Destination Choice Sets

23
DESTINATION CHOICE SETS An Inductive Longitudinal Approach Alain Decrop Louvain School of Management, Belgium Abstract: Choice or consideration set formation focuses on how consumers deal with the multitude of brands that are available in many product categories. This paper investigates the formation and evolution of destination choice sets over time through a longitudinal qual- itative study of vacation decision making. A typology of seven choice sets is presented (i.e., available set, awareness set, dream set, evoked set, exclusion set, surrogate set, unavailable set); the emergence process and connections of those sets are then explained. Choice sets prove to be continuous and to undergo turnarounds. The paper indicates that, to a large extent, final destination choice is driven by constraints and opportunities. Keywords: con- sumer behavior, choice sets, destination evaluation, interpretive research. Ó 2009 Elsevier Ltd. All rights reserved. INTRODUCTION Business research and practice often presents consumers’ decision making (DM) as a sequential process involving a series of steps from need recognition (awareness) to final choice (purchase) through the evaluation of products. In other words, decisions arise from cognitive, affective, and conative stages that lie at the core of hierarchy-of-effects (e.g., Lavidge & Steiner, 1961) and most consumer behavior models (e.g., Engel, Kollat, & Blackwell, 1973; Howard & Sheth, 1969). Researchers may use product alternatives and/or attributes as a refer- ence point when working with such models. This paper focuses on alternatives. Howard (1963) was the first to suggest that alternatives are grouped in an evoked set which includes the brands consumers consider acceptable for the next purchase. More broadly, consider- ation or choice set models focus on the way consumers first consider product or brand alternatives and then evaluate them in order to come to a final choice (Hauser & Wernerfelt, 1990; Howard & Sheth, 1969). These models assume a funneling categorization process through which consumers narrow the number of brand alternatives they are aware of down to a single choice. Choice set (CS) is used throughout Alain Decrop is Professor of Marketing at the Louvain School of Management and a member of CeRCLe (Center for Research on Consumption and Leisure) at the University of Namur, Belgium (Rempart de la Vierge 8, 5000 Namur, Belgium. Email: <alain.decrop@ fundp.ac.be>). His major research interests include consumer decision making and behavior, qualitative interpretive methods, and tourism marketing. Annals of Tourism Research, Vol. 37, No. 1, pp. 93–115, 2010 0160-7383/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. Printed in Great Britain doi:10.1016/j.annals.2009.08.002 www.elsevier.com/locate/atoures 93

description

destination

Transcript of Destination Choice Sets

  • Alain Decrop is Professor of Marketing at the Louvain School of Management and amember of CeRCLe (Center for Research on Consumption and Leisure) at the University ofNamur, Belgium (Rempart de la Vierge 8, 5000 Namur, Belgium. Email: ). His major research interests include consumer decision making and behavior,qualitative interpretive methods, and tourism marketing.

    Annals of Tourism Research, Vol. 37, No. 1, pp. 93115, 20100160-7383/$ - see front matter 2009 Elsevier Ltd. All rights reserved.

    Printed in Great Britain

    doi:10.1016/j.annals.2009.08.002www.elsevier.com/locate/atouresneed recognition (awareness) to final choice (purchase) through theevaluation of products. In other words, decisions arise from cognitive,affective, and conative stages that lie at the core of hierarchy-of-effects(e.g., Lavidge & Steiner, 1961) and most consumer behavior models(e.g., Engel, Kollat, & Blackwell, 1973; Howard & Sheth, 1969).Researchers may use product alternatives and/or attributes as a refer-ence point when working with such models. This paper focuses onalternatives. Howard (1963) was the first to suggest that alternativesare grouped in an evoked set which includes the brands consumersconsider acceptable for the next purchase. More broadly, consider-ation or choice set models focus on the way consumers first considerproduct or brand alternatives and then evaluate them in order to cometo a final choice (Hauser & Wernerfelt, 1990; Howard & Sheth, 1969).These models assume a funneling categorization process throughwhich consumers narrow the number of brand alternatives they areaware of down to a single choice. Choice set (CS) is used throughoutDESTINATION CHOICE SETSAn Inductive Longitudinal Approach

    Alain DecropLouvain School of Management, Belgium

    Abstract: Choice or consideration set formation focuses on how consumers deal with themultitude of brands that are available in many product categories. This paper investigatesthe formation and evolution of destination choice sets over time through a longitudinal qual-itative study of vacation decision making. A typology of seven choice sets is presented (i.e.,available set, awareness set, dream set, evoked set, exclusion set, surrogate set, unavailableset); the emergence process and connections of those sets are then explained. Choice setsprove to be continuous and to undergo turnarounds. The paper indicates that, to a largeextent, final destination choice is driven by constraints and opportunities. Keywords: con-sumer behavior, choice sets, destination evaluation, interpretive research. 2009 ElsevierLtd. All rights reserved.

    INTRODUCTION

    Business research and practice often presents consumers decisionmaking (DM) as a sequential process involving a series of steps from93

  • why such a rediscovery is necessary. First, extant literature on desti-nation CSs deals much more with the question as to why consumers

    94 A. Decrop / Annals of Tourism Research 37 (2010) 93115move to simplify and limit the number of brands they consider thanwith how consumers form their CSs (Brisoux & Laroche, 1981; Gruca,1989; Hauser & Wernerfelt, 1990). Moreover, many studies focus onthe structure and size of sets (for reviews, see Hauser & Wernerfelt,1990; Laroche & Parsa, 2000). In contrast, the literature is largely silentabout the process of forming a CS. Laroche, Kim, and Matsui (2003)attribute this paucity to the lack of sound theoretical bases and tothe difficulty to research such a topic. As a consequence, the categori-zation and evaluation processes underlying CSs remain largely unex-plored: It is unclear how an individual funnels the large number ofalternatives from the initial CS to select the final destination and whichprinciples are used in the process, at least from an empirical point ofview (Hong, Kim, Jang, & Lee, 2006, p. 750).Second, the few general CS models that have been proposed (e.g.,

    Brisoux & Laroche, 1980; Narayana & Markin, 1975; Turley & Leblanc,1995) are purely conceptual/theoretical and lack empirical roots.Moreover, those models rely on a hierarchical and sequential visionof DM that gradually reduces the number of alternatives, which is inline with the idea of a (bounded) rational decision maker (March &Simon, 1958). Most papers have viewed the formation of CSs as atwo-stage choice model wherein consumers first decide which brandsto consider and then, when a purchase situation arises, evaluate theremaining brands (Laroche et al., 2003). This vision makes little roomfor hedonistic, adaptive, and opportunistic perspectives that may berelevant in vacation DM (Decrop & Snelders, 2005). Finally, extant re-search fails to explore the dynamics of CSs both across and withinusage occasions. Nedungadi (1990) and Turley and Leblanc (1995)note that a static view dominates the CS literature. Shocker, Ben-Akiva,Boccara, and Nedungadi (1991, p. 186) regret such a limitation be-cause consideration sets are real, dynamic, changing with time andoccasion, and affected by consumer contexts and purposes.For these four reasons, reconsidering the formation of CSs is worth-

    while: Which types of CSs exist? How do they relate to each other? Howstable are they over time? How do they lead to choice? The presentstudy addresses such questions by investigating CSs in depth anddynamically, in the natural context of an actual vacation DM process.Before going into the empirical study, the next section presents extantliterature on CS in marketing and tourism research respectively.

