RESEARCH DOSSIER
Economic Preference Elicitation and Incentive-Compatible Research
Abstract. Economic preference elicitation translates decisions, bids, prices, and stated valuations into empirical evidence about choices and welfare-relevant judgments. This critical narrative research monograph reconstructs the historical development of incentive-compatible measurement from auction theory and the Becker–DeGroot–Marschak mechanism to modern experimental field protocols. It distinguishes formal truth-telling incentives from practical behavioral validity, compares choice and pricing tasks, examines willingness-to-pay and willingness-to-accept asymmetries, and evaluates hypothetical bias, risk and time preference estimation, belief-reporting rules, and random-payment designs. Original laboratory and field experiments are assessed alongside methodological analysis and quantitative evidence synthesis, including recent tests of payment architecture and distribution-sensitive bidding. The central finding is methodological: elicited values depend jointly on the intended construct, incentive mechanism, comprehension, preferences over uncertainty, framing, and research context. The monograph develops an evidence-sensitive framework for choosing elicitation methods, interpreting discrepancies, conducting preregistered comparisons, and studying valuation under human–artificial decision mediation. This is a critical narrative synthesis of existing scholarship; it introduces no new experiment, systematic-review protocol, or quantitative meta-analysis.
Keywords: economic preference elicitation; incentive compatibility; mechanism design; revealed preferences; stated preferences; Becker–DeGroot–Marschak mechanism; BDM; Vickrey auction; willingness to pay; willingness to accept; experimental economics; consumer valuation; preference reversal; hypothetical bias; risk elicitation; time preference; random incentive systems; field experiments; construct validity; psychometrics; Artificial Era
Economic Preference Elicitation as Scientific Measurement
Economic preference elicitation is the deliberate production of observable responses from which an investigator infers the relative attractiveness, monetary valuation, or temporal and risk properties of alternatives. A response may be a purchase at an offered price, a bid in an auction, an allocation across lotteries, a choice between delayed payments, or an answer to a carefully specified contingent-valuation question. Its scientific object is an inference from situated action to a construct whose meaning must be fixed in advance. A willingness-to-pay statement, for example, measures the maximum monetary sacrifice a person is prepared to make for a specified outcome under the operative budget, information, ownership, and transaction rules. It cannot be interpreted without those conditions. The methodological foundations in Becker et al. (1964) and Carson & Groves (2007) show why elicitation requires a response rule as well as a theory of what the response represents.
An experiment creates a measurement environment rather than simply recording an already completed market transaction. Participants encounter descriptions, prices, deadlines, probability displays, experimenter instructions, payment conditions, and expectations about what subsequent decisions will require. Each can affect attention, comprehension, strategy, or the value of the object. An observed bid therefore contains at least four distinguishable components: the participant's context-dependent evaluation, the incentive implications of reporting a number, errors or misunderstandings, and the design-specific transformation from evaluation to action. Consumer research on constructed choice Bettman et al. (1998) makes explicit that values may be assembled as people compare alternatives; the existence of a formal payment mechanism alone does not remove that psychological process.
The central research questions concern identification, validity, and transportability. Under what assumptions does a bid reveal the latent valuation the analyst wants? Which empirical patterns would demonstrate failure of those assumptions? Do mechanism variants preserve rankings, population means, or entire demand curves? Will the same instrument work with a novice purchaser, an experienced trader, and a participant facing uncertainty about their household resources? This monograph critically integrates theoretical mechanism analysis, original laboratory and field experiments, comparative elicitation studies, methodological critiques, and quantitative syntheses. It assesses findings according to design, construct, implementation, and population. It is a narrative scientific synthesis, with no preregistered literature-search protocol, newly collected observations, or newly estimated meta-analytic effect Harrison et al. (2004) Harrison & List (2004).
Historical Foundations of Revealed Preference and Experimental Valuation
Classical economic models often begin with choices made under constraints and infer a consistent ordering of alternatives. Their empirical challenge is that market behavior entangles preferences with income, prices, available options, transaction costs, habits, and information. Experimental economics changed this identification problem by designing transactions in which the response itself was rewarded according to specified rules. Vickrey's analysis of competitive sealed tenders Vickrey (1961) established a second-price logic under which a bidder's own report can determine allocation without determining the price conditional on winning. Becker, DeGroot, and Marschak Becker et al. (1964) subsequently supplied a practical single-response method for measuring a person's reservation value through a randomly determined transaction price.
These innovations answered different questions. An auction mechanism was intended to allocate goods and establish outcomes in a strategic environment; a BDM elicitation procedure was designed to elicit a single individual's monetary value without requiring an external rival. Both use an independence principle: the reported amount influences whether a transaction takes place, while a separate rule determines the payment. Their intellectual significance lies in replacing an open-ended verbal question with an experimentally implementable payoff schedule. Yet the theoretical result is conditional on the incentives and behavioral assumptions encoded in that schedule, including stable utilities, comprehension, and an absence of payoff-relevant features beyond the specified purchase Vickrey (1961) Becker et al. (1964).
A distinct psychological line emerged when choice and pricing tasks produced different orderings of gambles. Lichtenstein and Slovic Lichtenstein & Slovic (1971) described preference reversals in which selecting one lottery while assigning a higher price to another could coexist within the same person. Grether and Plott Grether & Plott (1979) subjected the phenomenon to experimental-economic scrutiny, seeking explanations in procedure, incentives, and theory. Their contribution made measurement disagreement a central substantive phenomenon: a finding can reveal the boundary of a preference model as well as an imperfection in an instrument. The history of elicitation is therefore the joint history of increasingly precise incentive design and increasingly demanding tests of whether elicited responses remain invariant across permissible descriptions and tasks.
Valuation, Utility, Demand, and the Construct Being Measured
A reservation price is a decision threshold defined for a particular transaction. For a buyer, willingness to pay is the highest price at which acquiring a specified item remains preferable to retaining the relevant money, given the conditions of the offer. For a seller, willingness to accept is the minimum compensation for surrendering an endowed item. The two measures are indexed to different reference positions and can differ because of wealth effects, reference-dependent evaluation, uncertainty, or implementation. An empirical investigation must therefore state whether its target is one person's threshold, a distribution of thresholds, a demand curve at posted prices, or an estimate of compensating variation for a policy. The reported dollar value is a construct-specific measurement, not a context-free cardinal unit of psychological utility Kahneman et al. (1990) Plott & Zeiler (2005).
Preference strength is also distinguished from the decision process that yields a response. Participants may attend to the most prominent feature, construct a temporary ranking, simplify probability information, or use cues from the range of prices provided by the experimenter. Bettman, Luce, and Payne Bettman et al. (1998) analyzed adaptive construction in consumer decisions; their approach predicts that elicitation tasks can recruit different cognitive computations even when they nominally target the same good. A preference inferred from a bid is consequently meaningful at the level of the specific task and can be compared across tasks only after the analyst specifies a measurement-invariance hypothesis.
Demand is a relationship between quantities or purchase probabilities and prices. A single WTP observation can contribute to demand estimation, but heterogeneity, multiple units, substitution, and budget constraints complicate the translation. Experiments that compare BDM reports with purchase decisions at individually randomized fixed prices directly test the empirical correspondence rather than assuming it Cole et al. (2020). The relevant estimands include a mean WTP, a median reservation price, a survival function of reported values, an actual purchase probability at each experimental price, and variation across populations. These quantities must be kept distinct: a mechanism may approximate average willingness to pay while distorting the tails of the distribution or the shape of estimated demand.
