11
The multiple roles of fit between brand alliance partners in alliance attitude formation Bendik Meling Samuelsen & Lars Erling Olsen & Kevin Lane Keller # Springer Science+Business Media New York 2014 Abstract This paper tests the assumption that consumersperceptions of fit between brand alliance partners can serve different roles in attitude formation depending on the level of elaboration given to a persuasive message about a brand alliance. We exper- imentally manipulate fit between brand concepts of real brands, situational involve- ment, and argument quality to test this assumption. A three-way interaction showed a positive main effect of fit on alliance attitude under low involvement and an interaction between fit and argument quality on alliance attitude under high involvement. Consequently, brand owners could expect more beneficial behavioral consequences of alliance attitudes if they are (a) based on alliances between brands with similar brand concepts, (b) backed with strong arguments, and (c) perceived as personally relevant by target consumers. The paper adds insight into how fit between the alliance partners can assume different roles as persuasion variables, thereby extending our understanding of theoretical mechanisms explaining when and why fit is important between brand alliance partners. Keywords Brand alliances . Perceived fit . Attitudes . Elaboration likelihood model 1 Introduction At first look, some brand alliances are puzzling, some are not. With sufficient funds, one can buy a Bentley car with Breitling instruments, or for less money, one can buy a Mark Lett DOI 10.1007/s11002-014-9297-y B. M. Samuelsen (*) Department of Marketing, BI Norwegian Business School, 0442 Oslo, Norway e-mail: [email protected] L. E. Olsen Department of Marketing, Oslo School of Management, P.O. Box 1195, Sentrum, 0107 Oslo, Norway e-mail: [email protected] K. L. Keller Tuck School of Business, Dartmouth College, 100 Tuck Hall, Hanover, NH 03755-9011, USA e-mail: [email protected]

The multiple roles of fit between brand alliance partners in alliance attitude formation

Embed Size (px)

Citation preview

Page 1: The multiple roles of fit between brand alliance partners in alliance attitude formation

The multiple roles of fit between brand alliancepartners in alliance attitude formation

Bendik Meling Samuelsen & Lars Erling Olsen &

Kevin Lane Keller

# Springer Science+Business Media New York 2014

Abstract This paper tests the assumption that consumers’ perceptions of fit betweenbrand alliance partners can serve different roles in attitude formation depending on thelevel of elaboration given to a persuasive message about a brand alliance. We exper-imentally manipulate fit between brand concepts of real brands, situational involve-ment, and argument quality to test this assumption. A three-way interaction showed apositive main effect of fit on alliance attitude under low involvement and an interactionbetween fit and argument quality on alliance attitude under high involvement.Consequently, brand owners could expect more beneficial behavioral consequencesof alliance attitudes if they are (a) based on alliances between brands with similar brandconcepts, (b) backed with strong arguments, and (c) perceived as personally relevant bytarget consumers. The paper adds insight into how fit between the alliance partners canassume different roles as persuasion variables, thereby extending our understanding oftheoretical mechanisms explaining when and why fit is important between brandalliance partners.

Keywords Brand alliances . Perceived fit . Attitudes . Elaboration likelihoodmodel

1 Introduction

At first look, some brand alliances are puzzling, some are not. With sufficient funds,one can buy a Bentley car with Breitling instruments, or for less money, one can buy a

Mark LettDOI 10.1007/s11002-014-9297-y

B. M. Samuelsen (*)Department of Marketing, BI Norwegian Business School, 0442 Oslo, Norwaye-mail: [email protected]

L. E. OlsenDepartment of Marketing, Oslo School of Management, P.O. Box 1195, Sentrum, 0107 Oslo, Norwaye-mail: [email protected]

K. L. KellerTuck School of Business, Dartmouth College, 100 Tuck Hall, Hanover, NH 03755-9011, USAe-mail: [email protected]

Page 2: The multiple roles of fit between brand alliance partners in alliance attitude formation

Breitling timepiece with a Bentley logo on the dial. However, would not an alliancebetween Hyundai and Breitling be more puzzling? Perhaps one might see Timex andHyundai as a better match?

