Chapter 3

 

3.1	The Development Process

Using a series of focus groups, we refined the draft survey elements from the scoping process and tested WTP elicitation formats. We conducted five formal focus groups using randomly selected subjects recruited by phone and paid for their participation: FG5 in Boulder (in April 1996), FG6 and FG7 in Denver (in May 1996), and FG8 and FG9 in Boston (in June 1996). The survey materials for the focus groups included a script read by the person administering the survey (the "administrator"), some visual materials, and a survey questionnaire. The visual materials were used to describe the set of Washington, DC, marble monuments, show weathering injuries, and illustrate the baseline injury time line and the beneficial effects of preservation. Appendix A contains the Boston focus group script and questionnaire.

The survey materials were modified after each of the first three focus groups. The same script and questions were used at both of the Boston groups. The Boston focus group moderator script and response booklet are included in Appendix A. These survey materials were submitted to a peer review process, which resolved a few remaining issues. Then, they were finalized for a group survey mode. Appendix B contains the final survey script, response booklet, and visual materials.

 

3.1.1

Refinements to Defining the Good

The injury photo series and time lines play an important role in defining the good because they illustrate how injuries might affect the monuments' appearance over time, and how long such changes might take with and without one of the preservation options. We finalized the photo series after a review by material injury experts. The experts recommended that the weathering injuries be made more comparable between the statue and the column capital. These changes were incorporated in the final digital enhancements. Final edits were also made to the booklet containing illustrations of Washington, DC, monuments and examples of erosion and chemical alteration injuries. These included adding more examples of marble monuments in Washington, DC, and improving the selection of photographs showing general erosion and alteration injuries.

During this phase we also tested whether respondents might be sensitive to the length of the baseline time line for injuries. Because the rate at which baseline injuries occur is uncertain, we used different time lines to represent injuries without preservation. In one of the baselines, the future conditions in the photograph time series occurred in 75 years and 150 years and the other baseline was 50 years and 100 years. The preservation options had the same percentage changes (25%, 50%, and 100%), but the positions on the time line varied according to the baseline (see Appendix B for the time line illustrations). Each focus group saw only one of the baselines; showing multiple baselines to a focus group would take too long, and risked confusing respondents or reducing the preservation program's credibility. We were not able to discern any difference in the valuation responses to the different baselines. However, the samples were too small to draw any conclusions about how different baseline injury assumptions might affect willingness to pay values.

 

3.1.2

WTP Elicitation Format

Instrument development activity during the formal focus groups concentrated primarily on the WTP exercises. These exercises were designed to provide the data for modeling WTP as a function of demographic characteristics and preservation effectiveness levels. We experimented with four common value elicitation methods across the focus groups: a payment card, a dichotomous choice or referendum, a ranking exercise, and pair-wise choices.

A payment card format obtains WTP information by showing a respondent a card with several dollar amounts on it, and asking her to circle the highest amount on the card that she is willing to pay for some good (examples include Desvousges et al., 1987; Cameron and Huppert, 1989; Chestnut and Rowe, 1990; Rowe et al., 1991; Lazo et al., 1992; Navrud et al., 1992; Schulze et al., 1995; and Poe and Welsh, 1996). In our study, a good is one of the optional levels of preservation effectiveness, e.g., Option A, so the amount circled is a lower bound on WTP for decreasing the rate of injury. Initially the values on our payment card ranged from $0 to $40, but the upper limit of the range was increased to $200 to avoid truncation bias because several respondents selected the "More than $40" payment card choice. Rowe et al. (1996) showed that truncation bias caused by setting upper limits too low was an important source of valuation bias; conversely, this study found no centering or range bias, which are thought to be potentially important sources of bias for payment card responses. A logarithmic function determined the intervals between values shown on the payment card (Rowe et al., 1996).

