Brand Extensions of Experiential Goods: Movie

Sequel Evaluations
SANJAY SOOD
XAVIER DRÈZE*

We examine movie sequels as brand extensions of experiential goods. Study 1

reveals a reversal of the traditional categorization model such that dissimilar ex-
tensions are rated higher than similar extensions. This reversal is moderated by
the name of the sequel; numbered sequels (Daredevil 2) are influenced by similarity
more than named sequels (Daredevil: Taking It to the Streets). Study 2 reveals
that the reversal arises because numbered sequels invoke a greater degree of
assimilation with the parent movie, thereby increasing consumers’ level of satiation
of experiential attributes. The Internet Movie Database (IMDb) provides external
validity for our results (study 3).

H ollywood has begun branding movies in a way similar

to that in which consumer-packaged-goods manufac-
turers brand their products. In this research we examine
contexts with predictions in experiential-goods contexts
such as movies. Past research has consistently found that
extensions in similar categories (e.g., line extensions) are
movie sequels as brand extensions in an experiential context. consistently rated higher than extensions in dissimilar cat-
With sequels, studios try to capitalize on the success of an egories (e.g., category extensions) in a physical-product con-
original movie by producing another film that reprises the text (e.g., Aaker and Keller 1990). In the context of movies,
same characters evolving in a new situation. As average we propose that experiential-attribute overlap (i.e., genre of
movie production and marketing costs continue to escalate the sequel) is more relevant than physical-attribute overlap
beyond $100 million (MPAA 2004), sequels represent an as a measure of similarity. Consistent with previous brand-
increasingly important new product introduction strategy for extension research, we expect that attitudes toward movie
the studios. Since 1980, sequels have grossed over $20 bil- sequels will be influenced by perceived similarity with the
lion at the box office. The average annual box-office revenue original movie. However, in contrast to traditional models,
for sequels has more than doubled to $1.9 billion annually we predict that this extension evaluation process reverses
this decade, compared to $718 million in the 1990s. Pre- such that consumers will prefer dissimilar (vs. similar)
movie sequels because consumers tend to satiate on expe-
dicting the success of any individual sequel, however, is
riential attributes (McAlister 1982). That is, consumers may
difficult. For every highly touted success such as Spiderman
prefer that experiential attributes such as the story line of
2, there are a greater number of disappointments such as
the sequel include different genres from the original because
Miss Congeniality 2 or Barbershop 2.
people do not want to see the parent movie again in the
In this research we examine individual sequels as brand sequel. Thus, increased assimilation with the parent brand
extensions using categorization models as our theoretical may also increase the likelihood that satiation occurs during
foundation. We contrast predictions in traditional-product sequel evaluations.
In addition, we examine whether the naming strategy used
*Sanjay Sood is assistant professor of marketing, UCLA Anderson
School of Management, 110 Westwood Plaza, Suite B414, Los Angeles, to launch the sequel can moderate the proposed effects of
CA 90095-1481 (sood@ucla.edu). Xavier Drèze is assistant professor of assimilation with the parent movie. Satiation on experiential
marketing, the Wharton School of the University of Pennsylvania, Phila- attributes such as the plot becomes more likely to the extent
delphia, PA 19104-6340 (xdreze@wharton.upenn.edu). Correspondence: that the parent movie is used as the basis for evaluations.
Sanjay Sood. The authors acknowledge the helpful input of the editor,
associate editor, and reviewers. In addition, the authors thank the Enter- The naming of the sequel is one method to vary the degree
tainment and Media Management Institute of UCLA for helping fund the of activation of the parent movie and therefore vary the
research. degree of assimilation with the sequel. In particular, we
suggest that sequels using a numbering title strategy (e.g.,
Dawn Iacobucci served as editor and Joseph Priester served as associate
editor for this article. Daredevil 2) will depend more upon perceived similarity to
the original movie than sequels using a naming title strategy
Electronically published October 9, 2006
(e.g., Daredevil: Taking It to the Streets) because the num-
352

䉷 2006 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 33 ● December 2006

All rights reserved. 0093-5301/2006/3303-0008$10.00
BRAND EXTENSIONS OF EXPERIENTIAL GOODS 353

