Carpenter and Nakamoto - Consumer Preference

GREGORY S. CARPENTER and KENT NAKAMOTO*

Market pioneers outsell later entrants in both consumer and Industrial markets.Entry barriers arising from preemptive positioning and switching costs have beenadvanced to explain this market share difference, termed "pioneering advantage."However, empirical studies show that pioneering advantages are present even inmature markets in which brands reposition and switching costs are minimal. In thesecoses, the authors argue that pioneering advantage can arise from the process bywhich consumers learn about brands ond form their preferences. This process canproduce a preference structure that favors the pioneer, making it difficult for laterenh-ants to "compete away" the pioneer's large market share, even if brands can

reposition and switching costs are minimal.

Consumer Preference Formation and PioneeringAdvantage

In many markets, leading brands outsell their rivalsfor years and sometimes decades. For example. IvorySoap dominated its market in 1923 and continued to doso in 1983 (Advertising Age 1983). Market share differ-ences are especially large for brands that entered earlyin the product life cycle, so-called "market pioneers" or"first-movers." Though many are not strictly the firstentrant (e.g.. Miller's Lite Beer was preceded 15 yearsearlier by Rheingold's Gablinger's) and many fail asBowmar did in the calculator market, surviving early en-trants or market pioneers appear to have significandy largershares than surviving later entrants in both consumer andindustrial markets (Robinson 1988; Robinson and For-nell 1985; Urban et al. 1986). This difference, termed"pioneering advantage," is well illustrated by Wrigley'schewing gum, which dominates its market more than sixdecades after its introduction.

'Gregory S. Carpenter is Associate Professor, Graduate School ofBusinesa, Columbia University. Kent Nakamoto is Assistant Profes-sor, Karl Eller Graduate School of Management. University of Ari-zona.

The authors are grateful for the helpful comments of Alan An-dreasen. Lee Cooper, Mike Hanssens, Morris Holbrook, Mike John-son. Barbara Kahn. Hal Kassarjian. Don Lehmann, Bob Meyer. DaveMontgomery. John O'Shaughnessy, Stan Omstein. Bill Robinson, BillRoss, and four JMR reviewers and for financial support from the Fac-ulty Research Fund. Institute for Marketing Studies, and the StrategyResearch Center, all of Columbia University, and the UCLA Aca-demic Senate.

This advantage is remarkable in many respects. It ap-pears resistant to competitors' actions, surviving the in-troduction of new brands, innovation by existing rivals,price competition from generics or imports, and shiftingconsumer tastes. Wrigley's advantage, for instance, haswithstood competitive attacks through product differen-tiation (e.g., Dentyne), innovation (e.g.. Carefree), pricecompetition from less-well-known brands, and shiftingconsumer tastes. This experience suggests that the mech-anism producing pioneering advantage somehow slowsthe natural forces of competition, making it difficult forlater entrants to "compete away" a pioneer's advantage.

Entry barriers arising from preemptive positioning andswitching costs have been advanced as one explanation.'If brands locate in a product space, as discussed by Ho-teiling (1929), an early entrant can preempt later onesby adopting the "best" market position, leaving smalleror less attractive segments for others (e.g.. Lane 1980;Prescott and Visscher 1977). If later entrants cannot re-position, they must offer "something extra" such as abargain price to gain market share, which makes com-petition for them more costly, discouraging entry andkeeping the pioneer's share large.

Switching costs created by uncertainty about later en-trants' quality or user skills can have a similar impacton competition. Schmalensee (1982) shows that if trial

'See Lieberman and Montgomery (1988) for an excellent review.

285

Journal of Marketing ResearchVol. XXVI (August 1989). 285-98

286 JOURNAL OF MARKETING RESEARCH, AUGUST 1989

is necessary to verify a brand's quality and the pioneerachieves a high rate of penetration, its product is lessrisky, forcing later entrants to cut price or offer somepremium to compensate for risk and induce trial. So longas trial of later entrants remains low and their qualityremains unknown, the pioneer will retain its high marketshare.

Brand-specific user skills also can create switchingcosts. If the product use requires a significant degree ofbrand-specific knowledge and if the pioneer achievessignificant penetration, the resulting base of pioneer-spe-cific knowledge will make trial of a later entrant costly.Stigler and Becker (1977) show that these user skills cancreate market share differences even if consumer tastesare identical and are fixed (see also Fomell, Robinson,and Wemerfelt 1985).

Empirical studies show that entry barriers are an im-portant source of pioneering advantage (e.g., Robinsonand Fomell 1985), but they also show that pioneeringadvantages are present in markets in which brands canreposition and switching costs are minimal. For exam-ple. Urban et al. (1986) show that Miller's Lite Beerretains its pioneering advantage, even though most con-sumers are aware of its major rivals (Budweiser andSchlitz), brands reposition, and user skills are unimpor-tant. In cases like this, a pioneer's advantage must arisefrom something other than preemptive positioning orswitching costs.

We argue that the process by which consumers leamabout brands and form preferences for them has an im-portant role in creating an advantage for pioneers in suchcases. Our explanation has two components. First, in theearly stages of many markets, consumers may know lit-tle about the importance of attributes or their ideal com-bination. For example. 100 years ago few people werelikely to have strong opinions about how sweet or car-bonated a cola should be. A successful early entrant canhave a major influence on how attributes are valued andon the ideal attribute combination. Cola-Cola, for ex-ample, may have had a significant impact in its earlyyears on the formation and evolution of individuals'preferences for colas. This influence can shift individ-uals' preferences to favor the pioneer over later entrants,leading to a market share advantage.

Second, we argue that this learning process can pro-duce a competitive advantage apart from influencing theconsumer's ideal combination of attributes: the pioneercan become strongly associated with the product cate-gory as a whole and, as a result, become the "standard"against which all later entrants are judged. Kleenex, Xe-rox, and Jelio are obvious examples. Being strongly rep-resentative, the pioneer is competitively distinct, whichmakes competing away its high share difficult for laterentrants, especially for low-priced copies or so-called "me-too" brands.

We develop this explanation, outlining the conditionsunder which it will operate, and deduce a set of testablehypotheses. We examine them for two novel productcategories. In both cases, we experimentally construct

an emerging market, varying the order of brand entryacross different experimental groups in addition to vary-ing the types of competitors that subsequently enter.Analysis shows that preferences are influenced by theorder of entry, as we predict. Moreover, the preference-formation process produces a preference structure thatmakes a pioneer's market share largely invulnerable tocompetitors even if switching costs are minimal and brandscan reposition. This advantage arises independently ofproduct characteristics and is due entirely to order of brandentry and the role of the pioneer in forming preferencesfor brands in the category.

Our analysis has important implications beyond ex-plaining pioneering advantage. First, we propose andprovide evidence that preferences are endogenous,evolving with the market as Kahneman and Snell (1988)and others suggest, in contrast to the widely held viewthat they are fixed (e.g., Stigler and Becker 1977). Amongother things, the analysis suggests that influencing pref-erences, not simply responding to them, may be an im-portant objective of marketing strategy. Second, we of-fer strategies for competing with market pioneers,including me-too strategies. Me-too strategies are gen-erally thought to be ineffective; however, our analysisshows they can be effective under certain conditions.

