Re-thinking Market Research PDF

Rethinking Market Research: Putting People Back In Author(s): Gerald Zaltman Source: Journal of Marketing Research, Vol. 34, No. 4 (Nov., 1997), pp. 424-437 Published by: American Marketing Association Stable URL: http://www.jstor.org/stable/3151962. Accessed: 07/09/2011 12:34 Your use of the JSTOR...

Rethinking Market Research: Putting People Back In Author(s): Gerald Zaltman Source: Journal of Marketing Research, Vol. 34, No. 4 (Nov., 1997), pp. 424-437 Published by: American Marketing Association Stable URL: http://www.jstor.org/stable/3151962. Accessed: 07/09/2011 12:34 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at. http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marketing Research. http://www.jstor.org  GERALDZALTMAN* Research methods are generally improved through new understand- ings of scientific procedure, validity, and reliability. Variations in these understandings-among knowledge communities as diverse as ethnog- raphers, statisticians, historians, and even practitioners and re- searchers-yield a rich set of innovations in quantitative tools, experi- mental designs, data collection instruments, observational methods, sampling procedures, and interpretive frameworks. Research methods, however, must be consonant not only with the way various communities view scientific inquiry, but also with fundamental characteristics of the thought and behavior of customers and managers. Most widely used methods in marketing lack the latter consonance. The author introduces particular insights about thought and behavior from multiple disciplines as design criteria for improving research methods. Rethinking Market Research: Putting People Back In The development of research methodology should be in marketing.A specific technique illustrates how current guided by knowledge about the nature of the phenomena methods might be improvedor new methods developed. that is studied:the thoughtsand behaviorsof managersand customers. Improving statistical and mathematical tools, PUTTINGPEOPLEBACKIN field and laboratoryexperimentaldesigns, survey designs, The premises that follow provide a sample of the quali- sampling techniques, and research reporting techniques, ties of manager and customer thought and behavior that when they neglect the natureof humanthoughtand behav- need to find expression in research methods. Readers are ior, leaves researchers overly prominent in the research invited to ask first, lest they be inclined to dismiss a premise process. Researchersmust engage managersand customers prematurely,"Would this premise have importantimplica- more actively in the researchundertakingby enabling them tions for conducting researchif it were true?"If the answer to representfully their thinking.This requiresusing knowl- is yes, it becomes appropriateto examine the evidence edge about cognitive and other processes to improve how cited. managersand customers are studied. In this article I review a numberof substantiveinsights Premise 1: ThoughtIs Image-Based,Not Word-Based that,collectively, imply a need to rethinkbasic assumptions Is thought dependenton words?...Or are our thoughts about thought and behavior that underlie existing research couched in some silent medium of the brain-a lan- methods. I argue that many substantiveresearch issues on guage of thoughtor "mentalese"-and merely couched which importantprogress has been made, such as emotion, in words whenever we need to communicatethem to a listener?...The idea that thought is the same thing as metaphor, nonverbal communication, and visual imagery, languageis an example of what can be called a conven- tend to be missed or misrepresentedby currentmethods of tional absurdity....There is no scientific evidence that inquiring into them. Greatersensitivity to the implications languages dramatically shape their speakers' way of of these insights is needed for basic researchmethodology thinking(Pinker 1994, pp. 56-58). Thought arises from images, which are topographically organizedneuralrepresentationsthat occur in the early sen- *GeraldZaltmanis Joseph C. Wilson Professorof Business Administra- sory cortices. When neurons are activated sufficiently, tion, School of Business Administration,HarvardUniversity. The author images can be experienced as conscious thought (Damasio thanksthreeJMRreviewersand Vijay Mahajanfor theirinsightfulandcon- 1994). An image, then, is an internalrepresentationused in structive comments on this work. Chris Argyris, Kathryn Braun, Sam informationprocessing (Kosslyn 1994). Because two-thirds Chun, and Kathy Randel also provided thoughtful critique and advice. of all stimuli reach the brain through the visual system Appreciation is extended to the Division of Research, HarvardBusiness School, for its generous supportof this work and to the MarketingScience (Kosslyn et al. 1990), images often are visual. Verbal lan- Institutefor arrangingearly validationstudies with membercompanies. guage plays an importantrole in the representation,storage, and communicationof thought(Bickerton 1990), but verbal Journal of MarketingResearch Vol. XXXIV (November 1997), 424-437 424 Rethinking Market Research 425 language is not the same as thought (Kosslyn and Koenig According to Gibbs (1992, p. 572), "The vast majorityof 1992). "Conceptualcapabilities develop in evolution well linguistic metaphors reflect underlying conceptualizations before speech" (Edelman 1992, p. 108). of experience in long-term memory that are already struc- This premisesuggests thatby developing methodsfor en- tured by metaphoricalschemes." Conceptual metaphorical gaging and/or monitoring imagic activity more directly, schemes stored in long-termmemory help us make sense of managers and customers can be moved closer to the way literal metaphors(Glucksberg 1991). their thought occurs and thereby provide more complete So basic are metaphorsto the representationof thought representationsof their thoughts and accounts of their be- that communicatorsand audiences alike often are unaware haviors.The imagic activity to be engaged involves both vi- of theiruse and thereforeof the significance of metaphorsin sual and nonvisual sensory mechanisms. the creation and expression of thought (Soyland 1994). Metaphorsnot only help us make sense of what we perceive, Premise 2: Most CommunicationIs Nonverbal but also direct our attentionaland perceptualprocesses (In- Although estimates vary, there is general agreementthat durkhya 1994). Observes Johnson (1987, p. 169), most human meaning is shared nonverbally (Patterson "Metaphoricalprojection is one fundamental means by 1991). A common rule of thumbis that approximately80% which we project structure,make new connections, and re- of humancommunicationis nonverbal(Weiser 1988). Bird- mold our experience." Burgess and Chiarello (1996) and whistell (1970) holds thatno more than 30% of the meaning Bottini and colleagues (1994) provide descriptions of the in a social exchange is conveyed by words, whereasMehra- neurologicalbasis for this. bian ( 1971) reportsthatonly 7% of the meaning in any mes- The process of imagining, that is, creating or arrivingat sage is contained in verbal language, and approximately what it is we know, shapes the content of knowledge. 