NCEE#

NATIONAL CENTER FOR

ENVIRONMENTAL ECONOMICS

Determining the Extent of Market and Extent of Resource for
Stated Preference Survey Design Using Mapping Methods

Shalini P. Vajjhala, Anna Mische John, and David A. Evans

Working Paper Series

Working Paper # 08-09
October, 2008

£

M'i

1 *

U.S. Environmental Protection Agency
National Center for Environmental Economics

5	"	1200 Pennsylvania Avenue, NW(MC 1809)

\	Washington, DC 20460

¦h

pro^^	http://www.epa.gov/economics


-------
Determining the Extent of Market and Extent of Resource for
Stated Preference Survey Design Using Mapping Methods

Shalini P. Vajjhala, Anna Mische John, and David A. Evans

NCEE Working Paper Series

Working Paper # 08-09
October, 2008

DISCLAIMER

The views expressed in this paper are those of the author(s) and do not necessarily represent
those of the U.S. Environmental Protection Agency. In addition, although the research described
in this paper may have been funded entirely or in part by the U.S. Evironmental Protection
Agency, it has not been subjected to the Agency's required peer and policy review. No official
Agency endorsement should be inferred.


-------
Determining the Extent of Market and Extent of Resource for
Stated Preference Survey Design Using Mapping Methods

Shalini P. Vajjhala, Anna Mische John, and David A. Evans*

Abstract

Determining the appropriate survey population and the commodity to be valued are
among the most fundamental design decisions for stated preference (SP) surveys. However, often
little information is available about who in the population holds measurable value for the resource
(the extent of the market) and their perceptions regarding the scope of the resource to be valued
(the extent of the resource). In this paper, we present a novel approach using cognitive mapping
interview techniques to shed light on these design questions. The method also provides ancillary
information that assists in the interpretation of information collected during focus groups and
through SP survey administration. The approach was developed and tested as part of an ongoing
study on environmental degradation associated with acidification in the Southern Appalachian
Mountain region. Although damage from acidification in the study region is broad, it is not clear
whether residents of this region care, in both a use and nonuse sense, about resources in their
states of residence, in neighboring states, on public lands, or more broadly across the region.

From a pilot study, we found that participants show a significant home-state preference in the
number and size of natural areas that they value within the larger Southern Appalachian Mountain
region. However, this preference is not strong enough to suggest that the market for improving
these resources is solely constrained to residents of the state in which the resource is located.

Key Words: Stated preference, cognitive mapping, extent of market, extent of resource,
definition of commodity

Subject Matter Classifications: Valuation Methods, Existence/Nonuse Values, Economic
Damages/Benefits

* Shalini Vajjhala is a fellow and Anna Mische John is a research assistant at Resources for the Future,
1616 P St. NW, Washington, DC 20036; David Evans is an economist at the U.S. Environmental Protection
Agency's National Center for Environmental Economics. Corresponding author: Shalini P. Vajjhala, (202)
328-5129, shalini@rff.org. This research was funded in part by EPA Science To Achieve Results Grant
RD-832422. The opinions expressed here are those of the authors and do not necessarily express the views
of the EPA. We thank Laura Chappell-Campbell and Susie Chung for their valuable research assistance and
comments. We thank Spencer Banzhaf, Dallas Burtraw, Alan Krupnick, Juha Siikamaki, David Simpson,
Matt Massey, and Joel Corona for their comments.


-------
Contents

1.	Introduction	1

2.	Spatial Design Decisions in SP Surveys	5

The Extent of the Market	6

The Extent of the Resource	9

3.	An Introduction to Mapping	11

4.	Study Design and Methodological Approach	14

5.	Acidification in the Southern Appalachians: An Application	18

Survey Population	21

Interview Process	22

6.	Maps, Data Analysis, and Study Results	25

The Extent of the Market	28

The Extent of the Resource	35

7.	Application of Mapping Results to SP Survey Design	37

8.	Conclusions	38

References	40

Appendix A: Mapping Interview Protocol	45

Appendix B: Debriefing Survey	49


-------
Determining the Extent of Market and Extent of Resource for
Stated Preference Survey Design Using Mapping Methods

Shalini P. Vajjhala, Anna Mische John, and David A. Evans
1. Introduction

Identifying the appropriate survey population and defining the resource to be
valued are among the most fundamental design decisions for stated preference (SP)
surveys. However, a researcher does not necessarily know the spatial distribution of those
who hold measurable value for the resource (the extent of the market) or the part of the
resource worth focusing on (the definition of the commodity or extent of the resource).1
Limited sampling resources preclude casting a large net and capturing every individual or
household that may value the resource in question. Furthermore, the information
available on the perceptions of the general population regarding the relevant scope of the
resource to be valued is often limited. These challenges are particularly true for resources
associated with significant nonuse values. These drivers, along with the desire for a
credible payment vehicle and the preferences of the survey sponsor, often result in the
use of convenient, implicit, or ad hoc definitions of the extent of the market and the
extent of the resource in SP surveys.

In this paper, we present a novel approach using cognitive mapping interview
techniques from the geography and psychology literature as a complement to traditional

1 We use the terms extent of the resource and definition of the commodity interchangeably to refer to the
geographic boundaries of a resource that could experience a relatively small change in quality (the
commodity described in an SP survey).

1


-------
focus group interviews for the development of SP surveys that assess willingness to pay
(WTP) for reducing environmental damages over large regions. The goal of this exercise
is both methodological—in that it demonstrates how to incorporate mapping into SP
studies—and applied. We develop and test the approach as part of an ongoing study on
environmental degradation associated with acidification in the Southern Appalachian
Mountains (SAM) (Krupnick 2004; Evans et al. 2008). Because the SAM region covers
parts of eight different states, it does not have clear jurisdictional boundaries that
correspond with the relevant ecological and economic ones. Further, even though
damages from acidification in the study region are broad, it is not clear how concerned
residents of this region are about improvements to degraded resources in their states of
residence or in neighboring states (extent of the market). In addition, it is also unclear if
residents care whether improvements are made on specific public lands or more broadly
distributed across the region (extent of the resource). This mapping study is designed help
answer these questions by informing survey design choices about the subpopulation(s) to
be interviewed and the resource(s) for which changes should be presented in the
valuation exercise.

Current methods to define the extent of the market and the extent of the resource
typically rely on convenient administrative boundaries and/or sample populations. These
approaches could fail to adequately address ambiguity surrounding the definition of the
market or the resource in the case of very large resources where individuals' prior
perceptions are poorly understood. We discuss the shortcomings of current approaches in
detail in the next section.

2


-------
In this study, we focus on the practical challenge posed by study constraints that
preclude broad survey strategies to learn the extent of the market for goods with
important nonuse values. In these cases, one could use an opened ended approach to
solicit unbiased preferences from individuals in a broad geographical area to learn the
extent of the market (e.g., "Tell me the natural resources you care about and where they
are."), but without careful framing this would be an inefficient exercise. Alternatively,
one could ask these individuals if they cared about improvements to a particular resource,
either through focus groups or a brief pilot SP survey. However, such an approach would
likely solicit either uninformative confirmations or require the costly development of a
credible and incentive compatible survey. Furthermore, one could not easily apply this
approach determining an appropriate extent of the resource for a survey addressing
changes over a broad area of interest, particularly when ambiguity of the extent of the
market and resource are confounded. We argue that mapping is a useful intermediate
method for identifying an appropriate extent of both the market and resource. A mapping
approach is more likely than these other approaches to determine whether preferences are
provincial in nature as it does not require the description of a particular payment or
management method to determine which resources respondents care about, but it offers
useful framing and limits the number of areas that a respondent can identify as those they
care about.2

2 Schlager and Ostrom (1992) highlight a similar challenge in the context of property rights regimes for
common property resources (using the example of inshore fisheries). They emphasize that research
definitions of property rights and the rules for their enforcement "shape perceptions of resource degradation
problems and the prescriptions recommended to solve such problems." They further note that ambiguity in

3


-------
Using a semi-structured format, our mapping protocol begins by providing study
participants with a base map of the region in a one-on-one interview format. Participants
are asked to add information to the maps about their use of the region and indicate their
preferences over particular areas or resources within the region. Respondents indicate
features including 1) places that they visited regularly while living in the region, 2) the
natural areas in the region that they cared about and/or were important to them, and 3)
any areas they perceived as degraded. The maps resulting from each interview are then
coded (in a process similar to transcribing an interview) to develop descriptive statistics
and allow for quantitative comparison of the sizes, types, and locations of the areas or
resources marked on participants' maps.

This information can be used to discern which natural areas are most salient to
individuals from different parts of the larger region and to characterize variations in the
sizes and locations of those areas. The mapping method is not a valuation exercise that
yields a cardinal measure that can be used in benefit-cost analysis; instead, the protocol
is designed to generate informative ordinal rankings of preferences that can be used to
inform the survey design. Furthermore, the mapping exercise does not require a large
information treatment or the use of a relatively inefficient (but unbiased and precise)
choice question for measuring value. In addition, because the protocol allows respondents
to report nonuse values, it is also preferable to relying on extent data describing resource
use. By providing spatial characterization of both the extent of the market and the extent

the understanding and definition of institutional arrangements surrounding a natural resource could
undermine the "analytical or prescriptive clarity" of research studies.

4


-------
of the resource, this approach demonstrates how mapping can inform the design of SP
surveys of both contingent valuation and conjoint forms.

