consumer choice between hedonic and utilitarian goods...consumer choice between hedonic and...

RAVI DHAR and KLAUS WERTENBROCH*

In this article, the authors examine how consumer choice between he-donic and utilitarian goods is influenced by the nature of the decision task.Building on research on elaboration, the authors propose that the relativesalience of hedonic dimensions is greater when consumers decide whichof several items to give up (forfeiture choices) than when they decidewhich item to acquire (acquisition choices). The resulting hypothesis thata hedonic item is relatively preferred over the same utiiitarlan item in for-feiture choices than in acquisition choices was supported in two choiceexperiments. In a subsequent experiment, these findings were extendedto hypothetical choices in which the acquisition and forfeiture conditionswere created by manipulating initial attribute-level reference states in-stead of ownership. Finally, consistent with the experimental findings, afield survey showed that, relative to market prices, owners of relatively he-donic cars value their vehicles more than do owners of relatively utilitar-ian cars. The authors discuss theoretical implications of these reference-dependent preference asymmetries and explore consequences for

marketing managers and other decision makers.

Consumer Choice Between Hedonic andUtilitarian Goods

Consumer choices are driven by utilitarian and hedonicconsiderations. Consumers choosing among new automo-biles, for example, may care about utilitarian features (e.g.,gas mileage) as well as about hedonic attributes (e.g., sportydesign). Research sugge.sts that these different considera-tions map onto independent components of product evalua-tions and attitudes and enable people to distinguish betweengoods according to their relative hedonic or utilitarian na-ture (Batra and Ahtola 1990; Mano and Oliver 1993).Broadly speaking, hedonic goods provide more experientialconsumption, fun, pleasure, and excitement (designerclothes, sports cars, luxury watches, etc.), whereas utilitar-ian goods are primarily instrumental and functional (mi-crowaves, minivans, personal computers, etc.; Hirschmanand Holbrook 1982; Strahilevitz and Myers 1998). If con-

*Ravi Dhar is Associate Professor of Marketing. Yale School ofManagement (e-mail: [email protected]) Klaus Wertenbroch is AssistantProfessor of Marketing, INSEAD (e mail: [email protected]).This article has benefiled from the comments of Steve Hoch, Eric Johnson,Daniel Kahneman. Ann McGill. Subrata Sen. Itamar Simonson, and threeanonymous JMR reviewers. The authors thank Mike Ziolkowski at theDecision Research Laboratory at the University of Chicago for his helpwith pan of Ihe data collection and Ecole des Hautes Etudes Commercialesfor funding, which we received as visiting researchers there. To interactwith colleagues on specific articles in this issue, see "Feedback" on theJMR Web site at wwwamaorg/pubs/jtnr.

sumers make choices between goods or characteristics withsuch different appeal, an account of consumer behaviorneeds to address the manner in which they make these fun-damental trade-offs.

In this article we examine consumer choice between twogoods, one of which is seen as superior on a hedonic di-mension and the other is seen as superior on a utilitarian di-mension. We compare preferences for these goods in an ac-quisition condition, in which the consumer chooses which ofthe two to acquire, and in a forfeiture condition, in which theconsumer chooses which of the same two items to give up.On the basis of the literature on the effect of elaboration onmessage evaluation (e.g., Tybout and Artz 1994), we pro-pose that greater spontaneous elaboration in forfeiturechoices increases the impact of hedonic aspects in overallevaluation. As a result, relative preferences for hedonic ascompared with the same utilitarian goods will be stronger inforfeiture than in acquisition choices. Consistent with ourunderlying theory, we show that the predicted asymmetrycan be attenuated using a thought-listing task that sup-presses the differential elaboration on the hedonic and utili-tarian dimensions.

We further propose that even in the absence of actual pos-session a choice can be framed as a forfeiture or as an ac-quisition decision on the basis of the attribute levels thatcharacterize a reference option. Consider, for example,someone who is debating between two apartments. One has

Journal of Marketinf- ResearchVol, XXXVll (February 2(KK)), 6()-7l 60

Hedonic and Utilitarian Goods 61

a nicer view (a relatively hedonic feature), but the other pro-vides a shorter commule to work (a relatively utilitarian fea-ture). If the person's current aparlmeni has a nice view anda short commute, the choice will he viewed as a forfeituredecision—a trade-off hetween a loss of quality of view anda loss of commuting convenience. In contrast, if the currentapartment has a poor view and a long commute, the choiceappears as an acquisition decision—a trade-off between again in quality of view and a gain in commuting conven-ience. We propose an increase in the relative preference forthe apartment that is superior on the hedonic dimensionwhen the decision is viewed as forfeiting a benefit ratherthan acquiring a benefit. We show that this asymmetry inpreferences due to the manipulation of the reference optioncan also be expressed in terms of differential loss aversionfor hedonic and utilitarian attributes (see Tversky and Kah-neman 1991).

The remainder of the article is organized as follows. Abrief review of prior research relevant to reference effectsand the role of elaboration in decision making leads to ourprediction of asymmetric preferences for hedonic and utili-tarian products in forfeiture and acquisition choices. Next,we test this prediction in three experiments involving realand hypothetical choices. As is illustrated in the apartmentexample, we use simple manipulations that determinewhether the hedonic-utilitarian trade-offs involve forfeitingor acquiring benefits. We then illustrate marketplace impli-cations of the experimental results in a field survey withused car data. We conclude with a discussion of the theoret-ical and managerial implications of our findings for pricing,promotion, and product modification strategies, which sug-gests that relative market shares for hedonic as comparedwith utilitarian products may depend on the frame of refer-ence used to evaluate these products.

PREFERENCE FOR HEDONJC AND UTILITARIANGOODS IN ACQUISITION VERSUS FORFEITURE

DECISIONS

Although the consumption of many goods involves bothdimensions to varying degrees (Batra and Abtola 1990),there is little doubt that consumers characterize some prod-ucts as primarily hedonic and others as primarily utilitarian.We define hedonic goods as ones whose consumption is pri-marily characterized by an affective and sensory experienceof aesthetic or sensual pleasure, fantasy, and fun (Hirschmanand Holbrook 1982). Utilitarian goods are ones whose con-sumption is more cognitively driven, instrumental, and goaloriented and accomplishes a functional or practical task(Strahilcvitz and Myers 1998). Similar to these findings onperceived product characteristics, recent work by Bazerman,Tenbrunsel, and Wade Benzoni (1998) suggests that we candistinguish between affective preferences ("wants") andcognitive or reasoned preferences ("shoulds") that underlieconsumer choice (see Shiv and Fedorikhin 2000;Wertenbroch 1998).' The want/should distinction is broadlycompatible with the distinction between bedonic and utili-tarian goods—items that are high on hedonic value arelikely to be subject to want preferences, and items that are

'Wenenbroch (1998) disdnguishes between "vice" and "virtue" goods,providing a formal conceptualization of goods that are subject to impulsivepreferences.

high on utilitarian value are likely to be subject to shouldpreferences. What has not been examined previously, how-ever, is whether evaluations of hedonic and utilitarian di-mensions and consequently the trade-offs between tbem aresystematically affected by tbe choice task.

Our focus on differences between acquisition and forfei-ture choices is motivated by the research on loss aversionthat demonstrates an asymmetry in evaluations dependingon the direction of the proposed trade, that is, whether agood is being acquired or forfeited relative to the con-sumer's present state (Kahneman, Knetsch, and Thaler1990; Tversky and Kahneman 1991). The conclusion fromthis body of work is that an item that is to be traded is gen-erally valued more when it is part of one's endowment thanwhen it is not. However, to the extent that both a bedonicand a utilitarian item are valued more wben they are for-feited than when they are acquired, the concept of loss aver-sion by itself does not provide any insight into relative as-sessments. Because acquisition and forfeiture choicespotentially involve different decision processes, we rely onthe compatibility principle that suggests that the evaluationof stimulus components may depend on the particular eval-uation task, affecting the decision maker's relative prefer-ences among the options (Shafir 1993; Tversky, Sattath, andSlovic 1988).