Formal Incentive Compatibility and Truthful Revelation
A mechanism is incentive compatible relative to a specified environment when responding according to the designated truthful strategy is optimal under the participant's assumed preferences and information. In a standard private-value second-price sealed-bid auction, truthful bidding is weakly dominant when each bidder values the object independently of the bidding report and the conventional payoff assumptions hold. A bidder who raises a report above the true reservation value risks winning at an unwanted price; a bidder who lowers it risks losing a profitable purchase. The rule matters because the amount paid by a winning participant is linked to another bid rather than their own. Vickrey's original formal analysis Vickrey (1961) establishes the strategic architecture that later elicitation studies adapt.
For a buyer in the Becker–DeGroot–Marschak procedure, the participant reports a maximum price b, an independent selling price p is drawn from a stated distribution, and the participant purchases at p if the qualifying condition p ≤ b is met. If the person's reservation value v is fixed and the participant wants to purchase at p exactly when p ≤ v, reporting b = v ensures that all and only desirable transactions take place. Reporting above v introduces a region of unwanted purchases; reporting below v sacrifices transactions that would have yielded positive utility. The proof is an argument about price contingencies and a fixed preference ordering. It is not a claim that all actual bidders report correctly or that their v is free of framing effects Becker et al. (1964) Harrison et al. (2004).
Truthful response is weak rather than strict at some values when price distributions have gaps or boundaries; different reports may generate the same allocation throughout the supported range. Risk attitude need not independently destroy the basic purchase-threshold logic under appropriately specified monotonic preferences, but income effects, learning costs, multi-unit valuation, inferred information from the price distribution, and repeated decisions can alter the objective. A precise methods section must state which assumptions yield the incentive-compatibility claim and whether the implemented experiment satisfies them. Carson and Groves Carson & Groves (2007) similarly show that the incentive properties of stated-preference questions depend on consequentiality and institutional circumstances rather than on attractive phrasing alone.
Identification, Behavioral Assumptions, and Participant Understanding
Identification asks whether distinct underlying values could produce the same observed response and whether the response function is known well enough to recover the intended construct. A price drawn only within a narrow interval may fail to distinguish valuations above its upper bound; participants with different high values make the same purchase decision at all prices they might face. Payment caps, endowments, exchange restrictions, and reporting grids can create additional censoring and discretization. The analyst should document the eligible price support, participation constraint, cash and good endowments, timing, and treatment of boundary bids. These design features shape both the empirical estimand and the degree of information carried by each observation Harrison et al. (2004) Wertenbroch & Skiera (2002).
Strategic truthfulness is only as operational as the participant's understanding of it. Subjects may mistakenly believe that a higher bid automatically makes them pay more, that a reported number is a suggestion rather than an executable contract, or that the investigator expects a socially desirable answer. Such beliefs can change reports even while the theoretical mechanism remains sound. Plott and Zeiler Plott & Zeiler (2005) explicitly investigated procedural misconceptions and training in the willingness-to-pay versus willingness-to-accept literature. Their evidence motivates comprehension checks, practice transactions, clear explanation of the independent price draw, and attention to the possibility that extensive instruction itself changes which task participants understand or value.
Statistical randomization addresses selection into experimental conditions but cannot by itself repair an invalid mapping from response to preference. Preregistration should identify the target construct, primary comparison, admissible bids, mechanisms, price supports, manipulation checks, and rules for excluding incomplete observations; interpretation should then separate deviations from truthful strategy, context-sensitive valuation, and ordinary response noise. The same observable misbid can arise from a misunderstanding, a changing value, or rational effort allocation. Recent work on BDM reliability Mamadehussene & Sguera (2023) illustrates the distinction by modeling how experimentally imposed price distributions can change the incentive to learn one's own preference, even when the basic truthfulness logic remains intact.
The Becker–DeGroot–Marschak Procedure in Laboratory Practice
The BDM mechanism creates an individual transaction in which a person announces a bid and faces a randomly determined price. The laboratory implementation must specify whether the item is purchased, sold, or exchanged, how the random price is generated, and which payment and possession rights become binding. In an ordinary buyer version, the independent price is drawn after a bid is recorded; if it falls at or below the bid, the buyer acquires the item at that drawn price. This yields an interpretable reservation threshold when the assumptions established by Becker et al. (1964) are satisfied. Because the mechanism does not require competing buyers, it is suitable for repeated product valuations, isolated consumer assessments, and settings where coordination of an auction market would be expensive.
Price support influences what can be observed and what participants may infer. Uniform sampling between declared lower and upper bounds gives a straightforward representation of transaction probabilities but may suggest an implicit reference range. A price outside a psychologically plausible interval can make a participant question the mechanism, and a restricted interval cannot precisely identify values beyond its limits. Even formally equivalent price distributions can change the expected value of deliberation. In a theoretical and experimental analysis, Mamadehussene and Sguera Mamadehussene & Sguera (2023) argue that a participant who must invest effort to discover a preference can rationally respond differently to the price distribution because the possible benefits of learning differ across distributions. Treating these findings as mere violations of rationality would miss the relevant research question.
Training and practice should target understanding of allocation and payments rather than coaching an investigator-preferred bid. A useful demonstration compares a true reservation threshold with several hypothetical reports, shows which prices would lead to purchase, and clarifies that the price paid is drawn independently. Comprehension items can test whether a high stated bid changes the price, whether a participant can decline a qualifying purchase, and whether the mechanism actually implements the stated consequence. These are procedural variables that require reporting and, where possible, experimental manipulation. The comparisons of BDM and auction procedures in Noussair et al. (2004), together with methodological arguments in Harrison et al. (2004), show why the elicitation name is insufficient documentation of the implemented treatment.
Second-Price Auctions and Strategic Price Independence
A sealed-bid second-price auction makes a bidder's own report consequential for winning while, conditional on winning, determining payment from another participant's bid. Under the familiar single-unit independent-private-values model, bidding one's reservation value is a weakly dominant strategy: understating value may forfeit a profitable outcome, and overstating value may result in purchase at a price above value. The strategic incentive differs from BDM because competitors jointly generate the price and can introduce uncertainty about others' behavior. The central result established by Vickrey (1961) remains a foundation for designing and interpreting incentive-compatible auction experiments.
An experiment using the Vickrey format still needs to disclose the number and identities of participants, the unit of allocation, tie-breaking, payment rules, and whether rival bidding can be observed across rounds. Strategic complexity can distract from the preference task when participants mistake a second-price mechanism for a first-price auction and shade bids toward an expected market-clearing price. Social motives and beliefs about other bidders can also matter when the utility of winning or losing is relational. The laboratory is therefore testing the combined environment of valuation, competition, and institutional understanding. A comparison with BDM identifies what the mechanism changes only if good, endowment, information, and transaction timing are otherwise held comparable Noussair et al. (2004) Harrison et al. (2004).
In a study comparing consumers' willingness to pay, Noussair, Robin, and Ruffieux Noussair et al. (2004) report differences between the BDM mechanism and the Vickrey auction under their particular implementation, showing that institutional form and training can affect revealed valuations. That finding warrants mechanism-specific replication rather than a universal ranking of elicitation devices. The proper scientific comparison reports not only average bids but also their distribution, response at treatment boundaries, comprehension, willingness to transact, and correspondence with independent fixed-price choices. A procedure that produces the highest apparent bids is not thereby the best measure of valuation; accuracy requires an external criterion, an induced value, or a defensible structural model.
Random nth-Price Auctions and Broader Participation
Standard second-price auctions may provide weak incentives for low-ranked bidders to reveal values when those bidders believe they have little prospect of trading. The random nth-price auction modifies the allocation procedure by selecting a trading threshold from participant ranks, with the objective of making marginal valuation consequential over a wider distribution of bidders. Shogren, Margolis, Koo, and List Shogren et al. (2001) formulated and examined this mechanism as an experimental elicitation alternative. Its purpose illustrates a general principle: a mechanism can be theoretically truth revealing yet provide unequal amounts of practical engagement, learning, or feedback across participants.
