Importance of Context in Brand Extension
Importance of Context in Brand Extension
Importance of Context in Brand Extension
Treating their brands as assets, many firms try to lever- of the decision context is essential for both theoretical and
age their equity by launching new products under estab- practical reasons. At a theoretical level, the decision context
lished brand names. Prior research has demonstrated that may change how consumers process and evaluate brand
consumers’ evaluations of these brand extensions depend extensions. At a practical level, understanding the effect of
primarily on two factors: the perceived quality of the parent the decision context can improve predictions of brand
brand and the perceived fit between the parent brand and the extension success, given that the decision context in the
extension category (Aaker and Keller 1990; Bottomley and marketplace not only varies between shopping situations
Holden 2001). While previous studies have demonstrated but also may differ from the decision context in brand
that the importance of fit can vary as a function of brand and extension concept testing.
consumer characteristics (Kim and John 2008; Yeung and The current research examines how the relative impor-
Wyer 2005), the current research examines how the relative tance of fit versus quality is affected by two basic features
importance of fit versus quality varies as a function of the of the consumer decision context: the presence of visual
consumer decision environment. Understanding the impact information and the presence of competing brands. Both
these features naturally vary across consumer decision envi-
ronments and have been shown to change how consumers
*Tom Meyvis is Associate Professor of Marketing, Stern School of Busi- weigh different attributes in their decisions. First, consumer
ness, New York University (e-mail: [email protected]). Kelly Gold-
smith is Assistant Professor of Marketing, Kellogg School of Management,
decision environments vary in terms of the opportunity to
Northwestern University (e-mail: [email protected]. visualize the product (McCabe and Nowlis 2003). For
edu). Ravi Dhar is George Rogers Clark Professor of Management and example, consumers can visually inspect the extension if it
Marketing, Yale School of Management, Yale University (e-mail: ravi. is a product on a shelf in a grocery store but not if it is a
[email protected]). Stephen Nowlis served as guest editor for this article, and
Ziv Carmon served as associate editor for this article.
product ordered in a restaurant or a product mentioned by a
friend. Furthermore, visual imagery has been shown to
influence how consumers respond to new product introduc- conceptualized as a function of overlapping category asso-
tions (Zhao, Hoeffler, and Dahl 2009) and change the ciations (Aaker and Keller 1990), the match between the
weight consumers assign to different types of information brand image and the extension category (Park, Milberg, and
(Pham, Meyvis, and Zhou 2001). Second, consumer deci- Lawson 1991), or overlapping benefit associations (Bro-
sion contexts also vary in terms of the opportunity to engage niarczyk and Alba 1994).
in brand comparisons, with consumers either evaluating the However, other studies have cast doubt on this assump-
extension product in isolation or choosing between the tion by demonstrating that favorably regarded brand names
extension product and other brands. Comparisons are an can directly increase extension evaluations, regardless of fit
essential feature of consumer choice (Dhar, Nowlis, and (Bottomley and Holden 2001), and that the effect of fit can
Sherman 1999), and the ability to engage in comparisons even disappear completely when consumers have sufficient
has been shown to systematically change the attributes to attribute information to base their evaluation on (Klink and
which consumers attend (Hsee et al. 1999). In summary, we Smith 2001). Furthermore, recent studies have shown that
examine how consumers’ evaluation of a brand extension how consumers react to differences in fit depends on their
(e.g., Heineken popcorn) is affected by two basic variations information processing style, including whether they process
in the decision context: the presence of visual cues (seeing information holistically or analytically (Monga and John
the bag of popcorn) and the availability of comparison 2007) and whether they tend to think abstractly or con-
brands (seeing Heineken in the context of other popcorn cretely (Kim and John 2008). Building on these prior
brands). demonstrations of the effect of processing style on exten-
To understand the process by which the decision context sion evaluations, we propose that common variations in the
can change the relative importance of brand–extension fit decision environment can systematically alter consumers’
versus parent brand quality, we build on recent work by processing style and, as a result, change their relative
Kim and John (2008), who demonstrate that consumers’ emphasis on fit versus quality.
sensitivity to brand–extension fit depends on consumers’
tendency to think abstractly. We propose that the decision CHANGING EXTENSION EVALUATIONS BY
context can change the level of abstraction at which con- CHANGING THE DECISION CONTEXT
sumers represent the extension and, as a result, change how Consumers evaluate brand extensions in a variety of deci-
they evaluate that extension. Specifically, we demonstrate sion contexts. We propose that the visual and comparative
that merely presenting a product picture or presenting the nature of the decision context can change how consumers
extension in the context of other brands in the product cate- mentally represent the extension. According to construal-
gory leads consumers to adopt a more concrete representa- level theory, people can represent objects at different levels
tion of the extension, which not only reduces the importance of abstraction, ranging from lower-level, concrete represen-
of extension fit but also increases sensitivity to parent brand tations that are contextualized and include incidental object
quality. As a result, decision contexts that allow consumers features to higher-level, abstract representations that are
to visualize the extension or compare it with other brands decontextualized and include only the core features of the
shift consumers’ preference from extensions of better-fitting object (Liberman and Trope 2008). We posit that the pres-
brands to extensions of higher-quality brands. ence of visual cues and comparison brands lead consumers
To illustrate, consider a consumer who is evaluating two to adopt a more lower-level, concrete representation of the
extensions in the deodorant category: a Nike deodorant (an brand extension.
extension of a high-quality brand that does not fit well with First, consider the effect of visual cues. Visual informa-
the deodorant category) and a CVS deodorant (an extension tion is readily imaginable and distinctive, whereas verbal
of a better-fitting but lower-quality brand). We propose that information is more pallid and decontextualized. For exam-
the consumer’s liking of the Nike deodorant extension rela- ple, when consumers contemplate the mere concept of a
tive to the CVS deodorant extension will increase in the “Nike deodorant,” their representation is not bounded by a
presence of other deodorant brands or in the presence of a particular context. However, adding visual product cues
visual cue (a picture of a deodorant stick). Before empiri- (however basic) will activate incidental features (e.g., “rec-
cally investigating this prediction, we briefly review prior tangular”) beyond the core features implied by the category
research on brand extension evaluations and discuss how (e.g., “odor protection”), thus facilitating the formation of a
consumers’ reactions to brand extensions can be altered by more vivid and specific image of the product. This suggests
changes in the decision context. that the presence of visual information can activate a lower-
level, more concrete representation. This view is consistent
CONSUMER EVALUATIONS OF BRAND EXTENSIONS: with recent findings that people tend to categorize objects
QUALITY VERSUS FIT into more categories when these objects are represented by
Although early studies on consumers’ response to brand pictures as opposed to words (Amit et al. 2008) and that
extensions acknowledged the importance of the perceived people process pictures more quickly when they represent
quality of the parent brand, they also suggested that this psychologically close objects but process words more
quality perception will transfer to a brand extension only if quickly when they represent psychologically distant objects
consumers perceive a good fit between the brand and the (Amit, Algom, and Trope 2009).
