorado Attorney General's Office, Denver,
Colorado.
Rowe, R.D., and W.D. Schulze. 1987. "Natural Resource Damages in the Colorado
Mountains: The Case of the Eagle Mine." Paper presented at the annual meetings of the
American Economic Association, Chicago, 111., December.
Rowe, R.D., A.M. Michelsen, and E.R. Morey. 1989. Fishing for Atlantic Salmon in
Maine: An Investigation into Angler Activity and Management Options. RCG/Hagler,
Bailly, Inc. report to the Bangor Hydro-Electric Company. Bangor, Maine.
RCG/Hagler, BaiHy, Inc.
-------
6-7
Samples, K.C., J.A. Dixon, and M.M. Gowen. 1986. "Information Disclosure and
Endangered Species Valuation." Land Economics 62(3):306-312.
Sanghvi, A. et al. 1989. Customer Value of Service Reliability: Residential. RCG/Hagler,
Bailly, Inc. report to Southern California Edison Company. El Monte, California.
Schulze, W.D., D.S. Brookshire, E.G. Walther, and K. Kelley. 1981. The Benefits of
Preserving Visibility in the National Parklands of the Southwest,. Vol. VIII. Methods
Development for Environmental Control Benefits Assessment. Report prepared for the
U.S. EPA, Washington, D.C.
Schulze, W.D., D.S. Brookshire, E. Walther, and K.K. MacFarland, M.A. Thayer, R.L.
Whitworth, S. Ben-David, W. Malm, and J. Molenar. 1983. "The Benefits of Preserving
Visibility in the National Parklands of the Southwest." Natural Resources Journal
23(1): 149-173.
Schulze, W.D. et al. 1990. Field and Laboratory Experiments on the Reliability of the
Contingent Valuation Method. Forthcoming, University of Colorado report to the U.S.
EPA under cooperative agreement #CR-812054.
Shuman, H., and S. Presser. 1981. Questions and Answers in Attitude Surveys:
Experiments on Question Form. Wording, arid Context. Academic Press.
Sinden, J.A., and A.C. Worrell. 1979. Unpriced Values: Decisions Without Market Prices.
New York: John Wiley & Sons.
Slovic, P. 1969. "Differential Effects of Real versus Hypothetical Payoffs on Choices
Among Gambles." Journal of Experimental Psychology 80:434-437.
Solmonson, S., G. Haas, and D. Ross. 1985. "Data Summary: 1983 Grand Canyon National
Park Visibility Project," in Assessment of Visibility Impairment on Visitor Eniovment and
Utilization of Park Resources. Cooperative Institute for Research in the Atmosphere,
Colorado State University. Fort Collins, Colorado.
Smith, V.K. 1972. "The Effect of Technological Change on Different Uses of
Environmental Resources," in Natural Environments: Studies in Theoretical and Applied
Analysis Edited by J. Krutilla, Johns Hopkins University Press, Baltimore.
Stoll, J.R., and L.A. Johnson. "Concepts of Value, Nonmarket Valuation, and the Case of
the Whooping Crane." Transactions of the Forty-ninth North American Wildlife and
Natural Resources Conference. 382-393.
RCG/Hagler, BaiHy, Inc.
-------
6-8
Stonefield, D.H. and D.W. Scott. 1989. "EPA's Visibility Protection Program: Ten Years
of Regulatory Development and Implementation." Paper presented at the Air and Waste
Management Association Specialty Conference on Visibility and Fine Particles, Estes Park,
Colorado. October 15-19.
Sutherland, R. J., and R.G. Walsh. 1985. "Effect of Distance on the Preservation Value of
Water Quality." Land Economics 61:281-91.
Tobin, R.P. 1958. "Estimation of Relationships of Limited Dependable Variables,"
Econometrics (26), pp. 24-36.
Tolley, G.A., A. Randall, G. Blomquist, R. Fabian, G. Fishelson, A. Frankel, J. Hoehn, R.
Krumm, E. Mensah, and T. Smith. 1986. "Establishing and Valuing the Effects of
Improved Visibility in Eastern United States." Report prepared for the U.S. EPA,
Washington, D.C.