Random nth-price designs demand especially clear descriptions of how many people trade and at what price. Participants must understand that the random rank and the clearing rule determine the outcome; otherwise differences across auction formats may reflect strategic misunderstanding. A meaningful comparison evaluates incentive properties conditional on the specified utility model, then separately measures real comprehension and bidding performance. In studies with unfamiliar consumer goods, practice rounds, monetary endowments, and exposure to previous trades can create learning effects that interact with the formal allocation rule. Such details belong in the experiment's reproducibility materials and causal analysis Shogren et al. (2001) Harrison & List (2004).
The method also emphasizes the relation between a sample statistic and the individual valuation it is meant to reveal. An auction with several bidders can estimate willingness-to-pay distributions among participants who understand the procedure, but recruitment and nonparticipation may make those bidders unrepresentative of consumers in the target market. Raising the chance of trade for low-ranked participants may help reveal values within the sample while leaving external validity unresolved. The larger methodological task is to separate properties of the transaction rule from properties of the participant population, including market experience, liquidity, product familiarity, and expectations about future opportunities to buy List (2003) Cole et al. (2020).
Real Payments, Endowments, and the Design of Consequential Tasks
Economic incentives require credible consequences. In a real-payment experiment, the participant receives or forfeits actual money, obtains the good, or participates in an enforceable future transaction according to the mechanism. A hypothetical task may communicate the same choice in words but does not impose the immediate budget constraint. A physical product may also carry additional utility when it can be handled, consumed, or resold, so real payment and real possession should be considered independently. Wertenbroch and Skiera Wertenbroch & Skiera (2002) developed point-of-purchase willingness-to-pay measurement in a consumer setting, demonstrating the importance of designing tasks close to the transaction being inferred.
An initial experimental endowment can influence reports if participants perceive it as earned income, unexpected bonus money, reimbursement, or property already incorporated into their budget. Whether the endowment can be taken away, converted into goods, or supplemented from personal funds changes feasible decisions. Mechanism instructions therefore should specify cash balance, market price, ownership rights, exchange opportunities, and final settlement. These considerations are particularly relevant when comparing willingness to pay with willingness to accept, because the researcher has altered the participant's reference position by assigning possession or money. The endowment-effect experiments Kahneman et al. (1990) and the later procedural investigation Plott & Zeiler (2005) indicate that endowment interpretations and elicitation rules require joint assessment.
The practical importance of payments is an empirical question rather than a doctrinal assumption. Holt and Laury Holt & Laury (2002) examined incentive magnitude in risky decisions, while the later study of payment mechanisms by Drichoutis, Palma, and Feldman Drichoutis et al. (2026) used induced-value tasks to compare payment conditions with details of the offered price range. Their published two-experiment report found that some architectural features affected misbidding even when studied payment differences were not detectable. This finding is specific to the tested designs and outcomes; it does not establish that hypothetical answers everywhere carry the same information as real transactions. Every applied study should state what kind of consequentiality it actually tested.
Laboratory Standardization and Experimental Software
Standardized laboratory implementation is essential when differences of a few monetary units are interpreted as differences in preference. Computerized experimental systems can hold instructions, randomization, lottery generation, time stamps, eligibility, payment formulas, and feedback constant across participants. Fischbacher's z-Tree system Fischbacher (2007) became influential as a practical platform for interactive economic experiments. A digital protocol nevertheless remains a measurement intervention: screens differ in salience, participants differ in computer familiarity, and technical failures can censor responses selectively. Source code, screenshots, random-seed practices, and reproducible allocation scripts provide a more informative record than the name of the software alone.
The experimental unit must be explicit. If subjects interact in auction groups, responses may be dependent because rival bids and market histories are shared. Treating all individual bids as independent observations can underestimate uncertainty. When random prices are drawn individually, transactions may be independent in the mechanism yet correlated through laboratory sessions, experimenter effects, or repeated decisions from the same person. A valid analysis accounts for the design's level of assignment, clustering, repeated-measures covariance, and the number of independently randomized units. Between-subject and within-subject design choices are reviewed by Charness et al. (2012), who emphasize the tradeoffs between precision and order or carryover effects.
Pretesting can identify whether the mechanism achieves its intended informational purpose before the main study. A pilot may include induced monetary values known to the researcher, comprehension questions, response-time checks, and a comparison of participant choices to the predicted transaction threshold. Field and laboratory performance can then be judged with prespecified accuracy metrics instead of a post hoc preference for whichever procedure produces the most convenient distribution. The account of field experiments by Harrison & List (2004) encourages documentation of how the laboratory's controlled mechanism interacts with real-world environments. The resulting design principle is reproducibility of incentives and observation, coupled with explicit tests of the human understanding on which those incentives operate.
Preference Reversal Between Choice and Pricing
A preference reversal occurs when different elicitation tasks imply incompatible orderings of the same alternatives under an assumed context-independent preference relation. In the classic gamble example, a person chooses a lottery with a comparatively high chance of a moderate payoff yet assigns a larger monetary selling price to a lottery with a smaller chance of a high payoff. A single stable ranking combined with consistent monetary evaluation would ordinarily predict agreement between choice and pricing under the relevant assumptions. Lichtenstein and Slovic Lichtenstein & Slovic (1971) documented reversals experimentally and initiated an enduring debate about response scales, task-sensitive attention, and rational choice.
Choice asks for a categorical comparison; pricing asks for a monetary equivalence. The numerical requirement in the latter may draw attention toward a gamble's maximum payoff or induce anchoring on plausible dollar amounts, while direct choice may make the chance of a favorable outcome more salient. Grether and Plott Grether & Plott (1979) investigated whether institutional features, incentives, or theoretical deficiencies could explain preference reversals. Their research illustrates an essential analytic distinction: a contradiction between two measured outcomes may arise because one task fails to reveal the intended latent ordering, because the person changes how value is constructed, or because the original single-ranking model inadequately describes decision behavior.
A strong experiment must hold stakes, endowments, gamble representations, and information constant while varying the response operation. It should counterbalance task order and measure comprehension rather than treating all reversals as identical errors. A reversal rate is sensitive to the chosen lottery pairs, the monetary elicitation rule, exclusions for indifference, and the handling of equal bids. Separately reporting the direction of reversal, individual variability, and magnitude of disagreement helps discriminate mechanisms. The scientific value of preference reversals lies in identifying where an inference from choice to valuation becomes task dependent, not in assigning a global degree of rationality to participants Lichtenstein & Slovic (1971) Bettman et al. (1998).
Preference Construction, Framing, and the Psychology of a Bid
Constructed-preference models treat valuation as a process that responds to goals, task demands, memory, attention, and the available alternatives. A consumer may hold relatively stable priorities yet assemble a specific monetary judgment only when asked to price a particular good. Bettman, Luce, and Payne Bettman et al. (1998) describe constructive consumer choice as an adaptive process in which decision strategies vary with complexity, accuracy goals, effort, and affective consequences. A bid can therefore be reproducible in one arrangement while shifting systematically when the comparison set, response scale, or reason for choosing changes. These shifts are empirical phenomena that a measurement theory must explain.
A formal incentive mechanism cannot ensure that the participant's evaluation process remains unaffected by its instructions. If an experiment presents prices ranging from a very low to a very high amount, participants may infer product quality, retail expectations, or experimenter norms from that range. Those inferences can influence the valuation the mechanism seeks to elicit. Mamadehussene and Sguera Mamadehussene & Sguera (2023) offer an account in which the price distribution changes incentives to invest cognitive effort in learning preferences, showing that distribution effects do not necessarily refute incentive compatibility. The distinction is between the truthfulness of reporting a value after deliberation and the endogenous determination of that value during deliberation.
