RAVI DHAR and ITAAAAR SIMONSON*

The authors propose that the attractiveness and choice probability of an alter-

native can be enhanced by making it the focus of a comparison (the focal option)

with a competing alternative. This proposition is supported in choice problems in-

volving alternatives about which consumers have information in memory (e.g., fro-

zen yogurt and fruit salad). The focal option was manipulated by asking respondents

how much more or less attractive one of the two (e.g., fruit salad) was. When

descriptions of alternatives' features were provided rather than retrieved from mem-

ory, a manipulation of the focal option had a weaker and less consistent effect on

preferences. Think-aloud protocols were used to gain insights into the effect of changing

the focal option on decision processes. The implications of the results for marketers'

communications strategies are discussed.

The Effect of the Focus of Comparison onConsumer Preferences

In classical preference theory, each consumer is as-sumed to have a well-defined preference order or utilityfunction, such that the consumer selects from any choiceset the altemative that offers the highest utility. This as-sumption implies that normatively equivalent proceduresfor assessing preferences should lead to the same pref-erence order (Tversky, Sattath, and Slovic 1988). Re-cent research, however, suggests that preferences are oftensensitive to the particular task and context characteristics(see Payne, Bettman, and Johnson 1992 for a review).

We examine the effect on preferences of shifting thefocus of attention to one of two considered options. Spe-cifically, building on the notions of selective attentionand loss aversion, we propose that the mere fact that analtemative is the focus of a comparison (referred to hereas the focal option) can make it appear more attractiveand enhance its choice probability (e.g.. Dunning andParpal 1989; Houston, Sherman, and Baker 1989; Tver-sky and Kahneman 1991). As an example, consider awoman in a restaurant who is debating whether to order

*Ravi Dhar is Assistant Professor, School of Organization andManagement, Yale University. Itamar Simonson is Assistant Profes-sor, Haas School of Business, University of California, Berkeley.

The authors are grateful to Jim Bettman, Danny Kahneman, AmosTversky, Jonathan Baron, and three anonymous JMR reviewers fortheir comments and suggestions.

frozen yogurt or a fruit salad for dessert. We proposethat her likelihood of preferring frozen yogurt to fruitsalad can be increased by asking her, before she makesthe decision, to think how much more or less attractivefrozen yogurt is to her. As the example illustrates, wetest our proposition by using a simple wording manip-ulation that determines which altemative is the focal op-tion. Such an effect might have significant implicationsfor marketers' communications strategies, suggesting thatthe choice probability of a promoted product can be en-hanced by making that product the focal option in com-parisons with other altematives.

We further propose that manipulations of the focal op-tion can influence preferences if consumers do not havepre-formed preferences between the altematives and needto retrieve information about them from memory. Con-versely, when altematives' descriptions are provided (e.g.,in a sales catalog), a manipulation of the focal option isnot expected to influence the manner in which altema-tives are compared and consumers' preferences.

A review of prior research relevant to the effect of thefocus of comparison on preferences led to several hy-potheses, which we tested in five studies. Using think-aloud protocols, we also examined the decision pro-cesses underlying the effect of the manipulation of thefocal option on preferences. We report the findings andconclude with a discussion of their implications for mar-keters.

430

Journal of Marketing ResearchVol. XXIX (November 1992), 430-40

EFFECT OF THE FOCUS OF COMPARISON O N CONSUMER PREFERENCES 431

INFLUENCES OF THE FOCUS OF COMPARISONON PREFERENCES

Interest in the impact on judgment and decision mak-ing of the manner in which altematives are comparedstarted in the context of similarity judgments and hasonly recently shifted to preference. Tversky (1977) pro-posed a model for the comparison process involved injudgments of similarity. This model is based on a fea-ture-matching process in which similarity judgments areassumed to be the result of a linear combination of theshared and distinctive features of the stimuli being com-pared. As Tversky noted, the model suggests that thedirection of compaiison of stimuli can influence their de-gree of perceived similarity. That is, when asked to as-sess the degree of similarity of a to b, one's focus ofattention is on the unique features of a, whereas whenassessing the similarity between b and a, one's focus ison b. That asymmetry leads, for example, to a greaterperceived similarity of North Korea to China than of Chinato North Korea, because people are aware of more uniquefeatures of China than of North Korea. Thus, task in-structions that infiuence the direction of comparison canaffect the perceived similarity between objects (e.g.,Johnson 1981). Tversky's work illustrates a generalprinciple, already observed by William James, that hu-man thought tends to be selective and pays more atten-tion to one part of its object than to another (e.g., oneof two compared altematives). This selectivity takes dif-ferent forms and has been observed in such diverse areasas perception, memory, and language (see Stone 1991for a review).

Houston, Sherman, and Baker (1989), building onTversky's research and work by Agostinelli et al. (1986),extended these findings to the domain of preference. Theyargued that, given the previous findings about the uniquerole of features in judgments of similarity, features thatare unique to the focal option should also exert a greaterinfluence on preferences. That is, when a person is com-paring one altemative with another, the relative prefer-ence for the focal option depends on whether the focaloption has unique good features or unique bad features.In their experiments, Houston and his coauthors as-sumed that, if altematives are presented sequentially, thesecond altemative will be the focal option unless re-spondents know in advance (before evaluating any al-temative) about the forthcoming preference task. Eachalternative in their experiments consisted of a list of ninepositive and nine negative features. Their results indi-cated that most subjects preferred the altemative pre-sented last when the two altematives had common badfeatures and unique good features, whereas the first al-temative was most preferred when the altematives hadunique bad features and common good features. Thus,people appear to focus on the unique features of the focaloption in determining the attractiveness of that option inrelation to another altemative.