extension category (Aaker and Keller 1990; Boush and Similar to visual cues, the presence of comparison brands
Loken 1991). In more general terms, a good fit between the in the extension category can also influence how consumers
brand and the extension category has been widely consid- mentally represent the brand extension. Comparisons with
ered a necessary condition for favorable consumer reactions other brands in the same product category will highlight the
(Völckner and Sattler 2006), regardless of whether fit is lower-level, incidental differences between the branded
The importance of the Context in Brand extension 3
products, while de-emphasizing the core features of the between a low-fit extension of a high-quality brand (e.g.,
product category, which are shared across brands and thus Nike deodorant) and a better-fitting extension of a lower-
irrelevant to preference (Tversky 1977). This contextualiza- quality brand (e.g., CVS deodorant). A shift in preference
tion and the accompanying emphasis on (distinguishing) from the extension of the better-fitting brand to the exten-
lower-level rather than (shared) higher-level features should sion of the higher-quality brand indicates an increase in the
result in a more concrete representation when consumers importance of quality relative to fit.
consider brand extensions in the context of other brands in In the first set of studies, we examine the effect of visual
the product category rather than in isolation. In summary, cues and find that adding a generic product picture increases
we propose that both visual cues and brand comparisons can preference for extensions of higher-quality, worse-fitting
shift consumers’ representation of brand extensions from a brands (Studies 1a and 1b). Furthermore, we show that this
schematic, abstract representation to a more detailed and shift in preference is caused by a more concrete thinking
concrete representation—thus moving these extensions psy- style because adding the product picture makes people think
chologically closer to the consumer (Liberman, Trope, and more concretely (Study 2) and thinking more concretely
Stephan 2007). shifts preference toward the extensions of the higher-quality
If the decision context changes how consumers represent brands (Study 3). In the second set of studies, we examine
the brand extension, how does it affect their extension the effect of comparisons and find that enabling brand com-
evaluations? Building on prior findings in the brand exten- parisons causes a similar shift in extension preferences
sion and psychological distance research streams, we expect (Studies 4a and 4b) and mental representations (Study 5).
that more concrete representations will increase the impor- Finally, we demonstrate that these shifts in extension pref-
tance of parent brand quality relative to brand–extension fit. erences indeed reflect a change in the relative importance
Kim and John’s (2008) recent work demonstrates that con- consumers attach to fit versus quality, as consumers’
crete mindsets are associated with reduced sensitivity to dif- thoughts shift from extension fit to brand quality (Studies
ferences in fit. Specifically, they show that people who tend 6a and 6b).
to think abstractly evaluate a high-fit extension of a brand
(Nike insoles) more favorably than a low-fit extension of STUDY 1A: THE EFFECT OF VISUAL CUES ON BRAND
that brand (Nike treadmills), whereas people who tend to EXTENSION PREFERENCES
think concretely do not show any such difference. This is In the first study, we examined the effect of visual cues
consistent with findings in the literature on psychological on consumers’ brand extension preferences. We predicted
distance indicating that more abstract mindsets tend to that adding a generic, nondiagnostic product picture would
increase reliance on normative ideals and general principles shift consumers’ preference from high-fit brand extensions
(Kivetz and Tyler 2007; Liberman, Trope, and Stephan to high-quality brand extensions.
2007). Thus, consumers in a more abstract mindset should
place greater weight on the normative appropriateness of Method
the extension—that is, whether the extension fits with the Two hundred twenty-seven college students completed this
image and skills of the parent brand. study, either without compensation after being approached
Conversely, consumers in a more concrete mindset on campus or for compensation in a lab session. Participants
should place greater weight on parent brand quality. People were randomly assigned to either the control condition or
in a concrete mindset tend to have an immediate temporal the picture condition. All participants were shown eight
focus, as well as a focus on lower-level features rather than product categories (two target categories and six filler cate-
higher-level principles (Liberman, Trope, and Stephan gories). Within each product category, they were asked to
2007). It follows that consumers in a concrete mindset choose between extensions of two brands. In the picture
should be primarily concerned about the immediate benefits condition, the two brands in each extension category were
that the brand extension can provide them with—benefits accompanied by the same generic picture of a typical prod-
that can be inferred from the quality of the parent brand. uct with the brand logo imposed on it, whereas in the con-
In summary, we propose that the presence of visual cues trol condition, only the brand logo was presented (for the
or comparison brands will activate more detailed, lower- picture stimuli, see Appendix A). As in subsequent studies,
level representations of the brand extension and, as a result, we counterbalanced the order in which the two brands in
shift consumers’ focus from brand–extension fit to parent each category were presented.
brand quality. We test this proposition by examining how Each target extension pair consisted of an extension of a
changes in the decision context influence people’s prefer- higher-quality national brand and an extension of a better-
ences between extensions that are superior on fit and exten- fitting store brand. We selected Nike deodorant and Speedo
sions that are superior on parent brand quality. camping gear as the low-fit, high-quality national brand
extensions and CVS deodorant and Kmart camping gear as
OVERVIEW OF THE EXPERIMENTS the better-fitting, lower-quality store brand extensions. This
As in previous brand extension research, participants in selection was based on a pretest in which participants (n =
our studies were presented with a set of brand extensions 23) rated the quality of the brands and the fit with the exten-
that they are asked to evaluate. Within this paradigm, we sion category on nine-point scales. According to this pretest,
systematically manipulated the presence of visual cues and participants perceived the Nike brand as significantly higher
the availability of comparison brands and examine how in quality than the CVS brand (MNike = 7.35, MCVS = 5.57;
these changes affect the importance of brand quality rela- t(22) = –4.54, p < .001) but as a significantly worse fit with
tive to extension fit. To infer the relative importance of deodorant (MNike = 4.23, MCVS = 7.66; t(21) = –5.33, p <
quality versus fit, we measure participants’ preference .001). Similarly, they perceived the Speedo brand as signifi-
4 JouRnal oF maRKeTinG ReseaRCh, ahead of Print
cantly higher in quality than the Kmart brand (MSpeedo = 6.45, stimuli, see Appendix B). Unlike in the previous study, no
MKmart = 3.95; t(21) = –4.34, p < .001) but as a significantly brand logos were presented in either condition.
worse fit with camping gear (MSpeedo = 3.77, MKmart = 7.05;
t(21) = 4.76, p < .001). Results and Discussion2
After making their choice for each product category, par- Adding a visual cue again shifted preference from better-
ticipants were presented with a list of the brands used in the fitting brands toward higher-quality brands. Presenting a
study and asked to circle any brand names they did not generic picture of the product category increased preference
know. We excluded participants who did not know one of for Apple suitcases over JanSport suitcases (PControl =
the parent brands for one of the replicates from the analyses 32.2%, PPicture = 46.8%; 2(1) = 5.05, p = .025), for Chick-
for that replicate. We used this same brand familiarity check fil-A guacamole over Taco Bell guacamole (PControl = 23.8%,
in all studies. Brand familiarity did not vary as a function of PPicture = 35.9%; 2(1) = 3.56, p = .059), for Mrs. Field’s
the manipulations for any of the replicates in any of the barbecue sauce over McDonald’s barbecue sauce (PControl =
studies. 16.3%, PPicture = 38.8%; 2(1) = 11.78, p < .001), and for
Nike razors over Bic razors (PControl = 27.1%, PPicture =
Results and Discussion1 40.0%; 2(1) = 3.92, p = .047). These results demonstrate
We expected that adding a visual cue would shift prefer- not only that the picture effect generalizes to choices
ence toward the extension of the higher-quality (but worse- between national brand extensions but also that pictures can
fitting) brand. Consistent with this prediction, we found that increase the relative preference for poor-fitting high-quality
presenting a generic picture of the product category signifi- brands over existing lower-quality products.
