Reverse Pricing and Online Price ElicitationStrategies in Consumer ChoiceCHE RNEVPRICE EL ICITATI ON ST RATE GIE S I N CHOICE

Alexander ChernevKellogg School of Management

Northwestern University

This research examines consumersrsquowillingness to pay in an online environmentSpecifically Icompare two price-elicitationstrategiesprice generation(ie ldquoname your pricerdquo) and price se-lection (ie ldquoselect your pricerdquo) Contrary to the common assumption that naming a price willbe preferredby consumersbecauseit offers the most flexibilityin articulatingonersquos willingnessto pay this research demonstrates that consumers often prefer to select rather than to generate aprice In a series of three experiments I show that the potentialunfavorableeffects of the price-generation task are associated with the absence of a readily available reference price range Ifurther demonstrate that the referenceprice range also has to be externallyprovidedand that in-ternally generated reference prices can as well eliminate the potential negative effect of theprice generationtask and strengthenconsumerpreferencesThese findingssupport the proposi-tion advanced in this research that a pre-choice articulation of reference prices can simplifyconsumer choice by imposing a structure consistent with the nature of the decision task

Price perception has been a focus of consumer researchersfor several decades Most of this research has employed astimulusndashresponse (SndashR) model which interprets the actualprices that consumers encounter as stimuli activating the per-ception process (Berkowitz amp Walton 1980 LichtensteinBloch amp Black 1988 Monroe amp Lee 1999) On being pre-sented with a given price consumers encode evaluate andstore this price while integratingit with the other nonprice in-formation The outcome of processing the price informationis a behavioral responsemdashfor example purchasing or notpurchasing the particular offering Thus the SndashR frameworkreflects the traditional marketing paradigm in which pricesare the stimuli presented to consumers who in turn processthe available information and act on it

With the development of the Internet and the increasingpopularityofonlinemarketplaceshoweverconsumersareof-fered more choicesnotonlyin terms of theproductassortmentbutalso in terms of productpricingOne such pricingstrategymost prominently popularized by the reverse auction pioneerPriceline asks consumers to name their own price for variousproductsand services (eg air ticketshotels rental cars) If aconsumersrsquoprice matches or exceeds the price set by the mer-chant the consumerrsquos price is accepted and the transaction iscompleted otherwise the consumerrsquos bid is rejected Unlike

traditional pricingmdashwhere the merchant sets the price andconsumersindicatetheiracceptanceof thispricebyeitherpur-chasing or not purchasing the productmdashin the Priceline sce-nario consumers themselveshave to set the price and the mer-chant indicates its evaluation of this price by either acceptingor not acceptinga consumerrsquos offer Because the price-settingfunctionsof theconsumerand themerchantare reversed inthiscase I refer to this scenario as reverse pricing

As a marketing phenomenon reverse pricing is similar toauction pricing in that consumers must explicitly state theirwillingness to pay for a given product (eg Sinha amp Green-leaf 2000)What makes the reverse pricingscenario analyzedin this article different from auction pricing is that (a) there isno apparent product scarcity products are commodities (egairline tickets) and multiple items are readily available and(b) reference pricesare oftennotreadilyavailableBecause thenumber of products is not visibly constrained consumers bidagainst the merchant not against one another Furthermorebecausethere isno scarcityconsumersalways have theoptionof walking away and purchasing the product elsewhereFinally in auctions there often is a clearly defined referencepricemdashsuch as an opening bid appraised value asking priceetcmdashwhereas in the reverse pricing scenario consumers areoften asked to name their price withoutan explicitlyavailablereference point

This research examineshow consumers articulateprices inthe context of an online reverse pricing scenario In this con-text I compare two elicitation procedures price generation

JOURNAL OF CONSUMER PSYCHOLOGY 13(1amp2) 51ndash62Copyright copy 2003 Lawrence Erlbaum Associates Inc

Request for reprints should be sent to Alexander Chernev KelloggSchool of Management Northwestern University 2001 Sheridan RoadEvanston IL 60208 E-mail achnwuedu

(ie ldquoname your pricerdquo) and price selection (ie ldquoselect yourpricerdquo) The former approach advanced by Priceline simplyasks consumers to state the price they are willing to pay for theproduct underconsiderationIn the latter approach presentedin this research consumersare presentedwith a set of possiblepricesandaskedto selectthepricetheyfindmostacceptable

The price-generationscenario is clearly more flexible thantheselectionscenariobecauseitallowsconsumerstopreciselyarticulate their willingness to pay Indeed in the generationscenario consumers have virtually unlimited degrees of free-dom to state their price whereas in the selection scenario theyare restricted by the set of prices presented to them Viewedfrom an economicsstandpointthis flexibilityis oneof the rea-sons why the ldquoname yourpricerdquo strategy would be consideredsuperior to the selection strategy assuming that consumershave establishedpreferences that can easily be translated intomonetary terms (eg Varian 1999) In fact the limited selec-tion constrains consumersrsquoability to adequately express theirwillingness to pay and as with any constrained optimizationwouldbeconsideredinferior toan unconstrainedoptimizationsuch as the price generation scenario

Yet because price generation assumes established prefer-ences and predetermined willingness to pay it can be arguedthat its impact on the consumer decision process will dependon the degree to which consumers are able to articulate theirproduct utility in monetary terms Building on this notion Ipropose that in the absence of a readily available task-specificcontext (eg reference price range) the generationstrategy islikely to be associatedwith a greater degree of uncertaintyandcognitiveeffort andas a result will be perceived to be inferiorto thesimplerprice selectiontaskThispropositionisalsosup-ported by the view of consumers as contingentdecision mak-ers whorather thanhavingasetofpredeterminedpreferencesform their preferences in the context of the specific decisiontask (Bettman Luce amp Payne 1998)

In the remainder of this article I develop the conceptualframework and derive specific predictions regarding the ef-fects of the generationand selectiontaskson theconsumerde-cision process Next I describe and report the results of threeexperiments that test these predictionsI concludeby discuss-ing the implications of the findings and providing directionsfor further research

PRICE ARTICULATIONIN CONSUMER CHOICE

To set the optimal price for a given product consumers mustexpress their overall evaluation as a monetary amount Re-cent research has proposed that consumers do this by a pro-cess of subjective interpolation whereby they search for apoint on the monetary scale that corresponds to their ex-pected utility from the product (Goldstein amp Einhorn 1987Janiszewski amp Lichtenstein 1999 see also Lynch

Chakravarti amp Mitra 1991) To illustrate consider Figure 1depicting a monetary scale with endpoints PMIN and PMAXwhere PMIN is a consumerrsquos perception of the lowest price forthe given product and PMAX is the highest expected priceThis monetary scale is linked to a utility scale with endpointsU(PMIN) and U(PMAX) where U(PMIN) is the utility at thelowest price and U(PMAX) is the utility at the highest ex-pected price

In this context the consumersrsquo task is to determine theprice (P) that corresponds to the utility that they expect to re-ceive from the product U(P) To express their utility in mone-tary terms consumers are assumed to establish a correspon-dence between the endpoints of the utility scale and theendpointsof the monetary scale Consequentlytheir willing-ness to pay stated in monetary terms is calculated to corre-spond to the proportional relations between U(PMIN)U(PMAX) and U(P) In other words consumers elicit theirwillingness to pay by equating the endpoints and matchingthe proportional change in the utility scale with a propor-tional change on the monetary scale

To illustrate consider a consumer who believes that thelowest airfare for a flight between Chicago and San Fran-cisco is $189 and the regular fare is $459 Based on the utilityderived from purchasing a ticket at these fares the consumerdetermines the utility from the particular offering under con-sideration which is then expressed in monetary terms (say$260) to correspond to relations between U($189) U($459)and U($260) Note that this subjective interpolation processhinges on the availability of a reference price range (egPMAX and PMIN) which is used as a benchmark in the pricearticulation process The importance of reference points inprice perception and evaluation has been underscored by nu-merous researchers in the field (Alba Broniarczyk Shimpamp Urbany 1994 Heath Chatterjee amp France 1995 Herr1989 Lichtenstein et al 1988 Mayhew amp Winer 1992Urbany amp Dickson 1991)

Yet on many occasions consumers do not have readilyavailable reference prices and as a result they lackbenchmarks needed to indicate their willingness to pay for agiven product In this case consumers must first evoke therelevant reference prices and then use these prices asbenchmarks in eliciting their willingness to pay To illustrate

52 CHERNEV

FIGURE 1 Reverse pricing as a subjective interpolation process

using the example presented in Figure 1 when PMIN andorPMAX are not present consumers first have to estimate thelikely values of PMIN andor PMAX After the range of possibleprices has been established consumers can assess the utilityassociated with these values [U(PMIN) and U(PMAX)] anddetermine the price they are willing to pay (P) based on theutility derived from this price [U(P)] As a result whenreference prices are not readily available price articulationcan be represented as a two-stage process whereby consum-ers first need to evoke the range of possible values and thenusing utilities of these values as benchmarks determine theirutility for the product and articulate this utility on a monetaryscale

Following this line of reasoning for consumers withoutreadily available reference price ranges price elicitation willbe associated with substantial effort because of the complexnature of the decision task involving reference price estima-tion in addition to the price-generation As a resultprice-generation tasks presented in a context where a refer-ence price range is not readily present are likely to be rela-tively more difficult compared to elicitation tasks with a sa-lient range of possible prices Building on this argument Ipropose that when a reference price range is not readilyavailable consumers faced with the more complex price-generation task are likely to be less confident in their deci-sions compared to consumers who are presented with the rel-atively simpler selection task Indeed the selection taskinherently presents consumers with a range of referenceprices and a consumerrsquos decision in this case is reduced toselecting one of the available price options This perceivedsimplicity of the decision task will also lead to an overallpreference for the selection task despite the fact that it ismore restrictive and hence less accurate This preference forthe selection-based price elicitation can be viewed as an ef-fortndashaccuracy tradeoff whereby consumers give up accuracyin expressing their willingness to pay in favor of making aless effortful decision (Payne Bettman amp Johnson 1993 Si-mon 1955)

In sum I propose that the availability of salient referenceprices moderates the impact of the price-elicitation task onconsumer preferences Specifically in the absence of areadily available reference price range consumers will likelytrade off the accuracy in expressing their willingness to paythat the price-generation task offers for the simpler althoughless flexible selection task In contrast in contexts wherereference prices are readily available consumers will be lesslikely to discount the price-generation task on the basis of itsrelative complexity and consequently will be less likely todisplay a stronger preference for the price-selection task

These predictionsare tested in a series of experiments thatcompare the two price-elicitation scenarios generationversus selection and examine consumer preferences forthese pricing strategies as a function of reference price avail-ability

EXPERIMENT 1

This experiment investigatesthe impact of the availabilityof areference priceonconsumerpreferencesfor a price-elicitationtask The goal is to offer preliminary evidence that in the ab-senceof establishedreference prices consumerswill prefer toexpress their willingness to pay using a selection rather than agenerationtaskThirty-tworespondentswere presentedwithahypothetical scenario in which they were asked to purchaseairline tickets from two online agencies Both agencies em-ployed reverse pricing but used different price-elicitationstrategies One agency asked consumers to generate (name) aprice whereas the other asked consumers to select from a listof 10 availableprices (from $169 to $439 in ascendingorderincreasing at $30 increments) This price-elicitation manipu-lationwas conductedwithinsubjectsIn additionrespondentsin the reference price condition were given a referencepricemdashthey were told that the regular airfare is$439mdashwhereas the others were not given any price informa-tion

Participants were randomly assigned to the conditionsof a2 (price-elicitation task generation vs selection) times 2 (refer-ence price availablevs not available)mixed factorial designThey were asked to indicate the price they were willing to payfor the ticketsand were told that if theirbidswere not success-ful they would have theoptionto purchase the ticketat the reg-ularpriceParticipantswere asked to indicatetheir confidencein the decisionas well as their expectationsof the likelihoodofsuccessof theirbidusinga 150-mmnongradedscalewith end-pointsNot confidentversus Very confidentand Not likely ver-susVery likely(seeBettmanJohnampScott1986for detailsonthe measurement procedure) Participantsrsquoresponses were re-cordedbymeasuringthedistancefrom the left endof thescaleand these responses were standardized to correspond to a100-point scale

The data show that in the absence of a reference price in-dividuals were more confident in the context of a selectiontask (M = 411) than in the context of a generation task (M =275) The direction of the effect remained the same evenwhen a reference price was present although respondents inthis condition were more confident compared to respondentsin the condition where a reference price was not available (M= 514 for the selection task vs M = 397 for the generationtask) The data show a significant main effect of the price-ar-ticulation task on decision confidence whereby the genera-tion task was associated with lower confidence compared tothe selection task F(1 30) = 3348 p lt 001 There was also amarginally significant main effect of the availability of refer-ence price participants were more confident in conditionswhere a reference price was readily available F(1 30) =377 p lt 10 Participantsrsquo evaluations of the likelihood ofsuccess of their bid followed a similar pattern Respondentsexpected prices derived from the selection task to have ahigher probability of being accepted compared to prices elic-

PRICE ELICITATION STRATEGIES IN CHOICE 53

ited through the generation task (M = 506 vs M = 327) F(130 ) = 4301 p lt 001 They also predicted a higher successprobability in conditions where a reference price was presentcompared to conditions where a reference price was notreadily available (M = 461 vs M = 372) F(1 30) = 423 p lt05

The data from this experiment suggest that consumers feelmore confident in the outcome of a selection rather than ageneration task This effect was also significant althoughless pronounced in the presence of a readily available refer-ence price Furthermore consumers perceived the selectiontask to have a higher likelihood of success compared to thegeneration task an effect more pronounced when a referenceprice was not readily available

Note that in Experiment 1 the potential effect of availabil-ity of a reference price range is confounded with the effectsassociated with the nature of the elicitation task Indeed con-sumer preference for a selection price mode can be the resultof an attempt to simplify the pricing decision by selecting apricing strategy with fewer possible alternatives Thus con-sistent with the effortndashaccuracy framework (Payne et al1993) one can argue that consumers will prefer the selectiontask because it offers a simpler (although less accurate) deci-sion and not necessarily because in this case a reference pricepoint is readily available This potential confound is ad-dressed in Experiment 2 by introducingan experimental con-dition (referred to as augmented generation) designed to dis-entangle the effects of reference price range availability anddecision effort associated with generating a precise pricepoint

EXPERIMENT 2

Building on Experiment 1 findings Experiment 2 examinesthe impact of the price-elicitation task on the strength of con-sumer preferences and how the availability of a referenceprice range moderates this impact This experiment investi-gates three scenarios price generation price selection and athird scenario referred to as augmented generation Partici-pants in the generation scenario were asked to name theirown price and were not provided with a specific price rangeRespondents in the selection condition were given a set of 10alternative prices to choose from Finally in the augmentedgeneration scenario (discussed in more detail later) partici-pants were asked to name their own price and were providedwith a price range

Method

Respondents were presented with the following scenario

Imagine that three of your classmates had to fly to SanFrancisco for a high-tech job fair They all decided tobuy their air tickets through a reverse pricing Web

travel agency that allows travelers to name their ownprice

Reverse pricing works as follows You tell the travelagency where and when you want to go and how muchyou are willing to pay The agency then searches for anairline willing to release seats at your price If theagency finds tickets at your price it will immediatelypurchase those tickets for you Because you get toname your own price tickets purchased through re-verse pricing Internet agencies cannot be changedtransferred or cancelled

The travel agency will make the best effort to makethe booking It will submit your and other customersrsquorequests to its airline partners who will then decide onthe lowest price to accept Everyone whose bid ishigher or equal to that price or higher will be granted aticket at the stated price Everyone else will have tomake their bookingsdirectly and pay the regular price

Next participants were told that Students A B and C are us-ing different reverse pricing travel agencies that employ dif-ferent procedures to allow customers to name their priceStudent A was asked to name the price he or she is willing topay (generation task) Student B was given a list of 10 pricesand asked to choose among them (selection task) and Stu-dent C was shown a scale that depicts the typical price rangeand asked to indicate the price he or she is willing to pay bypositioninga slider on the scale (augmented generation task)These price-generation tasks are presented in more detail inthe Appendix

Note that respondents in both generation and augmentedgenerationconditionswere offered agreaterdegreeof flexibil-ity in elicitingtheirwillingnessto paywhereas participantsinthe selection condition were limited to a list of 10 availableprices In addition respondents in the selection and the aug-mented generationconditionswere implicitlyprovidedwith aprice range (determined by the highest and the lowest pricesavailablefor selection)whereasparticipantsin thegenerationscenario were notgivena specificprice rangeAll respondentswere also informed that the regularairfare is$459mdasha manipu-lationdesignedtoprovidea more stringenttestof theobservedeffects (recall that the data from Experiment1 indicatethat theavailability of a reference price point is likely to mitigate theobserved reference price range effects)