Measurement practice should specify what invariance is scientifically expected. If research concerns a consumer's evaluation within a realistic store environment, responses to a different comparison set may represent real context sensitivity. If research claims to measure a general risk coefficient, instability under small and substantively irrelevant presentation changes challenges construct validity. Task variations should be theoretically justified, randomized, and examined with interaction tests and within-person consistency estimates. This approach places constructive choice research beside the formal analysis of truthful revelation: they investigate successive stages of how an observed answer is produced Bettman et al. (1998) Carson & Groves (2007).
Willingness to Pay, Willingness to Accept, and Ownership Effects
Willingness to pay and willingness to accept use distinct reference transactions. In a WTP task, a participant trades money for an item; in WTA, a participant trades an item for money. Their difference may reflect ordinary wealth effects, reference-dependent preferences, transaction uncertainty, attachment to ownership, or the institutional framing of a sale. Kahneman, Knetsch, and Thaler Kahneman et al. (1990) used experimental exchange and valuation tasks to investigate the endowment effect and challenged the prediction of symmetric valuations for certain goods. This research made procedural control and psychological reference points central to the interpretation of prices that participants were willing to offer or demand.
Plott and Zeiler Plott & Zeiler (2005) developed an influential procedural critique, arguing that aspects of training, misunderstanding, and elicitation can generate or amplify observed WTA–WTP gaps. In their implemented protocols, carefully designed instruction and practice altered the reported asymmetry. Their later examination of exchange asymmetries Plott & Zeiler (2007) extended the debate to the interpretation of apparent preference-based anomalies. These studies do not make every ownership effect disappear; they demonstrate the necessity of separating substantive reference dependence from possible artifacts of transactions, instructions, and response tasks. A comparison across studies must therefore report the full protocol rather than classifying findings solely by whether a gap was present.
Field evidence can test whether experience changes these outcomes. List List (2003) studied market experience and apparent anomalies, while later work List (2011) investigated market experience in a design that sought stronger causal identification. Both encourage a distinction between individuals with established trading practices and novices placed in unfamiliar experimental transactions. Experience may change understanding, search costs, attachment, strategic expectations, and the perceived normality of exchange. Researchers should define ownership clearly, measure trading familiarity, compare equivalent goods, and avoid interpreting the raw subtraction WTA minus WTP as a unique psychological mechanism without adequate controls.
Hypothetical Bias and Consequential Stated Preference
A hypothetical stated-preference question asks what a respondent would do under a described choice without necessarily carrying out the associated transaction. It can produce information about public goods, proposed services, and future products that are impossible to price directly in the market. The inferential hazard is that survey responses may be weakly constrained by actual purchasing power, attention, social expectations, or knowledge of the institutional decision. The relevant issue is therefore not simply whether money changed hands, but whether the respondent believes the answer can affect an outcome and whether the decision problem corresponds to the intended welfare question Carson & Groves (2007).
Murphy, Allen, Stevens, and Weatherhead Murphy et al. (2005) synthesized empirical work comparing hypothetical and actual stated valuations and documented substantial heterogeneity in hypothetical bias. Their meta-analysis drew on 28 studies and 83 observations; its reported median hypothetical-to-actual value ratio was approximately 1.35, with skewness and variation across designs. The median should not be presented as a universal correction factor. Product, elicitation method, respondent population, and payment conditions can jointly affect the difference. This is an evidence example of why a pooled estimate does not substitute for validating the chosen instrument in the substantive market where it will be used.
Carson and Groves Carson & Groves (2007) formalized incentive and informational properties of questions, placing emphasis on whether respondents understand the stated institutional consequences. A consequential referendum-like choice under a credible provision rule can have a different strategic structure from an open-ended question asking for a speculative number. Researchers should specify the provision scenario, payment vehicle, cost allocation, and credibility of implementation, then pretest beliefs about the decision's relevance. Such transparency supports interpretation of what was elicited and exposes when a convenient survey instrument measures hypothetical approval or expressive support rather than monetary reservation value.
Consumer Demand and Contextual Field Valuation
Consumers make real purchase decisions amid budgets, product bundles, social signals, and uncertain future needs. A laboratory bid isolates an experimentally manageable valuation but may omit these contextual determinants. Field studies move the transaction closer to practical behavior while introducing other sources of variation, including local availability, retailer trust, stockouts, weather, and market selection. Wertenbroch and Skiera Wertenbroch & Skiera (2002) examine willingness-to-pay measurement at the point of purchase, illustrating the ambition to connect a theoretically interpretable bid with consumer decisions made in an ordinary purchasing context.
The comparison between stated valuations and randomized posted prices is especially informative. At an offered price, the buyer either transacts or does not; if treatment assignment makes that price exogenous, the researcher can estimate a purchase probability for each price or a demand relationship. Cole, Fernando, Stein, and Tobacman Cole et al. (2020) conducted two field experiments involving agricultural information and rainfall insurance, comparing BDM demand elicitation with choice at individually randomized fixed prices. Their results varied across the two contexts, providing an empirical test of correspondence rather than assuming that BDM can automatically substitute for observed buying at a posted price.
External validity depends on the characteristics of both the target good and the decision maker. Insurance involves probabilities, trust, and contingent benefits; information services involve expectations about how knowledge will affect later productivity or behavior. Different cognitive demands can produce different mechanism performance even when the same formal incentive logic is used. A defensible generalization identifies which features were common across field implementations and which require replication: literacy, opportunity to deliberate, available liquidity, reference prices, and access to payment. The principle is that validity must be demonstrated for the claimed decision environment, with cross-study heterogeneity treated as information about mechanisms and populations Cole et al. (2020) Harrison & List (2004).
Eliciting Risk Attitudes With Monetary Lotteries
Risk preference is frequently estimated by observing choices between lotteries that differ in payoff distributions. Multiple price lists present a sequence of paired risky options and interpret the point at which a person switches as evidence about risk tolerance under an assumed utility model. Holt and Laury Holt & Laury (2002) developed an influential lottery-choice paradigm and studied the effects of incentive size on revealed risk aversion. This approach differs from asking a participant to announce a purchase bid for a product: a lottery task jointly engages attitudes toward risk, comprehension of probabilities, utility curvature, and the temporal framing of payment. The estimate is consequently conditional on the chosen parametric or nonparametric interpretation of lottery choices.
A switch point does not uniquely identify a stable, general risk trait. A respondent may misunderstand the randomization, apply probability weighting, treat experimental stakes as separate from household wealth, or vary attention when comparing near-indifferent options. Charness, Gneezy, and Imas Charness et al. (2013) review experimental elicitation methods and emphasize that different tasks can have distinct strengths and weaknesses. The investigator should report the probability display, amounts, choice set, admissible switch patterns, response editing rules, and treatment of apparent multiple switching. A risk coefficient derived from one instrument should be tested against alternative outcomes before being interpreted as a transportable personal disposition.
Survey-based risk questions provide another route when incentivized lotteries are infeasible. Dohmen and colleagues Dohmen et al. (2011) investigated self-reported attitudes toward taking risks and their relationships with incentivized tasks and consequential behaviors. Their research motivates multimethod validation, but correlation does not establish interchangeable measurement. Hypothetical tolerance for risk in general life, aversion to a specified financial loss, and curvature of utility over experimental stakes are different variables. The scientific aim is to determine which one predicts the outcome of interest and whether the prediction is stable across stakes, domains, and demographic groups.