Consumers are constantly exposed to brand alliances or co-brand initiatives like theseexamples. Previous research has shown that perceptions of fit between brand alliancepartners are important for consumers when forming alliance attitudes (e.g., Arnett et al.2010; James 2006; Lanseng and Olsen 2012; Park et al. 1996; Simonin and Ruth 1998;Swaminathan et al. 2012). The straightforward implication could be that both Breitling +Bentley and Hyundai + Timex can be seen as alliances with relatively high fit, as theformer combines symbolic brand concepts, while the latter combines functional brandconcepts (Park et al. 1986). Being a combination of a functional and a symbolic brandconcept, a Hyundai + Breitling alliance would be perceived as having lower fit.

However, the car + watch alliance examples raise some important questions still notaddressed in the brand alliance literature. Can consumers utilize perceptions of fitdifferently in the process of brand alliance attitude formation? Do consumers simplyaccept or reject alliances using fit as a cue, or do consumers use fit as a moremeaningful piece of information, i.e., as an argument to be evaluated? And finally,does fit increase or decrease processing of alliance information?

An approach to these questions can be found in dual process persuasion theories likethe Elaboration Likelihood Model (ELM; Petty and Cacioppo 1986) which posit thatsimilar attitudes can be developed by different processes. More specifically, any pieceof information may serve multiple roles depending on the recipient’s level of elabora-tion, with more elaboration given to issues perceived to have higher personal relevance(Petty and Cacioppo 1986).

The current study investigates the multiple roles of fit between brand alliancepartners. We report the results from an experiment with real brands representingvarying degrees of fit between their respective brand images. Personal relevance wasvaried through a situational involvement manipulation, and we varied argument qualityas a means to detect differences in elaboration (Petty and Cacioppo 1986) acrossdifferent levels of brand fit.

2 Theoretical background

2.1 Fit in brand alliances

Conceptually, fit is an instance where two objects share some commonalities (Aaker andKeller 1990). The assumption is that each object is represented with an associativeschema in memory (Anderson 1983), and fit exists to the degree that there are similar-ities between the two schemas in consumer memory. Categorization theory suggests thatan instance fits into a given category pending on the match between its attributes and thetypical attributes of the category as represented by the consumers’ schemas (Rosch1978). Consider a case where the retailer Old Navy wants to expand its portfolio with aline of wristwatches where they want to use both Old Navy and the producer’s brand(e.g., Timex for Old Navy). Old Navy does not extend into watches but joins forces withan established watch brand which brings its competence to the alliance. Generally,brands often form alliances outside their own product category, and fit must therefore be

Mark Lett

Page 3: The multiple roles of fit between brand alliance partners in alliance attitude formation

achieved through other means than product fit. In the Timex for Old Navy case, it islikely that judgment of fit would be based on more abstract associations (Lanseng andOlsen 2012). Even though consumers can assess fit holistically based on overallimpressions, we suggest that the abstract content of brand fit perceptions also mightvary depending on the positioning strategy pursued by the brand or brands in question(Bhat and Reddy 1998). At an abstract level, brand concepts are empirical manifesta-tions of brand positions (Park et al. 1986). Woods (1960) outlined six product classes,from which Park et al. (1986) outlined three brand concepts, describing the overallbenefit position of the brand in terms of functional, experiential, or symbolic benefits.One can say that Old Navy and Timex share similar functional concepts (high conceptfit), yet operate in completely different product categories. Still, fit can be assessed basedon the degree to which they share similar brand concepts (Lanseng and Olsen 2012).