In the Houston study, the unique and common features

of each altemative were manipulated experimentally, butin many choice situations it is difficult to sort preciselywhich features of altematives are common or unique.For example, in the choice between frozen yogurt andfruit salad for dessert, identifying the common and uniquefeatures is nontrivial and consumers are unlikely to en-gage in such analysis. However, even without such sort-ing of the unique features of altematives, the mere factthat an altemative is the focus of attention can often en-hance its perceived attractiveness. Because most alter-natives that consumers consider have more positive thannegative features (Howard 1977), the altemative that isthe focus of attention is likely to appear more attractive,particularly if the features of the altematives must be re-trieved from memory and are not extemally available.For example, in a choice between vacations in Rome andMaui, a consumer who focuses on Rome is likely to re-trieve its attractive features while paying less attentionto the advantages of Maui.

A second mechanism that might make the focal optionappear more attractive is related to the notion of lossaversion. Loss aversion indicates that losses loom largerthan corresponding gains, where both losses and gainsare assessed in relation to the same reference point(Tversky and Kahneman 1991). Loss aversion has beenobserved in both risky and riskless choice and can ac-count for a wide range of decision phenomena. The ref-erence point in relation to which the losses and gains aredefined may be the status quo option (Samuelson andZeckhauser 1988), the current endowment of the con-sumer (Kahneman, Knetsch, and Thaler 1991), or thepoint through which altematives are evaluated (Tverskyand Kahneman 1991).

In a comparison between two options, the referencepoint might be detennined by making one altemative thefocus of the comparison. For example, asking a con-sumer who is debating between frozen yogurt and fruitsalad to consider "How much more or less attractive isfrozen yogurt?" is likely to make that option the focusof the comparison. When the focal option is the refer-ence point, its advantages represent potential losses if itis not chosen, whereas its disadvantages represent po-tential gains. In other words, if the other (nonfocal) op-tion is selected, the consumer is losing the advantagesof the focal option but gaining its disadvantages. Be-cause losses loom larger than the corresponding gains,the focal option tends to be perceived as more attractiveand subsequently has a higher choice probability.

The preceding discussion is based on the assumptionthat the focal option serves as the reference altemativeand its features receive more attention in the preferenceformation process. However, this assumption may nothold if the descriptions of altematives are provided andare extemally available rather than retrieved from mem-ory when preferences are formed (Alba, Hutchinson, andLynch 1991: Biehal and Chakravarti 1986). That is, whenthe altematives' descriptions are extemally available,consumers are less likely to focus selectively on the fea-

432 JOURNAL OF AAARKETING RESEARCH, NOVEMBER 1992

tures of one of the options. For example, when evalu-ating the relative attractiveness of two altematives de-scribed in a sales catalog, consumers are likely to contrastthe desirability of the listed features of these altematives.In that case, phrasing the evaluation task with an em-phasis on one of the altematives ("How much more orless attractive is A?") is not expected to have much im-pact on the comparison process. We therefore predictthat changes of the focal option in comparisons betweenaltematives with extemally available descriptions will notinfiuence the relative preferences for these altematives.Our discussion leads to the following hypotheses.

Hia: In a judgtnent task involving a cotnparison betweentwo options about which consumers have informa-tion in memory, focusing attention on one altema-tive (the focal option) tends to enhance its attrac-tiveness.

Hit: An altemative that serves as the focal option in acomparison (as in H|a) subsequently has a higherchoice probability than it would have if the otheraltemative were the focal option.

H2: If the descriptions of altematives are provided andare extemally available when preferences are formed,being the focal option does not enhance the per-ceived attractiveness and choice probability of al-tematives.

STUDIES 1 AND 2

Method

Studies 1 and 2, designed to test H|a and H,b, weresimilar in design and choice problems. The problems usedin study 1 were replicated in study 2, in some cases withslight wording changes. Study 2 also included a choiceproblem involving desserts that was not used in study 1.Because of the similarities between the two studies, wedescribe them together.

The subjects in the two studies (combined) were 305undergraduate marketing (about 75%) and psychologystudents at the University of Califomia, Berkeley. Par-

ticipation was a course requirement. The respondents wereinformed that they would be presented alternatives indifferent categories and asked to indicate their prefer-ences. It was emphasized that there were no right or wronganswers and the researchers were interested only in thepersonal preferences of the participants. Then the scalethat was later used for measuring the relative attractive-ness of the altematives was explained briefly.

Each choice problem presented two options that wereidentified by their names, and respondents were askedto assume they had to choose between the two options.One problem involved a choice between the graduatebusiness schools of Harvard and Stanford (see Figure 1).A second problem, titled "Sweepstakes," pertained tothe selection of a vacation spot: "Assume that you haveentered a sweepstakes contest which has as its first prizea one week vacation. The vacation spot offered as thefirst prize is either Rome (Italy) or Maui (Hawaii). Inboth cases all expenses will be paid." A third problem(included only in study 2) involved the selection of adessert: "Assume that you and your friend have just haddinner and are thinking of having some dessert. The waitersuggests two alternatives: frozen yogurt and fresh fmitsalad (the cost and calorie content of the two dessertsare comparable)." A fourth problem involved a choiceof a restaurant: "Assume that you are invited for dinneron your birthday by a friend. You are asked to choosebetween eating at an Italian or a French restaurant (thecost and quality of the food are comparable for the tworestaurants)." In study 2, the options were Italian andChinese restaurants, with similar results.

Subjects were expected to have information about thealternatives in these problems in memory, but probablyno pre-formed preferences between them. After readingeach problem, subjects were asked the following ques-tion: "On the scale below, please indicate how much moreor less attractive to you is [the focal option]? (circle theappropriate number)." Following Dunning and Parpal(1989), we used a 19-point scale from —9 to +9. Above

Figure 1EXAMPLE OF A FOCAL OPTION MANIPULATION

Graduate SchoolAssume that you are planning to get an MBA. You have applied to the MBA programs at both the Harvard Business School

and the Stanford Business School.1. On the scale below, please indicate how much more or less attractive to you is an MBA at the Harvard Business School?