bering strategy relies more heavily on the original movie as image-related characteristics, or specific brand associations,
a basis for evaluations. improves extension evaluations.
Using a multimethod research approach, we examine movie- Our research diverges from past brand-extension research
sequel evaluations in three studies that provide supporting evi- in that we predict that, when the products being extended
dence regarding the effects of perceived similarity and naming are experiential and intangible in nature (e.g., movies), dis-
strategies. In study 1, we investigate sequels that differ in terms similar extensions will be preferred to similar extensions.
of perceived similarity and title-naming strategy using a tra- We base our predictions on categorization models with the
ditional laboratory experiment. In study 2, we incorporate a caveat that experiential attributes have a different basis for
computer-based experiment with response latencies to more evaluation compared to tangible attributes. According to the
closely examine the degree of assimilation during sequel eval- categorization model, assimilation with the parent brand im-
uations. In study 3, we analyze a movie database with over proves evaluations when extensions are similar because the
2.8 million consumer ratings of sequels launched over a 48 yr. activated parent-brand associations, typically search attrib-
period from 1957 to 2005 to provide external validity to our utes such as cavity protection for Crest, are favorable in
laboratory results and examine predictions beyond the labo- similar extension contexts such as mouthwash (Keller 1993).
ratory setting. For movie sequels, the parent-brand associations that come
to mind are likely to be experiential attributes such as the
original movie’s story line, its genre, and memorable scenes.
CONCEPTUAL DEVELOPMENT These attributes are typically featured in movie trailers and
television ads; hence, they should be relatively easy to recall.
Brand-Extension Research In contrast to physical goods, we suggest that experiential
attributes such as the story line and genre tend to satiate
Past research has conceptualized a brand as a category such that consumers prefer to experience something differ-
in memory (e.g., Aaker and Keller 1990; Boush and Loken ent in the sequel; hence, dissimilarity is preferred to simi-
1991). Brand-extension evaluations are moderated by the larity. Although high similarity provides a closer connection
perceived similarity between the parent-brand category in to the original film, in experiential contexts this process of
memory and the extension category (Gurhan-Canli and Ma- assimilation is more likely to result in satiation and may
heshwaran 1998; Keller and Aaker 1992). When perceived therefore lower sequel evaluations. For example, if the orig-
similarity is high, extensions are assimilated with the parent
inal movie is an action/adventure film, consumers may be
brand and affect is transferred from the parent brand to the
more attracted to a sequel that also includes a new genre
extension. Perceived similarity has been most often defined
such as a romance relative to a sequel that simply continues
as some form of physical relationship in terms of feature
the previous theme.
overlap. When feature overlap is high then the extension
The effects of satiation in experiential-goods settings are
category is similar and evaluations improve compared to
when feature overlap is low and the extension category is well established in psychology and consumer behavior.
dissimilar. For example, Crest is a brand of toothpaste; Coombs and Avrunin (1977) used physiological measures
hence, an extension to a similar category such as mouthwash to demonstrate that individuals become satiated with certain
is rated higher than an extension to a more dissimilar cat- attributes after consumption of an experiential good exceeds
egory such as shaving cream (Aaker and Keller 1990). a specified level. When consumers satiate on a product, they
Subsequent extension research has shown that perceived prefer to choose a product with different attributes on the
similarity can be defined in terms of intangible attributes next occasion (Lattin and McAlister 1985; McAlister 1982).
and brand-specific associations unrelated to the category. In Satiation has been identified as a mechanism underlying
some instances, intangible attributes influence extension variety-seeking behavior—defined as the preference for
evaluations more than tangible attributes. Physical similarity change when consumers choose sequences of experiential
is less important when the relationship between the parent goods (Kahn, Ratner, and Kahneman 1997; McAlister 1982).
and the extension is based on brand-concept consistency Consumers prefer to experience something new, perhaps as
(Park, Milberg, and Lawson 1991). For example, although a means to maintain an optimal level of stimulation (Berlyne
a kitchen timer is physically more similar to a watch than 1970; Raju 1982). Variety seeking therefore represents a
a pair of cuff links, the Rolex brand extends more favorably preference for dissimilarity over similarity when consumers
to the latter category because of the shared association with evaluate experiential-product experiences. This has been
status. Physical similarity has also been shown to be less shown with music (Ratner, Kahn, and Kahneman 1999),
important when the parent-to-extension relationship is based snack foods (Simonson 1990), and appetizers at restaurants
on a relevant brand-specific association (Broniarczyk and (Ratner and Kahn 2002). Within experiential goods, variety
Alba 1994). For example, although oatmeal is more similar seeking has been shown to be higher for movies versus beer
to cereal than lollipops, Froot Loops extends more favorably and soft drinks (Trivedi, Bass, and Rao 1994). Finally, in a
to the latter category because the color association provides 25 yr. review of the sensory-satiation literature, Inman
a shared connection that is relevant to lollipops. Thus, one (2001) proposed that consumers are more likely to seek
common theme in past brand-extension research is that high variety in experiential attributes such as taste than nonex-
perceived similarity, whether defined as physical attributes, periential attributes such as brand name. Surveys as well as
354 JOURNAL OF CONSUMER RESEARCH

