Introduction To Conjoint Analysis For Valuing Ecosystem .
Technical Memorandum Number EC 2008-03Introduction to Conjoint Analysisfor Valuing Ecosystem AmenitiesU.S. Department of the InteriorBureau of ReclamationFebruary 2008
Mission StatementsThe mission of the Department of the Interior is to protect andprovide access to our Nation’s natural and cultural heritage andhonor our trust responsibilities to Indian Tribes and ourcommitments to island communities.The mission of the Bureau of Reclamation is to manage, develop,and protect water and related resources in an environmentally andeconomically sound manner in the interest of the American public.
Technical Memorandum Number EC 2008-03Introduction to Conjoint Analysisfor Valuing Ecosystem AmenitiesDavid A. HarpmanTechnical Service CenterEconomics and Resource ManagementU.S. Department of the InteriorBureau of ReclamationFebruary 2008
AcknowledgementsThis document has benefited immeasurably from thepatient technical review and gracious assistance ofseveral of my colleagues. In alphabetical order, they are:Dr. J. Michael Bowker, U.S. Forest ServiceDr. Michelle A. Haefele, The Wilderness SocietyDr. Richard J. Lichtkoppler, U.S. Bureau of ReclamationDr. John B. Loomis, Colorado State UniversityDr. Bruce E. Peacock, National Park ServiceMr. Jonathan L. Platt, U.S. Bureau of ReclamationDr. Michael P. Welsh, Christensen AssociatesI’d also to thank the Winter Quarter 2008 EPM 4130Environmental Economics class at the University ofDenver for their many helpful comments.Any remaining errors are the sole responsibility of theauthor.Financial support for this project was provided by the U.S. Bureau ofReclamation’s Manuals and Standards Program.
ContentsContentsPageContents . iiiPurpose. 1What is Conjoint Analysis?. 1Synonyms . 1Nonmarket Goods . 2Stated and Revealed Preference . 2Some Definitions. 3Attribute . 3Level . 3Profile. 3Orthogonal . 3Origins and History . 4Steps in a CA Study . 5Characterize the Problem. 5Identify the Relevant Population . 5Attributes and Levels . 5Experimental Design. 6Full Factorial. 6Fractional Factorial . 7Randomized . 7Survey Development. 7Elicitation Formats. 8Collect Data . 9Estimate the Model . 9Interpret the Results . 9Economic Valuation Example. 9Background and Setting. 10Attributes. 11Cost . 11Humpback chub population index . 12Sediment quantity index . 12Attribute Levels . 12Program cost . 12Humpback chub population. . 13Sediment index. 13The Relevant Population. 13Experimental Design. 13Dichotomous Choice Conjoint. 14Survey Design. 14iii
ContentsWhat is Measured?. 15About the Data . 15Parameter Estimation . 16Estimated Logit Equation . 18More Notation. 18Graphing the Function . 19Consumer Surplus (CS) Measures . 20Confidence Intervals . 22Example Policy Analysis . 23No Action Consumer Surplus . 24Alternative X Consumer Surplus . 25Incremental Benefits . 26Interpretation of Results. 27Advantages of CA . 27Disadvantages of CA. 28Conclusions. 28Literature Cited . 29Appendix 1. The RUM Model. 33Appendix 2. Full-Factorial Design. 35Appendix 3. Total Economic Value. 39Description. 39TEV and Glen Canyon Dam . 39Appendix 4. Additional Background . 41The 1996 Glen Canyon Dam EIS . 41The LTEP EIS. 42More Recent Events. 43Appendix 5. Conditional Mean WTP. 45TablesTable 1.Table 2.Table 3.Table 4.Descriptive Statistics for Example Data . 16Measures of Consumer Surplus . 21Benefits of Alternative X. 27Historical and MLFF Operating Criterion . 41FiguresFigure 1. Example Conjoint Scenario. 15Figure 2. Plot of estimated probability function . 20Figure 3. Conditional consumer surplus . 22Figure 4. The benefit of increasing the chub population index is the differencebetween the areas under the two curves. 26iv
ContentsBoxesBox 1. Logit Regression Results for the Example. 17Box 2. Confidence Intervals for Conditional Mean Consumer Surplus . 23v
Conjoint AnalysisPurposeThe purpose of this document is to convey a conceptual and analyticunderstanding of the conjoint analysis (CA) methodology used for valuingecosystem amenities. An example application is described and solved in a stepby-step fashion. While by no means an exhaustive treatment of the subject, someof the difficulties and associated pitfalls are described. A number of usefulreferences are furnished for further study.What is Conjoint Analysis?In recent years, conjoint analysis (CA) has been employed to estimate the neteconomic value of natural resource amenities. This approach has its origins inbusiness marketing research and there are many applications in this context.Conjoint analysis is based on a primary survey of individuals utilizing a carefullydesigned survey instrument. Respondents are presented with differenthypothetical situations, described using their characteristics or attributes. andasked either to rank them or choose between them. Using the resultant surveydata, the probability that an individual will rank or choose any particular scenariois then estimated. The consumer surplus or net economic value of the amenitycan then be derived.SynonymsThere are an amazingly large number of terms for conjoint analysis. The usage ofthese terms seems to vary by discipline, with the context and nature of theapplication. The vast majority of business and marketing applications employ theterm conjoint analysis while economic studies use a variety of alternatedescriptors. Commonly encountered synonyms for conjoint analysis include thefollowing terms; contingent ranking (CR), attribute-based methods (ABMs),stated preference choice experiments (SPCEs) and choice-based experiments(CBEs).Because of economist’s focus on economic welfare and willingness to pay (WTP)measures, Holmes and Adamowizc (2003) suggest their use of attribute-basedmethods differs from other applications of conjoint analysis. Their view is hardlyuniversal however and the use of synonyms remains commonplace and confusing.1