Previous research suggests why trade-offs between bedo-nic and utilitarian dimensions depend on tbe task. For ex-ample, a choice (as opposed to rating) task generally favorsthe option that is higher on the utilitarian dimension. Tver-sky and Griffin (1991; Shafir, Simonson, and Tversky 1993)propose that decision makers search for reasons and argu-ments to justify their choices. Similarly, Tversky, Sattath,and Slovic (1988) show that alternatives that provide deci-sion makers with compelling and justifiable arguments aremore likely to be preferred in choice tasks. In line with thisview, Btihm and Pflster (1996) show that contexts that fos-ter justification enhance preferences for utilitarian features.Recently, Bazerman, Tenbrunsel, and Wade-Benzoni (1998)bave suggested that choice forces decision makers to tbcuson should preferences so that they are more likely to favormore utilitarian options. In summary, choice tasks enhancethe relative salience of utilitarian consequences in overallevaluation in both acquisition and forfeiture.

Yet because acquisition and forfeiture cboices representdifferent choice tasks, the evaluation of each stimulus willalso depend on differences in how consumers process thesetasks. We propose that forfeiture choices stimulate morespontaneous elaboration than do acquisition choices, and wesuggest two reasons for this differential elaboration. First, itis likely that the more time consumers have to examine andinteract with the object in the forfeiture condition, the morethey will tend to elaborate on tbe object's potential benefits(see Strahilevitz and Loewenstein 1998). Yet, although theextent to which consumers engage in elaboration dependson the time or resources available, it is unlikely to be thesole cause of differential elaboration. A second argument forgreater elaboration in the forfeiture condition builds on theliterature on counterfactual thinking. Counterfactuals aremental representations of alternative realities compared withthose actually obtained. Researcb distinguishes betweencomparisons of actual outcomes with more preferred alter-natives (i.e., upward counterfactuals) and comparisons with

62 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2000

less preferred alternatives (i.e., downward counterfactuals)and suggests that upward counterfactuals are spontaneouslygenerated more frequently than downward counterfactuals(Roese and Olson 1997).- Recent research has extendedthese ideas to prefactual thinking, that is, the imagination ofalternative possible outcomes prior to choice (Sanna 1996).These findings suggest that consumers are more likely toelaborate spontaneously on alternative future outcomeswben tbey have to forfeit an item (i.e., an upward prefactual)than when they acquire an item (i.e., a downward prefactual;see also Carmon and Ariely 2000). For example, someonewho has so far enjoyed a nice view and a short commute towork from his or her apartment but now has to forgo one ofthese two features in deciding between two new apartmentsis more likely to imagine what it is like not to have the viewand the commute and contrast this with the old apartmentthan someone who has not possessed these features in thepast and is about to acquire one of tbem. In summary, wepropose greater elaboration in forfeiture choices.

We also propose tbat the presence of such differentialspontaneous elaboration in the forfeiture choice conditionenhances the relative valuation of hedonic attributes. This isbased on two arguments. First, a well-documented findingin tbe literature is tbat elaboration on a positive stimulusmessage can enhance the favorableness of judgment (Ty-bout and Artz 1994). Thus, imagining the use of a superior.

reason for this asymmetry in counlerfactual thinking is ihal thenegative affecl associated wilh worse outcomes is more likely to trigger Iheimagination of (better) altematives to reality (Kahneman and Miller 1986;Roese 1997). A second, adaptive reason is thai people who experience neg-ative or unpleasant outcomes are more likely to focus on actions thai couldhave been taken to avoid these oulcomes (Lewin \9^^: Roese 1997).

positively valued item should increase its attractiveness (seeShiv and Huber 1999; Strahilevitz and Loewenstein 1998).In particular, elaboration increases the intluence of moreeasily imaginable attributes on product evaluations, makingthem more salient (Keller and McGill 1994; Sherman et al.1985; Shiv and Huber 1999). To the extent that hedonic at-tributes are more sensory and imagery-evoking (Maclnnisand Price 1987), tbe relative attractiveness of an item tbat issuperior on the bedonic dimension sbould thus be enhanced.Second, upward prefactual thinking induces negative emo-tions, because one is about to be worse off tban before(Roese 1997; Sanna 1999). To the extent that forfeiturechoices spontaneously trigger upward comparisons thathighlight (negative) affective consequences, respondentsmay be motivated to minimize the anticipated negativeemotions by retaining the more hedonic good. Figure I sum-marizes the proposed process, by which differential elabo-ration influences tbe relative salienee of bedonic and utili-tarian attributes.

This leads to the following hypothesis about choices be-tween hedonic and utilitarian goods. Consider consumerswho may have to choose one of two options neither of whichthey currently own; alternatively, consider consumers whomay have to forfeit one of two options, both of which theycurrently own. Although the two decisions are logicallyequivalent (i.e.. the choice sets are identical), we predict thathedonic attributes will be weigbed more beavily in relativeterms wben consumers are deciding which one of two op-tions to give up as opposed to which one of two options toacquire. We now test this hypothesis in three experimentsand a field survey. The first two experiments show how rel-ative preferences for hedonic and utilitarian goods can

Figure 1DIFFERENTIAL ELABORATION IN ACQUISITION AND FORFEITURE CHOICES

DECISION TASK

ACQUISITION CHOICE

iLess elapsed lime till task

Spontaneous prefactuals less likely—> Less spontaneous elaboration

Relative salience of andpreference for utilitarian features

FORFEITURE CHOICE

iMore elapsed time till task

- I -

Spontaneous prefactuals more likely—> More spontaneous elaboration

IHedonic features easier toimagine and elaborate on

Relative salience of andpreference for hedonic features


change as a result of whether subjects cboose which of themto acquire or which of them to forfeit. Experiment 2 also in-volves examining the role of elaboration in tbe relative as-sessment of hedonic and utilitarian goods. Experiment 3shows that our predictions for ownership-dependent acqui-sition and forfeiture choices also apply to reference depend-ence in tbe absence of actual possession. Finally, tbe fieldsurvey sbows that, relative to market prices, owners of rela-tively bedonic cars value their vehicles more than do own-ers of relatively utilitarian cars.

EXPERIMENT I: FORFEITURE VERSUS ACQUISITIONCHOICE BETWEEN HEDONIC AND UTILITARIAN

GOODS

Method

Fifty-one undergraduate and graduate students at a pri-vate Midwestern university were recruited for this experi-ment with flyers posted around the campus. The stimuliwere two gift certificates with $7 face values, one for an au-dio tape (or as partial payment for a compact disc ICD|) ofthe subjects' choice at a nearby local record store, the otherfor a lO-pack of brand-name computer disks at the nearbyuniversity book store. These stores were cbosen to equalizetransaction costs.

Tbe between-subjects experimental design consisted ofan acquisition condition and a forfeiture condition. Subjectswere randomly assigned to one of tbe two conditions. Thedependent variable was subjects' choices between the giftcertificates. Subjects in both conditions were shown the twocertificates when they entered the laboratory. In the acquisi-tion condition, they were told that they would first need tofill out a series of questionnaires and then could choose oneof the certificates as compensation. In contrast, subjects inthe forfeiture condition were told at the outset tbat theycould keep botb certificates as compensation. When theyhad completed the (unrelated) questionnaires, the experi-menter asked subjects in the acquisition condition to chooseone of the certificates. She informed subjects in tbe forfei-ture condition tbat there had been a procedural error wbenshe bad given away both gift certificates and therefore askedthem to return one certificate. After recording subjects'choices, the experimenter debriefed tbem about the purposeof the experiment and gave them back tbe gift certificatetbat they had just returned. Thus, subjects in the acquisitioncondition received one $7 certificate, whereas subjects intbe forfeiture condition ultimately received a total of $14worth of certificates.