scanner panel data verified a general pattern of behavior of plain how new members of a category (e.g., a sequel) be-
greater variety seeking among flavors than among brands, come integrated with the existing category in memory (e.g.,
consistent with the notion that consumers satiate on expe- the parent film). Most consistent new members of a category
riential attributes. Thus, although not commonly considered are assimilated with existing category knowledge such that
in extension research, satiation can provide some insight to similarities are encoded and recalled more frequently than
brand extensions in experiential contexts. differences (Sujan and Bettman 1989). In contrast, subtyped
instances are highly memorable for their differences from
HYPOTHESES the category rather than similarities. Subtyped instances are
therefore not merely integrated into the existing category
We predict that, consistent with satiation of experiential structure; rather, they are placed in a subcategory that is
attributes, consumers will rate dissimilar sequels higher than conceptually separate but linked to the original (Sujan and
similar sequels. In the movie industry, the experiential cat- Bettman 1989; Weber and Crocker 1983).
egories are identified by genres such as action/adventure, In a branding context, subtyping has been associated
comedy, and so on. Indeed, it is common for the entertain- with subbranded extensions. Subbranding is a form of
ment industry to describe movies by genre: video stores brand extension that combines a parent brand name with
organize rentals according to genres, and movie Web sites an individual name to form the name of a brand extension
often group films by genre. Hence, we defined similarity of (e.g., Courtyard by Marriott). Research suggests that ex-
the sequel via the genre described for the new film relative tensions that include the parent brand name only (e.g.,
to the original. Marriott) are more likely to be assimilated with parent-
In our studies we incorporate new genres in the written brand knowledge structure, while subbranded extensions
plot descriptions as a means to manipulate perceived sim- (e.g., Courtyard by Marriott) are more likely to be subtyped
ilarity of the sequel. Similar extensions introduced a new as distinct from the parent brand (Milberg, Park, and
story line for the sequel in the same genre as the original McCarthy 1997).
film; dissimilar extensions introduced a new story line that We suggest that the title strategy for sequels affects the
also included a different genre from the original film. For degree of assimilation of the sequel and consequently in-
example, Daredevil was an action movie that primarily fo- fluences the likelihood of satiation with the sequel’s story
cused on a superhero fighting crime in New York City. In line. Similar to a parent-name-only branding strategy, a
the context of a sequel, satiation may occur with the story numbered sequel title (e.g., Daredevil 2) relies heavily on
of the Daredevil as an action-oriented superhero. Satiation knowledge of the original movie (e.g., Daredevil) as a basis
therefore implies that consumers would prefer the sequel for evaluations of the sequel. In contrast, a named sequel
plot to be different in some substantive way so that the new title (e.g., Daredevil: Taking It to the Streets) relies less
movie would not be viewed as too close to the original. heavily on the original movie as a basis for evaluations
Adding a new genre to the sequel such as a romance element because the added part of the name cues novelty in the plot.
may provide the necessary variety to make the new film In addition, the extra phrase may help to subtype the sequel
seem more interesting. Note, however, that this new story as potentially offering a different experience than the orig-
element makes the film less similar to the original than if inal. If numbered sequels are more likely to be assimilated
the sequel were to update the story with more action centered with the original movie and subject to satiation than named
on the Daredevil character (which is presumably what peo- sequels, then perceived similarity should significantly affect
ple liked about the original movie). numbered sequels but not named sequels. Given that dis-
The name of the sequel should influence the degree of similarity should be more desirable than similarity with ex-
assimilation between the sequel and the original film, periential extensions, we derive the following hypothesis:
thereby moderating sequel evaluations. As the degree of
assimilation increases, sequel evaluations rely more heavily H1: There will be an interaction between naming strat-
on thoughts about the original movie; hence, assimilated- egy and perceived similarity in sequel evaluations.
sequel evaluations should be more subject to satiation. We Dissimilar extensions will be rated more favorably
investigate two naming strategies for sequel titles that may than similar extensions when a numbering strat-
result in varying degrees of assimilation during evaluations. egy is used; there will be no significant difference
One strategy, which we term the numbering strategy, simply in sequel evaluations when a naming strategy is
adds a number to the original title to signify that the new used.
movie is a sequel (e.g., Daredevil 2). An alternative strategy,
which we term the naming strategy, adds a phrase to the We examine the assimilation process more closely by
original title instead of simply numbering the sequel (e.g., including latency measures and manipulating the order of
Daredevil: Taking It to the Streets). Both of these title strat- presentation of information about the sequel. As discussed
egies are quite common, as evidenced by recent sequels such above, if assimilation is more prevalent with numbered se-
as Shrek 2 and Bridget Jones: The Edge of Reason. quels, then numbered (vs. named) sequel evaluations should
Categorization research on subtyping indicates that these rely more heavily on information stored in memory about
naming strategies may differ in the degree to which they the parent movie. Assimilation therefore involves faster
invoke the process of assimilation. Subtyping models ex- evaluation response times because consumers base opinions
BRAND EXTENSIONS OF EXPERIENTIAL GOODS 355