We begin our discussion with an examination of theleaming mechanism that produces a preference structurefavoring a pioneer. Next, we explain how that structureleads to a competitive advantage for market pioneers.Our experiments then are described and the results re-ported. In the final sections we discuss the results andtheir implications, extensions of the theory, and asso-ciated future work.

BUYER LEARNING AND PREFERENCEFORMATION

We consider preference formation for multiattributeproducts and services. We limit our analysis to productcategories for which the contribution of product attri-butes or features to overall brand value and the idealattribute combination is ambiguous. For example, if onepurchases a down quilt and receives from it a certainmeasure of value, the contribution of the percentage ofgoose down fill to that overall value is ambiguous. Asa result, the ideal percentage of goose down fill also isambiguous. Other examples include the flavoring of softdrinks, features of computer programs, and combina-tions of ingredients in vitamins. We exclude categoriesfor which the value of attributes and their ideal combi-nation is unambiguous.

Leaming about novel products such as these presentsa very difficult and complex problem for consumers.Knowledge about the category is initially minimal andconsumers are exposed to products sequentially^firstthe pioneer and subsequently later entrants. As a result,consumer preferences are likely to evolve through time,updated through heuristic judgment processes (Kahne-man and Snell 1988; Tversky and Kahneman 1974). Wefollow consumer ieaming about these products through

CONSUMER PREFERENCE FORA^ATION AND PIONEERING ADVANTAGE 287

three stages. First, we consider the consumer prior toexposure to any brand in the category. Second, we ex-amine the impact on buyer preferences of initial trials ofthe pioneer. Last, we consider the evaluation of multiplebrands that typically enter after a pioneer (including bothme-too brands and differentiated entrants).

In all three stages, we consider only surviving or so-called succes.^ful market pioneers and assume that theirsuccess is due in part to a high rate of market penetra-tion. We do not consider how pioneers achieve this highrate of penetration, but instead examine how their suc-cess produces a competitive advantage over rivals.

Preferences Prior to Trial

Prior to trial, preferences may be "weakly formed"because the category is novel and buyers know little, ifanything, about it. Even if buyers have objective infor-mation on brand attributes, the value of an individualattribute or the superiority of one attribute combinationover another may not be obvious (cf. Howard 1989;Howard and Sheth 1969). Hence individuals may belargely indifferent between altematives over some rele-vant range. In a spatial sense, the distribution of idealpoints or vectors across consumers—the taste distribu-tion—may be approximately uniform as shown in Figure1, a two-dimensional market with roughly uniform dis-tribution of ideal points (A) and the corresponding av-erage ideal point (B). Attribute weights used to valuebrands also will be tentative.

Updated Preferences

Buyers update their preferences through trial. In thecase of an emerging market, trial implies sampling thepioneering brand, perhaps repeatedly and for an appre-ciable period. For example. Cool Whip pioneered itsmarket and 18 years elapsed between its introduction andthe entry of a competing brand. However long, this pe-

Figure 1HYPOTHETICAL PRIOR (A) TASTE DISTRIBUTION AND (B)

CORRESPONDING PERCEPTUAL MAP WITH AVERAGE

IDEAL POINT

B

Attribute 1

Attribute1

IdeatPoint

Attribute 2

Figure 2

HYPOTHETICAL (A) TASTE DISTRIBUTION AND (B)

PERCEPTUAL MAP UPDATED AFTER PIONEERING ENTRY

Attribute 1

Attribute1

Pioneer

\ IdealPoint

Attribute 2Attribute 2

riod of single-brand use influences the preference struc-ture for the type of products we consider.

Buyers sample the pioneering brand and, lacking in-formation to the contrary, attribute a successful outcometo its attribute combination (Meyer 1987). In doing so,buyers may develop a naive theory relating brand fea-tures to value, which advertising and repeat purchasereinforce (Hoch and Ha 1986; cf. Deighton 1984). Thus,buyers leam through trial how to value attribute com-binations. Because their experience is limited to a singlebrand, however, they leam to value the pioneer's attri-bute combination and update their preferences accord-ingly.

In spatial terms, individuals shift their ideal points to-ward the pioneer's position to reflect their learning asshown in Figure 2. Both the taste distribution and av-erage ideal point shift toward the pioneer's location. Thisshift occurs independently of the characteristics of thepioneer. With different characteristics, the taste distri-bution would shift in another direction, provided the valueof brand attributes and the ideal attribute combinationsaie ambiguous. If both are well known to buyers, noshift will occur. Hence,

H,: The consumer taste distribution shifts toward the pi-oneering brand's position, provided the value of at-tributes and the ideal attribute combination are am-biguous.

Attribute weights are updated by a similar process.^Originally they are somewhat tentative, but are influ-

Attribute 2

^The distinction between attribute values and combinations or levelscan be made as follows: let v,, = ^.[x, - xf\ + 7,[y, - y*\ where v,is the value individual < attaches to brand j . {x^,y,) is the position ofbrand j in an .r-v space. (.v,*,y*) is individual I's ideal point, and p,and -y, capture the value of being closer to the ideal point. Here, pref-erences are characterized by both ideal points (xT,y*) and attributevalues (p,,7,).


enced by the pioneer if no objectively superior weightsexist. The pioneer defines the relative attribute impor-tance in brand evaluation and choice. For example, inthe market depicted in Figure 2, buyers may value at-tribute 1 highly because the pioneer is strong on that di-mension, but may find attribute 2 less valuable. A brandstrong on attribute 2 but weak on attribute 1 would bepoorly rated. If the pioneer were in the bottom right ofFigure 2B, the situation would be reversed. More for-mally,

H2: Attribute weights shift toward the relative attributestrengths of the pioneer, provided the contribution ofbrand attributes and the ideal attribute combinationare ambiguous.

This preference-learning process is well illustrated bythe Chesebrough-Ponds experience in the petroleum jellymarket. In 1880, Vaseline was introduced and advertisedas a healing agent of unsurpassed purity. SamplingVaseline, a translucent, highly pure gel, buyers learnedthat its attributes produced an effective wound prepara-tion and, generalizing from this observation, inferred thatthe effectiveness of a petroleum jelly lies in its translu-cence and purity. Subsequent trials and advertising ofcourse confirmed this conjecture. Thus, translucence cameto be favored over opacity (H,) and translucence as op-posed to, say, thickness gained more importance in brandevaluation (Hj).

Recent studies of category leaming lend further cre-dence to this leaming mechanism. A variety of organiz-ing principles have been advanced as bases for gener-alizing consumer knowledge, including scripts, schemas,explicit rules, and categories (Alba and Hutchinson 1987;Marks and Olson 1981; Meyer 1986; Sujan 1985). Elioand Anderson (1984) examined order effects in leamingof category structure. They found that category gener-alization and resulting category expectations are strong-est when leaming is based on a limited set of highlyrepresentative examples. Thus, expectations of subse-quent category members reflect prior experiences.Leaming from a single example such as a market pioneerrepresents an extreme version of this mechanism. Vari-ation in repeated trials of the pioneer is zero, so by de-fault the pioneer becomes highly representative and a goodexample of the category.