cantly increased preference for Nike deodorant over CVS Whereas this study demonstrates the robustness of the
deodorant (PControl = 50.3%, PPicture = 80.0%; 2(1) = 16.23, picture effect, the next two studies test the proposed mecha-
p < .001) and for Speedo camping gear over Kmart camp- nism. We have argued that visual cues increase preference
ing gear (PControl = 68%, PPicture = 80%; 2(1) = 4.67, p = for the extension of the higher-quality (but worse-fitting)
.031). Despite the subtlety of the manipulation, it had a sub- brand by inducing a more concrete representation that shifts
stantial effect on participants’ preferences: Simply provid- consumers’ emphasis from brand–extension fit to parent
ing the outline of a deodorant stick increased the choice brand quality. To test this proposed mechanism, we follow
share of Nike deodorant from 50% to 80%. In summary, the “chain of experiments” approach that Spencer, Zanna,
adding a picture of the product category shifted people’s and Fong (2005) propose: While the first study demon-
preferences from extensions of better-fitting store brands to strated the effect of visual cues on extension preferences,
extensions of higher-quality national brands. the second study tests the effect of visual cues on the pro-
posed mediator, and the third study tests the effect of the
STUDY 1B: THE EFFECT OF VISUAL CUES, proposed mediator on extension preferences. Specifically,
GENERALIZING BEYOND STORE BRANDS Study 2 tests whether the presence of a visual cue activates
Because all lower-quality brands used in the previous study a more concrete mindset, and Study 3 tests whether a more
were store brands, we attempted to replicate the effect using concrete mindset shifts people’s preference to the higher-
only national brands, thus ensuring that the picture effect is quality brand extension.
not dependent on idiosyncratic store brand characteristics.
STUDY 2: THE EFFECT OF VISUAL CUES ON
Method CONSUMERS’ MINDSET
Two hundred thirty-two college students participated in To test whether adding product pictures activates a more
the study for partial fulfillment of a course requirement. All concrete mindset, we first presented participants with a list
participants were shown 11 product categories (4 target of brand extensions that were either accompanied by a pic-
categories and 7 filler categories). The four target pairs con- ture of the product category or not, and then we measured
sisted of Chick-fil-A versus Taco Bell guacamole, Apple participants’ preference for abstract versus concrete repre-
versus JanSport suitcases, Mrs. Field’s versus McDonald’s sentations of a list of behaviors, a method commonly used
barbecue sauce, and Nike versus Bic razors. A separate to measure the level of abstraction of people’s current mind-
pretest (n = 93) using the same scales we used in the previ- set (e.g., Fujita et al. 2006).
ous study confirmed that the first brand in each pair was
perceived as higher quality but a worse fit with the exten- Method
sion category (all Fs (1, 92) > 15.44, ps < .001). Note that Ninety undergraduate students who participated in fulfill-
the four lower-quality brand extensions were not merely a ment of a course requirement were randomly assigned to
better fit with the product category but were existing prod- either the control condition or the picture condition. All par-
ucts. Including existing products enhanced the realism of ticipants were first shown a list of 18 brand extensions,
the procedure and enabled us to test whether visual cues including the target extensions used in the previous studies.
could increase preference for poor-fitting extensions of high- In the control condition, participants were only given the
quality brands over existing offerings by low-quality brands. brand names, whereas in the picture condition, the brand
Participants were again randomly assigned to either the names were accompanied by a generic picture of the exten-
control condition or the picture condition (for the picture sion product. Next, as part of a seemingly unrelated task,
1We removed data from participants who did not know one of the parent 2We removed data from participants who did not know one of the parent
brands for the deodorant (n = 57) or camping gear (n = 12) extensions from brands for the suitcase (n = 7), guacamole (n = 28), barbecue sauce (n =
the analysis for that replicate. 35), or razor (n = 21) extensions from the analysis for that replicate.
The importance of the Context in Brand extension 5
they were given a mindset measure developed by Fujita et scales as used in the previous studies, confirmed that the
al. (2006) based on Vallacher and Wegner’s (1989) behav- first brand in each pair was perceived as higher quality but a
ioral identification form. This measure consisted of a list of worse fit with the extension category (all Fs > 6.00, ps <
eight behaviors (e.g., sweeping the floor), each of which .05).
was accompanied by both a low-level description (e.g.,
moving a broom) and a high-level description (e.g., being Results and Discussion3
clean). For each behavior, participants were asked to circle Compared with participants in the abstract mindset
which of the two statements they thought best described that condition, those in the concrete mindset condition were
behavior. We used the number of high-level statements cir- significantly more likely to prefer Crest over Wal-Mart
cled as an indicator of the level of abstraction of the partici- facial moisturizer (PAbstract = 37.5%, PConcrete = 61.5%;
pant’s mindset. 2(1) = 4.60, p = .032) and marginally more likely to prefer
Häagen-Dazs over ShopRite cottage cheese (PAbstract =
Results and Discussion
41.7%, PConcrete = 62.5%; 2(1) = 2.97, p = .085). Further-
We expected that presenting participants with product more, collapsing across both replicates reveals that induc-
pictures would activate a more concrete mindset. Consistent ing a more concrete mindset significantly increased prefer-
with this prediction, we observed that participants in the ence for the extensions of the worse-fitting but higher-
picture condition circled significantly fewer high-level quality national brands (F(1, 65) = 5.22, p = .026).
statements than those in the control condition (MControl = These results are consistent with Kim and John’s (2008)
4.34, MPicture = 3.41; F(1, 88) = 6.19, p = .015). Whereas finding that consumers who tend to think more concretely
this result confirms that adding generic product pictures can are less sensitive to differences in fit between extensions of
lead consumers to think more concretely, we next test the same brand to multiple categories (i.e., Nike insoles vs.
whether thinking more concretely can shift consumers’ Nike treadmills). We extend their finding by directly manip-
emphasis from brand–extension fit to parent brand quality. ulating people’s mindsets and by varying quality as well as
STUDY 3: THE EFFECT OF CONSUMERS’ MINDSET fit, thus demonstrating that people who are put in a concrete
ON BRAND EXTENSION PREFERENCES mindset through an unrelated task shift their emphasis from
brand–extension fit to parent brand quality.
To examine the effect of consumers’ mindset on the rela- Together with Study 2, these results explain why simply
tive importance of brand–extension fit versus parent brand
adding a generic product picture can substantially alter con-
quality, we first induced either a concrete or an abstract
sumers’ brand extension preferences (as observed in Studies
mindset using a task unrelated to the main study. Next, par-
1a and 1b). Although the pictures in our studies did not add
ticipants were asked to choose between extensions of high-
any diagnostic information, they induced a more concrete
quality national brands and extensions of better-fitting but
mindset (Study 2), and in turn, a more concrete mindset
lower-quality store brands. We expected that participants in
shifted consumers’ emphasis from fit to quality (Study 3).
a concrete (rather than abstract) mindset would be more
likely to choose the extensions of the worse-fitting but The next set of studies examines whether a similar effect
higher-quality national brands. can be produced by another feature of the decision environ-
ment: the presence of other brands in the category.