The referendum format gives a respondent a dollar amount or price for a good and asks whether she is willing to vote pay that amount for the good (examples include Bishop and Heberlein, 1979; Hanemann et al., 1991; Grosclaude and Soguel, 1994; Carson et al., 1994; Carson et al., 1996). The Blue Ribbon Panel on CVM valuation recommended that the referendum format be used for natural resource damage assessment valuation studies because it poses a familiar task (i.e., voting for/against taxes or fee increases) (58 Fed. Reg. 4601, Jan. 15, 1993). However, there is some evidence that referendum questions may lead to inflated willingness to pay values because they force respondents into "moral dilemmas" of voting against a program they want to support, but at a lower willingness to pay amount than the one shown in the referendum question (Boyle and Bishop, 1988; Poe and Welsh, 1996).

Because this valuation format obtains limited information from each respondent, discrete choice modeling techniques are required to estimate willingness to pay values and samples must be large to obtain enough information for modeling (Cameron and James, 1987; Cameron, 1988). A follow-up referendum question with a higher or lower value can increase the amount of information per respondent (Hanemann et al., 1991). However, the survey must be administered in-person or by telephone to ensure the original referendum is answered prior to reading the follow-up referendum.

The modeling results are sensitive to the values shown in the referendum questions (Cameron and Huppert, 1991; Poe and Welsh, 1996). Identifying optimal values can require significant testing of survey design.

The referendum question used in focus groups asked participants if they would vote for a preservation program that achieved Option C (the 100% increase in the injury time line) if it increased their household taxes next year by $X. The dollar values ranged from $0.25 to $25, and the referendum question mentioned that the tax increase would only last one year.

Choice experiments show individuals two or more bundles of characteristics or attributes of a good, and asks them to rank the bundles. One of the attributes is a dollar value or price, which allows the estimation of willingness to pay and values for nonprice attributes. In the ranking format, the respondent partially or fully ranks the goods based on her preferences over the bundles. In Schulze et al. (1994), each respondent chose his or her most preferred bundle from five bundles of Superfund site cleanup attributes, which included residual health risk levels and cleanup costs. Then, each respondent selected a second most preferred bundle, assuming the first preference was not available.

We provided focus group respondents with a list of the three alternative preservation options and three prices. As a fourth alternative, we included the option "No preservation program," which had a price of $0. Respondents fully ranked the alternatives from 1 to 4, assigning a 1 to the alternative they most preferred and a 4 to the alternative they least preferred.

A pair-wise choice format presents a respondent with two alternative bundles and asks her to choose the bundle she likes better. The bundles include a dollar value or price along with other characteristics that differ across the bundles. Pair-wise choices utilize an ordinal selection method, i.e., choosing the more preferred bundle. This method differs from more typical conjoint methods that require respondents to rank multiple bundles or rate bundles using a cardinal scale to indicate intensity of preference (e.g., Johnson et al., 1995; and Roe et al., 1996). Examples of the pair-wise choice format include a study of landfill siting preferences by Opaluch et al. (1993), a comparison of stated and revealed preference models for a recreation site in Adamowicz et al. (1994), and a comparison of pair-wise choices and the referendum format in a study of moose habitat preferences (Adamowicz et al., 1996).

A bundle of attributes in our study comprised one of the preservation program options and a dollar amount or price. For example, Option A at $3, or Option B at $10 are bundles. Given a pair of bundles, e.g., Option B at $10 and Option C at $25, a respondent chooses whether she prefers Option B at $10, or Option C at $25. More complex bundles are common, e.g., bundles in Adamowicz et al. (1996) have seven attributes. However, the simplicity of our bundles avoids a potential criticism of more complex bundles. This criticism suggests that respondents might make their choices based on simplifying decision rules that ignore some attributes (Tversky, 1989).

We tested the four valuation formats in different focus groups and selected the pair-wise choice format for the survey based on the following evaluation criteria.