Trijonis, J., M. Pitchford, W. Malm, W. Watson, and R. Husar. 1990 forthcoming. "Causes
and Effects of Visibility Reduction: Existing Conditions and Historic Trends." State of the
Science Technical Paper #24 prepared for the National Acidic Deposition Assessment
Plan 1989-1990 Integrated Assessment. Washington, D.C.
U.S. Environmental Protection Agency. 1985. Developing Long-Term Strategies for
Regional Haze: Findings and Recommendations of the Visibility Task Force. Research
Triangle Park, North Carolina.
Vartia, Yrgo. 1983. "Efficient Methods of Measuring Welfare Change and Compensated
Income in Terms of Ordinary Demand Functions," Econometrica. Vol. 51, pp. 79-98.
Walsh, R.G., R.A. Gillman, and J.B. Loomis. 1982. Wilderness Resource Economics:
Recreation Use and Preservation Values. Prepared for the American Wilderness Alliance,
Denver, Colorado.
Walsh, R.G., J.B. Loomis, and R.A. Gillman. 1984. "Valuing Option, Existence, and
Bequest Demands for Wilderness" in Land Economics 60(1): 14-29.
Walsh, R.G., L.D. Sanders, and J.B. Loomis. 1985. Wild and Scenic River Economics:
Recreation Use and Preservation Values. Prepared for the American Wilderness Alliance,
Englewood, Colorado.
Winger, W. 1985. "The Economic Impact of Changes in Visibility on a Tourist-Oriented
Economy in Southwest Colorado." Report prepared for National Park Service, Air Quality
Division, Denver, Colorado.
RCG/Hagler, Bailly, Inc.
-------
APPENDIX A
Sample Mail Survey Instrument
RCG/Hagler, BaiHy, Inc.
-------
[031H
Managing Visibility at National Parks
What is Your Opinion?
i
A A
YV
$
Research conducted for
The Center for Economic Analysis
at the University of Colorado
-------
ABOUT YOUR VISITS TO NATIONAL PARKS
Q~1 Have you personally ever visited any national park (including monuments
arid seashore*)? (Circle number)
YES
NO -
If NO, than skip to Question Q-4
0-2 The enclosed map shows national parks where visibility is considered to
be an important resource. Have you personally ever visited any of the
national parks shown on the map in each of the following regions?
Ever visited a national park ir. this
region? (Circle YES or NO fc eacn
region)
If YES, how many days
have you visited in just
the last 2 years?
The Northwest YES > DAYS
NO
California? YES > DAYS
NO
The Southwest? YES > DAYS
NO
The Central U.S.? YES > DAYS
NO
The Northeast? YES > DAYS
NO
The Southeast? YES > DAYS
NO
0-3 Have you personally ever visited the following national parks?
Ever visited this park?
(Circle YES or NO for each park)
Yoseaite National YES
Park in California NO
If YES, how many days
have you visited in just
the last 2 years?
DAYS
Grand Canyon National...
Park in Arizona
YES
NO
DAYS
Shenandoah National YES
Park in Virginia NO
DAYS
-------
Q-4 Thinking about the next 5 years, about how likely are you to visit any of
the national parks shown on the map in each of the following regions?
