You are here:
Verbal Protocol Analysis of Cost Valuation Responses
The purpose of the project is to understand better how individuals interpret and respond to contingent valuation (CV) questions. The research will address three issues: the reliability of the referendum questions format, the important of reminding respondents about substitutes, and the sensitivity of CV estimates of the scope of potential natural resource injuries. An interdisciplinary team of economists and psychologists will direct the research. A verbal protocol methodology adapted from the field of cognitive psychology will be used to provide insights into thought processes that lead to observed CV response. These protocols involve the process of respondents "thinking aloud" while they answer the CV questions. This process yields insights on both the decision process that people use to answer questions and the information they use as part of the decision process. Protocols will follow a research design that will enable the results to be useful for both hypothesis testing and developing more general insights about people's response patterns. Surveys will be used to further test the findings of the verbal protocol research.
R822446-01Principal Investigators: Desvousages, WilliamTechnical Liaison:Research Organization: Research Triangle InstituteFunding Agency/Program: EPA/ORD/ExploratoryGrant Year: 1994Project Period: 1994-1996Cost to Funding Agency: $238,510
- Project Reports
- Executive Summary from Final Report
How do people interpret and respond to contingent valuation (CV) questions? How are the psychological processes that lead to a CV response affected by how you ask the valuation question and the sequence in which you ask the question? How stable are expressions of environmental valuations, including willingness-to-pay (WTP) amounts, across environmental goods? Should WTP amounts be viewed as "economic values" or as an expression of an attitude that is constrained to be on a dollar scale? How do people cope with emotionally difficult tradeoffs such as degree of environmental protection versus cost? In a world where expressions of preferences are often constructed, how should such preferences be measured? These are just some of the questions examined in the research supported by Environmental Protection Agency grant number R824310, which is summarized in this Final Technical Report.
The theoretical framework for the research was that expressions of preference, e.g., CV responses, are constructed – not revealed - at the time a valuation question is asked. Consequently, statements of preference will be highly sensitive to task variables such as response mode. The empirical approach was to combine both experimental methods and the process tracing method of verbal protocol collection in the investigation of preferences across a variety of public policy and consumer domains.
The following experiments were completed. In the first experiment, respondents were asked to value a natural resource preservation plan to protect salmon stocks in the Pacific Northwest. The respondents were randomly assigned to one of four value elicitation (response) formats: dichotomous choice, open-ended willingness-to-pay, and two payment card conditions anchored by low or high amounts. One fourth of the respondents in each response mode condition provided verbal protocols while generating their judgments. Second, a study was done where each respondent used multiple response modes to value a series of five different environmental goods. The goods included visibility improvement in the Grand Canyon National Park, protection of migrating waterfowl, establishment of oil spill response centers, the reintroduction of the red wolf into the Great Smoky Mountains National Park, and the preservation of salmon stocks in the Pacific Northwest. The five response modes were (1) open-ended willingness-to-pay, and ratings of (2) importance relative to other problems in society, (3) seriousness relative to other environmental problems, (4) use value and (5) existence value. Respondents were drawn from two geographical areas, Raleigh, NC and Austin, TX. A third set of experiments asked people to choose among a variety of goods that varied in terms of attributes such as safety and cost. The attributes were varied in terms of emotional tradeoff difficulty. Finally, a series of experiments were done that examined the relationships between monetary expressions of attitudes, and other
ways of expressing an attitude.
The major findings from the research include the following: First, a variety of psychological considerations and processes influence WTP amounts (e.g., fair share considerations), some of which are not consistent with the common economic assumptions about the basis of CV responses. Second, exactly how you ask the CV question matters substantially, both in terms of WTP amounts and the considerations influencing those amounts. In particular, dichotomous choice WTP amounts were significant higher than those obtained with other response modes. Third, the sequence in which a good is valued, given a set of goods to be valued, matters greatly. A good that is valued first in a sequence tends to viewed in a separate not a comparative sense, and therefore tends to be assigned a much larger WTP amount. Fourth, there is some significant stability of individual preferences among a set of environmental goods; although, the use of a WTP scale to capture such preferences is not as useful as other forms of attitude scales. Fifth, people clearly find tradeoffs between some pairs of attributes more emotionally difficult than tradeoffs between other pairs of attributes. People cope with more emotional tradeoffs by adopting lexicographic choice strategies that help avoid such tradeoffs. Sixth, people are better described as having stable attitudes, rather than stable economic values, in the domain of public policy concerns. Attitude scales often do better at reflecting those attitudes than WTP amounts.
Overall, the research findings support the constructed view of expressed preferences. Some guidelines (a "building code") for measuring constructed preferences are suggested. The focus of the guidelines is on doing more value assessments with fewer respondents. This reflects the belief that it is systematic error (i.e., cognitive process bias) rather than random error than impacts most constructed expressions of preference. The guidelines also stress the need to provide respondents with tools for thinking about values in addition to information about public policy problems and proposed solutions.