Though these results were obtained for perceptualjudgments, Barsalou (1985) describes a similar mecha-nism for evaluative judgments. For goal-derived cate-gories (groupings of products based on a common func-tion like "diet foods"), he shows that the typicality ofan item or its "goodness"—tantamount to an evalua-tion—increases with the number of times a person ex-periences the item as a member of that category and withthe proximity of the item's attributes to category ideals(e.g., zero calories for diet foods). In an emerging prod-uct category, ideals are inferred by the consumer fromexperience with the pioneer. Thus the pioneer's attri-butes naturally are placed close to the category ideals.

Furthermore, frequency of experience with the pioneerwill be greatest, by default (assuming high penetration),again increasing its goodness in the category. Togetherthese two factors make the pioneer highly representative.In the extreme, they make it uniquely representative ofthe category, as when the pioneer is synonymous withthe entire product class (e.g., Levi's, Kleenex, or Jello).'

PERCEPTUAL DISTINCTIVENESS ANDCOMPETITIVE ADVANTAGE

A pioneer typically is followed by the introduction ofcompeting brands. Later entrants include me-too orcopycat brands claiming to be just like the pioneer butless expensive and differentiated entrants offering sig-nificantly different attribute combinations. For example.Figure 3 shows our hypothetical market after the entryof three new competitors, a me-too brand and two dif-ferentiated entrants.

The leaming mechanism we propose predicts a marketshare advantage for the pioneer over differentiated laterentrants. Differentiated brands are less preferred becauseof a greater difference between their attribute combina-tion and the ideal one, all else equal, and the inferredattribute weights favor the pioneer.

However, this mechanism raises a fundamental issue:later entrants appear to have an incentive to reposition

'See Carpenter and Nakamoto (1987) fora more detailed discussionof the behavioral foundation for this leaming mechanism.

Figure 3HYPOTHETICAL PERCEPTUAL MAP AFTER MULTIPLE-

BRAND ENTRY

Attr ibute 1

Pioneer

Me-too

- -

, IdealPoint

LaterEntrant

LaterEntrant 2

1

Attribute 2

CONSUMER PREFERENCE FORMATION AND PIONEERING ADVANTAGE 289

toward the pioneer, increase their market share, andcompete away the pioneer's advantage. A similar ineen-tive has been shown in models of spatial competition ifprice-cutting is limited. What discourages brands fromdoing so and protects the pioneer's advantage?

PrototypicalityIn our framework, protection is provided by the pro-

totypicality or distinctiveness of the pioneer. Consumersappear initially to organize product knowledge aroundprototypical examples, using them as cognitive referents(Medin and Schaffer 1978; Sujan 1985). The pioneer hasa unique distinctiveness derived from its being represen-tative of the category. Being perceptually distinct, thepioneer overshadows brands positioned nearby, espe-cially me-too brands that often rely on the pioneer toestablish their identity. Thus, me-too brands suffer incomparisons with the pioneer despite their proximity tothe ideal point.

Moreover, as later entrants position closer to the pi-oneer, they become less distinct and the pioneer moredistinct, increasing the relative perceptual prominence ofthe pioneer, provided ideal attribute combinations andweights are ambiguous and if price differences betweenbrands are small in absolute terms {e.g., low-priced con-.sumer products). If the ideal attribute combination canbe determined objectively and prices are high, later en-trants may well be able to overcome the distinctivenessof the pioneer (e.g., personal computers). Therefore,comparisons drawn with the pioneer invariably favor it,so long as ambiguity remains, and produce a market shareadvantage for it.

Differences in distinctiveness translate directly intomarket share differences. The more similar the pioneerand its me-too, the greater the relative advantage of thepioneer and, thus, the greater its advantages over the me-too. The more dissimilar the two brands, the smaller thepioneer's relative advantage and hence the more nearlyequal their market shares. For example, in Figure 3, theadvantage of the pioneer over its me-too brand will in-crease the closer the me-too brand is positioned to thepioneer. Hence,

H3: The market share of the pioneer in relation to its me-too increases with the perceived similarity betweenthe two, provided individuals' attribute weights andideal points are ambiguous.

Furthermore, price-cutting by me-too brands has littleimpact on the advantage of the pioneer. Being closer tothe pioneer, the me-too brand is less distinct than thepioneer, so any price reduction has a smaller impact, allelse equal, than a similar price reduction by a more dif-ferentiated rival. For example, in Figure 3 a 10% pricecut by the me-too brand reduces the pioneer's share lessthan an equal price cut by either of the other later en-trants. More formally,

H4: The impact of a competitor's price reduction on thepioneer's market share increases with the perceiveddifferentiation of the competitor.

Pioneering brands, according to H, and H4, have aperceptually based competitive advantage that insulatesthem from competitors and may actually reverse com-petitive forces. Attempts to cut price and reposition to-ward the pioneer—to "compete away" its high share—increase the pioneer's advantage. Hence me-too brandsfail in part because they are overshadowed perceptually,not because of price competition as is often suggested(e.g.. Lane 1980). Me-too brands are unable to generatesufficient market share to remain viable, even with a low-price strategy, if they are perceived as very similar tothe pioneer. This situation puts intense pressure on theme-too brand to lower prices further to generate addi-tional volume, but not on the pioneer, so price-cuttingshould be asymmetric with the me-too lowering price butnot the pioneer. Federal Express is one example of thisphenomenon; Purolator Courier has aggressively at-tacked Federal Express principally on price, yet has haddifficulty building share while Federal Express' marketshare and price have remained relatively high (Carpenter1987).

Overcoming Distinctiveness

H3 and H4 also suggest a strategy for me-too brandsto overcome their principal competitive disadvantage—a lack of distinctiveness. A later entrant can diminishthe impact of the pioneer's distinctiveness and increaseits own by moving away from the pioneer. Doing so, itcan establish or develop a new location as a desirableone, shifting at least a portion of the taste distributiontoward its position and achieving a higher degree of rel-ative prominence.

Consequently an effective strategy for a me-too brandis to copy a differentiated brand, not the pioneer. Posi-tioning closer to a differentiated entrant helps developrecognition for the market segment and increases the rel-ative prominence of both the distinctive entrant and itscopycat. Furthermore, by decreasing the relative dis-tinctiveness of the pioneer, this segmentation strategy in-creases the market shares of both differentiated brandsat the pioneer's expense. The me-too brand in Figure 3could reduce the pioneer's advantage by copying laterentrant 1 or 2 rather than the pioneer. More formally,

H5: The market share advantage of the pioneer decreasesas the similarity between the differentiated later en-trant and its me-too increases, provided attributeweights and ideal combinations are ambiguous.

Summary

The preference-formation process produces a compet-itive advantage for pioneers in two ways. First, a pioneerdevelops the best position by shifting the taste distribu-tion toward its position and by influencing the attributeweights buyers use to evaluate brands. Shifting prefer-ences toward its own position generates a high share forthe pioneer and a smaller share for later entrants. How-ever, shifting the taste distribution and attribute weights


does not protect the pioneer's high share from compet-itors.

A second brand-specific effect does. Because the pi-oneer has a central role in category preference forma-tion, the pioneer becomes prototypical of brands in thecategory. Prototypicality protects the pioneer's high sharefrom competitors; attempts to position close to the pi-oneer and the ideal point and reduce price may, contraryto the prevailing view of competition, increase the pi-oneer's market share.