• How easy or difficult is the exercise for subjects? The referendum format is the easiest because respondents need only decide whether the good is "worth" the stated amount. The pair-wise choices were also easy for focus group participants. Participants found the payment card and the more complete ranking exercise hardest to complete.
• Which approach allows us to reasonably ask repeated valuation questions to the same subject to increase the amount of information we obtain from a given number of individuals? The pair-wise choice experiment format easily accommodates multiple valuation questions per respondent, e.g., Johnson et al., (1995) had respondents make 26 rating choices. Multiple referendum questions are also possible in terms of follow-up questions, but the number of WTP questions is significantly smaller. Payment cards are limited to one question per preservation program.
• Which approach would minimize the tendency for "cost calculation" that occurred in the focus groups? The term cost calculation refers to rejecting a dollar amount, which is less than or equal to willingness to pay, on the basis that if all households paid the amount, program funding would be excessive. The tendency to cost calculate seemed particularly strong with the referendum question. The payment card questions were also occasionally accompanied by comments like, "I think this is a fair amount for everyone to pay" and "It shouldn't cost more than this per person." The variation in prices in the pair-wise approach (as well as the accompanying instruction to focus on what the program would be "worth" to them) tended to focus respondents on the tradeoff between price and effectiveness, thereby reducing the tendency to calculate program costs.
• Which approach supports estimation of consumer surplus measures of value that are consistent with underlying utility theory? All of the approaches have been used to estimate consumer surplus and can be interpreted in ways consistent with utility theory. Payment cards provide the most straight-forward interpretation. Referenda have been widely used in contingent valuation. Pair-wise choices have been used, but not as frequently as the others.

Because the pair-wise choice format is a relatively new method for eliciting consumer surplus values, we added a payment card follow up as a second valuation format for comparison purposes. It should be noted that because the payment card follows the pair-wise choices, it does not provide a fully independent test of the two elicitation formats; it is possible that the values a respondent sees in the pairs might influence her response in payment card question. Ideally, a split sample approach would have been used, wherein some respondents answered the choice experiments and others answered the payment card. However, our anticipated sample of 275 individuals was not large enough to support a split sample.

Nevertheless, our choice of format should not be viewed as risky. Roe et al. (1996) suggest that binary choice conjoint methods such as the pair-wise choices used in our survey are similar to dichotomous choice methods. This suggests that the pair-wise choice methods should do as well as a dichotomous choice methods. However, they may also be subject to the same inherent problems. Several studies comparing WTP estimates across different hypothetical elicitation formats find that mean WTP from dichotomous choice questions are higher than mean WTP from open-ended questions (e.g., Kealy and Turner, 1993; and Brown et al., 1996). Brown et al. (1996) argued that plausible explanations for the higher dichotomous choice responses include "yea-saying" behavior (saying "yes" to reasonable values because actual WTP is uncertain) and conflicting response objectives (wanting to show support by saying "yes" versus wanting to say "no" to the stated value). We think that conjoint choice experiments may overcome these drawbacks by having respondents answer multiple valuation questions rather than one or two. Multiple response opportunities let a respondent show her support at prices she is willing to pay, while also showing that some prices are too high.

 

3.1.3

Payment Mechanism

Researchers agree that reliable valuation study should have a believable hypothetical market, which includes a plausible payment mechanism (see e.g., Fischoff and Furby, 1988; Mitchell and Carson, 1989; Arrow et al., 1993; and Hanemann, 1994). Earlier valuation studies using conjoint methods were able to incorporate user costs as payment mechanisms (e.g., Mackenzie (1993) used a hunting license fee, Adamowicz et al. (1994) used an entrance fee, and Roe et al. (1996) used travel cost). However, the nature of our preservation good precludes using a payment mechanism based on user costs. Furthermore, a tax payment mechanism caused respondents in focus groups to reject payment amounts for reasons that were not related to household WTP, e.g., total program costs could not justify per household costs of $x or more, or $x might be too much for poor households across the country. Consequently, we could not design a satisfactory mandatory payment mechanism, and a voluntary mechanism would invite embedding or free riding behavior.

In the end, we decided to explicitly state that the payment mechanism was undecided, and that the purpose of the survey was to find out how much preservation was worth to people. The survey emphasized that the respondent should focus on the trade-off between prices and preservation levels rather than who is paying what and how. The pair-wise choice elicitation format reinforced this trade-off. Instructions for the pair-wise choice treated the payment or "price" as mandatory:

For each alternative, you should imagine that the price is a one-time payment that your household would make if that preservation alternative were undertaken.

Our strategy was to make respondents feel as though the payment would be mandatory if a particular preservation option were undertaken, but avoid the problems associated with a tax mechanism. We included debriefing questions after the WTP section to detect the prevalence of embedding or strategic behavior. The responses to these questions indicate that most respondents did not engage in such behavior. About 90% of respondents said that they did not choose amounts larger than they would actually be willing to pay, and about 80% indicated that they were not free-riding¹.