of the sequel on preformed opinions about the original movie STUDY 1: SEQUELS AND PERCEIVED
(Boush and Loken 1991; Sujan and Bettman 1989). Con- SIMILARITY
versely, evaluation times should be slower for named sequels
because consumers rely more heavily on new information Method
learned about the sequel as opposed to preformed opinions
about the original (Milberg et al. 1997). In addition, this Participants and Design. Participants in this study
assimilation-based process implies that recall of the sequel were students at a large West Coast university. Two hundred
plot description will be reduced for numbered sequels be- thirty-eight participants were randomly assigned to four con-
cause evaluations rely more heavily on the original movie. ditions. They completed the questionnaire in exchange for
As a result, response times should be faster and recall of cash. The study was a 2 (similarity: similar or dissimilar)
the plot description should be lower as the degree of assim- # 2 (sequel title: numbered or named) factorial design with
ilation with the parent movie increases. two within-subject movie replicates.
Assimilation can also be examined by manipulating the Procedure and Materials. The first page of the ques-
presentation order of information. If the sequel title is pro- tionnaire informed participants that they would be evalu-
vided before the plot description, then the parent-movie ating movie sequels that were going to be released in the
category should be activated to a greater extent when the near future. The next page was headed by the title of the
sequel is numbered (vs. named), leading to a greater degree first sequel, which used either a numbered or a named title
of assimilation. If the sequel title is provided after the plot strategy (e.g., Daredevil 2 vs. Daredevil: Taking It to the
description, however, then the parent-movie activation is Streets). The title was followed by a brief description of the
equalized across naming strategies, and evaluations of sequel’s plot. The plot description was organized in three
numbered sequels should more closely resemble named sections. The first section indicated that the original movie’s
sequels. In summary, assimilation is more likely when a primary actors were also going to be part of the sequel. The
numbered title is provided before the description relative second section provided some brief details of the sequel’s
to when a numbered title is provided after the description. plot, closely following the genre of the original movie (e.g.,
In contrast, named sequels should not be as subject to action). These two sections comprised one paragraph and
assimilation, and therefore, order of presentation should constituted the entire plot description for the similar con-
not have as much of an effect. This leads to the following dition. In the dissimilar condition, a third section was added
three hypotheses: that described part of the plot that included a genre that was
different from the original movie (e.g., romance in addition
H2: There will be an interaction between naming strat- to action). An example stimulus is provided in the appendix.
egy and order of presentation in sequel evalua- After reading the plot description, participants evaluated
tions. Numbered extensions will be rated more the sequel on six scales: bad movie/good movie, forget it/
favorably when the title is presented after the de- must see, uninteresting/interesting, wait for rental/see open-
ing night, will be a flop/will be a hit, and sounds worse than
scription than when the title is presented before
most films/sounds better than most films. All were seven-
the description. There will be no significant dif-
point scales where higher numbers indicated more favorable
ference in sequel evaluations when a naming strat- evaluations.
egy is used. The next page included the title of the second sequel (e.g.,
Meet the Parents 2 or Meet the Parents: The Honeymoon)
H3: There will be an interaction between naming strat- followed by a plot description and the same method for
egy and order of presentation in reaction times of manipulating similarity. Finally, participants were asked to
sequel evaluations. Numbered extensions will be provide similarity ratings between the sequel and the original
rated faster when the title is presented before the as a manipulation check.
description than when the title is presented after
the description. There will be no significant dif- Results. The manipulation check revealed that sequels
ference in reaction times when a naming strategy described with the same genre were perceived to be more
is used. similar to the original movie than sequels described with
an added genre (Msim p 3.92 vs. Mdis p 4.65; F(1, 235) p
4.9, p ! .01).
H4: There will be an interaction between naming strat-
The six evaluation scales were combined into an overall
egy and order of presentation in recall of sequel sequel evaluation index for the analysis (Cronbach’s a p
descriptions. Numbered extensions will have a 0.95). A 2 # 2 # 2 (two replicates) mixed ANOVA was
higher recall when the title is presented after conducted to test the effect of the replicate factor. This
the description than when the title is presented analysis revealed a main effect of the movie replicate such
before the description. There will be no signif- that the sequel to Daredevil was rated lower than the se-
icant difference in recall when a naming strategy quel to Meet the Parents (MDD p 3.17 vs. MMTP p 4.40;
is used. F(1, 235) p 126.6, p ! .01) but did not reveal any signifi-
356 JOURNAL OF CONSUMER RESEARCH

cant interactions involving the replicates (either three way FIGURE 1

or two way); hence, the data were collapsed across repli-
SEQUEL EVALUATION BY SIMILARITY AND TITLE STRATEGY
cates. The analysis reveals a main effect of title strategy
(F(1, 235) p 21.1, p ! .01), a main effect of similarity
(F(1, 235) p 5.5, p ! .05), and an interaction between title
strategy and similarity (F(1, 235) p 4.9, p ! .05).
The results of the analysis are illustrated in figure 1. As
predicted, planned contrasts revealed that, for numbered
extensions, similar sequels were evaluated significantly
lower than dissimilar sequels (Msim p 3.18 vs. Mdis p 3.77;
F(1, 235) p 8.2, p ! .01). For named sequels, there was no
effect of similarity because similar sequels were evaluated
the same as dissimilar sequels (Msim p 4.05 vs. Mdis p
4.07; F(1, 235) ! 1, p 1 .2).