    CHOICE SET MODELS IN MARKETING AND TOURISM RESEARCH

    A range of theories in economics and psychology and a substantialnumber of empirical studies support CSs existence although theythis paper as a generic term to refer to different types of sets containingproduct alternatives people have in mind when making a decision.This research reconsiders the emergence of tourism destination CSs

    in a naturalistic interpretive perspective. Four major arguments justify

  • A. Decrop / Annals of Tourism Research 37 (2010) 93115 95are not directly observable (Hauser & Wernerfelt, 1990; Shocker et al.,1991). Consumers tend to reduce the number of brands in consider-ation due to the limitation of cognitive capacity (Miller, 1956) and tothe want to save information (Stigler, 1961). Authors such as Hauserand Wernerfelt (1990) and Roberts and Lattin (1991) view CS forma-tion as a tradeoff between utility and cost. CS models also benefit fromconsiderable attention in the marketing literature (Brisoux & Laroche,1980; Hastak & Mitra, 1996; Howard, 1977; Howard & Sheth, 1969;Narayana & Markin, 1975; Nedungadi, 1990; Spiggle & Sewall, 1987;Turley & LeBlanc, 1995). Most authors suggest two typical stages. First,the brand-consideration stage involves retrieval (a function of bothmemory and brand accessibility) to form an evoked set (Belonax,1979; Howard, 1963, 1977; Howard & Sheth, 1969), a considerationset (Roberts, 1989; Roberts & Lattin, 1991; Wright & Barbour, 1977),or an action set (Spiggle & Sewall, 1987), most of the time throughthe use of a simple heuristic. Next, the brand-evaluation stage pertainsto the evaluation of brands to arrive at a final choice (i.e., selection of asingle brand) on the basis of more elaborated heuristics.A few authors (e.g., Brisoux & Laroche, 1980; Shocker et al., 1991;

    Um & Crompton, 1990) add a brand-awareness stage before the con-sideration stage, which represents a perceptual stage wherein the con-sumer includes all the destinations s/he knows to arrive to anawareness set. Such a conceptualization works in a hierarchical man-ner: It is now a truism of marketing that brand awareness is a neces-sary precondition for choice (Nedungadi, 1990, p. 264); theconsumer has to evaluate these brands before making a final choice.In addition to the concepts of awareness set and consideration/evokedset that were the first to appear in the literature, researchers intro-duced other types of sets which are defined in Table 1. Shockeret al. (1991) present a series of alternative models of CS formationand change. They show that CSs are real, dynamic, changing with timeand occasion, and affected by consumer contexts and purposes.Tourism research models of destination choice are either in the

    form of CSs in input-output approaches of DM (Crompton, 1977;Um & Crompton, 1990; Woodside & Lysonski, 1989) or in the formof cognitive processes (Mathieson & Wall, 1982; Moutinho, 1987; vanRaaij & Francken, 1984). CS models focus on the evolution of vacationdestinations and/or plans in a series of CSs. Those models are in linewith the two- or three-stage conceptualization presented above. Theconsideration set (evoked set) is part of the perceived opportunityset (awareness set) and comprises all the destinations the vacationeris contemplating for his/her current vacation. As the latter is not omni-scient, the awareness set is itself only a part of the total opportunity setwhich entails all possible destination alternatives (Goodall 1991; Wood-side & Sherrell 1977). Choice then consists in an evaluation and selec-tion process through which the vacationer compares the destinationsincluded in the consideration set on the basis of alternatives and/orattributes.Crompton and his colleagues (Ankomah, Crompton, & Baker, 1996;

    Crompton, 1992; Crompton & Ankomah, 1993; Um & Crompton,

  • Table 1. A Review of Choice Set Types in the Marketing and TourismLiterature

    Choice set type Definition Authors

    Universal set Totality of all alternatives that couldbe obtained or purchased by anyconsumer under any circumstance

    Shocker et al. (1991)

    Total opportunity set All possible brands Woodside and Sherrell(1977)

    Awareness set Brands the consumer is aware of andcan remember

    Howard and Sheth (1969)

    Brands about which consumers havesome awareness

    Brisoux and Laroche (1981)

    Evoked set Brands that the consumer wouldconsider and on which s/hegathers information

    Howard (1963), Howardand Sheth (1969)

    The subset of brands that aconsumer considers buying out ofthe set of brands that he or she isaware of in a given product class

    Howard (1977)

    Brands acceptable to the consumer Belonax (1979)Consideration set Brands that a consumer will consider Wright and Barbour (1977),

    Roberts (1989)Brands that a consumer considersbuying in the near future

    Roberts and Lattin (1991)

    Action set Alternatives toward which theconsumer takes some action (e.g.,looking for information)

    Spiggle and Sewall (1987)

    Inert set Brands that are acceptable but notreally needed (have insufficientutility to be evaluated)

    Narayana and Markin(1975)

    Foggy set Brands consumers are aware of butare not processing

    Brisoux and Laroche(1981), Church, Larocheand Rosenblatt (1985)

    Inaction set Alternatives toward which theconsumer takes no action (e.g.,looking for information)

    Spiggle and Sewall (1987)

    Hold set Brands that consumers process onlypartially (because they are notappropriate)

    Brisoux and Laroche (1981)

    Brands about which consumers lacka final opinion (cannot saywhether they would accept orreject)

    Church et al. (1985)

    Reject set Brands that consumers discard Brisoux and Laroche (1981)Inept set Brands that are ruled unacceptable

    for some reasonNarayana and Markin(1975)

    Unavailable aware set Brands the consumer is aware of butthat are temporarily unavailablefor some reason

    Woodside and Lysonski(1989)

    96 A. Decrop / Annals of Tourism Research 37 (2010) 93115

  • considering due to a poor previous experience or negativeinformation).

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 97In addition to the structure of CSs, a few authors consider their sizeswhile others study the variables impacting their formation process. Asto evoked sets sizes, Woodside and Sherrell (1977), Thompson andCooper (1979), Woodside and Lysonski (1989), and Perdue and Meng(2006) respectively report means of 3.4, 2.7, 4.2, and 2.6 destinations.Bronner and de Hoog (1985) support Woodside and Sherrells (1977)proposition that vacationers make the effort to evaluate only a few(four two) alternatives among a much larger set of available choices.This proposition is in line with the larger study by Hauser and Werner-felt (1990) which shows that most evoked sets are small, ranging fromthree to six depending on the product category. Woodside and Lyson-ski (1989, pp. 1213) further mention that the average size of therespondents consideration [i.e., evoked] sets was significantly greaterthan the average number of countries mentioned in the respondentsinert, unavailable, and inept sets. More broadly, authors such asCrompton and Ankomah (1993), Hong et al. (2006), and Bothaet al. (1999) suggest three types of criteria affecting the funneling pro-cess that reduces the number of alternatives: personal motivations orpush factors, destination attributes or pull factors, and situational vari-ables or constraints.

    Study methods

    This paper is part of a larger study on vacation DM which followedthe destination choice process of 25 Belgian decision making units(DMUs) for a whole year. The authors selected informants likely toyield rich and varied information in order to maximize theory develop-ment (Strauss & Corbin, 1990). DMUs consisted of four types: six sin-gles (representing tourists who decide on their own), six couples(married or non-married), ten families with children, and three groupsof friends. Acknowledging any intention to go on summer vacation wasthe only criterion to be eligible for the study. In addition to DMUtypes, the sample varied as to group size, age range, educational back-ground, occupation, and vacation involvement.1990) introduce a distinction between the initial consideration set(awareness set), consisting of a large number of alternatives, and thelate consideration set (evoked set), a reduced set from which consum-ers choose their final destination. Botha, Crompton, and Kim (1999)and Hong et al. (2006) apply the same distinction in order to investi-gate the roles of categorization, affective image, and constraints informing destination CSs. The latter authors show that the late consid-eration set determines sequentially the intention to visit a particulardestination. Woodside and Lysonski (1989) further make a distinctionbetween evoked set, inert set (brands for which the consumer has aneutral evaluation), unavailable aware set (the vacationer is aware ofthose destinations but s/he is not considering them actively becauseof particular constraints), and inept set (brands the consumer is not