Our payment mechanism did not provide information about how much other households and businesses would have to pay; rather we asked respondents to consider only how much their own household would want to pay. It is possible for a respondent to over or under estimate her WTP if she cares about how the program affects other individuals' utilities (i.e., if nonpaternalistic altruism is a motivating factor for her WTP), but does not have information about how those utilities change. (Madariaga and McConnell, 1987). McClelland et al. (1992) suggested that type of situation may lead to double counting if total benefits include bequest values for the current generation and benefits for future generations. Because we are limiting our benefit estimate to the current generation, this is not a concern for our study.

 

3.1.4

Expert Panel Review

The instrument developed during the five focus groups was peer-reviewed by NAPAP participants and nonmarket valuation experts. A number of remaining design issues were resolved as a result of the peer review.

• Should the study use two different injury baselines? Although focus group participants' reactions did not appear to differ across the baselines, differences might emerge in larger samples. One advantage of having two injury baselines would be that the research can address the uncertainty regarding the true baseline, which improves the usefulness of the research as the science improves on this issue. An important disadvantage was that having two baselines complicated the modeling process, and required more data. Furthermore, we determined that two baselines in a single session would lead to fatigue, so we would have to split the sample. Given the small survey sample and uncertainty about whether two baselines provided sufficient variation for modeling purposes, the reviewers recommended that only one baseline be used. They also recommended using the 75 year and 150 year baseline, which would provide a more conservative valuation context.
• Can the payment vehicle be left unspecified? The valuation difficulty encountered most often during the focus groups was cost calculation. Mandatory payment vehicles such as taxes exacerbated this behavior, so we decided against a mandatory vehicle. However, voluntary vehicles generally encourage free riding behavior, so we also wanted to avoid them as well. Consequently, we recommended that the survey explicitly state that the method of payment had not been determined, and encourage respondents to focus on their households' willingness to pay regardless of the payment method. The reviewers recommended wording changes for the valuation section that emphasized household willingness to pay and replaced "cost" with "price." We also added follow-up questions to ascertain which respondents still based their responses on expected program costs.
• Are pair-wise choices the best value elicitation format to use? The reviewers suggested further examination of possible alternatives such as full or partial ranking of the four preservation options (i.e., Options A through C and the no preservation option), or adding a third alternative - either "No Preservation" or "Indifferent" - to each pair. They also recommended that a simple version of the valuation model be tested using the pair-wise data from the focus groups. Our subsequent investigation of alternative methods in the valuation literature did not show them to be necessarily superior to the pair-wise approach, and the model estimated with the focus group data generated WTP values that were comparable to the payment card responses. Further, we concluded that the pair-wise choices were easy enough for respondents to complete, and that the number of choices could increase without risking respondent fatigue. Finally, the model we estimated using focus group data produced reasonable results. Based on these conclusions, we retained the pair-wise choice approach. However, we doubled the number of pairs in the survey instrument from 5 to 10 to increase the amount of data per respondent.
• Should the payment card question be retained? A payment card WTP question was included in the focus group questionnaires as an alternative valuation question format. It also served as a tool to collect WTP information that helped us choose prices for the pair-wise alternatives. There was some indication that the payment card might obtain lower WTP responses than the pair-wise choices. Retaining the payment card question as a follow-up question after the pair-wise choices in the final survey instrument was discussed as possibly providing useful information for comparison to the results of the pair-wise choices, but that interpretation problems might arise if we found significant inconsistencies between the two approaches. The reviewers concluded that the benefits of additional information outweighed the risks of ambiguous findings. It was recognized that including both types of questions for each respondent was not a truly independent test of the second elicitation format unless we split the sample and had part answer the pair-wise choices and the rest answer the payment card. However, resource limitations made a split sample infeasible.

 

3.1.5

Survey Mode: Group Interviews

We administered our survey in-person. However, to maximize the sample size, we asked respondents to attend group sessions rather than conduct the survey at residences. At these sessions, a survey administrator presented the background and visual information to the group, and was available to answer questions. However, each individual read and completed her or his own survey questionnaire.