(Circle number of best response for each region)
DEFINITELY PROBABLY MIGHT PROBABLY DEFINITELY
WILL NOT WILL NOT VISIT WILL WILL
VISIT VISIT VISIT VISIT
California 1 2 3 4 5
The Southwest 1 2 3 4 5
The Southeast 1 2 3 4 5
0-5 Below are reasons some people have given for visiting national parks. How
important is each of these reasons to you personally? (Circle number of
best response for each reason)
NOT AT ALL SLIGHTLY SOMEWHAT VERY EXTREMELY
IMPORTANT IMPORTANT IMPORTANT IMPORTANT IMPORTANT
To experience unique
natural places
To experience unique
historic places
To do something enjoyable
with other people (for
example, family and friends)
To enjoy the vastness of
nature
To take part in outdoor
recreation, such as hiking,
fishing, or caaping
To have a change from ay
usual surroundings
Please list any other reasons
national parks:
you like to, or would like to, visit
2
-------
Q-6 Federal taxes are used to preserve and manage national park*. If you
personally could never visit a national park, would you want any of your
tax** spent to preserve and manage national parks? (Circle number)
1 DEFINITELY NO >
2 MAYBE NO
3 MAYBE YES
4 DEFINITELY YES
If definitely no. Skip to Question Q-8
Q-7 If you personally could never visit a national park in the future, how
important to you would each of the following reasons be to spend taxes to
preserve and manage national packs? (Circle number of best response for
each reason)
NOT AT ALL SLIGHTLY SOMEWHAT VERY EXTREMELY
IMPORTANT IMPORTANT IMPORTANT IMPORTANT IMPORTANT
So other members of my
family will have the
opportunity to visit these
areas now and in the future .1 2 3 4 5
So people outside my
family will have the
opportunity to visit these
areas now and in the future .1 2 3 4 5
So there will be areas
preserved in their natural
condition, even if no one
ever goes there
To allow scientific research
on nature or history 1 2 3 4 5
To preserve our national
heritage 1 2 3 4 5
So there is not
development everywhere 1 2 3 4 5
Do you have any other reasons? (Please list)
1 2 3 4 5
3
-------
ABOUT POLLUTION ISSUES FACING NATIONAL PARKS
Q-8 Below are sane types of effects that are happening or could happen in
national parks due to people's activities outside park boundaries. What
priority do you give to prevention of the following effects in national
parks due to human activities outside park boundaries? (Circle number of
best response for each effect)
LOW MEDIUM HIGH
PRIORITY PRIORITY PRIORITY
Air pollution decreasing the ability
to see scenic vistas 1 2 3
Air pollution injury to vegetation 1 2 3
Air pollution damage to historic
structures 1 2 3
Water pollution in streams or lakes
that harms fish or other aquatic
life 1 2 3
Water pollution that Buddies streams
or lakes but does not harm fish or
aquatic life 1 2 3
Park visitors being able to see or
hear mining or industrial activities
located outside park boundaries 1 2 3
Are there other types of effects of special concern to you? (Please list)
1 2 3
4
-------
ABOUT VISIBILITY IN AND AROUND NATIONAL PARKS
Throughout the U.S., air pollution from outside the parks cause* haze that
reduces how well a person can see in national parks and into scenic vistas
outside park boundaries.
The enclosed photographs show different levels of air pollution at three
national parks on days without rain or natural fog. The conditions at these
parks are typical of sumnertiae conditions at the national park* throughout
the region in which each park is located.
Photograph A shows almost no haze. This occurs on about 18 s uniaer days
each year (about 15* of the time).
Photograph B show* a little haze. This occurs on about 24 suamer days
each year (about 20% of the tine).
Photograph C show* average visibility condition*. Thi* occur* on about 48
suaner days each year (about 40% of the tiae).
Photograph D show* a lot of haze. Thi* occurs on about 30
each year (about 25% of the time).
r days
Q-9 If you were to visit a national park in each of these region*, you would
probably have average visibility like Photograph C. How do you think
having somewhat less than average haze due to air pollution, like
Photograph B rather than Photograph C, would affect your enjoyment of the
visit? (Circle number of best response for each region)
Having Visibility B Rather
Than Visibility C Would:
HAVE NO
EFFECT ON
ENJOYMENT
SOMEWHAT
INCREASE
ENJOYMENT
VERY MUCH
INCREASE
ENJOYMENT
California
(Use Yosenlte photos)
The Southwest
(Use Grand Canyon photos)
The Southeast
(Use Shenandoah photos)
5
-------
Q-10 Federal and state governments are considering changes to air pollution
control laws to protect and improve visibility in and around national
parks. These changes could affect everyone, even those who do not visit
the parks, because more air pollution controls could mean higher prices
for electricity, transportation, home heating, and for many other goods
and services, and could mean higher taxes.
How willing would you be to pay higher prices and taxes to support
visibility protection at national parks in the following regions?
(Circle number of best response for each region)
NOT AT ALL - SOMEWHAT EXTREMELY
WILLING WILLING WILLING
California 1 2 3 4 5
Southwest 1 2 3 4 5
Southeast 1 2 3 4 5
Please list any other regions of particular importance to you.