(Circle the appropriate number).Harvard Less Attractive Harvard More Attractive

Much Less Slightly Less Slightly More Much More- 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2 - 1 0 1 2 3 4 5 6 7 8 9

2. How much more or less do you prefer to do an MBA at the Harvard Business School?Harvard Less Preferred Harvard More Preferred

Much Less Slightly Less Slightly More Much More- 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2 - 1 0 1 2 3 4 5 6 7 8 9

3. Assume that you are admitted to both schools, which would you choose? (Check One)Harvard Stanford

EFFECT OF THE FOCUS OF COMPARISON O N CONSUMER PREFERENCES 433

the left/right side of the scale (from - 9 to - 1 or from1 to 9), the heading was "[focal option] less/more at-tractive," respectively. The identity of the focal optionwas manipulated between subjects, such that each optionwas the focus of the comparison in one version. The nextitem to which subjects responded was, "How much moreor less do you prefer the [focal option]?" A scale similarto the attractiveness measure was used, with the head-ings "[focal option]" less/more preferred" above the scale.Finally, the two options were listed (in the same orderin both conditions) and subjects were asked to indicatethe one they would choose.

Results

The results are summarized in Table 1. Consistent withH|a, in the graduate school problem, subjects tended to

rate higher the attractiveness and their preference for theschool that served as the focal option. In accordance withthese ratings and H|b, each school had a 20% it = 2.7,p < .05) greater share when it was the focal option thanwhen the other school was the focal option. A majorityof subjects who received the Harvard frame preferredHarvard, whereas a small majority of those with theStanford focus preferred the Stanford school.

In the sweepstakes problem, the manipulation of thefocal option again influenced the ratings and choices (16%share increase, t = 2.3, p < .05) as predicted by H|aand Hib. Similarly, in the dessert problem, the focal op-tion manipulation had a statistically significant effect onboth the ratings and subsequent choices as hypothesized(27% share increase, t = 3.4, p < .05). Finally, in therestaurant problem, the effect of the focal option manip-

Table 1STUDIES 1 AND 2. EFFECT OF FOCAL OPTION AAANIPULATION O N PREFERENCES IN MEMORY-BASED COMPARISONS

(standard errors in parentheses)

Problem 1: graduate school

Averageattractiveness

Averagepreference

Choice

Harvard is reference(n = 86)

3.3"(.51)3.0"(.54)

67%*33%

Harvard more

Harvard more

HarvardStanford

Problem 2: sweepstakes

Stanford is reference(n = 85)

1.3(.60)1.10(.62)

47%53%

Stanford more

Stanford more

HarvardStanford

Averageattractiveness

Averagepreference

Choice

Rome is reference(n = 86)

4.5"(.52)4.2"(.71)

78%"22%

Rome more

Rome more

RomeMaui

Problem 3: dessert

Maui is reference(n = 85)

.5(.69).3

(.62)62%

Rome more

Rome more

RomeMaui

Averageattractiveness

Averagepreference

Choice

Yogurt is reference(n = 66)

.2'(.69).1"

(.68)52%"

Yogurt more

Yogurt more

YogurtSalad

Problem 4: restaurant

Salad is reference(n = 68)

2.9(.58)2.9(.61)

25%75%

Salad more

Salad more

YogurtSalad

Averageattractiveness

Averagepreference

Choice

Italian is focal(n = 86)

3.4"(.48)3.7"(.60)

72%"

Italian more

Italian more

ItalianFrench

French is focal(n = 85)

.8(.52).9

(.54)61%39%

Italian more

Italian more

ItalianFrench

"The difference between conditions is statistically significant at the .05 level.'The difference between conditions is statistically significant at the .10 level.

434 JOURNAL OF AAARKETING RESEARCH, NOVEMBER 1992

ulation on the ratings, as predicted by H,a, was statis-tically significant and the effect on choice probability,11%, was marginally significant (r = 1.5, p < .10).

In study 1, we manipulated the focal option by askingsubjects, "How much more or less attractive is the [focaloption]?" Possibly the fact that the word "more" ap-peared before "less" created the impression that the focaloption altemative was expected to be more rather thanless attractive. To test this explanation, we replicated instudy 2 the business school and sweepstakes problems,using the wording "How much less or more attractive. . . " and "How much less or more do you prefer. . . . "For the business school problem, the difference in rat-ings posited in H,a was smaller, but still statistically sig-nificant (r = 2.8, /? < .01 for attractiveness; t = 1.9, p< .05 for preference). Similarly, the difference in choiceprobabilities was smaller and marginally significant (t =1.3, / ? < .10). In the sweepstakes problem, the resultswere almost identical to those of study 1 (t = 3.5, p <.01 for attractiveness; t = 2.5, p < .01 for preference;t = 1.8, p < .05 for choice).

Discussion

Studies 1 and 2 demonstrated that a manipulation ofthe altemative used as the focal option can have a sys-tematic effect on the evaluations and choice probabilitiesof altematives. Specifically, respondents tended to judgefocal altematives more favorably and were subsequentlymore likely to select them. All of the problems includedin studies 1 and 2 involved altematives about which re-spondents had information in memory but which wereprobably not previously compared. Study 3 examined theeffect of the focal option manipulation when descriptionsof altematives were provided rather than retrieved frommemory.