93% is communicatednonverbally.Much of the meaningof "Withoutimagination,nothing in the world could be mean- verbal language is determinedby nonverbalcues including ingful," observes Johnson (1987, p. ix). "Withoutimagina- haptics (touch), vocalics (paralanguage),proxemics (spatial tion, we could never make sense of our experience. Without cues), chronemics (temporalcues), and oculesics (eye con- imagination, we could never reason toward knowledge of tact, gaze, pupil dilation). In the event of an apparentcon- reality." Metaphors are central to imagination (Goldman tradiction,nonverbal cues tend to be believed over verbal 1986). Withoutmetaphorswe cannot imagine: they are the ones (Knapp 1980). engine of imagination. Verballanguagedeveloped only recently in the context of Sensitivity to metaphorsis thus even more importantbe- humanevolution, and written,phonetic-basedlanguagede- cause managersand consumers are likely to process infor- veloped even more recently (Ong 1982). Growing under- mation metaphoricallyeven when that is not the communi- standingof the role of the basic senses in learningand com- cator's intent. One implicationof this premise is that meth- municationand in the shaping of neural patternsreinforces ods designed to elicit and analyze metaphorssystematically the premise that nonverbal communication dominates could significantly augment knowledge gained from more thought (Howe 1991; Marks 1978). This is consistent with literal,verbocentricresearchapproaches. Premise 1. Most marketresearchtools are verbocentric.They rely on Premise 4: MetaphorsAre Importantin Eliciting Hidden literal language to collect, synthesize, and reportresponses Knowledge to surveys, ideas from focus groups, and literal trace data Metaphors are especially effective at surfacing hidden that is accumulatedby scanners.Even when available, non- knowledge (Glucksberg 1995; Shlain 1991). A number of verbalcues generallyare not processed systematicallyby re- investigatorsin the field of psychotherapyhave found that searchersand are reportedonly infrequently.Premise2 sug- explicit use of metaphorshelps patients make unconscious gests that it is importantto enrich verbocentricmethodolo- experiences progressivelymore conscious and communica- gies with techniques that accommodate nonverbal expres- ble (Ingram1994; Kopp 1995). Because they can elicit cog- sions of perception, learning, and thought that are found in nitive processes beyond those displayed by literal language, manager-and customer-createdvisual and othersensory im- metaphorsalso surface importantmental states that literal ages. This is even more importantbecause nonverbalcues language might altogether miss or underrepresent.In fact, are processed tacitly. metaphorsare storednot verbatimin memory,but abstractly in modality-free language (Burgess and Chiarello 1996). Premise 3: MetaphorIs Central to Thought They are the so-called "ground"-the abstractlyunderstood A metaphorinvolves understandingand experiencingone sharedpropertiesbetween vehicle and topic-that becomes thing in termsof another(Lakoffand Johnson 1980); it is the partof our image schema (Chandler 1991). perceptionof one thing as if it were a differentkind of thing Metaphorsare powerful because they hide as well as re- (Dent-Read and Szokolszky 1993). For example, tempera- veal thoughtsand feelings. The executive who sees him or turequality(hot) might be used to convey popularityor theft. herself as a lawn maintenancespecialist keeping weeds out Metaphors are central to understandingthe human mind of his or her organizationreveals a distancing attitude to- (Allbritton 1995; Honeck 1996). They both invoke and wardnew ideas while hiding the value of cultivatingchange. express nonverbal imagery. "Whatever else we are, we Researchmethods, then, must help reveal what a metaphor humans are metaphorizing animals" (Johnson 1995, pp. might be hiding. Skilled questioningis requiredin the use of 158-159). metaphorsto elicit deep meanings that are rendereddifficult The position that thinking is represented primarily to express or even hidden by standardprocedures.Standard throughmetaphorsis consistent with the interactionistview interview processes and focus group sessions are not that the creation of new thoughts is shaped by metaphors. amenable to such questioning. Given the increased likeli- 426 JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1997 hood of uncoveringimportantmental constructsby analyz- Premise 6: Emotionand Reason Are Equally Important ing metaphors (Forceville 1994), it is importantthat re- and Comminglein Decision Making searchtechniquesdo this. Few dispute the importanceof emotions in managerial Premise 5: CognitionIs Embodied and consumer decision making, yet most researchmethods are biased towardreason.Most methodscollect and present Metaphorsthat are based on bodily systems are centralto informationas if decisions were the result of logical infer- the expressionof abstractthought(Damasio1994).According ence and conscious processes.People areespecially likely to to Marks(1996) and others,the hiddenknowledgeelicited by display this quality of their decision making because it is metaphorsis rooted in physiologically-basedimage schema, what they are asked about, how they are encouraged to though distinguishing physical from cultural bases of respond, and what they can articulatemost readily. How- metaphorscan be difficult(Classen 1993). Johnson(1987, p. ever, reason and emotion are not independentand must be 79) argues that image schemataare recurringpatternsthat treated as commingling forces. Thus, the assumption that arisefromourbodily movementsandmanipulationor percep- decision making is driven largely by relatively overt and tual interactionwith objects;they "arethose recurringstruc- readily inspectedlogical inferenceis half rightand therefore turesof, or in, our perceptualinteractions,bodilyexperiences, also half wrong (Kahneman 1994). Multiple, complex rea- andcognitive operations"thatprovidecoherenceto our expe- soning systems working in concert are essential to normal riences. Thus, terms of juxtaposition(e.g., over) represent decision making. But multiplesystems of emotion also exist social hierarchy,direction,location,and conclusion,whereas (Harris 1993; LeDoux 1996) and are equally importantin color (e.g., blue) and orientation(e.g., down) reflectmood. normaldecision making (Isen 1993). The embodied cognition thesis receives compelling sup- Studies of patients with particularpatternsof braindam- port by researchersin many different areas (Danesi 1990; age reveal that when reasoningsystems are intact but emo- Thelen 1995). Metaphoricprocesses appearto originate in tional capacities damaged, poor decisions result; similarly, when emotional systems are intact and reasoningcapacities neurological substratestied to sensory-motor-affectivesys- tems (Marks 1996; Marksand Bornstein 1987). Ourabstract damaged, poor decisions also result (Damasio 1994; Koss- understandingis organizedby projectingonto it patternsof