Method
STUDY 4A: THE EFFECT OF BRAND COMPARISONS
Eighty undergraduate students and members of a national ON EXTENSION PREFERENCES
online panel participated in exchange for either monetary
compensation or lottery prizes. They were randomly To examine how brand comparisons can change con-
assigned to either the abstract or the concrete mindset con- sumers’ extension preferences, we presented participants
dition. We manipulated participants’ mindset using a task with the same target brand extension pairs used in Study 3,
adapted from Fujita et al. (2006). They were asked to either but now manipulated the ease of comparing the two brands
generate superordinate category labels (abstract mindset) or in each extension category. In two of the conditions, we
subordinate exemplars (concrete mindset) for 16 words, made brand comparisons easy by presenting the two brands
such as “singer,” “king,” “chair,” and “car.” For example, next to each other and asking participants either to choose
those in the abstract mindset condition were asked to indi- between the brands (choice condition) or to sequentially
cate what a chair is an example of (e.g., furniture), whereas evaluate the brands (proximal evaluation condition). In con-
those in the concrete mindset condition were asked to pro- trast, in the third condition (isolated evaluation), brand com-
vide an example of a chair (e.g., a desk chair). parisons were made more difficult, and thus less likely, by
Following the mindset manipulation, participants were separating the two brands with filler brands and a distrac-
asked to fill out a second, seemingly unrelated survey in tion task.4 We expected that being able to compare the two
which they were asked to imagine that they were shopping brands would activate a more concrete representation of the
for five products (two target categories and three filler cate-
gories). For each product category, they were presented 3We removed data from participants who did not know one of the parent
with two brands and asked to indicate which brand they brands for the cottage cheese replicate (n = 12) or the moisturizer replicate
would prefer in that category. As in previous studies, each (n = 1) from the analysis for that replicate.
4Because participants in the first two conditions assessed each brand in
target pair consisted of a high-quality, poor-fitting brand and
the context of the brand with which it was paired, they were assumed to be
a lower-quality, better-fitting brand: Crest versus Wal-Mart in joint evaluation mode (Hsee et al. 1999), whereas participants in the
facial moisturizers and Häagen-Dazs versus ShopRite cot- third condition were assumed to be in separate evaluation mode, as they
tage cheese. A separate pretest (n = 67), using the same considered each brand in isolation.
6 JouRnal oF maRKeTinG ReseaRCh, ahead of Print
extensions and thus benefit the extensions of the higher- easily comparable attributes (Nowlis and Simonson 1997),
quality, but worse-fitting brands. we propose that the presence of brand comparisons also
alters the decision maker’s mindset and, as a result,
Method increases the weight of attributes that are congruent with
Two hundred ten people at a large public transit station this activated mindset. Although both brand–extension fit
were asked to fill out a short questionnaire in exchange for and parent brand quality were easily evaluable (i.e., the
a small incentive. Participants were randomly assigned to national brand was clearly superior on quality and clearly
the choice condition, the proximal evaluation condition, or inferior on fit), the presence of brand comparisons increased
the isolated evaluation condition. All participants were pre- the weight of the attribute associated with concrete benefits
sented with two target brand extension pairs (Crest and Wal- (i.e., parent brand quality) at the expense of the attribute
Mart facial moisturizer, Häagen-Dazs and ShopRite cottage associated with abstract principles (i.e., brand–extension
cheese) as well as six filler extension pairs. fit).
Participants in the choice condition were asked to indi-
cate which of the two brands they would choose if they STUDY 4B: THE EFFECT OF BRAND COMPARISONS,
were shopping in that product category (similar to the pro- GENERALIZING BEYOND STORE BRANDS
cedure in the previous studies). In the two other conditions, Because both lower-quality brands in the previous study
participants evaluated each brand extension separately on were store brands, we attempted to replicate the effect using
two nine-point scales adopted from Broniarczyk and Alba only national brands. This will ensure that the effect of
(1994), measuring liking (1 = “dislike,” and 9 = “like”) and brand comparisons is not dependent on specific store brand
perceived performance (1 = “one of the worst,” and 9 = characteristics.
“one of the best”). For each participant, the brand with the
highest average rating on these two scales was inferred to Method
be that participant’s preferred brand for that extension cate- Three hundred two undergraduate and graduate students
gory. In the proximal evaluation condition, the two brands participated in this study either in fulfillment of a course
in each extension category (e.g., Crest and Wal-Mart facial requirement or for monetary compensation. Participants
moisturizer) were presented adjacent to each other, encour-
were randomly assigned to either the isolated evaluation
aging participants to compare the two brands as they formed
condition or the choice condition. The procedure in each
their evaluations. In contrast, in the isolated evaluation con-
condition was identical to that in the corresponding condi-
dition, to minimize the possibility of brand comparisons,
tion in Study 4a. We selected the brand extensions using the
participants first rated one brand from each of the eight
same pretest procedure as in the previous studies, with
extension pairs, then completed a brief distraction task
(unscrambling four unrelated anagrams), and only then exception of a change in the anchors of the quality scale (–4 =
rated the other brand from each extension pair. “extremely low quality,” 0 = “moderate quality,” and +4 =
“extremely high quality”). On the basis of the pretest results
Results and Discussion5 (n = 56), we selected two target pairs, consisting of Subway
Consistent with our prediction, enabling brand compari- versus McDonald’s milkshake machines and Applebee’s
sons significantly increased preference for the extensions of versus Taco Bell Mexican cookbooks. In each pair, the first
the higher quality, but worse-fitting brands. Participants in brand was perceived as higher-quality than the second brand
the isolated evaluation condition were less likely to prefer but a worse fit with the extension category (all F(1, 55) >
Häagen-Dazs cottage cheese over ShopRite cottage cheese 19.45, ps < .001).
(P = 34.0%) than participants in the proximal evaluation Results and Discussion
condition (P = 83.3%; 2(1) = 23.7, p < .001) or in the
choice condition (P = 61.4%; 2(1) = 8.55, p = .003). Fur- Enabling brand comparisons again increased the relative
thermore, participants in the isolated evaluation condition preference for the extensions of the higher-quality but
were also less likely to prefer Crest facial moisturizer over worse-fitting brands. Compared with participants who made
Wal-Mart facial moisturizer (P = 41.8%) than participants isolated evaluations, those who made choices were rela-
in the proximal evaluation condition (P = 64.3%; 2(1) = tively more likely to prefer Subway milkshake machines
5.23, p = .022) or in the choice condition (P = 59.4%; 2(1) = over McDonald’s milkshake machines (PEvaluation = 12.7%,
4.09, p = .043). In summary, participants showed an PChoice = 21.7%; 2(1) = 4.09, p = .043) and Applebee’s
increased preference for the extension of the worse-fitting Mexican cookbooks over Taco Bell Mexican cookbooks
but higher-quality brand when the decision environment (PEvaluation = 41.2%, PChoice = 58.9%; 2(1) = 8.72, p =
made it easier to engage in brand comparisons. .003). Thus, across a variety of brands and extension cate-
Whereas prior research has demonstrated that brand com- gories, we find that adjusting the decision environment to
parisons can change preferences by increasing the impact of enable brand comparisons shifts people’s preference from
extensions of better-fitting brands to extensions of higher-
5We removed data from participants who did not know one of the parent
quality brands. We have argued that the act of comparing
brands for the cottage cheese (n = 84) or moisturizer (n = 29) replicates the brands results in a more concrete representation of the
from the analysis for that replicate. In addition, we removed data from par- brand extension, which shifts consumers’ emphasis from
ticipants who gave identical average evaluations to both cottage cheese brand–extension fit to parent brand quality. In the next
brands (n = 19) or both facial moisturizer brands (n = 17) from the analysis
for that replicate because no preference could be inferred. Distributing
study, we test this proposed mechanism by examining
these participants equally over both brands does not affect the results (see whether enabling brand comparisons indeed activates a
also Novemsky and Dhar 2005; Nowlis and Simonson 1997). more concrete mindset.