Results and Dtscitssicm

Pretest. We chose the two gift certificates as stimuli onthe basis of the results of two pretests. In the first pretest,which we adopted from Strahilevitz and Myers (1998), sub-jects from the same population categorized several everydayconsumer goods as primarily utilitarian (defined as useful,practical, functional, something that helps achieve a goal,e.g., a vacuum cleaner), as primarily hedonic (defined aspleasant and fun. something that is enjoyable and appeals totbe senses, e.g., perfume), as both utilitarian and bedonic, oras neither. The majority of subjects classified music audiotapes and CDs as primarily hedonic (17 of 22 subjects, x~ =6.55, p < .02) and cotnputer disks as pritiiarily utilitarian (18of 22 subjects, X" = 8.91, /J < .01). A second prelest sbowed

that the gift certificates for these two kinds of items wereseen as equally attractive.

Experiment. We predicted that the relative preference forthe more hedonic item over the utilitarian item would begreater in the forfeiture condition. In support of tbis hypotb-esis, subjects were significantly more likely to give up tbedisk certificate (and therefore to prefer the music certificate)when they were faced with a decision of which item to for-feit than they were to select the music certificate when theywere faced with a decision of which item to acquire. Eighty-four percent of the subjects (21 of 25 subjects) preferred tbemusic certificate in tbe forfeiture condition compared with54% (14 of 26 subjects) in the acquisition condition (x- =5.382, p = .02). This suggests that the relative evaluation ofthe hedonic characteristics of goods is more favorable inchoice when the options represent potential losses thanwhen they represent potential gains.

EXPERIMENT 2: SUPPRESSING DIFFERENTIAL ELAB-ORATION IN FORFEITURE AND ACQUISITION

CHOICES BETWEEN HEDONIC AND UTILITARIANGOODS

Experiment I demonstrated the predicted interaction be-tween type of good and decision task. The purpose of thenext experiment is threefold. First, we want to replicate theresults of Experiment I with different products to show thegenerality of the effect. Second, we want to rule out the pos-sibility that the observed choice patterns arise from a differ-ence in consumers' uncertainty in their evaluations of hedo-nic and utilitarian goods.- For example, if consumers aremore uncertain about the precise value of hedonic than ofutilitarian goods, the decision may be seen as a choice be-tween a sure (utilitarian) and an uncertain (hedonic) out-come. Risk aversion would tben predict the pattern of pref-erences observed in Experiment 1. Because both goods werestandard market goods, there is no a priori reason to believethat subjects associated greater uncertainty with the morehedonic good. To rule out this explanation empirically, wepretested the stimuli by measuring subjects' uncertaintyabout their monetary valuations of the stimuli (Nowlis andSimonson 1997).

Third and most important. Experiment 2 involves using athought-listing task to examine whether the increased pref-erence for the hedonic good in forfeiture choices resultsfrom greater spontaneous elaboration on the hedonic object.On the basis of previous research, we suggest tbat requiringsubjects to list reasons for tbeir choices should diminish thiseffect for two reasons. First, if spontaneous elaboration fo-cuses decision makers on affective consequences, listingreasons should de-emphasize the hedonic relative to the util-itarian focus of subjects' evaluations. General evaluations ofattitude objects reflect more utilitarian components whensubjects think about reasons for their attitudes (Bohm andPfister 1996; Millar and Tesser 1986; Wilson et al. 1993).Second, to the extent that spontaneous elaboration favors thehedonic good because it is easier to imagine or elaborate on(see Keller and McGill 1994; Maclnnis and Price 1987), atask tbat forces elaboration on both items should suppressdifferential elaboration. Thus, the difference in subjects'

'Batenian and colleagues (1997). for example, find marginally greaterreference dependence for rarely bought, harder-to-e valu ate chocolates thanfor frequently bought soft drinks.


preferences between hedonic and utilitarian goods in the ac-quisition and forfeiture choices should be smaller when theylist reasons before choosing than when they do not.

MethodOne hundred fourteen undergraduate students from a pri-

vate Northeastern university were recruited in their collegedormitories. Subjects were run in small batches of severalindividuals. In return for their participation in an unrelatedquestionnaire study, they were offered a large pack ofM&M's chocolate candies and a UHU glue stick, each witha retail value of approximately $1.25.

The study design was a 2 x 2 between-subjects full facto-rial. Similar to Experiment I, the first factor manipulatedwhether subjects decided between the alternatives in a for-feiture or acquisition condition. In the acquisition condition,subjects were first shown the two items and were told thatthey would have to choose one of them as compensation atthe end of the study. Next, they filled out the unrelated ques-tionnaire and then made their choice. In the forfeiture con-dition, subjects were given both items at the beginning ofthe procedure and were told that these were theirs to keep ascompensation for their participation. Upon completion ofthe questionnaire, the experimenter informed these subjectsthat there had been a procedural error and asked them to re-turn one of the items. To prevent subjects in the forfeiturecondition from feeling misled to expect to receive twoitems, they were told that they would later be provided witha consolation item. After recording each subject's decisionof wbich Item to forgo, the experimenter debriefed subjectsand gave back the forfeited item. The second factor was in-tended to suppress differential spontaneous elaboration inthe forfeiture condition. Specifically, subjects were asked towrite down the reasons they would like to own M&M's andglue sticks. Subjects in the control group received no suchinstructions. Subjects were randomly assigned to the fourconditions.

Results and Discussion

Pretests. The stimuli had been selected on the basis of theresults of two pretests with samples from the same subjectpopulation. The first pretest was the same as the one used indesigning the stimuli for Experiment 1 and showed that themajority of subjects regarded M&M's as primarily hedonic(40 of 46 subjects, x ' = 25.13, /?< .001) and UHU gluesticks as primarily utilitarian (34 of 46 subjects, x~ = 10.52,p < ,001), In the second pretest, subjects stated their will-ingness to pay for a pack of M&M's (M = $,83) and for aUHU glue stick (M = $1.27; t(3l) = -3.70, p < .001, two-sided) and rated how confident they were in these valuationsof the two items. Subjects showed greater confidence intheir ability to evaluate M&M's (M - 6.59 on a nine-pointscale) compared with UHU glue sticks (M = 5.72; t(3l) =1.98, p < .06. two-sided). Thus, greater relative preferencefor M&M's in forfeiture choices cannot be explained bygreater uncertainty in evaluating M&M's compared with aglue stick.

Experiment. We predicted a relative increase in the pref-erence for the hedonic item in the forfeiture condition. Theresults are reported in Table I. Across the two without-reasons conditions, subjects showed a relatively strongerpreference for the hedonic good in forfeiture choice. The

Table 1RELATIVE CHOICE FREQUENCIES FOR HEDONIC AND

UTILITARIAN GOODS WITH AND WITHOUT REASONS IN

EXPERIMENT 2 (n = 114)

Rea.wns

Witlwul reasomM&M's (hedonic)Glue slick (uiilitarian)

Total n in each choice

With reasonsM&M's (hedonic)Glue slick (utilitarian)

Toial n in each choice

Forfeiture

\5%

26

62%38%

29

Acc/uisitmii

50%50%

30

55%45%

29

M&M's were preferred by 85% of the subjects (22 of 26) inthe forfeiture condition and by 50% of the subjects (15 of30) in the acquisition condition. This replicated the findingin Experiment I. We further predicted that the a.symmetry inpreferences between forfeiture and acquisition would be at-tenuated if subjects first provided reasons for their prefer-ences. Consistent with this prediction, 62% of the subjects(18 of 29) preferred the M&M's in the forfeiture conditioncompared with 55% (16 of 29) in the acquisition condition.We used a logit model to conduct an overall test of the mainand interaction effects. The dependent variable was a 0-1dummy variable, where I denoted preference for theM&M's. The independent variables were as follows: (I) adummy variable for task (I = acquisition), (2) a dummyvariable for the reasons manipulation (I = reasons listing),and (3) the interaction of these two main effects. Consistentwith the hypotheses, the coefficient for task was significant<PTA.SK = ~-^^' P < •^^)- ^^ was the coefficient for the inter-action (PTASK X RKASONS ~ -^6, p < .10). This result providesadditional evidence that hedonic characteristics loom largerin forfeiture choices. When .subjects engaged in an activitythat reduced the hypothesized difference in elaboration onthe two goods^for example, listing reasons for owningboth items—the choice differential was considerablysmaller.