Conjoint AnalysisNonmarket GoodsValues for goods traded in the market are called market values and are thetraditional measure of value associated with changes in water resourcemanagement. Familiar water resource examples are irrigation benefits andhydropower benefits. Values for goods which are not traded in the market (andthus not observable) are called nonmarket values. These may include changes inthe quantity and quality of recreation or changes in the intrinsic value of aresource.Recreation use is a commonly cited example of a nonmarket good. Certain typesof recreation uses, such as fishing and hunting, are termed consumptive uses. Acharacteristic of consumptive use is that once a good is used by one individual, itis unavailable for use by another individual. For example if a recreational anglercatches and keeps a fish, that fish is unavailable for other anglers to catch.Some recreation use activities, such as hiking, are termed nonconsumptive uses.Hiking, bird watching, wildlife viewing and similar activities do not require theconsumption of a resource. In the absence of crowding, other individuals can useor share in the use of the resource without diminishing it.Nonuse values are a special case in which the nonmarket good is the status of thenatural or physical environment. Nonusers, or individuals who never visit orotherwise use a natural resource may nonetheless be affected by changes in itsstatus or quality. Monetary expression of their preferences for these resources isknown as nonuse or passive-use economic value. Economists also use the termspassive-use value and intrinsic value to describe these preferences.Stated and Revealed PreferenceThere are two major classes of techniques for measuring the value of nonmarketgoods. These are the revealed preference approach and the stated preferenceapproach. Revealed preference approaches are based on the observed behavior ofconsumers. The observed behaviors reflect the decisions which people makeregarding activities that utilize or are affected by an environmental amenity.Reveal preference studies typically focus on measuring economic use value. Incontrast, stated preference methods elicit values directly from individuals, throughsurvey methods. The stated preference methods are suitable for measuring bothdirect use and nonuse or passive use values.2
Conjoint AnalysisSome DefinitionsAs with any topic, there is a unique vocabulary associated with conjoint analysis.Some commonly encountered terms are defined and a brief explanation isprovided below.AttributeAn attribute is characteristic or feature of a good which is of importance toconsumers when they make expenditure decisions. In the case of a vehiclepurchase decision, for example, pertinent attributes may be such things as theprice, the color, the number of doors, the size of the engine, off-road capabilityand so forth. For a natural resource amenity, characteristics such as the cost, thelevel of crowding, the catch rate, available access, and the view-shed in thesurrounding area may be pertinent attributes.LevelAttributes can be numerically described using levels. In the case of a vehicle, forexample, the size of the engine can be described using horsepower. For a naturalresource amenity, such as a fishing experience, the catch rate can be characterizedby the number of fish of a certain size which are landed per unit time.ProfileA short description of a hypothetical good using its attributes and the levels ofeach of those attributes is known as a profile. Synonyms include the termsalternative and treatment combination. Since many conjoint experiments are oneat-a-time choices, the term profile is more widely employed.OrthogonalStatistically, two vectors are said to be orthogonal or uncorrelated if their innerproduct is the zero vector (or matrix). This indicates the cross correlation of eachelement in the two vectors is zero. Typically, conjoint experiments areconstructed to ensure the levels of their attributes across profiles are orthogonal.3
Conjoint AnalysisOrigins and HistoryConjoint analysis is typically thought of as arising from business marketingresearch. As some authors point out though, marketing researchers borrowedheavily from earlier economics research. This includes the discrete choiceeconometrics work by Daniel McFadden, the 2000 Nobel Prize winner ineconomics (Orme 2006).In the business marketing context, students are often taught the term “conjoint”refers to respondents evaluating features of products or services, “CONsideredJOINTly” (Orme 2006). Several authors have suggested the term actually derivesfrom the verb “to conjoin” meaning “joined together” (Orme 2006).Conjoint measurement methods were first described in the mathematicalpsychology literature by Luce and Tukey (1964). Drawing on their work, PaulGreen applied the concept to complex purchasing decisions and the prediction ofbuyer behavior. He and coauthor Rao subsequently published what is recognizedby most authors