Neither component excludes a role in pioneering ad-vantage for preemptive positioning or switching costs.Both superior position and significant switching costs maycontribute to the advantage of the pioneer along with anyimpact of the pioneer on the evolution of preferences.However, the preference-formation process can producea pioneering advantage, as we have argued, even if thereare no superior positions, all brands are well known, andno user skills are necessary. Moreover, if we assumepreferences are fixed and brands are undifferentiated, entrybarrier explanations make no predictions about how en-try will affect preferences or about how the perceivedsimilarity of brands affects pioneering advantage.

EXPERIMENT I

Our first experiment is an initial test of the mechanismwe propose, designed to demonstrate its existence andexamine features of its operation. We focus on anemerging market in which a pioneer enters first, fol-lowed by later entrants including copycat and differen-tiated brands. In this first experiment, we test Hi, H3,and H5.

Method

To test these hypotheses we experimentally con-structed an emerging market. Six hypothetical variantsof a coniputer software package purporting to identifypotential sources of financial aid for students were de-signed. The brands differed on five dimensions—num-ber of financial aid sources identified, difficulty of use,time required to run the program, amount of documen-tation, and price. The six brands are profiled in Table1. Each brand was described in a paragraph ostensiblyparaphrasing advertising copy, including claims justi-

fying the superiority of a brand's attribute combination.No brand-specific skills were developed.

The basic structure of the market defined by thesebrands is shown in Figure 4. Two brands, G and K, wereused as market pioneers and are called ''reference" brands.For each subject in the experiment, one of these brandswas the pioneer and the other took the role of a distinc-tive follower. Brands P and W were (respectively) copy-cat brands to brands G and K. advertised as equivalentproducts at a lower price. These four are the brands ofcentral interest.

Forty-eight MBA students participated in the experi-ment as a course requirement. Subjects first read a shortdescription of the product's function, then a descriptionof the pioneer. Half of the subjects saw brand G firstand the other half saw brand K first.

At this point, the ambiguity of the ideal attribute com-bination was manipulated. After exposure to the pioneer,subjects were asked to imagine that they had purchasedthe product and had successfully obtained financial aidby using it. Half of the subjects then were asked to writea description of product characteristics that led to thesuccessful outcome. This was the ambiguous condition,in which subjects inferred criteria for good product per-formance. The other subjects read an analysis, purport-edly prepared by experts, specifying ideal attribute lev-els. This information made the me-too copy of thenonpioneer reference brand the optimum for all subjects.If brand G was the pioneer, the rule implied that brandW was the superior product; if brand K was the pioneer,brand P was superior. This manipulation constituted theobjective ideal point condition. All four combinations ofideal point revelation and pioneering brand were in-cluded in a 2 X 2 factorial design with 12 subjects giveneach combination.

After the subjects had read or written brand evalua-tions, they were asked to imagine that they were con-sidering buying the product again and were presentedfive new brands described as later entrants. Subjects thenwere asked to specify the ideal characteristics of theproduct on the five dimensions used to describe the prod-uct.

At this point, a one-hour delay was imposed duringwhich subjects pursued an unrelated task. The purposeof the delay was to avoid short-term memory effects on

Brand

GPMKW

c

Number ofsources

identifted

6-86-8

12-1418-2018-2024-26

Table 1BRAND PROFILES FOR EXPERIMENT 1

Requiredto run(min.)

404020909060

Instructionmanual(pages)

404065101025

1

Technicaldifficulty

ModerateModerateHighLow - i'LowModerate

Price ($)49.9529.9549.9549.9529.9549.95


Figure 4PRODUCT-MARKET CONFIGURATION (FOR THREE OF

FIVE AHRIBUTES)

Running Time(Minutes )

20

40

6 0

90 -

• M($ 49.95)

G($ 49.95)P($ 29.95)

C($49.95)

• K($ 49.95)WC$ 29.95)

6-8 12-14 18-20 24-26 Numberof

Sources

brand petxeptions. After the delay, subjects were askedto indicate brand preferences and to rate the sitnilarityof the brands. To rate preference, subjects allocated 100points over the six brands to reflect their relative pref-erences, which we treat as a surrogate for tnarket share.Finally, pairwise similarity ratings on an 11-point scalewere collected for all 15 brand pairs.

Results

We first examine overall preference shares of later en-trants and pioneers and the prototypicality of pioneers asmanipulation checks.

Pioneering advantage. Our analysis predicts that thepioneer will gain the largest preference share when theideal point is ambiguous, regardless of its position. Meanpreference shares for brands G and K, shown in Figure5 for each cell, support this prediction. When brand Gis the pioneer and the ideal point is ambiguous its shareis roughly twice as large as brand K's, as shown in Fig-ure 5A. When K is the pioneer the pattern is reversed;brand G receives a large share only when K is the pi-oneer and the ideal point is objective (in which case brandG's me-too, P, is specified as the ideal product), ben-efitting G. This situation gives rise to the observed in-teraction. A similar albeit weaker pattern is seen for brandK, as shown in Figure 5B.

To test the significance of these differences, we com-

Figure 5PREFERENCE SHARES FOR BRAND G AND BRAND K

BMarket Shareof Brand G

3 0

2 0

10

Mgrket Shareof Brand K

Ideal PointUnambiguous

Ideal PointAmbiguous

I

2 0

10

Ideal PointUnambiguous

Ideal PointAmbiguous

K PioneeringBrand

K PioneeringBrond


puted the difference in preference shares between brandsG and K and submitted it to an analysis of variance withpioneering and the ideal point revelation as factors. Theinteraction is significant (F, 44 = 6.88; p < .05). Further,for the ambiguous ideal point cells, the difference shouldincrease when G is the pioneer and decrease (becomingnegative) when K is the pioneer. A one-sided t-test forthis contrast is consistent with this prediction (t^^ = 1.91;p < .05). Therefore, pioneering brands have a largerpreference share, regardless of their characteristics, pro-vided the ideal point is ambiguous.

Prototypicality. Our analysis predicts that the pioneeris perceived as more prototypical, which shields it in onesense from competitors and yields it a significant ad-vantage in relative preference. This preference differ-ence arises from a fundamental difference in the percep-tual structure caused by pioneering. A brand should beperceived as more similar to others when it is the pi-oneer—more representative—but also it should domi-nate others—be more distinct.

To examine the prototypicality of pioneers, we use thenearest neighbor statistics developed by Schwarz andTversky (1980) and Tversky and Hutchinson (1986). Theypropose two measures—centrality and reciprocity—basedon similarity judgments and the concept of nearestneighbor to analyze perceptual structure. A brand's near-est neighbor is the brand to which it is perceived to bemost similar. The centrality of a brand is the number ofbrands for which it is the nearest neighbor; a brand per-ceived as highly similar to more brands is more central.The reciprocity of a brand is the rank of that brand inthe order of similarity of all brands to its nearest neigh-bor; a higher rank—greater relative similarity—indi-cates a more symmetric or reciprocal perception whereasa lower rank indicates an asymmetry in perception. Forexample, consider brand G in our hypothetical market.If pairwise similarity judgments indicate that G is mostsimilar to P of all brands, then G is P's nearest neighbor.If G is the nearest neighbor of P only, then G's centralityis 1. Further, if for brand G, P ranks second in similarityrelative to all others, then P's reciprocity is 2.