  • 98 A. Decrop / Annals of Tourism Research 37 (2010) 93115We interviewed informants in depth three times before the summervacation (in February, April, and June). The use of such a longitudinaldesign is justified by a few arguments. First, as mentioned previously, amajor objective of this study is to follow the evolution of CSs over timein order to bypass the limitation of extant CS studies that have beenundertaken in static contexts although CSs are dynamic in essence (Lar-oche et al., 2003; Shocker et al., 1991). Longitudinal research allows suchan understanding of patterns of change and of causal relationships overtime (Menard, 1991). The longitudinal design is further justified by a fewempirical studies (Dellaert, Ettema, & Lindh, 1998; Francken, 1978;Moutinho, 1987; van Raaij & Francken, 1984), which have shown thatvacation decision making is a long process stretching over a few monthsbefore the actual trip. Finally, a longitudinal designgenerally helps to im-prove the trustworthiness of qualitative studies (Lincoln & Guba, 1985;Menard, 1991) because a relation of trust develops with informants, be-cause data can be triangulated at different points in time, etc.During each interview, most members of the DMU were present,

    including children for families. In each interview, we asked informantsto talk about the same three central themes: general vacation and tra-vel behavior, expectations and motives, and current vacation projectsand considered destinations for the summer. Most informants sponta-neously mentioned a number of destinations when telling about theirsummer travel plans. In addition, in each interview series over time, weasked them a few questions in order to keep the discussion focused onCSs: Did you consider particular travel destinations yet? Which destina-tion(s)? How did you come to consider that (those) destination(s)? Isthis (are these) destination(s) available to you? [If more destinations]Which destination(s) do you prefer or are you most likely to go to? Arethere some destination(s) where you will never go? However, followingthe principles of grounded theory (see below) and interpretive re-search, questions were open ended and not based on existing theo-ries/hypotheses in order not to bias emerging results. We tried toavoid forcing the data and to keep enough room for discoveryand for the emergence of the informants own views.The analysis and interpretation of the interview transcripts was based

    on the grounded theory approach, which is a qualitative researchmethod that uses a systematic set of procedures to develop an induc-tively derived grounded theory about a phenomenon (Strauss & Cor-bin, 1990, p. 24). In this study, categories, patterns, and propositionsemerged from the data coding process and were not derived from ex-tant literature. We strived toward a balance between theory blindness,with the risk of missing important theoretical phenomena emergingfrom the data, and an over-detailed examination of the literature,which bears the risk of forcing the data through preconceived ideasand hypotheses (Glaser and Strauss, 1967). Only the sensitizing con-cepts of the study have been subjected to the initial literature review.We went into more details during and after the data collection andinterpretation process to connect our emerging findings to extant lit-erature. Coding involved three levels (i.e., open, axial, and selective)from the most descriptive to the most interpretive, from the most con-

  • EM

    F thecon ofCS hathelp omthe

    Typ

    times, its an infatuation, its. . . No, in this case, one should againremember that I had health constraints, thats why I couldnt make

    chose something easy...

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 99In this opening quote, one may see that a series of destinations nat-urally emerge in Martines discourse. These may be categorized in dif-ferent types of CS. First, Denmark and the Faroe Islands appear to beup my mind. If I cannot make up my mind, I have to choose a countrywhere I can make a last-minute decision. This means that I make it bymyself and therefore. . . If I take the Eastern countries, because I reallycould have gone to the Eastern countries but traveling alone in East-ern countries becomes very difficult, then . . . And yes, secondly, thereis the problem of language, the problem is that you are never sure tobe able to book beforehand, you can hardly trust people etc. I did notsee myself backpacking alone in the Eastern countries, thats why IInterviewer: Does it mean that you choose your destinations depending on theclimate?Martine: Not really, it depends on anything that can show up. Some-ERGING FINDINGS

    our themes are discussed in this section, that is, types of CS,tinuity of evoked sets, sizes of evoked sets, and the dynamicsformation. Each theme is introduced or illustrated by quotes tbetter appraise how the findings progressively emerged fr

    data through our analytical process.

    es of Choice Sets

    Interviewer: What about your vacation plans for this year?Martine (single, teacher, 37): As I already told you, Im consideringgoing to Denmark and the Faroe Islands because I really loved visitingIsland last year and the year before, no sorry the year still before (thusthree years ago when I went to Norvegia) and thus. . . [. . .] Or alterna-tively, if it really appears that I cannot leave at all [because Martinesmother suffers from health problems], there is a chance that Ill go toParis for one week-end but it will not be for a vacation then. . . It willonly be for two or three days.Interviewer: So, it seems that you have an overall preference for Northerndestinations?Martine: In summer, yes. Because I dont like big heats. Now itsdone, I will have to travel again. Maybe next year, that could be Peruor Bolivia because I also have those ideas in mind since a long timeand because these are countries you can also visit in June, July,August. In contrast, I cannot do Asia at all: thats the monsoon time,so I cannot leave. So its true that Im always constrained to look atcountries where its possible, which means a part of South Americaand North America.crete events to the most abstract processes. We used memos and dia-grams as well, to assist in interpreting and establishing connectionsamong concepts.

  • the evoked destinations that are available this year although Paris ismentioned as well as a kind of spare or replacement destination inthe event Martines mother health would deteriorate. Moreover, Peruor Bolivia are already considered for next year as dreamed destinationsfor a long time. In contrast, Asian countries are discarded and fall intoa kind of exclusion set because of poor climate conditions (i.e., themonsoon in the summer when Martine is traveling). South Americaand North America further appear to be possible destinations she isaware of. Finally, Martine considers the Eastern countries as anotheroption but those destinations appear to be temporarily unavailable be-cause as a single she is afraid to travel alone in those countries. So,when analysing the quote above, no less than six different CSs appearto coexist in Martines mind, i.e, an available set, a surrogate set, adream set, an exclusion set, an awareness set, and an unavailable set.Table 2 further describes the different types of CSs that progressively

    emerged from data analysis. The last column of Table 2 parallels eachemerging type with similar types presented in the literature discussedpreviously. Each set contains none, one, or more destinations. When

    Table 2. Types of CSs in Vacationers DM Processes

    Type of CS Emerging definition Stage in the DMprocess

    Equivalent term inextant literature

    Awareness Set Destinations the vacationer knowsbut does not express any affectionwith. This set reveals knowledgebut not intention.

    Consideration Awareness Set

    Evoked Set Destinations consideredspontaneously by the vacationerfor future (but not especially thenext) summer vacation.

    Evaluation (+) Evoked set

    100 A. Decrop / Annals of Tourism Research 37 (2010) 93115Surrogate Set Destinations that are not prioritizedby the vacationer but are kept asspare alternatives (one neverknows...).

    Evaluation (+/-) Hold set

    Exclusion set Destinations definitely rejected bythe vacationer (I would neverstay there).

    Evaluation (-) Reject or inept set

    Dream Set Destinations that are considered asideal places for traveling orvacationing but are permanentlyunavailable because of enduringstructural inhibitors.

    Constraint(structural)

    /

    Unavailable Set Destinations considered by thevacationer that are temporarilyunavailable because of particularsituational constraints.

    Constraint(situational)

    Aware unavailableset

    Available set Evoked destinations that really arefeasible after considering thevacationers constraints

    Constraint/final choice

    /

  • A. Decrop / Annals of Tourism Research 37 (2010) 93115 101looking at the relative importance of each set in the interview data (interms of how often the informant mentioned destinations), the evokedset proves to be the most important. The exclusion set and the unavail-able set are also important, although to a lesser extent. In contrast,informants less often mentioned destinations in the surrogate setand the awareness set.In order to avoid lexical confusion, the difference between surrogate

    set and exclusion set needs further explanation. The difference be-tween surrogate set and exclusion set is a result of the evaluation ofdestination alternatives (see third column of Table 2). The surrogateset involves destinations that are positively evaluated but are not priori-tized by the vacationer. They are temporarily shelved as spare alternativesthat could be chosen in the last resort if a better solution is not found,as illustrated by Martines words in the opening quote: alternatively, ifit really appears that. . ., there is a chance that. . . In contrast, theexclusion set contains destinations permanently rejected by the vaca-tioner because they are negatively evaluated. Most of the time this rejec-tion is the result of a general lack of interest (New York does notinterest me at all) or of an excessive weakness of the destination onparticular attributes (such as climate for Asia in Martines case) thatinformants highly value. Motives and disliked activities may also leadto the exclusion of particular destinations (e.g., a few informants avoidany beach destination, like Spain). Variety seeking and emotional fac-tors also come into play. Finally, exclusion may be a function of thetype of trip which is considered. For example, an interviewed family ex-cludes destinations like Poland and Russia for vacationing (relaxing)but would not reject them for traveling (visiting).The distinction between unavailable set and dream set also requires