Time Preference, Discounting, and Joint Identification
Intertemporal preference tasks ask participants to choose between outcomes delivered at different dates. An apparently high discount rate can reflect impatience, limited liquidity, distrust of delayed payment, expected inflation, risk of nonpayment, or valuation of money under changing circumstances. Andersen, Harrison, Lau, and Rutström Andersen et al. (2008) showed the importance of eliciting risk and time preferences jointly when interpreting experimental discounting, because utility curvature changes the mapping from observed intertemporal choices to estimated discount parameters. Under a nonlinear utility model, inferring time discounting while assuming linear utility can yield an inaccurate structural estimate.
Payment implementation is particularly consequential in delayed-reward tasks. If a participant receives immediate cash in one condition and a promised transfer later in another, the two options differ in settlement risk and convenience as well as delay. The experiment must specify identical payment channels, credible commitment, timing windows, and handling of failed transfers. An incentive-compatible task with respect to a stylized discounting model may therefore identify a composite of time, uncertainty, and trust unless those dimensions are controlled or independently measured. Researchers should define the estimand before choosing a discount function and compare model-implied choices with held-out decisions Andersen et al. (2008).
Temporal and risk estimation also reveal a general limitation of treating the participant's choices as direct readings of numerical parameters. The same data can often be fitted by alternative functional forms with different implications beyond the observed options. Extrapolating a discount rate estimated over a short interval to long-term retirement saving or public policy requires assumptions about functional stability and context. The appropriate analytic record includes model fit, uncertainty intervals, sensitivity to utility and probability assumptions, out-of-sample predictions, and heterogeneity in payment credibility. Comparable critiques apply to risk preference scales reviewed by Charness et al. (2013) and Dohmen et al. (2011).
Probability Beliefs and Proper Scoring Rules
A forecast about the likelihood of an event is a belief, whereas a willingness-to-pay bid is a valuation conditional on beliefs and preferences. The two should be measured separately whenever an experimental explanation requires identifying both. Proper scoring rules link a reported probability to an eventual monetary payoff and can encourage honest probability reporting under specific assumptions. Yet risk preferences affect how the participant values the scoring-rule payments, so strict propriety under a risk-neutral expected-payoff criterion may not translate into truthfulness for every risk-averse subject. Offerman, Sonnemans, van de Kuilen, and Wakker Offerman et al. (2009) analyze corrections for risk attitudes in scoring-rule elicitation.
The distinction matters for insurance, environmental risk, health decisions, and algorithmic advice. An increase in the bid for an insurance contract can reflect a higher subjective chance of loss, higher aversion to that loss, stronger trust in the insurer, or changes in expected liquidity. Inferring risk preference from WTP alone would conflate these determinants. Combining belief elicitation, structured lottery tasks, and randomized information treatments can improve identification, provided the additional measurements do not substantially change beliefs or preferences through the act of questioning. The study of field demand by Cole et al. (2020) illustrates why insurance provides a demanding context for evaluating elicitation mechanisms.
A credible design records both the elicitation rule and the theoretical conditions under which the report is meaningful. Participants should know whether probability reports are scored with a binary outcome, how payments are calculated, whether forecasts are independent, and whether the experiment can verify the event. In analysis, separate calibration and discrimination of belief forecasts from their strategic truthfulness and from a participant's utility function. This conceptual separation helps prevent a generic claim of 'incentive compatibility' from masking distinct mathematical requirements for beliefs, prices, and lottery choices Offerman et al. (2009) Carson & Groves (2007).
Random Incentive Systems and Repeated Experimental Decisions
Experiments often collect many decisions but pay for only one randomly selected decision. The intention is to make each response consequential while avoiding aggregate income effects and complex portfolio incentives. The random incentive system is sometimes justified by an expected-utility argument in which an independently chosen decision is paid and the participant treats each potential payoff accordingly. Its behavioral performance depends on what subjects believe about random selection, whether they integrate decisions into a broader mental portfolio, and the utility framework used. Harrison and Swarthout Harrison & Swarthout (2014) discuss the methodological issues raised by alternative payment protocols.
Paying every round can make stakes more credible or encourage learning but may create wealth changes, hedging opportunities, and dependence across choices. Paying one selected round limits some of these effects while potentially reducing attention if selection feels remote. A between-subject comparison of payment regimes can test whether differences appear in bids, violations of known values, switching patterns, response times, or attrition. Drichoutis, Palma, and Feldman Drichoutis et al. (2026) directly compared multiple variants in induced-value tasks, finding that some payment features predicted less than theoretically incidental aspects of choice architecture. Their result is evidence about the implemented experimental environments and does not by itself resolve the optimal protocol for all populations.
Selection probabilities and sampling procedures should be specified in plain language and implemented reproducibly. A participant who expects hidden discretionary payment or doubts the random selection will face a different incentive system from the one described in the methods paper. Experimental reports should include the actual number of paying decisions, whether order was randomized, whether participants learned outcomes as they proceeded, and whether the same subject could receive multiple products. Transparency is especially important for field experiments where prizes, electronic transfers, and delayed settlement may introduce compliance differences. These are measurement conditions that can be studied causally with randomized protocol variation Harrison & Swarthout (2014) Charness et al. (2012).
Psychometric Reliability and Cross-Task Measurement Invariance
Reliability concerns how much of the variation in repeated measurements reflects reproducible differences rather than measurement noise. A single BDM bid recorded once can be consequential and yet provide limited information about test–retest stability. Repeating the task allows estimation of within-person dispersion, but repetition also creates learning and anchoring, and immediate retesting is not equivalent to observing preferences weeks later. For risk and time tasks, the object of measurement must be fixed before evaluating reliability: stable rankings of alternatives, a structural parameter, or prediction of consequential behavior are different targets. The multimethod evidence discussed in Dohmen et al. (2011) and Andersen et al. (2008) motivates this distinction.
Cross-task validity asks whether different instruments preserve a theoretically specified construct. If BDM, a Vickrey auction, and posted-price purchase decisions are intended to recover the same reservation-value distribution, researchers can test agreement in level, rank, and shape of demand. They should predefine acceptable disagreement and consider whether task order alters reported values. The Noussair et al. (2004) comparison shows the relevance of mechanism and training; the Cole et al. (2020) field comparisons show the relevance of product and setting. Agreement in group means may coexist with large individual-level errors, so average correspondence and individual concordance must be analyzed separately.
Statistical precision does not guarantee construct validity. A laboratory can produce narrowly dispersed bids because its price range encourages anchoring, because participants coordinate on a salient benchmark, or because an item truly has similar value for everyone. Reliability estimates need a design that can distinguish these mechanisms, such as randomized price supports, counterbalanced protocols, repeated observation, induced-value calibration, or prediction of externally observed transactions. Evidence about systematic differences across task versions should be retained rather than averaged away. A measurement instrument earns a broad validity claim through converging, discriminating, and consequential evidence, not merely through the incentive label attached to its instructions Mamadehussene & Sguera (2023) Bettman et al. (1998).
Comparative Evidence for Mechanism Performance
Choosing a mechanism requires evidence about its performance against the researcher’s target quantity. Induced-value experiments are powerful for detecting reporting errors because a participant's monetary value is set by the experiment, so the researcher can compare the optimal threshold with the observed bid. Their strength is also a limitation: imposed values do not reproduce the effort, uncertainty, familiarity, and personal significance involved in discovering a real preference. Field comparisons sacrifice knowledge of each person's latent value but can benchmark elicited demand against actual purchase probabilities. These designs identify different claims and should be integrated rather than treated as interchangeable tests Harrison et al. (2004) Cole et al. (2020).
Drichoutis, Palma, and Feldman Drichoutis et al. (2026) reported a published study of two experiments with an aggregate sample exceeding 3,000 participants. In the first, researchers varied the payment architecture around a BDM-like induced-value selling task. They found that induced-value magnitude and the range used for random prices mattered for misbidding, whereas the payment variation and certainty of payment tested did not produce corresponding differences. The second used a second-price auction and found similar qualitative patterns with less misbidding. These results sharpen the inference that formal incentive design must be paired with empirical protocol evaluation. The observable outcome was misbidding relative to known induced values, rather than accuracy of an unknowable natural preference.