The range of prices in the selection and the augmentedgeneration conditions was identical and was designed to beof negligible diagnostic value to consumers For that pur-pose a pretest was conducted in which 24 participants fromthe same population were presented with the experimentalscenario and were asked to indicate the lowest price at whichthey would expect to find tickets for the specified route Theaverage of the lowest available prices suggested by the pre-test sample was $184 Based on these data the lower end ofthe range of available prices in the selection and augmentedgeneration tasks was set at $189mdashin line with respondentsrsquo

54 CHERNEV

expectations The upper end of the price range was set toequal the regular airfare ($459) which was provided to par-ticipants in all experimental conditions

When presented with the experimental scenario respon-dents were asked to evaluate the relative difficulty of stu-dentsrsquo decisions by positioning Students A B and C alonga 150-mm line with endpoints ldquoVery easyrdquo and ldquoVery diffi-cultrdquo This procedure yielded two measures of decision dif-ficulty a ranking of the relative difficulty of the pricing de-cisions based on their ordering on the scale and a moreprecise metric measure of the decision difficulty obtainedby measuring the absolute distance from the left end of thescale (Very easy) to the marks corresponding to the per-ceived decision difficulty of each of the three pricing sce-narios

Next participants were asked to evaluate each studentrsquosconfidence in their decision as well as their perception of theprobability of each student getting a ticket at their statedprice Participant responses were measured on the same typeof scale as the decision difficulty scale described previously(endpointsNot confident at all and Very confident for the de-cision-confidencescale and Low probabilityand High prob-ability for the probability of a successful bid) Finally re-spondents were asked to indicate their preference for each ofthese pricing methods (ldquoIf you have to use one of these agen-cies which one would you userdquo)

Participants were tested in groups following their regularclass meetings They worked at their own pace and were de-briefed on completion of the experiment As an incentive fortheir participation several drawings were conducted for cashprizes of $50

Results

The data show a preference for price-elicitation tasks thatoffer a readily available price range The generation taskwas associated with the lowest confidence Sixty-eight per-cent of the responses pointed to the generation task as in-stilling the least confidence compared to 22 for the selec-tion task and 10 for the augmented generation task c2(2)= 552 p lt 001 The difference between the generation andboth the selection and augmented generation was signifi-cant at the 001 level c2(1) = 272 c2(1) = 546 and thedifference between the selection and the augmented genera-tion tasks was marginally significant c2(1) = 275 p lt 10The confidence evaluations revealed through the metricmeasure of scale distances were consistent with the rank-ings data The generation task was associated with the low-est confidence (M = 416 on a 100-point scale) and the se-lection and the augmented generation tasks were associatedwith higher confidence (M = 561 and M = 567 respec-tively) The difference between the generation and the se-lection task was significant F(1 48) = 873 p lt 01 as wasthe difference between the generation and the augmentedgeneration task F(1 48) = 1125 p lt 005 The difference

between the selection and the augmented generation taskwas non-significant F(1 48) lt 1

Decision-difficulty data displayed a pattern similar tothe decision confidence evaluations whereby higher levelsof confidence were associated with easier decisions Spe-cifically 62 of the respondents found the generation taskto be the most difficult 24 considered the selection taskto be the most difficult and only 6 viewed the augmentedgeneration as the most difficult c2(2) = 6148 p lt 001The difference between the generation and the selectiontask was significant c2(1) = 1902 p lt 001 as were thedifferences between the generation versus the augmentedgeneration c2(1) = 6071 p lt 001 and the augmented gen-eration versus the selection tasks c2(1) = 694 p lt 01Metric measures of decision difficulty also show a signifi-cant decrease in the perceived decision difficulty in condi-tions where reference prices were present Once again thegeneration task was perceived the most difficult (M = 603)compared to the selection and the augmented generationtasks (M = 431 and M = 411 respectively) The differencebetween the generation and the selection task was signifi-cant F(1 44) = 1621 p lt 001 as was the difference be-tween the generation and the augmented generation taskF(1 44) = 1916 p lt 001 The difference between the se-lection and the augmented generation tasks wasnonsignificant F(1 44) lt 1

With respect to evaluating the success probability of theirbid participantsranked the generation task as having the low-est probability of yielding a successful outcome (Figure 2)The selection task was perceived as most likely to yield a suc-cessfulbid(70 of the responses)followedby theaugmentedgeneration task (26) Only 4 of the respondentsperceivedthegenerationtask tohavethehighestprobabilityofyieldingasuccessful outcome The difference in rankings between theselection and the augmented generation task (70 vs 26)

PRICE ELICITATION STRATEGIES IN CHOICE 55

FIGURE 2 Predicted success of the different price elicitation tasks(Experiment 2)

was significant c2(1) = 2406 p lt 001 as was the differencebetween the generation and the augmented generation taskc2(1) = 1049p lt 005The metric distancemeasures revealedsimilar distribution of participantsrsquoevaluations of the proba-bilityof a successfuloutcomeThe average rating of the likeli-hood of a successful bid through a generation task was M =337 compared to M = 642 for the selection task F(1 48) =8122 p lt 001 and M = 578 for the augmented generationtask F(1 48)= 4823p lt 001The differencebetween the se-lection and the augmented generation task was significant aswell F(1 48) = 427 p lt 05

In addition to the projective measures of generation par-ticipants were asked to indicate their own preference for theprice-elicitation task The majority of the respondents (56)chose the augmented generation task 36 indicated a pref-erence for the selection task and only 16 indicated thatthey would prefer the generation task (Figure 3) The differ-ence between the augmented generation and the selectiontask was significant c2(1) = 419 p lt 05 and the differencebetween the selection and the augmented generationwas alsosignificant c2(1) = 548 p lt 05

DISCUSSION

Experiments 1 and 2 document that the impact of the natureof the price-articulation task is moderated by the referenceprice availability In Experiment 1 participants were pre-sented with an external reference price representing thehigher end of the price range (PMAX) and in Experiment 2respondents were given the lower end of the price range(PMIN) as well In both experiments providing participantswith reference price information had a significant effect ontheir decision process Specifically respondentsrsquo responsesindicated higher levels of confidence associated with the se-

lection compared to the generation task The selection taskwas also rated higher in terms of its predicted likelihood ofsuccess this task was preferred by consumers as wellThese data also demonstrate the robustness of the observedpreference for the selection task Recall that all participantswere already given a reference price (PMAX) and yet theystill indicated stronger preference for the elicitation taskswhen the entire price range was readily available

The augmented generation task was introduced in this ex-periment to test the propositionthat the hypothesizedeffect ofthe nature of the decision task on consumer preferences is afunction of reference price availabilityIn fact if the presenceofa readilyavailablepricerangeis thefactordrivingconsumerpreference for the selection task then making the price rangereadily available in the augmented generation conditionshould also increase consumer preference for the generationtask Moreover because respondents in both generation andaugmentedgenerationconditionswere offered thesame flexi-bility in expressing their willingness to pay the observed ef-fectscouldnotbedirectlyattributedto thedifferenceinelicita-tionflexibilitybetweenthegenerationand theselectiontasks

The data from the selection and the augmented genera-tion tasks offer further insights into the nature of price ar-ticulation Because the price ranges and mid-range valuesin the selection and the augmented generation tasks wereidentical the only difference between these tasks was theelicitation flexibilitymdashthe degree of freedom consumershave to select the price that most adequately reflects theirwillingness to pay These tasks however were ranked asmarkedly different in terms of the probability of successwhereby nearly three times as many of the respondents in-dicated their expectation that the selection task would likelylead to a more successful outcome compared to the aug-mented generation task Interesting though when indicatingtheir own preference for a price-elicitation mechanism par-ticipants showed a significantly stronger preference for theaugmented generation task (56 vs 36) These data es-sentially suggest a reversal of preferences Although con-sumers evaluated the selection task as having the highestlikelihood of resulting in a successful bid when given achoice they preferred the augmented generation task Infact of all respondents who indicated that the selection taskis likely to be the most successful 46 indicated that ifgiven a choice they would prefer the augmented generationtask thus effectively reversing their preferences

One possible explanation for the observed data is thatwhen presented with the choice of a price-generationmethod consumers were not willing to sacrifice the elicita-tion accuracy associated with the augmented generationtask for the simplicity associated with the selection taskyet when presented with a scenario in which the price-gen-eration mode has already been selected they evaluated it ashaving a higher probability of success This finding impliesthat when forecasting their own ability to articulate a suc-

56 CHERNEV

FIGURE 3 Price elicitation task preferences (Experiment 2)

cessful bid consumers essentially have overweighed theirown ability to generate a successful bid This data pattern isconsistent with the research on decision confidence indicat-ing that in many scenarios individuals are likely to displaya tendency toward overconfidence in their decisions(Allwood amp Montgomery 1987 Griffin amp Tversky 1992Kahneman amp Tversky 1973 Mahajan 1992)

Conceptually this article argues that the observedpreference for the selection task is driven to a large degreeby the complex two-stage nature of the decision taskwhich involves (a) eliciting a readily available price rangeand then (b) generating a price reflecting consumersrsquo will-ingness to pay As a result consumers are likely to discountthe preference-articulation flexibility offered by the genera-tion task and prefer the simpler although more restrictiveselection task This prediction was tested by introducing theaugmented generation condition which simplified thedecision task by offering consumers a readily availableprice range Consistent with this proposition the data showthat the advantage of the selection task is eliminated bymaking the price range readily available Participants in theaugmented generation condition were more confident intheir decisions predicted that prices generated through theaugmented generation task would most likely result in asuccessful bid and indicated a higher overall preferencefor this task as well

Note that so far the impact of the availability ofreference price range was tested by giving respondents arange of possible values in the augmented generation taskThis range was set in a way that minimizes its diagnosticvalue specifically by a pretest asking participants to gener-ate the lowest expected price and then setting the experi-mental range consistent with respondentsrsquo expectationsNevertheless one can argue that providing an external pricerange helped participants validate their initial price rangeexpectations raising consumer confidence in their intuitiveprice range estimates This increased confidence in the va-lidity of their intuitive estimates of the lower end of theprice range could in turn have confounded the resultscontributing to the increase in respondentsrsquo decision confi-dence

To address this possible confound in Experiment 3 theavailability of a reference price range is manipulated byasking participants prior to the choice task to indicate theirexpectation of the lowest possible prices Thus unlike inExperiment 2 where the reference price range was providedby the experimental stimuli in Experiment 3 this range isinternally generated by respondents This manipulation of-fers an alternative strategy to test the theory advanced inthis research by varying the source of the reference pricerange If the theory is valid its predictions should holdwhen the reference price ranges are internally generated aswell The specifics of Experiment 3 and the experimentalresults are presented in more detail in the following section

EXPERIMENT 3

Experiment 3 aims to provide further evidence that the per-ceived advantage of the selection over the generation task iscaused by the complexity of the decision process associatedwith the absence of readily available reference prices In thisexperiment participants were asked to generate price rangesbefore articulatingtheirwillingnessto pay This manipulationallows respondents to have reference prices readily availablefor use as benchmarks in the price-elicitation process Thusthe goal is to show that a simple restructuring of the decisiontask in a generation-ready format can affect consumer prefer-ences for the price-articulationstrategy even in the absence ofexternally provided reference price information

Method

Participants were asked to imagine that they must fly to SanFrancisco for a high-tech job fair and were asked to book aircar and hotel through an online reverse pricing travel agencyOverall the initial scenario was similar to the one used in Ex-periment 2 except that (a) Experiment 3 respondents had tomake decisions for themselves (rather than in the context of aprojective task) and (b) participants had to make three deci-sions in each of three product categories (air car hotel)

Prior to stating their price some of the respondents wereasked to indicate the lowest possible price available for book-ing in each of the three categories For example participantswere told ldquoThe regular price for an airline ticket to San Fran-cisco is $459 What do you think is the lowest price that atravel agency could possibly negotiate with an airline for aflight to San Francisco at any time (regularly priced at$459)rdquo The purpose of this manipulation was for respon-dents to articulate a reference point that would be readilyavailable for the price-generation task This was thepre-choice articulation condition In contrast participants inthe control condition were informed of the regular price butwere not asked to indicate their expectation of the lowestavailable price Thus the only difference between the twoconditionswas that respondents in the pre-choice articulationcondition were asked to indicate their expectationof the low-est available price in each of the three product categoriesNote that participants in the pre-choice articulation conditiondid not receive any additional information compared to therespondents in the control condition

The experimental design was 2 (reference price availabil-ity pre-choice articulation vs control) times 3 (price elicitationtask generation vs selection vs augmented generation)Thepre-choice articulation manipulationwas conductedbetweenparticipantsrsquo and the price-elicitation task was conductedwithin-subjects Each participant had to elicit three pricesone in each product category (air car and hotel) To avoidpossible confounds between the product categories used in

PRICE ELICITATION STRATEGIES IN CHOICE 57

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

(ie ldquoname your pricerdquo) and price selection (ie ldquoselect yourpricerdquo) The former approach advanced by Priceline simplyasks consumers to state the price they are willing to pay for theproduct underconsiderationIn the latter approach presentedin this research consumersare presentedwith a set of possiblepricesandaskedto selectthepricetheyfindmostacceptable

The price-generationscenario is clearly more flexible thantheselectionscenariobecauseitallowsconsumerstopreciselyarticulate their willingness to pay Indeed in the generationscenario consumers have virtually unlimited degrees of free-dom to state their price whereas in the selection scenario theyare restricted by the set of prices presented to them Viewedfrom an economicsstandpointthis flexibilityis oneof the rea-sons why the ldquoname yourpricerdquo strategy would be consideredsuperior to the selection strategy assuming that consumershave establishedpreferences that can easily be translated intomonetary terms (eg Varian 1999) In fact the limited selec-tion constrains consumersrsquoability to adequately express theirwillingness to pay and as with any constrained optimizationwouldbeconsideredinferior toan unconstrainedoptimizationsuch as the price generation scenario

Yet because price generation assumes established prefer-ences and predetermined willingness to pay it can be arguedthat its impact on the consumer decision process will dependon the degree to which consumers are able to articulate theirproduct utility in monetary terms Building on this notion Ipropose that in the absence of a readily available task-specificcontext (eg reference price range) the generationstrategy islikely to be associatedwith a greater degree of uncertaintyandcognitiveeffort andas a result will be perceived to be inferiorto thesimplerprice selectiontaskThispropositionisalsosup-ported by the view of consumers as contingentdecision mak-ers whorather thanhavingasetofpredeterminedpreferencesform their preferences in the context of the specific decisiontask (Bettman Luce amp Payne 1998)

In the remainder of this article I develop the conceptualframework and derive specific predictions regarding the ef-fects of the generationand selectiontaskson theconsumerde-cision process Next I describe and report the results of threeexperiments that test these predictionsI concludeby discuss-ing the implications of the findings and providing directionsfor further research

PRICE ARTICULATIONIN CONSUMER CHOICE

To set the optimal price for a given product consumers mustexpress their overall evaluation as a monetary amount Re-cent research has proposed that consumers do this by a pro-cess of subjective interpolation whereby they search for apoint on the monetary scale that corresponds to their ex-pected utility from the product (Goldstein amp Einhorn 1987Janiszewski amp Lichtenstein 1999 see also Lynch

Chakravarti amp Mitra 1991) To illustrate consider Figure 1depicting a monetary scale with endpoints PMIN and PMAXwhere PMIN is a consumerrsquos perception of the lowest price forthe given product and PMAX is the highest expected priceThis monetary scale is linked to a utility scale with endpointsU(PMIN) and U(PMAX) where U(PMIN) is the utility at thelowest price and U(PMAX) is the utility at the highest ex-pected price

In this context the consumersrsquo task is to determine theprice (P) that corresponds to the utility that they expect to re-ceive from the product U(P) To express their utility in mone-tary terms consumers are assumed to establish a correspon-dence between the endpoints of the utility scale and theendpointsof the monetary scale Consequentlytheir willing-ness to pay stated in monetary terms is calculated to corre-spond to the proportional relations between U(PMIN)U(PMAX) and U(P) In other words consumers elicit theirwillingness to pay by equating the endpoints and matchingthe proportional change in the utility scale with a propor-tional change on the monetary scale