If, as we have suggested, the increased evaluation of he-donic characteristics is due to differential spontaneous elab-oration in the forfeiture condition, the imagined impact offorfeiting a hedonic item should be greater than the imag-ined impact of forfeiting an equivalent utilitarian item.Thus, in a brief follow-up study, wo directly compared theimagined impact of forfeiting a hedonic and an equally at-tractive utilitarian good by having subjects evaluate hypo-thetical outcomes. This approach is similar to previous re-search on evaluations of imagined outcomes (e.g.,Kahneman and Miller 1986; Schkade and Kahneman 1998).We used ratings instead of choice to reinforce the centralfinding from Experiments 1 and 2 with a different evalua-tion mode. Because the objects in each of the two item pairsin these experiments had been shown to be equally attractivein acquisition choices, we focused on the imagined impactof forfeiture only.

In a within-subjects design, subjects evaluated the lossexperienced by two hypothetical consumers. A and B, one ofwhom was described as having lost the more hedonic of twopreviously owned goods, whereas the other was described as


having lost the more utilitarian of the same two goods."* Inone scenario, they had each won the two $7 gift certificatesdescribed in Experiment I (i.e., one for an audio tape or CDof tbe person's eboice, the other for a 10-pack of disks). Ahad subsequently lost the music certificate but not the diskcertificate, whereas B had lost the disk certificate but not themusic certificate. In another scenario, A and B had each wona small bag of M&M's and a UHU glue stick. A had subse-quently lost the M&M's but not the glue stick, whereas Bhad lost the glue stick but not the M&M's. Note that thesescenarios paralleled the forfeiture conditions in Experiments1 and 2, except that forfeiture resulted from a loss in cir-cumstances beyond the target person's control (theft orbreakage) rather than from choice.

Sixty-seven subjects compared on nine-point rating scales(I) which of the target persons felt worse and (2) whichmissed the lost prize more (I = A who lost the music certifi-cate/M&M's, 9 = B who lost the disk certificate/glue stick).Subjects predicted that the target person who lost the hedo-nic music certificate would feel worse (t = -6.17, p < .0001)and miss the item more (t = -9.16, p< .0001) than would theperson who lost the utilitarian disk certificate. Similarly,subjects predicted that the person who lost the hedonicM&M's would feel worse (t = -3.18, p < .01) and miss theitem more (t - -2.47, p < .05) than would the person wholost the utilitarian glue stick. These results cannot be ex-plained by greater overall preferences for the M&M's or themusic certificate, because tbe two utilitarian items wereevaluated at least as bighly as the corresponding hedonicitems in the pretests. Instead, hedonic characteristics be-come more salient when subjects imagine tbe impact of for-feiture independent of choice.

EXPERIMENT J: REFERENCE EFFECTS IN CHOICESBETWEEN HEDONIC AND UTILITARIAN GOODS

Experiments I and 2 created acquisition and forfeiturechoices by manipulating actual ownership of a hedonic anda utilitarian good. Subjects either owned both and had togive one up or owned neither and had to choose one. Weused this design to obtain externally valid findings for actualconsumer goods of real monetary value. Recent researchsuggests that asymmetric valuations can also occur in theahsence of physical possession (Sen and Johnson 1997;Tversky and Kahneman 1991). Specifically, when con-sumers are provided with a reference point, they may eval-uate alternatives with respect to that reference point. Thus, achoice between the same two alternatives can be framed asa torfeiture or as an acquisition decision depending on theattribute levels that characterize a reference alternative.Evidence of a shift in preference due to a manipulation ofthe reference option would extend the scope of our previousfindings beyond the realm of ownership effects. We test thishypothesis using hypothetical choice problems betweencomparable alternatives tbat are described at the attributelevel.

The asymmetry in preferences due to a reference pointsbift can be expressed in terms of relative loss aversion forbedonic and utilitarian attributes. Consider the four stimulusitems in Figure 2. Choice option h is characterized by a bigbscore in the hedonic attribute and a low .score in the utilitar-

^Similar results were obtained with a bet ween-subjects design.

Figure 2REFERENCE DEPENDENCE FOR A HEDONIC AND A

UTILITARIAN PRODUCT ATTRIBUTE

Hedonicattribute

Utilitarianattribute

ian attribute. Choice option f is characterized by tbe reversescores. A superior reference item s has high scores in bothattributes, and an inferior reference item i has low scores inbotb. When the reference item is s, so that the decision iswhich of two superior attribute levels to forfeit, consumershave a relatively stronger preference for h over f (as shownby indifference curve U ) than when the reference item is i(as illustrated hy the steeper indifference curve Uj).

Our hypothesis implies that the ratio of the choice shareof h to the choice share of f is greater in forfeiture choices.These ratios can be transformed into a coefficient A.(if ofrelative loss aversion for hedonic and utilitarian goods asfollows:

(1)Pr(h > J) Pr{hPr(f

Pr(h

Pr(f

Pr(f

sh)

Pr(h y if)

Pr(f >- ih)

where >-s and y^ denote strong preference, given a superior(s) or an inferior (i) reference item, and A., and X( are tbe pa-rameters of loss aversion for h and f (see Tversky andKahneman 1991). Our prediction of stronger preferences forthe hedonic good when the reference item is superior im-plies that the relative loss-aversion coefficient i f is greaterthan I. Because such differential loss aversion may be afunction of attribute importance (Tversky and Kahneman1991), we design eboice problems such that the utilitarianattributes are at least as important as the hedonic attributes.

Method

Subjects were 141 undergraduate students at a privateNortheastern university. There were four choice problems(within subjects, in counterbalanced order), eacb witb tworeference item conditions (superior versus inferior; betweensubjects). Tbe reference options were designed according toFigure 2. In eacb problem, subjects decided between two al-


ternatives, one of which was superior in a utilitarian attrib-ute (point f in Figure 2), and the other was superior in a be-donic attribute (point h in Figure 2), all else being equal. Anexample is provided in the Appendix. The alternatives wereas follows:

•apartments: utilitarian attribute, distance to work (10 tninutesversus 45 tninutes); hedonic attribute, view from the apartment(breathtaking view of sunset and city skyline versus view of aparking lot).

•coworkers: utilitarian attribute, reliahility (very reliable versusnot very reliable); hedonic attribute, fun to work with (a lot offun versus somewhat arrogant).

•college luncfi pians: utilitarian attribute, walking distance locafeteria (5 minutes versus 10 minutes); hedonic attribute.dessert menu (cookies, pastry, and fresh fruit for dessert versusno dessert).

•sfiatnpoos: utilitarian attribute, cleansing efficacy (very effec-tive cleansing agent versus moderately effective cleansingagenl); hedonic attribute, softness of hair (hair feels soft andsilky versus hair feels dry after shampooing).

We had conducted a pretest to ensure that these pairs ofattributes differed in their hedonic and utilitarian contentand that the more hedonic attributes were not seen as moreimportant than the corresponding utilitarian attributes.Thirty-five subjects rated the relative hedonic and utilitariancontent as well as the importance of each attribute used inthe four problems. Using a measure adopted from Leclerc,Scbmitt, and Dube (1994), the hedonic ratings were an-chored at 1 = utilitarian and 9 = hedonic, where the termsutilitarian and hedonic were defined as in the pretests inExperiments 1 and 2, and the importance ratings were an-chored at I = not at all important and 9 = very important.

A brief cover story for eacb problem manipulated the ref-erence item. In the superior reference item condition (equiv-alent to a forfeiture condition; point s in Figure 2), subjectswere instructed to imagine themselves as currently consum-ing an alternative that was characterized by superior valuesin botb attributes (e.g., a 10-minute drive to work and abeautiful view of the sunset from their current apartment). Incontrast, m the inferior reference item condition (equivalentto an acquisition condition; point i in Figure 2), they wereasked to imagine themselves as currently consuming an al-ternative that was characterized by inferior values in both at-tributes (e.g., a 45-minute drive to work and a view of aparking lot from their apartment). In botb conditions, theywere told that they then had to switch to one of the two de-cision alternatives (e.g., because they had to move out oftheir current apartment). Subjects were randomly assignedto the two conditions. The dependent variable was subjects'cboices.