These "group surveys" take advantage of some of the beneficial features of more traditional in-person surveys because an administrator is present to explain the topic, present visual material, answer questions, and make certain the questionnaire is complete. Yet, the self-administered response format decreases the likelihood of socially desirable response behavior, which occurs when individuals provide answers that they think are socially acceptable.

We selected the group survey administration method after careful consideration of three alternatives: mail, mail/phone combination, and in-person (using small groups rather than individuals). Extensive reliance on visual materials precluded phone-only surveys. The pair-wise choice format would also be difficult to administer over the phone without first mailing respondents written materials that described the preservation program options and their pair-wise choice sets.

To determine which method would best serve our survey needs, we compared the three alternatives based on the three criteria below. Table 3-1 summarizes the comparisons.

• Was the random sample spatially limited?
• Could the survey contain complex material such as multiple time lines?
• Was additional effort required to develop final survey materials?

The instrument developed during focus group work was too lengthy for a mail survey. The survey administrator's script, which presented important information about marble monuments, injuries, and preservation programs, was much too long for a standard-length mail survey. Furthermore, the survey contained more than 50 questions, including 12 valuation exercises and several open-ended questions. This is a large number of questions for a mail survey. Furthermore, some of the concepts in the survey such as the multiple time lines were potentially too complex or confusing for a standard-sized mail instrument. The survey could have been trimmed to a more manageable length, but doing so would have required more resources than were available.

Similarly, the mail/phone combination would entail a significant development effort to transform the survey administrator script and visual materials into a mailable information packet. The development needs of the mail/phone survey would be somewhat less extensive than the mail survey because the actual survey would have been administered over the phone, which lessened the risk of confusion somewhat. Nevertheless, the investment would have been significant, and was beyond the capabilities of the this study.

Because survey materials were essentially ready for in-person administration, and we thought that it was necessary to present the topic at the same level of complexity as in the focus groups, we selected the group survey option. Group administration was more cost effective than individual interviews because it saved a significant amount of time; 272 people were surveyed in 16 group sessions rather than 272 individual interview sessions. An important limitation of this approach is that sampling is subject to locational constraints because it is limited to individuals or households located near a survey location. Respondents must be willing and able to travel to the interview site, so survey participation is likely to be lower than for a mail survey. Consequently, it is important that the inducements to participate be attractive, and that the representativeness of the final sample be ascertained. Chapter 4 addresses these topics.

Another concern about group surveys is that interactions among respondents might lead to third party influences. For example, one respondent who comments out loud that the preservation program is a waste of money might influence the valuation choices of other respondents. Chapter 4 describes the procedures we implemented to minimize interactions between respondents.

 

3.2	Final Instrument

We designed our survey to obtain willingness to pay estimates for the benefits of slowing the rate of injuries to a discrete subset of marble monuments. The survey administrator followed a script (in Appendix B), to present information and materials that familiarized respondents with marble monuments in Washington, DC, and the general types of injuries caused by weathering. The survey administrator also presented the injury time line and photograph time series, which illustrated how the injuries to marble monuments in Washington, DC could be expected to worsen on average over time. Afterward, the moderator described three preservation program options that slowed the injury time line to varying degrees by partially protecting monuments from the weathering effects of air pollution.

Throughout the survey administrator's presentation, each respondent answered questions in her or his individual response booklet (Appendix B). These responses provided attitudinal data about topics such as whether monuments are important or whether individuals cared about weathering injuries. After the survey administrator explained the two pair-wise choice examples, the respondents completed the valuation and demographic sections of the response booklet, and the survey ended. In the valuation section, the respondents completed 10 pair-wise choices, answered a payment card question and a disembedding follow up question. They also answered several debriefing questions about their WTP choices. Each survey instrument section is described below.

 

3.2.1

Instrument Section A: Introduction to Outdoor Marble Monuments in Washington, DC

The final survey began by familiarizing respondents with monuments in general, and marble and other building materials used for monuments. Then it described the set of outdoor marble monuments in Washington, DC by using photographs of some of these monuments and providing a list of monuments with a map showing their locations in the city. Survey participants had an opportunity to review these materials, and responded to questions about whether they had visited monuments in Washington, DC, and if so, whether they were in Washington, DC on business or vacation during their most recent visit to the monuments. We asked the visitation question to find out whether visiting the monuments affected willingness to pay.