lyn 1994). The separationof reason and emotion, however convenient, is misleading: "The lower levels in the neural sensory experience (McAdams and Bigand 1993). To un- edifice of reason,"writes Damasio (1994, p. xiii), "are the derstandquantity abstractly,for example, a person might same ones that regulatethe processingof emotions and feel- draw on the physical experience of verticality (e.g., up, ings.... Emotion,feeling, and biological regulationall play a down), which, in turn,might be attendedby judgmentalas- role in human reason." Reason and emotion are not to be sociations ("up"has many more positive associations than considered opposites. Although emotions sometimes ob- "down"). scure this, theircomplexity and biological roots are what en- An interesting approachto the physiological aspect of able them to play a crucial role in rationality (De Sousa metaphorsinvolves synesthesia,or cross-modalityperception, 1987). Emotions also shape the tacit metaphorizingprocess in which one mode of sensoryperceptionis transferredto an- of reasoning,wherebypast experienceis used as a metaphor other(Marksand Bornstein1987). When asked,most people to guide current actions and future plans (Bottini et al. will characterizea sneeze as brighterthana cough. Otherex- 1994). amples includecolored speech perceptionand the association This premise argues for the use of researchmethods that of soft and low pitchedsoundswith darkcolors. Associations enable both reason and emotion to surface,and to do so in a betweensensorymetaphorsarepatterned;thatis, differentin- way that reflects their commingling. It holds, further,that dividualswithinthe same culturedisplaythe same systematic these two categories are each representedby multiple sys- connectionsbetween dimensionsof specific modalities. tems and should not be treatedas opposite each other. ClassenandHowe (1996, p. 87) observethat,"As a basisfor experiencingoneself, one's society, andthe world,the senses, Premise 7: Most Thought,Emotion,and Learning Occur like the body, are always of culturalsignificance,whetherex- WithoutAwareness tensively elaboratedor ostensibly suppressed.Every society Emotions are biological functions of the nervous system; has a sensory order,and to every sensory orderthere corre- differentgroupsof emotions are handledby differentneural sponds a particularsocial and symbolicorder."Differentcul- systems (LeDoux 1996). Feelings, the conscious experience turesapproachthe sensoriumdifferently(Classen1993;Stoller of emotions, are only the tip of the iceberg. Most emotions 1989)andinteractionsamongsensesdifferaccordingly(Howe and cognitive functions, which guide thoughtand behavior, andClassen 1991).Oneculturemightfavorsound/visioncom- occur withoutawareness(Plutchik 1993; Shimamura1994); binations,whereasanotherfavors sound/taste.Moreover,the that is, most mental life is tacit. practicalvalue of a particularsense may correspondto its cul- Cautions LeDoux (1996, p. 32), "We have to be very turalor symbolic importance(Emanatian1995). careful when we use verbal reportsbased on introspective This premise suggests that a richer representationof ab- analyses of one's own mind as scientific data."His and oth- stract thought can be achieved by incorporatinginto re- er observations about the origins of verbal reports(Nisbett searchmethodsthe elicitationand probingof metaphorstied and Wilson 1977) suggest a need for prudencein the use of to motor and perceptualsystems (e.g., the senses) and phys- introspective reports, survey interviews, group meetings, ical movement and sense of spatial orientation.There will and questionnaires.Even the impact of consciously experi- be commonalities in the meaning of even broad,embodied enced stimuli on attitudesand goals is unconscious and thus metaphorswithin cultures. might be missed as a source of influence on thoughtand be- Rethinking Market Research 427 havior (Janiszewski 1988). Other reasons for exercising The frontallobe appearsto have a centralrole in planning caution in the use of verbal report data in the absence of eye movements on the basis of mental maps (Hubel 1988). stimuli specifically designed to activate images can be "We humansdo not just sit thereand wait for informationto found (Seger 1994; Weiser 1993; Ziller 1990). come in; we actively seek informationand actively test hy- Concerns about reliance on standard verbal (oral and potheses. If one has an inkling of whatone is seeing, one can written) reportsby customers and managersare reinforced seek out confirmatoryevidence by looking for particular by the growing body of research on memory distortion parts to observe" (Kosslyn and Koenig 1992, p. 99). "Be- (Schacter 1995, 1996; for an important treatment of the cause our eyes register fine detail only with a very small process in marketingcontexts, see Braun 1997). It is of con- foveal region of the visual field," explains Hochberg(1972, siderable importancethat what is rememberedis a creative p. 65), productof priorexperience, currentbeliefs, and futurepre- conscious plans.This does not mean thatcritical incidentre- we must learnaboutthe visual world by a succession of glances in different directions. Such glances are made porting is not useful. It has been of value in marketing(see by saccadic eye movements, whose end-points are Park, Iyer, and Smith 1989). The research on memory re- decided before the movement is initiated(i.e., saccades construction simply implies a need for caution in imple- are ballistic movements):where one looks is decided in mentingthe procedureand interpretingresults(Russo, John- advance....Where we directour foveas when we look at son, and Stephens 1989). picturesis guided by the hypotheses that are generated One importantfunctionof higher-orderconsciousness, or by what we see in peripheralvision.... The integration awarenessof awareness,is to facilitatethe makingof choic- of the successive glimpses that we receive when scan- es. Consciousness, then, is a central construct for under- ning a picture must depend on our ability to fit each standing decision makingjust as are values, attitudes,con- view into some "mentalmap,"into a cognitive structure cepts, and norms (Graham1993). However, as importantas that stores the information(p. 68). it is, consciousness is the end resultof a largely unconscious This occurs so rapidlythatthere is seldom conscious aware- system of neuralinformationprocessing(LeDoux 1996) and ness of the process. associated implicit learning (Seger 1994). Consciousness Our mental maps are made up of iconic imagery. What and unconsciousnessare naturalbiological phenomenaulti- mately to be explained by natural laws (Chalmers 1996; people attendto always reflects an inner map-a set of neu- Searle 1992), particularlyas they involve neurobiological ronalgroupings-that is used unconsciouslyto organizeand understandsensory experience (Weiser 1993). The most ca- processes relatedto vision and imagery(Edelman 1992; Ze- ki 1993). sual examinationof eye fixation patternsdemonstratesthe Because most mental life is unconscious (Baars 1988), use of knowledge, belief, and expectation to guide eye Premise 7 stresses the