Discussion
The first study investigated movie sequels as a type of
brand extension in an experiential-goods context. Two find-
ings diverged from traditional brand-extension research on
packaged goods. First, when evaluating numbered sequels,
consumers preferred dissimilar sequels to similar sequels. a key, the computer screen revealed a second section de-
Assuming that the numbered sequel is more likely to be termined by the order manipulation. Participants in the title-
assimilated with the parent movie, this indicates that con- first condition read the title of the sequel first (either num-
sumers prefer sequels that differ in a substantive way from bered or named) and then pressed a key to read the plot
the original. In addition, similarity did not have any influ- description. Participants in the title-last condition read the
ence if the sequel used a naming strategy. This is interesting plot description first followed by the sequel title. In this
because it seems as though sequel evaluations can be im- study, only the Daredevil sequel was evaluated (the release
proved either through a substantive change in content via of Meet the Fockers prevented us from using Meet the Par-
additional genres or by a superficial change in the title (i.e., ents again). After reading the title and description, partici-
naming vs. numbering). pants pressed a key to provide their movie evaluations on
We proposed that the interaction found in the first study the same six dimensions used in the first study. Following
emerges because the numbered sequels were more likely to the evaluations, participants were given a surprise recall task
be assimilated with the original movie. As a result, when where they were asked to record as many details about the
evaluating a numbered movie sequel, information about the sequel plot description as they could remember.
sequel is more likely to be subject to the effects of satiation.
We investigate this assimilation account more closely in Results. The six evaluation measures were collapsed to
study 2 by examining process measures including response provide an overall index (Cronbach’s a p 0.91). As pre-
times and incidental recall. In particular, as stated in hy- dicted, there was a significant interaction between presen-
potheses 2–4, we expect that the order of title presentation tation order and title strategy (F(1, 167) p 4.1, p ! .05).
(i.e., before or after the plot description) will reduce the This interaction is illustrated in figure 2A. Consistent with
degree of assimilation for numbered but not named sequels. the assimilation-based model of evaluations, when the title
was provided before the description, the numbered sequel
was rated significantly lower than the named sequel
STUDY 2: ASSIMILATION AND ORDER (Mnum p 2.64 vs. Mname p 3.12; F(1, 167) p 4.4, p ! .05).
OF PRESENTATION There were no differences in evaluations when the title was
provided after the description. No other effects were sig-
Method nificant.
Participants and Design. One hundred seventy-three The response-latency measures of the evaluation task re-
students at a large East Coast university completed the sur- veal the same interaction between presentation order and
veys in exchange for course credit. The study included a 2 title strategy (F(1, 167) p 7.0, p ! .01; see fig. 2B). Con-
(title strategy: numbered or named) # 2 (presentation order: sistent with our prediction, evaluation response times were
title first or title last) between-subjects factorial design. Par- significantly faster when the numbered title was presented
ticipants were randomly assigned to a condition. first compared to when the named title was presented first
(Mnum p 19.1 sec. vs. Mname p 27.1 sec.; F(1, 167) p
Procedure. Participants completed the survey on a per- 18.8, p ! .01). Mirroring the pattern of sequel evaluations,
sonal computer. The first screen introduced the task of eval- there was no significant difference in response times if the
uating a movie sequel that will be released. After pressing description was presented first, but there was a significant
BRAND EXTENSIONS OF EXPERIENTIAL GOODS 357

FIGURE 2 cantly reduced when the numbered title was presented first
compared to when the named title was presented first
SEQUEL EVALUATION (A), EVALUATION RESPONSE TIME (B),
AND PLOT RECALL (C) BY PRESENTATION ORDER AND (Mnum p 24.4 words vs. Mname p 35.6 words; F(1, 143) p
TITLE STRATEGY 6.0, p ! .05). Once again, there was no significant difference
in recall if the description was presented first. No other
effects were significant. We should note that in this analysis,
only 149 data points were used because 24 participants did
not provide an answer.

Discussion
The pattern of interactions between presentation order and
naming strategy suggests that numbered-sequel evaluations
involved a greater degree of assimilation, indicating greater
reliance on the original movie as a basis for evaluations.
When the numbered title was shown after the sequel de-
scription, respondents took longer to evaluate the sequels,
they recalled more about the sequels, and they evaluated the
numbered sequels more favorably. In contrast, evaluations
of sequels with named titles seemed to invoke assimilation
to a lesser degree because order of presentation did not affect
evaluations. Consistent with a more piecemeal processing
strategy, named sequels took longer to evaluate, and recall
of sequel information was higher relative to a numbered
sequel.
Summing across these two studies, the results thus far
suggest that the naming-title strategy for sequels dominates
the numbering strategy. Especially for similar story lines,
named sequels tend to have higher evaluations. Since both
types of movie-sequel-naming strategies are commonly used
in practice, we obtained a database of movie sequels to
examine the external validity of our experimental results.
Although there are many factors that influence the accep-
tance of sequels (e.g., marketing budgets, number of screens,
and competitive films with similar plots), we can examine
whether perceived similarity and title-naming strategy ac-
tually lead to different sequel evaluations in the marketplace.

STUDY 3: INTERNET MOVIE DATABASE

In order to explore the external validity of the present set
of results, we obtained a database of sequels from the In-
ternet Movie Database (IMDb; http://www.imdb.com). The
sequel database includes movies spanning a 48 yr. period
(from 1957 to 2005). We extracted all of the movie sequels
released in this period, the original movies, the year of re-
lease, the movie genre, and a user rating for each movie.
We then removed movies that were erroneously coded as
sequels and movies for which the numbering scheme was
ambiguous (e.g., Teen Wolf Too). This left us with a final
database of 317 sequels.
In addition to testing our conceptual factors in the mar-
ketplace, we could test new predictions that we did not
main effect of title strategy (F(1, 167) p 8.0, p ! .01), where examine in the laboratory experiments. According to our
named sequels take longer to evaluate than numbered sequels. model of movie-sequel evaluations, a sequel is more likely
Finally, the recall measures also revealed the same in- to be successful if it is named (vs. numbered). The financial
teraction between presentation order and title strategy success of movies is difficult to gauge, given the competitive
(F(1, 143) p 4.1, p ! .05; see fig. 2C). Recall was signifi- and market-dynamic effects affecting box office (Ainslie,
358 JOURNAL OF CONSUMER RESEARCH