STUDY 3

Study 3 tested H2 and also examined the effect of thefocal option manipulation on recall of the altematives'features. The focal option manipulation was predicted tolead to better recall of the features of the focal option,but that effect was expected to be small. When evalu-ating the attractiveness of the focal option in relation tothe nonfocal option, respondents were expected to elab-orate more on the features of the former and later to havebetter recall of these features. In contrast, because re-spondents received written descriptions of the altema-tives and rehearsed their features in anticipation of a re-call test, the differences in actual recall between the focaland nonfocal options were expected to be small.

Method

Procedure. The subjects were 151 undergraduate stu-dents enrolled in a marketing class. Participation in theexperiment was part of a course requirement. There weretwo differences between the task in study 3 and that inthe previous studies. First, the two altematives in each

problem were described in the questionnaire. With theexception of one category in which the described optionshad familiar brand names, the altematives were referredto as brand AAA and brand BBB. A second differencein study 3 was that in the first three categories (35mmcamera, cassette player, and answering machine), in ad-dition to rating the relative attractiveness and making achoice, respondents performed a recall task.

In these three categories, subjects sequentially re-ceived the descriptions (including pictures) of altema-tives, each on a separate handout, and were told thatlater they would be asked questions about these alter-natives. For each altemative, subjects were given twominutes to read the attribute information. They then in-dicated the relative attractiveness of and their preferencefor the focal option, using the same manipulation andquestions as in studies 1 and 2 (e.g., "How much moreor less attractive is altemative BBB?"), followed by achoice between the two options. Next, subjects were givena recall task and asked to list as many features as theycould remember for each of the two altematives. Withineach focal option condition, the order in which alter-natives were listed in the recall part was manipulatedbetween subjects. This manipulation was included to ac-count for the possibility that the recall task for one al-temative would inhibit recall for the second.

After completing the task for the first three productcategories, subjects evaluated two altematives in fouradditional categories to test Hj without a recall task. Inthese categories, the descriptions of the altematives werecontained in the questionnaire (rather than on separatehandouts), such that subjects could refer back to theproduct information when assessing the relative attrac-tiveness of altematives.

Stimuli. As indicated, the features and pictures (takenfrom Consumer Reports and a sales catalog) of the al-tematives included in study 3 were presented in the ques-tionnaire (or, for the first three problems, on handouts).Each altemative was described in terms of five or sixfeatures and subjects were asked to evaluate the attrac-tiveness of altematives (with the same manipulation asin studies 1 and 2) and make choices. Figure 2 showsthe altematives in one category (without the pictures).

Results

As can be seen in Table 2, contrary to H2, the focaloption manipulation did infiuence the attractiveness andpreference ratings in favor of the focal option. This ef-fect, however, tended to be weaker than that found instudies 1 and 2. In terms of choice, consistent with H2,the sheu'es of the focal options increased in only three ofthe seven categories. Even in these three categories, theshare increases associated with the focal option manip-ulation were smaller than those found in studies 1and 2.

To examine the impact of the focal option manipula-tion on recall of features, we conducted an analysis ofvariance with each altemative providing one observa-

EFFECT OF THE FOCUS OF COMPARISON O N CONSUMER PREFERENCES 435

Figure 2

STUDY 3: EXAMPLE OF PROVIDED DESCRIPTIONS

BRAND "AAA"VIDEO CASSETTE RECORDER WITH REMOTE

[picture of brand AAA]Brand AAA Price: $259.00

The Brand "AAA" video cassette recorder has the following features:On Screen Display: provides detailed instructions for programming on the screen itself;4-event/l month programmable timer: makes for easy recording for any day of the month;Special effects: includes freeze frame for great still shots;Variable speed slow motion: allows for greater enjoyment of sports programs.

BRAND "BBB"VIDEO CASSETTE RECORDER WITH REMOTE

[picture of brand BBB]Brand BBB Price: $209.00

The Brand "BBB" video cassette recorder has the following features:Digital Visual Tracking: automatically adjusts for optimum picture quality;Auto Eject: ejects tape at end of play and switches off power supply;Auto Index: marks tape, making it easy to find the start of each recording during rewind or fast forward;2-week timer: allows you to reserve programs for recording two weeks in advance.

Table 2STUDY 3: EFFECT OF FOCAL OPTION AAANIPULATION

O N PREFERENCES WHEN ALTERNATIVE DESCRIPTIONSARE PROVIDED

(standard errors in parentheses)

Category

MemoryCamera

Cassetteplayer

Answeringmachine

NonmemoryPC printer

Typewriter

VCR

Color TV

Effect onattractiveness

rating'

3,42"(.59)1,43"(.64)2.18"(.59)

2.12"(.64)2,17"(,67),90

(.73)2.33"(,66)

Effect onpreference

rating''

3,48"(,62)1,43"(,70)1.89"(.65)

1,79"(.67)2,08"(.75),71

(.82)2,08"(.71)

Effeci onchoice'

11%"(,066)

- 7 %(,076)7%(,076)

- 2 %(,080)

10%(.075)

- 1 %(,079)

0%(.075)

'Effect on attractiveness rating is the difference in average ratingsbetween the two conditions (with sign reversal in one condition). Forexample, if the average rating is 3 when A is focal and - 1 when Bis focal, the effect size is 2.

""Effect on preference rating is the difference in average ratings be-tween the two conditions.

""Effect on choice is the increase in choice share when the aitemativeis the focal option.

"Statistically significant at the ,05 level.

tion. The dependent measure was the number of featurescorrectly recalled for that aitemative. A listed attributewas coded as correctly recalled if it was included, as is,in the description of the aitemative. The independentvariables included four 0-1 dummy variables. One vari-able received a value of 1 if the aitemative was the focaloption. A second variable received a value of 1 if theaitemative was listed first in the recall task. A third vari-able received a value of 1 if the aitemative was the onechosen. This variable was included on the basis of priorresearch suggesting that jjeople have better recall for thealtematives they choose (Biehal and Chakravarti 1982;Johnson and Russo 1981). The fourth variable receiveda value of 1 if the aitemative was evaluated first, to ac-count for a possible effect on recall of the order in whichaltematives' descriptions were considered.