In the next section, we review the economics literature on determining the extent
of the market and defining the commodity in an SP study, and describe in detail the
motivation for using mapping analysis as an introductory component of SP survey
instrument design. We then provide a review of the mapping literature and outline the
potential contributions of this method to the economics literature and to SP research
(Section 3). Next, we detail the elements of our methodological approach and mapping
study design (Section 4). We then present a pilot application evaluating the extent of the
market and the extent of the resource for damages from acidification in the SAM region
(Section 5), and highlight our analyses and findings and compare our results to the extent
of the market suggested by a well-known national recreational use survey (Section 6).
Finally, we conclude with a discussion of the strengths and weaknesses of the method for
wider application to different resources and regions, highlighting several areas for further
study (Sections 7 and 8).

2. Spatial Design Decisions in SP Surveys

Two parallel questions in SP survey design are (a) how to determine the extent of
the market (and thus whom to survey) and (b) how to define the commodity of interest.
These decisions must be made jointly because a respondent's preference concerning a
resource could depend on its size, quality, and location. In contrast to the question of the
appropriate extent of the market, the definition of the appropriate commodity receives
relatively limited treatment in the SP literature and is discussed largely as a design feature

5


-------
that is at the discretion of the researcher.3 We review each of these design questions
separately in this section, and consider issues related to their joint determination in our
discussion of the pilot application.

The Extent of the Market

Well-known studies have demonstrated the importance of the choice of the
sample population in estimating total WTP (Smith, 1993a). Nevertheless, reviews of SP
design methods provide little detail on how to identify the relevant population from
which to estimate WTP other than to encourage the researcher to carefully consider the
issue (Bateman et al. 2002; Champ and Welsh 2007; Freeman 2003; Mitchell and Carson
1989). The limited guidance provided is often simply a suggestion to consult data from
existing surveys that encompass the entire potential population of interest. Such data are
unavailable in many cases and could prove to be unreliable when extrapolated to different
applications in other surveys. To demonstrate this, we show how the mapping study
reveals significantly greater interest in a resource than is indicated by a large-scale
recreational use survey. This suggests that the mapping survey can be used to identify the
presence of significant non-use (or perhaps latent use) values.

Loomis (2000) summarizes the issues surrounding the determination of the
relevant population for an SP study. His starting point is the well-known Samuelson

3 The importance of the definition of the commodity has received some attention regarding the effects of
the completeness of its description in SP surveys (Boyle 2003) and considerable treatment in analyses of
scope responsiveness (Smith and Osborne 1996). However, here we are interested in it as a survey design
choice.

6


-------
condition for the optimal level of a public good. But the Samuelson condition alone is an
insufficient guide because individual use and existence values for different environmental
resources are not confined to geographic or political boundaries, and almost anyone could
conceivably have some value for a resource in question. This is particularly true for
resources with significant nonuse benefits, given that physical proximity is relatively
unimportant for their enjoyment.4 Also, our interest in nonuse values essentially
precludes the use of traditional methods of identifying the boundaries of a market by
exploiting changes in the prices of substitutes and complements (Whinston 2007; Smith
1993b).

Loomis (2000) further identifies a number of SP studies that use a convenient
population sample where the extent of the market is assumed to be generally proximate to
the resource itself. Often such a convenient design choice is made for good reasons. A
researcher typically has little information about who values the resource ex ante and
therefore develops a definition of the extent of the market (at least for sampling purposes)
based on other objectives and study constraints. For example, Banzhaf et al. (2006)
define their survey population as all New York State households to allow for the use of
an incentive-compatible payment vehicle (state income tax) that corresponded with a
credible management agency (New York State) for the resource in question (ecosystem

4 More limiting is the legal question of who has standing with respect to managing the commodity in
question and who might have to pay for that commodity. But such legal distinctions do not provide a guide
in many WTP estimates associated with benefit-cost analysis. See Schlager and Ostrom (1992) for
examples of overlaps in management of common property natural resources.

7


-------
quality in the Adirondack Park).5 Although New York State residents are likely to have
considerable WTP for improvements to this resource, residents of neighboring states are
also likely to receive similarly significant benefits from improvements to the resource.
Given the sampling budget and the design complexities associated with surveying a
larger population, it was decided that estimating the WTP of the residents of neighboring
states was outside the scope of the study.

In his empirical analysis, Loomis (2000) demonstrates that an important bias may
be introduced by limiting the geographical extent of the market to convenient
jurisdictional boundaries, particularly when considering nonuse values (see also Loomis
1996). Mitchell and Carson (1989) call this problem population choice bias. Others have
found a significant reduction in WTP with increasing distance from the resource,
indicating the importance of the assumed choice of the spatial extent of the market for
surveying purposes (Banzhaf et al. 2006; Johnson et al. 2001; Bateman et al. 2006;
Hanley et al. 2003). Many of these investigators also found a discrete drop in the distance
gradient of WTP at the political boundary of the jurisdiction in which the resource lies.
Note that these studies have the benefit of using data from an SP survey cast over a broad
area; the drop in WTP found by these studies could be the result of provincial preferences
for resources and may, therefore, stand as a true description of WTP. Alternatively, the

5 Although this study only estimated the WTP for improvements to the Adirondacks from reduced acid
deposition from New York State residents, the information gathered by the survey is still relevant when
conducting a benefit-cost analysis for the reduction of acid deposition precursors. The total WTP of all
New York residents can be viewed as a lower bound of the total value of the improvement to the resource.

8


-------
drop could be caused by a lack of credibility of the payment vehicle or the described
method of resource management.

Overall, studies that find a WTP price gradient over space have the advantage of
a pre-existing survey. In the absence of such data, the mapping approach can be used ex
ante to improve the researcher's understanding of the potential extent of the market for
focus group testing, survey construction, and sampling protocol, and also to avoid
potential pitfalls like the sampling bias that Loomis (2000) describes. Furthermore, by not
being bound to a particular resource management regime, the open-ended approach of the
mapping protocol could allow the researcher to identify a population that may benefit
from the policy intervention which might otherwise be hidden by other components of an
SP survey.

The Extent of the Resource

Most SP studies take the definition of the commodity, which technically is the
change in the quality of a particular resource, as given. This is typically a sensible
approach as the motivation for conducting the survey is to understand the economic
welfare consequences of the changes resulting from a particular course of action.
However, those changes may be realized over a large geographical area, and describing
the entire extent of the change could be impractical or undesirable for a particular survey.
For example, the boundaries of the natural resource in question might not correspond to
the administrative and jurisdictional boundaries under which resource management

9


-------
decisions are made.6 Thus, other elements of the survey design, such as the choice of a
credible payment vehicle, may be compromised by the use of an excessively broad
geographic description of the resource.7 In these cases, the best option could be to
partition the resource into components and have separate surveys eliciting WTP for the
improvements to each component.8 For the purposes of this study, defining the extent of
large resources refers to formally evaluating (a) whether potential survey participants
identify with the resource in its entirety or with specific features or subregions within the
resource, (b) where they perceive the boundaries of the resource to be within the larger
region, and (c) how these boundaries compare with other established political or
geographic boundaries, such as state or national park borders.

Even with a particular geographic description of a resource, the extent to which
individuals are able to identify with very large-scale resources is unclear. With very large
resources, research has shown significant embedding effects, where respondents to SP

6	Overlaps in management by different state and federal authorities could mean that a specific resource is
encompassed by a jurisdictional boundary at one level of government but is managed by a higher level of
government. As described below, this is the case with respect to measuring WTP for reductions in acidic
deposition in the SAM region. See Schlager and Ostrom (1992) for a detailed discussion of the different
bundles of management rights (among other types of rights) associated with common property natural
resources.

7	When we refer to a credible payment vehicle, we mean one that, in addition to providing an incentive to
reveal one's true WTP for the resource improvements described in the survey, is associated with a credible
method of managing and funding the program that achieves those improvements.

8	In this case it may even be worthwhile to administer each survey to the subpopulation that finds changes
to the particular component of the resource salient. However, such an approach may come at the cost of
statistical power in estimating WTP bounds or for survey reliability and validity tests. More fundamentally,
one might object to partitioning the entire commodity this way (whether or not a common population is
used) as the qualities of neighboring or similar resources may affect the WTP for a particular change in
another resource. This problem is common to most methods for valuing changes that result from policies
that affect a large region. The mapping method we describe in this paper provides a sense for how
important partitioning the commodity may be to estimating the welfare effects of such policies.

10


-------
surveys are incongruently willing to pay the same amount for improvements to a large
resource as they are for improvements to a smaller component of the same resource.
Fischhoff et al. (1993) describe several different methodological reasons for these effects
and outline strategies for overcoming such biases. Problems can arise if respondents
consider only a limited subregion that they associate with the larger resource or if they
cognitively truncate a larger area based on their prior experiences and perceptions
(Fischhoff et al. 1993). Therefore, even if the extent of the resource is predetermined by
other considerations of the study, survey respondents' prior perceptions of the commodity
must be understood to ensure correct interpretation and evaluation of the reliability of
WTP responses from the survey.

3. An Introduction to Mapping

Mapping is a technique widely applied and tested both in geography and
psychology literature that has the potential to reduce ambiguity about the extent of the
market and the extent of the resource in valuation research. In psychology, the process of
cognitive mapping has been examined over many decades. Beginning in the 1940s with
Tolman's (1948) landmark study that first recognized the term cognitive mapping, both
geographers and psychologists have conducted experiments and studies to understand
how individuals perceive different types and scales of spaces and to characterize
systematic biases and distortions in map representations (Tversky 1981, 1991, 1992). The
approaches of these two fields to mapping can be distinguished primarily by the focus on
internal maps in psychology versus external maps or representations of spaces in
geography (Downs and Stea 1973, 1977; Golledge andZannaras 1973; Golledge 1976).