Drèze, and Zufryden 2005), the importance of secondary an ANOVA on the log of the ratio of the sequel rating
markets such as international box office and DVD sales or relative to its parent. Results indicate significant main effects
rentals (Neelamegham and Chintagunta 1999), and the great for similarity (F(1, 316) p 5.0, p ! .05) and naming strat-
variability in production costs. However, we can measure egy (F(1, 316) p 4.2, p ! .05). The interaction is not sig-
the success of a sequel by the likelihood of releasing sub- nificant (F(1, 316) p .5, p 1 .4). The results are illustrated
sequent sequels. That is, we posit that a studio will only in figure 3. Due to the colinearity inherent to field data, the
commission a second sequel if the first sequel was deemed statistics reported here are based on the Type III sums of
to be successful. It would not make business sense to pro- squares (SS) rather than Type I. This is a more conservative
duce a third or a fourth film in a series if the predecessors test of our hypotheses (Type I SS yield F’s of 8.8, 5.1, and
flopped. Thus, the likelihood of releasing a third or higher .5 for similarity, naming, and the interaction, respectively).
numbered sequel should be greater when the title of the To examine the prediction regarding the likelihood of
previous sequel is named versus numbered. future sequels, we ran a logistic regression using the pres-
In order to examine these predictions, we coded the movies ence of a sequel to the sequel as a dependent variable. That
in the IMDb according to the experimental factors in study is, we coded a sequel as successful if it spawned further
1. For the title-naming factor, we coded each sequel as having sequels (e.g., Beverly Hills Cop 3 or Batman Forever). Since
a numbered title if the only change in name compared to the the database extends to 2005, we took precautions when
preceding movie in the series was the presence of a number estimating the likelihood of future sequels. For instance, the
(e.g., Spiderman 2); otherwise, the sequel was coded as a reason why there is no Spiderman 3 is more likely due to
named title (e.g., Bridget Jones: The Edge of Reason). For the fact that Spiderman 2 came out in 2004 rather than to
the genre factor, we compared the genres listed for each sequel the lack of success of the first sequel (Spiderman 2 grossed
to the genres listed for its predecessor (a movie can be of $370 million domestically). Hence, we dropped recent mov-
multiple genres, e.g., action/comedy). If the sequel was de- ies from our analysis and only considered movies launched
scribed as having genres that were not present in the prede- before 2000. This left us with 250 movies in the database.
cessor, it was coded as a dissimilar extension; otherwise, the The 5 yr. cutoff was chosen in light of the fact that 75% of
sequel was coded as being a similar extension. sequels are released within 5 yr. of their predecessor.
Finally, we incorporated user ratings as the dependent The results of the logistic regression are consistent with
variable. Ratings are provided by users who visit the IMDb the ANOVA performed on ratings. We find significant main
Web site. Each movie can be rated on an 11-point scale effects for both similarity (x 2 p 7.6, p ! .01) and naming
ranging from zero to 10, with higher numbers indicating (x 2 p 3.9, p ! .05) but no interaction (x 2 p .1, p 1 .5).
more favorable ratings. The number of user ratings provided
for each movie varies; however, the totals are considerable Discussion
since the Web site is extremely popular. There is an average
number of 8,826 ratings for each sequel in our database, or Consistent with study 1, sequels that were dissimilar to
a total of about 2.8 million user ratings in the entire database. their parent movie (i.e., sequels with a different genre) re-
Although the database allows us to examine the degree ceived higher consumer ratings than sequels that were sim-
of correspondence between the laboratory results and the ilar to their parent movie. Also consistent with study 1,
marketplace, it is important to note that the field study is named sequels received higher consumer ratings. In fact,
constrained relative to the first two studies. Most important,
the lack of experimental control in the real-world data dic- FIGURE 3
tates caution in directly comparing the results across meth- IMDb MOVIE RATING BY SIMILARITY AND TITLE STRATEGY
odologies. In our experiments, we kept the descriptions con-
stant such that the same story line could be tested in all
treatment conditions. In contrast, the movies in the database
belong to only one condition, and therefore, the cell means
compare different movies with different story lines. Simi-
larly, the laboratory experiments controlled movie quality;
the sequels were all the same except for the conceptual
manipulations. The database, however, includes movies of
differing quality across the cells. If movie studios act stra-
tegically when releasing sequels, then there may be some
systematic reasons for naming a sequel differently or com-
municating a different genre that would dilute the strength
of our conceptual factors studied in the laboratory.
To control for differences in quality across movies and
account for the fact that each movie belongs to only one
cell in study 1, we coded the ratings in terms of the relative
change from one movie to the next. That is, we performed
BRAND EXTENSIONS OF EXPERIENTIAL GOODS 359