The effect of being the focal option on the number ofcorrectly recalled features was positive and statisticallysignificant (F = 4.1, p < .05). On average, there wasan increase of .2 correctly recalled attributes when anoption was the focal option. Respondents also had betterrecall for the altematives they chose (F = 6.6, p < .05).The effects of the order of the options in the recall taskand the order in which altematives were evaluated werenot significant (p > .5 andp > .2, respectively). A sim-ilar analysis with the total number of features recalledrather than just those recalled correctly indicated that noneof the effects were significant. This result might refiectthe tendency of subjects to list the same number of fea-tures for both altematives considered, even when theyhad difficulty remembering those features.

436 JOURNAL OF AAARKETING RESEARCH, NOVEMBER 1992

Discussion

The results of study 3 indicate that, even when alter-natives' descriptions are provided, the perceived attrac-tiveness of the focal altemative is enhanced. The resultsalso indicate that the focal options are associated withbetter recall, though that difference is rather small. Ap-parently, the manipulation causes respondents to elabo-rate more on the focal option even when the descriptionsof both alternatives are provided. However, this effectis weaker than that found in studies 1 and 2, where al-ternatives' features were retrieved from memory, and thereis no consistent effect on subsequent choices. ,

One limitation of these results is that the product cat-egories included in study 3 were different from those instudies 1 and 2. Consequently, the observed differencesbetween these studies might be explained by the choiceof product categories rather than any inherent differencesin the evaluation process. Study 4 was designed to ex-amine this altemative explanation by contrasting mem-ory-based and provided-description-based (or stimuli-based) decisions, with the same product categories acrossconditions.

STUDY 4

Method

The subjects were 132 undergraduate marketing stu-dents. Participation was part of a course requirement.Each respondent was asked to rate altematives and makechoices in two categories, graduate schools and vacationspots. One version of each problem was similar to thatused in study 1, involving two altematives that wereidentified by their names (Harvard or Stanford and Romeor Maui, respectively). In the other version, two alter-natives (altematives A and B) were described. The pro-vided attribute descriptions of each altemative were de-rived from a pilot study (designed primarily for a differentpurpose) in which 260 undergraduate marketing studentswere asked to list the advantages and disadvantages ofthe specific school and vacation spot altematives (e.g.,of spending a one-week vacation in Rome vs. Maui).The most common responses from that study were usedin constructing the descriptions of the altematives for study4. For example, Rome was described as "a city in West-em Europe, historical monuments, ancient architecture,good shopping, good ethnic cuisine." Finally, as in theprevious studies, subjects entered relative attractivenessand preference ratings and then made a choice.

Results

As can be seen in Table 3, the results of study 1 werereplicated in the problem versions that included the namesof the schools and vacation spots. Consistently, the focaloption manipulation shifted the ratings and choices infavor of the focal option, though because of the smallsample size, the effect in the sweepstakes problem wasnot statistically significant. In contrast, when altema-tives' descriptions were provided, the effect of the focal

option manipulation on ratings and subsequent choiceswas consistently weaker.

Thus, study 4 replicated the results of the previousstudies with the same product categories across task con-ditions. These findings are consistent with the assump-tion that the focal option manipulation influences deci-sion processes differently in the two types of problems.Specifically, as in our previous discussion, the focal op-tion is assumed to receive more attention if altemativeevaluation is memory-based. Conversely, when alter-natives' descriptions are extemally available, the effectof the focal option manipulation on decision processesis weaker, consistent with the results of studies 3 and 4.

STUDY 5

The effect of the focal option manipulation on deci-sion processes was investigated by means of think-aloudprotocols. Respondents were expected to focus their at-tention on the focal option when forming memory-basedjudgments. This prediction can be tested by counting thenumber of thoughts on the focal and nonfocal options.In addition, the mechanisms that were hypothesized tounderlie the effect of the focal option manipulation, aswell as the types of altematives used in our research,suggest that respondents would be more likely to con-sider positive than negative features of altematives. Forexample, respondents would be more likely to talk aboutthe quality of education they would receive at Harvardor Stanford than to talk about the cost of that education.

Method

The subjects were 37 undergraduate students enrolledin a marketing class. Participation was part of a courserequirement. Each subject first received detailed instmc-tions on the task and the importance of thinking aloudwhile responding to problems. It was emphasized thatthere were no right or wrong answers, and that the re-searchers were interested only in the preferences andthoughts of the participants. Before starting the actualtask, subjects were given one problem to practice think-ing aloud while being recorded. As in the previous stud-ies, the focal option was manipulated between subjectsin the two versions of the questionnaire.

The task included five choice sets used in prior stud-ies, including three problems in which altematives wereidentified by name—the graduate school, sweepstakes(vacation spot), and restaurant (Chinese vs. Italian)problems—and two problems involving a cassette playerand a typewriter with altematives that were described inthe questionnaire. By using different product categoriesfor the five problems, we were able to obtain from eachsubject protocols of both memory-based and provided-description-based choices.