importanceof studying in their own movement (Kosslyn and Koenig 1992). What has been de- right unconscious but accessible events and processes scribed for the visual system also applies to other sensory among managersand customers. To attend to unconscious systems that provide our understandingof our world. Thus, but accessible mind states, research methods must engage what we know-our mental models-and how what we people in ways that enable them to bring unconscious states know is representedby metaphorliterally influences what to a level of awareness.Metaphorsare basic mechanismsfor we sense, and what we sense influences what we know. doing this. Mental models are themselves metaphorsfor context-spe- cific pattems of neural associations that influence how we Premise 8: Mental Models Guide the Selection and attendto and process informationabout those contexts. Processing of Stimuli Mental models are not private affairs (Gergen 1994). Sperber(1994, pp. 53ff) speaks of an "epidemiologyof rep- Thought occurs as topographically organized neurons fire. These groups of neurons represent basic constructs resentation,"whereby competing informationis introduced into a common environmentin which multiple factors influ- (Damasio 1990). Sufficient activationof one neuronalgroup will cause other groups with which it has established path- ence its likelihood of successful social distributionand re- tention. If successful, the informationbecomes the basis for ways to become active; one thought leads literally to another, which may lead back again to the earlier thought. culturally shared mental models at local and sometimes Sets of connected neuronalgroups constitutemental models global levels (Cole 1996). Domain-specific mental models (Eimas and Galaburda1990). The cerebralcortex alone (an are thus communally shared and change as individual area that, if made flat, would be aboutthe size and thickness sources of variationgain social prominencethrougha com- of a table napkin)contains about ten million billion neurons petitive process of contagion (Bargh 1990; Scheper and with an even greater number of opportunitiesfor connec- Faber 1994). tions (ten followed by millions of zeros). An extraordinary The implicationof this premise is that researchmethods number of mental models can be created and maintained. need to go beyond identifyingrelevantconstructs.They must These mental models are largely dormant;when they are identify associations among these constructs as expressed active, it is usually at a level below consciousness. This initially by the people researchersstudy. Because subtle and reflects our limited capacity to be conscious of more than not so subtledifferencesmightexist in the thoughtprocesses one cluster of thoughtat a time and explains why verballan- thatconnect the same pairof constructsin differentpeople, it guage, which enables us to focus on finite informationto is importantthat associations among constructs be elicited keep from getting overloaded, only has the appearanceof directly from customers and managers.Like constructs,the being dominant. As noted subsequently, mental models thoughtsthatrelatethem to one anothercan have differentdi- interactwith various perceptualsystems, such as vision. mensionsthatcannotbe providedby quantitativetools alone. 428 JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1997 Premise 9: DifferentMentalModels May Interact methods that explore them.1 The technique involves an People's mental models describe, for example, their intensive exploration of self-generated metaphors on the approaches to ill-structured problems, their dispositions partof managersand customers. towardsnackfoods, and whatthey look for in a workout.An individual person has thousandsof such models (Schacter Implementationof the Procedure 1996), each of which contains neuronalgroups (that repre- Upon qualifyingfor participationin a project,participants sent constructs) linked by one or more pathways to other are providedwith a set of instructionsand guidelinesfor col- neuronal groups. For example, the mental model for each lecting images, and a two-hourinterviewis scheduled.Top- topic area just mentioned may contain a neuronal cluster ics and participantsamples vary broadly.For example, par- representingthe construct "escape."The similarity of this ticipantsin a recently completed projectfor a manufacturer construct across models accommodates what Edelman of medical equipment needed to have a chronic medical (1992) calls reentry, a process whereby the construct of problem,be 65 or more years of age, and live independently. "escape"as it pertainsto snack foods might activate (or be Some consumer topics have included how women picture activatedby) the notion of "escape"pertainingto a workout. theirday, marriedcouples' thoughtsand feelings aboutusing Thus, a sufficiently strong activationof a mental model in the telephone at home, and the experience of using com- one context may activate throughreentry(their sharingthe puter-mediatedcommunicationfor personal ends. Among construct "escape") a mental model in another context. managers,topics have includedhow ill-structuredproblems are approached,views on the firm's majorchallenges, how Moreover,the mannerin which a model for snack foods is clients see the company, how the futureof customerbehav- activated might be influenced by other coactive models. ior is envisioned, and how marketingstrategyis developed. These models, then, are fluid and connected, not fixed and Because a morecomplete understandingof customersand isolated (Millgram 1997). The process of reentry suggests a more fundamentalor managersrequirestools that engage their nonverbal,espe- cially visual, channels of thought and communication,par- global neuronal structure. For example, a generic mental model of "escape"might be understoodin terms of its own ticipants are asked to take photographsand/or collect im- connected sets of neuronal structures."Escape"as a con- ages from magazines, books, newspapers,or other sources thatexpress for them the meaningof the researchtopic. Par- structmight have multipledimensionsconnectedby various ticipantsare usually given 7-10 days to collect images. This pathways:there might be different versions of the generic time period allows people with less immediateaccess to vi- model of "escape"for differentcontexts.This is virtuallyas- sual images time to find meaningfulstimuli. Having partic- sured by the plasticity of the brainand by the fact that dif- ipants collect pictures they define as relevant gives them ferent pathways are involved in each instance of recall that controlover the choice of stimuli used in the interview.Par- leads to the retrievalof technicallydifferentconstructs.The ticipant control yields several benefits. Participant-generat- deconstructedversion of "escape"for snack foods would re- ed picturesare richerin meaning because what the eye per- semble, but not be the same as, the version of "escape"in- ceives and encodes when viewing images such as magazine volved in what people look for in a workout. advertisementsis guided by existing customer knowledge, Constructs arise from