ABOUT VISIBILITY AT NATIONAL PARKS IN THE SOUTHWEST
0-11 The photographs for Grand Canyon National Park show typical visibility
conditions in the national parks of the Southwest. In the future,
visibility at these national parks could improve or worsen depending on
how much air pollution control is undertaken. How important do you think
the following goals are for protecting visibility at national parks in
the Southwest? (Circle number of oest answer for both goals)
NOT AT ALL SOMEWHAT EXTREMELY
IMPORTANT IMPORTANT IMPORTANT
Improving visibility at some or
all parks 1
Preventing visibility from getting
worse at some or all parks 1
6
-------
WHAT IS THE VALUE OF PROTECTING VISIBILITY
AT NATIONAL PARKS IN THE SOUTHWEST?
New air pollution controls being considered for the protection of visibility
at national parks in the Southwest could mean higher prices and higher taxes
throughout the country. The next questions concern how much obtaining
improvements and preventing worsening in visibility at national parks in the
Southwest would be worth to your household.
These questions concern only visibility at national parks in the Southwest and
assume there will be no change in visibility at national parks in other regions.
Other households are being asked about visibility, human health and vegetation
protection in urban areas and at national parks in other regions. For these
questions, assume you could be sure that any change would occur next year and
continue forever, and all households now and in the future would also pay the
most it is worth to them to protect visibility.
q-12 With additional air pollution controls, average visibility conditions in
and around all national parks in the Southwest could improve. What is the
most your household would be willing to pay every year in increased prices
and taxes to have average visibility improve from Grand Canyon Photograph C
to Photograph B at all national parks in the Southwest? (Circle best
answer)
$0.00
$2
$8
$25
$60
$150
$400
$0.50
$3
$10
$30
$75
$200
$500
$1.00
$4
$15
$40
$100
$250
$750
$1.50
$5
$20
$50
$125
$300
MORE THAN
Q-13 What is the most your household would be willing to pay every year in
increased prices and taxes to have average visibility improve from Grand
Canyon Photograph C to Photograph A at all national parks in the Southwest?
(Circle best answer)
$0.00
$2
$8
$25
$60
$150
$400
$0.50
$3
$10
$30
$75
$200
$500
$1.00
$4
$15
$40
$100
$250
$750
$1.50
$5
$20
$50
$125
$300
MORE THAN
7
-------
Q14 It is also possible that some additional air pollution controls may be
needed just to keep visibility at national parks in the Southwest from
getting worse. What is the most your household would be willing to pay
every year in increased prices ana taxes to prevent average visibility at
all national parks in the Southwest from becoming like Photograph D for
Grand Canyon rather than like Photograph C. (Circle best answer)
$0.00
$2
$8
$25
$60
$150
$400
$0.50
$3
$10
$30
$75
$200
$500
$1.00
$4
$15
$40
$100
$250
$750
$1.50
$5
$20
$50
$125
$300
MORE THAN
Q-15 Please provide any information that helps explain your answers to Questions
12, 13, and 14 above. You may also use the back page of the questionnaire.
Q-16 We understand it may be difficult to determine the most you are willing to
pay for changes in visibility at national parks. Would you say your
answers to Questions 12, 13, and 14 are: (Circle number of best answer)
1 VERY ACCURATE?
2 WITHIN THE BALLPARK?
3 SOMEWHAT INACCURATE?
4 PROBABLY VERY INACCURATE?
8
-------
Q17 Would you say the dollar amounts you gave in answer to Questions 12, 13 and
14 are: (Circle number of best answer)
1 BASICALLY FOR THE STATED CHANGES IN VISIBILITY AT THE NATIONAL PARKS
j2 SOMEWHAT FOR THE STATED CHANGES IN VISIBILITY AND SOMEWHAT TO HELP WITH
OTHER NEEDS AT THE NATIONAL PARKS
3 BASICALLY TO HELP THE NATIONAL PARKS AND ARE NOT RELATED TO THE STATED
CHANGES IN VISIBILITY
4 OTHER (Please specify)
About what percent of your dollar answers is for visibility at
national parks in the Southwest? (Circle number)
NONE SOME HALF MOST ALL
0% 10% 20% 30% 401 501 60% 70% 801 90% 100%
Q-18 Of the money you would be willing to pay to control haze in and around
national parks in the Southwest, about what percent do you think should
be spent to control haze in and around Grand Canyon National Park?