11


-------
More recently, these fields have coalesced around research on defining the
theoretical underpinning of digital mapping tools, such as Geographic Information
Systems (GIS), to develop and evaluate whether the tools are responsive to how
individuals navigate and think about spaces (Kaplan 1976; Tversky 1993; Mark and
Frank 1996). Although significant research has been conducted on variations in spatial
perceptions and comprehension at different scales—from the very small (a single room)
to the very large (continent-level)—little spatial cognition research to date has focused on
perceptions of large-scale natural resources (Klett and Alpaugh 1976; Evans et al. 1981;
Brown 2004). For this reason, we bring together elements from both the geography and
psychology literature on natural resources to develop an interview method to elicit a
cognitive map of a large, natural environment.9

The method applied here extends traditional participatory mapping techniques
using a semi-structured interview format to elicit survey respondents' individual maps of
a region and their perceptions and preferences over particular areas or resources within a
region (Lynch 1960; Chambers 1994); thus, the method provides a quantifiable
justification for follow-on SP instrument design. This method is proposed for determining
which areas are valued within a larger, shared natural resource and for identifying any

9 Cognitive maps, cognitive collages, participatory maps, and spatial mental models are all terms that have
been used in the geography and psychology literature to describe maps (spatial representations) created
using interview- or survey-based methods for eliciting and recording spatial data. Specific examples
include sketch mapping, scale mapping, and transect walking. Maps resulting from these and related
processes vary greatly in form, content, and detail, ranging from basic diagrams of spatial relationships,
drawings on the ground with sticks or chalk, paper sketches, base maps marked with stickers, three-
dimensional physical site models, and GIS- and computer-generated images (Lynch 1960; Downs and Stea
1977; Chambers 1994; Craig etal. 2002; Tversky 1993).

12


-------
systematic variations, preferences, or biases in how different subsets of a survey
population might differentially perceive or value parts of a larger resource. Using
mapping as a complement to traditional SP methods provides an opportunity to elicit
perceptions of the resource from a subset of the largest potential survey population. This,
in turn, provides a way to develop a baseline spatial characterization and assessment of
both use and nonuse values that allow the researcher make more informed decisions
about the commodity to describe and the population to survey.10

As we demonstrate, information from the mapping interviews provides more
general (ordinal, not cardinal) data than would typically be gathered from an SP survey
focused on a specific researcher-defined resource. However, because the mapping
protocol questions are not structured around questions assessing WTP (e.g., a
referendum) nor limited to a single type or cause of environmental damages, they offer
additional flexibility in eliciting information on the extent of the market for a particular
resource. This method is also an improvement over the use of recreational and market
data to understand the extent of the resource as it does not preclude the identification of
nonuse values.

10 Although the mapping protocol identifies the presence of use and non-use values, it does not clarify the
tradeoff the respondent is willing to make for the benefits associated with any particular policy
intervention. We are not suggesting the mapping approach as a method for measuring welfare changes.

13


-------
4. Study Design and Methodological Approach

As a whole, our mapping pilot study consists of the following phases: defining
research questions; developing a base map; designing an interview protocol to elicit (a)
patterns of travel in the region, (b) areas that respondents care about, and (c) areas of
perceived environmental deterioration; designing a debriefing survey. Because this
approach is intended to be streamlined, the focus is on gathering essential baseline
information from a small subset of the potential survey population as early in the survey
design process as possible. To most effectively contribute to the larger SP survey, we
take a mental models interview approach (Morgan et al. 2001) and use a semi-structured
format with selected open-ended questions to elicit responses regarding the sizes, types,
and locations of the natural resources that individuals value within a large region. The
pilot application described here is not a freestanding research effort. Although the
instrument design framework outlined in this section can be used for full mapping
studies, here we have modified the approach and tailored it for application to SP survey
design.

Within an integrated mapping-SP methodology, some basic research questions
will probably be consistent across all studies, including the following: Who cares about
the resource? Where do individuals perceive the boundaries of the resource? Is this
perception consistent across the survey population, or do subpopulations differ
systematically in how they view and value the resource? Do people value parts or
subregions within a resource, and are embedding biases and related biases likely when
considering these areas? Many additional questions could be asked, and the proposed

14


-------
approach is simply intended to serve as an outline for a wider range of applications.

These questions are intended to complement insights from other sources of existing data
(as with the design of any survey) to elicit information on how residents of a large region
might value a shared resource. For example, in some regions, recreation data might
suggest that individuals use resources close to home most frequently or, conversely, that
a single highly visited or high-profile area dominates a larger region. In contrast,
mapping provides an independent verification of these data that also incorporates nonuse
values.

The next steps in the design of the mapping study are to develop a base map and a
mapping interview protocol that build on the main hypothesis of the larger study. There
are a wide range of studies that provide examples of how spatial information can be
elicited and recorded on base maps with varying levels of detail, ranging from blank
sheets of paper and simple line drawings to complex GIS maps (see Lynch 1960; Downs
and Stea 1973, 1977; Francescato and Mebane 1973; Orleans 1973; Saarinen 1973;
Golledge 1976; Tversky 1991; Chambers 1994; Craig et al. 2002; Brown 2004; and
Vajjhala 2005 for examples). Because the base map is the main focus of the mapping
interview and the primary medium through which interview responses will be recorded, it
is crucial that the map and protocol be developed in parallel. Critically, the base map
must be sparse and must serve primarily as a broad frame of reference. The map should
not have so much information or text that it appears to be complete, leading participants
to re-create features already on the map or to refrain from adding information altogether
because they assume it is already there.

15


-------
The framing of the base map (which features are included and at what scales) is
highly likely to influence the scale of participants' responses (Downs and Stea 1977). For
this reason, the overall area should be defined broadly enough to encompass the resource
being studied and relevant surrounding areas while still leaving room for new
information, such as areas and boundaries at the edges of a study region that participants
might identify. Similarly, the printed maps used in interviews should be sufficiently
large, such as 18 by 24 inches, to allow participants to add information clearly at different
scales. We cannot emphasize enough that, to be effective, each base map must be tailored
to the questions being asked, the region being evaluated, and the context of the larger
study; each base map must also be extensively pretested.11

As a complement to the base map, the interview protocol contains three main
sections (see Appendix A for complete protocol). In the first section of the interview,
participants are asked about patterns of use and travel in the study region to elicit basic
spatial information and allow individuals to grow accustomed to the process of adding
information to the base map in response to interview questions. This information reveals
whether the areas that respondents care about are also areas that they frequently use or
visit. In the second section of the protocol, the main focus of the study, individuals are
asked to think about natural areas that they value (in a colloquial sense). This set of
questions asks respondents to identify areas that they care about, whether or not they

11 For this study, pretests were conducted with a small set of participants to evaluate survey responses
given different features (with and without major highways or shading for reference) and levels of detail on
the base map.

16


-------
actively use those areas. All questions are designed to be broad enough to allow for early
identification of key resources and areas that may have nonuse values, but not so broad
that the information of primary concern (i.e., preferences related to natural areas) would
not be mentioned. The responses to this set of questions provide the key results described
below.

In the third section of the protocol, respondents are asked whether they are aware
of any areas and causes of environmental deterioration in the region. These questions are
intended to determine whether respondents are aware that areas in the region have been
affected by specific environmental threats as well as the extent to which they perceive
that these threats are affecting the areas they marked on their maps as areas that they care
about. Furthermore, we may identify the presence of any common scientific
misconceptions about the status or quality of the resource or any threats to it that could
complicate the interpretation of focus group or survey results (Fischhoff et al. 1993).

A number of different, valid approaches can be used to structure these three sets
of questions and, depending on the goal of the larger study, questions could focus on
eliciting (a) a boundary for the region as a whole, (b) specific points people care about
within the region, or (c) broader areas or subparts of the region. To allow for more
efficient coding of the collected data, differently colored markers and pens can also be
used to differentiate types of places added to the maps. In all cases, the questions should
be sufficiently broad to allow for follow-up once the participant has responded. The
interview process and protocol are discussed further in the context of our application to
the SAM region.

17


-------
The final element of the proposed methodology is a short written debriefing
survey including demographic questions as well as basic ranking and follow-on questions
relevant for the larger SP study (see Appendix B for complete debriefing survey). The
data from the survey are intended to validate and complement the results of the mapping
interview and allow for comparative analysis of differences in preferences based on
demographic characteristics of the population. For example, do individuals who lived in
the study region for the longest mark the areas that they value as larger regions on their
maps than those who have lived in the region for less time?

Taken as a whole, this methodology, consisting of research question
identification, base map design, mapping interview implementation, and debriefing
survey completion, is intended to be part of a larger effort, and the methods can be
applied with as much or as little detail as a project requires. Because the goal of this
approach is to provide a structured framework for informing the survey design and
interpreting the results of focus groups used to evaluate draft surveys and the actual
survey administration, this section deliberately presents a very basic, streamlined
approach to integrating mapping and an SP survey design to inform, without duplicating,
information elicited in the larger survey about natural resources, perceived damages, and
WTP for improvements. In the next section, we place the proposed methodology in
context and discuss an application focused on the SAM.