 

3.2.2

Instrument Section B: Nature of Value of Monuments

Understanding why people value monuments helps understand the potential loss from air pollution injuries. Typical reasons given by focus group participants for valuing monuments fell into one of three categories: historical or educational value, aesthetic value, and cultural or patriotic value. Focus groups also indicated that passive use values such as bequest value tended to be more important than their own direct use value. The final survey contained a number of rating questions to obtain information from respondents on how important to them are various listed reasons for having monuments. We asked these questions to determine whether attitudes toward the existence and importance of these monuments differed in any systematic way such as across age groups or between visitors and nonvisitors, and to learn more about why people care about monuments.

 

3.2.3

Instrument Section C: Injuries to Monuments over Time

This section first introduced respondents to two injury concepts, erosion and chemical alteration, which were described using a combination of text and photographs. It is important to note that although we are using the term injury here for clarity purposes, the survey did not use the term "injury." Instead, it referred to "changes in the appearance" of the monuments.

Following the general description of injuries and before any mention of an injury baseline, the survey response booklet asked respondents to indicate the time frame over which they cared about the appearance of monuments. This question provides attitudinal information about what time frames people think about when they consider the future physical condition of these monuments. Including such questions allows us to test whether willingness to pay is sensitive to such time frames. Respondents also read a list of reasons why some people might care about changes in the appearance of monuments, and rated the importance of each reason to themselves on a scale of 1 to 5. The list of reasons generally reflected the reasons for valuing monuments in survey Section B. Similar to the responses in Section B, these ratings can be used to determine if answers differ systematically with demographic characteristics.

A complete description of the preservation program required that respondents first understand how the appearance of the monuments would change over time without the program. We accomplished this by showing them a baseline injury time line accompanied by photographs that illustrated current conditions for a statue and a common architectural element (i.e., a column capital), and expected average conditions of the marble monuments in Washington, DC, in 75 and 150 years. Respondents learned that the photographs represented the average condition of the marble monuments, and that natural and pollution-related weathering effects would vary for individual monuments depending on a number of factors such as the age of the monument and its exposure to the environment. They also saw a large poster illustrating the time line associated with the photo series (see Figure 2-2).

Finally, respondents learned that there were several mitigation strategies for the injuries, ranging from general maintenance to active protection, and they ranked these options. This exercise reminded individuals that there are substitutes for a preservation program, and it obtained their preferences over responses prior to introducing the program. Respondents were also asked whether they thought that replacing a monument (with an accurate replica) could fully restore its original value, or if some value was lost when the original monument was gone. Answers to this question provide some insight into how the public views the importance of the original material and the nature of damages such as whether part of the cultural value is embodied in the original material.

 

3.2.4

Instrument Section D: Preservation Program Options and Description

We then contrasted the baseline injury time line with the longer injury time lines achieved by three preservation program options. Each option's time line was described three ways. First, we noted the resulting percentage increase in time before the illustrated levels of injury would occur (a 25% increase for Option A, a 50% increase for Option B, and 100% increase for Option C). We also explained how the percent increase lengthened the injury time periods in terms of years; e.g., Option A's 25% increase moved the illustrated level of injuries from the original time line of 75 and 150 years to 95 and 190 years, respectively. Finally, we illustrated these changes in the number of years before each level of injury would be reached using large posters with time lines and photographs of the statue and column positioned on the time lines to visually illustrate the original injury timing and an option's effect of slowing down the rate at which injuries occur. Figure 2-2 contains a small replica of the poster board for Option A, and Appendix B shows replicas for all options.