(Circle best answer)
NONE SOME HALF MOST ALL
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Q-19 About what percent of the dollar amount you stated you would be willing
to pay for improving visibility conditions in and around national parks
in the Southwest can be explained by the following reasons? (Answers
should total to 100%)
8 SO MY HOUSEHOLD AND I COULD ENJOY CONDITIONS AS NATURAL AS
POSSIBLE ON VISITS TO NATIONAL PARKS IN THE SOUTHWEST
8 SO OTHERS, NOW AND IN THE FUTURE, COULD ENJOY CONDITIONS AS
NATURAL AS POSSIBLE ON VISITS TO NATIONAL PARKS IN THE SOUTHWEST
8 TO HAVE CONDITIONS AS NATURAL AS POSSIBLE AT NATIONAL PARKS IN
THE SOUTHWEST, EVEN IF NO ONE WERE TO EVER VISIT
% OTHER (please explain)
"=T5r%
9
-------
ABOUT YOU AND YOUR HOUSEHOLD
Q-20 "Where do you live? (Circle number of best answer)
1 A LARGE METROPOLITAN AREA (Over 1 million people)
2 A LARGE CITY (100,000 to 1 million people)
3 A SHALL CITY OR TOWN (10,000 to 100,000 people)
4 A VERY SMALL TOWN OR RURAL AREA (under 10,000 people)
Q-21 How far do you live from the nearest national park? (Circle number)
1 LESS THAN 50 MILES 4 200-499 MILES
2 50-99 MILES 5 500-999 MILES
3 100-199 MILES 6 1000 MILES OR MORE
7 I DON'T KNOW
Q-22 What is the name of the nearest national park?
Q-23 Your sex? (Circle number)
1 MALE 2 FEMALE
Q24 Your present age?
YEARS
Q-25 Are you presently? (Circle number of best, answer)
1 EMPLOYED FULL-TIME 4 UNEMPLOYED
2 EMPLOYED PART-TIME 5 RETIRED
3 FULL-TIME HOMEMAKER 6 STUDENT
Q-26 Including yourself, how many members of your household are in each
age group? (If none, write "0")
UNDER 18 YEARS Of AGE
18 TO 64 YEARS OLD
65 YEARS AND OVER
Q-27 How much formal education have you completed? (Circle number)
1 NO FORMAL EDUCATION
2 SOME GRADE SCHOOL
3 COMPLETED GRADE SCHOOL
4 SOME HIGH SCHOOL
5 COMPLETED HIGH SCHOOL
10
SOME COLLEGE OR TRADE SCHOOL
COMPLETED TRADE SCHOOL
COMPLETED COLLEGE
SOME GRADUATE WORK
ADVANCED COLLEGE DEGREE
Q-28 What was the approximate annual gross income (before taxes) received
in 1987 by you and family members living with you? (Circle number)
1 UNDER 510,000
2 $10,000-19,999
3 $20,000-29,999
4 $30,000-39,999
$40,000-49,999
$50,000-59,999
$60,000-69,999
$70,000-79,999
9 $80,000-89,999
10 $90,000-99,999
11 MORE THAN 5100,000
12 CHOOSE NOT TO ANSWER
10
-------
Is there anything we may have overlooked? Please use this space for additional
comments you would like to make about managing natural resources and visibility
at national parks.
Please return questionnaire to:
Managing The Parks
RCG/Hagler, Bailly
P.O. Drawer O
Boulder, Colorado 80306-1906
Your help is greatly appreciated. If you wish to receive a summary of results,
print "results requested" on this page. We will see that you receive it.
SW3
-------
APPENDIX B
Sample Letters
RCG/Hagler. Bailly. Inc.
-------
RCG/Hagler, Bailly, Inc.