    further comments. This distinction is a reflection of the interventionof constraints in vacationers DM process (see third column of Ta-ble 2) and more precisely of the difference between situational andstructural constraints/inhibitors (Decrop, 1999). The unavailable setis composed of temporarily rejected destinations due to the interven-tion of situational constraints. These pertain to a particular decisionsituation and become more important when summertime isapproaching. For example, in the third interview of Jacqueline andRoger below, the couple no longer considers going to Bretagne orAuvergne because of their childrens poor school results. Time (re-lated to occupation), money, and accompaniment (e.g., being alonein Martines case) are the major momentary situational inhibitors. Ofcourse, this is not to say that those destinations will not re-enter theirevoked set another year. In contrast, the dream set includes destina-tions that are permanently out of reach because of enduring structuralconstraints. An alternative that falls into the dream set is not likely tore-enter the vacationers evoked set in the following years, at least aslong as the structural constraint is present. For example, an older sin-gle female informant would love to go to India but is not able toachieve that dream because of her poor health. Occupation, familysituation, and economic status are the major structural inhibitors inour data.

  • back to the surface again. Moreover, data show that the temporal ori-gin of many evoked projects and destinations lies in the last vacationexp theFaro

    DM

    102 A. Decrop / Annals of Tourism Research 37 (2010) 93115and I plan a bit beforehand but these are not precise projects. I dontlive in the hope of realizing. . .

    Size of Destination Evoked Sets

    This section focuses on evoked sets as it appears to be the mostimportant CS type in our data. Table 3 illustrates the evolution ofthe number of projects and spontaneously evoked destinations foreach DMU over the three series of interviews. The evoked set includesalternative destinations, in the sense of the one or the other, or a com-Us are considering only one vacation project at a time:

    Marie-France (couple, retired, 53): Personally I dont do very muchplanning. I have friends who plan a lot in advance: trips next yearand then in two years, well put a bit of money aside, so we can goto such a place. . . But I dont do that at all: the vacation comeshappen.

    Memory and learning (through experience) thus appear to contrib-ute to the continuity of evoked sets. In the same way, plans and desti-nations do not always pertain to the current year but also to the comingyears as illustrated by Martine who considers traveling to Peru or Boli-via next year (see opening quote). Furthermore, most informants donot only have vacation ideas and plans for the summer vacation butalso for other periods of the year. In short, more projects in differentstates of progress coexist in vacationers mind. Instead of speaking ofdifferent evoked sets, data suggest DMUs possess one ongoingmacro-evoked set. Indeed, the destinations included in one evokedset do not pertain to only one and the same consumption situation(or purchase decision) bounded in time and space. One should definethe macro-evoked set as a set of product alternatives which are all pos-sible in the near future. This finding is probably typical of vacation andtravel decisions which may involve planning and variety seeking overextended time periods. Macro-evoked sets allow consumers to solvetradeoffs more easily and not to give up desirable alternatives forever:if it is not this year, it will be for next year. In contrast, a minority oferience. For example, Martine first considered Denmark ande Islands while she was traveling in Iceland one year before:

    Martine (teacher, 37): Denmark, a rucksack, the Faroe Islandsthis iswhat Ive been thinking about ever since Ive been to Iceland. Once Isaw it, I said to myself the Faroe Islands must be a bit like that. NowIve been thinking about that for a whole year but maybe it will notOngoing macro-evoked sets

    The continuity of evoked sets is another major finding of this study.While being interviewed, informants often evoke vacation destinationsthat stem from earlier CSs. Previous years destinations or plans, whichthey have abandoned or postponed due to situational or structural con-straints, or to stronger preferences for other alternatives, are coming

  • Table 3. Case by Case Evolution of the Number of Vacation Projects andEvoked Destinations (One Line for Each Project)

    DMU id. t1 t2 t3 t41

    Projects Evokedset

    Projects Evokedset

    Projects Evokedset

    Projects Evokedset

    1 2 3 2 3 2 2 (-) NO -1 2 (+) 1 (-) YES YES

    2 2 2 1 YES NO (\)1 2 (+) 1 3 (+) NO -

    3 1 1 1 1 YES YES4 1 2 1 2 1 1 (-) YES NO (\)5 0 0 1 0 1 0 YES NO (\)6 0 0 1 1 1 2 (+) YES YES7 1 3 2 2 (-) 2 2 YES YES

    1 1 YES YES8 0 0 1 1 YES YES9 1 2 2 4 (+) 1 3 (2\) YES YES10 0 0 1 2 2 1 (-) YES YES

    1 YES YES11 1 4 1 4 (2\) 1 1 (-) YES YES12 1 2 1 1 (-) YES YES13 1 1 1 1 YES 2 (+)14 1 2 1 2 (-) 1 1 (-) YES YES15 1 2 1 1 (\) 1 1 (\) NO -16 1 2 1 3 (2\) 1 1 (-) YES NO17 1 4 1 2 (-) 2 1 (-) YES YES

    1 (\) YES YES18 1 1 1 1 1 1 YES 2 (+)19 2 1 2 1 2 1 NO -

    1 2 (+) 2 NO -20 1 2 1 1 (-) 1 1 YES YES21 1 3 (2 ) 1 1 (-) 1 2 YES 3 (+)22 2 1 2 2 (+) 0 - YES NO (\)

    2 2 (\)23 2 1 1

    2 1 (-) YES YES24 1 3 1 2 (1\) 1 1 (-) YES YES25 2 1 1 0 - NO -

    1 1 (\) NO -

    (+): Extension of the evoked set in comparison with the former interview.(-): Reduction of the evoked set in comparison with the former interview.(\): Modification of the evoked set in comparison with the former interview (number ofalternatives that have changed).: Combination of different alternatives in the same vacation plan.YES/NO: indicates whether or not an evoked vacation project/destination has actually beenachieved.1 Informants were contacted again after the summer vacation in order to check whether ornot they actually carried out the project and destination they evoked in the former interview.

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 103

  • and n)but Uand ofevok theinvo ci-sion restab ngfour 1.

    104 A. Decrop / Annals of Tourism Research 37 (2010) 93115The first group is made of DMUs with a large evoked set (three or fourdestinations), which decreases over time (sub-sample size: n = 6).These vacationers are highly involved, they have less personal con-straints (singles, couples, or older families) but make their decisionsvery late (less than one month before departure). The content of plansalso shows less stability: newer plans or destinations often appear overtime. The second group of DMUs have an initially small evoked set(one or two destinations), which decreases to zero or one destinationin t2 or t3 (n = 7). Early DM characterizes this group (i.e., they choose adestination once and for all or they make the decision not to go onvacation). These vacationers are older singles, highly involved in travel-ing. The third group of informants include DMUs that have an initiallysmall evoked set (t1) that expands (t2) and then decreases (t3) (n = 5).These informants show little stability regarding the content of theirvacation plans, most of the time because of children. The final decisionsecond series of interviews (mean t2 = 1.96/DMU or 1.58/plathey decrease slightly in the third series (mean t3 = 1.72/DM1.29/plan). Overall, data show that the number and evolutioned destinations is a function of involvement (the higherlvement, the larger and the less stable the evoked set), and detiming (the earlier the final decision, the smaller and the mole the evoked set). Further analyses lead to a distinction amogroups of informants with different patterns described in Figurebination of destinations, that is an inclusion of more destinations inthe same vacation plan. The number of evoked destinations rangesfrom zero to four in the data. Most of the time, informants evoke onlyone or two destinations and never spontaneously mention more thanfour alternatives while being interviewed. This finding is in line withprevious studies presented in section two.Two emerging factors explain the limited size of destination evoked

    sets. First, destination is not always the major vacation decision itemwhen compared with accommodation, period and transport decisions(Author, 1999); as a result, informants do not process that item inmuch depth. Second, destination proves to be a hyper-complex prod-uct. Belonax and Mittelstaedt (1978) show that having more choice cri-teria leads to larger evaluation costs and hence to smaller evoked sets.While being asked whether they had thought of destinations otherthan three evoked places (i.e., Turkey, Prague, and Saint Petersburg),the parents of a large family answered:

    Louis (teacher, 59): No, its already complicated enough to thinkabout three. I dont think we thought about another one.Jacqueline (teacher, 54): Its already complicated enough! And afterall, we dont have any other desires: the same ones always come back.Prague, we will go there too, but maybe one time at Easter, for a weekor. . . We realize that maybe there are less tourists if we go in the off-peak season, outside the main summer vacation time. Its easier, itscloser. No, we havent thought of other things.