5. Acidification in the Southern Appalachians: An Application

The SAM region is a large, mountainous area surrounding the Appalachian
Mountain chain that stretches from Alabama and Georgia in the south to Virginia and

18


-------
West Virginia in the north. The full region covers approximately 37 million acres
(SAMAB 1996), encompassing parts of eight different states and multiple ecosystems,
land uses, and management authorities, including National Park Service and Forest
Service lands, state parks, public recreation areas, private property, and agriculture land,
among others. The region is anchored by—Great Smoky Mountain National Park
(GSMNP) in North Carolina and Tennessee and Shenandoah National Park (SNP) in
Virginia. These two parks and the surrounding high-elevation forest and stream resources
in the region are currently at risk for significant damages from acid deposition; this issue
has emerged as a policy priority for the affected states and the region as a whole (SAMI
2002; Mische John et al. in preparation).12

The scale of this resource makes both environmental evaluation and policymaking
difficult. The absence of estimates of the economic value of improvements in ecological
systems to date has hindered policymakers' attempts to set efficient environmental
regulation and policy goals. The larger SP study (to which this mapping pilot study is
designed to contribute) is intended to characterize the potential damages to forests and
streams from acid deposition based on the best available science, produce WTP estimates
for environmental improvements that can be achieved with further reduction in acid
deposition across the region, and provide sufficiently robust and detailed results to inform

12 The SP surveys to be informed by this pilot study will not necessarily be limited to describing
improvements in these parks. Indeed, the mapping interviews are also intended to determine whether it is
necessary to estimate the WTP for improvements to ecosystems in national, state, and private forests as
well as national parks.

19


-------
future regulatory analysis.13 The multiple jurisdictional boundaries within the SAM
region pose the greatest challenge for survey implementation in this region. Because the
affected resources are distributed across multiple states and management authorities,
there is no single convenient, well-defined, and widely-used description of the affected
area that clearly identifies the primary survey population from which to sample or
corresponds to how people might perceive the resource.

Figure 1 is the base map used in this pilot study. The extent of the map was
defined based on existing data on the ecosystems, acidification, and recreational patterns
in the SAM region. This base map provides very basic information, including state
borders and names, major highways, and light shading highlighting publicly managed
forests and water resources in the region (participants are not told what the shaded areas
represent and any respondents who asked during the course of the mapping interview
were simply told that the shading indicated some forest and water features). The map was
deliberately designed to minimize the amount of text and to avoid serious framing and
anchoring effects.14 Additionally, the map is titled "Interior Southeastern United States"
to avoid (a) leading participants to focus too narrowly on the Appalachian Mountains or
any social, cultural, or political association with that term and (b) shifting the focus to
competing resources that are less relevant to the study, such as the Atlantic Coast.

13	This study builds off of the Banzhaf et al. (2006) study, which estimated WTP for ecological
improvements from reduced acidification in the Adirondack^. The benefits-transfer components of the
study will evaluate the transfer of WTP estimates between the SAM and Adirondack Mountains regions.

14	See Kahneman et al. (1982) for a general discussion of these effects and Tversky (1981, 1991, 1992) for
specific examples of distortions in perceptions of space and spatial information.

20


-------
Survey Population

In this pilot application and test of the methodology, our interview protocol
focused on eliciting and characterizing responses regarding the types, sizes, and locations
of places that individuals value in the larger SAM region. Given the limited time and
resources for this effort within the larger study, we chose to sample from two states in the
SAM region, North Carolina and Virginia. These two states were selected for three main
reasons. First, each state contains a large portion of the high-elevation resources affected
by acid deposition and a large share of the population in the states that compose the
region. Second, they are separated by sufficient geographic distance to sample and
evaluate distinct survey subpopulations. Third, each state encloses part of a large national
park affected by acid deposition—constituting a sufficiently large portion of the SAM
region to provide different characterizations of the full region and resource—as well as
other state and federal lands.

Comparing individuals from these two states allowed us to evaluate whether they
(a) cared most about the resources within their own state of residence, (b) valued the
SAM area in its entirety, or (c) valued multiple subregions across the entire resource. We
hypothesized, first, that current and former residents of the region would focus on the
GSMNP and the SNP as high-profile resources and, second, that residents of the state
containing a particular resource would value that resource more than residents of other
states in the region. To test these hypotheses, we recruited a convenience sample of
current and former residents of North Carolina and Virginia currently living in the
Washington, DC, area.

21


-------
Participants were recruited through online advertisements; 16 participants from
North Carolina and 15 from Virginia were chosen from approximately 135 responses to
our solicitation.15 All respondents were screened to select participants who had lived in
the study region for a minimum of 5 years since the age of 16. Additionally, to avoid
overlap between groups of participants from the two states, candidates were screened to
eliminate any prospective volunteers who had lived in both North Carolina and Virginia.
Gender balance and geographic distribution within each state were also considered when
recruiting participants. All interviews were conducted from October 2006 to January
2007.

The final sample included approximately equal numbers of men and women from
each state; participants were 23 to 66 years old, with an average age of approximately 34
years. The sample included participants from a wide range of educational backgrounds,
ranging from "some college, but no degree" to "postgraduate degree," with the majority
of participants holding bachelor's or associate's degrees. Median household income
across all participants was in the $50,000 to $84,999 bracket.

Interview Process

Each mapping interview was scheduled and conducted individually. The process
was described as a "mapping interview" and no mention was ever made of the
Appalachian Mountains, environmental degradation, or acid deposition in soliciting

15 Morgan et al. (2001) note that typical sample sizes for exploratory mental model studies similar to this
pilot study range from 20 to 30 when interviews are conducted within population groups expected to share
relatively similar beliefs.

22


-------
participants or conducting interviews. Each interview began with a brief, general
introduction to the goals of the study (to make a map of natural resources in the region
that were important to the respondent) as outlined in detail in the protocol. Study
participants were provided with colored markers and the base map shown in Figure l.16
Significant emphasis was placed on the process of demarcating visited and valued areas
on the map. Participants were repeatedly reminded to carefully consider the sizes and
shapes of locations that they added to their maps as well as the relationships among the
locations.

Interviews were typically between 45 minutes and 1 hour in duration. In the semi-
structured interview protocol, participants were first asked to 1) identify places on the
map that they visited regularly or thought of as a significant destination while living in
the region, 2) add to their maps the center points and boundaries of five natural areas in
the region that they cared about most, and 3) identify any areas in the region that have
experienced environmental deterioration or improvement and to identify the cause of
these changes.17 After adding the center point for a valued natural area, participants were
prompted to carefully examine and explain what defined the size and boundary of the
marked area. For example, prompts included "I noticed that you didn't include this

16	The interviewer handed each participant one colored marker at the beginning of each section of
questions, corresponding with a pre-established coding system. This system was used to avoid confusion
and prevent participants from demarcating visited and valued areas in the wrong color-coded categories.

17	Participants were required to demarcate exactly five areas that they care about on their maps.
Respondents were not explicitly asked to rank these areas during the mapping interview; however, the
order in which places were added served as an implicit ranking. Explicit rankings were elicited in the
debriefing survey to allow for comparison with these implicit rankings elicited during the mapping
interviews. No requirements were placed on the minimum or maximum number of degraded areas that
could be added to each map.

23


-------
(town/highway/etc.) in the area you marked, do you consider it part of this resource? If
not, what defines the start of this edge for you?" Respondents were given time after
adding each area to consider its size, boundary, location, and relationship to other areas
and were allowed to make any corrections or changes. At all points during the interview,
participants were asked to respond to questions both by adding information to their maps
and by explaining their responses verbally; this allowed the interviewer to follow up,
record relevant details, and add prompts for clarification or greater detail. Finally, each
interview was followed by a short written survey with demographic questions and
additional questions about places they marked on their maps.

Figure 1. Base Map of the Southern Appalachian Mountain (SAM) Region

24


-------
The pairing of the mapping interview and debriefing survey also allowed for
quantitative evaluation of respondents' preferences for maintaining or improving the
environmental quality of different valued resources using implicit (by order of addition
on their maps) and explicit rankings (written survey questions). While not sufficiently
explicit to measure the tradeoffs individuals would be willing to make for improvements
to any of the resources included on the map, the mapping protocol provides an empirical
basis for choices made regarding the extent of the market and extent of the resource
imposed in a SP survey design and sampling protocol.

6. Maps, Data Analysis, and Study Results

Results from the 30 mapping interviews and written surveys were transcribed and
coded after all interviews were completed.18 Data compiled from the maps included
counts of places visited and valued by state, type of resource (forest, water, or other), size
of resource, and order in which places were added to the maps, among other more
specific attributes. Figures 2 and 3 are examples of the types of maps collected during the
study that show the diversity in the types and sizes of natural areas that individuals
marked as places that they cared about.

18 All interviews and map coding for this study were completed by a single interviewer/transcriber.

Because this methodology is intended to be applied as an exploratory study with a small sample size, we do
not address any issues of intercoder reliability that might emerge in larger applications of the approach.

25


-------
Figure 2. Virginia Participant Map Showing Major Areas Visited and Traveled (Black),
Five Most Valued Natural Areas (Multiple Colors), and Degraded Areas (Red)

INTERIOR SOUTHEASTERN UNITED STATES

Tennessee

South
Carolui

Alabama

NORTH

Figure 3. North Carolina Participant Map Showing Major Areas Visited and Traveled
(Black), Five Most Valued Natural Areas (Multiple Colors), and Degraded Areas (Red)

INTERIOR SOUTHEASTERN UNITED STATES

west
Virginia

Kentucky

itffcrth
c^rolina;

.Tenneuw

South
-Carolina

Alabama

Georgia

NORTH

26


-------
At prompts to consider what defined the boundary of an area they cared about,
participants highlighted a wide variety of defining characteristics for specific resources
and areas. For example, a large majority of participants demarcated a natural area
overlapping (in part or whole) the GSMNP or SNP and/or referred to an area specifically
as "the Smokies" or "the Shenandoah." However, the sizes and shapes of the resources
varied significantly. Figures 4 and 5 illustrate these differences and respectively show the
SNP and the GSMNP (hatched areas) overlaid with the boundaries drawn by three
selected participants (dashed outlines) for valued areas, which they generally marked as
overlapping either national park and/or specifically labeled as "the Smokies" or "the
Shenandoah."