neuronal clusters that take on beliefs, or expectations(i.e., their mental models). So rapid- meaning by virtue of their association with otherconstructs ly does this occur that there is seldom conscious awareness (neuronalclusters). Pathwaysthus become importantto es- of the process. Using researcher-suppliedpictures inhibits tablishing the meaning of "escape."The more a particular the richness of this stimulus selection function of mental cluster (e.g., generic "escape")is operativeamong different models. Having participantscollect stimuli increases the mental models, the more importantthatconstructis in gen- likelihood thatimportantbut previouslyunconsideredissues eral, even if its precise meaning varies from one model to will be uncovered.And by affordingparticipantstime in ad- another.Because such a constructis so centralto thought,it vance of their interview to process implicitly the images is importantthat researchmethods be capable of establish- they deem relevant, the pool of importantconstructsto be ing core constructs. The core constructs identified in this surfaced during the interview expands. This processing is way might representdeep metaphorsthat are powerful, so- likely to be unconscious, though the meanings that result cially shared,and orienting thoughtstructures. can become explicit duringthe interview. Premises8 and9 implya needfor methodsthatfacilitatethe constructionof socially sharedmentalmodels.These methods An Illustration:ApproachingIII-StructuredProblems mustfacilitatethe elicitationof hiddenas well as moreevident An eight-step interviewprocess with illustrationsfrom an constructsandenable people to representthe thoughtprocess- ongoing projectabouthow senior executives and academics es thatconnect constructs.Premise9 stressesthe importance approachill-structuredproblemsis described subsequently. of identifyingcore constructsthat might be involved in kin- Ill-structuredproblemsare defined as nonroutineproblems dred mental models (e.g., consumptionof snack foods and that have no clear best solution. Their precise naturemay be motivationsto workout) thatmay influenceone another. unclear;it may only be evident that a problemexists. This HELPINGCUSTOMERSAND MANAGERS JTheZaltmanMetaphorElicitationTechnique is available for use with- REPRESENTTHOUGHT out restrictionfor academic researchpurposes by teaching faculty. Train- A specific technique that incorporates the foregoing ing materialsare availableand I am pleased to provide additionalguidance to other academics. The technique is patented, and nonacademic use premises and related insights into a researchmethod is pre- requirespriorauthorization.The techniquehas undergonea series of vali- sented in the spirit of encouragingstill other ways to build dations studies and has been tested with more than 20 firms and 2500 cus- substantiveknowledge about thoughtand behaviorinto the tomers and managers. Rethinking Market Research 429 topic should be of interestto JMR readersfor the following that,he or she is asked to sort them into meaningfulsets and reasons: (1) managers are an important focus of market indicate whether any pictures in a given set say the same research;(2) there is little groundedresearchon this topic; thing and could be put aside. (3) important ill-structured problems often involve cus- Step 4: Constructelicitation. The inclusion and design of tomers; and (4) understandingcustomers depends on more this step is based on the fundamentalimportanceof catego- than customer data-it depends to a significant degree on rizationto comprehendingexperience (Lakoff 1987; Rosch how a manager thinks (i.e., on his or her habits of mind 1978). It also is based on evidence that people possess self- when approachingill-structuredproblems). Understanding reflective mechanismsfor evaluating and reevaluatingtheir the mind of the customer requires an understandingof the state of mind and ongoing processing (Koriat 1994; Nelson mind seeking the understanding.Studyingapproachesto ill- and Narens 1994). The Kelly RepertoryGrid technique has structuredproblems is one way of gathering insight about been used to demonstratethe presenceof these mechanisms the managerialmind. even among preschool children (Gelman and Wellman This example involves a fairly abstractprocess thatis im- 1991). A modified version of the Kelly RepertoryGrid is portantto the participants,but to which they had previously used with ladderingtechniquesto elicit constructs(Gengler given little, if any, attention.To date, 36 executives and 6 and Reynolds 1995; Kelly 1963). These established meth- academics have been interviewed about their approachto ods are adaptedto enhance the methods' ability to identify ill-structuredproblems.During the interview, every partici- and contributeto the understandingof the basic constructs pant has reporteddiscovering ideas central to his or her ap- throughmetaphor. proach that were not evident before. They have also ac- The interviewerrandomlyselects three of a participant's knowledged the unexpected prominence of emotional fac- picturesand asks how any two are similar and yet different tors in the way they addresssuch problems. from the third with respect to their relation to the research Step 1: Storytelling.Participantsare asked to describethe topic. Initially this surfaces one or two constructs. Various salient content of each picture they bring to the interview. questions are used to elicit other constructs that are the Because human memory and communication are story- antecedentsand/orconsequencesof the initial constructs.In based (Schank 1990), and because participantshave invest- response to three images consisting of a child diving off a ed time and energy in thinkingaboutthe topic and collecting high boardinto a pool, sharkschasing a swimmer,and a per- appropriateimages, they come to the interview with a par- son lost in a maze, an executive indicatedthat the lattertwo ticular story they want to tell. During this step the inter- pictures were similar with respect to his approach to ill- viewer probes for the deeper meaning of the visual structuredproblems in that they indicated anxiety and a metaphors. Pictures, then, should serve as effective entry desire to escape, whereas the child diving representedcon- points for exploring customer concepts (Ball and Smith fidence and a sense of direction. The process of randomly 1992; Weiser 1988). Picturestypically representbasic con- selecting three pictures and laddering on the constructs cepts thatembody extensive informationand defining attrib- elicited continues until the constructs that are surfaced utes. The stories that accompanyvisual metaphorsare high- become redundant.Normally, four triads are required to ly revealing;they crystallizethe essence of a pictureas a rep- reach this point. resentationalmedium for bundles of related thought (Dis- Step 5: Metaphorelaboration. This step relies on proce- sanayake 1988, 1995; Schank 1990). Frequently,the salient dures used in art therapy.The intervieweruses specific cri- or relevant content of pictures identified by participantsis teria to select two or three images. A participantmight