- Project Status Reports
This project was transferred to Grant R824310 with a project period of October 1, 1995 to April 30, 1998 and a new title: How People Respond to Contingent Valuation Questions.
Objectives of Research:
The purpose of this project is to increase our understanding of how people interpret and respond to contingent valuation (CV) questions. The theoretical framework is that preferences for objects of any novelty and complexity are often constructed – not merely revealed – in the generation of a response to a valuation task.
A series of experiments explore how CV responses are sensitive to the question format, the nature of the environmental good being evaluated, difficulties in making tradeoffs, and the importance of reminding respondents about substitutes. A special feature of the research is the use of the verbal protocol methodology to provide greater insights into the cognitive processes leading to an observed willingness-to-pay response. The results of the project should advance our understanding of contingent valuation reasoning and contribute to the development of the verbal protocol methodology as applied to CV research. More generally, the project is expected to yield insights into the reliability and validity of expressed valuation methods as a tool in making natural resource decisions.
The following major studies (projects) have been completed or are underway. Listed under the description of each major study or project are the related publications and presentations.
Study 1. Elicitation Format and Constructive Preferences: A Verbal Protocol Analysis of the Willingness-to-Pay (WTP) for the Preservation of a Natural Resource.
An experiment has been completed in which 604 respondents were asked to value a natural resource preservation plan to protect salmon stocks in the Pacific Northwest. The respondents were randomly assigned to one of four elicitation (response) formats: dichotomous choice, open ended willingness-to-pay, and two payment card conditions anchored by low or high amounts. Twenty-six percent of the respondents in each response mode condition provided verbal protocols while generating their judgments. Respondents also answered a number of manipulation check, attitude, and demographic questions.
Preliminary analyses indicated the following. First, the estimated willingness-to-pay amounts varied greatly with response mode, in particular, the dichotomous choice response mode generated significantly higher WTP amounts. Second, the initial dichotomous choice bid had a significant, and positive, effect on the WTP responses. Third, a wide variety of thoughts seemed to underlie the WTP responses. Fourth, the frequency of different types of thoughts varied by response format. For example, 25% of the respondents in the dichotomous choice format mentioned concerns for broader environmental issues, while only 13% mentioned such concerns in the other response modes.
Presentations / Papers. A presentation of based on this study was given at Camp Resources, August, 15, 1997. A paper based on this study is in preparation.
Study 2. Trade-off Difficulty and Choice.
A series of experiments were conducted on how people cope with more difficult trade-offs, such as cost versus safety considerations in the purchase of a good, and the impact of such coping on observed choice patterns. This work was done in the domain of consumer decisions. Across three different experiments that involved more than 160 subjects it was found that people increasing used a more lexicographic type choice strategy that emphasized the quality versus the price (cost) attribute when the quality attribute was rated as more inherently emotion-laden. The implications of attribute-level trade-off difficulty when interpreting choice data are discussed in a paper based on these experiments.
Presentations / Papers. A presentation based on this study was given at the Subjective Probability, Utility, and Decision Making conference held in August, 1997, in Leeds, England. A paper based on this study has been completed and submitted for journal publication.
Study 3. The Stability of Preferences Across Environmental Goods and Response Modes.
This study involves approximately 300 respondents using a mixture of response modes to value a series of five different environmental goods. The goods included visibility improvement in the Grand Canyon National Park (Air), protection of migratory waterfowl in the Central Flyway, preservation of salmon stocks in the Pacific Northwest, establishment of oil spill response centers, and the reintroduction of the red wolf into the Great Smoky Mountains National Park. The response modes are either a contingent valuation response (WTP) or a set of general scales, e.g., the importance of the problem. A key, and innovative, feature of the experimental design is that each respondent judges all five commodities using WTP in one session and then judges all five commondies using the rating scales in another session. This design enables a richer and more direct assessment of preference reliability and validity than in previous CV studies. Further, half the respondents are from the Raleigh area of North Carolina and half are from the Austin area of Texas. This design feature allows us to test the effects of distance from the good (red wolf reintroduction or bird protection) on WTP responses.
Progress. This data collection part of this experiment is finished. Analysis of the data is ongoing.
Project 4: Measuring Preferences in a Constructive World: Towards a Building Code.
A paper that argues that a constructive view of preferences requires a fundamentally different approach to the measurement of preferences than that which is implied by the more traditional view of preference expression as revealing well-defined preferences has been completed. In this paper the outline of a "building code" for preference construction and measurement is offered. Including in that building code is the need to provide respondents with the tools for thinking about difficult tradeoffs. Also discussed in that paper is how a "building code" for constructing preferences must be sensitive to the purpose of the valuation exercise, e.g., whether measurement is for the prediction or the design of future decisions.
Presentations / Papers. A version of this paper was presented at a National Science Foundation symposium on the elicitation of preferences at UC Berkeley in August, 1997. The paper was also presented at the subjective probability, utility, and decision making conference held in August, 1997, in Leeds, England. The paper is under review at a journal. Finally, the ideas in this paper are also contained in a second paper that offers a comprehensive review of consumer choice behavior. That paper is also under journal review.