To illustrate consider a consumer who believes that thelowest airfare for a flight between Chicago and San Fran-cisco is $189 and the regular fare is $459 Based on the utilityderived from purchasing a ticket at these fares the consumerdetermines the utility from the particular offering under con-sideration which is then expressed in monetary terms (say$260) to correspond to relations between U($189) U($459)and U($260) Note that this subjective interpolation processhinges on the availability of a reference price range (egPMAX and PMIN) which is used as a benchmark in the pricearticulation process The importance of reference points inprice perception and evaluation has been underscored by nu-merous researchers in the field (Alba Broniarczyk Shimpamp Urbany 1994 Heath Chatterjee amp France 1995 Herr1989 Lichtenstein et al 1988 Mayhew amp Winer 1992Urbany amp Dickson 1991)

Yet on many occasions consumers do not have readilyavailable reference prices and as a result they lackbenchmarks needed to indicate their willingness to pay for agiven product In this case consumers must first evoke therelevant reference prices and then use these prices asbenchmarks in eliciting their willingness to pay To illustrate

52 CHERNEV

FIGURE 1 Reverse pricing as a subjective interpolation process

using the example presented in Figure 1 when PMIN andorPMAX are not present consumers first have to estimate thelikely values of PMIN andor PMAX After the range of possibleprices has been established consumers can assess the utilityassociated with these values [U(PMIN) and U(PMAX)] anddetermine the price they are willing to pay (P) based on theutility derived from this price [U(P)] As a result whenreference prices are not readily available price articulationcan be represented as a two-stage process whereby consum-ers first need to evoke the range of possible values and thenusing utilities of these values as benchmarks determine theirutility for the product and articulate this utility on a monetaryscale

Following this line of reasoning for consumers withoutreadily available reference price ranges price elicitation willbe associated with substantial effort because of the complexnature of the decision task involving reference price estima-tion in addition to the price-generation As a resultprice-generation tasks presented in a context where a refer-ence price range is not readily present are likely to be rela-tively more difficult compared to elicitation tasks with a sa-lient range of possible prices Building on this argument Ipropose that when a reference price range is not readilyavailable consumers faced with the more complex price-generation task are likely to be less confident in their deci-sions compared to consumers who are presented with the rel-atively simpler selection task Indeed the selection taskinherently presents consumers with a range of referenceprices and a consumerrsquos decision in this case is reduced toselecting one of the available price options This perceivedsimplicity of the decision task will also lead to an overallpreference for the selection task despite the fact that it ismore restrictive and hence less accurate This preference forthe selection-based price elicitation can be viewed as an ef-fortndashaccuracy tradeoff whereby consumers give up accuracyin expressing their willingness to pay in favor of making aless effortful decision (Payne Bettman amp Johnson 1993 Si-mon 1955)

In sum I propose that the availability of salient referenceprices moderates the impact of the price-elicitation task onconsumer preferences Specifically in the absence of areadily available reference price range consumers will likelytrade off the accuracy in expressing their willingness to paythat the price-generation task offers for the simpler althoughless flexible selection task In contrast in contexts wherereference prices are readily available consumers will be lesslikely to discount the price-generation task on the basis of itsrelative complexity and consequently will be less likely todisplay a stronger preference for the price-selection task

These predictionsare tested in a series of experiments thatcompare the two price-elicitation scenarios generationversus selection and examine consumer preferences forthese pricing strategies as a function of reference price avail-ability

EXPERIMENT 1

This experiment investigatesthe impact of the availabilityof areference priceonconsumerpreferencesfor a price-elicitationtask The goal is to offer preliminary evidence that in the ab-senceof establishedreference prices consumerswill prefer toexpress their willingness to pay using a selection rather than agenerationtaskThirty-tworespondentswere presentedwithahypothetical scenario in which they were asked to purchaseairline tickets from two online agencies Both agencies em-ployed reverse pricing but used different price-elicitationstrategies One agency asked consumers to generate (name) aprice whereas the other asked consumers to select from a listof 10 availableprices (from $169 to $439 in ascendingorderincreasing at $30 increments) This price-elicitation manipu-lationwas conductedwithinsubjectsIn additionrespondentsin the reference price condition were given a referencepricemdashthey were told that the regular airfare is$439mdashwhereas the others were not given any price informa-tion

Participants were randomly assigned to the conditionsof a2 (price-elicitation task generation vs selection) times 2 (refer-ence price availablevs not available)mixed factorial designThey were asked to indicate the price they were willing to payfor the ticketsand were told that if theirbidswere not success-ful they would have theoptionto purchase the ticketat the reg-ularpriceParticipantswere asked to indicatetheir confidencein the decisionas well as their expectationsof the likelihoodofsuccessof theirbidusinga 150-mmnongradedscalewith end-pointsNot confidentversus Very confidentand Not likely ver-susVery likely(seeBettmanJohnampScott1986for detailsonthe measurement procedure) Participantsrsquoresponses were re-cordedbymeasuringthedistancefrom the left endof thescaleand these responses were standardized to correspond to a100-point scale

The data show that in the absence of a reference price in-dividuals were more confident in the context of a selectiontask (M = 411) than in the context of a generation task (M =275) The direction of the effect remained the same evenwhen a reference price was present although respondents inthis condition were more confident compared to respondentsin the condition where a reference price was not available (M= 514 for the selection task vs M = 397 for the generationtask) The data show a significant main effect of the price-ar-ticulation task on decision confidence whereby the genera-tion task was associated with lower confidence compared tothe selection task F(1 30) = 3348 p lt 001 There was also amarginally significant main effect of the availability of refer-ence price participants were more confident in conditionswhere a reference price was readily available F(1 30) =377 p lt 10 Participantsrsquo evaluations of the likelihood ofsuccess of their bid followed a similar pattern Respondentsexpected prices derived from the selection task to have ahigher probability of being accepted compared to prices elic-

PRICE ELICITATION STRATEGIES IN CHOICE 53

ited through the generation task (M = 506 vs M = 327) F(130 ) = 4301 p lt 001 They also predicted a higher successprobability in conditions where a reference price was presentcompared to conditions where a reference price was notreadily available (M = 461 vs M = 372) F(1 30) = 423 p lt05

The data from this experiment suggest that consumers feelmore confident in the outcome of a selection rather than ageneration task This effect was also significant althoughless pronounced in the presence of a readily available refer-ence price Furthermore consumers perceived the selectiontask to have a higher likelihood of success compared to thegeneration task an effect more pronounced when a referenceprice was not readily available

Note that in Experiment 1 the potential effect of availabil-ity of a reference price range is confounded with the effectsassociated with the nature of the elicitation task Indeed con-sumer preference for a selection price mode can be the resultof an attempt to simplify the pricing decision by selecting apricing strategy with fewer possible alternatives Thus con-sistent with the effortndashaccuracy framework (Payne et al1993) one can argue that consumers will prefer the selectiontask because it offers a simpler (although less accurate) deci-sion and not necessarily because in this case a reference pricepoint is readily available This potential confound is ad-dressed in Experiment 2 by introducingan experimental con-dition (referred to as augmented generation) designed to dis-entangle the effects of reference price range availability anddecision effort associated with generating a precise pricepoint

EXPERIMENT 2

Building on Experiment 1 findings Experiment 2 examinesthe impact of the price-elicitation task on the strength of con-sumer preferences and how the availability of a referenceprice range moderates this impact This experiment investi-gates three scenarios price generation price selection and athird scenario referred to as augmented generation Partici-pants in the generation scenario were asked to name theirown price and were not provided with a specific price rangeRespondents in the selection condition were given a set of 10alternative prices to choose from Finally in the augmentedgeneration scenario (discussed in more detail later) partici-pants were asked to name their own price and were providedwith a price range

Method

Respondents were presented with the following scenario

Imagine that three of your classmates had to fly to SanFrancisco for a high-tech job fair They all decided tobuy their air tickets through a reverse pricing Web

travel agency that allows travelers to name their ownprice

Reverse pricing works as follows You tell the travelagency where and when you want to go and how muchyou are willing to pay The agency then searches for anairline willing to release seats at your price If theagency finds tickets at your price it will immediatelypurchase those tickets for you Because you get toname your own price tickets purchased through re-verse pricing Internet agencies cannot be changedtransferred or cancelled

The travel agency will make the best effort to makethe booking It will submit your and other customersrsquorequests to its airline partners who will then decide onthe lowest price to accept Everyone whose bid ishigher or equal to that price or higher will be granted aticket at the stated price Everyone else will have tomake their bookingsdirectly and pay the regular price

Next participants were told that Students A B and C are us-ing different reverse pricing travel agencies that employ dif-ferent procedures to allow customers to name their priceStudent A was asked to name the price he or she is willing topay (generation task) Student B was given a list of 10 pricesand asked to choose among them (selection task) and Stu-dent C was shown a scale that depicts the typical price rangeand asked to indicate the price he or she is willing to pay bypositioninga slider on the scale (augmented generation task)These price-generation tasks are presented in more detail inthe Appendix

Note that respondents in both generation and augmentedgenerationconditionswere offered agreaterdegreeof flexibil-ity in elicitingtheirwillingnessto paywhereas participantsinthe selection condition were limited to a list of 10 availableprices In addition respondents in the selection and the aug-mented generationconditionswere implicitlyprovidedwith aprice range (determined by the highest and the lowest pricesavailablefor selection)whereasparticipantsin thegenerationscenario were notgivena specificprice rangeAll respondentswere also informed that the regularairfare is$459mdasha manipu-lationdesignedtoprovidea more stringenttestof theobservedeffects (recall that the data from Experiment1 indicatethat theavailability of a reference price point is likely to mitigate theobserved reference price range effects)

The range of prices in the selection and the augmentedgeneration conditions was identical and was designed to beof negligible diagnostic value to consumers For that pur-pose a pretest was conducted in which 24 participants fromthe same population were presented with the experimentalscenario and were asked to indicate the lowest price at whichthey would expect to find tickets for the specified route Theaverage of the lowest available prices suggested by the pre-test sample was $184 Based on these data the lower end ofthe range of available prices in the selection and augmentedgeneration tasks was set at $189mdashin line with respondentsrsquo

54 CHERNEV

expectations The upper end of the price range was set toequal the regular airfare ($459) which was provided to par-ticipants in all experimental conditions

When presented with the experimental scenario respon-dents were asked to evaluate the relative difficulty of stu-dentsrsquo decisions by positioning Students A B and C alonga 150-mm line with endpoints ldquoVery easyrdquo and ldquoVery diffi-cultrdquo This procedure yielded two measures of decision dif-ficulty a ranking of the relative difficulty of the pricing de-cisions based on their ordering on the scale and a moreprecise metric measure of the decision difficulty obtainedby measuring the absolute distance from the left end of thescale (Very easy) to the marks corresponding to the per-ceived decision difficulty of each of the three pricing sce-narios

Next participants were asked to evaluate each studentrsquosconfidence in their decision as well as their perception of theprobability of each student getting a ticket at their statedprice Participant responses were measured on the same typeof scale as the decision difficulty scale described previously(endpointsNot confident at all and Very confident for the de-cision-confidencescale and Low probabilityand High prob-ability for the probability of a successful bid) Finally re-spondents were asked to indicate their preference for each ofthese pricing methods (ldquoIf you have to use one of these agen-cies which one would you userdquo)

Participants were tested in groups following their regularclass meetings They worked at their own pace and were de-briefed on completion of the experiment As an incentive fortheir participation several drawings were conducted for cashprizes of $50

Results

The data show a preference for price-elicitation tasks thatoffer a readily available price range The generation taskwas associated with the lowest confidence Sixty-eight per-cent of the responses pointed to the generation task as in-stilling the least confidence compared to 22 for the selec-tion task and 10 for the augmented generation task c2(2)= 552 p lt 001 The difference between the generation andboth the selection and augmented generation was signifi-cant at the 001 level c2(1) = 272 c2(1) = 546 and thedifference between the selection and the augmented genera-tion tasks was marginally significant c2(1) = 275 p lt 10The confidence evaluations revealed through the metricmeasure of scale distances were consistent with the rank-ings data The generation task was associated with the low-est confidence (M = 416 on a 100-point scale) and the se-lection and the augmented generation tasks were associatedwith higher confidence (M = 561 and M = 567 respec-tively) The difference between the generation and the se-lection task was significant F(1 48) = 873 p lt 01 as wasthe difference between the generation and the augmentedgeneration task F(1 48) = 1125 p lt 005 The difference

between the selection and the augmented generation taskwas non-significant F(1 48) lt 1

Decision-difficulty data displayed a pattern similar tothe decision confidence evaluations whereby higher levelsof confidence were associated with easier decisions Spe-cifically 62 of the respondents found the generation taskto be the most difficult 24 considered the selection taskto be the most difficult and only 6 viewed the augmentedgeneration as the most difficult c2(2) = 6148 p lt 001The difference between the generation and the selectiontask was significant c2(1) = 1902 p lt 001 as were thedifferences between the generation versus the augmentedgeneration c2(1) = 6071 p lt 001 and the augmented gen-eration versus the selection tasks c2(1) = 694 p lt 01Metric measures of decision difficulty also show a signifi-cant decrease in the perceived decision difficulty in condi-tions where reference prices were present Once again thegeneration task was perceived the most difficult (M = 603)compared to the selection and the augmented generationtasks (M = 431 and M = 411 respectively) The differencebetween the generation and the selection task was signifi-cant F(1 44) = 1621 p lt 001 as was the difference be-tween the generation and the augmented generation taskF(1 44) = 1916 p lt 001 The difference between the se-lection and the augmented generation tasks wasnonsignificant F(1 44) lt 1

With respect to evaluating the success probability of theirbid participantsranked the generation task as having the low-est probability of yielding a successful outcome (Figure 2)The selection task was perceived as most likely to yield a suc-cessfulbid(70 of the responses)followedby theaugmentedgeneration task (26) Only 4 of the respondentsperceivedthegenerationtask tohavethehighestprobabilityofyieldingasuccessful outcome The difference in rankings between theselection and the augmented generation task (70 vs 26)

PRICE ELICITATION STRATEGIES IN CHOICE 55

FIGURE 2 Predicted success of the different price elicitation tasks(Experiment 2)

was significant c2(1) = 2406 p lt 001 as was the differencebetween the generation and the augmented generation taskc2(1) = 1049p lt 005The metric distancemeasures revealedsimilar distribution of participantsrsquoevaluations of the proba-bilityof a successfuloutcomeThe average rating of the likeli-hood of a successful bid through a generation task was M =337 compared to M = 642 for the selection task F(1 48) =8122 p lt 001 and M = 578 for the augmented generationtask F(1 48)= 4823p lt 001The differencebetween the se-lection and the augmented generation task was significant aswell F(1 48) = 427 p lt 05

In addition to the projective measures of generation par-ticipants were asked to indicate their own preference for theprice-elicitation task The majority of the respondents (56)chose the augmented generation task 36 indicated a pref-erence for the selection task and only 16 indicated thatthey would prefer the generation task (Figure 3) The differ-ence between the augmented generation and the selectiontask was significant c2(1) = 419 p lt 05 and the differencebetween the selection and the augmented generationwas alsosignificant c2(1) = 548 p lt 05

DISCUSSION

Experiments 1 and 2 document that the impact of the natureof the price-articulation task is moderated by the referenceprice availability In Experiment 1 participants were pre-sented with an external reference price representing thehigher end of the price range (PMAX) and in Experiment 2respondents were given the lower end of the price range(PMIN) as well In both experiments providing participantswith reference price information had a significant effect ontheir decision process Specifically respondentsrsquo responsesindicated higher levels of confidence associated with the se-

lection compared to the generation task The selection taskwas also rated higher in terms of its predicted likelihood ofsuccess this task was preferred by consumers as wellThese data also demonstrate the robustness of the observedpreference for the selection task Recall that all participantswere already given a reference price (PMAX) and yet theystill indicated stronger preference for the elicitation taskswhen the entire price range was readily available

The augmented generation task was introduced in this ex-periment to test the propositionthat the hypothesizedeffect ofthe nature of the decision task on consumer preferences is afunction of reference price availabilityIn fact if the presenceofa readilyavailablepricerangeis thefactordrivingconsumerpreference for the selection task then making the price rangereadily available in the augmented generation conditionshould also increase consumer preference for the generationtask Moreover because respondents in both generation andaugmentedgenerationconditionswere offered thesame flexi-bility in expressing their willingness to pay the observed ef-fectscouldnotbedirectlyattributedto thedifferenceinelicita-tionflexibilitybetweenthegenerationand theselectiontasks