P. 0. Drawer 0
Boulder, Colorado 80306-1906
303/449 5515 ¦ Fax: 303/443 5684
August 22, 1988
Dear
The national parks have been set aside as special resources.
Yet, the management and protection of national parks involve
costs to each of us. Therefore, decisions about how to protect
and manage the parks should consider the opinions of all people
in the country.
I have a favor to ask of you. In about a week you will receive
in the mail a questionnaire and color photographs of national
parks. The questionnaire asks about your visits to national
parks and asks your opinions about managing and protecting
national park resources. Your opinions are important, even if
you do not visit national parks.
Because we can send only a few questionnaires, we have
scientifically selected households to reflect the opinions of
citizens from around the country. Your response is very
important.
RCG/Hagler, Bailly, Inc. is a professional research firm hired to
help conduct this study for the University of Colorado. You will
receive your questionnaire from us along with a postage paid
return envelope. The results will be provided to the National
Park Service, and to all other interested parties.
Thank you in advance for any help you can provide.
Project Manager
Management. Economic, and Technical Counsel
¦
Washington - New York - Cambridge - Boulder - Brussels
-------
RCG/Hagler, Bailly, Inc.
P.O. Drawer O
Boulder, Colorado 80306 1906
303/449 551 S^fax: 303T443 5684
September 6, 1988
Dear
Here is the questionnaire I told you about in my letter a few days
ago. People have filled it out say it takes about 20 to 25
minutes to complete (sometimes more, sometimes less). Your responses
will help provide an understanding of what all citizens want and
don't want with regard to protection of the resources at national
parks.
Your questionnaire should be filled out by either the male or female
head of household. Your response is very important because you are
part of a relatively small sample of people from around the country
who have received this survey. The questions do not require
scientific knowledge, only that you consider and answer each question
as well as you can.
Your response will be confidential. Results will only be reported
statistically, such as "20% have visited Yosemite National Park." The
enclosed form has a number for mailing purposes only, so we may check
your name off the follow-up mailing list when you return the
questionnaire to us.
The results will be made available to the National Park Service and t
all other interested parties. If you would like, we will send you a
summary of the results. Simply write "results requested" on the back
page of the survey and I will see that you get them.
RCG/Hagler, Bailly has been hired to help conduct this survey, so your
completed questionnaire should be sent directly to our office. A
postage paid, self-addressed envelope is enclosed for your
convenience. I will be happy to answer any questions you might have.
Please write or call. We appreciate your assistance.
Project Director
P.S. Since we know that your time is valuable, we offer the enclosed
gift as a token of our appreciation for your help.
Management, Economic, and Tecftrur ~i Counsel
Washington New Cambridge Boulder Brussels
-------
September 7, 1988
Last week a questionnaire was mailed to you seeking your
opinions about the preservation and management of the
national parks. Your name was drawn from a scientific
random sample of U.S. citizens.
If you have already completed the questionnaire and
returned it to us, please accept our sincere thanks. If
not, please do so today. Because it has been sent to only
a small number of households around the country, it is
extremely important that your opinions be included in the
study to accurately represent the opinions of all citizens
nationwide.
If you did not receive the questionnaire, or if it got
misplaced, please call me collect at (303) 449-5515, and I
will put another one in the mail to you today.
Robert D. Rowel/
RCG/Hagler, Bailly, Inc.
Project Director
-------
RCG/Hagler, Bailly, Inc.
P.O. Drawer O
Boulder, Colorado 80306 -,1906
303/449 5515 Fax: 303/443 5684
September 23, 1988
Dear
Three weeks ago I wrote to you asking for your opinions about
the protection and management of national parks. As of today,
we have not received your completed questionnaire.
The University of Colorado and RCG/Hagler, Bailly, Inc. are
conducting this research to help decision makers in government
and industry better understand how citizens like you want the
air quality resources in and around national parks to be managed.
I am writing to you again because the opinions of each household
selected are important to this study. For the results to be
truely representative of households from around the country, it
is essential that each person in the sample return the
questionnaire. Even if you feel that you know very little about
air pollution at national parks, your opinion is valuable and
still counts.