We suggest that when encountering persuasive messages for an alliance like Timexwatches for Old Navy (and importantly, the watch carries both names), consumers willreact with an initial perception of fit between the two brands’ concepts. Subsequently,this perception of fit is utilized when forming an attitude toward the brand alliance. Inthe case of Old Navy + Timex, we expect consumers to perceive fit as being relativelyhigher because the two brands both follow a functional brand concept strategy (Parket al. 1986). If Old Navy chose Omega as alliance partner, perceived concept fit shouldbe considerably lower, as Omega follows a more symbolic brand concept strategy. It iscrucial to acknowledge that from a theoretical point of view, we assume that the level offit (from lower to higher) affects the attitudinal response and that the brand conceptcombinations are empirical bases of fit of particular relevance to brand alliances. Otherbases of fit, not used in the current study, could for example be complementarity inusage situations, overlap in user groups, expertise, and so on. We do not a priori haveany theoretical reason to suggest that high fit obtained through two functional conceptsworks through other mechanisms than high fit obtained through two symbolic con-cepts. Our central question then becomes as follows: how is this fit perception utilizedin the attitude formation process? The basic proposition made in the current research isthat fit will assume different roles depending on level of situational involvement.

2.2 The moderating role of situational involvement

Situational involvement in the ELM literature is different from task involvement. Theformer describes the extent to which the issue (i.e., the brand alliance) at hand hasrelevance for the recipient’s needs, goals, and desires (Petty et al. 1983). The latterdescribes engagement in a task (e.g., solving a riddle) that need not be personallyrelevant and does not necessarily motivate the recipient to scrutinize a message to reacha judgment about the true merits of an issue (Petty et al. 1983). By putting recipients indifferent situations, one can manipulate the extent to which they perceive the issue asmore or less personally relevant, hence the term situational involvement. Managerswho consider using brand alliances in their strategy focus more on situational involve-ment than task involvement. Target groups differ in their needs, goals, and desires and,thus, differ in how personally relevant they perceive the brand alliance. Managers musttherefore take these differences in personal relevance into account when choosingalliance partners. Reflecting this priority, we therefore focus on situational involvementand not task involvement in the current study.

Mark Lett

Page 4: The multiple roles of fit between brand alliance partners in alliance attitude formation

Under different levels of involvement, the ELM posits that variables in thepersuasion context can serve multiple roles. That is, “variables can affect theamount and direction of attitude change by (a) serving as persuasive arguments,(b) serving as peripheral cues and/or (c) affecting the extent or direction ofissue and argument elaboration” (Petty and Cacioppo 1986, p. 16). Which rolemight fit assume?

First, in order to assess fit as an argument (role a), one should consider that in ELM,the term argument refers to any information in a message that permits a person toevaluate the message target (e.g., issue, object, person, etc.) along whatever dimensionscentral for that person (Petty and Cacioppo 1986, p. 18). Brand alliance fit can serve therole as argument if it provides some meaningful information that can be elaboratedupon in attitude formation. For example, “the benefits of brand A fit with brand B’sbenefits, so together the benefits of A + B > A or B independently.” Such elaboration islikely to increase with higher situational involvement. If fit works as an argument, weshould thus observe an interaction between fit and situational involvement, with morepositive responses to fit in high involvement, than low involvement conditions.

Second, to assess fit as a cue (role b), one should consider that in the ELM,peripheral cues refer to stimuli in the persuasion context that can affect attitudes withoutprocessing of message arguments (Petty and Cacioppo 1986, p. 18). Given initialpositive attitudes towards alliance brands A and B, a cue effect would predict thatthe mere observation of (higher) fit between these brands should prompt a positivealliance attitude, i.e., a main effect of fit. A simple inference like “Brand A and B fittogether, and as long as I like brand A and brand B, I like the new A + B” illustratessuch an effect.

Finally, it could also be the case that fit by itself induces recipients toscrutinize the brand alliance message (role c above). If fit between brand alliancepartners facilitates elaboration, higher fit should increase recipients’ sensitivity tothe quality of the information in the brand alliance message(s) they receive (c.f.Petty and Wegener 1998). The inference could be “An alliance between A andB? Let me see what they have to say.” In this case, fit is not elaborated upon asa piece of information itself; it does not produce like or dislike by itself, ratherfit leads the recipient to elaborate on other parts of the message. With this logic,we would see an interaction between fit and argument quality (varied acrossconditions to detect differences in elaboration).