The survey administrator described the preservation program options as a one-time "light" cleaning, followed by a coating treatment that partially protects the monuments from air pollution. Focus group research showed that people believed the various coating treatments would be effective, but they preferred that some cleaning precede coating applications. We told respondents that the initial cleaning did not improve the monuments' appearance, but allowed the coating to adhere to the surface. The coating was said not to harm the monument. Respondents were told that multiple coating applications were no more effective than the original application to prevent respondents from thinking that two applications of the Option A coatings (25%) were as effective as one Option B coating (50%). Finally, the survey said that the treatments could only be performed once, i.e., future generations could not repeat the treatment to generate additional benefits. This approach avoided complicating the WTP analysis with concerns such as double counting benefits, because future generations cannot pay for themselves. It raises the issue of how altruistic motives affect WTP responses. Some focus group respondents expressed a concern about tax mechanisms because they did not want to force poor people to pay for preservation. This concern with others' well-being indicates that altruistic motives are partly individualistic, which means that costs incurred by other people affect an individual's own WTP (Madariaga and McConnell, 1987). Consequently, our WTP estimates could under estimate (or over estimate) actual WTP to the extent that payments by others increased (or decreased) an individual's own WTP.

Throughout the focus groups and scoping surveys, respondents were receptive to the idea of a coating treatment and seemed to accept the plausibility that such a procedure existed and could achieve the changes in rates of injury as described. Occasionally, concerns arose about whether the coating was toxic, or whether it would harm the monuments. Our final description was worded to allay likely concerns about the initial cleaning and coating treatment and to focus respondents' attention as quickly as possible on the changes in the time lines rather than the details of the treatment artifice.

We also addressed the issue of substitutes for the monuments and for the preservation program so that their willingness to pay responses would more accurately reflect the value of the preservation options in light of available substitutes. In describing the good, we reminded respondents that only the 100 marble monuments were included in the program; several hundred DC monuments made of other materials would not be included in the preservation program. Recall from Section 3.2.3 that we also listed alternative methods for responding to monument injuries such as replacement or removal.

After describing the preservation program options, we included a rating to help respondents place the preservation good into perspective. The rating exercise listed 10 issues of varying importance, from "maintaining economic growth" to "using energy-efficient compact florescent light bulbs," and included a preservation program for marble monuments in Washington, DC. The respondents rated each issue on a scale of 1 to 5, where 1 meant not at all important and 5 meant very important. The value of an exercise such as this is that it moderates any strong emotional responses that respondents have after hearing a lot of information about injuries to monuments by reminding them that there are many issues such as improving education and reducing crime that are more important. Thus, it offsets tendencies to overestimate willingness to pay. The ratings also show us where monument preservation lies on along a spectrum of social issues.

We did not include a lot of detail regarding how a preservation program would be implemented. The cleaning/coating treatment processes were described in general terms, and only distinguished by the different levels of protection they achieved. The survey noted that the program details had not been finalized, and that the survey was part of the planning process because planners wanted to find out whether the public thought that preserving the monuments was important. However, the survey noted that the program would be administered by the Marble Monument Trust, which was described as a nonprofit organization specializing in monument preservation. We used a nonprofit agency because earlier focus group respondents voiced suspicion about whether the federal government could accomplish the preservation program, or whether funds would be spent appropriately.

 

3.2.5

Instrument Section E: Valuation of Preservation Program Options

In this section, respondents completed 10 pair-wise choices, a willingness to pay payment card question, and several followup questions regarding their valuations responses.

The survey reminded respondents several times that the payment for the pair-wise choices and the payment card was a one-time payment made by their household. We chose to use a lump-sum payment rather than periodic (e.g., monthly or annual) payments mainly to reinforce the idea that the preservation program was a one-time program; i.e., the coating treatment would not be repeatedly applied over time. A drawback of using a lump-sum payment mechanism is that it may result in lower WTP values; respondents may give lump-sum WTP values that are lower than the present value of periodic WTP values using market discount rates. Nevertheless, we thought the pair-wise choices would be easier to perform with a lump-sum payment rather than a periodic payment because the respondent already had to make intertemporal comparisons between preservation options.

Prior to answering the valuation questions, survey participants were reminded that the preservation program only involved the approximately 100 marble monuments in Washington, DC, and that about 350 other Washington DC monuments made of other materials would not be affected by the program. Respondents were also reminded that the dollar amounts they selected represented one-time payments that would reduce household expenditures on other goods and services.