    When looking at their evolution, evoked sets sizes are relatively sta-ble between the first (mean t1 = 1.93 destinations/DMU or 1.62/plan)

  • Figure 1. Evolution Patterns of Evoked Sets Sizes in Time

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 105is taken very late... or is never made. Finally, a fourth group of DMUs ischaracterized by a low (zero or one) and stable number of evoked des-tination alternatives (n = 7). This group is characterized by low vaca-tion involvement, a substantial number of constraints, early (or last4. Low involvement, lot of constraints, early or last-minutedecisions (loyalty)

    4. Low involvement, lot of constraints, early or last-minutedecisions (loyalty)

    4. Low involvement, lot of constraints, early or last-minutedecisions (loyalty)

    4. Low involvement, lot of constraints, early or last-minutedecisions (loyalty)

    4. Low involvement, lot of constraints, early or last-minutedecisions (loyalty)min

    The

    Infromlyset1 (February) t2 (April) t3 (June)mean=1.62 mean=1.58 mean=1.29

    1. Highly involved, less personalconstraints, latedecisions

    2. Highly involved, older and single vacationers, earlydecisions

    3. Families, lot ofconstraints(children), late (or no) decision

    t1 (February) t2 (April) t3 (June)mean=1.62 mean=1.58 mean=1.29

    1. Highly involved, less personalconstraints, latedecisions

    2. Highly involved, older and single vacationers, earlydecisions

    3. Families, lot ofconstraints(children), late (or no) decision

    t1 (February) t2 (April) t3 (June)mean=1.62 mean=1.58 mean=1.29

    1. Highly involved, less personalconstraints, latedecisions

    2. Highly involved, older and single vacationers, earlydecisions

    3. Families, lot ofconstraints(children), late (or no) decision

    t1 (February) t2 (April) t3 (June)mean=1.62 mean=1.58 mean=1.29

    1. Highly involved, less personalconstraints, latedecisions

    2. Highly involved, older and single vacationers, earlydecisions

    3. Families, lot ofconstraints(children), late (or no) decision

    t1 (February) t2 (April) t3 (June)mean=1.62 mean=1.58 mean=1.29

    1. Highly involved, less personalconstraints, latedecisions

    2. Highly involved, older and single vacationers, earlydecisions

    3. Families, lot ofconstraints(children), late (or no) decisionute) DM, and loyalty (people going to the same place).

    dynamics of CS formation

    order to give readers a better grasp on emerging findings, we startthe case of a married couple (Jacqueline and Roger) and we ana-how their CSs evolved over the three series of interviews.

    First interview (February 14):Interviewer: Do you already have some projects in mind for your next summervacation?Jacqueline (housewife, 58): Oh, we had an absurd idea but I thinkthat we are now back to reality. It was to go to Mexico or Brazil.Interviewer: What do you mean by absurd?Jacqueline: Because I think its too expensive.Roger (high magistrate, 63): Yes. Moreover, we have to go there inJuly or August.Jacqueline: For Brazil, temperature is OK.Roger: Yes but for Mexico at that time, the period doesnt fit.Jacqueline: Therefore, it was Brazil. But Brazil is too expensive. Thatswhy we come closer, we come closer.Roger: Turkey is the country we are now considering.Jacqueline: Ive been to get a brochure and we look at it a little bit.[. . .] Or alternatively, if my husband would like to take his car, we

  • F ue-line icoand re.

    106 A. Decrop / Annals of Tourism Research 37 (2010) 93115been considered.Jacqueline: And Ireland! [. . .] It means that if we were heading forcold, it was Island or Ireland. My husband would like to visit Nordiccountries but Im afraid people are so clean, too organized. . . I dontknow. . . I prefer more colorful and joyful people.Interviewer: Are there some factors that limit your destination choicescurrently?Jacqueline: Yes, as far Im concerned, climate for example for all ofSouth-eastern Asia. Its too hot, a humid heat, I dont like it. More-over, in Thailand, those stories about pedophiles, prostituted smallgirls. . . Id feel rather uncomfortable there.

    Second interview (April 9):

    Interviewer: could you please tell me about your vacation plans for thissummer?Jacqueline: [. . .] As far as destinations are concerned, wed like to goto Bretagne or to Auvergne. Because it seems that in Auvergne, thereis still more authenticity if we were to stay in a farm (but maybe thatsa false idea, I may be mistaken). But anyway, we always have goodmemories when we come back from Auvergne.Interviewer: In the first interview, you told me about Turkey and Italy; did yougive up those two destinations?Jacqueline: Actually for Turkey, we changed our mind (wrongly orrightly) because if we dont rent a car in Istambul, we have to catchanother flight to go to the Mediterranean coastline and if we take apackage tour as offered in calatogs, we will end up in a tourist para-dize. [. . .] Moreover, I do worry about my husband who is sufferingfrom back ache; Id feel guilty would something happen to him. [. . .]In contrast, if we go to civilized countries, should it be any problem,there is always an hospital somewhere and the possibility to be under-stood. [. . .] Italy is not definitively excluded but as we already go toRome now. . . I mean. . . my husband loves the sea but the Mediterra-nean sea is not really the sea for us. Its not wild enough, its too flat,too blue. I prefer a sea that moves, that splashes; as far Im concerned,I like the Atlantic ocean, and my husband too. Thats why we have togo to the South-east [of France], to Bretagne or to the North sea.Interviewer: So, Turkey and Italy are ruled out for this year?Jacqueline: For this year, yes. Especially, since weve got the taxes topay. . .

    Third interview (June 22):

    Interviewer: Could you please tell me about your vacation plans for thissummer?Jacqueline: Actually, they are avorted for the good reason that my hus-band has decided to retire, on the one hand, and because of the poorschool results of our children on the other hand. [. . .] Thats why weopted for theNorth sea.Wehave rent an apartment fromone to 15 July.

    rom this opening quote, we see that in the first interview, Jacqand Roger mention dreamed or absurd destinations (MexBrazil) that seem out of reach because of price and temperatuthought about going to Italy, in the Marches region. Because in Italy,we never get bored, there is always something to see and the food isnice. [. . .] So as you can see, we did consider many things!Roger: Oh yes, many things have been considered. Island has also

  • They evoke more reasonable destinations as well (Turkey, Italy, Ice-land and Ireland). Alternatives are also already excluded at that stage(South-eastern Asia because of climate and ethical concerns). In thesecond interview, plans have changed: Bretagne and Auvergne areevoked and further appear to be the two available alternatives left. Tur-key is now part of their unavailable set because of Rogers health prob-lems and budget constraints; Italy is shelved because the couple isalready going on a city-trip to Rome. During the third interview, it ap-pears that plans still have changed: a surrogate destination not men-tioned in the former interviews (i.e., the Belgian North Sea) emergesas final choice. In conclusion, for Jacqueline and Roger, as it is alsothe case for more DMUs in our sample, the final chosen destinationhas little to do with the destinations mentioned during the first (andsometimes even the second) interview (see Figure 2).Two major explanations of such turnarounds emerge from data anal-

    ysis. On the one hand, plans show an increasing level of realism overthe three series of interviews. In t1, informants are inclined to dreamaloud when mentioning possible summer vacation destinations. Whileaware of actual and potential constraints, they try to convince them-selves that the project is still feasible. That dream dimension is far