Comparative evidence should be read at the level of the observed manipulation. A finding that a payment protocol had no detectable effect does not prove equivalence, because precision, effect sizes, sample composition, and choice of outcome govern which differences could have been observed. Conversely, a significant price-support effect can reflect behavioral interpretation or effort allocation rather than failure of the mechanism's formal incentive theorem. Researchers should report confidence intervals, randomized unit, compliance, bid distributions, and a predeclared smallest consequential difference. Independent replications with new goods, different payment conditions, and broader populations would clarify how far the 2026 induced-value findings can be transported Drichoutis et al. (2026) Mamadehussene & Sguera (2023).
Field Experiments and Revealed Purchase Benchmarks
The practical test of a demand instrument is whether it predicts purchase behavior under realistic transactions. In their two field experiments, Cole and colleagues Cole et al. (2020) compared BDM elicitation against randomly offered fixed prices for an agricultural information product and rainfall index insurance. The strategy uses randomized prices to estimate an observed purchase response without taking elicited bids as ground truth. Field implementation also exposes heterogeneity in understanding, budget constraints, and product complexity that may be suppressed in laboratory recruitment. The paper illustrates the advantage of triangulating mechanism-generated reports with a distinct incentive-compatible purchase benchmark.
The correspondence of observed demand curves depends on what is held constant. A bid elicited before lengthy explanation is not necessarily comparable with an immediate yes-or-no purchase after participants inspect the product. Training, price distribution, product descriptions, sales interface, trust, and perceived resale opportunities can act as treatment components. An evaluator should document these features and, where feasible, randomize a common informational treatment before mechanism assignment. In a field comparison, attrition and missing purchase decisions must be analyzed by arm, especially if the mechanism is more cognitively demanding for certain respondents. Both intention-to-treat and protocol-complier analyses can be informative when their estimands are clearly stated Cole et al. (2020) Harrison & List (2004).
Consumer willingness to pay is especially unstable when the commodity is unfamiliar, an insurance contract requires probability integration, or future benefits are difficult to imagine. Market experience can alter both preferences and the ability to express them through unfamiliar rules. Studies of experience in exchange markets List (2003) List (2011) point to this dual effect. The design consequence is to compare instruments with respect to external outcomes and participant characteristics, allowing mechanism-by-experience interactions rather than treating inconsistent bids from novices as universal flaws. Such heterogeneous effects are central to the validity of field translation.
Learning, Repeated Exposure, and Procedural Experience
Repetition has several competing effects. Practice may correct misunderstandings, improve mastery of bidding and price lotteries, and reduce inattentive answers. It may also alter valuations by increasing familiarity, revealing price norms, or encouraging strategic interpretations that were absent initially. Plott and Zeiler's studies Plott & Zeiler (2005) Plott & Zeiler (2007) make training central to the interpretation of ownership-related valuation gaps, while the market experience research of List (2003) emphasizes that accumulated transactional knowledge is not equivalent to a one-minute explanation. A study should state whether it seeks initial consumer valuation, a value after informed learning, or a stable preference under repeated opportunities.
Czajkowski, Hanley, and LaRiviere Czajkowski et al. (2015) investigated how experience can affect preferences for environmental public goods, extending the question of learning to repeated stated choice environments. This research suggests that preferences measured after exposure can be systematically different from those recorded before participants acquire information or understand the consequences. Time trends alone are insufficient to attribute a change to learning: fatigue, income changes, social information, and attrition may move in the same direction. Randomized opportunities for practice, counterbalanced items, and comparison of learning curves across mechanisms can distinguish parts of these effects.
The practical protocol should separate instructional training, preference discovery, and market exposure. An induced-value practice task can verify understanding without requiring participants to reveal private consumer valuations; repeated experience with actual goods can then be studied as a distinct treatment. Researchers should record whether practice outcomes are financially consequential and how feedback was delivered. When a protocol changes after early participants show misunderstandings, that change must be documented, and data from different versions should not be silently pooled. Such reporting improves replication and clarifies whether a mechanism is robust to the kinds of learning that real-world implementation inevitably produces Noussair et al. (2004) Mamadehussene & Sguera (2023).
Causal Identification and Statistical Analysis of Elicitation Designs
A causally interpretable comparison requires an explicitly defined outcome and a treatment contrast. The treatment may be the BDM mechanism rather than a posted-price choice, the range of random prices, presence of practice, real versus hypothetical payment, or the frequency with which decisions are paid. Random assignment identifies the effect of the assigned protocol under the experiment's implementation and interference assumptions; it does not directly establish which underlying psychological mechanism caused a difference. Factorial designs can estimate interactions between payment, training, price framing, and participant experience when their sample size and assignment scheme support those analyses Charness et al. (2012) Drichoutis et al. (2026).
Outcome measurement should align with the scientific claim. With induced values, an error measure can be the deviation of bid from the experimenter's value or the frequency of decisions with negative expected transaction value. With field products, demand at a specified price, realized purchase, and follow-through are often preferable benchmarks. Analyses of bids may require handling censoring, bounded distributions, clustering across sessions, and repeated observations. Comparing only mean bids can miss a mechanism that compresses reports around a price reference while reproducing the population mean. Distributional tests and individual-level predictive validation make the scientific claim more precise Cole et al. (2020) Harrison et al. (2004).
Preregistration is especially helpful when many plausible comparisons can be calculated. A plan should specify primary outcomes, sampling frame, power assumptions, treatment assignments, stratification, excluded responses, and intended models for heterogeneity. The final report should present uncertainty, missing-data patterns, and violations of planned protocols, distinguishing exploratory analyses from confirmatory tests. Cross-task claims become stronger when the analyst can forecast out-of-sample purchases or independently test the proposed measurement model on participants who were not used to choose its parameters. The literature on experimental design Charness et al. (2012) and field methods Harrison & List (2004) provides a rationale for these requirements.
Distributional Validity, Population Differences, and Ethical Consequences
An elicitation device can perform well on average yet generate unequal error rates. Numeracy, language proficiency, financial stress, digital literacy, familiarity with auctions, disability-related accessibility needs, and experience with the product may alter comprehension and the ability to express a value. The relevant comparison is whether experimental treatments yield equivalent measurement relationships across groups, after accounting for legitimate differences in resources and preferences. This requires estimating differential protocol effects, not merely comparing group averages. Risk-measurement research Dohmen et al. (2011) and field valuation studies Cole et al. (2020) show why instrument validation must include the populations to whom conclusions will be generalized.
Ethics and scientific design meet when an experimental bid can trigger a real financial loss or purchase. Investigators should ensure informed consent covers the binding transaction, identify maximum possible payment, provide clear alternatives and appropriate withdrawal procedures, and comply with institutional research oversight. Vulnerable participants may interpret probabilities or transaction rules differently, so comprehension and accessibility are relevant to equitable research participation. The investigator also has obligations concerning confidentiality of individual bids, transparent compensation, avoidance of deceptive scarcity, and respectful treatment of the participant's financial circumstances. The legitimacy of the transaction is a precondition for interpreting it as a freely made choice Carson & Groves (2007) Harrison & List (2004).
Public policy applications demand even more careful inference. A valuation task can support the design of a fee, benefit, or intervention only when the reported value represents the appropriate welfare-relevant construct. In hypothetical surveys, consequentiality and payment vehicles shape meaning; in real purchase experiments, cash constraints can limit the purchasing expressions of people who genuinely value the good. Monetary WTP is jointly influenced by preference and ability to pay. Analysts should therefore distinguish valuation differences from distributional welfare judgments and explicitly report the ethical implications of using bids to allocate scarce benefits or determine access to essential services Murphy et al. (2005) Carson & Groves (2007).