In defining the boundaries of these areas, some participants referenced ecosystem
characteristics or some other feature that identifies a specific "type" of environment, such
as the sandhills of North Carolina, which one participant stated "are a separate area
because the topography and vegetation are different from the area around it." Still other
participants marked the borders of their valued areas based on locations where changes in
natural features occur, such as the increasing "hilliness" west of Asheville, North
Carolina, marking the start of the Smoky Mountains. Participants also used distances
from cities, highways, and state boundaries to denote the start or edge of a natural area, or
the ownership or management (public versus private) to clarify why they had drawn a
boundary at a specific location. In several cases, valued areas overlapped one another
partially or completely (like the two areas in Figure 3 near GSMNP), highlighting the
potential for mapping to help with early identification of potential embedding problems.

27


-------
Figure 4. Sizes and Locations of Valued Areas Marked as Overlapping the SNP or
Labeled as "the Shenandoah" on Three Selected Participants' Maps (Dashed Lines)
Relative to the Actual SNP Boundaries (Hatched Area)

Figure 5. Sizes and Locations of Valued Areas Marked as Overlapping the GSMNP or
Labeled as "the Smokies" on Three Selected Participants' Maps (Dashed Lines)
Relative to the Actual GSMNP Boundaries (Hatched Area)

The Extent of the Market

To examine how participants' valued areas vary by state, coded data from all

maps were used to conduct basic statistical analyses, such as comparing the counts of

28


-------
visited and valued places added across all maps. Participants from both North Carolina
and Virginia added an average of 18 places that they cared about or had visited in the
study region. As Figure 6 shows, North Carolinians added more places to their maps, on
average, and a majority of these places were within North Carolina. Former Virginia
residents marked fewer places, on average, and their maps also reflected greater out-of-
state travel for the areas marked. Because the population surveyed includes only former
residents of either state, we expect that, as individuals who have moved out of state, study
participants are likely to be more highly traveled than other residents of the region. As a
result, the balance between within-state and out-of-state additions to each map likely
represents an upper bound for out-of-state additions, and we would expect a stronger
home-state preference to emerge among current residents of the region.

Figure 6. Number and Locations of Valued and Visited Areas on Respondents' Maps

NC	VA

The home-state preference is even stronger when we focus exclusively on the five
natural areas respondents marked on their maps. Across all participants, more than 60

29


-------
percent of the center points and areas of the valued natural places were located entirely
within the participant's home state. This average was slightly higher for respondents from
North Carolina who, on average, indicated that four of their five valued places were
within North Carolina.

Participants were also far more likely to demarcate and/or label (as either the
"Shenandoah" or the "Smokies" or similar label) an area corresponding to the national
park in their home states than the park outside their states. Across all participants, 53
percent of North Carolinians and 67 percent of Virginians only included an area
corresponding to the park in their home states (see Figure 7), and one-third of all
participants included both parks. The participants who included both park areas on their
maps always ranked the park area within their own state higher than the other park.19
None of the participants marked only the park outside of their home state, and all but one
participant from each state included an area that encompassed part or all of the national
park in their home states among the five areas that they valued.

These results support the hypothesis that people are more likely to value natural
resources in their own states than those located in other states; this is consistent with
findings from the SP literature on the appropriate extent of the market discussed above.
This suggests that the extent of the market in the SAM region is affected by state

19 Although we emphasized to respondents that they should refer to their preferences, perspectives, and
behavior when they were living in the study region when responding to interview questions, a higher
percentage of North Carolinians may have marked both parks because they now live closer to the resources
in Virginia.

30


-------
boundaries, but not to the extent that values for natural resources held by residents in the
region are limited solely to the resources in their home states.

Another way to identify the extent of the market for a resource for a stated
preference survey is to review other sources of survey data (e.g., Bateman et al., 2002).
However, often the only relevant surveys available collect recreational use data. By
definition these are limited to users and therefore cannot identify the extent of the market
for goods with nonuse values. In order for the mapping approach to be useful in
determining the extent of the market for a resource for which there are significant nonuse
values, it must provide more actionable information than is available from use data. By
comparing the mapping results to information gathered from a sophisticated recreation
survey, we provide evidence that the mapping approach identifies potential nonuse, or
latent use values, and thus provides a better picture of the extent of the market.

A candidate recreational survey for the SAM area is the National Survey of
Fishing, Hunting, and Wildlife-Associated Recreation (FHWAR). The FHWAR is based
on a nationally representative sample of households and is conducted by the U.S. Census
Bureau on behalf of the U.S. Fish and Wildlife Service (U.S. DOI 2001).

31


-------
Figure 7. Percentage of Participants Who Marked a Valued
Area That Included the GSMNP and/or SNP on Their Maps

Table 1 reports the total number of respondents to the
FHWAR survey from North Carolina and Virginia, as well as
the number of those respondents who "hunted," "fished," or
"took trips to observe wildlife," in these states in 2001. We see
that, regardless of the activity, the percentage of respondents
in each state that recreate in the other state is small.
Furthermore, these values are much smaller than the
percentage of respondents to the mapping survey who value an
area outside their home state. As we see in Figure 7, 40
percent of respondents from North Carolina valued a natural
area corresponding to the SNP in Virginia, and 27 percent of
respondents from Virginia valued a natural area corresponding
to the GSMNP in North Carolina.20 This may be because there
are important nonuse values that the mapping survey is
picking up. It also may be because the mapping survey does
not limit the respondent to a particular time period when
asking if which resources are valued. The values expressed on

All Participants

27%

NC Participants

0%

VA Participants

~	Included BOTH Park Areas

~	Included NEITHER Park Area

~	Included ONLY the Smokies

~	Included ONLY the Shenandoah

20 For our purposes one limitation of the FHWAR survey is that it does not ask about recreation in specific
areas within states (e.g., a particular park) or types of areas (e.g., mountainous regions). However, one
could interpret the share of residents of a state that recreate in another state as an upper bound on the
percentage of that state's residents that recreate in the resource of interest (in our application, the parts of
the SAM region in either North Carolina or Virginia).

32


-------
the mapping survey may reflect anticipated use of out-of-state resources. In contrast, the
FHWAR data only captures use over the previous year and therefore does not provide

nformation.21















Table 1:1

Recreation by State





State of Number of

Activity

Number of

Total

Percent of total

residence respondents*51



recreators that

recreators*

days







recreate in:



recreated in:





NC

VA



NC

VA





FHWAR

survey









Fishing

118

9

119

98%

2%

NC 2123

Hunting

76

4

77

97%

3%



Wildlife

37

3

37

98%

2%



Fishing

33

123

140

4%

96%

VA 2044

Hunting

4

86

87

3%

97%



Wildlife

18

65

72

9%

91%





Mapping survey









Fishing

14

0

14

100%

0%

NC 17

Hunting

4

0

4

100%

0%



Wildlife

15

5

15

94%

6%



Fishing

0

7

7

0%

100%

VA 14

Hunting

1

2

2

2%

98%



Wildlife

4

11

11

4%

96%

*A recreator is one that engages in the associated activity in either state. In the mapping
survey respondents were asked their recreation in a typical year, while in the FHWAR study
they were asked about their recreation in 2001.

**For the FHWAR, this is the total number of individuals for which data was collected (as
opposed to households, which was the survey unit for the first wave).

It is possible that the mapping respondents are not representative in their
recreation activities compared to the FHWAR sample. If the members of our convenience

21 One could estimate a model of likely visitation frequency using observed data on recreators and non-
recreators collected as part of the FHWAR survey to predict the number of potential users of out-of-state
resource use by members of a particular state. However, there presumably are significant unobserved
differences between those observed recreating and those who are not and so it may be difficult to
extrapolate the recreating sample to the non-recreating sample in the survey.

33


-------
sample are more apt to recreate out-of-state, then this may be explaining their preferences
for out-of-state resources. As part of the written debriefing survey, we asked respondents
the average number of days in a typical year that they recreated in their home states and
in their neighboring states.22 We see in Table 1 that the respondents to the mapping
survey are more avid recreators than those in the FHWAR survey sample. However, with
the important exception of wildlife watching, they are not more likely to engage in these
activities outside their home state. Furthermore, we see in the rightmost columns of Table
1 that the percentage of days recreated in the home state is comparable between the
respondents to the FHWAR survey and to the mapping survey for each activity. For the
wildlife watching activity, we may then be seeing that the typically number of days of
wildlife watching out-of-state occurs in concentrated periods (say over a week) at
infrequent intervals. In either case, the recreation data appear to miss non-use values,
infrequent use values, or both.23 The disparity between the percentage of days recreated
in each state and the percentage of respondents who care about a resource outside their
state provides further evidence that the mapping protocol provides nonuse and latent use
value information that may not be apparent from recreation use data. Even more directly,
we see that the percentage of respondents in our sample who recreate out-of-state in the

22	We asked about recreation in a typical year rather than a particular year like in the FHWAR survey
because some of our respondents no longer live in their home state. Otherwise the recreation questions in
the debriefing survey are very similar to those in the FHWAR survey.