then not evident to their interviewers.In describingher approach be asked to imagine widening the frame of one of the pic- to ill-structuredproblems,a vice presidentof marketingfor tures in any direction or dimension and to describe what a beverage company ascribedto a pictureof a personcross- would enter the picturethat would reinforce (or sometimes ing a makeshiftrope bridge in an OutwardBound-like set- contradict) its meaning for them. Participantsare usually ting the importanceof trustingother people in highly uncer- asked a variety of questions that furtheralter the picture. tain situations.Otherideas she associated with this construct These questions are determinedin advance by the research relatedto anxiety, confidence, failure, and recovery. team dependingon the researchgoal. The participantis en- Excellentdiscussions of the substantivevalue and the reli- couraged to explore additional thoughts and feelings em- ability and validity of research using pictures in consumer bedded in the way he or she has chosen to alter the images. behavior,sociology, psychology, psychotherapy,and anthro- This process of visual elaborationdisrupts the equilibrium pology abound(e.g., Collier and Collier 1986; Denzin 1989; established by the pictures and stimulates emotional re- Grady 1996; Prosser 1996; Wallendorfand Amould 1991). sponses that create existential issues for the participantthat Given that pictures are powerfully expressive, thoughtsare involve constructsin both conscious and preconsciousstates image-based,and visual stimuli is so prominentin the brain, (Leyton 1992; Plutchik 1993). the high value of picturesin researchis not surprising. A research director for a telecommunicationsorganiza- Step 2: Missed images. Because pertinentpictures may tion broughta photographof a setting sun with a richly col- not be available within the time period of the assignment, ored sky to reflect the feeling of peace and serenity at suc- participantsare asked to describe pictures they wanted to cessfully resolving an ill-structured problem. Asked to find but could not. This enables people with less access to widen this picture, he added, to his surprise, a number of images to specify desired stimuli. Most participantsdo not unpleasantfeaturesin the foregroundto illustratethe messy, reportany missed images; if they do, a drawing "note"of difficult things that others do not see that must be done these images is made and used as partof the basic stimuli. before the beautifulsunset occurs. This surfacedfeelings of Step 3: Sorting. Participantsare invited to bring between frustrationfrom trying to define the problem and getting 12 and 15 images; if a participantarrives with more than "buy-in"from others to a particularproblemdefinition. 430 JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1997 Step 6: Sensory images. Sensory metaphorsare powerful then you pop out the other side and instead of running, mechanisms for bringingunconscious thoughtto a level of screaming,fearful, you're running,screaming,happy. awareness at which verbal articulationcan occur (Lakoff 1994; Turner 1994). Although the senses (taste, touch, Step 8: The digital image. In the final step the participant creates a summary image or montage that expresses the smell, color, sound, emotional feeling) vary in their ability to be recreated (Engen 1991), all help recall experience topic understudy. This step, like Step 7, reflects the notion that a person's assumptions influence interpretationsby (Damasio 1994). A sensory event "is integrated into the mental representationof an experience" (Engen 1991, p. "forecasting"how a story will end (Collins 1991). The story told in the digital image brings forthassumptions,framesof 114). In this step, participantsare asked to use nonvisual senses to convey what is and is not representativeof the con- reference, and decision rules, some expressed for the first time duringthe creationof the digital image and some raised cept being explored. Participantsprovidea rich and consistentset of responses duringearlier steps. Creationof the digital image actively distortsand violates duringthis step, and patternsemerge across participantsin a the previously symmetricimages. It is precisely this process given project.Emphasishere is not on specific responses,but ratheron understandingwhy or how responsesrelateto a par- of recreationthroughdistortionthatelicits new ideas, inten- ticular research topic. Linguistic analyses of imagery and sifies feelings, and deepens insights (Leyton 1992). When their reconstructedimages are complete, participantsare understandingthe hedonics involved can provide important additionalinsights aboutcustomerthinkingand behavior. asked to tell the stories they represent as if they were In the ill-structuredproblemproject,one executive indi- explaining the images to someone who had not been present cated that her approachto ill-structuredproblems had the duringany partof their interviews. Because creative capacity is far more basic and inherent physical feeling of being on a roller coaster. Another indi- as a cognitive process than is generally acknowledged cated that his approachwas not the color light blue because that was too peaceful. A thirdperson said it was "the taste (Finke, Ward, and Smith 1992; Smith, Ward, and Finke of aspirin dissolving in your mouth when you didn't take 1995), virtually every participanthas a successful experi- ence in this step. This step necessitatesthe involvementof a enough water;once you started,you had to follow through." Step 7: The vignette. In this step participantsare asked to specialist in digital imaging techniquesto serve as the par- imagine a shortmovie thatdescribestheirthoughtsand feel- ticipant's"hands."The participant'spicturesare scannedin- to the computer,and image managementsoftwareis used to ings about the topic. This step is based partly on ideas in- create the montage.The focus of this process is to help par- volving psychodramaand partlyon the fact thatdifferentar- eas of the brain are active when engaging moving rather ticipants express their thinking (Lanham 1993) ratherthan than still images (Hubel 1988; Zeki 1993), which suggests to develop an image that is aesthetically pleasing. Techni- that different ideas might emerge with the activationof dif- cians are trainedto avoid leading participantswhen convey- ferent areas of the brain (Collins 1991). Different brain ing what the technology permits. However, they do prompt structuresand functions contributedifferentiallyto the cre- participantsto explain the relevance of particularimage ation of the mind and the thoughtsgeneratedtherein.Final- editing decisions. For example, the orientationof, proximi- ly, this step activatesthe voyeur's gaze, which is directedto- ty to, and consonance with otherelements of a cutoutplace- ward "a valued, culturalend and structuredby personaland ment can be quite meaningful.The overall structuralpattern social motives" (Denzin 1995, p. 1). New constructsand as- might hold significance, as might the sequence in which im- sociations among constructsoften are elicited in this step. age elements are introducedinto