Pretest. The pretest results supported our manipulation ofthe relative hedonic and utilitarian content and importanceof the attributes. First, subjects distinguished clearly be-tween hedonic and utilitarian attributes in all four cases.Distance to work was seen as a highly utilitarian attribute ofapartments (M = 1.80), whereas the view from an apartmentwas rated as highly hedonic (M - 7.86; t = -16.52, p <.0001). Similarly, a coworker's reliability was seen as utili-tarian (M - 1.6), whereas fun in working with a coworkerwas seen as hedonic (M = 7.6; t = -14.71, p < .0001).

Distance to the cafeteria was a utilitarian attribute (M =2.63), whereas presence of the dessert menu was hedonic(M = 7.46; t = -8.32,/7< .0001). A shampoo's cleansing ef-ficacy was utilitarian (M = 2.23), whereas the softness ofone's hair was hedonic (M = 6.91; t = -9.67, p < .0001).

Second, across attribute pairs, the attributes that wererated as relatively more hedonic were never rated as moreimportant than the corresponding utilitarian attributes. Forapartments, distance to work (M = 6.82) was rated as moreimportant than the view (M = 6.06; t = 2.05, p < .05). Forcoworkers, reliability was rated as more important (M =7.74) than whether the coworker was fun to work with (M =6.17; t = 4.28, p < .0001). For lunch plans and shampoosthere was no significant difference in attribute importanceratings (at p < .20). Thus, tbe pretest results rule out that thegreater preference for the hedonic good in forfeiture choicesis confounded with greater importance of hedonic attributes.

Experiment. We predicted an increase in relative prefer-ence for the hedonically superior alternative in the superiorreference item condition compared with the inferior refer-ence item condition. The individual choice shares are re-ported in Tahle 2 and are discussed here for the apartmentproblem. In the apartment problem, 64% of the subjects se-lected the apartment that had the better view over the apart-ment characterized by tbe shorter commute wben the cur-rent apartment bad a breatbtaking view of tbe sunset andcity skyline and was a 10-minute drive from work (superiorreference item). In contrast, only 52% of the subjects chosethat apartment when the existing apartment was describedas overlooking a parking lot and being located 45 minutesfrom work (inferior reference item; )^f= 1.64). Assbown inTable 2, similar results were obtained across the four eboiceproblems.

We used a logit model to conduct an overall test of tbis ef-fect. The dependent variable was a O-I dummy variable,where I denoted preference for the item superior in the he-donic attribute. The independent variables were a dummyvariable for reference item (I - superior) and tbree dummyvariables for the individual choice problems. The results arepresented in Table 3. As predicted, subjects were signifi-

Table 2RELATIVE CHOICE FREQUENCIES FOR OPTIONS THAT ARE

SUPERIOR IN THE HEDONIC OR THE UTILITARIAN ATTRIBUTE

IN EXPERIMENT 3 (n= 141)

Opiioii.s:

Apartments (AB

CD

l.uiuii plansEF

ShainpiHis (>GH

Oiniim h Superior in:

\i,i = 1.64)Distance to work (utilitarian)View from apartment (hedonic)

Reliability (utilitarian)Fun to work with (hedonic)

Distance to cafeteria (utilitarian)Dessert menu (hedonic)

•hi = 2 . 4 1 )

Cleansing efficacy (utilitarian)Softness of hair (hedonic)

Total n in each choice

Reference Item

Superior

36%64%

52%48%

40%60%

2 1 %79%

70

Inferior

48%52%

63%37%

63%37%

39%61%

71


Table 3 K MLOGIT ANALYSIS PARAMETER ESTIMATES FOR PREDICTING

THE PROBABILITY OF PREFERRING THE ALTERNATIVE THAT

IS SUPERIOR IN THE HEDONIC ATTRIBUTE IN EXPERIMENT 3

InterceptSuperior reference itemApartmentsLunch plansShampoo

Log-likelihoodChi-squaredfn

ParameierEslimtiie

-.66*.70*«.65***.24

I.I6*«

-388.5340.35* •4

564

SumdardError

.20

^M.25

.0001 ; /?< .01 .

jS»«*y-more likely to ehoose the alternative that was supe-rior in the hedonic attribute when the decision was madegiven a superior reference item than when it was made givenan inferior reference item (PSUPERIOR REFERENCE ITEM = -70;/7<.0OOl).

This result lends further support to our hypothesis that he-donic consequences loom larger in forfeiture choices. Insteadof inducing actual losses and gains of alternatives, this ex-periment induced acquisition and forfeiture frames by askingsubjects to choose hetween two items that provided improve-ments or deerements in attributes relative to a two-dimen-sional reference item. This experiment extends our previousfindings by demonstrating asytnmetric evaluations of hedo-nic and utilitarian goods that result not just from a manipula-tion of ownership but from using a stated comparator.

FIELD SURVEY: MARKETPLACE IMPLICATIONS OFTHE ASYMMETRY IN FORFEITURE AND ACQUISITION

CHOICES

We obtained the results so far using both actual and hy-pothetical options under controlled laboratory conditions.Although the use of real products enhanced the external va-lidity of the findings, we wanted to explore the implicationsof these results for consumers' valuations of goods in themarketplace. A direct implication ol" the Undings is that, incomparison with potential buyers, owners of hedonic goodsshould be more reluctant to forgo these (i.e., demand higherselling prices) than are owners of comparable utilitariangoods. As a consequence, buyer-seller price gaps should belarger for hedonic than for utilitarian goods.

We test this hypothesis in a tleld survey using an open-ended contingent valuation measure (see Mitchell and Car-son 1989), because we have previously focused on showingthe effect in purely choice-based designs. Automobiles areparticularly appropriate for this kind of comparison, becausetbey differ in hedonic versus utilitarian content and are of-ten advertised along these dimensions. Moreover, used carmarket price data are publicly available. We predict thatowners of more hedonic cars will demand higher sellingprices (willingness to accept or WTA) relative to potentialbuyers' willingness to pay (WTP) than will owners of moreutilitarian cars. Because nonowners* (i.e., buyers") WTP isbuilt into aggregate market prices, tbe ratio of WTA to mar-

ket prices serves as a conservative approximation ofbuyer-seller price gaps.-**

Method

Two hundred seventeen incoming MBA students at a pri-vate Southeastern university filled out a questionnaire dur-ing an orientation event (before any coursework), in wbichthey were asked which car (make and model), if any, theycurrently owned. Subjects were asked to imagine that theywere to sell their car in the next 30 days and to state the min-imum selling price (WTA) they would demand. They thenrated their cars on the hedonic and utilitarian dimensionsseparately on nine-point scales. The questionnaire also in-cluded the following control variables: whether subjectswere leasing or financing tbeir vehicle (coded as an indica-tor variable) to account for a possible effect of legal owner-ship on valuation, the year the car was built, the car's ap-proximate current mileage, the price at which they hadbought tbe car, and the perceived uniqueness of the ear on anine-point rating scale.f' Last, subjects stated if they wereaware of the current value of the car from Keliey's Bluebook(an authoritative price list used by many car dealers and in-surance companies to determine used car values), as thatmight reduce any possible differences between reservationand market prices. In addition to the survey responses, wealso detemiined as an approxitiiation of the market's WTPthe current secondhand market price from Keliey's Bluebookfor each vehicle.


We predicted that the ratio of WTA-to-market priceswould be higher for hedonic than for utilitarian cars. The re-sults support this predicti()n and are reported in Table 4. Ourrespondents owned 30 different brands of vehicles.Excluding those respondents who stated that they wereaware of the Bluebook prices of their cars had no significant

WTA-WTP gups imply undertfading (Kahneman. Knetsch.and Thaler 1990}. market pnces capture WTP only ol those nonowners whodo buy and therefore exceed average WTP across all nonowners, includingthose who do not buy.

""Ageand mileage served asconlrols for any systemalic differences in de-preciation and usage behavior between cars viewed as utilitarian and hedo-nic. Because WTA-WTP discrepancies may be motivated in part by the dif-ficulty of finding substitutes, including original purchase prices anduniqueness ratings ensures that the hypothesized difterence in buyer-sellergaps for hedonic and utilitarian cars is not just due to incotne and substitu-tion effects (Hanemann 1991).