23	Admittedly the FHWAR could be structured to ask about recreation in a typical year which would
address the infrequency of use issue. Such an approach would come with other problems, such as recall
bias, and so is understandably inappropriate given the purposes of the FHWAR survey.

34


-------
convenience sample is smaller than the percentage that values an out of state resource
corresponding to the GSMNP or the SNP.

The Extent of the Resource

Similar to the counts and locations of valued areas used to evaluate the extent of the
market, we use the sizes of the mapped valued areas to examine the extent of the resource.
Defining five size categories based on the approximate sizes of areas on all maps, with 1
corresponding to the smallest area (less than 500 square miles) and 5 corresponding to the
largest areas (those greater than 8,000 square miles), the sizes of all valued areas were
estimated for each map and were assigned codes denoting their size categories. A majority
of valued areas across all maps and participants were in the smallest two size categories, 1
or 2; however, most participants included a variety of area sizes on their maps.

Of those participants who marked a natural area partially or entirely overlapping
the GSMNP or SNP and/or referred to an area specifically as "the Smokies" or "the
Shenandoah" on their maps, most drew the area such that it was significantly larger than
the GSMNP or SNP boundaries, as illustrated by Figures 4 and 5. Interestingly, of those
participants who included both parks on their maps, participants delimited the area
representing or including the park outside their home states such that it was significantly
larger than the corresponding park within their home states. Of Virginians who included
an area corresponding to the GSMNP on their maps, 75 percent drew it larger than the
park boundaries compared with 50 percent of North Carolinians; 100 percent of North
Carolinians who included an area corresponding to the SNP on their maps drew it larger
than the park boundaries compared with 77 percent of Virginians. Additionally, as Figure

35


-------
8 illustrates, the average sizes of both the GSMNP and the SNP areas marked on
participants' maps were substantially larger than the average size of all other valued areas
that respondents marked on their maps.

Figure 8. Average Sizes of Valued Areas Demarcated or Labeled as GSMNP or SNP

Average Size of Valued Areas

=¦ 5



- 4

c

TO
Q.

o

TO

a.



o
o
<

CO

~ NC Participants
¦ VA Participants

Average - All Areas Great Smoky Mountain Shenandoah National Average without
National Park	Park	GSMNP and SNP

This relationship between the distance from a resource and its perceived size raises
the hypothesis that respondents may assign disproportionate value to a less familiar and
more distant resource (based on a skewed perception of scale and assuming that scale is
correlated with WTP). This may even be true if researchers feel that they have clearly
defined the extent of a resource to be valued if the information in the survey is not
completely accepted by the respondent. This hypothesis contrasts with findings that WTP
for improvements to such resources typically drops with distance; however, further
research is required to evaluate potential confounding effects of map distortions and spatial

36


-------
biases. Together, these initial analyses and results suggest that the extent of the resource for
the SAM region is significantly larger than current national park boundaries; however,
residents do not necessarily value the region as a single large, contiguous resource. Instead
it appears that, although a majority of residents assign a greater value to resources within
their home states, a significant percentage of residents value other subparts of the resource
of interest, such as the GSMNP and the SNP; thus, further exploration of the extent of the
resource through focus groups for the acidification WTP study is warranted.

7. Application of Mapping Results to SP Survey Design

The mapping analysis has informed the ongoing design of the larger SP survey for
the SAM region in a number of ways. Although the mapping exercise does not provide a
particular decision rule as to the appropriate extent of the market or resource, it does
suggest that resource improvements outside of one's home state are much more important
than the recreational demand data suggest. Therefore, as it stands, the survey will
probably be administered in the four-state region where damages from acidification are
the most extensive (Virginia, North Carolina, Tennessee, and West Virginia) and describe
improvements to high-elevation forests and streams over this entire area. Fortunately,
focus group testing has dispelled one of our more significant concerns—that respondents
might not find a multi state compact for intervening and improving the quality of forests
and streams credible.24 Focus group participants were asked to vote on a program for
reducing the effects of acidification; this program would raise their income taxes and

24 Juha Siikamaki, Alan Krupnick, and Susie Chung of Resources for the Future have led these focus groups.

37


-------
would require a simple majority in each state to go into effect.25 During this process,
participants also voiced concerns about ecological connectivity and strongly felt that,
absent improvements over the entire resource, no improvement in any single area would
be sustainable. Therefore, even if they only cared for part of the resource, they felt that it
would not be possible for the area they cared about to improve absent improvements
elsewhere. These results, in combination with the results of the mapping study, motivated
an expansive definition of the extent of the resource.

8. Conclusions

Taken as a whole, this pilot application develops and illustrates a methodology for
integrating mapping into SP survey design. Application of the proposed methodology to
the SAM region reveals that the method does allow for a preliminary analysis of the extent
of the market and the extent of the resource when the resources of interest have important
nonuse values; however, the scope of this study would need to be significantly expanded
to further test the efficacy of the method for other applications and locations. In this pilot
application, initial results reveal that individuals value a greater number of resources in
their home states and assign higher rankings to home-state resources than to those in other
states in the region. Based on this finding, we would expect that an SP instrument that
focuses only on resources or damages in a single state would generate higher average
WTP values from residents of that state than from residents of other states.

25 Similar to Banzhaf et al. (2006), the intervention that leads to improvements in resource quality is a
program to add lime to streams and forests. A program that would directly reduce emissions raises issues of
who should pay and the incentive compatibility of the payment vehicle.

38


-------
Other analyses also show that using only national park or national forest
boundaries in an SP survey may underestimate the extent of the resource people value.
Participants who indicated that they valued a mountainous area in North Carolina or
Virginia overwhelmingly identified areas larger than corresponding national park
boundaries for the GSMNP and SNP, respectively. Finally, participants from both states
showed a consistent pattern of marking more distant valued areas as geographically larger
than more proximate ones. This suggests an interesting avenue for further research on the
spatial perceptions associated with drops in distance-weighted WTP.

Although this mapping exercise is clearly useful for understanding the market for
resources that have important nonuse values, we also feel that this method would easily
transfer to a choice among alternative recreational sites in a recreational demand model
(Parsons and Hauber, 1998). This pilot study also suggests avenues for new experimental
research to examine, for example, the effect of conducting an SP survey on respondents'
perceptions of the sizes and boundaries of the resources they value, or conversely, the
effect of a prior mapping exercise on respondents' WTP in a follow-on SP survey.

Overall, the results presented here are based on a pilot study, and the conclusions with
respect to perceptions of large resources warrant further study. At this stage, by providing
a preliminary spatial characterization of both the extent of the market and the extent of
the resource, this approach demonstrates how mapping can both inform the design of SP
surveys and aid in the interpretation of WTP results.

39


-------
References

Banzhaf, H.S., D. Burtraw, D.A. Evans, and A. Krupnick. 2006. Valuation of Natural
Resource Improvements in the Adirondacks. Land Economics 82(3): 445-64.

Bateman, I., R.T. Carson, B. Day, W.M. Hanemann, N. Hanley, T. Hett, A. Jones, G.
Loomes, S. Mourato, E. Ozdemiroglu, D.W. Pearce, R. Sugden, and J. Edward
Swanson. 2002. Economic Valuation with Stated Preference Techniques: a
Manual. Cheltenham, UK: Elgar.

Bateman, I.J., B.H. Daya, S. Georgioua, and I. Lake. 2006. The Aggregation of

Environmental Benefit Values: Welfare Measures, Distance Decay and Total
WTP. Ecological Economics 60: 450-60.

Boyle, K.J. 2003. Contingent Valuation in Practice. In A Primer on Nonmarket

Valuation, edited by Champ, P.A., K.J. Boyle, and T.C. Brown. Boston, MA:
Kluwer Academic Press, 111-170.

Brown, G. 2004. Mapping Spatial Attributes in Survey Research for Natural Resource

Management: Methods and Applications. Society & Natural Resources 18(1): 17-
39.

Chambers, R. 1994. The Origins and Practice of Participatory Rural Appraisal. World
Development 22(7): 953-69.

Champ, P. A., and M.P. Welsh. 2007. Survey Methodologies for Stated Choice Studies. In
Valuing Environmental Amenities Using Stated Choice Studies, edited by B.J.
Kanninen. Dordrecht: Springer, 21-42.

Craig, W.J., T.M. Harris, and D. Weiner. 2002. Community Participation and Geographic
Information Systems. New York: Taylor & Francis.

Downs, R.M., and D. Stea. 1973. Image and Environment: Cognitive Mapping and
Spatial Behavior. New Brunswick, NJ: Aldine Transaction Publishers.

40


-------
Downs, R.M., and D. Stea. 1977. Maps in Minds: Reflections on Cognitive Mapping.
New York: Harper and Row.

Evans, D.A., H.S. Banzhaf, D. Burtraw, A.J. Krupnick, and J. Siikamaki. 2008. Valuing
Benefits from Ecosystem Improvements Using Stated Preference Methods: an
Example from Reducing Acidification in the Adirondacks Park. In Saving
Biological Diversity: Balancing Protection of Endangered Species and
Ecosystems, edited by Askins, R. A., G.D. Dreyer, G.R. Visgilio, and D.M.
Whitelaw. New York: Springer, 101-117.

Evans, G.W., D.G. Marrero, and P.A. Butler. 1981. Environmental Learning and
Cognitive Mapping. Environment and Behavior 13(1): 83-104.