Summing up, we believe that fit can serve different roles dependent on situationalinvolvement. Under low situational involvement, we expect fit to work as a cue (a maineffect), with higher fit leading to more positive alliance attitudes. With increasingsituational involvement, fit obtains more informational value in its own right, produc-ing increased argument quality sensitivity. Consequently, we would expect an interac-tion between fit and argument quality under high situational involvement. Formally, wepropose the following hypothesis:

H1: There will be a three-way interaction between alliance partners’ brand fit,situational involvement, and argument quality on alliance attitude. Under lowsituational involvement, alliance attitude will be more positive under high com-pared to low fit. Under high situational involvement, an interaction between fit andargument quality will show higher argument quality sensitivity in high-fit

Mark Lett

Page 5: The multiple roles of fit between brand alliance partners in alliance attitude formation

compared to low-fit conditions, leaving alliance attitude most positive in the strongargument high-fit condition.

3 Method

The purpose of the experiment was to test the assumption that the effects of fit betweenthe concepts of brand alliance partners on brand alliance attitude are moderated by therecipients’ level of involvement.

To test this assumption, we developed a 2 (brand concept fit: high vs. low)×2(situational involvement: high vs. low)×2 (argument quality: strong vs. weak) betweensubjects factorial design.

3.1 Pretests

Our hypothesis required two conditions of brand fit: high vs. low. In a realworld, high fit could arguably exist with functional-functional, symbolic-symbolic, and experiential-experiential alliances. We have not theorized whythese different combinations would predict different effects since our currentconcern is the level of fit and not on what dimensions these levels of fit arebased. Consequently, in this initial test of brand alliance fit’s potential multipleroles, we sought to establish different levels of brand alliance fit by combining abrand with a functional concept with an alliance partner predominantly associat-ed with either a functional or symbolic concept (Bhat and Reddy 1998; Parket al. 1986). To identify candidate brands, 16 brands were elicited from inter-views with 15 participants from the same sampling frame but different to thosewho participated in the main experiment. They produced a list consisting of eightassumingly functional and eight assumingly symbolic brands. In a second pretest,160 students assessed these brands’ functional vs. symbolic profile using theBhat and Reddy (1998) scale. Each participant rated two brands, yielding 40observations for each brand. Based on this pretest, we chose the men’s wearretail chain Dressmann (equivalent to Old Navy) as the base brand. Dressmannas a functional brand is highly familiar to the target population, and it is acommonplace strategy that clothing chains extend their portfolios with accesso-ries like jewelry and watches. Consequently, wristwatches were chosen as prod-uct category, and Casio (functional–high fit) and Rolex (symbolic–low fit) werechosen as alliance partners.

3.2 Participants and procedure main study

Two hundred thirty-one business school students (52.4 % male, Mage=25) participatedvoluntarily and were randomly assigned to one of eight conditions. Participantsreceived booklets to read and fill out without flipping back and with no time con-straints. The first two pages contained a cover story and attitude measures for a varietyof brands. Page three presented a new cover story, designed to manipulate involvement,and the fourth page contained an advertisement informing participants of the new brand

Mark Lett

Page 6: The multiple roles of fit between brand alliance partners in alliance attitude formation

alliance. On the following pages, participants rated attitudinal responses towards thebrand alliance, attitudes towards the ad, and manipulations checks. Upon leaving thepremises, participants were thanked and debriefed.

3.3 Manipulations and measurement

Based on the pretest, the functional brand Dressmann was chosen as a base brand. Highfit was obtained by combining Dressmann with Casio (also functional), whereas low fitwas obtained by combining Dressmann with Rolex (highly symbolic). In the stimulimaterial, the alliance would be presented as Casio for Dressmann or Rolex forDressmann. According to composite brand alliance theory (Park et al. 1996), Casioand Rolex would then be header brands, and Dressmann would be the modifier. Weused the term “for” not “by” to clarify who were producing for whom. In a pretest, 23participants were asked to rate their familiarity with the three brands on a seven-pointscale. The result was an equivalent high familiarity for the three brands (MDressmann=6.39, MCasio=6.00, and MRolex=6.09). The brands were also equally well-liked on aseven-point scale (MDressmann=4.15,MCasio=3.95, andMRolex=4.05). Consequently, wecould assume that the brands were equally likable and well-known.