In the event that our questionnaire was misplaced, a
replacement is enclosed. If you should have any questions,
please call me collect at (303) 449-5515.
Your assistance is greatly appreciated.
Sincerely,
/ /
Robert'D. Rowe //
Project Manager '
Management, Economic, and Technical Counsel
¦
Washington New York Cambridge Boulder Brussels
-------
RCG/Hagler, Bailly, Inc.
P O Drawer O
Boulder. Colorado 80306-1906
303/449 5515 ¦ Fax. 303/443 5684
November 2, 1988
Dear
Some weeks ago I wrote asking for your help on a survey concerning visibility
at National Parks. So far, we have not received your completed survey.
Several people have called me saying they will probably never visit the parks
we asked about. Even if you will never visit these parks, decision makers in
government must learn if you feel it is important or unimportant to protect or
improve visibility at national parks and why you feel this way. The decisions
they make can affect the taxes and prices you pay, regardless of how often you
visit the parks.
Other people have told me some of the questions are hard to answer. These are
also hard questions for decision makers to answer without your input. We are
asking you to answer the best you can. if you feel your answers are
inaccurate, please tell us by writing this next to your answers.
Because you are part of a small group scientifically selected to represent all
of the public's opinions, your response is very important to us. In case you
have misplaced your survey, we have enclosed another. If you have any
questions, please call me, collect, at (303) 449-5515. As I have indicated,
all your responses will be held strictly confidential.
Sincerely,
Project Manager, "Managing the Parks"
RDR; jlw
Management. Economic, and Technical Counsel
Washington - New York - Cambridge - Boulder - Brussels
-------
APPENDIX C
Sample Telephone Follow-up Survey Instrument
RCG/Hagler, Bailly, Inc.
-------
PARVU PHONE FOLLW-UP Version 2
Focus Region: Grand Canyon/Southwestern U. S.
Codes: 8 , 88 = Don't Know; 9,99 = Missing or Refused
Hello, is this ? My name is , and I'm calling from CIC Research.
We're working with the University of Colorado.
Several weeks ago we sent you a printed survey about visibility at the national
parks. I'm calling as part of a short telephone follow-up to the mail
questionnaire (but this is not the mail questionnaire.) This follow-up is
being done to find out how people who did not respond to the mail survey differ
from those who completed the survey. This information is important to help us
understand the responses to the mail survey. We would appreciate if you could
answer a few questions that will take about 5 minutes.
ID
1-3
Q1 It this a convenient time to talk?
NO
YES
Q2 Have you ever personally visited any national park, (including
monuments and seashores)? 1 NO 2 YES
u
Skip to Q4
Q3 Have you personally ever visited Grand Canyon National Park?
1 NO 2 YES > (If Yes) How many days have you
visited this park in just the
last two years? DAYS
AGREE
VISIT
VYOS
DYOS
Q4 In the next 5 years, do you think you "Definitely Will
Visit", "Might Visit", or "Definitely Will Not Visit" a
national park in the southwestern United States?
Examples of parks in the Southwest are Bryce Canyon,
Zion, Saguaro, Petrified Forest, and Carlsbad Caverns
DEFINITELY WILL
REFUSED
MIGHT 5_
DON'T KNOW
WILL NOT
FVISIT
1
-------
Q5Do you have the national park photographs we mailed you?
1 NO > Skip to Q9.
2 NO, BUT I REMEMBER THEM. > Skip to Q9.
3 YES > I would like to ask you a few short
questions about the photographs. Could
you get them?
1 NO ? YES
In the future, visibility at national parks of the southwestern U.S.
could improve or wor_ -n, depending upon how much air pollution
control is undertaken. But any changes in air pollution control
could affect everyone through higher prices for electricity,
transportation, home heating and for many other goods and services
or through higher taxes.
YES, THEY HAVE PHOTOS TO LOOK AT
The next questions concern only the pictures for Grand Canyon,
which are typical of conditions at all national parks in the south,
western U.S. The typical summer visibility is like Photograph C.
Q6 Do you think it is "Not At All Important", "Somewhat
Important", or "Extremely Important" to prevent visibility
from getting worse, like Photo D, at national parks in the
Southwest?