to be the seminal article on the topic of conjoint analysis (Greenand Rao 1971).Early conjoint analyses were based on the so-called “full-profile” approach.These were typically implemented by using specially designed conjoint carddecks. Each card in the deck described a product profile. Respondents wererequired to sort the deck from the most desirable profile to the least desirableprofile. The size of the deck reflected an orthogonal design and increased rapidlywith the number of attributes and the number of levels. For practical reasons, thislimited the number of attributes and levels which could be investigated.Researchers soon found that better response data could be obtained by askingrespondents to rate (on a scale of, for example, 1 to 10) the desirability of eachcard.Richard Johnson (1974) invented a clever method of making pairwise trade-offswhich is used in experiments to this day. This allowed respondents to focus ontwo attributes at a time. Johnson formed one of the preeminent firms in this field,Sawtooth Software, and went on to develop a process called adaptive conjointanalysis (ACA). The ACA approach dynamically narrows the number of conjointquestions posed to a respondent based on the pattern of their previous responses.In the late 1980s, the evolution of conjoint analysis drew upon the emerging fieldof discrete choice analysis pioneered by McFadden and others. Discrete choicemethods allow conjoint questions to be constructed in a manner that is morerealistic and natural to respondents. Although the associated econometricmethods are much more complex, it can allow for a more rigorous modeling ofattribute interactions.4
Conjoint AnalysisMore comprehensive and informative accounts of the history and evolution ofconjoint analysis can be found in Orme (2006), Holmes and Adamowizc (2003)and other sources.Steps in a CA StudyAlthough there are many variations in approach, a number of practitioners seemto agree on an eight-step approach to conjoint analysis. These eight steps aredescribed below.Characterize the ProblemThe first step in undertaking a conjoint analysis is identifying the problem andcharacterizing its salient features. For a traditional marketing study, this mightinclude identifying the focus of the exercise in terms of product features,packaging or price and how that might effect market share or total productpurchases. In the natural resource economics context, the analyst should identifythe geographic scope and the range of economic values potentially affected bychanges in amenity services.Identify the Relevant PopulationAn important aspect of any primary survey exercise is the identification of thepopulation which could be affected by the proposed management action. Theidentified relevant population forms the sample frame which should be targetedby the survey effort. For less well known but locally important
Reveal preference studies typically focus on measuring economic use value. In contrast, stated preference methods elicit values directly from individuals, through survey methods. The stated preference methods are suitable for measuring both direct use and nonuse or passive use values. 2
Conjoint analysis is a statistical analysis technique that has been used in the field of mar- . Sawtooth Software, Inc. offers several conjoint packages including Conjoint Value Analysis (CVA), Choice-Based Conjoint (CBC) analysis, and Adaptive Conjoint Analysis (ACA). We elected to us
more often than nonmetric conjoint analysis. In the SAS System, conjoint analysis is performed with the SAS/STAT procedure TRANSREG (transformation regression). Metric conjoint analysis models are fit us-ing ordinary least squares, and nonmetric conjoint analysis models are fit using an alternating least squares algorithm (Young, 1981; Gifi, 1990).
Conjoint Analysis The commands in the syntax have the following meaning: ¾With the TITLE - statement it is possible to define a title for the results in the output window ¾The actual Conjoint Analysis is performed with help of the procedure CONJOINT. There are various subcommands within this procedure:-The PLAN subcommand tells CONJOINT .
SPSS 8.0 is a powerful software package for data management and analysis. The Conjoint option is an add-on enhancement that provides a comprehensive set of procedures for conjoint analysis. The procedures in Conjoint must be used with the SPSS Base system and are completely integrated into that system. The Conjoint option includes:
Conjoint analysis is sometimes referred to as “trade-o ” analysis because respondents in a conjoint study are forced to make trade-o s between product features. In this sense, conjoint analysis is able to infer the “true” value structures that influence consumer decision making
Bayesian Conjoint Analysis, Choice-Based Conjoint, Water Park, Pricing 1. Introduction Since the 1970s conjoint analysis has been intensively used by tourism managers to design and price competing multi-a
of conjoint analysis methods have been proposed - see for example (Sawtooth Software). Since the early 1970s conjoint analysis continues to be a very popular approach with hundreds of commercial applications per year (Wittink and Cattin 1989). In conjoint analysis designing questionnai
Bruksanvisning för bilstereo . Bruksanvisning for bilstereo . Instrukcja obsługi samochodowego odtwarzacza stereo . Operating Instructions for Car Stereo . 610-104 . SV . Bruksanvisning i original