The data from the selection and the augmented genera-tion tasks offer further insights into the nature of price ar-ticulation Because the price ranges and mid-range valuesin the selection and the augmented generation tasks wereidentical the only difference between these tasks was theelicitation flexibilitymdashthe degree of freedom consumershave to select the price that most adequately reflects theirwillingness to pay These tasks however were ranked asmarkedly different in terms of the probability of successwhereby nearly three times as many of the respondents in-dicated their expectation that the selection task would likelylead to a more successful outcome compared to the aug-mented generation task Interesting though when indicatingtheir own preference for a price-elicitation mechanism par-ticipants showed a significantly stronger preference for theaugmented generation task (56 vs 36) These data es-sentially suggest a reversal of preferences Although con-sumers evaluated the selection task as having the highestlikelihood of resulting in a successful bid when given achoice they preferred the augmented generation task Infact of all respondents who indicated that the selection taskis likely to be the most successful 46 indicated that ifgiven a choice they would prefer the augmented generationtask thus effectively reversing their preferences

One possible explanation for the observed data is thatwhen presented with the choice of a price-generationmethod consumers were not willing to sacrifice the elicita-tion accuracy associated with the augmented generationtask for the simplicity associated with the selection taskyet when presented with a scenario in which the price-gen-eration mode has already been selected they evaluated it ashaving a higher probability of success This finding impliesthat when forecasting their own ability to articulate a suc-

56 CHERNEV

FIGURE 3 Price elicitation task preferences (Experiment 2)

cessful bid consumers essentially have overweighed theirown ability to generate a successful bid This data pattern isconsistent with the research on decision confidence indicat-ing that in many scenarios individuals are likely to displaya tendency toward overconfidence in their decisions(Allwood amp Montgomery 1987 Griffin amp Tversky 1992Kahneman amp Tversky 1973 Mahajan 1992)

Conceptually this article argues that the observedpreference for the selection task is driven to a large degreeby the complex two-stage nature of the decision taskwhich involves (a) eliciting a readily available price rangeand then (b) generating a price reflecting consumersrsquo will-ingness to pay As a result consumers are likely to discountthe preference-articulation flexibility offered by the genera-tion task and prefer the simpler although more restrictiveselection task This prediction was tested by introducing theaugmented generation condition which simplified thedecision task by offering consumers a readily availableprice range Consistent with this proposition the data showthat the advantage of the selection task is eliminated bymaking the price range readily available Participants in theaugmented generation condition were more confident intheir decisions predicted that prices generated through theaugmented generation task would most likely result in asuccessful bid and indicated a higher overall preferencefor this task as well

Note that so far the impact of the availability ofreference price range was tested by giving respondents arange of possible values in the augmented generation taskThis range was set in a way that minimizes its diagnosticvalue specifically by a pretest asking participants to gener-ate the lowest expected price and then setting the experi-mental range consistent with respondentsrsquo expectationsNevertheless one can argue that providing an external pricerange helped participants validate their initial price rangeexpectations raising consumer confidence in their intuitiveprice range estimates This increased confidence in the va-lidity of their intuitive estimates of the lower end of theprice range could in turn have confounded the resultscontributing to the increase in respondentsrsquo decision confi-dence

To address this possible confound in Experiment 3 theavailability of a reference price range is manipulated byasking participants prior to the choice task to indicate theirexpectation of the lowest possible prices Thus unlike inExperiment 2 where the reference price range was providedby the experimental stimuli in Experiment 3 this range isinternally generated by respondents This manipulation of-fers an alternative strategy to test the theory advanced inthis research by varying the source of the reference pricerange If the theory is valid its predictions should holdwhen the reference price ranges are internally generated aswell The specifics of Experiment 3 and the experimentalresults are presented in more detail in the following section

EXPERIMENT 3

Experiment 3 aims to provide further evidence that the per-ceived advantage of the selection over the generation task iscaused by the complexity of the decision process associatedwith the absence of readily available reference prices In thisexperiment participants were asked to generate price rangesbefore articulatingtheirwillingnessto pay This manipulationallows respondents to have reference prices readily availablefor use as benchmarks in the price-elicitation process Thusthe goal is to show that a simple restructuring of the decisiontask in a generation-ready format can affect consumer prefer-ences for the price-articulationstrategy even in the absence ofexternally provided reference price information

Method

Participants were asked to imagine that they must fly to SanFrancisco for a high-tech job fair and were asked to book aircar and hotel through an online reverse pricing travel agencyOverall the initial scenario was similar to the one used in Ex-periment 2 except that (a) Experiment 3 respondents had tomake decisions for themselves (rather than in the context of aprojective task) and (b) participants had to make three deci-sions in each of three product categories (air car hotel)

Prior to stating their price some of the respondents wereasked to indicate the lowest possible price available for book-ing in each of the three categories For example participantswere told ldquoThe regular price for an airline ticket to San Fran-cisco is $459 What do you think is the lowest price that atravel agency could possibly negotiate with an airline for aflight to San Francisco at any time (regularly priced at$459)rdquo The purpose of this manipulation was for respon-dents to articulate a reference point that would be readilyavailable for the price-generation task This was thepre-choice articulation condition In contrast participants inthe control condition were informed of the regular price butwere not asked to indicate their expectation of the lowestavailable price Thus the only difference between the twoconditionswas that respondents in the pre-choice articulationcondition were asked to indicate their expectationof the low-est available price in each of the three product categoriesNote that participants in the pre-choice articulation conditiondid not receive any additional information compared to therespondents in the control condition

The experimental design was 2 (reference price availabil-ity pre-choice articulation vs control) times 3 (price elicitationtask generation vs selection vs augmented generation)Thepre-choice articulation manipulationwas conductedbetweenparticipantsrsquo and the price-elicitation task was conductedwithin-subjects Each participant had to elicit three pricesone in each product category (air car and hotel) To avoidpossible confounds between the product categories used in

PRICE ELICITATION STRATEGIES IN CHOICE 57

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

using the example presented in Figure 1 when PMIN andorPMAX are not present consumers first have to estimate thelikely values of PMIN andor PMAX After the range of possibleprices has been established consumers can assess the utilityassociated with these values [U(PMIN) and U(PMAX)] anddetermine the price they are willing to pay (P) based on theutility derived from this price [U(P)] As a result whenreference prices are not readily available price articulationcan be represented as a two-stage process whereby consum-ers first need to evoke the range of possible values and thenusing utilities of these values as benchmarks determine theirutility for the product and articulate this utility on a monetaryscale

Following this line of reasoning for consumers withoutreadily available reference price ranges price elicitation willbe associated with substantial effort because of the complexnature of the decision task involving reference price estima-tion in addition to the price-generation As a resultprice-generation tasks presented in a context where a refer-ence price range is not readily present are likely to be rela-tively more difficult compared to elicitation tasks with a sa-lient range of possible prices Building on this argument Ipropose that when a reference price range is not readilyavailable consumers faced with the more complex price-generation task are likely to be less confident in their deci-sions compared to consumers who are presented with the rel-atively simpler selection task Indeed the selection taskinherently presents consumers with a range of referenceprices and a consumerrsquos decision in this case is reduced toselecting one of the available price options This perceivedsimplicity of the decision task will also lead to an overallpreference for the selection task despite the fact that it ismore restrictive and hence less accurate This preference forthe selection-based price elicitation can be viewed as an ef-fortndashaccuracy tradeoff whereby consumers give up accuracyin expressing their willingness to pay in favor of making aless effortful decision (Payne Bettman amp Johnson 1993 Si-mon 1955)

In sum I propose that the availability of salient referenceprices moderates the impact of the price-elicitation task onconsumer preferences Specifically in the absence of areadily available reference price range consumers will likelytrade off the accuracy in expressing their willingness to paythat the price-generation task offers for the simpler althoughless flexible selection task In contrast in contexts wherereference prices are readily available consumers will be lesslikely to discount the price-generation task on the basis of itsrelative complexity and consequently will be less likely todisplay a stronger preference for the price-selection task

These predictionsare tested in a series of experiments thatcompare the two price-elicitation scenarios generationversus selection and examine consumer preferences forthese pricing strategies as a function of reference price avail-ability

EXPERIMENT 1

This experiment investigatesthe impact of the availabilityof areference priceonconsumerpreferencesfor a price-elicitationtask The goal is to offer preliminary evidence that in the ab-senceof establishedreference prices consumerswill prefer toexpress their willingness to pay using a selection rather than agenerationtaskThirty-tworespondentswere presentedwithahypothetical scenario in which they were asked to purchaseairline tickets from two online agencies Both agencies em-ployed reverse pricing but used different price-elicitationstrategies One agency asked consumers to generate (name) aprice whereas the other asked consumers to select from a listof 10 availableprices (from $169 to $439 in ascendingorderincreasing at $30 increments) This price-elicitation manipu-lationwas conductedwithinsubjectsIn additionrespondentsin the reference price condition were given a referencepricemdashthey were told that the regular airfare is$439mdashwhereas the others were not given any price informa-tion

Participants were randomly assigned to the conditionsof a2 (price-elicitation task generation vs selection) times 2 (refer-ence price availablevs not available)mixed factorial designThey were asked to indicate the price they were willing to payfor the ticketsand were told that if theirbidswere not success-ful they would have theoptionto purchase the ticketat the reg-ularpriceParticipantswere asked to indicatetheir confidencein the decisionas well as their expectationsof the likelihoodofsuccessof theirbidusinga 150-mmnongradedscalewith end-pointsNot confidentversus Very confidentand Not likely ver-susVery likely(seeBettmanJohnampScott1986for detailsonthe measurement procedure) Participantsrsquoresponses were re-cordedbymeasuringthedistancefrom the left endof thescaleand these responses were standardized to correspond to a100-point scale

The data show that in the absence of a reference price in-dividuals were more confident in the context of a selectiontask (M = 411) than in the context of a generation task (M =275) The direction of the effect remained the same evenwhen a reference price was present although respondents inthis condition were more confident compared to respondentsin the condition where a reference price was not available (M= 514 for the selection task vs M = 397 for the generationtask) The data show a significant main effect of the price-ar-ticulation task on decision confidence whereby the genera-tion task was associated with lower confidence compared tothe selection task F(1 30) = 3348 p lt 001 There was also amarginally significant main effect of the availability of refer-ence price participants were more confident in conditionswhere a reference price was readily available F(1 30) =377 p lt 10 Participantsrsquo evaluations of the likelihood ofsuccess of their bid followed a similar pattern Respondentsexpected prices derived from the selection task to have ahigher probability of being accepted compared to prices elic-

PRICE ELICITATION STRATEGIES IN CHOICE 53

ited through the generation task (M = 506 vs M = 327) F(130 ) = 4301 p lt 001 They also predicted a higher successprobability in conditions where a reference price was presentcompared to conditions where a reference price was notreadily available (M = 461 vs M = 372) F(1 30) = 423 p lt05

The data from this experiment suggest that consumers feelmore confident in the outcome of a selection rather than ageneration task This effect was also significant althoughless pronounced in the presence of a readily available refer-ence price Furthermore consumers perceived the selectiontask to have a higher likelihood of success compared to thegeneration task an effect more pronounced when a referenceprice was not readily available

Note that in Experiment 1 the potential effect of availabil-ity of a reference price range is confounded with the effectsassociated with the nature of the elicitation task Indeed con-sumer preference for a selection price mode can be the resultof an attempt to simplify the pricing decision by selecting apricing strategy with fewer possible alternatives Thus con-sistent with the effortndashaccuracy framework (Payne et al1993) one can argue that consumers will prefer the selectiontask because it offers a simpler (although less accurate) deci-sion and not necessarily because in this case a reference pricepoint is readily available This potential confound is ad-dressed in Experiment 2 by introducingan experimental con-dition (referred to as augmented generation) designed to dis-entangle the effects of reference price range availability anddecision effort associated with generating a precise pricepoint

EXPERIMENT 2

Building on Experiment 1 findings Experiment 2 examinesthe impact of the price-elicitation task on the strength of con-sumer preferences and how the availability of a referenceprice range moderates this impact This experiment investi-gates three scenarios price generation price selection and athird scenario referred to as augmented generation Partici-pants in the generation scenario were asked to name theirown price and were not provided with a specific price rangeRespondents in the selection condition were given a set of 10alternative prices to choose from Finally in the augmentedgeneration scenario (discussed in more detail later) partici-pants were asked to name their own price and were providedwith a price range

Method

Respondents were presented with the following scenario

Imagine that three of your classmates had to fly to SanFrancisco for a high-tech job fair They all decided tobuy their air tickets through a reverse pricing Web

travel agency that allows travelers to name their ownprice

Reverse pricing works as follows You tell the travelagency where and when you want to go and how muchyou are willing to pay The agency then searches for anairline willing to release seats at your price If theagency finds tickets at your price it will immediatelypurchase those tickets for you Because you get toname your own price tickets purchased through re-verse pricing Internet agencies cannot be changedtransferred or cancelled

The travel agency will make the best effort to makethe booking It will submit your and other customersrsquorequests to its airline partners who will then decide onthe lowest price to accept Everyone whose bid ishigher or equal to that price or higher will be granted aticket at the stated price Everyone else will have tomake their bookingsdirectly and pay the regular price

Next participants were told that Students A B and C are us-ing different reverse pricing travel agencies that employ dif-ferent procedures to allow customers to name their priceStudent A was asked to name the price he or she is willing topay (generation task) Student B was given a list of 10 pricesand asked to choose among them (selection task) and Stu-dent C was shown a scale that depicts the typical price rangeand asked to indicate the price he or she is willing to pay bypositioninga slider on the scale (augmented generation task)These price-generation tasks are presented in more detail inthe Appendix

Note that respondents in both generation and augmentedgenerationconditionswere offered agreaterdegreeof flexibil-ity in elicitingtheirwillingnessto paywhereas participantsinthe selection condition were limited to a list of 10 availableprices In addition respondents in the selection and the aug-mented generationconditionswere implicitlyprovidedwith aprice range (determined by the highest and the lowest pricesavailablefor selection)whereasparticipantsin thegenerationscenario were notgivena specificprice rangeAll respondentswere also informed that the regularairfare is$459mdasha manipu-lationdesignedtoprovidea more stringenttestof theobservedeffects (recall that the data from Experiment1 indicatethat theavailability of a reference price point is likely to mitigate theobserved reference price range effects)

The range of prices in the selection and the augmentedgeneration conditions was identical and was designed to beof negligible diagnostic value to consumers For that pur-pose a pretest was conducted in which 24 participants fromthe same population were presented with the experimentalscenario and were asked to indicate the lowest price at whichthey would expect to find tickets for the specified route Theaverage of the lowest available prices suggested by the pre-test sample was $184 Based on these data the lower end ofthe range of available prices in the selection and augmentedgeneration tasks was set at $189mdashin line with respondentsrsquo

54 CHERNEV

expectations The upper end of the price range was set toequal the regular airfare ($459) which was provided to par-ticipants in all experimental conditions

When presented with the experimental scenario respon-dents were asked to evaluate the relative difficulty of stu-dentsrsquo decisions by positioning Students A B and C alonga 150-mm line with endpoints ldquoVery easyrdquo and ldquoVery diffi-cultrdquo This procedure yielded two measures of decision dif-ficulty a ranking of the relative difficulty of the pricing de-cisions based on their ordering on the scale and a moreprecise metric measure of the decision difficulty obtainedby measuring the absolute distance from the left end of thescale (Very easy) to the marks corresponding to the per-ceived decision difficulty of each of the three pricing sce-narios

Next participants were asked to evaluate each studentrsquosconfidence in their decision as well as their perception of theprobability of each student getting a ticket at their statedprice Participant responses were measured on the same typeof scale as the decision difficulty scale described previously(endpointsNot confident at all and Very confident for the de-cision-confidencescale and Low probabilityand High prob-ability for the probability of a successful bid) Finally re-spondents were asked to indicate their preference for each ofthese pricing methods (ldquoIf you have to use one of these agen-cies which one would you userdquo)

Participants were tested in groups following their regularclass meetings They worked at their own pace and were de-briefed on completion of the experiment As an incentive fortheir participation several drawings were conducted for cashprizes of $50

Results

The data show a preference for price-elicitation tasks thatoffer a readily available price range The generation taskwas associated with the lowest confidence Sixty-eight per-cent of the responses pointed to the generation task as in-stilling the least confidence compared to 22 for the selec-tion task and 10 for the augmented generation task c2(2)= 552 p lt 001 The difference between the generation andboth the selection and augmented generation was signifi-cant at the 001 level c2(1) = 272 c2(1) = 546 and thedifference between the selection and the augmented genera-tion tasks was marginally significant c2(1) = 275 p lt 10The confidence evaluations revealed through the metricmeasure of scale distances were consistent with the rank-ings data The generation task was associated with the low-est confidence (M = 416 on a 100-point scale) and the se-lection and the augmented generation tasks were associatedwith higher confidence (M = 561 and M = 567 respec-tively) The difference between the generation and the se-lection task was significant F(1 48) = 873 p lt 01 as wasthe difference between the generation and the augmentedgeneration task F(1 48) = 1125 p lt 005 The difference

between the selection and the augmented generation taskwas non-significant F(1 48) lt 1