1 NOT AT ALL ^ SOMEWHAT 5 EXTREMELY
Q7 Would you support new air pollution controls to
prevent typical summer visibility from decreasing from
Photo C to Photo D at all national parks in the Southwest
if it cost your household $25 a year?
1 NO 2 YES > Skip to Q8.
Would you support new air pollution control regulation to
prevent typical summer visibility from decreasing from
Photo C to Photo D at all national parks in the soutwest
if it cost your household $5 a year?
1 NO 2 YES
$ (Provided, if any)
(Why/why not? If provided)
HAVE
10
GET
11
WORS1
12
WTP1
13
WTP2
14
WTP1S
15-1
WHY
18-19
2
-------
Q8 Do you think it is "Not At All Important", "Somewhat
Important", or "Extremely Important " to improve visibility so
it should be like Photo B, at national parks in the Southwest?
1
NOT AT ALL
3
SOMEWHAT
EXTREMELY
Q9 Would you support new air pollution controls to
increase typical summer visibility from Photo C to
Photo B at all national parks in the Southwest if it cost
your household $25 a year?
IMPR1
20
NO
_2_
YES
(provided, if any)
WTP$3
WTP2$
21
22-24
Skip to Q14.
NO, THEY DO NOT HAVE PHOTOS
Q10 Do you think it is "Not At All Important", "Somewhat
Important", or "Extremely Important" to prevent visibility
from getting somewhat worse at national parks in the south-
western United States?
Examples of parks in the Southwest are Bryce Canyon,
Zion, Saguaro, Petrified Forest, and Carlsbad Caverns.
NOT AT ALL
SOMEWHAT
EXTREMELY
Qll Would you be willing to pay any more in increased prices
or taxes to support new air pollution controls that would
prevent typical visibility from becoming somewhat worse
at all national parks in the Southwest?
1
NO
2
YES
3
Don't Know
Q12 Do you think it is "Not At All Important", "Somewhat
Important", or "Extremely Important" to improve visibility
at national parks in the Southwest?
WORS2
25
WTP4
26
1
NOT AT ALL
3
SOMEWHAT
5
EXTREMELY
IMPR2
27
3
-------
Q13 Would you be willing to pay any more in increased prices
or taxes for new air pollution controls that would somewhat
improve typical visibility at all national parks in the south-
west?
_L
NO
2
YES
3
Don't Know
Now, just a few last questions to help group your responses with
those of others:
Q14 Do you live in:
1 A LARGE METROPOLITAN AREA of Over 1 million people
2 A LARGE CITY of 100,000 to 1 million people
3 A SMALL CITY OR TOWN of 10,000 to 100,000 people
4 A VERY SMALL TOWN OR RURAL AREA of Under 10,000 people
Q15 How far do you live from the nearest national park? Would
you say it is:
1 LESS THAN 50 MILES
2 50-99 MILES
3 100 - 199 MILES
(7 I DON'T KNOW)
4 200 - 500 MILES
5 500 - 1,000 MILES
6 MORE THAN 1,000 MILES
Q16 What is the name of the nearest national park?
Q17 What is your present age?
Q18 Including yourself, how many people are in your household?
Q19 What was your total household income in 1987 before taxes
and deductions? Would it be:
1 - UNDER $2 0,000
2 - BETWEEN $2 0,000 AND $4 0,000
3 - BETWEEN $40,000 AND $60,000
4 - BETWEEN $60,000 AND $80,000
5 - OVER $8 0,000
6 - REFUSED TO ANSWER
Thank you. That's all the questions I have,
WTP5
28
WHERE
29
MILES
30
PARK
AGE
31-33
34-35
HH
36-37
INC
38
4
-------
Interviewer, add:
Q18 Sex: 1 - MALE
2 - FEMALE
Q19 Comment:
Q2 0 Interviewer number
Q21 Language
0 NO LANGUAGE BARRIER
¦ 1 POSSIBLE LANGUAGE BARRIER
2 DEFINITE LANGUAGE BARRIER
Q22 Other Comments:
SEX
39
COMM1
40-41
INT #
42
LANG
43
COMM2
44-45
5
------- |