The importance of the Context in Brand extension 7
STUDY 5: THE EFFECT OF BRAND COMPARISONS ON extension preferences by activating a more concrete repre-
CONSUMERS’ MINDSET sentation of the extension.
In Study 5, participants first saw a series of brand exten- The studies thus far show that changes in the decision
sions, either in the same format as in the choice condition environment can systematically shift consumers’ prefer-
(and were asked to compare the brands) or in the same for- ences from extensions of better-fitting brands to extensions
mat as in the isolated evaluation condition (and were asked of higher-quality brands. We have interpreted this prefer-
to separately evaluate each brand). After they completed ence shift as reflecting a shift in relative importance from
this task, we measured the level of abstraction of their cur- brand–extension fit to parent brand quality. The next two
rent mindset. We expected that participants who could com- studies seek direct evidence of this shift in emphasis by ask-
pare the brands would be in a more concrete mindset than ing participants to explain their extension preferences and
those who considered each brand separately. measuring the relative number of thoughts about fit versus
quality. First, in Study 6a, we replicated the effect of visual
Method cues on extension preferences (Study 1) and asked partici-
One hundred four undergraduate students participated in pants to describe how they made their choice. We expected
the study for course credit. Participants were randomly that presenting a picture of the product category would
assigned to either the choice or the isolated evaluation con- result in a relative increase in thoughts about brand quality
dition. The procedure was similar to the corresponding con- rather than extension fit.
ditions in Study 4a, including the selection of target and STUDY 6A: THE EFFECT OF VISUAL CUES ON
filler brand extensions and the order in which they were pre- THOUGHTS ABOUT QUALITY VERSUS FIT
sented. However, unlike in Study 4a, participants were not
asked to explicitly choose or evaluate the brands (because Method
explicit responses could interfere with the subsequent mind- Two hundred seventy-four undergraduate students par-
set measure); instead, they were shown the extensions on a ticipated either in fulfillment of a course requirement (n =
computer screen and asked to “think about which product 180) or for monetary compensation (n = 94). They were ran-
you would choose” (in the choice condition) or to “think domly assigned to either the control or the picture condition.
about how you would evaluate each product” (in the iso- The procedure was identical to the procedure used in Study
lated evaluation condition). To make sure that participants 1, except two differences. First, after their choice, partici-
actually followed these instructions (in the absence of pants were asked to “describe how you made your choice.”
explicit responses), we took an a priori measure by includ- Second, because we assumed that the thought listings would
ing an instructional manipulation check (see Oppenheimer, interfere with subsequent questions, we presented partici-
Meyvis, and Davidenko 2009). We omitted participants who pants with only one product category (Nike vs. CVS
failed the instructional manipulation check from the analy- deodorant).
sis, yielding a useable sample of 89 participants. This proce-
Results and Discussion6
dure did not differentially affect the two conditions (2 < 1).
After viewing all brand extensions, participants were We replicated the picture effect: Adding the deodorant out-
asked to fill out a modified version of Vallacher and Weg- line significantly increased participants’ preference for Nike
ner’s (1989) behavioral identification form (adjusted to deodorant over CVS deodorant (PControl = 69.5%, PPicture =
relate to the products shown in the comparison manipula- 81.1%; 2(1) = 4.68, p = .030). Next, we examined the
tion). This mindset measure consisted of a list of 12 con- thought listings to test whether this shift in preference was
cepts (e.g., eating cottage cheese), each of which was driven by a corresponding shift in concerns about quality
accompanied by both a low-level description (e.g., taking versus fit. Two coders who were blind to the hypotheses and
spoonfuls out of a tub) and a high-level description (e.g., conditions recorded the number of thoughts related to par-
taking in nutrition). For each concept, participants were ent brand quality (e.g., “Nike makes quality products”) and
asked to mark the description that best described that con- brand–extension fit (e.g., “CVS knows personal hygiene
cept. We used the number of high-level statements marked products”) for each respondent. The coders agreed on 87%
by each participant as an indicator of the level of abstrac- of the classifications; the remaining ones were resolved
tion of that participant’s mindset. through discussion. Consistent with our predictions, pre-
senting the picture significantly increased the number of
Results and Discussion thoughts related to quality (MControl = .60, MPicture = .91;
Participants’ mindset was reliably influenced by whether F(1, 256) = 14.77, p < .001) and marginally decreased the
they thought about choosing between brands or about sepa- number of thoughts related to fit (MControl = .30, MPicture =
rately evaluating each brand. Participants in the choice con- .19; F(1, 256) = 3.00, p = .056). Thus, the presence of the
dition selected significantly fewer abstract statements (M = visual cue shifted people’s emphasis from fit to quality
5.51) than participants in the isolated evaluation condition (quality–fit difference scores: DControl = .31, DPicture = .72;
(M = 6.42; F(1, 87) = 4.34, p = .040). Thus, people who F(1, 256) = 15.04, p < .001).
compared brands within a product category adopted a more To test whether this change in thought patterns accounts
concrete mindset than people who separately evaluated the for the effect of the picture manipulation on the brand
brands. Together with the prior finding that concrete mind- choices, we also conducted a mediation analysis. As we
sets shift consumers’ preference from high-fit extensions to
high-quality extensions, these results support the hypothesis 6We removed data from participants who did not know one of the parent
that engaging in brand comparisons changes consumers’ brands (n = 15) from the analysis.
8 JouRnal oF maRKeTinG ReseaRCh, ahead of Print
mentioned previously, the picture reliably influenced both resolved through discussion. Consistent with our predic-
participants’ brand choice and their thoughts about quality tions, participants in the choice condition listed significantly
versus fit. Furthermore, participants’ relative thoughts influ- fewer fit-related thoughts (MChoice = .16, MEvaluation = .44;
enced their brand extension choices (2(1) = 16.86, p < F(1, 130) = 11.08, p = .001) and marginally more quality-
.001), and when this effect is controlled for, the picture related thoughts (MChoice = .51, MEvaluation = .31; F(1, 130) =
effect is no longer significant, while the thought effect 3.22, p = .075). In other words, the presence of brand com-
remains reliable (picture effect: 2(1) = 1.76, p = .185; parisons shifted participants’ emphasis from fit to quality,
thought effect: 2(1) = 14.46, p < .001). Finally, the Sobel as indicated by a corresponding shift in thoughts (quality–fit
test indicates that this reduction in the picture effect is reli- difference scores: DChoice = .35, DEvaluation = –.13; F(1, 130) =
able (z = 2.82, p = .005). 14.32, p < .001).