Following the recommendations from Petty et al. (1983), we manipulated situationalinvolvement prior to advertisement exposure. Across conditions, all participants readon the page preceding the advertisement a header saying “IMPORTANTINFORMATION!” followed by an encouragement to read the following materialcarefully. This was done to keep task involvement constant across groups. High-involvement participants read that they would see an ad for a bold new alliance, soonto be marketed in their domestic market. In addition, because their answers were soimportant, 100 participants would receive an exemplar of the new product. Low-involvement participants read that the product was to be test-marketed 2 years fromnow, most probably in Southern Europe. Through these means, we sought situationallyto induce a feeling of higher vs. lower personal relevance of the brand alliance byadhering to the recommendations of Petty et al. (1983).

The verbal arguments in the advertisement contained seven statements, out ofwhich six were typical product attributes (glass quality, waterproof rating, casingmaterial, warranty, country of origin, energy source), and the seventh was more like aslogan. The strong and weak argument versions of these six product attributes (Pettyet al. 1983) varied in strength. Twenty-eight participants rated the perceived quality ofeach message claim (either strong or weak) on a seven-point scale (1=very weak, 4=neutral, 7=very strong), and the claims were presented together with a picture of thewristwatch without the brand name. Combined into an index, the strong argumentclaims had M=5.81, significantly higher than the scale midpoint (t=17.86, p<.001),whereas the weak argument claim index had M=1.96, significantly lower than thescale midpoint (t=27.57, p<.001).

The advertisement featured both brands’ logos (Rolex for Dressmann vs. Casio forDressmann). The verbal content varied as described above according to the argumentquality manipulation. The ad also featured a picture of the same male model acrossconditions wearing a wristwatch. A second picture in the ad showed the watch with thenames of Casio and Dressmann vs. Rolex and Dressmann, respectively, on the dial; allother visual aspects were held constant across conditions.

Mark Lett

Page 7: The multiple roles of fit between brand alliance partners in alliance attitude formation

We measured attitude towards the brand alliance with three semantic differentialscales (1=bad/unfavorable/negative, and 7=good/favorable/positive; Petty et al. 1983).

4 Results

4.1 Manipulation checks

To assess the argument quality manipulation, two questions were asked in the last partof the booklet: to what extent did you perceive the information presented in theadvertisement as being “weak–very strong” and “bad–good”? On seven-point scalesanchored 1 and 7. The items were collapsed into an index (Cronbach’s alpha=.92).Participants in the strong argument condition rated the information significantly stron-ger than did participants in the weak argument condition (Mstrong=4.65 vs. Mweak=3.13, F(1,230)=75.22, p<.001).

Perceptions of fit between brand alliance partners were assessed through three items:“to me, the alliance [brand] for Dressmann makes sense,” “[Brand] for Dressmann is anatural alliance,” and “[Brand] and Dressmann fit well together.” An index of thesethree items (Cronbach’s alpha=.89) revealed that fit was perceived as significantlyhigher in the high-fit compared with the low-fit condition (Mhigh fit=4.92 vs. Mlow fit=2.38, F(1,230)=268. 39, p<.001).

We used the Zaichkowsky (1994) involvement scale to test the situational involve-ment manipulation. An involvement index (Cronbach’s alpha=.89) revealed that thehigh-involvement participants felt significantly more personally involved than the low-involvement participants (Mhigh=2.76 vs. Mlow=2.44, F(1,230)=4.23, p<.05).

4.2 Test of hypothesis

We hypothesized that a three-way interaction would show a positive main effect of fit inthe low situational involvement condition, and an interaction between fit and argumentquality in the high situational involvement condition where sensitivity to argumentquality should be higher for high compared to low fit.