For our study, the centrality and reciprocity of a brandshould vary systematically with pioneering. We expect'pioneering to increase a brand's centrality. Being moreprototypical, a brand should have more near neighborswhen it is the pioneer than when it is not. Furthermore,we expect pioneering to reduce a brand's reciprocity.Because the pioneer is more prototypical, its nearestneighbor (its me-too) will be perceived as more similarthan it would be if both were later entrants (cf. Tverskyand Hutchinson 1986).

Analysis of these statistics shows a significant impactof pioneering in the ambiguous condition. Measures ofcentrality and reciprocity of the reference brands werecomputed for each subject and treated as repeated mea-sures in an analysis of variance with pioneering and idealpoint ambiguity as between-subjects factors. The inter-action of pioneering and ideal point ambiguity is signif-icant (F,,44 - 4.45, p < .05).

The pattern of nearest neighbor statistics for brands Gand K is shown in Figure 6. For the ambiguous condi-tion, mean centrality for brand G rises from 1.15 to 1.33when it pioneers the market and brand K's mean cen-trality increases similarly from 1.52 to 1.68. Also, aspredicted, mean reciprocity for each brand falls when itis the pioneer, from 1.47 to 1.06 for G and from !.21to 1.04 for K. This pattern of results is consistent withthe prototypicality effect we hypothesize. Pioneeringbrands evidently are perceived as more central and thepioneer's me-too brand is perceived as relatively moresimilar than is the case when the same two brands enterthe market later.

These analyses indicate that our manipulations pro-duced (1) an advantage arising from the order of entryand (2) a difference in the perceived prototypicality ofbrands for the pioneering brand. Both effects are limitedto the ambiguous condition. To test our hypotheses ex-plicitly, we next examine the resulting patterns of pref-erence in greater detail.

Ideal point shifts. H, predicts that the perceived idealproduct will depend on whether brand G or K is seenfirst and on whether the ideal point is ambiguous. As atest we computed mean values for the ideal number ofsources identified and ideal time required to run. Figure7 shows the locations of brands G, P, K, and W and thefour average ideal points for each subject group for aportion of the product space. Subscripts denote whichbrand entered first (G or K) and whether the ideal pointwas ambiguous (A) or unambiguous (U).

The pattern of ideal points is as we predict. If the idealattribute combination is ambiguous, the ideal point shiftstoward the position of the pioneer—whether it is G orK—as shown in Figure 7 by I^A being closer to brandG and I^A being closer to brand K. If the ideal attributecombination is unambiguous, the idea! point shifts to-ward the objectively superior brand (here, brand P if Kwas the pioneer and brand W if G was the pioneer) asshown in Figure 7 by IGU being closer to brand K andIKU being closer to brand G. Interestingly, not all buyers'ideal points shifted to the objectively superior one; somebuyers continued to prefer the pioneer, even in the faceof strongly contradictory evidence.

We analyzed the underlying data by computing the ab-solute difference between reported ideal values and thosefor brand G and brand K for three characteristics (num-ber of sources, running time, and documentation length)which we denote DG* and DKI (/ = 1, 2, 3 for the threecharacteristics). The difference between these values wascomputed (that is, DK. - Dc) and analyzed via MAN-OVA with pioneering and ideal point ambiguity as fac-tors.

The pattern of results suggests that, when the idealattribute combination is ambiguous, the ideal points shifttoward the position of the pioneer as indicated in Figure7. The only significant effect is an interaction betweenpioneering and ideal point ambiguity (F341 = 4.87;/7 <.01). The effect is eliminated if "optimal" attribute lev-els are revealed. In univariate analyses of the ambiguous


Figure 6IMPACT OF PIONEERING ON BRAND CENTRALITY AND REOPROCITY

BrandCentrality

2.0

1.5

1.0

-

r • -•

: \ ^ • • ' .

1

^^ -̂« Brand

^ ^ ^ Brand

I

K

6

BBrand

Reciprocity

2.0

1.5

1.0

Brand G

Brand K

K PioneeringBrand

PioneeringBrond

Figure 7PERCEPTUAL MAP WITH IDEAL POINT POSITIONS BY

^ . . EXPERIMENTAL CONDITION

Running Time(Minutes)

40

60

90

• G

» •

• • •

1

.P

• KU

•

1 . .

. . . ^

• ^ G A

•

«

^KA

• K,W

6-8 12-14 18-20 Numberof

Sources

cases alone these differences are not statistically reliable.However, that effect tnay be due in part to the fact thatthey were collected before the delay and the recency ofexposure to later entrants may have weakened the pi-oneering effect.

These results suggest that, as in the case of preferenceshares, ideals tended to favor the pioneer regardless ofthe position it adopted. This situation occurs so long asthe ideal attribute combination is ambiguous. Objectiveideal point information appears to overwhelm experi-ence. This finding suggests, contrary to models ofpreemptive positioning, that pioneers develop the bestpositions rather than simply preempting them. More-over, preferences seems to be at least partly endogenous,rather than fixed as commonly assumed. Preference evo-lution appears to depend on the order of brand entry inthe market.

Pioneer versus later entrants. We test H^ and H5, whichmake predictions about the preference share differencesbetween the pioneers and later entrants, by first calcu-lating the difference in preference shares between pi-oneers and me-too brands as share{G) - share(P) andshare(K) — share(W). Terming this difference "relativeadvantage" and denoting it by RA{G) and RA{K), re-spectively, we model it as

(1) RA{j) - a, +


where:

RA{f) = relative advantage of brand j = G or K,S(G,P) = the similarity of brands G and P,

S(K,W) = the similarity of brands K and W, andPc = I if G is the pioneer.

H3 and H5 make predictions about the signs of coef-ficients in equation I. When brand G is the pioneer,equation 1 is

RAiG) = (a, + a^) + (aj + as)S(G,P) + (uj + ae) SiK,W).

When the ideal point is ambiguous, H3 and H5 imply thatbrand G should gain relative advantage from pioneering(a, + a4 > 0), from greater similarity to brand P (a2 +a5 > 0), and from greater dissimilarity between brandsK and W (a^ + as < 0). When brand K is the pioneer,equation 1 reduces to

RA(K) = a, + a2 SIG,P) + aj5(A',W).

When the ideal point is ambiguous, brand K should gainrelative advantage from increasing dissimilarity betweenbrands G and P (ai < 0) and from increasing similaritybetween brands K and W (03 > 0).

We estimate equation 1 for each condition separately.The results for the ambiguous condition are reported inTable 2. Overall the models predict well when the idealpoint is subjective; R^s range from .35 to .47. For brandG when the ideal point is subjective, these coefficientsimply the following model.