The protocols were analyzed by two independentjudges. Specifically, the judges were instructed to countthe number of positive and negative thoughts about thetwo altematives in each of the five problems. The in-terjudge reliability was 87%. Disagreements were re-

EFFECT OF THE FOCUS OF COMPARISON O N CONSUMER PREFERENCES 437

Table 3STUDY 4: FOCAL OPTION MANIPULATION—MEMORY-BASED VERSUS PROVIDED DESCRIPTIONS

(standard errors in parentheses)

Problem: graduate school

Memory-based descriptions

Averageattractiveness

Averagepreference

Choice

Averageattractiveness

Averagepreference

Choice

Harvard is reference(n = 28)

Stanford is reference(n = 28)

3.5"(.74)3.5"(.86)

66%"34%

School A(n

Harvard more

Harvard more

HarvardStanford

is reference= 38)

.2(.82).7

(.78)

54%

Stanford more

Stanford more

HarvardStanford

Provided descriptions

School B is reference(n = 38)

1.6"(.89).9

(.92)57%43%

School A more

School A more

School ASchool B

Problem: sweepstakes (vacations)

.5(.95).6

(.95)47%53%

School B more

School B more

School ASchool B

Averageattractiveness

Averagepreference

Choice

Averageattractiveness

Averagepreference

Choice

Memory-based descriptionsRome is reference

(n = 28)Maui is reference

(n = 28)3.9

(1.00)4.0(.95)

79%21%

Vacation A(n =

Rome more

Rome more

RomeMaui

is reference-• 38)

Provided descriptions

2.5(1.05)2.6(.85)

69%31%

Vacation B(n =

Rome more

Rome more

RomeMaui

is reference38)

1.2(85)1.2(90)

41%

Vacation A more

Vacation A more

Vacation AVacation B

.6(.80).7

(.85)59%41%

Vacation A more

Vacation A more

Vacation AVacation B

°The difference between (focal option) conditions is statistically significant at the .05 level."The difference between (focal option) conditions is statistically significant at the .10 level.

solved by discussion. To test the predicted effect of thefocal option manipulation on decision processes, wecompared the number of thoughts on the focal optionwith the number on the other altemative. We also ex-amined the number of positive and negative thoughts oneach altemative.

Results

The effect of the focal option manipulation on the rat-ings and choices replicated the results of the previousstudies. In the graduate school, sweepstakes, and res-taurant problems, there was a statistically significant shiftin the attractiveness and preference ratings in favor of

the focal option (p < .05 for all). The increase in choiceprobability due to the focal option manipulation was 26%in the graduate school problem, 10% in the sweepstakesproblem, and 25% in the restaurant problem.

In the problems with provided descriptions, the focaloption manipulation had a statistically significant effecton the ratings and increased the choice probability by7% in the cassette player category. In the typewriter cat-egory, the direction of the effect on the ratings was re-versed (not statistically significant) and there was nochange in the choice probabilities of the two altematives.In sum, similar to the previous findings, the results showa rather consistent effect of the focal option manipulation

438 JOURNAL OF AAARKETING RESEARCH, NOVEMBER 1992

Table 4STUDY 5 (PROTOCOLS): EFFECT OF FOCAL OPTION AAANIPULATION O N NUMBER OF THOUGHTS

FOR AND AGAINST ALTERNATIVES

(standard errors in parentheses)

Problem type

Memory-basedchoice sets

Fully describedchoice sets

Average number of thoughtson focal option

Positive

.85'(.11)1.49"(.15)

Negative

.31(.05).71

(.13)

Total

1.17"(.11)2.20(.17)

Average number of thoughtson nonfocal option

Positive

.50"(.07)1.33"(.16)

Negative

.18(.04).56

(.09)

Total

.68(.08)1.89(.19)

'The difference between average number of positive and negative thoughts is statistically significant at the .01 level."The difference between average total number of thoughts on the focal and nonfocal options is statistically significant at the .01 level.

on ratings and subsequent choices when judgments werememory based, but a weak and inconsistent effect whenthe descriptions of altematives were provided.

Table 4 reports the average number of positive andnegative thoughts on the focal and nonfocal alternativeswithin the two types of problems. In the graduate school,sweepstakes, and restaurant problems combined, the av-erage numbers of positive, negative, and total thoughtsabout the focal altemative were significantly greater thanthe average numbers of such thoughts about the nonfocalaltemative (t = 2.3, / = 2.2, and t = 3.6, respectively,p < .05 for all). This finding is consistent with the as-sumption that the focal option manipulation causes re-spondents to focus their thoughts on the focal altema-tive. Particularly noteworthy is the finding that, despitethe tendency to prefer the focal option, respondents ex-pressed more negative thoughts about that altemative thanabout the nonfocal one. In addition, as expected, morepositive than negative thoughts were expressed about boththe focal and nonfocal altematives (for both, p < .01).

These results indicate that, when altematives wereevaluated from memory, the focal option received moreattention and positive features were more salient. In con-trast, in categories with provided descriptions (cassetteplayer and typewriter), there was on average only a smallincrease in the number of positive, negative, and totalthoughts when an altemative was the focal option, andthis effect was not statistically significant (p > .10).

GENERAL DISCUSSION

Consumers usually are assumed to have well-definedattitudes and preferences for altematives offered to them.A further assumption is that, to change consumer pref-erences, marketers should employ various means of per-suasion (e.g., Fishbein and Ajzen 1975; Petty, Ca-cioppo, and Schumann 1983). Recent research, however,indicates that consumers' preferences are often fuzzy anduncertain (e.g., March 1978; Payne, Bettman, and John-son 1992), making them susceptible to various other in-fluences. We examined effects on consumer preferencesthat are produced by influencing the manner in whichaltematives are compared. Specifically, the results in-

dicate that shifting the focus of comparison to an alter-native can enhance that altemative's perceived attrac-tiveness. This effect was obtained for altematives aboutwhich respondents had information stored in memory butprobably had no pre-formed preferences between them.

We also examined situations in which descriptions ofalternatives are provided (e.g., in a sales catalog) andare extemally available when relative preferences areformed. In these cases, the focal option tended to receivehigher attractiveness ratings, but we found no consistenteffect on subsequent choices (see also Sanbonmatsu,Kardes, and Gibson 1991). This result suggests that, evenwhen alternative descriptions are provided, the focal op-tion manipulation can lead consumers to pay more at-tention to the focal option, though the magnitude of thateffect was small.