An Integrated Framework for Selecting Elicitation Mechanisms
A defensible preference-elicitation study begins with a construct rather than with a favored experimental device. For reservation values of an existing consumer good, an appropriately implemented BDM mechanism, auction, or randomized posted-price benchmark may be suitable, depending on the target population and institutional constraints. For risk attitudes, monetary lotteries and self-reports address related but distinct constructs; for time preferences, joint identification with utility and payment credibility becomes essential; for public goods, consequential stated-choice designs may be the only feasible approach. The relevant matching criterion is the inference the analyst intends to make. Formal mechanism conditions Becker et al. (1964) Vickrey (1961) and the survey incentive analysis Carson & Groves (2007) define different admissible mappings.
Mechanism selection then requires evaluating how participants encounter the task. A good device should have understandable transaction rules, an explicit price or probability support, a plausible relation to the real decision, and a verifiable payoff implementation. Before data collection, researchers can compare candidate protocols on their theoretical incentive properties, likely comprehension burden, behavioral calibration results, implementation cost, and external validity. No single score can substitute for these separate judgments. The comparative studies in Noussair et al. (2004), Cole et al. (2020), and Drichoutis et al. (2026) indicate that institutional form and incidental-seeming features can change observed performance across settings.
The final step is a prespecified validation strategy. If the scientific purpose is a demand curve, compare elicited valuations with independently randomized purchases over the relevant price range. If the purpose is individual measurement, estimate test–retest behavior and prediction across tasks. If the purpose is a theoretical mechanism test, create induced values and verify that participants understand the opportunity structure. For each, report heterogeneity and uncertainty, and preserve the distinction between failure to reject an effect and evidence that two methods are sufficiently equivalent. This framework converts methodological pluralism into cumulative evidence while retaining the particular constraints of each study Harrison et al. (2004) Charness et al. (2012).
Critical Synthesis of Theory and Empirical Evidence
The evidence supports three levels of conclusion. Formal analyses establish conditional incentive properties: BDM and second-price auctions provide reasons for truth-telling under their relevant assumptions. Controlled experiments establish what participants actually report under specific instructions and payment regimes. Field studies assess whether those reports correspond to transactions and demand in consequential environments. These levels address complementary questions, and disagreement among them can locate a substantive feature of preference formation, a procedural defect, or a boundary in the model. Neither theoretical dominance nor observed correspondence in one setting is a general guarantee of universally valid valuation Becker et al. (1964) Vickrey (1961) Cole et al. (2020).
Preference reversals and ownership-related valuations reveal the theoretical importance of task variation. The original reversal research Lichtenstein & Slovic (1971) Grether & Plott (1979), experiments on endowment and exchange Kahneman et al. (1990) Plott & Zeiler (2005), and constructive choice theory Bettman et al. (1998) collectively show that response format, learned procedures, and reference conditions can change what the task expresses. Their interpretations remain contested because task effects can be compatible with several mechanisms. The scientific synthesis therefore treats a task difference as a result to explain through controlled comparisons and competing predictions. It does not label every difference an irrational error or a stable valuation change.
Recent experimental studies sharpen the distinction. Distribution-sensitive BDM bids have been modeled through endogenous preference-learning effort Mamadehussene & Sguera (2023). The 2026 induced-value study Drichoutis et al. (2026) found important effects of price support and value magnitude while several tested payment features had no detectable effect. Field evidence Cole et al. (2020) showed that correspondence with posted-price demand varies by commodity and implementation. Quantitative synthesis of hypothetical valuation Murphy et al. (2005) demonstrated heterogeneity rather than a universal correction. The combined pattern warrants a broad conclusion: preference elicitation is jointly a mechanism-design problem, a psychological measurement problem, and an empirical validation problem.
Artificial Mediation of Economic Preference Elicitation
In the Artificial Era, many valuation tasks are delivered through digital interfaces, recommendation systems, and conversational agents. These systems can help participants compare alternatives, understand random prices, and review consequences, but they can also shape salience, suggest reference values, and introduce framing through the order of explanation. An artificial intermediary that asks for willingness to pay may therefore become part of the experimental treatment. Existing research on constructed choice Bettman et al. (1998) and distribution-sensitive BDM behavior Mamadehussene & Sguera (2023) supplies a scientific rationale for testing this influence, while direct evidence must be collected for particular artificial systems and populations.
A useful design would randomize explanatory assistance, agent wording, recommendation visibility, and price-distribution displays while keeping the financial consequence constant. It would measure understanding of the transaction rule, bid accuracy with induced values, subsequent purchase behavior, and the stability of responses after the agent is removed. In a repeated interaction, recommendation histories and memory of earlier bids could alter what subjects perceive as their own valuation. Such effects should be interpreted as possible changes in information, attention, or constructed preferences until the study can distinguish them. The field benchmark approach used by Cole et al. (2020) can provide one empirical route to testing artificial mediation against actual purchases.
The scientific horizon extends to the organization of decisions across human and artificial systems. A consumer may delegate exploration, comparison, or the execution of a bid while retaining the practical consequences of a transaction. The research object is then a distributed elicitation process whose observable components include prompts, model output, human confirmation, settlement, and subsequent behavior. Identifying who supplied a numeric report, what information the system provided, and which agent made the final economically consequential choice becomes a methodological requirement. These propositions are research directions for the Artificial Era, not claims that current evidence has already established a new psychological mechanism of artificial preference.
Unresolved Problems and Priorities for Future Research
One unresolved question is how to distinguish stable preference change from changed incentives to discover a preference. In BDM settings, the draw distribution can affect attention and deliberation even when truthful reporting remains conditionally optimal. Studies that measure the time and effort spent evaluating alternatives, experimentally vary information acquisition costs, and compare predictions across price supports could test competing accounts more sharply Mamadehussene & Sguera (2023). A second question concerns the boundary of incentive compatibility under realistic budgeting, multiple purchases, and repeated interaction. Structural models should state whether each transaction is evaluated separately or within an evolving household portfolio.
A third priority is calibrated multimethod measurement across populations. Researchers should compare the same participants under bids, binary purchases, auctions, and suitably consequential stated choices while counterbalancing task sequence and preserving comparable goods and information. The analysis should separate average accuracy, distributional accuracy, within-person rank stability, and prediction of future choice. Cross-cultural samples, financial-resource variation, low-numeracy populations, and accessible interfaces would identify whether an instrument is invariant enough for its intended comparisons. Existing work on risk measures Dohmen et al. (2011), experiment design Charness et al. (2012), and real-world BDM benchmarking Cole et al. (2020) provides empirical starting points.
A fourth priority is preregistered replication of apparently incidental choice-architecture effects. Price ranges, fonts, numeric response grids, examples, feedback, and payment salience may all affect reports, but it is necessary to establish which effects generalize to meaningful outcomes and which disappear with experience or different goods. The experiments of Drichoutis et al. (2026) provide testable hypotheses; the earlier training and exchange debate Plott & Zeiler (2005) Plott & Zeiler (2007) explains why procedural details may be scientifically decisive. Future evidence syntheses should code the mechanism, target construct, training, payment, price support, and product characteristics separately rather than aggregating every 'incentive-compatible' instrument into one category.
Conclusion
Economic preference elicitation is a systematic attempt to convert consequential decisions into evidence about valuations, rankings, demand, and intertemporal or risky tradeoffs. Its major technical achievement is the design of mechanisms in which a truthful response follows from an explicit payoff rule. The BDM procedure, second-price auctions, and related formats demonstrate how the analyst can align a report with the participant's transactional interest. Their historical importance remains secure even as experiments reveal deviations, comprehension difficulties, and context-dependent values. The appropriate scientific claim attaches the mechanism to a precisely specified choice environment and set of assumptions Becker et al. (1964) Vickrey (1961).