Thus, presenting a product picture shifted participants’ Next, we tested whether these changes in participants’
emphasis from fit to quality, which in turn shifted their pref- thoughts mediated the effect of the brand comparisons on
erence from the deodorant by the better-fitting brand (CVS) their extension preferences. As we mentioned previously,
to the deodorant by the higher-quality brand (Nike). Next, the brand comparison manipulation reliably influenced both
in Study 6b, we examine whether facilitating brand com- extension preferences and the relative number of quality
parisons can produce a similar shift in relative importance. versus fit thoughts. Furthermore, participants’ thoughts reli-
We expected that participants who chose between brand ably influenced their brand preferences (b = –1.41, 2(1) =
extensions, rather than evaluated them in isolation, would 16.36, p < .001), and when this effect is controlled for, the
mention more thoughts about parent brand quality and effect of the brand comparison manipulation is no longer
fewer thoughts about brand–extension fit. significant (b = –.13, 2(1) < 1), whereas the effect of par-
ticipants’ thoughts remains reliable (b = –1.38, 2(1) =
STUDY 6B: THE EFFECT OF BRAND COMPARISONS 13.75, p < .001). Finally, the Sobel test indicates that this
ON THOUGHTS ABOUT QUALITY VERSUS FIT reduction in the comparison effect is reliable (z = 2.26, p =
Method .02). In summary, the results from these last two studies
Two hundred eighty-seven undergraduate students par- demonstrate that the previously observed changes in exten-
ticipated in exchange for monetary compensation. They sion preferences in response to visual cues and brand com-
were randomly assigned to either the choice or the isolated parisons indeed result from a systematic shift in consumers’
evaluation condition. The procedure was identical to the emphasis, a shift away from the abstract principle of exten-
procedure used in the corresponding conditions of Study 4a, sion fit and toward more immediate concerns about brand
except for three differences. First, after indicating their pref- quality.
erence, participants were asked to “list the thoughts you GENERAL DISCUSSION
relied on when making your evaluations (choices).” Second,
to avoid interference of the thought listing with subsequent Consumers’ evaluations of brand extensions have been
questions, we presented participants with only one product shown to depend on the quality of the parent brand and the
category (Crest vs. Wal-Mart facial moisturizer). Third, the fit between the brand and the extension category. The cur-
choice condition also included the option of choosing nei- rent research demonstrates how subtle changes in the deci-
ther of the brands. We added this option to make sure that sion environment can change the relative importance of
any difference between the choice and the evaluation condi- these two factors by changing how consumers represent the
tions could not be due to people in the choice condition brand extension. We proposed that adding visual cues or
being forced to establish a preference. comparison brands activates a more concrete mindset,
which in turn shifts consumers’ focus from the fit of the
Results and Discussion7 brand extension to the quality of the parent brand. Consis-
First, we replicated the effect of brand comparisons on tent with this hypothesis, presenting a nondiagnostic picture
extension preferences. Compared with participants in the of the product category (Studies 1a and 1b) and enabling
isolated evaluation condition, those in the choice condition brand comparisons (Studies 4a and 4b) both shifted partici-
were more likely to prefer Crest facial moisturizer over pants’ preferences toward extensions of higher-quality but
Wal-Mart facial moisturizer (PIsolated = 52.0%, PChoice = worse-fitting brands. Furthermore, in line with our account
69.1%; 2(1) = 3.84, p = .05). Second, we examined the of these effects, these same changes in the decision context
thought listings to test whether this shift in preference was activated a more concrete mindset (Studies 2 and 5), and in
driven by a corresponding shift in concerns about quality turn a more concrete mindset shifted preferences from
versus fit. Two coders who were blind to the hypotheses and extensions of better-fitting brands to extensions of higher-
conditions recorded the number of thoughts related to par- quality brands (Study 3). Finally, these preference shifts
ent brand quality (e.g., “Crest is a brand you can trust”) and indeed reflect a change in consumers’ emphasis from fit to
to brand–extension fit (e.g., “Crest has no expertise in skin quality, as the increased preference for the high-quality
care products”) for each respondent. The coders agreed on brand extensions was mediated by an increase in the num-
93.1% of the classifications; the remaining ones were ber of thoughts about quality relative to the number of
thoughts about fit (Studies 6a and 6b).
7We removed data from participants who did not know one of the parent Theoretical and Managerial Implications
brands (n = 38) from the analysis. In addition, we excluded data from par-
ticipants who gave identical average ratings to both brands in the evalua-
Although prior brand extension research has examined
tion condition (n = 21) and participants who circled the “neither” option in various influences on the extension evaluation process, it
the choice condition (n = 98) because no preference could be inferred. has not examined how consumers’ brand extension evalua-
The importance of the Context in Brand extension 9
tions vary as a function of the decision environment. Our fit and quality effects—as is the case when managers try to
finding that the decision environment systematically predict the success of a proposed extension.
changes how consumers process and evaluate brand exten- In addition to predicting consumers’ reactions to exten-
sions has important implications for both managerial and sions, managers can use our findings to enhance consumers’
academic branding research. Specifically, given that deci- extension evaluations by communicating more effectively.
sion environments vary in the marketplace, researchers For example, a high-quality brand that is introducing an
examining the potential success of a brand extension should extension to a distant category would benefit from a com-
consider the nature of the consumer decision context in their munication context that encourages brand comparisons
studies. (e.g., through comparative advertising) and presents the
For example, consumers sometimes evaluate extensions extension as vividly as possible. Conversely, a lower-quality
in the absence of other brands (e.g., when buying yogurt at brand that is introducing an extension in an adjacent cate-
McDonald’s) or in the absence of visual cues (e.g., when gory should promote isolated evaluations of their product
purchasing a Virgin Atlantic plane ticket online). This (e.g., by placing it in an end-of-aisle display) to reduce the
absence of brand comparisons and visual cues is particu- emphasis on parent brand quality.
larly likely when consumers are evaluating an extension The current research also provides a theoretical contribu-
without actually shopping for it (e.g., when hearing about tion to the emerging literature on construal-level effects in
the brand extension from a radio advertisement or from a consumer choice by demonstrating another factor that can
friend). For such cases, brand extension studies that include influence the level of abstraction at which people represent
brand comparisons and visuals in the decision context will a decision. Prior research has shown that people use more
tend to overestimate consumers’ liking of poor-fitting exten- abstract representations with increasing temporal, social,
sions of high-quality brands. and spatial distance (Liberman, Trope, and Stephan 2007),
However, in most shopping situations, consumers can when the event is hypothetical rather than real (Wakslak et
visually inspect the extension and compare it with other al. 2006), or when people are chronically disposed to think
brands in the product category (e.g., when shopping in a more abstractly (Vallacher and Wegner 1989). Our results
grocery store). In those cases, brand extension studies that indicate that in addition to these inherent characteristics of
fail to match the rich context of this environment will tend the target or the person, the level of abstraction of people’s
to underestimate the potential of poor-fitting extensions of representations depends on the subtle features of the deci-
high quality brands. This is particularly relevant for aca- sion environment. The same consumer can represent a given
demic studies of brand extensions, which usually do not brand extension at an abstract level when the extension is
include comparisons or visuals. Indeed, although some brand presented verbally and in isolation, but at a more concrete
extension studies involve choices between brands (e.g., level when the extension is accompanied by a visual cue or
Broniarczyk and Alba 1994; Swaminathan, Fox, and Reddy by competing brands in the same category.