Theoretical Implications

Our research demonstrates that preferences betweenaltematives can be influenced systematically by using asimple wording manipulation that causes respondents touse one option as the focus of comparison. These fmd-ings extend the growing literature on the effect of thedirection of comparison on preferences (e.g.. Dunningand Parpal 1989; Houston, Sherman, and Baker 1989;Sanbonmatsu, Kardes, and Gibson 1991). Two mecha-nisms were hypothesized to account for the effect of thefocal option manipulation. First, a tendency to focus onthe features of the focal option can enhance its attrac-tiveness and subsequent choice probability if the optionhas primarily positive features. A second explanation isbased on the principle of loss aversion and the assump-tion that the focal option serves as the reference alter-native and thus appears more attractive. Unfortunately,there are no process measures that can "prove" the effectof loss aversion on preferences in any particular case.The power of loss aversion as an explanation is derivedfrom its ability to account for a wide variety of decisionphenomena that are inconsistent with standard theoriesof choice (Tversky and Kahneman 1991).

Both the attention focus and loss aversion mechanismswere likely to be operating and contributing to the ob-

EFFECT OF THE FOCUS OF COMPARISON O N CONSUMER PREFERENCES 439

served results, though separating the two may be diffi-cult. The two explanations could be disentangled by us-ing choice problems in which the altematives' negativefeatures are more salient than their positive features. Inthat case, the attention focus explanation leads to theprediction that the focal option would appear less at-tractive, because its negative features are the focus ofattention. According to loss aversion, in contrast, thepositive features of the reference option (which are po-tential losses if that option is not selected) would tendto loom larger than its negative features. Thus, accord-ing to the loss aversion explanation, an option whosenegative features are more salient than its positive fea-tures would still tend to appear more attractive when itis the focal option than when the other option is the focaloption.

We tested several problems in which the negative fea-tures were expected to be more salient. In one problem,subjects evaluated two flights, one nonstop but more ex-pensive and the other requiring two stops but less ex-pensive. In another problem, the choice was between twoblood pressure medications with different side effects (mildheadache vs. mild indigestion). A third choice was be-tween an apartment requiring a longer commute and anapartment that was in worse condition. We tested theseproblems twice; in the first study the task wording was"How much more or less attractive is . . .?" and in thesecond it was "How much more or less unattractive is. . . .?" However, we did not find a consistent pattem ofresults, and differences were not statistically significant.The fact that the focal option manipulation did not havea significant effect in these tests might reflect the con-flicting effects of loss aversion and attention focus onpreferences. Future research should further investigatethe impact of changing the focal option when alterna-tives' negative features are more salient. Finding appro-priate altematives for such tests may be difficult, be-cause merely counting the number of positive and negativefeatures is not sufficient and because of the difficulty ofassessing a priori the degree of loss aversion on variousattributes.

Marketing Consequences

Our findings suggest that marketers can affect con-sumer preferences by influencing the manner in whichthe product they promote is compared with competingaltematives. In particular, when designing comparativeads, shelf talkers, and personal communications, mar-keters could encourage consumers to consider the ad-vantages and disadvantages of their product in relationto a competitor (rather than the competitor's dis/advan-tages in relation to their product). Furthermore, the re-sults suggest that it might be more effective to use apromoted brand as the focal option in a comparison withother brands (e.g., using comparative advertising) thanto feature the promoted brand exclusively. That is, con-sumers may feel more confident if brand evaluations arebased on a comparison with other altematives (assuming

the comparison is perceived as credible) than if a brandis evaluated in isolation.

A limitation of our research is that it was conductedin the laboratory and employed questions that subjectsin an experiment may expect to answer but consumerswould not. Marketers therefore should find an appro-priate way to translate the laboratory instructions intoevaluation guidelines that focus attention on the alter-native they are trying to promote. For example, a sales-person may try to focus a customer's attention andevaluation process on the relative advantages and dis-advantages of the brand he or she wants to sell ratherthan on those of the competitor.

The finding that an option that is the focus of a com-parison tends to appear more attractive and to have ahigher choice probability provides another reason to pro-mote brand awareness and recall. Specifically, whenpurchase decisions are being made, brands associated withgreater awareness and better recall of features are morelikely to be the focus of comparisons with other alter-natives and thus appear more attractive. Finally, mar-keters should also take the focal option effect into con-sideration when presenting various altematives within theirproduct line. For example, the findings suggest that if acar with options (e.g., cruise control) is the standardpackage (e.g., "How much more attractive to you is thecar with the options?"), a consumer is likely to perceivegreater benefits associated with the added options and tobe more likely to prefer the car with the options than heor she would if the car without the options were the stan-dard.

REFERENCES

Agostinelli, Gina, Steven J. Sherman, Russell H. Fazio, andEliot S. Hearst (1986), "Detecting and Identifying Change:Additions Versus Deletions," Journal of Experimental Psy-chology: Human Perception and Performance, 12 (4), 445—54.

Alba, Joseph W., J. Wesley Hutchinson, and John G. Lynch,Jr. (1991), "Memory and Decision Making," in Handbookof Consumer Behavior, Thomas Robertson and Harold Kas-sarjian, eds. New York: Prentice-Hall, Inc., 1-49.

Biehal, Gabriel and Dipankar Chakravarti (1982), "Informa-tion-Presentation Format and Learning Goals as Determi-nants of Consumers Memory Retrieval and Choice Processes,"Journal of Consumer Research, 8 (March), 431—41.

and (1986), "Consumers' Use of Memory andExtemal Information in Choice: Macro and Micro Perspec-tives," Journal of Consumer Research, 12 (March), 382-405.