Fischhoff, B., M.J. Quadrel, M. Kamlet, G. Loewenstein, R. Dawes, P. Fischbeck, S.
Klepper, J. Leland and P. Stroh. 1993. Embedding Effects: Stimulus
Representation and Response Mode. Journal of Risk and Uncertainty 6(3): 211-
34.

Francescato, D. and W. Mebane. 1973. How Citizens View Two Great Cities: Milan and
Rome. In Image and Environment: Cognitive Mapping and Spatial Behavior,
edited by Downs, R.M., and D. Stea. New Brunswick, NJ: Aldine Transaction
Publishers, 131-147.

Freeman, A.M. 2003. The Measurement of Environmental and Resource Values.
Washington, DC: RFF Press.

Golledge, R.G. 1976. Methods and Methodological Issues in Environmental Cognition
Research. In Environmental Knowing: Theories, Research, and Methods, edited
by Moore, G.T. and R.G. Golledge, Stroudsburg, PA: Dowden, Hutchinson &
Ross, 300-313.

Golledge, R., and G. Zannaras. 1973. Cognitive Approaches to the Analysis of Human
Spatial Behavior. In Environment and Cognition, edited by Ittleson, W. New
York: Academic Press, 59-94.

41


-------
Johnson, R.F., R.W. Dunford, W.H. Desvousges, and M.R. Banzhaf. 2001. The Role of
Knowledge in Assessing Nonuse Damages: a Case Study of the Lower Passaic
River. Growth and Change 32: 43-68.

Hanley, N., F. Schlapfer, and J. Spurgeon. 2003. Aggregating the Benefits of

Environmental Improvements: Distance-Decay Functions for Use and Non-use
Values. Journal of Environmental Management 68: 297-304.

Kahneman, D., P. Slovic, A. Tversky. (eds.). 1982. Judgment under Uncertainty:
Heuristics and Biases. New York: Cambridge University Press.

Kaplan, R. 1976. Way-finding in the Natural Environment. In Environmental Knowing:
Theories, Research, and Methods, edited by Moore, G.T. and R.G. Golledge,
Stroudsburg, PA: Dowden, Hutchinson & Ross, 46-57.

Klett, F.R. and Alpaugh, D. 1976. Environmental Learning and Large Scale

Environments. In Environmental Knowing: Theories, Research, and Methods,
edited by Moore, G.T. and R.G. Golledge, Stroudsburg, PA: Dowden, Hutchinson
& Ross, 121-130.

Krupnick, A.J., S. Banzhaf, D. Burtraw, B. Cosby, C.T. Driscoll, D. Evans, J. Siikamaki.
2004. Valuation of Regional Ecological Response to Acidification and
Techniques for Transferring Estimates.

http://cfpub.epa.gov/ncer_abstracts/index.cfm/fuseaction/display.abstractDetail/ab
stract/7726/report/0 (accessed 16 January 2008).

Loomis, J. 1996. How Large is the Extent of the Market for Public Goods: Evidence from
a Nationwide Contingent Valuation Survey. Applied Economics 28:779-82.

Loomis, J. 2000. Vertically Summing Public Good Demand Curves: an Empirical
Comparison of Economic versus Political Jurisdictions. Land Economics
76(2):312-21.

Lynch, K. 1960. The Image of the City. Cambridge, MA: MIT Press.

42


-------
Mark, D.M., and A.U. Frank. 1996. Experiential and Formal Models of Geographic
Space. Environment and Planning, B 23: 3-24.

Mische John, A., D. Burtraw, D. Evans, H.S. Banzhaf, A. Krupnick, and J. Siikamaki. A
Summary of the Science of Acidification in the Southern Appalachian Mountain
Region. Discussion paper (in preparation). Washington, DC: Resources for the
Future.

Mitchell, R.C., and R.T. Carson. 1989. Using Surveys to Value Public Goods: the
Contingent Valuation Method. Washington, DC: RFF Press.

Morgan, M.G., B. Fischhoff, A. Bostrom, C.J. Atman. 2001. Risk Communication: a
Mental Models Approach. New York: Cambridge University Press.

Orleans, P. 1973. Differential Cognition of Urban Residents: Effects of Social Scale on
Mapping. In Image and Environment: Cognitive Mapping and Spatial Behavior,
edited by Downs, R.M., and D. Stea. New Brunswick, NJ: Aldine Transaction
Publishers, 115-130.

Parsons, G.R. and A.B. Hauber. 1998. Spatial Boundaries and Choice Set Definition in a
Random Utility Model of Recreation Demand. Land Economics 74(1): 32-48.

SAMAB (Southern Appalachian Man and the Biosphere Cooperative), 1996. Southern
Appalachian Assessment. http://samab.org/data/SAA_data.html.

SAMI (Southern Appalachian Mountains Initiative). 2002. Final Report. Asheville, NC:
SAMI.

Saarinen, T. F. 1973. Student Views of the World. In Image and Environment: Cognitive
Mapping and Spatial Behavior, edited by Downs, R.M., and D. Stea. New
Brunswick, NJ: Aldine Transaction Publishers, 148-161.

Schlager, E., and E. Ostrom. 1992. Property-Rights Regimes and Natural Resources: a
Conceptual Analysis. Land Economics 68(3): 249-62.

Smith, V.K. 1993a. Welfare Effects, Omitted Variables, and the Extent of the Market.
Land Economics 69(2): 121-31.

43


-------
Smith, V.K. 1993b. Nonmarket Valuation of Environmental Resources: an Interpretive
Appraisal. Land Economics 69(1): 1-26.

Smith, V.K., and L.L. Osborne. 1996. Do Contingent Valuation Estimates Pass a "Scope"
Test? A Meta-Analysis. Journal of Environmental Economics and Management
31(3): 287-301.

Tolman, E.C. 1948. Cognitive Maps in Rats and Men. The Psychological Review 55(4):
189-208.

Tversky, B. 1981. Distortions in Memory for Maps. Cognitive Psychology 13: 407-33.

Tversky, B. 1991. Spatial Mental Models. The Psychology of Learning and Motivation 27:
109-45.

Tversky, B. 1992. Distortions in Cognitive Maps. Geoforum 23: 131-38.

Tversky, B. 1993. Cognitive Maps, Cognitive Collages, and Spatial Mental Models. In
Spatial Information Theory: a Theoretical Basis for GIS Proceedings COSIT '93,
edited by Frank, A.U., and I. Campari. Lecture Notes in Computer Science,
Volume 716. Berlin: Springer, 14-24.

U.S. DOI (Department of the Interior), Fish and Wildlife Service, and U.S. Department
of Commerce, U.S. Census Bureau. 2001. 2001 National Survey of Fishing,
Hunting, and Wildlife-Associated Recreation.

http://www.census.gov/prod/www/abs/fishing.html (accessed July 1, 2006).

Vajjhala, S.P. 2005. Mapping Alternatives: Facilitating Public Participation in

Development Planning and Environmental Decision Making. Ph.D. dissertation,
Carnegie Mellon University, Engineering and Public Policy, Pittsburgh, PA.

Whinston, M.D. 2007. Antitrust Policy toward Horizontal Mergers. In Handbook of

Industrial Organization, Volume 3, Edited by Armstrong, M., and R. Porter. New
York: Elsevier, 2369- 2440.

44


-------
Appendix A: Mapping Interview Protocol

Survey number

INTERIOR SOUTHEASTERN U.S. MAPPING INTERVIEW

Briefly introduce the project to the subject. Begin with a description of the type of maps the
project is trying to collect and how the subjects' participation is important. The process
should take less than 60 minutes. Participants will add information onto the base map in
response to a series of interview questions. The primary goal is to gather information about
the areas individuals "value" in the region. Emphasize that drawing skills or map-making
skills are not required; however, the subject should carefully consider the sizes, shapes, and
boundaries of the areas they add to the map and how they relate to one another.

Hello my name is Anna. I'm here today from Resources for the Future. We are conducting a study on the
Interior S outheastern U.S. We would like you to help us by takingpart in a mapping exercise. You don't
have to have any experience with drawing or map-making, so please don't worry! What I would like you to
think carefully about is the natural places you care about in this region. I'm going to ask you to add
information onto the base map in front of you, and I'd like you to think about the si^es, shapes, and
boundaries of these areas as you add them to your map and also how they relate to one another.

The goal of this whole interview is for you to create a map of the areas you care about, the places that are
important to you, and spaces that you value in the Interior S outheastern United States. The base map in
front of you shows parts of 8 states in this region, and there are colored markers here for you to use. First,
take a minute to look over this base map. Do you have any questions?

As I askyou questions I'd like you to answer each out hud (there is a tape recorder here) and also to answer
each question by adding the areas and locations that you identify on to your map. If you aren't sure about a
specific answer — don't worry-you can always go back and add places, change your map, or make corrections.
The point isjustfor you to carefully identify the natural areas that are important to you in this region and
draw these areas on your map. Do you have any questions about the project? Okay, let's begin.

Places You Go [Black]

1. The first few things I am going to ask you are general questions about where you
used to live in this region and any major places you went to regularly. For all of these
questions, I would like you to focus only on the time during which you lived in this
region. This is very important, so I really want to emphasize that I would like you to
you think only about places you went when you lived in this area.

45


-------
2.	First, I'd like you to start by taking a look at the base map in front of you and begin
by finding the general location for where you used to live (this doesn't have to be
exact, just take your best guess). Here is a BLACK marker. Using this marker, mark
the location of your home (when you lived in the region) on the map. You can use
any symbol you would like to identify your home, and please write the name of the
town and your former zip-code (if you remember it) next to your symbol.