named sequels that were dissimilar in genre received eval- ation. Although the title-order manipulation in study 2 is
uations that were almost as high as the parent movie. In consistent with this explanation, it would be helpful to have
addition, in the database we could further examine the long- direct evidence of satiation. Future research should examine
term effects of naming strategy by examining the likelihood satiation in more detail. In addition, we have operationalized
of future sequels. As implied by studies 1 and 2, named the experiential attributes in terms of a movie’s story line.
sequels were more likely than numbered sequels to be fol- Other experience-based categories such as television and
lowed by another sequel. music may provide interesting points of comparison to
There are important differences between the database re- movie sequels. For example, in television, spin-offs are often
sults and the results of studies 1 and 2. First, the interaction created on the basis of an original hit show. Similar to
between similarity and title-naming strategy failed to reach movies, there are highly visible successes such as Frasier
significance in the database. In addition to our earlier cau- as well as highly visible disappointments such as Joey. It
tions regarding direct comparisons, in our laboratory studies would be interesting to compare and contrast the determi-
respondents rated sequels that they had not seen (in fact, nants of success in a broader set of experiential categories.
the sequels did not exist yet); in the database, the ratings The results have several theoretical and managerial im-
were presumably provided by people who had seen the plications. First, dissimilarity is preferred to similarity when
movie. This, no doubt, lessens the impact of the naming consumers evaluate experiential brand extensions presum-
factor for similar movies because the actual movie content ably because of assimilation with the parent movie. Since
will reveal the truth, whereas our evaluations capture ex- sequels are released years after the original movie, this result
pectations. Second, in our studies respondents were ran- indicates that assimilation-based satiation may extend over
domly assigned to a condition. Random assignment was not a much longer period of time than previously considered.
possible in the database, and therefore, the results could be Previous research has typically focused on satiation with
due to the fact that individual differences between people time horizons of seconds or days, not years as in our studies.
lead to differences in the types of movies they choose to Note that this is predicted rather than experienced satiation,
see and how they rate those movies afterward. reflecting that consumers have intuitions about the effects
Nevertheless, the main effects of similarity and sequel- of satiation (Ratner and Novemsky 2003).
naming strategy factors emerged despite these differences. Second, the experimental results suggest that the title of
Even more important, our prediction regarding the likeli- the sequel can influence how consumers process information
hood of future named sequels also reached significance in about the movie. Numbered sequels relied heavily on the
the database. original movie as a basis for evaluations, and a simple
change to a named sequel seemed to diminish the degree
GENERAL DISCUSSION of assimilation. Investigations of naming strategy show that
similar superficial changes to a name can lead to profound
Three studies examined movie sequels as a type of ex-
periential brand extension. In contrast to traditional cate- influence on evaluations. Naming effects have occurred in
gorization models, our results revealed a reverse pattern for brand-extension evaluations (Milberg et al. 1997), ingredient
the effects of perceived similarity based on the degree of branding (Desai and Keller 2002), and children’s evaluations
assimilation and the subsequent likelihood of satiation. of brand extensions (Zhang and Sood 2002). Given this
Study 1 revealed that evaluations of numbered movie se- substantial impact that the brand name serves in terms of
quels improved when the sequels were dissimilar as opposed evaluations, future research should more systematically in-
to similar, whereas named sequels did not depend on sim- vestigate how the components of a brand name affect in-
ilarity. Study 2 showed that the likelihood of satiation is formation processing.
dependent upon activation of the original movie; numbered In terms of managerial implications, the results suggest
sequels were more likely to be assimilated with the original that studios could improve the reception of sequels by using
movie, leading to lower evaluations, faster response times, naming strategies for the title. In this article we did not
and reduced recall of the sequel’s plot. Study 3 found two manipulate the type of naming strategy used, but the results
of the experimental main effects actually extended to real- imply that the type of name itself may have an effect. Spe-
world sequel evaluations over a 48 yr. period. Dissimilar cifically, our titles cued respondents about something dif-
sequels were rated higher than similar sequels, and sequels ferent in the sequel (e.g., Daredevil: Taking It to the Streets),
with named titles were rated higher than sequels with num- and this cue was reinforced in the actual plot description.
bered titles in the IMDb. In addition, named (vs. numbered) However, other naming strategies may not cue something
sequels were more likely to result in the launch of yet an- new about the sequel and therefore would presumably be
other sequel. Summing across these three studies, we con- more likely to lead to assimilation (e.g., The Matrix Re-
clude that for experiential attributes, the evaluations of brand loaded). It would be interesting to know if a superficial name
extensions reverse the traditional pattern because consumers change can lead to very different ratings of otherwise iden-
evidently value dissimilarity over similarity. tical experiences. Note that we examined moderately dis-
There are several limitations of the current research that crepant sequels; if the sequel descriptions were highly dis-
should be further investigated. Collectively, the three studies crepant (e.g., completely new plot), then a different pattern
reported provide indirect evidence for the process of sati- of evaluations may have emerged.
360 JOURNAL OF CONSUMER RESEARCH