Dunning, David and Mary Parpal (1989), "Mental AdditionVersus Subtraction in Counterfactual Reasoning: On As-sessing the Impact of Personal Actions and Life Events,"Journal of Personality and Social Psychology, 57 (I), 5—15.

Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, In-tention and Behavior: An Introduction to Theory and Re-search. Reading, MA: Addison-Wesley Publishing Com-pany.

Houston, David A., Steven Sherman, and Sara M. Baker (1989),

440 JOURNAL OF MARKETING RESEARCH, NOVEMBER 1992

"The Influence of Unique Features and Direction of Com-parison on Preferences," Journal of Experimental SocialPsychology, 25 (3), 121-41.

Howard, John A. (1977), Consumer Behavior: Application ofTheory. New York: McGraw-Hill Book Company.

Johnson, Eric J. and J. Edward Russo (1981), "Product Fa-miliarity and Learning New Information," Advances in Con-sumer Research, Vol. 8, Kent B. Monroe, ed. Ann Arbor,MI: Association for Consumer Research.

Johnson, Michael (1981), "Context Effects in Product Per-ception," in Advances in Consumer Research, Vol. 8, KentB. Monroe, ed. Ann Arbor, MI: Association for ConsumerResearch, 1112-15.

Kahneman, Daniel, Jack Knetsch, and Richard Thaler (1991),"The Endowment Effect, Lx)ss Aversion, and the Status QuoBias ," Journal of Economic Perspectives, 5 ( I ) , 193-206 .

March, James G. (1978), "Bounded Rationality, Ambiguity,and the Engineering of Choice," Bell Journal of Economics,9 (2), 587-608.

Payne, John W., James R. Bettman, and Eric J. Johnson (1992),"Behavioral Decision Research: A Constructive ProcessingPerspective," Annual Review of Psychology, 43.

Petty, Richard E., John T. Cacioppo, and David Schumann

(1983), "Central and Peripheral Routes to Advertising Ef-fectiveness: The Moderating Role of Involvement," Journalof Consumer Research, 10 (September), 134-48.

Samuelson, William and Richard Zeckhauser (1988), "StatusQuo Bias in Decision Making," Journal of Risk and Un-certainty, 1 (1), 7-59.

Sanbonmatsu, David M., Frank R. Kardes, and Bryan D. Gib-son (1991), "The Role of Attribute Knowledge and OverallEvaluations in Comparative Judgment," Organizational Be-havior and Human Decision Processes, 48 (1), 131-46.

Stone, Maria (1991), "Attention and Focus of Comparison,"working paper. Psychology Department, University of Cal-ifornia, Berkeley.

Tversky, Amos (1977), "Features of Similarity," Psycholog-ical Review, 84 (4), 327-52.

and Daniel Kahneman (1991), "Loss Aversion inRiskless Choice: A Reference Dependent Model," Quar-terly Journal of Economics, 106(4), 1039-61.

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

Reprint No. JMR294103

U.S. Postal Service Statement of Ownership,Management, and Circulation (Required by 39 U.S.C. 3685)

1 A. Title of Publication: /MR lounat of Marketing gesearchIB. Publication No. 0022-24372. Date of Filim: Sept. 29, 19923. Frequency of Issue: Quarterly3A. No. of Issues Published Annually: 438. Annual Subscription Price: $704. Complete Mailing Address of Known OfOce of Publication: 250 S. Wacker Dr., Suite 200, Chi-

cago, Cook County, I I 60606-58t95. Complete Mailing Address of the Headquarters or General Business Offices of the Publisher:

250 S. Wacker Dr., Suite 200, Chicago, II60606-58196. Full Names and Complete Mailing Address of Publisher, Editor, and Managing Editor:

Croup Publisher: leffrey Heilbrunn, 250 S. Wacker Dr., Suite 200, Chicago, IL 60606-5B19;Editor: Barton A. WeiU, College of Business Administration, University of Fkirida, 204 Math-erly Hall, Cainsville, FL 32611Executive Editor: Thomas E. Caruso, 250 S. Wacker Dr., Suite 200, Chicago, IL 60606-5819;

7. Otvner: American Marketing Association, 250 S. Wacker Dr., Suite 200, Chicago, IL 60606-5819

8. Known Bondholders, Mortgagees, and Other Security Holders Owning or Holding 1 % orMore of Total Amount of Bomb, Mortgages, or Other Securities (If there are none, so state):None

9. For Completion by Nonprofit Organizations Authorized To Mail at Special Rates (DMM Sec-tion 423.12 only). The purpose, function, and nonprofit status of this organization and theexempt status for Federal income tax purposes Have Not Changed During Preceding 12Months

10. Extent and Nature of Circulation:

A. Total No. Copies (Net Press Run)B. Paid and/or Requested Circulation

1. Sales throu^ dealers and carriers, streetvendors, arid counter sales

2. Mail subscriptions (Paid and/or requested)C. Total Paid and/or Requested

Circulation (sum of lOBt and 10B2)D. Free Distribution hy Mail, Carrier,

or Other Means—Samples, Complimentary, andOther Free Copies

E. Total Distribution (Sum of C and D)F. Copies Not Distributed

1. Office use, left over, unaccounted,spoiled after printing

2. Returns from news agentsG. TOTAL (Sum of E, F1 and 2—Should

equal net press run shown in A)

Average no.copies each

issue during pre-ceding 12 months

8,059

407,101

7,141

8367,977

Actual no.copies of singleissue published

nearest tofiling date

8,085

07,055

7,055

9097,964

8,059 8,085I certify that the statements made by me atrave are correct and complete. Signature and titleof editor, publisher, husiness manager, or owner: Thomas E. Caruso, Executive Editor