RA{G) = 38.3 + 3.40 S{G,P) - 7.03 S(K,W),

which shows that increasing similarity between G and Pincrea.ses brand G's relative advantage, as does greaterdissimilarity between brands K and W. For brand K theimplied model is

RA(K) = 14.6 - 6.51 S{G,F) + 1.83 S{K,W),

which demonstrates a similar pattern. The revelation ofthe ideal point eliminates these relationships. Pioneering

Table 2REGRESSION MODEL RESULTS FOR RELATIVE

ADVANTAGE AS A FUNCTION OF BRAND SIMILARITIES

V^HEN QUALITY IS SUBJECTIVE

Variable

ConstantSimilarity (G.P)Similarity (A", HOPG

Pc X similarity (G,P)Pc X similarity (K.W)

Relative

Brand G

7.60-.93-.3730.70"4.33"

-6.66'

advantage

Brand K

14.60-6 .51 '

1.83-41.60"

9.54'-2.86

> < .05."/j < 10.'p < .01.

and brand similarity are virtually useless for predictingrelative advantage when the ideal point is objectively re-vealed; R^s drop to under . 10.

These results support H, and H5. Prominence or pro-totypicality achieved through market pioneering appearsto offer some shield against competitors, reducing theirability to "compete away" the pioneer's share. Estab-lishing a differentiated location as a niche, however, doesreduce the relative prominence of the pioneer and thusits relative advantage.

Discussion

The results of the first study are very suggestive. Ashypothesized, preferences shift in favor of the pioneerbut only when the ideal attribute combination is ambig-uous, which is consistent with the buyer-learning mech-anism we propose. On the basis of this preference shift,later entrants are perceived as less ideal. Therefore, or-der of entry affects the structure of consumer preferencesfor brands in the category, yielding pioneers a superiorposition and a substantially higher share of buyers'choices.

The finding that similarity between the nonpioneeringreference brand and its me-too decreased the relative ad-vantage of the pioneer (and also its overall share) is alsosuggestive. As the similarity of these brands increases,the perceptual mass at a location distinct from that of thepioneer increases. A natural consequence would be forthe consumer to split the market into subcategories, onecentered around the pioneer and the other formed arounddifferentiated brands. In terms of conceptualizing pi-oneer advantages, this finding suggests that pioneer ad-vantages are to some extent local, restricted to percep-tual locations relatively "nearby." What is "nearby"depends on the concentration of brands in different areas.A greater concentration of brands at a differentiated lo-cation increases the likelihood that those brands will besubcategorized and the product space segmented in themind of the consumer. Indeed, a common brand intro-duction strategy is to introduce a new attribute that dis-tinguishes the new brand in an attempt to establish a newsubcategory (e.g., toothpaste in a pump container, de-signer jeans, decaffeinated sugar-free colas).

These perceptual effects suggest that a later entrant'scompetitive advantage depends on developing a suffi-cient level of distinctiveness. Greater prominence andgreater concentration at a differentiated position increasethe ability of later entrants to compete with the pioneer.The problem for the pioneer's me-too brand is that itdiffers only in terms of price, which may not be a mean-ingful difference, particularly when adjusted for per-ceived quality. Instead, differentiating the product by of-fering a more significant advantage seems necessary,which is consistent with Bond and Lean's (1977) studyof the prescription drug industry in which later entrantswere able to overtake the pioneers by offering new ben-efits.


EXPERIMENT 2

The goal of experiment 2 is to provide a stronger testof the impact of learning on preference structure. In de-sign, it is similar to experiment I in which order of brandentry is varied, but here preferences are measured ex-plicitly via conjoint analysis to test H2 and H4.

Method

The procedure was similar to that in experiment I.Subjects were exposed to one brand in a novel productclass, down quilts. Told that they had purchased a prod-uct from the category and it had performed satisfactorily,subjects were asked to explain why and state whether ornot they would purchase from the category again shouldthe need arise. For two-thirds of the subjects, this stepwas followed by exposure to a second brand; the othersubjects were exposed to no other products. Finally, allsubjects were shown eight hypothetical brand profiles,all different from brands already used, and asked to rankthem, which formed a full-profile conjoint task.

Four brands used in the experimental market differedin terms of four attributes: type of down fill (mature gooseand duck or duck only), fill rating ( a numerical indicatorof warmth), type of quilt cover (cotton or synthetic), andprice. Brand A was filled with mature goose and duckdown; brand B was filled with goose down only. Thecover of brand A was said to be cotton, whereas brandB's cover was synthetic but tightly stitched to preventbunching of the fill. Both had fill ratings higher than 500(good) and both were priced at $395. The other twobrands, C and D, were me-too brands said to be the sameas A and B respectively but priced at $335.

The pioneering brand was manipulated; subjects saweither A or B. We also manipulated the type of secondentrant; subjects saw either the other distinctive brand,a me-too brand, or no second entrant. Hence we had a2 x 3 factorial design with pioneering brand (A vs. B)and type of later entrant (differentiated brand, me-too,none). Fifty-five MBA students participated in the studyfor course credit.

For the conjoint tasks, eight unnamed brands wereprofiled in terms of the same four attributes. The onlydifference was that a specific fill rating (525 or 575) wasincluded. Each attribute assumed two levels, giving 16hypothetical brands. None of these was identical to brandsA, B, C, and D unless the inferior fill level was asso-ciated with the distinctive brand or the superior level wasassociated with the me-too. The eight brands formed aone-half orthogonal fraction of the full set. Subjects rankedbrands from 1 to 8, 1 being best, and were asked to becareful in ranking both poor and good brands.

Results

To assess the impact of leaming on preferences, wefirst analyzed preferences for brands A and B as a func-tion of attributes. Rankings for the eight hypotheticalbrands were regressed on dummy variables representing

the four attributes for each subject separately, so regres-sion parameters indicate the predicted change in rank re-sulting from a change in attributes."

Pioneering advantage. The impact of pioneering canbe deduced by analyzing how rankings vary with the pi-oneering variable. We calculated rankings for brands Aand B for each subject, assuming a 525 fill rating. Pre-dicted rankings by the type of market pioneer are shownin Figure 8. We see that the combination of the pioneeris always preferred to other combinations, regardless ofwhich brand entered first.

To analyze the underlying data, we calculated the dif-ference in rankings predicted between brands A and Band used it as a dependent variable in an ANOVA withpioneering and the type of second entrant as factors. Theresults show that pioneering is the only significant factor(F|,49 = 20; f> < .001). When brand A is the pioneer,the predicted difference in rank is -1 .75 , indicating thatbrand A would have a lower or more preferred rank.When brand B is the pioneer, the difference is 1.57, in-dicating that brand B would be preferred. Therefore, theresulting preference structure produces a pioneering ad-vantage.

Though the dependent variable here is ordinal, metric analyses ofsuch data are robust, differing little from nonmetric analyses (Car-mone. Green, and Jain 1978; Churchill 1987).

Figure 8PREFERENCE RANKING BY EXPERIMENTAL CONDITION

PredictedRonk

7

6

5

4

Mixed FillCotton

Goose FillSynthetic

Mixed Fill.CottonCover

Goose Fill,SyntheticCover

Pioneer'sAttributes


Attribute weights. We examined the impact of pi-oneering on attribute weights by analyzing differencesin the attribute weights (regression coefficients) acrossindividuals. The two pioneers differed on two dimen-sions, type of fill and cover. For these attributes, meanweight differences are consistent with H2, showing a shiftin favor of the pioneer. Figure 9 shows that for subjectswho saw a pioneer with goose down fill only, changingto mixed fill lowers relative preference by 1.31. Simi-larly, for subjects whose pioneer had a cotton cover,changing to a synthetic cover increases the preferencerank by 1.77. showing a drop in relative preferences.