the collage. The interview- Because of the added dimension of movement, this step er seeks to understandthe meaning. accommodates expressions of embodied cognition that are One executive's digital image and descriptionare present- related to biological motor systems and spatial orientation. ed in Figure 1. To presentmore fully the vocalic quality(par- The research directorof a major consulting company con- alanguage)of the description,it is usually given in the par- structedthe following vignette: ticipant'sown voice on video tape or interactiveCD-ROM. It would startwith me running,screamingdown the CONSTRUCTINGTHE CONSENSUSMAP streetat thethoughtof solvinganotherproblem...."No, A constructby itself possesses little innatemeaning.Con- no! I don't want to do it!" I want to run away from it. structs acquire meaning primarily through causal associa- Before you go in there,you just thinkit's darkand scary tions with other constructs (Edelman 1992; Gergen 1994b). and unpleasant and very stressful-lost sleep, lots of work, lots of hours. But it's not so much that you're Consequently,the data must be aggregated(see Laddermap afraidyou won't be able to solve the problem-it's just for an effective software procedure). Transcripts, audio thatyou're afraidto solve it becauseof the work and the tapes, images, and interviewers' notes are examined for con- intensity of the effort required.But then I force myself structs and construct pairs. The resulting consensus map to enter the land of the problem,going inside the pyra- depicts the most important set of constructs and the connec- mid. This is the incubator.This is where you have all tions among them that influence customer and manager per- the intensity and concentrationand focus and you're ception, understanding,and behavior. just really working on the problem. Once you get in A rule of thumb used in developing the final consensus there, you're no longer afraid because you're so map is that a given construct must be cited by half or more wrappedup into it and there's a sense of enjoyment in of the participants in a project and be associated directly actually doing it when you go througha process of ges- tation for a couple of days where there might be... all with another such construct by one-third of the participants. kinds of things streaminginto this place so they're all A completed consensus map usually includes between 25 like thought rays and workingrays and stress rays. And and 30 constructs and represents 85% of the constructs sur- Rethinking Market Research 431 Figure 1 SOLVING ILL-STRUCTUREDPROBLEMS Transcriptand Imagefor StephanHaeckel* This is a representationof the way I thinkabout messy problems at two levels: generally, as a humanbeing, and specifically, as Steve Haeckel. Humansextractsignals from theirenvironment (represented by the background photo of the universe) (anomaly detection) and filter them througha set of concepts and perceptions(frames of reference). The carousel is a metaphorfor the brain, and the slides in it symbolize individual percepts and concepts that extract specific patterns (pattern recognition) and colors from the "whitelight"of signals constantlybombarding us. This reduces data glut by several orders of magnitude and is the first step in creating meaning. The patterns are then interpretedby mental processes (frames of reference) in a manner governed but not dictated by our DNA (representedby the extract from a print-outof part of the human genome). We translate these interpretationsinto action (solution formulation) by making decisions and acting upon them, thereby creating real world objects (symbolized by the Step Pyramid) and/or intellectual objects (exemplified by the excerpt from the score of Beethoven's Ninth Symphony). At a personal level, the universe representsmy habit of trying to get the largest, most expansive context to frame an ill-structuredproblemwhen it presents itself (problem sensing, problem definition). The universe is also meantto reflect a disconcerting awareness that this perspective is only one of an infinity of possible perspectives, and unique in its particularsto me (values, di- stance). Next, I test this overarching structure with several different filters (the individualslides in the carouselof my mind) (switching rules) that I have accumulatedthroughexperience, borrowed from others, and developed by learning. These slides are selected from all of the slides I have collected on the basis of intuitive and subjective judgments about their quality and relevance. The results, projectedon the "screen"of my particular DNA-governed consciousness, are a set of candidate understandingsand conclusions about the natureof the problem. I then typically force myself to articulatethese conclusions in writtenform-which is why I used the textual representationof DNA ratherthan a more graphical helix icon. It has been my experience since elementaryschool that this is the all things foul would wear the brows of grace, yet outcome is a result of conscious and subconscious most reliable way for me to make a self- grace must still look so." (Left on the cuttingroom choices made along the way about (1) what assessment of the quality of my thinking (di- floor in the editing process leading to the signals to pay attention to (anomaly detection, stance, integrity). It seems that I'm not aware of composition of this digital image is a close-up problem-sensing), (2) what frames to use in what I'm thinking until I say it, not sure of what photograph of Grace Kelly's eyebrows. So filtering meaning out of these signals (frames of I'm saying until I write it, and it's often only when importantdoes thatquote seem to be in explaining reference, problem definition), (3) the I've readwhat I've writtenthatI'm sorryor happy how I deal with messy problems-i.e., trying not mechanisms internalto me for deciding what the that I thought it. to reject the obvious and the apparentlyinsincere highest order potential solution is (integrity, An importantcriterion for me in making this or banal out of hand-that I actually contrived to values, switching rules), and (4) acting on that assessment is how big the idea seems to be smuggle it into the ZMET interviewby using this decision by expressing it as an idea (reflective (passion). Some of the most intrinsically terriblevisual pun.) intelligence, trust). satisfying and exciting moments of my life are So I strive for the biggest idea possible The fact that I chose Beethoven's Ninth and associated with encounters with huge ideas (passion), which on reflection may explain why I Imhotep's 4700 year old pyramid-the first and (reflective intelligence). I can remember the feel compelled to search initially for the largest still one of the largest man-made artifacts-is a place and people around me when I first was possible context-the broadest canvas on which reflection of my personal conviction that these exposed to the hubris and implications of the to paint. But the result is always a textual/verbal rank among the very largest and most successful second law of thermodynamics,and the insight expression of a conclusion about the essential ideas in human history for solving importantill- encapsulatedby the line from Macbeth, "Though natureof the problemand how to deal with it. The structuredproblems. *StephanHaeckel, Directorof StrategicStudies, IBM Advanced Business Institute.Printedwith permission. 