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 107less present in the next two series of interviews. Considered destina-tions become more realistic as consumers take situational inhibitorsinto account to a larger extent. As a consequence, the unavailableset grows during the DM process whereas the size of the available setdecreases. Moreover, the unavailable set appears to be less stable overtime than the exclusion set and than the dream set. Sometimes, situa-tional inhibitors even lead vacationers to choose a surrogate destina-tion that was not part of evoked sets.On the other hand, opportunities may arise such as an invitation

    from relatives or an advertized special offer. Such opportunities result

    Awareness set

    Surrogate set : t3 Belgian North sea

    Evoked set : t1 Mexico, Brazil, Turkey, Italy, Iceland, Ireland, (t2 Bretagne, Auvergne)

    Dream set : t1 Mexico, Brazil

    Unavailable set : t2 Turkey, Italy

    Available set: t2 Bretagne, Auvergne

    Exclusion set : t1 South-eastern Asia (e.g., Thailand)

    Final Choice: t3 North sea

    Figure 2. The Formation of Destination CSs (The Case of Jacqueline andRoger)

  • in the sudden awareness of previously unknown destinations. The vaca-tioner may finally prefer these new alternatives and choose them overother alternatives, as illustrated by the following quote:

    Thierry (couple, medical rep, 28): But if an opportunity comes outfrom now to the end of the year (from now to vacation time, sorry),it is very possible... that we will take this opportunity. I dont know,if we have an opportunity to go... to Italy because for one reason oranother, we have a house, or rather there is a friend who rents onehouse and who proposes us to go with him, well we will go to Italy.

    Figure 3 summarizes the studys findings. The formation of CSs mayinclude four dimensions (we prefer this term to stages as the processis not always sequential). Awareness is the first dimension: vacationersare either aware of existing destinations or not aware. Awareness mostof the time results from ones own experience or from informationgathered opportunistically from external sources. Next, consumersevaluate the destinations they are aware of. Destinations evaluated pos-itively will fall into either the evoked set (preference or expectation le-vel) or the surrogate set (tolerance level). Destinations evaluatednegatively become the exclusion set. The third dimension involves tak-ing constraints into account. Evoked destinations end up into the

    108 A. Decrop / Annals of Tourism Research 37 (2010) 93115 MARKET Consideration Evaluation Constraints Choice

    Figure 3. The Formation of Destination CSsExclusion set - dream set when vacationers face one or more structural constraint(s)or into the unavailable set if they are confronted in one or more situ-ational inhibitor(s). In contrast, destinations are included in the avail-able set if no constraint is experienced. The final choice is made eitherfrom the available set, the surrogate set, or straight from the awarenessset. Vacationers may choose a spare or surrogate destination when theavailable set decreases to no alternative at all (due to the interventionof situational inhibitors). Sometimes DMUs become aware of new des-tinations not considered at first because of propositions from their

    All existing destinations

    Awareness set

    Unawareness set

    Surrogate set

    Evoked set

    Dream set

    Unavailable set

    Available set Final Choice

    +

    +/-

  • tions have been considered and evaluated, structural constraints and

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 109situational factors are taken into account; this reduces the number ofevoked destinations to an available set which includes only the alter-natives that are feasible.Moreover, our emerging model suggests that the choice process may

    be shortened through direct connections between awareness set andmarketing and/or social environment (see the dotted arrow from theunawareness set to the awareness set in Figure 3). Finally, vacationersmay choose an opportunistic destination rather than an alternativefrom the available set or surrogate set (see the dotted arrow fromthe awareness set to choice).

    DISCUSSION AND CONCLUSION

    This paper focuses on the formation and evolution of CSs. Theemerging typology of Table 2 partly confirms and extends the tradi-tional conceptualizations of Narayana and Markin (1975), and Brisouxand Laroche (1980) adapted in tourism by Um and Cromptons(1990), and Woodside and Lysonskis (1989). One may parallel thepresentation of an exclusion set, respectively with Brisoux and Lar-oches reject set and Narayana and Markins inept set. In the sameway, the concepts of a surrogate set and of an unavailable set are notfar from Brisoux and Laroches hold set, and Woodside and Lysonskisunavailable aware set respectively. However, the surrogate set involvesmore processing and the formation of positive attitudes towards prod-uct alternatives, which the hold set does not. In contrast, extant litera-ture has never considered ideas of an available set and a dream setbefore, although these two types of sets may play major roles in desti-nation choice such as described earlier.Moreover, our study shows that the number of destinations is quite

    stable over time and ranges from zero to four with means close to twodestinations/DMU (1.5/plan). Such a finding is in line with the generalidea that most evoked sets are small, ranging from three to six depend-ing on the product category (Hauser & Wernerfelt, 1990) and with thespecific contention that vacationers evaluate only a few destinations(i.e., four two) among a much larger set of available alternatives(Bronner & de Hoog, 1985; Perdue & Meng, 2006; Woodside & Sher-rell, 1977). Evoked sets sizes may be even smaller in this study due tothe naturalistic method employed, as informants were to evoke alterna-tives for an actual rather than hypothesized DM situation.Traditional conceptualizations present the formation of CSs along

    two typical sequential stages: the brand-consideration stage and thebrand-evaluation stage (Brisoux & Laroche, 1980; Hastak & Mitra,1996; Howard, 1977; Howard & Sheth, 1969; Narayana & Markin,1975; Nedungadi, 1990; Spiggle & Sewall, 1987; Um & Crompton,1990). However, the emerging conceptualization of Figure 3 is a littlemore subtle since it suggests that three stages or dimensions ratherthan two may lead consumers to their final choice. More specifically,our model adds a constraint dimension as third stage: after destina-

  • decision making (Payne, Bettman, & Johnson, 1993) which has been

    110 A. Decrop / Annals of Tourism Research 37 (2010) 93115underused in investigating vacation DM.Third, this paper helps better understand the dynamics of CS forma-

    tion. As far as we know, this research is the first to examine the forma-tion of destination CSs longitudinally as the same informants havebeen interviewed up to three times. Such a longitudinal approachhighlights the growing level of realism in vacation plans and destina-tions over time. A shift from dream to reality is at play in the formationof CSs, which is comparable to Mansfelds (1994) adaptation of thevalue stretch concept (Rodman 1963). Vacationers may move froma preference/ideal value level (dreamed but not necessarily availablefinal choice or between surrogate set and choice. This straight recourseto the awareness set highlights that evaluation is not always necessaryfor choice. Focusing on the stage of CS formation prior to evaluationand choice, Nedungadi (1990) provides empirical evidence that brandchoice is significantly altered outside the traditional evaluation-basedroute, through variation in the retrieval and consideration of brands(p. 273). He suggests a memory-based choice dependent on brandaccessibility. For a brand to be included in the CS, the consumer mustrecall that brand and fail to recall other brands that s/he might other-wise prefer.