Decision-difficulty data displayed a pattern similar tothe decision confidence evaluations whereby higher levelsof confidence were associated with easier decisions Spe-cifically 62 of the respondents found the generation taskto be the most difficult 24 considered the selection taskto be the most difficult and only 6 viewed the augmentedgeneration as the most difficult c2(2) = 6148 p lt 001The difference between the generation and the selectiontask was significant c2(1) = 1902 p lt 001 as were thedifferences between the generation versus the augmentedgeneration c2(1) = 6071 p lt 001 and the augmented gen-eration versus the selection tasks c2(1) = 694 p lt 01Metric measures of decision difficulty also show a signifi-cant decrease in the perceived decision difficulty in condi-tions where reference prices were present Once again thegeneration task was perceived the most difficult (M = 603)compared to the selection and the augmented generationtasks (M = 431 and M = 411 respectively) The differencebetween the generation and the selection task was signifi-cant F(1 44) = 1621 p lt 001 as was the difference be-tween the generation and the augmented generation taskF(1 44) = 1916 p lt 001 The difference between the se-lection and the augmented generation tasks wasnonsignificant F(1 44) lt 1

With respect to evaluating the success probability of theirbid participantsranked the generation task as having the low-est probability of yielding a successful outcome (Figure 2)The selection task was perceived as most likely to yield a suc-cessfulbid(70 of the responses)followedby theaugmentedgeneration task (26) Only 4 of the respondentsperceivedthegenerationtask tohavethehighestprobabilityofyieldingasuccessful outcome The difference in rankings between theselection and the augmented generation task (70 vs 26)

PRICE ELICITATION STRATEGIES IN CHOICE 55

FIGURE 2 Predicted success of the different price elicitation tasks(Experiment 2)

was significant c2(1) = 2406 p lt 001 as was the differencebetween the generation and the augmented generation taskc2(1) = 1049p lt 005The metric distancemeasures revealedsimilar distribution of participantsrsquoevaluations of the proba-bilityof a successfuloutcomeThe average rating of the likeli-hood of a successful bid through a generation task was M =337 compared to M = 642 for the selection task F(1 48) =8122 p lt 001 and M = 578 for the augmented generationtask F(1 48)= 4823p lt 001The differencebetween the se-lection and the augmented generation task was significant aswell F(1 48) = 427 p lt 05

In addition to the projective measures of generation par-ticipants were asked to indicate their own preference for theprice-elicitation task The majority of the respondents (56)chose the augmented generation task 36 indicated a pref-erence for the selection task and only 16 indicated thatthey would prefer the generation task (Figure 3) The differ-ence between the augmented generation and the selectiontask was significant c2(1) = 419 p lt 05 and the differencebetween the selection and the augmented generationwas alsosignificant c2(1) = 548 p lt 05

DISCUSSION

Experiments 1 and 2 document that the impact of the natureof the price-articulation task is moderated by the referenceprice availability In Experiment 1 participants were pre-sented with an external reference price representing thehigher end of the price range (PMAX) and in Experiment 2respondents were given the lower end of the price range(PMIN) as well In both experiments providing participantswith reference price information had a significant effect ontheir decision process Specifically respondentsrsquo responsesindicated higher levels of confidence associated with the se-

lection compared to the generation task The selection taskwas also rated higher in terms of its predicted likelihood ofsuccess this task was preferred by consumers as wellThese data also demonstrate the robustness of the observedpreference for the selection task Recall that all participantswere already given a reference price (PMAX) and yet theystill indicated stronger preference for the elicitation taskswhen the entire price range was readily available

The augmented generation task was introduced in this ex-periment to test the propositionthat the hypothesizedeffect ofthe nature of the decision task on consumer preferences is afunction of reference price availabilityIn fact if the presenceofa readilyavailablepricerangeis thefactordrivingconsumerpreference for the selection task then making the price rangereadily available in the augmented generation conditionshould also increase consumer preference for the generationtask Moreover because respondents in both generation andaugmentedgenerationconditionswere offered thesame flexi-bility in expressing their willingness to pay the observed ef-fectscouldnotbedirectlyattributedto thedifferenceinelicita-tionflexibilitybetweenthegenerationand theselectiontasks

The data from the selection and the augmented genera-tion tasks offer further insights into the nature of price ar-ticulation Because the price ranges and mid-range valuesin the selection and the augmented generation tasks wereidentical the only difference between these tasks was theelicitation flexibilitymdashthe degree of freedom consumershave to select the price that most adequately reflects theirwillingness to pay These tasks however were ranked asmarkedly different in terms of the probability of successwhereby nearly three times as many of the respondents in-dicated their expectation that the selection task would likelylead to a more successful outcome compared to the aug-mented generation task Interesting though when indicatingtheir own preference for a price-elicitation mechanism par-ticipants showed a significantly stronger preference for theaugmented generation task (56 vs 36) These data es-sentially suggest a reversal of preferences Although con-sumers evaluated the selection task as having the highestlikelihood of resulting in a successful bid when given achoice they preferred the augmented generation task Infact of all respondents who indicated that the selection taskis likely to be the most successful 46 indicated that ifgiven a choice they would prefer the augmented generationtask thus effectively reversing their preferences

One possible explanation for the observed data is thatwhen presented with the choice of a price-generationmethod consumers were not willing to sacrifice the elicita-tion accuracy associated with the augmented generationtask for the simplicity associated with the selection taskyet when presented with a scenario in which the price-gen-eration mode has already been selected they evaluated it ashaving a higher probability of success This finding impliesthat when forecasting their own ability to articulate a suc-

56 CHERNEV

FIGURE 3 Price elicitation task preferences (Experiment 2)

cessful bid consumers essentially have overweighed theirown ability to generate a successful bid This data pattern isconsistent with the research on decision confidence indicat-ing that in many scenarios individuals are likely to displaya tendency toward overconfidence in their decisions(Allwood amp Montgomery 1987 Griffin amp Tversky 1992Kahneman amp Tversky 1973 Mahajan 1992)

Conceptually this article argues that the observedpreference for the selection task is driven to a large degreeby the complex two-stage nature of the decision taskwhich involves (a) eliciting a readily available price rangeand then (b) generating a price reflecting consumersrsquo will-ingness to pay As a result consumers are likely to discountthe preference-articulation flexibility offered by the genera-tion task and prefer the simpler although more restrictiveselection task This prediction was tested by introducing theaugmented generation condition which simplified thedecision task by offering consumers a readily availableprice range Consistent with this proposition the data showthat the advantage of the selection task is eliminated bymaking the price range readily available Participants in theaugmented generation condition were more confident intheir decisions predicted that prices generated through theaugmented generation task would most likely result in asuccessful bid and indicated a higher overall preferencefor this task as well

Note that so far the impact of the availability ofreference price range was tested by giving respondents arange of possible values in the augmented generation taskThis range was set in a way that minimizes its diagnosticvalue specifically by a pretest asking participants to gener-ate the lowest expected price and then setting the experi-mental range consistent with respondentsrsquo expectationsNevertheless one can argue that providing an external pricerange helped participants validate their initial price rangeexpectations raising consumer confidence in their intuitiveprice range estimates This increased confidence in the va-lidity of their intuitive estimates of the lower end of theprice range could in turn have confounded the resultscontributing to the increase in respondentsrsquo decision confi-dence

To address this possible confound in Experiment 3 theavailability of a reference price range is manipulated byasking participants prior to the choice task to indicate theirexpectation of the lowest possible prices Thus unlike inExperiment 2 where the reference price range was providedby the experimental stimuli in Experiment 3 this range isinternally generated by respondents This manipulation of-fers an alternative strategy to test the theory advanced inthis research by varying the source of the reference pricerange If the theory is valid its predictions should holdwhen the reference price ranges are internally generated aswell The specifics of Experiment 3 and the experimentalresults are presented in more detail in the following section

EXPERIMENT 3

Experiment 3 aims to provide further evidence that the per-ceived advantage of the selection over the generation task iscaused by the complexity of the decision process associatedwith the absence of readily available reference prices In thisexperiment participants were asked to generate price rangesbefore articulatingtheirwillingnessto pay This manipulationallows respondents to have reference prices readily availablefor use as benchmarks in the price-elicitation process Thusthe goal is to show that a simple restructuring of the decisiontask in a generation-ready format can affect consumer prefer-ences for the price-articulationstrategy even in the absence ofexternally provided reference price information

Method

Participants were asked to imagine that they must fly to SanFrancisco for a high-tech job fair and were asked to book aircar and hotel through an online reverse pricing travel agencyOverall the initial scenario was similar to the one used in Ex-periment 2 except that (a) Experiment 3 respondents had tomake decisions for themselves (rather than in the context of aprojective task) and (b) participants had to make three deci-sions in each of three product categories (air car hotel)

Prior to stating their price some of the respondents wereasked to indicate the lowest possible price available for book-ing in each of the three categories For example participantswere told ldquoThe regular price for an airline ticket to San Fran-cisco is $459 What do you think is the lowest price that atravel agency could possibly negotiate with an airline for aflight to San Francisco at any time (regularly priced at$459)rdquo The purpose of this manipulation was for respon-dents to articulate a reference point that would be readilyavailable for the price-generation task This was thepre-choice articulation condition In contrast participants inthe control condition were informed of the regular price butwere not asked to indicate their expectation of the lowestavailable price Thus the only difference between the twoconditionswas that respondents in the pre-choice articulationcondition were asked to indicate their expectationof the low-est available price in each of the three product categoriesNote that participants in the pre-choice articulation conditiondid not receive any additional information compared to therespondents in the control condition

The experimental design was 2 (reference price availabil-ity pre-choice articulation vs control) times 3 (price elicitationtask generation vs selection vs augmented generation)Thepre-choice articulation manipulationwas conductedbetweenparticipantsrsquo and the price-elicitation task was conductedwithin-subjects Each participant had to elicit three pricesone in each product category (air car and hotel) To avoidpossible confounds between the product categories used in

PRICE ELICITATION STRATEGIES IN CHOICE 57

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

ited through the generation task (M = 506 vs M = 327) F(130 ) = 4301 p lt 001 They also predicted a higher successprobability in conditions where a reference price was presentcompared to conditions where a reference price was notreadily available (M = 461 vs M = 372) F(1 30) = 423 p lt05

The data from this experiment suggest that consumers feelmore confident in the outcome of a selection rather than ageneration task This effect was also significant althoughless pronounced in the presence of a readily available refer-ence price Furthermore consumers perceived the selectiontask to have a higher likelihood of success compared to thegeneration task an effect more pronounced when a referenceprice was not readily available

Note that in Experiment 1 the potential effect of availabil-ity of a reference price range is confounded with the effectsassociated with the nature of the elicitation task Indeed con-sumer preference for a selection price mode can be the resultof an attempt to simplify the pricing decision by selecting apricing strategy with fewer possible alternatives Thus con-sistent with the effortndashaccuracy framework (Payne et al1993) one can argue that consumers will prefer the selectiontask because it offers a simpler (although less accurate) deci-sion and not necessarily because in this case a reference pricepoint is readily available This potential confound is ad-dressed in Experiment 2 by introducingan experimental con-dition (referred to as augmented generation) designed to dis-entangle the effects of reference price range availability anddecision effort associated with generating a precise pricepoint

EXPERIMENT 2

Building on Experiment 1 findings Experiment 2 examinesthe impact of the price-elicitation task on the strength of con-sumer preferences and how the availability of a referenceprice range moderates this impact This experiment investi-gates three scenarios price generation price selection and athird scenario referred to as augmented generation Partici-pants in the generation scenario were asked to name theirown price and were not provided with a specific price rangeRespondents in the selection condition were given a set of 10alternative prices to choose from Finally in the augmentedgeneration scenario (discussed in more detail later) partici-pants were asked to name their own price and were providedwith a price range

Method

Respondents were presented with the following scenario

Imagine that three of your classmates had to fly to SanFrancisco for a high-tech job fair They all decided tobuy their air tickets through a reverse pricing Web

travel agency that allows travelers to name their ownprice

Reverse pricing works as follows You tell the travelagency where and when you want to go and how muchyou are willing to pay The agency then searches for anairline willing to release seats at your price If theagency finds tickets at your price it will immediatelypurchase those tickets for you Because you get toname your own price tickets purchased through re-verse pricing Internet agencies cannot be changedtransferred or cancelled

The travel agency will make the best effort to makethe booking It will submit your and other customersrsquorequests to its airline partners who will then decide onthe lowest price to accept Everyone whose bid ishigher or equal to that price or higher will be granted aticket at the stated price Everyone else will have tomake their bookingsdirectly and pay the regular price

Next participants were told that Students A B and C are us-ing different reverse pricing travel agencies that employ dif-ferent procedures to allow customers to name their priceStudent A was asked to name the price he or she is willing topay (generation task) Student B was given a list of 10 pricesand asked to choose among them (selection task) and Stu-dent C was shown a scale that depicts the typical price rangeand asked to indicate the price he or she is willing to pay bypositioninga slider on the scale (augmented generation task)These price-generation tasks are presented in more detail inthe Appendix

Note that respondents in both generation and augmentedgenerationconditionswere offered agreaterdegreeof flexibil-ity in elicitingtheirwillingnessto paywhereas participantsinthe selection condition were limited to a list of 10 availableprices In addition respondents in the selection and the aug-mented generationconditionswere implicitlyprovidedwith aprice range (determined by the highest and the lowest pricesavailablefor selection)whereasparticipantsin thegenerationscenario were notgivena specificprice rangeAll respondentswere also informed that the regularairfare is$459mdasha manipu-lationdesignedtoprovidea more stringenttestof theobservedeffects (recall that the data from Experiment1 indicatethat theavailability of a reference price point is likely to mitigate theobserved reference price range effects)

The range of prices in the selection and the augmentedgeneration conditions was identical and was designed to beof negligible diagnostic value to consumers For that pur-pose a pretest was conducted in which 24 participants fromthe same population were presented with the experimentalscenario and were asked to indicate the lowest price at whichthey would expect to find tickets for the specified route Theaverage of the lowest available prices suggested by the pre-test sample was $184 Based on these data the lower end ofthe range of available prices in the selection and augmentedgeneration tasks was set at $189mdashin line with respondentsrsquo

54 CHERNEV

expectations The upper end of the price range was set toequal the regular airfare ($459) which was provided to par-ticipants in all experimental conditions

When presented with the experimental scenario respon-dents were asked to evaluate the relative difficulty of stu-dentsrsquo decisions by positioning Students A B and C alonga 150-mm line with endpoints ldquoVery easyrdquo and ldquoVery diffi-cultrdquo This procedure yielded two measures of decision dif-ficulty a ranking of the relative difficulty of the pricing de-cisions based on their ordering on the scale and a moreprecise metric measure of the decision difficulty obtainedby measuring the absolute distance from the left end of thescale (Very easy) to the marks corresponding to the per-ceived decision difficulty of each of the three pricing sce-narios

Next participants were asked to evaluate each studentrsquosconfidence in their decision as well as their perception of theprobability of each student getting a ticket at their statedprice Participant responses were measured on the same typeof scale as the decision difficulty scale described previously(endpointsNot confident at all and Very confident for the de-cision-confidencescale and Low probabilityand High prob-ability for the probability of a successful bid) Finally re-spondents were asked to indicate their preference for each ofthese pricing methods (ldquoIf you have to use one of these agen-cies which one would you userdquo)

Participants were tested in groups following their regularclass meetings They worked at their own pace and were de-briefed on completion of the experiment As an incentive fortheir participation several drawings were conducted for cashprizes of $50

Results

The data show a preference for price-elicitation tasks thatoffer a readily available price range The generation taskwas associated with the lowest confidence Sixty-eight per-cent of the responses pointed to the generation task as in-stilling the least confidence compared to 22 for the selec-tion task and 10 for the augmented generation task c2(2)= 552 p lt 001 The difference between the generation andboth the selection and augmented generation was signifi-cant at the 001 level c2(1) = 272 c2(1) = 546 and thedifference between the selection and the augmented genera-tion tasks was marginally significant c2(1) = 275 p lt 10The confidence evaluations revealed through the metricmeasure of scale distances were consistent with the rank-ings data The generation task was associated with the low-est confidence (M = 416 on a 100-point scale) and the se-lection and the augmented generation tasks were associatedwith higher confidence (M = 561 and M = 567 respec-tively) The difference between the generation and the se-lection task was significant F(1 48) = 873 p lt 01 as wasthe difference between the generation and the augmentedgeneration task F(1 48) = 1125 p lt 005 The difference

between the selection and the augmented generation taskwas non-significant F(1 48) lt 1

