Académique Documents
Professionnel Documents
Culture Documents
West
hile a decade and a half of work building on Hoffman and Novak's (1996) analysis of computermediated environments has informed management
of online media, much of this work suggests that consumers
interact with brands online in ways similar to what they do
offline. That is, consumers join online brand communities
for many of the same reasons they join offline brand communities (e.g., Algesheimer, Dholakia, and Herrmann 2005;
Muniz and O'Guinn 2001; Schau, Muiz, and Amold 2009;
Thompson and Sinha 2008).
However, social media practitioners now seek best practices for contexts in which brick-and-mortar research is
largely inapplicable. Specifically, social media can make
the identity of a brand's supporters transparent to prospective consumers in ways that have no offline analog. Before
the advent of social networking, consumers could only
guess at the identities of other brand supporters on the basis
of advertising or the identity of spokespeople. In contrast,
in the social media world, consumers viewing a brand page
are likely to see pictorial information about other people
Rebecca Walker Naylor is Assistant Professor of Marketing (e-mail:
naylor_53@fisher.osu.edu), and Patricia M. West is Associate Professor
of Marketing (e-mail: west_284fisher.osu.edu). Fisher College of Business, The Ohio State University. Cait Poynor Lamberton is Assistant Professor of Marketing, Katz Graduate School of Business, University of
Pittsburgh (e-mail: cpoynorkatz.pitt.edu). Order of authorship is arbitrary; all authors contributed equally to this article. The authors thank Darren Dahl, Jeff Inman, Greg Allenby, and Andrew Hayes for their comments
on various aspects of this research.
105
Journal of Marketing
Voiume 76 (November 2012), 105-120
choosing not to reveal the identity of a brand's online supporters, a case we call "ambiguous MVP." If a brand displays the identity of its supporters, in social media settings,
this information is typically conveyed through profile photographs that display brand supporters' demographic characteristics.' Relative to a target consumer, displayed MVP
may be demographically similar or demographically dissimilar, or it may present a heterogeneous mix of similar
and dissimilar consumers. We compare the effects of maintaining ambiguous MVP with that created by each of these
types of identified MVP. In doing so, we show when it is
more beneficial to reveal the identity of current brand supporters to prospective customers and when to retain ambiguity about the brand's support base.
From a practical perspective, this research contributes
to the limited academic research investigating how firms
can best configure their social networks to meet strategic
objectives. For example. Tucker and Zhang (2010) demonstrate that displaying the number of sellers and buyers in
online exchanges can change business-to-business listing
and buying behavior. However, such findings provide limited managerial guidance because they do not compare the
effects of displaying the number of members on a particular
site with the range of other actions a manager may consider
when deciding how or whether to display online supporters.
From a theoretical perspective, our work challenges social
influence theory (SIT; Latane 1981), which suggests that
virtual exposure to unknown others should exert little social
influence. Furthermore, we provide novel insights into
social influence effects created by heterogeneous groups
and ambiguous others, for which the traditional reference
group literature (e.g., Bearden and Etzel 1982; Bearden,
Netemeyer, and Teel 1989; Berger and Heath 2007;
Bumkrant and Cousineau 1975; Childers and Rao 1992;
Escalas and Bettman 2003) is largely silent. In addition, we
show the importance of joint versus separate evaluation
mode (Hsee et al. 1999; Hsee and Leclerc 1998) as a moderator of the influence of ambiguous MVP. Finally, our
findings yield a road map for brand managers to use when
deciding whether to reveal the identities of their online supporters or to retain ambiguity according to (1) the demographic composition of existing supporters relative to targeted new supporters and (2) whether the brand is likely to
be evaluated singly or in combination with competing
brands.
inferred commonality; writing style, expression of priorities, or shared interest could prompt reasonable consumers
to generalize to other facets of preference. In contrast, the
present study considers the effect of the pictorial MVP of
consumers. In this case, consumers have affiliated with the
brand but have done so without any persuasive intent.
Moreover, they have not provided written product information or recommendations to try the brand that would ground
inferences of commonality.
Despite these differences, we expect that similar MVP
will generate high levels of inferred commonality with a
brand's user base. We base this expectation in management
research suggesting that demographic similarity (which we
refer to simply as "similarity" and can be observed from
photographs) leads to inferences of deeper-level commonality. That is, even in the absence of any other information,
demographically similar individuals are presumed to share
personality traits, values, and attitudes (Cunningham 2007).
This inferred commonality prompts the individual to raise
his or her evaluation of the brand.