    Theoretical implications

    The papers findings entail four major contributions to extant liter-ature. The first contribution is to generate an empirical model of CSformation that refines traditional conceptualizations by Narayanaand Markin (1975), Brisoux and Laroche (1980), and Turley and Le-blanc (1995). In contrast with those purely conceptual hypothetico-deductive models, the papers Figure 3 has been generated inductivelyand analytically from empirical data related to actualas opposed tohypotheticalDM processes. Moreover, our emerging model is spe-cific to tourism; it involves local and substantive theory (Strauss & Cor-bin, 1990) empirically rooted in a destination choice setting. Incontrast, authors such as Um and Crompton (1990) or Woodsideand Lysonski (1989) have borrowed and adapted the general/formaltheory of traditional CS conceptualizations to a tourism context.Second, our findings indicate that to a large extent, the formation of

    CS is a constraint- and opportunity-driven process, which has been ne-glected by the CS literature so far. Constraints consideration leadsconsumers to categorize evoked brands as part of either the available,unavailable, or dream set. Such a distinction has not been introducedby former models. A few authors already investigated the role of con-straints in vacation DM and destination choice (e.g., Gilbert & Hudson,2000; Hong et al., 2006; Perdue & Meng, 2006). However, they did notintroduce constraints consideration as a stage in itself in the formationprocess of CS. Moreover, our data suggest that opportunities areresponsible for the appearance of new alternatives that consumersmay finally choose. Such results are in line with the idea of adaptive

  • the likelihood that products receive consideration. The cueing of spe-

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 111cific product alternatives by contacts with friends and acquaintances orwith promotional and other marketing activity (e.g., sales personnel)may also affect retrieval from memory and thus the formation of choicesets (p. 190). Managers should enhance their brands accessibility(ease of retrieval) and desirability at the same time.Moreover, practitioners should try to develop effective plans and ac-

    tions in order to facilitate the transfer of their brands from the dreamor unavailable set into the available set of their target customers. Forexample, the cruise industry has made considerable efforts these lastyears to change the perception that a cruise is an expensive elite productthat only richpeople can afford; operators have loweredprices, theyhaveoffered families children-friendly packages, and they have extended dis-tribution channels. Furthermore, destination managers should makesure that their country, region, or city is part of vacationers surrogateset of destinations. For example, one could use a slogan such as Whynot the Belgian coast? in advertisements targeting Belgian vacationerswho may choose that nearby popular spot as a surrogate destination ifthey realize that they lack the time or the budget to go abroad.Tour operators, travel agents and destination managers should also

    consider the opportunistic and constraint-driven way in which potentialdestinations) to an expectation level (realistically available destina-tions) and finally to a tolerance level (surrogate destinations that rep-resent an acceptable minimum) as far as plans evolve and summertimeapproaches. The longitudinal data also suggest many turnarounds invacation plans and destination CSs. Evoked sets composition and sizefluctuates in time (see the four patterns of Figure 1) and cannot belimited to a linear reduction such as often presented in extantliterature.Finally, the CS perspective used in this paper highlights the continu-

    ity of vacation DM. More projects are considered simultaneously withdifferent time horizons, multiple DMUs, and so forth. The coexistenceof those projects in the vacationers mind leads to the formation ofmacro-evoked sets of destinations. So destination choice proves to bea continuous process where thinking, dreaming, talking about vaca-tions, and gathering information is ongoing. Such a finding enhancesthe hedonic and experiential dimensions of the DM and CS formationprocesses, which is far from the vision of a (bounded) rational decisionmaker depicted in most CS models (Brisoux & Laroche, 1980; Hauser& Wernerfelt, 1990; Howard & Sheth, 1969; Narayana & Markin, 1975;Spiggle & Sewall, 1987).

    Managerial Implications

    Travel professionals are likely to make better predictions fromchoice models that recognize different types of sets rather than focus-ing on the consideration (evoked) set only. The predictive validity ofevoked sets considered alone proves to be very limited indeed. Shockeret al. (1991) stress that research on the formation of CSs helps improve

  • Limitations and Suggestions for Future Research

    112 A. Decrop / Annals of Tourism Research 37 (2010) 93115Of course, this study entails a few limitations which open avenues forfuture research. First, set models may oversimplify the reality of choices.Set models imply a binary logic in which a (destination) choice alterna-tive is part of a set or not.While good conceptual andoperational reasonslead to using such models, fuzzy classifications could be more realisticand outperform set models from a methodological perspective. Ofcourse, a mathematical modeling approach would then be needed.For example, aprobabilisticmodel couldbedevelopedwherein the alter-natives likelihood of being chosen would depend on explanatory vari-ables such as awareness, availability, and/or congruence with criteriaof a dream destination. Moreover, as in other conceptualization efforts,the number of and the names given to the different CSs may seem some-what arbitrary. In this paper, we tried to limit such subjectivity by gener-ating the various CSs systematically and analytically from the empiricaldata and by giving readers the opportunity to validate the typology bythemselves through the interview quotes. Moreover, the particular num-bering andnamingofCSs are of lesser importance than theway theyhelpto better understand the processes through which consumers come toselect or reject choice alternatives.Next, a series of factors influencing the formation of CSs, such as the

    contextual variables listed in Figure 3, should be investigated in moredepth. For example, the length and type of trip may affect the level ofplanning and hence the composition of the CS: a long-distance triptourists form their CSs. On the one hand, operators and agents couldbenefit from the opportunistic nature of many vacationers by keepingin touch with them, and making them offers such as early booking orlast-minute discounts. Active steps are particularly beneficial becausevacationers are waiting for and not purposefully searching for informa-tion. On the other hand, managers should also help remove particularvacationers constraints. They could understand how those constraints,imposed by individual goals and other personal circumstances, interactwith available alternatives and other environmental factors. Moreover,travel agents should not overwhelm vacationers with propositions andalternatives as destination evoked sets are of limited sizes.Finally, industry people should incorporate the dynamism and conti-

    nuity of many CSs in their marketing strategies and decisions. Most vaca-tioners are involved in more than one vacation project at a time withdifferent needs, desires, and expectations, which leads to the formationof a macro-evoked set entailing both available destinations for the cur-rentproject andunavailableordreamdestinations that they could recon-sider in future plans. Consequently, vacationers should not be put inexclusive segments: an alternative that has once been rejectedmay re-en-ter the evoked set at a later time or in another choice situation. For exam-ple, parents with young children could have shelved Egypt at one timebut might evoke that country again as a possible destination a few yearslater when children are older or have left the family nest.

  • could increase the research requirements and thus lower the number

    price. In the same way, Perdue and Meng (2006) show that the reasonsmentioned by respondents for destination selection differ from the

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 113reasons cited for rejection, and contend that a non-compensatory heu-ristic is used in ski destination choice. They close their paper by sug-gesting a duality in the formation of CS: the observed reasons forselection may be necessary conditions to get into consideration sets,while the observed reasons for rejection reflect the actual choice(Perdue & Meng, 2006, p. 347).As a final limitation, this studys emerging propositions are qualita-

    tive, interpretative and context-bound. The papers findings, tablesand figures have been built inductively from the study of a phenome-non situated in a particular context. Refering to Strauss and Corbin(1990), this is a local and substantive theory, in contrast with a generaland formal theory. A substantive theory emerges from the study of aphenomenon situated in a particular situational context, whereas a for-mal theory develops when the phenomenon is being examined undermany types of situations. If the analytical generalization of emergingfindings is possible to some extant, statistical generalization is not pos-sible because of the theoretical (non-random) sampling procedureand the small sample size. The papers emerging propositions couldbe tested on larger representative samples if statistical generalizationwere desired.

    REFERENCES

    Ankomah, P., Crompton, J., & Baker, D. (1996). Influence of cognitive distance invacation choice. Annals of Tourism Research, 23, 138150.

    Belonax, J. (1979). Decision rule uncertainty, evoked set size, and informationvariability. Advances in Consumer Research, 6, 232235.

    Belonax, J., & Mittelstaedt, R. (1978). Evoked set size as a function of number ofchoice criterion and information variability. Advances in Consumer Research, 5,4851.

    Botha, C., Crompton, J., & Kim, S. (1999). Developing a revised competitiveposition for Sun/Lost City, South Africa. Journal of Travel Research, 37, 341352.

    Brisoux, J., & Laroche, M. (1980). Proposed consumer strategy of simplification forcategorising brands. J. Summey & R. Taylor, eds., Evolving Marketing Thoughtfor 1980 (pp. 112114). Carbondale, IL: Southern Marketing Association.of alternatives in the CS when compared with a short-distance trip.Focusing on choice heuristics or strategies underlying the formationof CSs at different stages of Figure 3 is another possible extension ofthis research. Of course, such a focus on decision heuristics would re-quire a merging of the product alternatives approach to DM chosen inthis paper with an (multi-)attributes perspective.In tourism, a few authors (Perdue & Meng, 2006; Turley & Leblanc,

    1995) suggest that the attributes affecting the selection of destinationalternatives may be different from those affecting rejection. Turley andLeblanc (1995) assume that inclusion in the CS is based on a satisficingrule (i.e., meeting a minimum level on a few primary attributes) whilerejection is then based on facilitating attributes such as availability and

  • Brisoux, J., & Laroche, M. (1981). Evoked set formation and composition: An

    114 A. Decrop / Annals of Tourism Research 37 (2010) 93115empirical investigation under a routinized response behavior situation.Advances in Consumer Research, 8, 357361.