Decision-difficulty data displayed a pattern similar tothe decision confidence evaluations whereby higher levelsof confidence were associated with easier decisions Spe-cifically 62 of the respondents found the generation taskto be the most difficult 24 considered the selection taskto be the most difficult and only 6 viewed the augmentedgeneration as the most difficult c2(2) = 6148 p lt 001The difference between the generation and the selectiontask was significant c2(1) = 1902 p lt 001 as were thedifferences between the generation versus the augmentedgeneration c2(1) = 6071 p lt 001 and the augmented gen-eration versus the selection tasks c2(1) = 694 p lt 01Metric measures of decision difficulty also show a signifi-cant decrease in the perceived decision difficulty in condi-tions where reference prices were present Once again thegeneration task was perceived the most difficult (M = 603)compared to the selection and the augmented generationtasks (M = 431 and M = 411 respectively) The differencebetween the generation and the selection task was signifi-cant F(1 44) = 1621 p lt 001 as was the difference be-tween the generation and the augmented generation taskF(1 44) = 1916 p lt 001 The difference between the se-lection and the augmented generation tasks wasnonsignificant F(1 44) lt 1

With respect to evaluating the success probability of theirbid participantsranked the generation task as having the low-est probability of yielding a successful outcome (Figure 2)The selection task was perceived as most likely to yield a suc-cessfulbid(70 of the responses)followedby theaugmentedgeneration task (26) Only 4 of the respondentsperceivedthegenerationtask tohavethehighestprobabilityofyieldingasuccessful outcome The difference in rankings between theselection and the augmented generation task (70 vs 26)

PRICE ELICITATION STRATEGIES IN CHOICE 55

FIGURE 2 Predicted success of the different price elicitation tasks(Experiment 2)

was significant c2(1) = 2406 p lt 001 as was the differencebetween the generation and the augmented generation taskc2(1) = 1049p lt 005The metric distancemeasures revealedsimilar distribution of participantsrsquoevaluations of the proba-bilityof a successfuloutcomeThe average rating of the likeli-hood of a successful bid through a generation task was M =337 compared to M = 642 for the selection task F(1 48) =8122 p lt 001 and M = 578 for the augmented generationtask F(1 48)= 4823p lt 001The differencebetween the se-lection and the augmented generation task was significant aswell F(1 48) = 427 p lt 05

In addition to the projective measures of generation par-ticipants were asked to indicate their own preference for theprice-elicitation task The majority of the respondents (56)chose the augmented generation task 36 indicated a pref-erence for the selection task and only 16 indicated thatthey would prefer the generation task (Figure 3) The differ-ence between the augmented generation and the selectiontask was significant c2(1) = 419 p lt 05 and the differencebetween the selection and the augmented generationwas alsosignificant c2(1) = 548 p lt 05

DISCUSSION

Experiments 1 and 2 document that the impact of the natureof the price-articulation task is moderated by the referenceprice availability In Experiment 1 participants were pre-sented with an external reference price representing thehigher end of the price range (PMAX) and in Experiment 2respondents were given the lower end of the price range(PMIN) as well In both experiments providing participantswith reference price information had a significant effect ontheir decision process Specifically respondentsrsquo responsesindicated higher levels of confidence associated with the se-

lection compared to the generation task The selection taskwas also rated higher in terms of its predicted likelihood ofsuccess this task was preferred by consumers as wellThese data also demonstrate the robustness of the observedpreference for the selection task Recall that all participantswere already given a reference price (PMAX) and yet theystill indicated stronger preference for the elicitation taskswhen the entire price range was readily available

The augmented generation task was introduced in this ex-periment to test the propositionthat the hypothesizedeffect ofthe nature of the decision task on consumer preferences is afunction of reference price availabilityIn fact if the presenceofa readilyavailablepricerangeis thefactordrivingconsumerpreference for the selection task then making the price rangereadily available in the augmented generation conditionshould also increase consumer preference for the generationtask Moreover because respondents in both generation andaugmentedgenerationconditionswere offered thesame flexi-bility in expressing their willingness to pay the observed ef-fectscouldnotbedirectlyattributedto thedifferenceinelicita-tionflexibilitybetweenthegenerationand theselectiontasks

The data from the selection and the augmented genera-tion tasks offer further insights into the nature of price ar-ticulation Because the price ranges and mid-range valuesin the selection and the augmented generation tasks wereidentical the only difference between these tasks was theelicitation flexibilitymdashthe degree of freedom consumershave to select the price that most adequately reflects theirwillingness to pay These tasks however were ranked asmarkedly different in terms of the probability of successwhereby nearly three times as many of the respondents in-dicated their expectation that the selection task would likelylead to a more successful outcome compared to the aug-mented generation task Interesting though when indicatingtheir own preference for a price-elicitation mechanism par-ticipants showed a significantly stronger preference for theaugmented generation task (56 vs 36) These data es-sentially suggest a reversal of preferences Although con-sumers evaluated the selection task as having the highestlikelihood of resulting in a successful bid when given achoice they preferred the augmented generation task Infact of all respondents who indicated that the selection taskis likely to be the most successful 46 indicated that ifgiven a choice they would prefer the augmented generationtask thus effectively reversing their preferences

One possible explanation for the observed data is thatwhen presented with the choice of a price-generationmethod consumers were not willing to sacrifice the elicita-tion accuracy associated with the augmented generationtask for the simplicity associated with the selection taskyet when presented with a scenario in which the price-gen-eration mode has already been selected they evaluated it ashaving a higher probability of success This finding impliesthat when forecasting their own ability to articulate a suc-

56 CHERNEV

FIGURE 3 Price elicitation task preferences (Experiment 2)

cessful bid consumers essentially have overweighed theirown ability to generate a successful bid This data pattern isconsistent with the research on decision confidence indicat-ing that in many scenarios individuals are likely to displaya tendency toward overconfidence in their decisions(Allwood amp Montgomery 1987 Griffin amp Tversky 1992Kahneman amp Tversky 1973 Mahajan 1992)

Conceptually this article argues that the observedpreference for the selection task is driven to a large degreeby the complex two-stage nature of the decision taskwhich involves (a) eliciting a readily available price rangeand then (b) generating a price reflecting consumersrsquo will-ingness to pay As a result consumers are likely to discountthe preference-articulation flexibility offered by the genera-tion task and prefer the simpler although more restrictiveselection task This prediction was tested by introducing theaugmented generation condition which simplified thedecision task by offering consumers a readily availableprice range Consistent with this proposition the data showthat the advantage of the selection task is eliminated bymaking the price range readily available Participants in theaugmented generation condition were more confident intheir decisions predicted that prices generated through theaugmented generation task would most likely result in asuccessful bid and indicated a higher overall preferencefor this task as well

Note that so far the impact of the availability ofreference price range was tested by giving respondents arange of possible values in the augmented generation taskThis range was set in a way that minimizes its diagnosticvalue specifically by a pretest asking participants to gener-ate the lowest expected price and then setting the experi-mental range consistent with respondentsrsquo expectationsNevertheless one can argue that providing an external pricerange helped participants validate their initial price rangeexpectations raising consumer confidence in their intuitiveprice range estimates This increased confidence in the va-lidity of their intuitive estimates of the lower end of theprice range could in turn have confounded the resultscontributing to the increase in respondentsrsquo decision confi-dence

To address this possible confound in Experiment 3 theavailability of a reference price range is manipulated byasking participants prior to the choice task to indicate theirexpectation of the lowest possible prices Thus unlike inExperiment 2 where the reference price range was providedby the experimental stimuli in Experiment 3 this range isinternally generated by respondents This manipulation of-fers an alternative strategy to test the theory advanced inthis research by varying the source of the reference pricerange If the theory is valid its predictions should holdwhen the reference price ranges are internally generated aswell The specifics of Experiment 3 and the experimentalresults are presented in more detail in the following section

EXPERIMENT 3

Experiment 3 aims to provide further evidence that the per-ceived advantage of the selection over the generation task iscaused by the complexity of the decision process associatedwith the absence of readily available reference prices In thisexperiment participants were asked to generate price rangesbefore articulatingtheirwillingnessto pay This manipulationallows respondents to have reference prices readily availablefor use as benchmarks in the price-elicitation process Thusthe goal is to show that a simple restructuring of the decisiontask in a generation-ready format can affect consumer prefer-ences for the price-articulationstrategy even in the absence ofexternally provided reference price information

Method

Participants were asked to imagine that they must fly to SanFrancisco for a high-tech job fair and were asked to book aircar and hotel through an online reverse pricing travel agencyOverall the initial scenario was similar to the one used in Ex-periment 2 except that (a) Experiment 3 respondents had tomake decisions for themselves (rather than in the context of aprojective task) and (b) participants had to make three deci-sions in each of three product categories (air car hotel)

Prior to stating their price some of the respondents wereasked to indicate the lowest possible price available for book-ing in each of the three categories For example participantswere told ldquoThe regular price for an airline ticket to San Fran-cisco is $459 What do you think is the lowest price that atravel agency could possibly negotiate with an airline for aflight to San Francisco at any time (regularly priced at$459)rdquo The purpose of this manipulation was for respon-dents to articulate a reference point that would be readilyavailable for the price-generation task This was thepre-choice articulation condition In contrast participants inthe control condition were informed of the regular price butwere not asked to indicate their expectation of the lowestavailable price Thus the only difference between the twoconditionswas that respondents in the pre-choice articulationcondition were asked to indicate their expectationof the low-est available price in each of the three product categoriesNote that participants in the pre-choice articulation conditiondid not receive any additional information compared to therespondents in the control condition

The experimental design was 2 (reference price availabil-ity pre-choice articulation vs control) times 3 (price elicitationtask generation vs selection vs augmented generation)Thepre-choice articulation manipulationwas conductedbetweenparticipantsrsquo and the price-elicitation task was conductedwithin-subjects Each participant had to elicit three pricesone in each product category (air car and hotel) To avoidpossible confounds between the product categories used in

PRICE ELICITATION STRATEGIES IN CHOICE 57

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

expectations The upper end of the price range was set toequal the regular airfare ($459) which was provided to par-ticipants in all experimental conditions

When presented with the experimental scenario respon-dents were asked to evaluate the relative difficulty of stu-dentsrsquo decisions by positioning Students A B and C alonga 150-mm line with endpoints ldquoVery easyrdquo and ldquoVery diffi-cultrdquo This procedure yielded two measures of decision dif-ficulty a ranking of the relative difficulty of the pricing de-cisions based on their ordering on the scale and a moreprecise metric measure of the decision difficulty obtainedby measuring the absolute distance from the left end of thescale (Very easy) to the marks corresponding to the per-ceived decision difficulty of each of the three pricing sce-narios

Next participants were asked to evaluate each studentrsquosconfidence in their decision as well as their perception of theprobability of each student getting a ticket at their statedprice Participant responses were measured on the same typeof scale as the decision difficulty scale described previously(endpointsNot confident at all and Very confident for the de-cision-confidencescale and Low probabilityand High prob-ability for the probability of a successful bid) Finally re-spondents were asked to indicate their preference for each ofthese pricing methods (ldquoIf you have to use one of these agen-cies which one would you userdquo)

Participants were tested in groups following their regularclass meetings They worked at their own pace and were de-briefed on completion of the experiment As an incentive fortheir participation several drawings were conducted for cashprizes of $50

Results

The data show a preference for price-elicitation tasks thatoffer a readily available price range The generation taskwas associated with the lowest confidence Sixty-eight per-cent of the responses pointed to the generation task as in-stilling the least confidence compared to 22 for the selec-tion task and 10 for the augmented generation task c2(2)= 552 p lt 001 The difference between the generation andboth the selection and augmented generation was signifi-cant at the 001 level c2(1) = 272 c2(1) = 546 and thedifference between the selection and the augmented genera-tion tasks was marginally significant c2(1) = 275 p lt 10The confidence evaluations revealed through the metricmeasure of scale distances were consistent with the rank-ings data The generation task was associated with the low-est confidence (M = 416 on a 100-point scale) and the se-lection and the augmented generation tasks were associatedwith higher confidence (M = 561 and M = 567 respec-tively) The difference between the generation and the se-lection task was significant F(1 48) = 873 p lt 01 as wasthe difference between the generation and the augmentedgeneration task F(1 48) = 1125 p lt 005 The difference

between the selection and the augmented generation taskwas non-significant F(1 48) lt 1

Decision-difficulty data displayed a pattern similar tothe decision confidence evaluations whereby higher levelsof confidence were associated with easier decisions Spe-cifically 62 of the respondents found the generation taskto be the most difficult 24 considered the selection taskto be the most difficult and only 6 viewed the augmentedgeneration as the most difficult c2(2) = 6148 p lt 001The difference between the generation and the selectiontask was significant c2(1) = 1902 p lt 001 as were thedifferences between the generation versus the augmentedgeneration c2(1) = 6071 p lt 001 and the augmented gen-eration versus the selection tasks c2(1) = 694 p lt 01Metric measures of decision difficulty also show a signifi-cant decrease in the perceived decision difficulty in condi-tions where reference prices were present Once again thegeneration task was perceived the most difficult (M = 603)compared to the selection and the augmented generationtasks (M = 431 and M = 411 respectively) The differencebetween the generation and the selection task was signifi-cant F(1 44) = 1621 p lt 001 as was the difference be-tween the generation and the augmented generation taskF(1 44) = 1916 p lt 001 The difference between the se-lection and the augmented generation tasks wasnonsignificant F(1 44) lt 1

With respect to evaluating the success probability of theirbid participantsranked the generation task as having the low-est probability of yielding a successful outcome (Figure 2)The selection task was perceived as most likely to yield a suc-cessfulbid(70 of the responses)followedby theaugmentedgeneration task (26) Only 4 of the respondentsperceivedthegenerationtask tohavethehighestprobabilityofyieldingasuccessful outcome The difference in rankings between theselection and the augmented generation task (70 vs 26)

PRICE ELICITATION STRATEGIES IN CHOICE 55

FIGURE 2 Predicted success of the different price elicitation tasks(Experiment 2)

was significant c2(1) = 2406 p lt 001 as was the differencebetween the generation and the augmented generation taskc2(1) = 1049p lt 005The metric distancemeasures revealedsimilar distribution of participantsrsquoevaluations of the proba-bilityof a successfuloutcomeThe average rating of the likeli-hood of a successful bid through a generation task was M =337 compared to M = 642 for the selection task F(1 48) =8122 p lt 001 and M = 578 for the augmented generationtask F(1 48)= 4823p lt 001The differencebetween the se-lection and the augmented generation task was significant aswell F(1 48) = 427 p lt 05

In addition to the projective measures of generation par-ticipants were asked to indicate their own preference for theprice-elicitation task The majority of the respondents (56)chose the augmented generation task 36 indicated a pref-erence for the selection task and only 16 indicated thatthey would prefer the generation task (Figure 3) The differ-ence between the augmented generation and the selectiontask was significant c2(1) = 419 p lt 05 and the differencebetween the selection and the augmented generationwas alsosignificant c2(1) = 548 p lt 05

DISCUSSION

Experiments 1 and 2 document that the impact of the natureof the price-articulation task is moderated by the referenceprice availability In Experiment 1 participants were pre-sented with an external reference price representing thehigher end of the price range (PMAX) and in Experiment 2respondents were given the lower end of the price range(PMIN) as well In both experiments providing participantswith reference price information had a significant effect ontheir decision process Specifically respondentsrsquo responsesindicated higher levels of confidence associated with the se-

lection compared to the generation task The selection taskwas also rated higher in terms of its predicted likelihood ofsuccess this task was preferred by consumers as wellThese data also demonstrate the robustness of the observedpreference for the selection task Recall that all participantswere already given a reference price (PMAX) and yet theystill indicated stronger preference for the elicitation taskswhen the entire price range was readily available

The augmented generation task was introduced in this ex-periment to test the propositionthat the hypothesizedeffect ofthe nature of the decision task on consumer preferences is afunction of reference price availabilityIn fact if the presenceofa readilyavailablepricerangeis thefactordrivingconsumerpreference for the selection task then making the price rangereadily available in the augmented generation conditionshould also increase consumer preference for the generationtask Moreover because respondents in both generation andaugmentedgenerationconditionswere offered thesame flexi-bility in expressing their willingness to pay the observed ef-fectscouldnotbedirectlyattributedto thedifferenceinelicita-tionflexibilitybetweenthegenerationand theselectiontasks