A composite measure of brand alliance attitude (Cronbach’s alpha=.93) served as adependent variable in a 2 (brand concept fit: high vs. low)×2 (situational involvement:high vs. low)×2 (argument quality: strong vs. weak) ANOVA; see results in Table 1.

First, there was a significant main effect of fit, with higher fit producing in generalmore positive brand alliance attitude than low fit (Mhigh fit=4.60 vs. Mlow fit=3.22,F(1,223)=70.23, p<.001). Second, the main effect of situational involvement was notsignificant (Mhigh involvement=3.93 vs. Mlow involvement=3.89, F(1,223)<1.0). Third,argument quality had a significant main effect (Mstrong=4.52 vs. Mweak=3.30,F(1,223)=55.14, p<.001) on brand alliance attitude, with stronger arguments givingmore positive attitudes. More importantly, these main effects were qualified by asignificant three-way interaction (F(1,223)=8.27, p<.005).

The pattern of results is visualized in Fig. 1a (low situational involvement) and Fig. 1b(high situational involvement). The interaction between brand concept fit and argumentquality was significant for high involvement (F(1,111)=8.22, p=.005) but not low involve-ment (F(1,112)=1.38, n.s.). Essentially, this shows that brand concept fit only increased

Mark Lett

Page 8: The multiple roles of fit between brand alliance partners in alliance attitude formation

sensitivity to the argument quality manipulation for high-involvement subjects. To sum-marize, these results lend support to our hypothesis and our general prediction that fitassumes different roles depending on elaboration likelihood. Figure 1b shows that at equalargument quality, fit ads considerably to brand alliance attitude when arguments arestrong. It does not seem like fit penalizes weak arguments, but rather rewards strongarguments. This pattern is not evident under low situational involvement (Fig. 1a).Concept fit had a main effect on brand alliance attitude within low involvement (Mlow

fit=3.09,Mhigh fit=4.7, F(1,112)=49.89, p<.001) and high involvement (Mlow fit=3.36,Mhigh fit=4.49, F(1,111)=23.27, p<.001). Argument quality also showed a significantmain effect within low involvement (Mweak=3.58,Mstrong=4.2, F(1,112)=7.47, p=.007)and high involvement (Mweak=3.03, Mstrong=4.83, F(1,112)=59.11, p<.001).

Planned contrasts show that within the low fit, low situational involvement condi-tion, brand alliance attitude was significantly different between strong and weakargument conditions (Mstrong=3.53 vs. Mweak=2.64, F(1,223)=7.69, p<.01). Therewas no significant difference in the high fit, low situational involvement condition(Fig. 1a). In the high situational involvement condition (Fig. 1b), the difference inbrand alliance attitude was significantly lower in the low fit, weak arguments comparedto low fit, strong arguments condition (Mstrong=3.93 vs. Mweak=2.79, F(1,223)=11.64,p<.005). This difference was also significant in the high-fit condition (Mstrong=5.73 vs.Mweak=3.25, F(1,223)=58.85, p<.001).

5 Discussion

Previous research has pointed to the importance of fit in brand alliances (James 2006;Lanseng and Olsen 2012; Park et al. 1996; Simonin and Ruth 1998). Still, none of thesestudies have examined how fit between the alliance partners can serve different roles inexplaining attitudinal responses depending on elaboration likelihood. In the currentstudy, we obtained support for our prediction that fit can serve multiple roles. Theseresults are important for several reasons.

First, the reported study adds insight into how fit between the alliance partners can playdifferent roles as persuasion variables, and thereby extending our understanding oftheoretical mechanisms explaining when and why fit is important between brand alliance

Table 1 Brand alliance attitude scores

Low-situational involvement High situational-involvement

Mean SD (n) Mean SD (n)

Strong arguments

High fit 4.87 0.98 (29) 5.73 0.94(30)

Low fit 3.53 1.59 (30) 3.93 1.40 (27)

Weak arguments

High fit 4.52 0.90 (27) 3.25 1.45 (29)

Low fit 2.64 1.27 (30) 2.79 1.20 (29)