Table 4ORDINARY LEAST SQUARES REGRESSION RESULTS FOR

AUTOMOBILE SELLING-PRICE PREMIUMS IN THE FIELD

SURVEY (R2 = .22, p<.0001)

InterceptDifferenceLeasing (yes = 1)Bought priceYear builtMilesUnique

ParameterEsiirnaie

1.361.017*.024.160**

-.012- .085***-.<H)8

StandardError

1.135.005.030.041.019.OM,008


effect on the results. We regressed the ratio of reservationprices to Bluehook prices on the natural logs of the originalpurchase price and mileage, as well as on the year the carwas built, subjects' perceived uniqueness ratings, and acomposite measure of subjects' ratings of the hedonic andutilitarian characteristics oftheir vehicles. This measure wasthe difference between each individual's hedonic and utili-tarian ratings. Thus, values could range from -8 (indicatingpurely utilitarian vehicles) to 0 (indicating vehicles tbat areseen as both utilitarian and hedonic or as neither) to +8 (in-dicating purely hedonic vehicles).

As predicted, the greater the net hedonic content of tbe ve-hicle, the higber the ratio of respondents' stated selling reser-vation priees to market prices (PUIFFERENCE = f*' ' P ^.001). Owners of hedonic cars were more reluctant to partwith them than were owners of utilitarian cars.'' Among tbecontrol variables, only mileage (PMILES = -085;p< .05) andoriginal purebase prices (PBOUGHT.PRICE - '^0; p < .(XX)1)affected the ratio of WTA to market prices, whicb suggestspossible usage rate and income effects on buyer-seller pricegaps for used cars. Tbere was no evidence of multicollinear-ity in the independent variables. In summary, tbis study il-lustrates marketplace implications of our experimental re-sults that show tbat hedonic aspects loom larger in forfeiture(e.g., selling) tban acquisition (e.g., buying). Includinguniqueness ratings and original purcbase prices in our analy-sis controls for the rival explanation that hedonic cars, whichmigbt be more expensive tban utilitarian cars, are seen bytbeir owners as unique collectibles with bigh investmentvalue. We note, however, tbat this nonexperimental fieldstudy can only provide suggestive evidence consistent withour bypotbesis but naturally cannot confirm it.

GENERAL DISCUSSION

Previous research has sbown that consumer perceptionsand preferences bave both bedonic and utilitarian dimen-sions. We demonstrate a fundamental asymmetry in bowconsumers trade off these dimensions in acquisition and for-feiture choices. Our data consistently show an increase inthe weight of tbe hedonic aspects in forfeiture cboices.Experiments I and 2 manipulated real ownership of two dif-ferent pairs of products. Subjects in both experiments showa relative increase in the preference for tbe bedonic good inforfeiture compared witb acquisition choices. Experiment 2also showed that this effect is moderated by the relativesalience of bedonic considerations in tbe forfeiture condi-tion. Tbis was predicted on the basis of tbe notion tbat tbeincreased opportunity for spontaneous elaboration in forfei-ture enhances the evaluation of hedonic goods. Experiment3 replicated the preference asymmetry by inducing a forfei-ture frame tbrough a simple attribute-level reference pointmanipulation instead of imposing real losses or gains onsubjects. Finally, the field survey sacrificed experimentalcontrol to illustrate marketplace implications of the asym-metry in forfeiture and acquisition choices. Owners of he-donic cars valued their vehicles more tban did owners ofutilitarian cars, relative to market prices.

Tbe series of studies limits the effect of alternative ac-counts of why consumers may be more reluctant to part witb

•'The results reflecl Bluehimk prices for cars in g(X)d condilion and aresimilar for vehicles in excellent condilion.

bedonic tban witb utilitarian goods. Work by Belk (1988)suggests tbat consumers develop symbolic relationshipswitb tbeir possessions. If tbese relationsbips are stronger forhedonic tban for utilitarian possessions, consumers migbtreasonably value sucb options more over time. However, theduration of ownership in Experiments 1 and 2 appears toobrief for such differences in relationsbips to develop.^ A re-lated argument can be derived from Hanemann (1991), inwbicb he argues tbat consumers' true selling prices (WTA)are a function of the substitutability and tradability of tbegood to be traded. If hedonic goods are more unique and ir-replaceable (e.g., a bridal gown), perhaps because we de-velop emotional attachments to them over time, consumersmight be more reluctant to forfeit tbem. Altbougb possiblein general, tbese arguments do not apply to Experiments 1and 2, in wbicb tbe alternatives used were widely availablemarket goods. Moreover, we controlled for the effect of sub-stitutability in tbe field survey by including perceiveduniqueness as a covariate in the analysis. In practice, tbesealternative processes are likely to enhance tbe strengtb of tbephenomenon, providing promising areas of further research.

Similarly, another rationale for greater preferences forutilitarian items in acquisition choices can be derived fromKabn and Meyer (1991), In wbicb tbey show that tbe sub-jective importance of attributes tbat are seen as enbancing orpreserving a status quo can be altered by the level of attrib-ute uncertainty. Specifically, they show that increasing tbisattribute uncertainty increases the weight of preserving at-tributes and diminishes the weight of enhancing attributes. Ifpeople consider utilitarian goods a means of preserving ben-efits in day-to-day life (e.g., a fork is a means of avoidingeating with one's bands) and bedonic goods are thought ofas tools that provide enhancements (e.g.. wine is a tool forenhancing the quality of a meal), then the weight of utilitar-ian (i.e., preserving) attributes would increase in acquisitioncboices if these are characterized by greater attribute uncer-tainty.y However, it is highly unlikely that there was a dif-ference in attribute uncertainty between tbe acquisition andforfeiture conditions in our experiments. We used marketgoods, with whicb subjects in botb conditions bad similarprior experience, and tbe opportunity for incremental learn-ing in tbe experiments was limited.

The asymmetry in preference for the hedonic good be-tween tbe acquisition and forfeiture frames is potentiallyconsistent with findings on tbe omission bias (Baron and Ri-tov 1994; Spranca, Minsk, and Baron 1991). Baron and col-leagues report that consequences arising from action orcboice induce greater feelings of responsibility than do con-sequences arising out of inaction or omission. It tbere isgreater guilt associated with choosing a hedonic item (seeStrahilevitz and Myers 1998) and if retaining bedonic goodsinduces less guilt than acquiring them, hedonic items maybe relatively less preferred in acquisition cboices. However,tbere are several reasons this argument does not provide avalid rival explanation for our findings. The hedonic stimuliin our studies are fairly regular consumption items. More

''Recenl research suggests that even short incremenis in duration of own-ership may affect absolute valuations (Strahileviiz and Loewenslein 1998)but makes no predictions about differential evaluations. Further researchmight examine whether such differences exist for hedonic and utilitariangotxis.

''We thank an anonymous reviewer for this suggestion.


important, subjects in the forfeiture condition also made ac-tive choices instead of receiving an item as the result of in-action. Thus, both conditions should have induced an equaldegree of rcsponsihility and guilt in suhjects. Furthermore,the data described at the end of Experiment 2 showed thatthe loss of a hedonic good was evaluated worse, even whenno choice occurred and when it was described as the resultot circumstances not under the protagonist's control (i.e.,theft or breakage).

Finally, we tested this rival explanation directly by exam-ining whether acquisition choices are seen as inducing moreguilt than are forfeiture choices. Following Spranca, Minsk,and Baron {1991), we asked subjects to evaluate the degreeof guilt felt by two hypothetical persons for choosing a he-donic good. A forfeiture condition was created by describinga person who had mistakenly received both a hedonic{M&M's or CD certificate) and a utilitarian {glue stick ordisk cerlificale) prize in a lottery and then had to forfeit oneof them when the mistake was di.scovered. In an acquisitioncondition, a person was simply described as having won achoice hetween the same two items. Both persons were de-picted as having selected the hedonic item. A random sam-ple of 80 passengers at a regional airport rated on a nine-point scale the amount of guilt the target persons felt as aresult of selecting the hedonic item in the acquisition andforfeiture conditions. The mean guilt ratings for selectingthe hedonic item were not significantly different in the for-feiture and the acquisition conditions {M = 2.69 and M =2.53; F{l,78) - .25, p < \}. These results do not support acorrelation between responsibility and guilt and a greaterpreference for hedonic items in forfeiture choices.