The impact across all individuals was assessed by anANOVA of regression parameters with pioneering andtype of competition as factors. For these two attributes,the effect of pioneering is the only significant factor (f, 49= 8.12;/? < .01 for fill and F,.49 = 12.63;/? < .001 forcover). Therefore, for both attributes, changing levelsaway from the pioneer decreases preference, supportingH,.

Pioneer versus later entrants. H4 predicts that the im-pact of a competitor price cut will increase as the com-petitor becomes more differentiated. We tested that pre-diction by examining each individual's price coefficientand comparing across experimental groups. We pre-dicted that price coefficients will be greatest (in absolutevalue) when the pioneer is followed by a differentiatedentrant and smallest when the pioneer is followed by noother brand.

Differences in attribute weights for price are consis-

Figure 9DECREASE IN PREFERENCE IN RESPONSE TO CHANGES

IN AnRIBUTES

Decreose InPreference

1

2.0

1.5

1.0

0.5

0

-0.5

Mixed Fill.CottonCover

Goose ToMixed Fill

\

\Cotton ToSynthetic

1

Goose Fill.SyntheticCover

Cover

Pionoer'sAttributes

Figure 10IMPACT OF PRICE CUT ON PREFERENCES FOR PIONEER

Decrease InPreference

2.5

2.0

1.5

1.0

0.5

±None Me-To Distinct Type of

LaterEntront

tent with this hypothesis, as shown in Figure 10. Whenno second entrant was introduced, a price cut producedan increase in preference of .667; when a me-too brandwas introduced, preference increased 1.53; when a dis-tinctive entrant was introduced, preference increased 2.23.These differences are significant, as demonstrated byanalysis of variance (F249 = .368; p < .05). Therefore,the impact of price grows with the differentiation of thesecond entrant, supporting H4.

Discussion

In this experiment we focused on the impact of pi-oneering on the overall preference structure, rather thanon the advantage of a specific brand. As predicted byour learning model, pioneering has a significant effecton preference structure. Derived anribute preferences fa-vor attributes associated with the pioneer, whichever brandentered first. Thus, the pioneer strongly biases categorypreferences through prior exposure and successful out-comes.

This bias produces an advantage for the pioneer, pro-vided penetration and trial are sufficient. Across bothexperimental groups, the pioneer is consistently the mostpreferred brand, yielding, in an actual market, the larg-est share. This fmding is consistent with our initial re-sults in experiment 1.

A differentiated competitor also has the strongest ef-fect on the preference structure, as experiment 1 sug-gests. The more similar a me-too brand and a differen-tiated entrant, the lower the pioneer's share, suggestingthat sufficient "mass" elsewhere in the market could seg-ment it and thus affect the pioneer. In our second ex-periment, the type of second entrant had a similar effect


on the price sensitivity of the pioneer. The pioneer be-came increasingly price sensitive as the second entrantbecame increasingly differentiated. This finding sug-gests that more diverse competition (i.e., greater dissim-ilarity among brands) should produce more price com-petition.

Moreover, it suggests that price is least effective atstealing share from the pioneer for a me-too brand andmost effective for a differentiated entrant. Cross-priceelasticities., influenced by the perceptual prominence ofthe pioneer, therefore may vary inversely with interbranddistances. The implication is that me-too brands fail notbecause of price-cutting as some analysts argue (e.g..Lane 1980), but because price-cutting is ineffective forclosely positioned brands.

IMPUCATIONS AND EXTENSIONSOur theory suggests that pioneering advantage arises

in part because of the impact of early, successful entryon the preferences of buyers. The pioneer frames per-ceptions of the category and profoundly infiuences theformation of preferences when attribute weights and theideal attribute combination are ambiguous. The pioneershifts the preference distribution toward its own positionand becomes prototypical of the category. The result isa superior position for the pioneer and a degree of in-sulation from similarly positioned brands. Our two ex-periments provide some insight into the operation of thatmechanism.

ImplicationsOne important implication of our analysis is that pi-

oneering advantage depends in part on perceptual prom-inence. Achieving perceptual prominence may providetwo advantages to the pioneer. First, it may insulate thepioneer from brands that attempt to copy its position;closer to the pioneer, rivals lose sales. Second, promi-nence may reduce the impact of competitors' prices onthe pioneer's market share. Pioneers appear least sen-sitive to prices of competitors that are most similar tothem, which suggests an inverse relationship betweencross-price elasticities and distance between brands.

This fmding implies that me-too strategies are unlikelyto succeed if attribute weights and the ideal attributecombination are ambiguous. Positioned close to the pi-oneer, me-too brands are overshadowed by the pioneerand the pioneer is least price sensitive. Therefore, me-too strategies are ineffective because they position a brandin the least advantageous location and rely on the leastpowerful competitive weapon-—price. However, seg-menting the market, copying a differentiated entrant ratherthan the pioneer, appears to be an effective late entrystrategy. It enables both late entrants to develop a degreeof prominence, reduce their own price sensitivity, andincrease the price sensitivity of the pioneer.''

'See Carpenter (!987) for a more detailed discussion of the man-agerial implications of this and other mechanisms for producing pi-oneering advantage.

More generally, our theory suggests that evolution ofcompetition in a market and the fortunes of any one branddepend critically on the timing of entry and the actionstaken once entered. The structure of preferences that pre-vails in mature markets may depend on the order of brandentry and the success of the brands. Many preferencestructures are possible, depending on the order of brandentry. As a result, the success of a brand may dependcritically on the sequence of entries—who enters priorand subsequent to its introduction. This view is a sub-stantial departure from the traditional view in which suc-cessful brands cater closely to buyer wants and in whichbrand entry leaves preferences unaffected.

Limitations, Speculations, and Future Work

The primary goal of the studies reported here was totest the plausibility of our theory. The limitations of ourlaboratory setting are obvious and our conclusions mustbe, to a certain extent, speculative. Nevertheless, theyargue strongly for the importance of consumer-basedmechanisms in explaining pioneering advantage.

In addition, our conclusions suggest areas for futureinvestigation. First, pioneering advantages appear to bepersistent, lasting in the case of Wrigley's for more thansix decades. One pxjssible explanation is that buyer pref-erences, once developed, can be viewed as a productschema (Bettman 1986), which can be difficult to alterand persistent over time, even when evidence supportingit is shown to be inaccurate or false (Fiske and Taylor1984). Such a phenomenon is exemplified by the con-tinued success of Coke Classic, even though a majorityof consumers prefer "new" Coke in blind taste tests. Ex-ploring the persistence of these schemas as one expla-nation for the persistence of pioneering advantages wouldbe fruitful.

Second, in many categories dominant brands like Crestin the toothpaste market appear to have many of the ad-vantages of pioneers, chiefly perceptual prominence.Examining how these advantages arise and showing sim-ilarities to the pioneering case would be interesting andimportant work.

Third, developing a theory of pioneering advantagesuggests another important extension^—^a theory of over-taking pioneering brands. How, and under what condi-tions, can pioneers be overtaken? The conventionalthinking seems to be that later entrants must differen-tiate, offer "something extra." Exactly how one does sois unclear. Future research should focus on the require-ments for successful differentiation strategies.