2001), most studies present respondents with extensions While our research examined how extension preferences
that are to be evaluated either in isolation or in the context shift as a function of changes in the decision context and in
of other extensions of the same brand. For example, partici- consumers’ mindset, we speculate that these changes in con-
pants are asked to indicate their attitudes toward Lufthansa text and mindset also correspond to a change in perspective.
suitcases (Yeung and Wyer 2005) or toward Häagen-Dazs In particular, the abstract mindset associated with the
popcorn, cottage cheese, and candy bars (Aaker and Keller absence of both visual and competitive contexts may be
1990). Furthermore, these extensions are typically only similar to the perspective of a marketing manager, whereas
described verbally (i.e., without any visual cues). Thus, for the concrete mindset may be more similar to the perspective
shopping situations in which brand comparisons and visual of a typical shopper. Indeed, because taking a manager’s per-
cues are available, most academic studies of extension spective implies a consideration of other consumers’ pur-
evaluations would underestimate the importance of parent chase intentions, managers will experience a greater psy-
brand quality and overestimate the importance of brand– chological distance to the decision and thus form more
extension fit. abstract representations (Liberman, Trope, and Stephan
Our findings should not be taken to imply that a good fit 2007). In other words, the impoverished decision environ-
between the brand and the extension category cannot be ment typical of many marketing surveys may lead survey
critically important for favorable extension evaluations. respondents to adopt a mindset that is closer to a managerial
Indeed, even in conditions in which visual cues and com- perspective than to a shopper’s perspective.
parison brands were present, a significant number of partici- To examine this possibility, we conducted an additional
pants still preferred the extension of the lower-quality but study in which we asked undergraduate students (n = 270)
better-fitting brand. Thus, regardless of the nature of the to indicate their purchase intentions for the same target and
decision context, many people still value fit sufficiently to filler brand extensions used in Study 3 (1 = “definitely not
compensate for the difference in parent brand quality. Fur- interested,” and 9 = “definitely interested”). Approximately
thermore, because our stimuli did not include any extreme half the respondents were first asked to take the perspective
examples of low-fit brand extensions, such as Nike cookies of a brand manager and indicate whether it would be a good
or Häagen-Dazs backpacks, the current research cannot idea for the brand to introduce the extension (1 = “terrible
speak to the potential of these extreme extensions. As such, idea,” and 9 = “great idea”). The other participants were
rather than questioning the conceptual insights gained from simply asked to imagine being a consumer shopping for
previous extension research, our results demonstrate the these products. As we expected, first taking the brand
importance of taking the decision environment into account manager’s perspective reliably shifted consumers’ prefer-
whenever it is important to assess the specific magnitude of ences from higher-quality national brands to better-fitting
10 JouRnal oF maRKeTinG ReseaRCh, ahead of Print
store brands (Crest facial moisturizers to Wal-Mart facial analyses of Studies 6a and 6b demonstrate the role of qual-
moisturizers: F(1, 265) = 9.52, p = .002; Häagen-Dazs cottage ity and fit perceptions in changing participants’ brand exten-
cheese to ShopRite cottage cheese: F(1, 177) = 6.15, p = .014). sion preferences. Furthermore, the results of the thought
Thus, priming people with the outside perspective of a mar- protocol studies also help address two potential limitations
keting manager has a similar effect on extension prefer- of the current research.
ences, as does removing comparison brands and visual cues. First, whereas the two sets of brands used in our studies
Finally, although we have focused on consumers’ reac- do not systematically differ on any other dimensions, they
tions to brand extensions, the effects documented here differ on both fit and quality. Although this implies that we
should generalize to other consumer judgments that are can only draw conclusions about changes in the relative
guided by abstract principles. For example, our framework importance of fit versus quality, the thought protocol stud-
would predict that, in general, the presence of visual infor- ies suggest that both factors are affected. In Study 6a, the
mation and competing brands should reduce the importance presence of a product picture significantly increased the
consumers place on principles of fit, including the fit number of quality thoughts and marginally decreased the
between a brand and a celebrity endorser and the fit between number of fit thoughts, whereas in Study 6b, the presence
a brand’s personality and consumers’ self-concept. A con- of competing brands marginally increased the number of
sumer may generally dislike the idea of a family snack brand quality thoughts and significantly decreased the number of
being endorsed by a likeable but infamous rock star. How- fit thoughts. These findings indicate that changes in the
ever, when considering purchasing the product in the gro- decision environment can affect both the perceived impor-
cery store, a consumer may be more influenced by his or her tance of quality and the perceived importance of fit, though
liking of the endorser than by the lack of fit with the brand. the actual magnitude of each effect may depend on the
nature of the context manipulation.
Limitations and Further Research
Second, although we have argued that the decision envi-
An inherent limitation to the use of real brands as stimuli ronment affects consumers’ preferences by changing the
is that the brands can differ on other dimensions than the importance of fit and quality, it is possible that it changes
ones we intended to manipulate, that is, other than brand– consumers’ perceptions of fit and quality instead. Again, the
extension fit and parent brand quality. This is even more of thought protocol studies help address this issue. In these
an issue given that several of our studies used store brands studies, the effect of the decision context manipulations on
as the lower-quality brands and national brands as the people’s preference was mediated by the relative number of
higher-quality brands. To address this issue, we conducted a quality-related versus fit-related thoughts. This indicates
follow-up survey (n = 26) with participants from the same that the effect of the decision environment is at least partly
population used in our studies to test for any unintended due to changes in the relative importance of these two fac-
systematic differences between the lower-quality brands tors. However, whether the changes in consumers’ decision
and the higher-quality brands. First, participants indicated context can affect perceptions of fit and quality as well as
how familiar they were with each of the parent brands (1 = the relative importance of fit versus quality remains a ques-
“not at all familiar,” and 9 = “very familiar”), revealing that tion for further research.
two sets of brands did not differ in familiarity (Mlow quality =
7.51, Mhigh quality = 7.29; F < 1, not significant [n.s.]). Sec- REFERENCES
ond, participants indicated for each of the parent brands to Aaker, David A. And Kevin L. Keller (1990), “Consumer Evalua-
what extent they agreed that it “makes a lot of different tions of Brand Extensions,” Journal of Marketing, 54 (January),
products” (1 = “completely disagree,” and 9 = “completely 27–41.
agree”). Not surprisingly, the four low-quality store brands Amit, Eleanor, Daniel Algom, and Yaacov Trope (2009), “Distance-
were perceived as offering more diverse products than the Dependent Processing of Pictures and Words,” Journal of
corresponding high-quality national brands (Mlow quality = Experimental Psychology: General, 138 (August) 400–415.