3.	Have you lived in other places in this region ? Ifso add these "homes" to the map as well and label
them too.

4.	Now I am going to ask you a few questions about some of your activities and trips in
this region during the time you lived here.

5.	Think carefully about where you used to go outside of the town or city where you
used to live in the region. These are places that you might have gone somewhat
frequently, that were not part of your everyday routine, like work or the grocery

store. Please mark each place and label it. (Give the subject time to add a few places) For
example, are there any specific places you used to go at least once a month or a few
times a year?

How about places you might have gone for (say these prompts one at a time, give the subject
enough time to think about it between each prompt and either add places or say "no ")...

" General recreation / outdoor activities? Parks? Campsites?

Hunting? Fishing? Hiking?

"	Observing wildlife/photography

"	Vacations or other travel?

"	Trips to visit family/friends?

"	Seasonal activities? White water rafting? Fruit picking?

6. Look back on the places that you already have drawn on your map. Would you like

to add any places in any of these other states (pointgenerally to blank areas on the
subject's map)}

Is there any place that you went to often or think is important that is not already on
the map? If there is, add it now.

Places You Value [Multiple Markers]

Ok, now I'm going to askyou to switch markers. Don't wory if there are places thatyou've forgotten to add
up to this point, you can always go back and add these places. Remember this process is not aboutfinding

46


-------
exact locations, instead its more importantfor you to think carefully about the si^e of each area you add to
the map and where the edges are relative to the other points you've drawn on your map

Now I would like you to think about the areas that you cared about in this region during the time you
lived here. I would like you to focus on natural places, not cities or man-made destinations like a
family farm orfriend's home. Instead I would like you to think about any type of natural environment or
area that was personally important to you. This can include places you like and value even if
you never went there often or at all. It can even include places that you wanted to go to.
but haven't visited

I am going to ask you to addyourfve most important areas to the map one at a time. Before you add
any information to your map, ty to think about how these paces relate to one another. Each area can be
as big or as small as is important to you.

7.	Now I'd like you to use the RED marker, and start by thinking about the first of
these places you value. Begin by marking the center of this place on your map with
an X and labeling it.

8.	Now think carefully about the size of this whole "place" on your map. How big is
the area that you cared about and think is important? What defines this area around
the center point you selected? Now I want you now to draw a boundary of this space
and explain what features define or form the edges or boundaries of this area. What
makes up the edge of this area?

9.	Why did you choose to add this area as a natural place you cared about?

10.	Now here is a BLUE marker. I would like you to repeat this same process for the
next place you care about and think is important. Start again by marking the center
of this second natural place that you value on your map with an X and label it.

11.	Again I'd like you to think carefully about the size of this place on your map and also
how it relates to the first area you added. Draw the boundary of this area.

What defines this bounday? (Take notes here at each of these explanations)

12.	Why did you choose to add this area as a natural place you cared about?

Continue with the third (ORANGE), fourth (YET J OW) andfifth (GREEN)places, always
asking the subject to mark the center of each place on your map with an X and a label. Remind
the subject to consider how each new place relates to the others already on the map, and then ask the
subject to carefully draw the bounday of each area.

47


-------
13. With the orange/yellow/green marker mark the center/boundary of a

third/fourth/fifth natural area that you most care about, even if you never visited it.

Why didjou choose that as the center?

I noticed that you included/ avoided

Does that overlapping area include this other area you've marked over here?

14.	Excellent. Now I want you to take a minute to look over your whole map. Is there
anything you would like to add or to change about these places you value or their
boundaries? Do you think that anything should be bigger or smaller? Are the center
points where you would like them to be?

15.	Have you visited any of these places on your map? If so, when/what for/how often?

(Take notes here)

Negative and Deteriorated Places

1. For the last drawing section I would like you to use this PURPLE marker. Look
closely at the region on your map and the five places that are important to you, and
think about any major changes you saw during your time in the region. Are there any
areas that you think were degraded or have deteriorated significantly while you lived
in the region? If so, mark these areas on the map, and explain why you think these
areas are degraded and what the causes are?

Okay, Congratulations-you're almostfinished! I would like you to just take one final look at your
map, and see if there is anything missing or anythingyou would like to change. Do you have any
questions for me?

As a final wrap up, I have a brief written survey that I would like you to complete about
the places on your map and some general demographic questions. This shouldn't take
more than 5 minutes. As you go through the survey feel free to ask me any questions you
might have, and you can just hand it to me when you are finished.

Once they've handed in their survey, explain the payment, etc.

Give them the letter and let them know that if thy have any questions or would like to follow-up they
can contact us at the email/phone on the letter, and thank them very much for their time...

48


-------
Appendix B: Debriefing Survey

INTERIOR SOUTHEASTERN UNITED STATES

1.	Of the five important natural places that you marked on your map, please rank
these places in order of importance from 1 being most important to 5 least
important. (Please write your answers in the spaces below using the same names
that you used on your map. Feel free to refer back to your map.)

1	.	

2	.	

3	.	

4	.	

5	.	

2.	If you were to add any other natural place outside of this map region to the list
of natural places that you care about, what would it be? (Please write your
answer in the space below)

3.	How would you rank this place relative to the other natural places you ranked
above? (Check only one box.)

~ Above number 1

~

Between numbers

1

and

2

~

Between numbers

2

and

3

~

Between numbers

3

and

4

~

Between numbers

4

and

5

~ Below number 5

49


-------
How many years did you live in the entire region represented on your map since
you were 16 years old?

~	Less than 5 years

~	5-15 years

~	16-30 years

~	More than 30 years

Please write the name of your home state (from the map) in the space below.

How many years has it been since you last lived in this state?

	years (Please write the total number of years in the space to the

left.)

How many years total did you live in this state only?

~	Less than 5 years

~	5-15 years

~	16-30 years

~	More than 30 years

50


-------
8. If you were given $100 to distribute for making environmental improvements in
the region on your map, how would you divide this money across the 8 states in
the region shown on your map? (Please write a number in the blank next to each
state, the total for all states should add up to $ 100.)

Alabama
Georgia
Kentucky
North Carolina
South Carolina
Tennessee
Virginia
West Virginia

9. When you lived in the area on your map, did you ever hunt in your home state?

~	No (SKIP TO Question 11)

~	Yes (Continue to Question 10)

10. If yes, on average about how many days per year (from 1 day to 365 days)

did you hunt in your home state? (Write your answer in average number of
days per year in the blank to the left.)

	 Days

11. When you lived in this area, did you ever hunt in any of the other states in the
region outside of your home state?

~	No (SKIP TO Question 13)

~	Yes (Continue to Question 12)

51


-------
12. If yes, on average about how many days per year (from 1 day to 365 days)
did you hunt in each of the areas below? (Write your answers in average
number of days per year in the blanks below.)

in the entire region outside of your home state
in Virginia only
in North Carolina only
in Tennessee only

13. When you lived in the area on your map, did you ever fish in your home state?

~	No (SKIP TO Question 15)

~	Yes (Continue to Question 14)

14. If yes, on average about how many days per year (from 1 day to 365 days)
did you fish in your home state? (Write your answer in average number of
days per year in the blank below.)

	 days

15. When you lived in this area, did you ever fish in any of the other states in the
region outside of your home state?

~	No (SKIP TO Question 17)

~	Yes (Continue to Question 16)

52


-------
16. If yes, on average how many days per year (from 1 day to 365 days) did
you fish in each of the areas below? (Write your answers in average number
of days per year in the blanks to the left.)

in the entire region outside of your home state
in Virginia only
in North Carolina only
in Tennessee only

17. When you lived in the area represented on the map, did you ever take any trips
at least one mile from your home to observe wildlife in your home state?

~	No (SKIP TO Question 19)

~	Yes (Continue to Question 18)

18. If yes, on average about how many trips at least one mile from your home
did you make to observe wildlife in your home state? (Write your answer
in number of trips in the blank below.)

	 Trips per year

19. When you lived in this area, did you ever take any trips to observe wildlife in
any of the other states in the region outside of your home state?

~	No (SKIP TO Question 21)

~	Yes (Continue to Question 20)

53


-------
20. If yes, on average how many trips per year did you make to observe

wildlife in each of the areas below? (Write your answers in trips per year in
the blanks to the left.)

in the entire region outside of your home state
in Virginia only
in North Carolina only
in Tennessee only

21. Would you describe yourself as an environmentalist?

~	Yes, definitely

~	Yes, somewhat

~	No

22. What is the maximum tax increase for your household that you would accept to
pay for making improvements to parks and the natural environment in the
entire area represented on your map? (Write your answer in the box below.)

I would accept a tax increase of at most $ 	 per year

for the next 10 years to pay for this program.

23. Please write your age in the space to the right. 	 years

54


-------
24.

What is your gender?

~	Male

~	Female

25. What is the highest degree or level of education that you have completed?

~	Less than high school

~	Graduated from high school - Diploma or Equivalent (GED)

~	Some college, no degree

~	Bachelor's degree or Associate degree

~	Postgraduate degree

26. Please indicate the category that best represents your total household income in
the past 12 months before taxes. Was it...

~	Less than $19,999

~	$20,000-$34,999

~	$35,000-$49,999

~	$50,000-$84,999

~	$85,000-$124,999

~	$125,000 or more

Thank you for completing this survey/

Please hand-in your completed survey to your map interviewer.

55


-------