432 JOURNAL OF MARKETINGRESEARCH, NOVEMBER 1997 faced by any one participant.At most, datafrom four or five help of a technicianjust as the digital image is now con- participants,randomly selected, are generally required to structedas partof the interview.The use of animationin this generateall of the constructsin a consensus map (Zaltman process may permitmore complete expression of embodied and Coulter 1995). This should not be surprising.The mind cognition. is not the possession of the individual.It grows from inter- The researchtechniquediscussed here is labor intensive, personalassociations and other interactionswithin a socio- and it is likely that other efforts to build on these same cultural world (Bargh 1990; Gergen 1994a; McClamrock premises might share this quality initially. A major chal- 1995; Resnick 1991; Sperber 1994). These associations lenge is how to implementthese methods and develop ways arise in a sociocultural context that produces commonly of analyzing the resultingdata more efficiently without sac- sharedmeaningsfor objects or events found withinthatcon- rificing effectiveness. A relatedissue involves the represen- text (Clemen and Winkler 1985; Morrisonand Schmittlein tativeness of the relationshipsarising between constructs.It 1991). Emotionsalso are socially constructed(Oatley 1993; is suggested here that these relationshipsrepresenthypothe- Saarni 1993). Therefore,the constructsin a consensus map ses whose representativenessof a largerpopulationmust be that are based on a small sample of participantscan be rep- establishedin other ways. Validationstudies to date support resentativeof a largerpopulation.At the same time, the re- the contention that the constructselicited are representative lationshipsbetween constructsmust be viewed with caution of a much largerpopulation(appropriateto the sample). In- and tested with a largersample. sights about the representativenessof the associations be- Figure 2 presents a subset of the consensus map of key tween constructshave been encouragingbut still are inade- constructs,based on 36 interviews from the approachto ill- quate. It is yet to be determinedhow many people inter- structuredproblems study. The map reflects the thoughts viewed with this methodneed to specify a given relationship and feelings of senior executives and academics who have between constructs for that relationship to be considered substantial experience addressing ill-structuredproblems. representative.The numberis likely to be in excess of that Less experiencedmanagerswould likely producea different which is needed to establish relevantconstructs.Some vali- map. A discussion of the importantrelationships among dation studies have involved comparisons with large sur- these constructsis not feasible here. Normally, these maps veys (in excess of 35,000 respondentsin one case), but the are presentedelectronicallyso thata personcan "click"with survey dataprovidedso far only showed relevantconstructs, a mouse on a constructor connection between them to gen- not associations or pathwaysamong them. erate a sample of illustrative images accompanied by de- scriptions of their meaning delivered in participants'own New Applications voices. The use of physiological measuresof consumerresponse is familiar to marketing (Bagozzi 1991; Olson and Ray FUTURERESEARCHDIRECTIONS 1983; Rothschild and Hyun 1990). Electroencephalograph Furtherresearchneeds to follow several directions. One (EEG) measures, skin reactance, vocalics, and pupillogra- generaldirectionconcernsfurtherresearchand development phy, for example, have been used with varying degrees of with respectto the premisesand steps discussed here. A sec- success. Newer technologies such as positron emission ond general directionconcerns new uses of these ideas and tomography(PET) scans and functionalmagneticresonance procedures.Both are discussed subsequently. imaging (fMRI) have creatednew opportunitiesto measure with far greaterprecision than previously possible, mental ImprovingMetaphorElicitation events that were not amenable to ready observation until Considerablymore research is needed into the complex recently (Cabeza and Nyberg 1997; Kosslyn et al. 1993). interplayof verbaland nonverbalrepresentationsof thought New technologies expected in the next few years, coupled just as more is needed in how best to leverage the image with the rapidaccumulationof knowledge aboutbrainstruc- basis of thought. For example, the technique used here for ture and functioning, promise even more exciting opportu- illustrativepurposes stresses visual stimuli. It may be that nities to act on several of the premisesdiscussed at the out- for some populationsand topics, such as entertainmentor set of this article. transportation,stress should be placed on other perceptual In a study now underwayat HarvardUniversity and the systems, such as audition.People might collect examples of MassachusettsGeneral Hospital, Stephen M. Kosslyn and I music and other sounds as the basis for the initial story- are using PET scans to assess the impactof threealternative telling. Also, more work is possible with respect to the marketing stimuli (relating to automobile dealerships) de- analysis of data from Step 6 involving sensory metaphors. veloped by Lewis Carboneof ExperienceEngineeringfor a This would be especially appropriatein cross-culturalstud- division of GeneralMotors.The constructsinvolved in these ies. To date, little work has been done using this technique stimuli (aboutwhich study participantsalso complete a writ- (and Step 6 data in particular)in othercountriesor with cul- ten questionnaire)include anxiety, trust,and comfort. turallydiverse groupson a comparativebasis. The technique Subsequentwork will use PET and fMRI to identify and is based on processes that are universal in a culturalsense, assess constructs, monitor sequential events (forward and though the specifics uncoveredin differentculturalsettings backward) consistent with Premises 8 and 9, and identify (e.g., the type of metaphorsused, the meanings they hide and evaluate salient metaphors.It also will be used to eval- and reveal) would likely differ dependingon the topic. Step uate the salience, character(positive, negative), and memo- 7, the vignette or movie, representsanotherfruitfuldirection rableness of metaphorscontained in digital images created for research and development. Currently,people verbalize in Step 8, vignettes developed in Step 7, and other images, the movie scenes they imagine. Newer softwaremay permit such as those articulatedin Step 6. PET and fMRI present the participantto constructa vignette on a computerwith the special opportunitiesto identify deep or root metaphorsthat Rethinking Market Research 433 Figure 2 PARTIALCONSENSUS MAP: MANAGER'S APPROACH TO ILL-STRUCTUREDPROBLEMS ANOMALY DETECTION TRUST - I PROBLEM SENSING DI-STANCE i Vt STRESS ~ V^