7.67, Mhigh quality = 5.52; F(1, 24) = 18.8, p < .001). How- ———, ———, ———, and Nira Liberman (2008), “Thou Shalt Not
ever, for the remaining six (non–store brand) replicates, Make Unto Thee Any Graven Image: The Distance Dependence
of Representation,” in The Handbook of Imagination and Men-
there was no difference in perceived brand breadth between tal Simulation, Keith Markman, Bill Klein, and Julie Suhr, eds.
the low-quality national brands and the high-quality New York: Psychology Press, 53–68.
national brands (Mlow quality = 5.58, Mhigh quality = 5.88; F < Bottomley, Paul A. and Stephen J.S. Holden (2001), “Do We Really
1, n.s.). Finally, given that a few of the brand extensions Know How Consumers Evaluate Brand Extensions? Empirical
already existed, we also asked participants to circle any Generalizations Based on Secondary Analysis of Eight Studies,”
extensions they had “personally tried before.” For five of Journal of Marketing Research, 38 (November), 494–500.
the ten replicates, not a single participant indicated having Boush, David M. and Barbara Loken (1991), “A Process-Tracing
had any experience with either the low-quality or the high- Study of Brand Extension Evaluation,” Journal of Marketing
quality brand extension, ruling out any explanation based Research, 28 (February), 16–28.
on prior product experience. Broniarczyk, Susan M. and Joseph W. Alba (1994), “The Impor-
In summary, across the ten replicates used in our studies, tance of the Brand in Brand Extensions,” Journal of Marketing
Research, 31 (May), 214–28.
we did not identify any consistent differences between the Dhar, Ravi, Stephen M. Nowlis, and Steven J. Sherman (1999),
two sets of brands other than the intended difference in “Comparison Effects on Preference Construction,” Journal of
brand–extension fit and parent brand quality, thus ruling out Consumer Research, 26 (December), 293–306.
any possible alternative accounts based on differences in Fujita, Kentaro, Yaacov Trope, Nira Liberman, and Maya Levin-
brand familiarity, perceived brand breadth, or prior experi- Sagi (2006), “Construal Levels and Self-Control,” Journal of
ence. In addition, the thought protocols and mediation Personality and Social Psychology, 90 (March), 351–67.
The importance of the Context in Brand extension 11
appendix a
sTudy 1a: sTimuli used in The PiCTuRe CondiTion
Hsee, Christopher K., George F. Loewenstein, Sally Blount, and Liberman, Nira and Yaacov Trope (2008), “The Psychology of
Max H. Bazerman (1999), “Preference Reversals Between Joint Transcending the Here and Now,” Science, 322 (November),
and Separate Evaluations of Options: A Review and Theoretical 1201–1205.
Analysis,” Psychological Bulletin, 125 (September), 576–90. ———, ———, and Elena Stephan (2007), “Psychological Dis-
Kim, Hakkyun and Deborah Roedder John (2008), “Consumer tance,” in Social Psychology: Handbook of Basic Principles,
Response to Brand Extensions: Construal Level as a Moderator Vol. 2, Arie W. Kruglanski and E. Tory Higgins, eds. New York:
of the Importance of Perceived Fit,” Journal of Consumer Psy-
Guilford Press, 353–83.
chology, 18 (2), 116–26.
McCabe, Deborah B. and Stephen M. Nowlis (2003), “The Effect of
Kivetz, Yifat and Tom R. Tyler (2007), “Tomorrow I’ll Be Me: The
Effect of Time Perspective on the Activation of Idealistic Versus Examining Actual Products or Product Descriptions on Consumer
Pragmatic Selves,” Organizational Behavior and Human Deci- Preference,” Journal of Consumer Psychology, 13 (4), 431–39.
sion Processes, 102 (March), 193–211. Monga, Alokparna Basu and Deborah Roedder John (2007), “Cul-
Klink, Richard R. and Smith, Daniel C. (2001), “Threats to the tural Differences in Brand Extension Evaluation: The Influence
External Validity of Brand Extension Research,” Journal of of Analytic Versus Holistic Thinking,” Journal of Consumer
Marketing Research, 38 (August), 326–35. Research, 33 (March), 529–36.
12 JouRnal oF maRKeTinG ReseaRCh, ahead of Print
appendix B
sTudy 1B: sTimuli used in The PiCTuRe CondiTion
Novemsky, Nathan and Ravi Dhar (2005), “Goal Fulfillment and Swaminathan, Vanitha, Richard J. Fox, and Srinivasan K. Reddy
Goal Targets in Sequential Choice,” Journal of Consumer (2001), “The Impact of Brand Extension Introduction on
Research, 32 (December), 396–404. Choice,” Journal of Marketing, 65 (October), 1–15.
Nowlis, Stephen M. and Itamar Simonson (1997), “Attribute–Task Tversky, Amos (1977), “Features of Similarity,” Psychological
Compatibility as a Determinant of Consumer Preference Rever- Review, 84 (July), 327–52.
sals,” Journal of Marketing Research, 34 (May), 205–218. Vallacher, Robin R. and Daniel M. Wegner (1989), “Levels of Per-
Oppenheimer, Daniel, Tom Meyvis, and Nicolas Davidenko sonal Agency: Individual Variation in Action Identification,”
(2009), “Instructional Manipulation Checks: Detecting Satisfic- Journal of Personality and Social Psychology, 57 (4), 660–71.
ing to Increase Statistical Power,” Journal of Experimental Völckner, Franziska and Henrik Sattler (2006), “Drivers of Brand
Social Psychology, 45 (4), 867–72. Extension Success,” Journal of Marketing, 70 (April), 18–34.
Park, C. Whan, Sandra Milberg, and Robert Lawson (1991), Wakslak, Cheryl J., Yaacov Trope, Nira Liberman, and Rotem
“Evaluations of Brand Extensions: The Role of Product Feature Alony (2006), “Seeing the Forest When Entry Is Unlikely:
Similarity,” Journal of Consumer Research, 18 (September), Probability and the Mental Representation of Events,” Journal
185–93. of Experimental Psychology: General, 135 (November),
Pham, Michel, Tom Meyvis, and Rongrong Zhou (2001), “Beyond 641–53.
the Obvious: Chronic Vividness of Imagery and the Use of Yeung, Catherine W.M. and Robert S. Wyer Jr. (2005), “Does Lov-
Information in Decision Making,” Organizational Behavior and ing a Brand Mean Loving Its Products? The Role of Brand-
Human Decision Processes, 84 (March), 226–53. Elicited Affect in Brand Extension Evaluations,” Journal of
Spencer, Steven J., Mark P. Zanna, and Geoffrey T. Fong (2005), Marketing Research, 42 (November), 495–506.
“Establishing a Causal Chain: Why Experiments Are Often Zhao, Min, Steve Hoeffler, and Darren Dahl (2009), “The Role of
More Effective than Mediational Analyses in Examining Psy- Imagination-Focused Visualization on New Product Evalua-
chological Processes,” Journal of Personality and Social Psy- tion,” Journal of Marketing Research, 46 (February), 46–55.
chology, 89 (6), 845–51.