Each pair-wise choice showed the respondent a pair of alternatives, and the respondent was instructed to select the alternative that she preferred. An alternative was a combination of one of the preservation options (Option A, Option B, Option C, or No Option) and a dollar value that the household would pay as a one-time payment for that option. The dollar values could be selected from the following: $0 (for No Option only); $0.25; $1; $3; $7; $10; $15; $25; $50. As an example, a respondent might see a pair like:

Preservation Option C for $25

or

Preservation Option B for $7

We designed 20 versions of the survey such that all plausible combinations of alternatives appeared on at least one version. Plausible combinations excluded pairs with more effective options for a lower dollar amount, e.g., Option B at $3 and Option A at $7. However, a survey version could show Option B at $3 in one pair and Option A at $7 in another pair. Each of the 20 versions had five pairs containing the No Preservation at $0 alternative, which was paired with an Option A, Option B, and Option C alternative at least once. Each version's remaining five pairs combined Options A through C, with pairs A & C, A & B, and B & C, appearing at least once in every version.

To help respondents understand the selection process, the pair-wise section began with two example pairs that were presented by the survey administrator. The first pair comprised Option C at $25 and No Preservation Option at $0. The No Preservation Option referred to the baseline injury time line. The second example pair had Option B at $10 and Option C at $15. For each pair, the survey administrator provided reasons for why people might select either option. By including these reasons, we let participants know that there were good reasons for selecting either alternative in a pair. Figure 3-1 shows the survey administrator's instructions for the example pairs.

After the example pairs, respondents were instructed to complete the remainder of the response booklet by themselves. First they completed the ten pair-wise choices, and provided written explanations for why they selected some answers. They then answered a follow-up question about their response strategy, indicating whether they always chose the most effective option, the least expensive option, or whether they chose the most appealing alternative based on the price and effectiveness trade-offs.

The next survey question was a payment card question that asked respondents to select their maximum willingness to pay for Preservation Option C rather than have no preservation option. Figure 3-2 shows the instructions in the response booklet for the payment card.

As Figure 3-2 illustrates, respondents were reminded that the purpose of the exercise was to find out how much their households were willing to pay as a one-time payment for a particular preservation option, regardless of how much other households would pay and how funds would be raised. Similar reminders preceded the pair-wise choices as well.

Following the payment card question was an open-ended question that respondents used to explain why they selected a particular value from the payment card. Then respondents provided additional insight into their valuation rationale by rating how closely 12 debriefing statements described their thoughts at the time they were making valuation choices. Examples of these statements include: "My household should not have to pay for this type of preservation;" "I don't like the programs because they include monuments that I think are not important;" "I have doubts about the information presented;" and "The amounts I chose reflect what I think to be my fair share of the costs of the Preservation Options." Responses to these statements help to interpret the respondent's WTP responses.

The valuation section concluded with a verification question for the payment card exercise. This question asked a respondent to consider whether the value she selected from the payment card might be more or less than her household's actual maximum willingness to pay for the preservation program, and gave some reasons for why that might be the case:

• Sometimes people select an amount greater than the most they would be willing to pay for Preservation Option C for these 100 marble monuments because they are thinking, for example, about more monuments or more effective preservation programs.
• Sometimes people select an amount less than the most they would be willing to pay for Preservation Program C because they are thinking, for example, more about the "fair share" for each household to pay than about the value of Preservation Option C to their household.

If a respondent indicated that she had over or underestimated willingness to pay, then she completed a new payment card and an open-ended question explaining the reason for the change.

 

3.2.6

Instrument Section F: Demographic Information

Following the valuation exercises, respondents provided sociodemographic information about themselves (age, gender, and educational attainment), and about their households (size, number of children, and annual income). This information was also obtained in the telephone recruitment survey, but we included it in the in-person survey because some respondents were reluctant to provide personal information over the telephone. Demographic characteristics are important determinants of WTP. For example, willingness to pay may be constrained by ability to pay (i.e., household income).

1. In the debriefing section, respondents indicated how strongly each attitudinal statement reflected their thinking while completing the valuation section on a scale of 1 (Does Not Reflect My Thinking) to 5 (Closely Reflects My Thinking).The percentages noted here are based on the respondents selecting either a 1 or a 2 as their response.

Contact me via email | Document Index | Home Page | Top of Page

Last Update: 1-9-98