    Bronner, F., & de Hoog, R. (1985). A recipe for mixing decision ingredients.European Research, 13, 109115.

    Church, N. J., Laroche, M., & Rosenblatt, J. A. (1985). Consumer brand catego-rization for durables with limited problem solving: An empirical test andproposed extension of the Brisoux-Laroche model. Journal of EconomicPsychology, 6, 231253.

    Crompton, J. L. (1977). A systems model of the tourists destination selection decisionprocess with particular reference to the role of image and perceived constraints.Unpublished Doctoral Dissertation, Texas A&M University, USA.

    Crompton, J. (1992). Structure of vacation destination choice sets. Annals ofTourism Research, 19, 420434.

    Crompton, J., & Ankomah, P. (1993). Choice set propositions in destinationdecisions. Annals of Tourism Research, 20, 461476.

    Decrop, A. (1999). Commitments and opportunities: Judgment and decision making byvacationers. Namur: Presses Universitaires de Namur.

    Decrop, A., & Snelders, D. (2005). A grounded typology of vacation decisionmaking. Tourism Management, 26, 121132.

    Dellaert, B., Ettema, D., & Lindh, C. (1998). Multi-Faceted tourist travel decisions:A constraint-based conceptual framework to describe tourists sequentialchoices of travel components. Tourism Management, 19, 313320.

    Engel, J., Kollat, D., & Blackwell, R. (1973). Consumer behavior. New York: Holt,Rinehart and Winston.

    Francken, D. (1978). Het vakantie-besluitvormingsproces. Internal report. Breda:Netherlands Research Institute for Tourism and Recreation.

    Gilbert, D., & Hudson, S. (2000). Tourism demand constraints: A skiingparticipation. Annals of Tourism Research, 27, 906925.

    Glaser, B., & Strauss, A. (1967). The discovery of grounded theory. Chicago: Aldine.Goodall, B. (1991). Understanding holiday choice. In C. Cooper ed. Progress in

    tourism, recreation and hospitality management (Vol. 3, pp. 5977). London:Belhaven Press.

    Gruca, T. (1989). Determinants of choice set size: An alternative method ofmeasuring evoked sets. Advances in Consumer Research, 16, 515521.

    Hastak, M., & Mitra, A. (1996). Facilitating and inhibiting effects of brand cues onrecall, consideration set, and choice. Journal of Business Research, 37, 121126.

    Hauser, J., & Wernerfelt, B. (1990). An evaluation cost model of consideration sets.Journal of Consumer Research, 16, 393408.

    Hong, S.-K., Kim, J.-H., Jang, H., & Lee, S. (2006). The roles of categorization,affective image and constraints on destination choice: An application of theNMNL model. Tourism Management, 27, 750761.

    Howard, J. (1963). Marketing management, analysis and planning. Homewood, IL:Richard D. Irwin.

    Howard, J. (1977). Consumer behavior: An application of theory. New York: McGraw-Hill.

    Howard, J., & Sheth, J. (1969). The theory of buyer behavior. New York: John Wiley andSons.

    Laroche, M., Kim, C., & Matsui, T. (2003). Which decision heuristics are used inconsideration set formation?. Journal of Consumer Marketing, 20(3), 192209.

    Laroche, M., & Parsa, H. (2000). Brand management in hospitality: An empiricaltest of the Brisoux-Laroche model. Journal of Hospitality and Tourism Research,24, 199222.

    Lavidge, R., & Steiner, G. (1961). A model for predictive measurements ofadvertising effectiveness. Journal of Marketing, 25(October), 5962.

    Lincoln, Y., & Guba, E. (1985). Naturalistic inquiry. Beverly Hills, CA: SagePublication.

    Mansfeld, Y. (1994). The Value stretch model and its implementation indetecting tourists class-differentiated destination choice. In R. V. Gasser & K.Weiermair (Eds.), Spoilt for choice. Decision making processes and preference change

  • of tourists: intertemporal and intercountry perspectives (pp. 6079). Thaur:Kulturverlag.

    Miller, G. (1956). The magical number seven, plus or minus two: Some limits onour capacity for processing information. The Psychological Review, 63(March),

    A. Decrop / Annals of Tourism Research 37 (2010) 93115 1158197.Moutinho, L. (1987). Consumer behaviour in tourism. European Journal of

    Marketing, 21(10), 244.Narayana, C., & Markin, R. (1975). Consumer behavior and product performance:

    An alternative conceptualization. Journal of Marketing, 39(Fall), 16.Nedungadi, P. (1990). Recall and consumer consideration sets: Influencing choice

    without altering brand evaluations. Journal of Consumer Research, 17, 263276.Payne, J., Bettman, J., & Johnson, E. (1993). The adaptive decision-maker. Cambridge:

    Cambridge University Press.Perdue, R., & Meng, F. (2006). Understanding choice and rejection in destination

    consideration sets. Tourism Analysis, 11(6), 337348.Roberts, J. (1989). A grounded model of consideration set size and composition.

    Advances in Consumer Research, 16, 749757.Roberts, J., & Lattin, J. (1991). Development and testing of a model of

    consideration set composition. Journal of Marketing Research, 28(November),429440.

    Shocker, A., Ben-Akiva, M., Boccara, B., & Nedungadi, P. (1991). Consideration setinfluences on consumer decision-making and choice: Issues, models, andsuggestions. Marketing Letters, 2(3), 181197.

    Spiggle, S., & Sewall, M. (1987). A choice sets model of retail selection. Journal ofMarketing, 51, 97111.

    Stigler, G. (1961). The economics of information. Journal of Political Economy,69(June), 213225.

    Strauss, A., & Corbin, J. (1990). Basics of qualitative research: Grounded theoryprocedures and techniques. Newbury Park: Sage.

    Thompson, J., & Cooper, R. (1979). Additional evidence on the limited size ofevoked and inept sets of travel destinations. Journal of Travel Research,18(Winter), 2325.

    Turley, L., & LeBlanc, R. (1995). Evoked sets: A dynamic process model. Journal ofMarketing Theory and Practice, 3(2), 2836.

    Um, S., & Crompton, J. (1990). Attitude determinants in tourism destinationchoice. Annals of Tourism Research, 17, 432448.

    van Raaij, W., & Francken, D. (1984). Vacation decisions, activities and satisfac-tions. Annals of Tourism Research, 11, 101112.

    Woodside, A., & Lysonski, S. (1989). A general model of traveler destinationchoice. Journal of Travel Research, 27(Spring), 814.

    Woodside, A., & Sherrell, D. (1977). Traveler evoked, inept, and inert sets ofvacation destinations. Journal of Travel Research, 16(Winter), 1418.

    Wright, P., & Barbour, F. (1977). Phased decision strategies: Sequels to initialscreening. M. Starr & M. Zeleny, eds., Multiple criteria decision making(pp. 91109). Amsterdam: North Holland.

    Submitted 10 December 2008. Final Version 18 June 2009. Accepted 13 August 2009.Refereed anonymously. Coordinating Editor: Muzzo Uysal

    Available online at www.sciencedirect.comMarch, J., & Simon, H. (1958). Organizations. New York: Wiley.Mathieson, A., & Wall, G. (1982). Tourism: Economic, physical and social impacts.

    Harlow: Longman.Menard, S. (1991). Longitudinal research. London: Sage.

    DESTINATION CHOICE SETSINTRODUCTIONCHOICE SET MODELS IN MARKETING AND TOURISM RESEARCHStudy methods

    EMERGING FINDINGSTypes of Choice SetsOngoing macro-evoked setsSize of Destination Evoked SetsThe dynamics of CS formation

    DISCUSSION AND CONCLUSIONTheoretical implicationsManagerial ImplicationsLimitations and Suggestions for Future Research

    References