Summing across studies, we would expect that naming Inman, J. Jeffrey (2001), “The Role of Sensory-Specific Satiety in
changes will mostly affect opening-week box-office results Attribute-Level Variety Seeking,” Journal of Consumer Re-
because the moviegoers who purchase a ticket on opening search, 28 (June), 105–20.
day are in a similar position to that of the participants of Kahn, Barbara E., Rebecca K. Ratner, and Daniel Kahneman
(1997), “Patterns of Hedonic Consumption over Time,” Mar-
studies 1 and 2. These consumers know the movie title and
keting Letters, 8 (1), 85–96.
probably have seen a trailer or read a description of the Keller, Kevin L. (1993), “Conceptualizing, Measuring, and Man-
movie. In contrast, moviegoers who purchase a ticket on aging Customer-Based Brand Equity,” Journal of Marketing,
subsequent weeks will have a chance of being exposed to 57 (March), 1–22.
word of mouth and, thus, might have an opinion about the Keller, Kevin L. and David A. Aaker (1992), “The Effects of
movie that is based on the movie content more than on the Sequential Introduction of Brand Extensions,” Journal of
movie title. The responses of these moviegoers then are more Marketing Research, 29 (February), 35–50.
likely to resemble the results of study 3. Interestingly for Lattin, James M. and Leigh McAlister (1985), “Using a Variety-
the studios, the naming strategy may be a useful tool for Seeking Model to Identify Substitute and Complementary Re-
improving the opening-weekend reception. lationships among Competing Products,” Journal of Market-
ing Research, 22 (August), 330–39.
McAlister, Leigh (1982), “A Dynamic Attribute Satiation Model
APPENDIX of Variety-Seeking Behavior,” Journal of Consumer Research,
9 (September), 141–50.
SEQUEL PLOT DESCRIPTION EXAMPLE Milberg, Sandra J., C. Whan Park, and Michael S. McCarthy
(1997), “Managing Negative Feedback Effects Associated
DAREDEVIL 2 with Brand Extensions: The Impact of Alternative Branding
Strategies,” Journal of Consumer Psychology, 6 (2), 119–40.
Ben Affleck and Jennifer Garner reprise their roles in the MPAA (2004), “U.S. Entertainment Industry: 2003 MPA Market
next installment following the crime-fighting adventures of Statistics,” Motion Picture Association of America, http://
Daredevil Matt Murdock and Elektra Natchios. This time the www.mpaa.org.
Daredevil gets caught in a turf battle between two rival gangs Neelamegham, Ramya and Pradeep Chintagunta (1999), “A Bayes-
in Hell’s Kitchen, New York. The gangs show no mercy, ian Model to Forecast New Product Performance in Domestic
fighting each other with automatic weapons, fire bombs, and and International Markets,” Marketing Science, 18 (2), 115–36.
samurai-style swords. The Daredevil must harness his special Park, C. Whan, Sandra Milberg, and Robert Lawson (1991), “Eval-
uation of Brand Extensions: The Role of Product Level Sim-
powers in new ways in order to keep the streets of New York
ilarity and Brand Concept Consistency,” Journal of Consumer
safe. Research, 18 (September), 185–93.
In addition to the action, in this movie the story line focuses Raju, Puthankurissi S. (1982), “Optimum Stimulation Level: Its
more on the romance between Daredevil and Elektra, adding Relationship to Personality Demographics and Exploratory
a new dimension to the Daredevil story [dissimilar condition]. Behavior,” Journal of Consumer Research, 7 (3), 272–82.
Ratner, Rebecca K. and Barbara E. Kahn (2002), “The Impact of
Private versus Public Consumption on Variety-Seeking Be-
REFERENCES havior,” Journal of Consumer Research, 29 (2), 246–57.
Ratner, Rebecca K., Barbara E. Kahn, and Daniel Kahneman
Aaker, David A. and Kevin L. Keller (1990), “Consumer Evalu-
(1999), “Choosing Less-Preferred Experiences for the Sake
ations of Brand Extensions,” Journal of Marketing, 54 (Jan-
of Variety,” Journal of Consumer Research, 26 (1), 1–15.
uary), 27–41.
Ratner, Rebecca K. and Nathan Novemsky (2003), “The Time
Ainslie, Andrew, Xavier Drèze, and Fred Zufryden (2005), “Mod-
eling Movie Lifecycles and Market Share,” Marketing Sci- Course and Impact of Consumers’ Erroneous Beliefs about
ence, 24 (3), 508–17. Hedonic Contrast Effects,” Journal of Consumer Research,
Berlyne, Daniel E. (1970), “Novelty, Complexity, and Hedonic 29 (4), 507–16.
Value,” Perception and Psychophysics, 8 (5A), 279–86. Simonson, Itamar (1990), “The Effect of Purchase Quantity and
Boush, David M. and Barbara Loken (1991), “A Process Tracing Timing on Variety-Seeking Behavior,” Journal of Marketing
Study of Brand Extension Evaluations,” Journal of Marketing Research, 27 (May), 150–62.
Research, 28 (February), 16–28. Sujan, Mita and James R. Bettman (1989), “The Effects of Brand
Broniarczyk, Susan M. and Joseph W. Alba (1994), “The Impor- Positioning Strategies on Consumers’ Brand and Category
tance of the Brand in Brand Extension,” Journal of Marketing Perceptions: Some Insights from Schema Research,” Journal
Research, 31 (May), 214–28. of Marketing Research, 26 (November), 454–67.
Coombs, Clyde H. and George S. Avrunin (1977), “A Theorem Trivedi, Minakshi, Frank M. Bass, and Ram C. Rao (1994), “A
on Single-Peaked Preference Functions in One Dimension,” Model of Stochastic Variety-Seeking,” Marketing Science, 13
Journal of Mathematical Psychology, 16 (3), 261–66. (3), 274–97.
Desai, Kalpesh K. and Kevin L. Keller (2002), “The Effects of Weber, Renee and Jennifer Crocker (1983), “Cognitive Processes
Ingredient Branding Strategies on Host Brand Extendability,” in the Revision of Stereotypic Beliefs,” Journal of Personality
Journal of Marketing, 66 (1), 73–93. and Social Psychology, 45 (5), 961–77.
Gurhan-Canli, Zeynep and Durairaj Maheshwaran (1998), “The Effects Zhang, Shi and Sanjay Sood (2002), “‘Deep’ and ‘Surface’ Cues:
of Extensions on Brand Name Dilution and Enhancement,” Jour- Brand Extension Evaluations by Children and Adults,” Jour-
nal of Marketing Research, 35 (November), 464–73. nal of Consumer Research, 29 (June), 129–41.