The data from the selection and the augmented genera-tion tasks offer further insights into the nature of price ar-ticulation Because the price ranges and mid-range valuesin the selection and the augmented generation tasks wereidentical the only difference between these tasks was theelicitation flexibilitymdashthe degree of freedom consumershave to select the price that most adequately reflects theirwillingness to pay These tasks however were ranked asmarkedly different in terms of the probability of successwhereby nearly three times as many of the respondents in-dicated their expectation that the selection task would likelylead to a more successful outcome compared to the aug-mented generation task Interesting though when indicatingtheir own preference for a price-elicitation mechanism par-ticipants showed a significantly stronger preference for theaugmented generation task (56 vs 36) These data es-sentially suggest a reversal of preferences Although con-sumers evaluated the selection task as having the highestlikelihood of resulting in a successful bid when given achoice they preferred the augmented generation task Infact of all respondents who indicated that the selection taskis likely to be the most successful 46 indicated that ifgiven a choice they would prefer the augmented generationtask thus effectively reversing their preferences

One possible explanation for the observed data is thatwhen presented with the choice of a price-generationmethod consumers were not willing to sacrifice the elicita-tion accuracy associated with the augmented generationtask for the simplicity associated with the selection taskyet when presented with a scenario in which the price-gen-eration mode has already been selected they evaluated it ashaving a higher probability of success This finding impliesthat when forecasting their own ability to articulate a suc-

56 CHERNEV

FIGURE 3 Price elicitation task preferences (Experiment 2)

cessful bid consumers essentially have overweighed theirown ability to generate a successful bid This data pattern isconsistent with the research on decision confidence indicat-ing that in many scenarios individuals are likely to displaya tendency toward overconfidence in their decisions(Allwood amp Montgomery 1987 Griffin amp Tversky 1992Kahneman amp Tversky 1973 Mahajan 1992)

Conceptually this article argues that the observedpreference for the selection task is driven to a large degreeby the complex two-stage nature of the decision taskwhich involves (a) eliciting a readily available price rangeand then (b) generating a price reflecting consumersrsquo will-ingness to pay As a result consumers are likely to discountthe preference-articulation flexibility offered by the genera-tion task and prefer the simpler although more restrictiveselection task This prediction was tested by introducing theaugmented generation condition which simplified thedecision task by offering consumers a readily availableprice range Consistent with this proposition the data showthat the advantage of the selection task is eliminated bymaking the price range readily available Participants in theaugmented generation condition were more confident intheir decisions predicted that prices generated through theaugmented generation task would most likely result in asuccessful bid and indicated a higher overall preferencefor this task as well

Note that so far the impact of the availability ofreference price range was tested by giving respondents arange of possible values in the augmented generation taskThis range was set in a way that minimizes its diagnosticvalue specifically by a pretest asking participants to gener-ate the lowest expected price and then setting the experi-mental range consistent with respondentsrsquo expectationsNevertheless one can argue that providing an external pricerange helped participants validate their initial price rangeexpectations raising consumer confidence in their intuitiveprice range estimates This increased confidence in the va-lidity of their intuitive estimates of the lower end of theprice range could in turn have confounded the resultscontributing to the increase in respondentsrsquo decision confi-dence

To address this possible confound in Experiment 3 theavailability of a reference price range is manipulated byasking participants prior to the choice task to indicate theirexpectation of the lowest possible prices Thus unlike inExperiment 2 where the reference price range was providedby the experimental stimuli in Experiment 3 this range isinternally generated by respondents This manipulation of-fers an alternative strategy to test the theory advanced inthis research by varying the source of the reference pricerange If the theory is valid its predictions should holdwhen the reference price ranges are internally generated aswell The specifics of Experiment 3 and the experimentalresults are presented in more detail in the following section

EXPERIMENT 3

Experiment 3 aims to provide further evidence that the per-ceived advantage of the selection over the generation task iscaused by the complexity of the decision process associatedwith the absence of readily available reference prices In thisexperiment participants were asked to generate price rangesbefore articulatingtheirwillingnessto pay This manipulationallows respondents to have reference prices readily availablefor use as benchmarks in the price-elicitation process Thusthe goal is to show that a simple restructuring of the decisiontask in a generation-ready format can affect consumer prefer-ences for the price-articulationstrategy even in the absence ofexternally provided reference price information

Method

Participants were asked to imagine that they must fly to SanFrancisco for a high-tech job fair and were asked to book aircar and hotel through an online reverse pricing travel agencyOverall the initial scenario was similar to the one used in Ex-periment 2 except that (a) Experiment 3 respondents had tomake decisions for themselves (rather than in the context of aprojective task) and (b) participants had to make three deci-sions in each of three product categories (air car hotel)

Prior to stating their price some of the respondents wereasked to indicate the lowest possible price available for book-ing in each of the three categories For example participantswere told ldquoThe regular price for an airline ticket to San Fran-cisco is $459 What do you think is the lowest price that atravel agency could possibly negotiate with an airline for aflight to San Francisco at any time (regularly priced at$459)rdquo The purpose of this manipulation was for respon-dents to articulate a reference point that would be readilyavailable for the price-generation task This was thepre-choice articulation condition In contrast participants inthe control condition were informed of the regular price butwere not asked to indicate their expectation of the lowestavailable price Thus the only difference between the twoconditionswas that respondents in the pre-choice articulationcondition were asked to indicate their expectationof the low-est available price in each of the three product categoriesNote that participants in the pre-choice articulation conditiondid not receive any additional information compared to therespondents in the control condition

The experimental design was 2 (reference price availabil-ity pre-choice articulation vs control) times 3 (price elicitationtask generation vs selection vs augmented generation)Thepre-choice articulation manipulationwas conductedbetweenparticipantsrsquo and the price-elicitation task was conductedwithin-subjects Each participant had to elicit three pricesone in each product category (air car and hotel) To avoidpossible confounds between the product categories used in

PRICE ELICITATION STRATEGIES IN CHOICE 57

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

was significant c2(1) = 2406 p lt 001 as was the differencebetween the generation and the augmented generation taskc2(1) = 1049p lt 005The metric distancemeasures revealedsimilar distribution of participantsrsquoevaluations of the proba-bilityof a successfuloutcomeThe average rating of the likeli-hood of a successful bid through a generation task was M =337 compared to M = 642 for the selection task F(1 48) =8122 p lt 001 and M = 578 for the augmented generationtask F(1 48)= 4823p lt 001The differencebetween the se-lection and the augmented generation task was significant aswell F(1 48) = 427 p lt 05

In addition to the projective measures of generation par-ticipants were asked to indicate their own preference for theprice-elicitation task The majority of the respondents (56)chose the augmented generation task 36 indicated a pref-erence for the selection task and only 16 indicated thatthey would prefer the generation task (Figure 3) The differ-ence between the augmented generation and the selectiontask was significant c2(1) = 419 p lt 05 and the differencebetween the selection and the augmented generationwas alsosignificant c2(1) = 548 p lt 05

DISCUSSION

Experiments 1 and 2 document that the impact of the natureof the price-articulation task is moderated by the referenceprice availability In Experiment 1 participants were pre-sented with an external reference price representing thehigher end of the price range (PMAX) and in Experiment 2respondents were given the lower end of the price range(PMIN) as well In both experiments providing participantswith reference price information had a significant effect ontheir decision process Specifically respondentsrsquo responsesindicated higher levels of confidence associated with the se-

lection compared to the generation task The selection taskwas also rated higher in terms of its predicted likelihood ofsuccess this task was preferred by consumers as wellThese data also demonstrate the robustness of the observedpreference for the selection task Recall that all participantswere already given a reference price (PMAX) and yet theystill indicated stronger preference for the elicitation taskswhen the entire price range was readily available

The augmented generation task was introduced in this ex-periment to test the propositionthat the hypothesizedeffect ofthe nature of the decision task on consumer preferences is afunction of reference price availabilityIn fact if the presenceofa readilyavailablepricerangeis thefactordrivingconsumerpreference for the selection task then making the price rangereadily available in the augmented generation conditionshould also increase consumer preference for the generationtask Moreover because respondents in both generation andaugmentedgenerationconditionswere offered thesame flexi-bility in expressing their willingness to pay the observed ef-fectscouldnotbedirectlyattributedto thedifferenceinelicita-tionflexibilitybetweenthegenerationand theselectiontasks

The data from the selection and the augmented genera-tion tasks offer further insights into the nature of price ar-ticulation Because the price ranges and mid-range valuesin the selection and the augmented generation tasks wereidentical the only difference between these tasks was theelicitation flexibilitymdashthe degree of freedom consumershave to select the price that most adequately reflects theirwillingness to pay These tasks however were ranked asmarkedly different in terms of the probability of successwhereby nearly three times as many of the respondents in-dicated their expectation that the selection task would likelylead to a more successful outcome compared to the aug-mented generation task Interesting though when indicatingtheir own preference for a price-elicitation mechanism par-ticipants showed a significantly stronger preference for theaugmented generation task (56 vs 36) These data es-sentially suggest a reversal of preferences Although con-sumers evaluated the selection task as having the highestlikelihood of resulting in a successful bid when given achoice they preferred the augmented generation task Infact of all respondents who indicated that the selection taskis likely to be the most successful 46 indicated that ifgiven a choice they would prefer the augmented generationtask thus effectively reversing their preferences

One possible explanation for the observed data is thatwhen presented with the choice of a price-generationmethod consumers were not willing to sacrifice the elicita-tion accuracy associated with the augmented generationtask for the simplicity associated with the selection taskyet when presented with a scenario in which the price-gen-eration mode has already been selected they evaluated it ashaving a higher probability of success This finding impliesthat when forecasting their own ability to articulate a suc-

56 CHERNEV

FIGURE 3 Price elicitation task preferences (Experiment 2)

cessful bid consumers essentially have overweighed theirown ability to generate a successful bid This data pattern isconsistent with the research on decision confidence indicat-ing that in many scenarios individuals are likely to displaya tendency toward overconfidence in their decisions(Allwood amp Montgomery 1987 Griffin amp Tversky 1992Kahneman amp Tversky 1973 Mahajan 1992)

Conceptually this article argues that the observedpreference for the selection task is driven to a large degreeby the complex two-stage nature of the decision taskwhich involves (a) eliciting a readily available price rangeand then (b) generating a price reflecting consumersrsquo will-ingness to pay As a result consumers are likely to discountthe preference-articulation flexibility offered by the genera-tion task and prefer the simpler although more restrictiveselection task This prediction was tested by introducing theaugmented generation condition which simplified thedecision task by offering consumers a readily availableprice range Consistent with this proposition the data showthat the advantage of the selection task is eliminated bymaking the price range readily available Participants in theaugmented generation condition were more confident intheir decisions predicted that prices generated through theaugmented generation task would most likely result in asuccessful bid and indicated a higher overall preferencefor this task as well

Note that so far the impact of the availability ofreference price range was tested by giving respondents arange of possible values in the augmented generation taskThis range was set in a way that minimizes its diagnosticvalue specifically by a pretest asking participants to gener-ate the lowest expected price and then setting the experi-mental range consistent with respondentsrsquo expectationsNevertheless one can argue that providing an external pricerange helped participants validate their initial price rangeexpectations raising consumer confidence in their intuitiveprice range estimates This increased confidence in the va-lidity of their intuitive estimates of the lower end of theprice range could in turn have confounded the resultscontributing to the increase in respondentsrsquo decision confi-dence

To address this possible confound in Experiment 3 theavailability of a reference price range is manipulated byasking participants prior to the choice task to indicate theirexpectation of the lowest possible prices Thus unlike inExperiment 2 where the reference price range was providedby the experimental stimuli in Experiment 3 this range isinternally generated by respondents This manipulation of-fers an alternative strategy to test the theory advanced inthis research by varying the source of the reference pricerange If the theory is valid its predictions should holdwhen the reference price ranges are internally generated aswell The specifics of Experiment 3 and the experimentalresults are presented in more detail in the following section

EXPERIMENT 3

Experiment 3 aims to provide further evidence that the per-ceived advantage of the selection over the generation task iscaused by the complexity of the decision process associatedwith the absence of readily available reference prices In thisexperiment participants were asked to generate price rangesbefore articulatingtheirwillingnessto pay This manipulationallows respondents to have reference prices readily availablefor use as benchmarks in the price-elicitation process Thusthe goal is to show that a simple restructuring of the decisiontask in a generation-ready format can affect consumer prefer-ences for the price-articulationstrategy even in the absence ofexternally provided reference price information

Method

Participants were asked to imagine that they must fly to SanFrancisco for a high-tech job fair and were asked to book aircar and hotel through an online reverse pricing travel agencyOverall the initial scenario was similar to the one used in Ex-periment 2 except that (a) Experiment 3 respondents had tomake decisions for themselves (rather than in the context of aprojective task) and (b) participants had to make three deci-sions in each of three product categories (air car hotel)

Prior to stating their price some of the respondents wereasked to indicate the lowest possible price available for book-ing in each of the three categories For example participantswere told ldquoThe regular price for an airline ticket to San Fran-cisco is $459 What do you think is the lowest price that atravel agency could possibly negotiate with an airline for aflight to San Francisco at any time (regularly priced at$459)rdquo The purpose of this manipulation was for respon-dents to articulate a reference point that would be readilyavailable for the price-generation task This was thepre-choice articulation condition In contrast participants inthe control condition were informed of the regular price butwere not asked to indicate their expectation of the lowestavailable price Thus the only difference between the twoconditionswas that respondents in the pre-choice articulationcondition were asked to indicate their expectationof the low-est available price in each of the three product categoriesNote that participants in the pre-choice articulation conditiondid not receive any additional information compared to therespondents in the control condition

The experimental design was 2 (reference price availabil-ity pre-choice articulation vs control) times 3 (price elicitationtask generation vs selection vs augmented generation)Thepre-choice articulation manipulationwas conductedbetweenparticipantsrsquo and the price-elicitation task was conductedwithin-subjects Each participant had to elicit three pricesone in each product category (air car and hotel) To avoidpossible confounds between the product categories used in

PRICE ELICITATION STRATEGIES IN CHOICE 57

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

cessful bid consumers essentially have overweighed theirown ability to generate a successful bid This data pattern isconsistent with the research on decision confidence indicat-ing that in many scenarios individuals are likely to displaya tendency toward overconfidence in their decisions(Allwood amp Montgomery 1987 Griffin amp Tversky 1992Kahneman amp Tversky 1973 Mahajan 1992)

Conceptually this article argues that the observedpreference for the selection task is driven to a large degreeby the complex two-stage nature of the decision taskwhich involves (a) eliciting a readily available price rangeand then (b) generating a price reflecting consumersrsquo will-ingness to pay As a result consumers are likely to discountthe preference-articulation flexibility offered by the genera-tion task and prefer the simpler although more restrictiveselection task This prediction was tested by introducing theaugmented generation condition which simplified thedecision task by offering consumers a readily availableprice range Consistent with this proposition the data showthat the advantage of the selection task is eliminated bymaking the price range readily available Participants in theaugmented generation condition were more confident intheir decisions predicted that prices generated through theaugmented generation task would most likely result in asuccessful bid and indicated a higher overall preferencefor this task as well

Note that so far the impact of the availability ofreference price range was tested by giving respondents arange of possible values in the augmented generation taskThis range was set in a way that minimizes its diagnosticvalue specifically by a pretest asking participants to gener-ate the lowest expected price and then setting the experi-mental range consistent with respondentsrsquo expectationsNevertheless one can argue that providing an external pricerange helped participants validate their initial price rangeexpectations raising consumer confidence in their intuitiveprice range estimates This increased confidence in the va-lidity of their intuitive estimates of the lower end of theprice range could in turn have confounded the resultscontributing to the increase in respondentsrsquo decision confi-dence

To address this possible confound in Experiment 3 theavailability of a reference price range is manipulated byasking participants prior to the choice task to indicate theirexpectation of the lowest possible prices Thus unlike inExperiment 2 where the reference price range was providedby the experimental stimuli in Experiment 3 this range isinternally generated by respondents This manipulation of-fers an alternative strategy to test the theory advanced inthis research by varying the source of the reference pricerange If the theory is valid its predictions should holdwhen the reference price ranges are internally generated aswell The specifics of Experiment 3 and the experimentalresults are presented in more detail in the following section

EXPERIMENT 3

Experiment 3 aims to provide further evidence that the per-ceived advantage of the selection over the generation task iscaused by the complexity of the decision process associatedwith the absence of readily available reference prices In thisexperiment participants were asked to generate price rangesbefore articulatingtheirwillingnessto pay This manipulationallows respondents to have reference prices readily availablefor use as benchmarks in the price-elicitation process Thusthe goal is to show that a simple restructuring of the decisiontask in a generation-ready format can affect consumer prefer-ences for the price-articulationstrategy even in the absence ofexternally provided reference price information