Fourth, our analysis suggests an important new di-mension to marketing strategy—^achieving a competitiveadvantage by influencing consumer tastes rather than re-sponding to them. It implies a process of competitionthat differs markedly from the traditional one in whicha better mousetrap displaces the poorer one. Instead,competition may center on a battle over poorly formedpreferences and perceptions, with the winner receivinga potentially valuable asset—a favorable preferencestructure. The implications of such a process are far-

298 JOURNAL OF /MRKETING RESEARCH, AUGUST 1989

reaching and await much closer investigation of the roleof the mechanism described here in the more complexprocess driving market behavior.

CONCLUSIONWe propose an explanation for pioneering advantage

by examining the role of leaming in the formation ofpreferences. We suggest that pioneering advantage, un-der certain conditions, depends importantly on biases inbuyers' preferences arising from the preference-forma-tion process. When the contribution of attributes and theideal combinations are ambiguous and penetration is suf-ficient, trial of the pioneer has an important role in theformation of preferences for all brands. All are com-pared with the pioneer, the ideal brand is perceived asclose to it, and the pioneer is perceived as prototypical—representative yet competitively distinct. In this situationthe pioneer occupies a favorable perceptual position thatis difficult to imitate and costly to compete against,yielding a powerful competitive advantage.

REFERENCESAdvertising Age (1983). "Study: Majority of 25 Leaders in 1923

Still on Top" (September 19), 32.Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimen-

sions of Consumer Expertise." Journal of Consumer Re-search, 13 (March), 411-54.

Barsalou, Lawrence W. (1982). "Context-Independent andContext-Dependent Information in Concepts," Memory andCognition, 10 (January), 82-93.

(1985), "Ideals, Central Tendency, and Frequency ofInstantiation as Determinants of Graded Structure in Cate-gories," Journal of Experimental Psychology: Learning,Memory, and Cognition, 11 (October), 629-55.

Bettman, James R. (1986). "Consumer Psychology." AnnualReview of Psychology, 37, 257-89.

Bond. Ronald S. and David F. Lean (1977), Sales. Promotion,and Product Differentiation in Two Prescription Drug Mar-kets. Washington. DC: Federal Trade Commission.

Carmone, Frank J.. Paul E. Green, and Arun K. Jain (1978),"Robustness of Conjoint Analysis: Some Monte Carlo Re-sults." Journal of Marketing Research. 15 (May), 300-3.

Carpenter. Gregory S. (1987). "Market Pioneering and Com-petitive Positioning Strategy," Annales des Telecommuni-cations. 42 (November-December), 699-709.

and Kent Nakamcto (1987), "Market Pioneering,Leaming, and Preference," in Advances in Consumer Re-search, Vol. 15. M. Houston, ed. Prove, UT: Associationfor Consumer Research.

Churchill, Gilbert A.. Jr. (1987), Marketing Research: Meth-odological Foundations. Chicago: Dryden Press.

Deighton. John (1984), "The Interaction of Advertising andEvidence," Journal of Consumer Research. II (December),763-70.

Elio, Renee and John R. Anderson (1984), "The Effects ofInformation Order and Leaming Mode on Schema Abstrac-tion," Memory and Cognition, 12 (January), 20-30.

Fiske, Susan and Shelley E. Taylor (1984), Social Cognition.Reading, MA: Addison-Wesley Publishing Company.

Fomell. Claes, William T. Robinson, and Birger Wemerfelt(1985), "Consumptive Experience and Sales Promotion Ex-penditure," Management Science, 31 (September), iO84-105.

Hoch. Stephen J. and Young-Won Ha (1986), "ConsumerLeaming: Advertising and the Ambiguity of Product Ex-perience," Journal of Consumer Research, 13 (September),221-33.

Hotelling, Harold (1929), "Stability in Competition," Eco-nomic Journal, 39, 41-57.

Howard, John A. (1989), Consumer Behavior in MarketingStrategy. Englewood Cliffs, NJ: Prentice-Hall, Inc.

and Jagdish N. Sheth (1969), The Theory of ConsumerBehavior. New York: John Wiley & Sons, Inc.

Kahneman, Daniel and Jackie Snell (1988), "Predicting Util-ity." in Insights in Decision Making. R. Hogarth, ed. Chi-cago: University of Chicago.

Lane, W. J. (1980). "Product Differentiation in a Market withEndogenous Sequential Entry," Bell Journal of Economics,11 (Spring), 237-60.

Lieberman, Marvin B. and David B. Montgomery (1988),"First-Mover Advantages." Strategic Management Journal.9 (January-February). 41-58.

Marks, Lawrence J. and JerTy Olson (1981), "Toward a Cog-nitive Structure Conceptualization of Product Familiarity,"in Advances in Consumer Research. Vol. 8, Kent B. Mon-roe, ed. Ann Artwr. Ml: Association for Consumer Re-search.

Medin, Douglas L. and Marguerite M. Schaffer (1978), "Con-text Theory of Classification Learning," Psychological Re-view. 85 (May), 207-38.

Meyer, Robert J. (1986), "A Theory of the Inductive Leamingof Multiattribute Preferences." in Perspectives on Method-ology in Consumer Research. D. Brinberg and R. J. Lutz,eds. New York: Springer Verlag.

(1987), "The Leaming of Multiattribute JudgmentPolicies," Journal of Consumer Research. 14 (September).155-73.

Prescott, Edward C. and Michael Visscher (1977). "Sequen-tial Location Among Firms with Foresight." Bell Journal ofEconomics. 8 (Autumn), 378-93.

Robinson. William T. (1988), "Sources of Market PioneerAdvantages: The Case of Industrial Goods Industries,"Journal of Marketing Research, 25 (February), 87-94.

and Claes Fomell (1985), "Sources of Market Pi-oneering Advantages in Consumer Goods Industries," Jour-nal of Marketing Research. 22 (August), 305-18.

Schmalensee, Richard (1982), "Product Differentiation Ad-vantages of Pioneering Brands." American Economic Re-view. 72 (June), 349-65.

Schwarz. Gideon and Amos Tversky (1980), "On the Rec-iprocity of Proximity Relations." Journal of MathematicalPsychology. 22 (December), 157-75.

Stigler, George J. and Gary S. Becker (1977), "De GustibusNon Est Disputandum." American Economic Review, 67(March). 76-90.

Sujan. Mita (1985). "Consumer Knowledge: Effects of Eval-uation Strategies Mediating Consumer Judgments," Journalof Consumer Research. 12 (June), 31-46.

Tversky, Amos and J. Wesley Hutchinson (1986), "NearestNeighbor Analysis of Psychological Spaces," PsychologicalReview. 93 (January). 3-22.

and Daniel Kahneman (1974), "Judgment Under Un-certainty: Heuristics and Biases," Science. 185. 1124-31.

Urban, Glen, Theresa Carter, Steve Gaskin, and Zofia Mucha(1986), "Market Share Rewards to Pioneering Brands: AnEmpirical Analysis and Strategic Implications," Manage-ment Science, 32 (June), 645-59.

o. JMR263102

Carpenter and Nakamoto - Consumer Preference

Documents

Transcript of Carpenter and Nakamoto - Consumer Preference