The limitations of the present research point out promis-ing areas for further research. One relates to the choice prob-lems that were used. The choice sets in our experimentswere limited to two alternatives. Greater complexity andtask realistn in acquisition and forfeiture choices from a setof more than two alternatives may induce different decisionprocesses that may change the proposed eftect. In addition,our outcome- rather than process-oriented methodologydoes not examine the thoughts that subjects spontaneouslygenerate in the two conditions and that are predicted to me-diate the asymmetry in relative evaluations. Further researchcould examine think-aloud protocols to provide additionalsupport for the existence and effect of differential elabora-tion in acquisition and forteiture choices. Last, it would beinteresting to test whether the preference asymmetry we ob-served is due solely to the intrinsic properties of hedonic andutilitarian goods or whether the effect extends to other prod-uct features because of more general differences in the easewith which these can be elaborated on.

Theoretical Implications

Tbis fundamental asymmetry in how consumers trade offhedonic and utilitarian product attributes in domains oflosses and gains also sheds light on the discussion of thecauses of loss aversion and the processes hy which it oper-ates (e.g., Hanemann 1991; Sen and Johnson 1997).Although the phenomenon itself is well established, rela-tively little is known about the exact processes that underlieasymmetric valuations of gains and losses. To address thisgap, we need a systematic analysis of key moderators thatdrive differences in reference dependence and loss aversion

across categories of goods and attributes. Such differenceshave been shown, for example, by Hardie, Johnson, andFader (1993), who demonstrate greater loss aversion forproduct quality than for price, and by lrwin (1994), whofinds greater loss aversion for environmental (public) goodstban market {private) goods.

Tbe present findings contribute to this discussion in twoways. First, the endowment effect and buyer-seller pricegaps may arise from a differential focus on the hedonic andutilitarian aspects of a traded good if owners/sellers aremore likely than nonowners/buyers to engage in sponta-neous elaboration in determining their evaluations of tbatgood. This has several implications for further research re-lated to loss aversion. For example, we could examine theeffect of asking buyers to imagine the actual experience withthe good to be traded, which should attenuate buyer-sellerprice gaps. Furthemiore, the variation in loss aversion{measured as X) across goods can be examined to determineif it is correlated with the hedonic content of these goods andwith ease of elaboration.

Second, we designed Experiment 3 so that we could com-pare loss-aversion coefficients for hedonic (A.|,) and utilitar-ian (Xf) attributes by computing a relative loss-aversion co-efficient Xf,t directly from observed choice shares. Thechoice-based nature of this design enables re.searchers to es-timate relative loss aversion without having to determine thesize of the individual eoeffieients from the usual WTP andWTA measures that previous research has used. To the ex-tent that consumer preferences in markets are revealedthrough choices (i.e., joint evaluations), WTP and WTA(i.e., separate evaluations) potentially introduce a source ofmeasurement error and bias {see Hsee 2000). Further re-search can use the present choice based design to derive po-tentially more valid estimates of relative loss aversionacross attributes and commodity types.

Managerial Implications

Managerial implications of the findings are straightfor-ward. At a strategic level, if competing firms are forced tocut existing product attribute or service levels {see Sen andMorwitz 1996), consumers may he more reluctant to acceptcuts on the more hedonic dimensions. In contrast, addingthe same hedonic benefits may have relatively less impacton market share than would adding more utilitarian benefits.Similar implications may hold for bargaining situations thatinvolve trade-offs between hedonic and utilitarian benefits.For example, labor unions may be more likely to rejectmanagement proposals to cut funding for cotnpany-ownedvacation retreats {a hedonic benefit) than to reject proposalsfor a slight increase in the number of working hours (a util-itarian feature) but may value similar improvements inworking hours relatively more than increased funding of va-cation retreats.

Our results also suggest implications for pricing and pro-motion strategies. Marketers ought to be able to charge pre-miums for hedonic goods to which consumers have adaptedin some manner when the consumers are faced with a deci-sion to discontinue consumption. For example, all else be-ing equal, marketers may he able to add a bedonic premiumto the buyout option price at whicb lessees of luxury orsports cars can buy their vehicles at the end of the leaseterm. Alternatively, we suspect that buyout rates are higher

70 JOURNAL OF MARKETING RESEARCH. FEBRUARY 2000

for these hedonic cars than for more utilitarian ones, such ascompact cars or minivans. Introductory special offers are of-ten used to acquire new customers. Our results suggest thatacquisition through trial periods and samples may be rela-tively more effective for hedonic (e.g., cahle television) thanfor utilitarian (e.g., encyclopedias) goods. All else beingequal, this may make low introductory price offers espe-cially attractive for hedonic goods. More generally, our re-sults also indicate that secondhand markets involving pri-vate sellers may be less efficient for hedonic than forutilitarian goods, because owners of hedonic goods may berelatively more reluctant to sell at prices that potential buy-ers are willing to offer.

Asymmetric preferences due to a simple task manipula-tion raise the question of which frame is more appropriatewhen consumers make purchase decisions. From a norma-tive perspective, trade-offs between hedonic and utilitarianalternatives to derive overall evaluations should be made in-dependently of particular reference items, which makes ei-ther frame suspect. Descriptively speaking, the answer tothis question may depend on a consumer's propensity to fo-cus on forgone alternatives. If consumers tend to elaborateon what might have been, choosing the more hedonic optionmay make them happier. However, if out of sight is out ofmind for consumers, the more utilitarian option may be thebetter choice.

APPENDIX

We provide examples of one of the stimuli (apartments) inthe inferior reference item (corresponding to acquisition;upper panel) and superior reference item (corresponding toforfeiture; lower panel) conditions in Experiment 3:

Apartments

Imagine that for the past year you have been renting aone bedroom apartment that has the following features:

• overlooks a large parking lot.• is a 45-niinute drive from your place of work.

Now you have to move out of this apartment, and youface a decision of renting one of the two apartments de-scribed below. Both apartments have one bedroom and aresimilar in all other respects (for example, monthly rent,safety).

View Distance lo Work

Your presemapartment

Apanment I

Apartnienl [l

Overlooks alarge parking lot

Breathtaking viewof sunset andcity skyline

Overlooks a largeparking lot

4.'>-minute drive

45-minule drive

lO-minuIe drive

Apartments

Imagine that for the past year you have been renting aone-bedroom apartment that has the following features:

•has a breathtaking view of the sunset and city skyline.•is a lO-minute drive from your place of work.

Now you have to move out of this apartment, and youface a decision of renting one of the two apartments de-scribed below. Both apartments have one bedroom and aresimilar in all other respects (for example, monthly rent,safety).

View Distance to Work

Your presentapartment

Apartment I

Apartment II

Breathtaking viewof sunset andcity skyline

Brealhtaking viewof sunset andcity skyline

Overlooks a largeparking lot

lO-minute drive

45-minute drive

10-minute drive

REFERENCESBaron, Jonathan and liana Ritov (1994), "Reference Points and

Otnission Bias," Organizational Beliavior and Human DecisionProcesses, 59 (3). 475-98.

Bateman, Ian, Alistair Munro, Bruce Rhtxies, Chris Starmer, andRobert Sugdcn (1997), "A Test of the Theory of Reference-De-pendent Preferences," Quarterly Journal of Economics. 112(May), 479-505.

Batra, Rajeev and Olli T. Ahtola (1990), "Measuring the Hedonicand Utilitarian Sources of Con.sumcr Attitudes," Marketing Let-ters, 2 (2), 159-70.

Bazerman, Max H., Ann E. Tenbrunscl, and Kimberly Wade-Ben-zoni (1998), "Negotiating with Yourself and Losing: Under-standing and Managing ContTicting Internal Preferences,'Mca(/-emy of Managemenl Review, 23, 225-41.