Method

Participants were asked to imagine that they must fly to SanFrancisco for a high-tech job fair and were asked to book aircar and hotel through an online reverse pricing travel agencyOverall the initial scenario was similar to the one used in Ex-periment 2 except that (a) Experiment 3 respondents had tomake decisions for themselves (rather than in the context of aprojective task) and (b) participants had to make three deci-sions in each of three product categories (air car hotel)

Prior to stating their price some of the respondents wereasked to indicate the lowest possible price available for book-ing in each of the three categories For example participantswere told ldquoThe regular price for an airline ticket to San Fran-cisco is $459 What do you think is the lowest price that atravel agency could possibly negotiate with an airline for aflight to San Francisco at any time (regularly priced at$459)rdquo The purpose of this manipulation was for respon-dents to articulate a reference point that would be readilyavailable for the price-generation task This was thepre-choice articulation condition In contrast participants inthe control condition were informed of the regular price butwere not asked to indicate their expectation of the lowestavailable price Thus the only difference between the twoconditionswas that respondents in the pre-choice articulationcondition were asked to indicate their expectationof the low-est available price in each of the three product categoriesNote that participants in the pre-choice articulation conditiondid not receive any additional information compared to therespondents in the control condition

The experimental design was 2 (reference price availabil-ity pre-choice articulation vs control) times 3 (price elicitationtask generation vs selection vs augmented generation)Thepre-choice articulation manipulationwas conductedbetweenparticipantsrsquo and the price-elicitation task was conductedwithin-subjects Each participant had to elicit three pricesone in each product category (air car and hotel) To avoidpossible confounds between the product categories used in

PRICE ELICITATION STRATEGIES IN CHOICE 57

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

the experiment and the price-elicitation tasks their combina-tions were counterbalanced across respondents

Thegenerationandselectionmanipulationswere similar tothose in the first two experiments Participants were asked toeither name theirown price or to select theirprice from a set of10 availableprices The augmented generation condition wassomewhat different Respondentswere givena scale of values(identical to the values of the options in the selection condi-tion) and were asked to (a) mark on the scale the price theywere willingtopay andthen(b) write theexactprice theywerewilling to pay Thus from a purely cognitiveperspectivepar-ticipants in the augmented generation condition had the mosteffortful task because they had to express their willingness topay twiceonce using a scale and once by writingdown the ex-act amount they were willing to pay

After stating their willingness to pay respondents wereasked to indicate their confidence in the decision they justmade their perceived decision difficulty and their perceivedlikelihoodof getting a bookingat the stated price Participantswere tested in groups following their regular class meetingsThey worked at their own pace and were debriefed upon com-pletion of the experiment As an incentive for their participa-tionseveraldrawingsforcash prizesof$50were conducted

Results

This research argues that when a reference price range is notreadily available consumers prefer the simpler but more con-straining selection task to the more flexible yet more complexgeneration task I also proposed that the presence of apre-choice articulation task in which consumers were asked

to elicit a reference pricewould moderate thiseffect The datafrom the generationand selection tasksare consistentwith thetheoretical predictions Respondents were less confident intheirdecisionsandexpectedthemto havea lower likelihoodofsuccess in the context of a generation rather than a selectiontask These data are presented next

The data show that the prechoice price articulation had asignificant impact on how the nature of the price elicitationtask affected the strength of consumer preferences (Figure 4)Participants in the control conditionwere less confidentwhennaming their own price (M = 299) than when choosinga pricein thecontextof a selectiontask (M = 482)F(1 66) = 1758plt 0011 In contrastrespondentswho were askedtoarticulateaprice range prior to the choice task indicated no difference intheir confidence between the generation and selection condi-tions (M = 473 vs M = 505) F(1 66) lt 1 This interactionef-fect was significantF(1 66)= 647p lt 05 indicatingthat theprechoicearticulationwas indeed moderatingthe effect of theprice-elicitation task on consumer preferences

Participants in the control condition also perceived thegeneration task as more difficult than the selection task (M= 471 vs M = 305) F(1 66) = 139 p lt 001 see Figure5 This pattern of results was significantly different F(166) = 101 p lt 005 for respondents in the prechoice articu-lation condition who perceived the selection and the

58 CHERNEV

FIGURE 4 Decision confidence data (Experiment 3)

1Note that although in Experiment 2 participants were asked to makecomparative judgments yielding relative measures in Experiment 3 respon-dents were asked to provide their responses on separate scales For that rea-son in Experiment 3 rank-order data were not readily available and only par-ticipantsrsquo metric scale responses were analyzed

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

generation task to be similarly difficult (M = 471 vs M =305) F(1 66) lt 1

Finally there was a significant difference in participantsrsquopredictionsof the likelihoodofsuccessof the two priceelicita-tion tasks F(1 66) = 712 p lt 01 see Figure 6 Specificallyrespondents in the control condition rated the generation task

as significantly less likely to lead to a successful outcome asthe selection task F(1 66) = 2513 p lt 001 In contrast thecorrespondingdifference for participantsin the pre-choicear-ticulationconditionwas non-significantF(166)=23pgt10

With respect to the augmented-generation task it was pre-dicted that similar to the selection task respondents in the

PRICE ELICITATION STRATEGIES IN CHOICE 59

FIGURE 5 Decision difficulty data (Experiment 3)

FIGURE 6 Predicted success likelihood data (Experiment 3)

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

augmented generation task would not be affected by thepre-choice articulation manipulation because an externalprice range was readily availableto participantsin both condi-tions Yet the data displayed a pattern that was significantlydifferent from the pattern displayed in the selection task Spe-cifically respondents in the control condition reported beingequally confident in their decisions in the selection and theaugmented generation tasks (M = 46 vs M = 482) F(1 66) lt1 They perceived these tasks to be equally difficult (M = 285vs M = 305) F(1 66) lt 1 and the augmented generationtaskto have a lower success probability compared to the selectiontask (M = 467 vs M = 557) F(1 66) = 435p lt 05The asso-ciated interactionswere significantas follows F(266)= 401p lt 05 for the confidencedataF(2 66) = 587p lt 001 for thesuccess likelihooddata and F(2 66) = 556p lt 01 for the de-cision difficulty data

In contrast participants in the prechoice articulation con-dition indicated higher confidence in the context of an aug-mented generation task (M = 622 vs M = 505) F(1 66) =857 p lt 005 This task was also associated with higher suc-cess likelihood(M = 66 vs M = 563) F(1 66) = 603 p lt 05and was perceived to be less difficult as well (M = 213 vs M= 381) F(1 66) = 1627 p lt 001 The associated interac-tions were significant as follows F(1 66) = 550 p lt 05 forthe confidence data F(1 66) = 1022 p lt 005 for the successlikelihood data and F(1 66) = 602 p lt 05 for the decisiondifficulty data

Discussion

The pattern of the data in the control conditionof Experiment3 is similar to the data from the first two experiments Re-spondents were less confident in decisions made in the con-text of a price-generation task compared to the selection andthe augmented generation conditions they expected the gen-eration task to have the lowest probability of yielding a suc-cessful outcome and also rated the generation task as rela-tively the most difficult In contrast participants in thepre-choice articulation condition displayed a distinctly dif-ferent pattern of responses whereby the relative disadvan-tage of the generation task observed in the control conditionwas less pronounced These findings are consistent with theexperimental hypothesis that the availability of a referenceprice range plays an important role in the price elicitationprocess

Experiment 3 also documents that to have an effect on thepreferenceelicitationprocess thesourceofthereferencepricerange does not necessarily need to be external internallygen-erated reference prices can facilitate the price-elicitationpro-cess as well This finding is especiallyimportant in lightof theargument advancedin thediscussionfollowingExperiment2The data from Experiment 3 clearly show that even when re-spondentsare notprovidedwithanyadditionalinformationre-garding the reference price ranges availability of these price

rangesat the timeof theprice-elicitationtaskcanhaveasignif-icant impact on consumersrsquo decision processes

An unpredicted yet interesting finding of this experimentis that the prechoice articulation condition participants weremore confident when presented with the augmented genera-tion task compared to both the selection and generation tasksRespondents in this condition also perceived the augmentedgeneration task to be less difficult and to have the highest like-lihood of success compared to the participants in the controlconditionThere are at least two factors that make this findingnoteworthyFirst recall that theaugmentedgenerationtask re-quired the most effort from respondents who were asked toelicit their price twice At the same time participants in thiscondition were provided with an external price range as a re-sultgeneratingan internal reference price shouldhave had noeffect Yet the data revealed a significant increase in thestrength of respondent preference for this condition

One possibleexplanationfor these data is that thevery pro-cessof thinkingaboutthe lowestpossiblepricerefinedpartici-pantsrsquo preferences about their willingness to pay Thus it ispossiblethattheprechoiceprice-elicitationprocessresultedina reference point that was very precise (eg $391 as opposedto $390) and respondents were searching for the same level ofaccuracy in the price-elicitation task that followed To illus-trate a consumerwho otherwisemightbe willingtopaysome-where around $280 for an airline ticket when asked to gener-ate the lowest possible price will indicate a specific numbersay $267 because she is seeking to express her willingness topay with the same degree of precision In this context the dif-ference between the selection and the augmented generationtaskcanbe attributedto thegreaterelicitationaccuracy associ-ated with the augmented generation task More general thesedata suggest that the price-elicitation task can be beneficialwhen there is acompatibilitybetweenthedegreetowhichcon-sumers have articulated willingness to pay and the elicitationaccuracy offered by the pricing task

GENERAL DISCUSSION

This research examines how consumers generate prices andspecifically the role of the price-elicitationtask and referenceprice availability in the price-elicitation process Contrary topopular belief that more choice is always better this researchdemonstrates that consumers often prefer a price-elicitationtask that offers less flexibilityand is more restrictive in allow-ing consumers to express their willingness to pay More im-portant this research identifies preconditions that moderatethe impact of the price-elicitation task on consumer prefer-ences Specifically I show that consumer price-generationprocessesare moderatedby the presence of a readily availablereference price This reference price can be either externallyprovided (Experiment 2) or internally generated (Experiment

60 CHERNEV

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

3) whichpointstoa moregeneralconstructunderlyingthedif-ferential impact of the selection and generation tasks

Data reported in this research are consistent with theview that consumers are often uncertain about the specificvalues of product attributes but are more certain about howtheir preferences tend to compare with other consumers inthe population (Prelec Wernerfelt amp Zettelmeyer 1997Wernerfelt 1995) Following this line of reasoning it ispossible that when a price range is readily available con-sumers who do not know their absolute preferences usetheir relative preferences to determine the price they arewilling to pay For example when presented with a rangeof prices from PMIN to PMAX in addition to estimating theirutility for paying these prices as suggested by the subjec-tive interpolation model depicted in Figure 1 consumersmight suggest a price that is reflective of their perception ofhow their individual characteristics (risk profile expertisedeal proneness) compared to those of the other individualsthus using the available price range as a proxy for the distri-bution of preferences in the population

Data presented in this research can also be linked to thenotion that consumers derive informational input from theexperienced ease or difficulty with which relevant materialcomes to mind (Waumlnke Bless amp Biller 1996 WaumlnkeBohner amp Jurkowitsch 1997) Thus when reference pricesare not readily available the consumer decision process israther complex and price articulation is likely to be per-ceived as more difficult consequently leading to lower de-cision confidence The price-generation task could alsohave a negative affect that impacts an individualrsquos evalua-tion of the alternatives under consideration (Garbarino ampEdell 1997) Thus as the complexity of the decision taskincreases so does the negative affect associated with thetask lowering the overall attractiveness of the alternativeunder consideration

Research presented here has important managerial impli-cations First it demonstrates that giving consumers morechoice in setting their own price might not always be the beststrategy and that on many occasions the method of namingonersquos own price is perceived as suboptimal More importantthis research shows that consumer preference for a price-elic-itation method is a function of the degree to which referenceprices are readily available This finding implies that firmscan benefit from offering customers reference price points(eg minimal bidding price) to use as benchmarks in theprice-elicitation process Alternatively firms can benefitfrom asking consumers to articulate a price range prior to theprice-generation task Finally firms can offer customers achoice of several price-elicitation strategies letting consum-ers self-select the one they feel most comfortable with Alikely result of implementing such strategy is that consumerswho perceive themselves as experts in a given product cate-gory will end up naming their own prices whereas nonexpertsmight prefer the simpler price-selection task

ACKNOWLEDGMENTS

I thank Dawn Iacobucci Angela Lee Vincent Nijs and BrianSternthal for their constructive comments This research waspartially funded by the McManus Research Chair that I re-ceived

REFERENCES

Alba J W Broniarczyk S M Shimp T A amp Urbany J E (1994) Theinfluence of prior beliefs frequency cues and magnitude cues on con-sumersrsquo perceptions of comparative price data Journal of Consumer Re-search 21(2) 219ndash235

Allwood C M amp Montgomery H (1987) Response selection strategiesand realism of confidence judgments OrganizationalBehavior amp HumanDecision Processes 39(3) 365ndash383

Berkowitz E N amp Walton J R (1980) Contextual influences on con-sumer price responses An experimental analysis Journal of MarketingResearch 17(3) 349ndash358

Bettman J R John D R amp Scott C A (1986) Covariation assessmentby consumers Journal of Consumer Research 13(3) 316ndash326

Bettman J R Luce M F amp Payne J W (1998) Constructive consumerchoice processes Journal of Consumer Research 25(3) 187ndash217

Garbarino E C amp Edell J A (1997) Cognitive effort affect and choiceJournal of Consumer Research 24(2) 147ndash158

Goldstein W M amp EinhornH J (1987)Expression theory and the prefer-ence reversal phenomena Psychological Review 94(2) 236ndash254

Griffin D amp Tversky A (1992 July) The weighing of evidence and thedeterminants of confidence Cognitive Psychology 24(3) 411ndash435

Heath T B Chatterjee S amp France K R (1995) Mental Accounting andChanges in Price The Frame Dependence of Reference DependenceJournal of Consumer Research 22(1) 90ndash97

Herr P M (1989)Priming price Prior knowledgeand context effects Jour-nal of Consumer Research 16(1) 67ndash75

Janiszewski C amp Lichtenstein D R (1999) A Range Theory Account ofPrice Perception Journal of Consumer Research 25(4) 353ndash369

Kahneman D amp Tversky A (1973) On the psychologyof predictionPsy-chological Review 80(4) 237ndash251

Lichtenstein D R Bloch P H amp Black W C (1988) Correlates of priceacceptability Journal of Consumer Research 15(2) 243ndash252

LynchJGChakravartiDamp MitraA (1991)Contrasteffects in consumerjudgmentsChanges in mental representations or in the anchoringof ratingscales Journal of Consumer Research 18(3) 284ndash297

Mahajan J (1992) The Overconfidence Effect in Marketing ManagementPredictions Journal of Marketing Research 29(3) 329ndash342

Mayhew G E amp Winer R S (1992) An empirical analysis of internal andexternal reference prices using scanner data Journal of Consumer Re-search 19(1) 62ndash70

Monroe K B amp Lee A Y (1999) Remembering Versus Knowing Issuesin Buyersrsquo Processing of Price Information Journal of the Academy ofMarketing Science 27(2) 207ndash225

Payne J W Bettman J R amp Johnson E J (1993) The adaptive decisionmaker New York Cambridge University Press

Prelec D Wernerfelt B amp Zettelmeyer F (1997) The role of inference incontext effects Inferring what you want from what is available Journal ofConsumer Research 24(1) 118ndash125

Simon H A (1955) A behavioral model of rational choice The QuarterlyJournal of Economics 69 99ndash118

Sinha A R amp Greenleaf E A (2000) The impact of discrete bidding andbidder aggressiveness on sellersrsquo strategies in open english auctions Re-serves and covert shilling Marketing Science 19(3) 244ndash265

PRICE ELICITATION STRATEGIES IN CHOICE 61

Urbany J E amp Dickson P R (1991) Consumer normal price estimationMarket versus personal standards Journal of Consumer Research 18(1)45ndash51

Varian H R (1999) Intermediate microeconomics A modern approach(5th ed) New York Norton amp Co

Waumlnke M Bless H amp Biller B (1996)Subjective experience versus con-tent of information in the construction of attitude judgments Personalityand Social Psychology Bulletin 22 1105ndash1113

Waumlnke M Bohner G amp Jurkowitsch A (1997) There are many reasonsto drive a BMW Does imagined ease of argument generation influence at-titudes Journal of Consumer Research 24(2) 170ndash177

Wernerfelt B (1995) A Rational reconstruction of the compromise effectUsing market data to infer utilities Journal of Consumer Research 21(4)627ndash633

Accepted by Dawn Iacobucci

62 CHERNEV

APPENDIXAn Overview of Price-Elicitation Strategies in Experiment 2