Mark Lett

Page 9: The multiple roles of fit between brand alliance partners in alliance attitude formation

a

b

3.53

4.87

2.64

4.52

0

1

2

3

4

5

6

7

Low fit High fit

Strong arguments Weak arguments

3.93

5.73

2.793.25

0

1

2

3

4

5

6

7

Low fit High fit

Strong arguments Weak arguments

0

Fig. 1 Patterns of alliance attitudes of a low situational involvement condition and b high situationalinvolvement condition

Mark Lett

Page 10: The multiple roles of fit between brand alliance partners in alliance attitude formation

partners in line with the information processing paradigm. An important finding in thecurrent study is that higher fit could not negate the negative impact of weak messagearguments for the more involved participants. In other words, high fit might not besufficient for positive brand alliance attitudes for involved consumers. For example,perceptions of high fit of the brand alliance between Nike and iPod might not be enoughfor active runners, if Nike does not also provide strong arguments to support the quality oftheir running gear. For less involved consumers on the other hand, high fit could besufficient for positive brand alliance attitudes even when arguments are weak. Forexample, Disney’s cartoon characters co-promoted with Kellogg’s cereals would fit wellbased on their experiential images, but Disney can hardly provide strong arguments in thecereal category. However, concept fit might be sufficient for children. A practical impli-cation of these findings is that brand owners should be cautious in selecting brand allianceinitiatives based solely on fit. If it is likely that the consumers find the brand alliancepersonally relevant, and the brand alliance in itself cannot be supported by strongattributes and benefits (arguments) of relevance for the target market, fit in itself betweenthe alliance partners will not be sufficient. Less involved consumers respond just as well toa low-fitting alliance with strong arguments, as more highly involved consumers respondto a high-fitting alliance backed with weak arguments. If well-fitting alliance partners arehard to find, low-fitting partners in alliance backed with strong arguments can be just asviable, especially if one considers the opportunity to increase fit perceptions by exposingthe target audience repeatedly to the alliance message (Lane 2000).

Second, the current study finds that brand alliance attitudes were significantly higherfor strong arguments than for weak arguments in the low-fit conditions. These findingsare consistent with prior research on how incongruent information can increase cogni-tive elaboration (e.g., Gürhan-Canli and Maheswaran 1998; Maheswaran and Chaiken1991; Mandler 1982; Meyers-Levy and Tybout 1989). Low-fitting alliances wouldprobably trigger curiosity as to why these two brands have joined together in analliance (Lee and Schumann 2004). Curiosity could lead to elaboration to resolve thelow-fit perceptions and thereby increase argument quality sensitivity. Brand managersshould take these findings into account and be especially careful with low-fitting brandalliances if these are not supported by strong arguments. As shown by both Meyers-Levy and Tybout (1989) and Gürhan-Canli and Maheswaran (1998), brand managersshould keep incongruence moderate, since highly incongruent brand alliances thatcannot be made sense of most likely will result in less favorable responses.

The restrictive operationalization of fit and use of only two product categoriespotentially limit the current study’s generalizability. There are many bases of fit. Thecurrent study conceptualizes fit as degree of similarities between two schemas inconsumer memory, using the overall brand concepts (Park et al. 1986) as bases of fit.However, fit between alliance partners could be achieved in many ways—e.g., com-plementarity in usage situations, overlap in user groups, and expertise to name a few.Future research should apply other bases of fit and theorize how these combinationscould moderate the current study’s results. Could for example high product fit based oncomplimentary products by itself provide strong arguments for the brand alliance? Ishigh fit in a symbolic brand alliance less sensitive to high-involvement elaborationsince it is harder to assess the merits of symbolic arguments? These and other researchquestions decomposing the fit variable should be fruitful avenues for future research.Another limitation in the current study is that task involvement is kept constant. Future

Mark Lett

Page 11: The multiple roles of fit between brand alliance partners in alliance attitude formation

research could benefit from manipulating both task and situational involvement in thesame study. In addition, multiple roles of fit could affect not only brand allianceattitudes but also post-alliance attitudes towards the individual brands in the brandalliance. Future research should therefore look into spillover effects on partner brandsunder different levels of elaboration.