Belk. Russell W, (1988). "Possessions and the Extended Self,"Journal of Consumer Research. 15 (September). 139-68.

Bohm. Gisela and Hans-Riidiger Pfister (1996), "Instrumental orEmotional Evaluations: What Determines Preferences?" AdaPsychologica, 93 (1-3), 135-48.

Carmon, Ziv and Dan Ariely (2000), "Focusing on the Forgone:How Value Can Appear So Different to Buyers and Sellers."Journal of Consumer Research, forthcoming.

Hanemann, W. Michael (1991), "Willingness to Pay and Willing-ness to Accept: How Much Can They Differ?" American Eco-nomic Review. 81 (June), 635-47.

Hardic, Bruce G., Eric J. Johnson, and Peter S. Fader (1993),"Modeling Loss Aversion and Reference Dependence Effects onBrand Choice," Marketing Science. 12 (Fall), 378-94.

Hirschman, Elizabeth C. and Morris B. Holbrook (1982), "Hedo-nic Consumption: Emerging Concepts, Methods and Proposi-tions," Journal of Marketing. 46 (Summer), 92-101.

Hsee, Christopher K. (2(X)0), "Attribute Evaluability and Its Impli-cations for Joinl-Separate Evaluation Reversals and Beyond," inChoices. Values, and Frames, Daniel Kahneman and Amos Tver-sky, eds. New York: Cambridge University Press, forthcoming.

Irwin, Julie R. (1994), "Buying/Selling Price Preference Reversals:Preference for Environmental Changes in Buying Versus SellingModes," Organizational Behavior and Human DecisionPmce55^5, 60(3), 431-57.

Kahn, Barbara E, and Robert J. Meyer (1991), "Consumer Multi-attribute Judgments Under Attribute-Weight Uncertainty," Jour-nal of Consumer Research, 17 (March), 508-22.

Kahneman, Daniel. Jack L. Knetsch, and Richard H. Thaler (1990),"Experimental Tests of the Endowment Effect and the CoaseTheorem," Journal of Political Economy. 98 (6), 1325^8.

and DaleT Miller (1986), "Norm Theory: Comparing Re-ality to its Alternatives," Psychological Review. 93 (2), 136-53.

Keller, Punam Anand and Ann L. McGill (1994), "Differences inthe Relative Intluence of Product Attributes Under AlternativeProcessing Conditions: Attribute Importance Versus Attribute


Ease of Imagability," Journal of Consumer Psychology, 3 (1),29-19.

Leclere, France, Bernd H. Schmitt, and Laurette Dub6 (1994),"Foreign Branding and its Effects on Product Perceptions andAttitudes." Journal of Marketing Research, 31 (May), 263-70.

Lewin, Kurt (1935), A Dynamic Theory of Personality: Selected Pa-pers, D.K. Adams and K.E. Zener, trans. New York: McGraw-Hill.

Maclnnis, Deborah J. and Linda L. Price (1987), "The Role of Im-agery in Information Processing: Review and Extensions," Jour-nal of Consumer Research, 13 (March), 473-91.

Mano, Haim and Richard L. Oliver (1993), "Assessing the Dimen-sionality and Structure of the Consumption Experience: Evalua-tion, Feeling, and Satisfaction," Journal of Consumer Research,20 (December), 451-66.

Millar. Murray G. and Abraham Tesser (1986), "Effects of Affec-tive and Cognitive Focus on the Attitude-Behavior Relation."Journal of Personality and Social Psychology, 51 (2), 270-76.

Mitchell, Robert C. and Richard T. Carson (1989). Using Surveysto Value Public Goods: The Contingent Valuation Method. Bal-timore: Johns Hopkins University Press.

Nowlis, Stephen M. and Itamar Simonson (1997), "Attribute-TaskCompatibility as a Determinant of Consumer Preference Rever-sals," Journal of Marketing Research, 34 (May), 205-18.

Roese, Neal J. (1997), "Counterfactual Thinking." PsychologicalBulletin. 12! (i), 133^8.

and James M. Olson (1997), "Cou nterf actual Thinking:The Intersection of Affect and Function," in Advances in Exper-imental Social Psychology. Vol. 29, Mark P. Zanna, ed. SanDiego: Academic Press, 1-59.

Sanna, Lawrence J. (1996), "Defensive Pessimism, Optimism, andSimulating Alternatives: Some Ups and Downs of Prefactualand Counterfactual Thinking," Journal of Personality and SocialPsychology, 71 (5), 1020-36.

(1999), "Mental Simulations, Affect, and Subjective Con-fidence: Timing Is Everything." Psychological Science, 10(July), 339^5.

Schkade, David A. and Daniel Kahneman (1998), "Does Living inCalifornia Make People Happy? A Focusing Illusion in Judg-ments of Life Satisfaction," Psychological Science, 9 (Septem-ber), 340-^6.

Sen, Sankar and Eric J. Johnson (1997), "Mere-Possession Effectswithout Possession in Consumer Choice," Journal of ConsumerResearch, 24 (June), 105-17.

and Vicki G. Morwitz (1996). "Is It Better to Have Lovedand Lost Than Never to Have Loved at All? The Effects ofChanges in Product Features Over Time," Marketing Letters, 7(3), 225-35.

Shafir, Eldar (1993), "Choosing Versus Rejecting: Why Some Op-tions Are Both Better and Worse Than Others," Memory & Cog-nition,2\ (July), 546-56.

, Itamar Simonson. and Amos Tversky (1993), "Reason-Based Choice," Cognition, 49 (October/November), 11-36.

Sherman, Steven J,. Robert B. Cialdini, Donna F. Schwartzman,and Kim D. Reynolds (1985). "Imagining Can Heighten orLower the Perceived Likelihood of Contracting a Disease: TheMediating Effect of Ease of Imagery," Personality and SocialPsychology Bulletin, II (March), 118-27.

Shiv, Baba and Alexander Fedorikhin (2000), "Heart and Mindin Conflict: The Interplay of Affect and Cognition in Con-sumer Decision Making," Journal of Consumer Research,forthcoming.

and Joe! Huber (1999). 'The Impact of Anticipating Satis-faction on Choice." working paper. College of Business, Uni-versity of Iowa.

Spranca, Mark, Elisa Minsk, and Jonathan Baron (1991), "Omis-sion and Commission in Judgment and Choice," Journal of Ex-perimental Social Psychology, 27 (1), 76-105.

Strahilevitz, Michal A. and George F. Loewenstein (1998), 'TheEffect of Ownership History on the Valuation of Objects." you/--nal of Consumer Research, 25 (December), 276-89.

and John G. Myers (1998), "Donations to Charity as Pur-chase Incentives: How Well They Work May Depend on WhatYou Are Trying to Se!!," Journal of Consumer Research, 24(March). 434-^6.

Tversky, Amos and Da!e Griffin (1991), "Endowment and Con-trast in Judgments of Wel!-Being," in Subjective Weil-Being:An Interdisciplinary Perspective. Fritz Strack, Michae! Ar-gyle, and Norbert Schwarz, eds. Oxford: Pergamon Press,101-18.

and Daniel Kahneman (199!). "Loss Aversion in RisklessChoice: A Reference-Dependent Model," Quarterly Journal ofEconomics, 106(4), 1039-61.

-, Shmuel Sattath, and Pau! Slovic, (1988). "ContingentWeighting in Judgment and Choice," Psychological Review, 95(3), 371-84.

Tybout. A!ice M. and Nancy Artz (1994), "Consumer Fsycho!ogy,"Annual Review of Psychology. 45, 131-69.

Wertenbroch, Klaus (!998), "Consumption Se!f-Control by Ra-tioning Purchase Quantities of Virtue and Vice," Marketing Sci-ence. 17(4), 317-37,

Wi!son, Timothy D., Douglas J. Lisle, Jonathan W. Schooler, SaraD. Hodges, Kristen J. Klaaren, and Suzanne J. LaFleur (!993),"Introspecting About Reasons Can Reduce Post-Choice Satis-faction," Personality and Social Psychology Bulletin, 19 (3),331-39.

consumer choice between hedonic and utilitarian goods...consumer choice between hedonic and...

Documents