The accumulated behavioral evidence establishes a more demanding account of measurement. Choice and pricing tasks may yield preference reversals, ownership and reference conditions can influence WTP and WTA, hypothetical answers can diverge from actual transactions, and the ability to understand or learn a value can depend on how prices are displayed. The strongest inference arises when formal theory, induced-value calibration, controlled task comparison, and independent market behavior converge. Apparent contradictions among studies often expose differences in product, context, incentives, and estimand that must be documented rather than averaged away Grether & Plott (1979) Murphy et al. (2005) Cole et al. (2020) Drichoutis et al. (2026).
The resulting methodological principle is that incentive compatibility specifies a conditional truth-telling property, while scientific validity is earned through empirical identification and reproducible validation of the intended psychological and economic construct. Research should begin by defining that construct, choose a defensible transaction mechanism, establish comprehension and credible consequences, examine robustness to framing and training, and test generalization through external choices. This principle organizes classical experimental economics, modern consumer psychology, and emerging digitally mediated preference research within a single evidence-sensitive approach. It provides the enduring foundation for economic valuation studies that seek both formal rigor and psychologically meaningful interpretation.
References
- Andersen, S., Harrison, G. W., Lau, M. I., & Rutström, E. E. (2008). Eliciting risk and time preferences. Econometrica, 76(3), 583–618. https://doi.org/10.1111/j.1468-0262.2008.00848.x
- Becker, G. M., DeGroot, M. H., & Marschak, J. (1964). Measuring utility by a single-response sequential method. Behavioral Science, 9(3), 226–232. https://doi.org/10.1002/bs.3830090304
- Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive consumer choice processes. Journal of Consumer Research, 25(3), 187–217. https://doi.org/10.1086/209535
- Carson, R. T., & Groves, T. (2007). Incentive and informational properties of preference questions. Environmental and Resource Economics, 37, 181–210. https://doi.org/10.1007/s10640-007-9124-5
- Charness, G., Gneezy, U., & Kuhn, M. A. (2012). Experimental methods: Between-subject and within-subject design. Journal of Economic Behavior & Organization, 81(1), 1–8. https://doi.org/10.1016/j.jebo.2011.08.009
- Charness, G., Gneezy, U., & Imas, A. (2013). Experimental methods: Eliciting risk preferences. Journal of Economic Behavior & Organization, 87, 43–51. https://doi.org/10.1016/j.jebo.2012.12.023
- Cole, S., Fernando, A. N., Stein, D., & Tobacman, J. (2020). Field comparisons of incentive-compatible preference elicitation techniques. Journal of Economic Behavior & Organization, 172, 33–56. https://doi.org/10.1016/j.jebo.2020.01.021
- Czajkowski, M., Hanley, N., & LaRiviere, J. (2015). The effects of experience on preferences: Theory and empirics for environmental public goods. American Journal of Agricultural Economics, 97(1), 333–351. https://doi.org/10.1093/ajae/aau087
- Dohmen, T., Falk, A., Huffman, D., Sunde, U., Schupp, J., & Wagner, G. G. (2011). Individual risk attitudes: Measurement, determinants, and behavioral consequences. Journal of the European Economic Association, 9(3), 522–550. https://doi.org/10.1111/j.1542-4774.2011.01015.x
- Drichoutis, A. C., Palma, M. A., & Feldman, P. J. (2026). Incentives and payment mechanisms in preference elicitation. Experimental Economics, First View, 1–17. https://doi.org/10.1017/eec.2026.10042
- Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10, 171–178. https://doi.org/10.1007/s10683-006-9159-4
- Grether, D. M., & Plott, C. R. (1979). Economic theory of choice and the preference reversal phenomenon. American Economic Review, 69(4), 623–638. https://www.jstor.org/stable/1808708
- Harrison, G. W., Harstad, R. M., & Rutström, E. E. (2004). Experimental methods and elicitation of values. Experimental Economics, 7, 123–140. https://doi.org/10.1023/B:EXEC.0000026975.48587.f0
- Harrison, G. W., & List, J. A. (2004). Field experiments. Journal of Economic Literature, 42(4), 1009–1055. https://doi.org/10.1257/0022051043004577
- Harrison, G. W., & Swarthout, J. T. (2014). Experimental payment protocols and the bipolar behaviorist. Theory and Decision, 77, 423–438. https://doi.org/10.1007/s11238-014-9447-y
- Holt, C. A., & Laury, S. K. (2002). Risk aversion and incentive effects. American Economic Review, 92(5), 1644–1655. https://doi.org/10.1257/000282802762024700
- Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1990). Experimental tests of the endowment effect and the Coase theorem. Journal of Political Economy, 98(6), 1325–1348. https://doi.org/10.1086/261737
- Lichtenstein, S., & Slovic, P. (1971). Reversals of preference between bids and choices in gambling decisions. Journal of Experimental Psychology, 89(1), 46–55. https://hdl.handle.net/1794/22312
- List, J. A. (2003). Does market experience eliminate market anomalies? Quarterly Journal of Economics, 118(1), 41–71. https://doi.org/10.1162/00335530360535144
- List, J. A. (2011). Does market experience eliminate market anomalies? The case of exogenous market experience. American Economic Review, 101(3), 313–317. https://doi.org/10.1257/aer.101.3.313
- Mamadehussene, S., & Sguera, F. (2023). On the reliability of the BDM mechanism. Management Science, 69(2), 1166–1179. https://doi.org/10.1287/mnsc.2022.4409
- Murphy, J. J., Allen, P. G., Stevens, T. H., & Weatherhead, D. (2005). A meta-analysis of hypothetical bias in stated preference valuation. Environmental and Resource Economics, 30, 313–325. https://doi.org/10.1007/s10640-004-3332-z
- Noussair, C., Robin, S., & Ruffieux, B. (2004). Revealing consumers’ willingness-to-pay: A comparison of the BDM mechanism and the Vickrey auction. Journal of Economic Psychology, 25(6), 725–741. https://doi.org/10.1016/j.joep.2003.06.004
- Offerman, T., Sonnemans, J., van de Kuilen, G., & Wakker, P. P. (2009). A truth serum for non-Bayesians: Correcting proper scoring rules for risk attitudes. Review of Economic Studies, 76(4), 1461–1489. https://doi.org/10.1111/j.1467-937X.2009.00557.x
- Plott, C. R., & Zeiler, K. (2005). The willingness to pay–willingness to accept gap, the “endowment effect,” subject misconceptions, and experimental procedures for eliciting valuations. American Economic Review, 95(3), 530–545. https://doi.org/10.1257/0002828054201387
- Plott, C. R., & Zeiler, K. (2007). Exchange asymmetries incorrectly interpreted as evidence of endowment effect theory and prospect theory? American Economic Review, 97(4), 1449–1466. https://doi.org/10.1257/aer.97.4.1449
- Shogren, J. F., Margolis, M., Koo, C., & List, J. A. (2001). A random nth-price auction. Journal of Economic Behavior & Organization, 46(4), 409–421. https://doi.org/10.1016/S0167-2681(01)00165-2
- Vickrey, W. (1961). Counterspeculation, auctions, and competitive sealed tenders. Journal of Finance, 16(1), 8–37. https://doi.org/10.1111/j.1540-6261.1961.tb02789.x
- Wertenbroch, K., & Skiera, B. (2002). Measuring consumers’ willingness to pay at the point of purchase. Journal of Marketing Research, 39(2), 228–241. https://doi.org/10.1509/jmkr.39.2.228.19086