Acknowledgments The authors would like to thank student Martin Stenstad for his efforts in data collectionand Luk Warlop for comments on an earlier version of this manuscript.

References

Aaker, D. A., & Keller, K. L. (1990). Consumer evaluations of brand extensions. Journal of Marketing, 54(1),27–41.

Anderson, J. R. (1983). The architecture of cognition. Cambridge: Harvard University Press.Arnett, D. B., Laverie, D. A., & Willcox, J. B. (2010). A longitudinal examination of the effects of retailer-

manufacturer brand alliances: the role of perceived fit. Journal of Marketing Management, 26(1–2), 5–27.Bhat, S., & Reddy, S. K. (1998). Symbolic and functional positioning of brands. Journal of Consumer

Marketing, 15(1), 32–43.Gürhan-Canli, Z., & Maheswaran, D. (1998). The effects of extensions on brand name dilution and

enhancement. Journal of Marketing Research, 35(November), 464–473.James, D. O. (2006). Extension to alliance: Aaker and Keller’s model revisited. Journal of Product and Brand

Management, 15(1), 15–22.Lane, V. R. (2000). The impact of ad repetition and ad content on consumer perception of incongruent brand

extensions. Journal of Marketing, 64(2), 80–91.Lanseng, E. J., & Olsen, L. E. (2012). Brand alliances: the role of brand concept consistency. European

Journal of Marketing, 46(9), 1108–1126.Lee, E.-J., & Schumann, D. W. (2004). Explaining the special case of incongruity in advertising: combining

classic theoretic approaches. Marketing Theory, 4(1/2), 59–90.Maheswaran, D., & Chaiken, S. (1991). Promoting systematic processing in lowmotivation settings: the effect

of incongruent information on processing and judgment. Journal of Personality and Social Psychology,61(July), 13–25.

Mandler, G. (1982). The structure of value: accounting for taste. In M. S. Clark & S. T. Fiske (Eds.), Affect andcognition: the 17th Annual Carnegie Symposium on Cognition (pp. 3–36). Hillsdale: Lawrence Erlbaum.

Meyers-Levy, J., & Tybout, A. M. (1989). Schema congruity as a base for product evaluation. Journal ofConsumer Research, 16(June), 39–54.

Park, C. W., Jaworski, B. J., & MacInnis, D. (1986). Strategic brand concept image management. Journal ofMarketing, 50(4), 134–145.

Park, C. W., Jun, S. Y., & Shocker, A. D. (1996). Composite branding alliances: an investigation of extensionand feedback-effects. Journal of Marketing Research, 33(4), 453–466.

Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: central and peripheral routes toattitude change. New York: Springer Verlag.

Petty, R. E., & Wegener, D. T. (1998). Matching versus mismatching attitude functions: implications forscrutiny of persuasive messages. Personality and Social Psychology Bulletin, 24(3), 227–240.

Petty, R. E., Cacioppo, J. T., & Schumann, D. W. (1983). Central and peripheral routes to advertisingeffectiveness: the moderating role of involvement. Journal of Consumer Research, 10(2), 135–146.

Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categoriza-tion (pp. 27–48). Hillsdale: Lawrence Erlbaum Associates.

Simonin, B. L., & Ruth, J. A. (1998). Is a company known by the company it keeps? Assessing the spillovereffects of brand alliances on consumer brand attitudes. Journal of Marketing Research, 35(1), 30–42.

Swaminathan, V., Reddy, S., & Dommer, S. (2012). Spillover effects of ingredient branded strategies on brandchoice: A field study. Marketing Letters, 23(1), 237–251.

Woods, W. A. (1960). Psychological dimensions of consumer decision. Journal of Marketing, 24(3), 15–19.Zaichkowsky, J. L. (1994). The personal involvement inventory: reduction, revision, and application to

advertising. Journal of Advertising, 23(4), 59–70.

Mark Lett