In contrast, previous literature has suggested that consumers exposed to dissimilar MVP will infer little commonality with the brand's users and will express lower evaluations for the brand than for a brand with similar MVP.
Importantly, the reference group literature suggests that
even if the dissimilar brand supporters are not explicitly
dissociative (i.e., members of groups with whom consumers
do not want to be associated) (White and Dahl 2006,2007),
consumers may avoid similar purchase patterns simply due
to demographic dissimilarity (Berger and Heath 2008).
Work on non-target market effects also suggests that seeing
dissimilar individuals can lead consumers to infer low levels of conunonality (Aaker et al. 2000). Thus, if MVP indicates that the brand is liked by people whom target consumers perceive as dissimilar, target consumers should infer
less commonality between themselves and the brand's user
base and adjust their liking for the brand downward compared with when MVP is similar.
Is Ignorance Bliss? Ambiguous MVP
Given that displaying MVP dissimilar to a target consumer
may lower brand evaluations in comparison with similar
MVP, perhaps displaying ambiguous MVP is the firm's
safest decision. Ambiguous MVP involves the display of
others about whom no or very limited identifying demographic information is provided. Thus, ambiguity may be
manifest by not showing any pictures of brand supporters,
showing only supporters who have not provided a picture,
or showing photos of brand supporters whose identity has
been obscured.
Prior research offers little guidance regarding the use of
ambiguous MVP. Some literature suggests that when people
encounter unidentified others, they infer little commonality
with them. Sassenberg and Postmes (2002), for example,
show that when people know nothing about other group
members, they report low levels of liking and low perceptions of group cohesiveness. These authors ground their
findings in social categorization theory (Turner et al. 1987),
which argues that individuals who cannot be placed in a
person's in-group will be subject to any stereotypes common to out-groups. However, recent research on ambiguity
adopts an information-processing perspective and comes to
a different conclusion. This work suggests that in the
absence of externally provided information about others,
consumers anchor on the self to infer that ambiguous others
are like them. Because of these inferences, Naylor, Lamberton, and Norton (2011) demonstrate that an ambiguous
online reviewer is more persuasive than a dissimilar
reviewer and equally as persuasive as a similar reviewer.
How does this research translate to the present context?
Note that MVP does not involve extended formation of inand out-groups or interaction among members. Rather, it
involves only incidental, passive exposure to other consumers. Given the nature of MVP exposure, we propose
that the information-processing explanation, rather than the
social categorization perspective, is likely to hold. In other
words, when MVP is ambiguous, consumers will project
their own characteristics onto the brand's user base (thus
inferring commonality), emerging with a level of affinity
like that generated by similar MVP but greater than that created by dissimilar MVP.
Heterogeneous MVP
Although the difference between similar and dissimilar
MVP can be predicted according to prior literature, existing
theory fails to explain responses to heterogeneous MVP.
Understanding reactions to heterogeneous groups is important, because it is possible that a brand will not present supporters that are uniformly similar or dissimilar to the target,
either because doing so is out of their control or because
their objectives include extension into previously uru-epresented market segments. Diverse groups do not form a
cohesive "reference group" in the traditional sense, and thus
the reference group literature has little to say on this point.
Some prior research has suggested that diverse groups may
be interpreted in the same manner as a group perceived to
be uniformly dissimilar (i.e., that the group's preferences do
not match the target's). For example, Jehn, Northcraft, and
Neale (1999) argue that diversity in a workgroup cues individuals to expect opinions and behaviors that diverge from
their own. Similarly, Chatman and Flynn (2001) show that
demographic heterogeneity within a workgroup initially
leads to low levels of cooperation. However, some
researchers advise broad inclusion of a wide range of consumers as members of social networking sites (Dholakia
and Vianello 2009), arguing that heterogeneity could indicate a brand's wide range of features or suggest broad
appeal.
Consistent with these recommendations, we predict that
the MVP of a heterogeneous mix of supporters can be a
strength rather than a weakness for firms, albeit for different reasons than those Dholakia and Vianello (2009) propose. We base our prediction in the idea that individuals
tend to be particularly sensitive to incidental similarities
between themselves and others, showing more positive attitudes toward a product in the presence of even superficial
similarities (e.g., Jiang et al. 2010). Furthermore, work on
the self-referencing effect shows the positive effect of self-
relatedness for information processing (e.g., Perkins, Forehand, and Greenwald 2005), such that individuals show
enhanced attention and cognitive fluency for information
perceived as self-congruent. If such effects hold in a social
media context, individuals will be more influenced by the
presence of even a small number of individuals in an MVP
array who are similar to themselves (i.e., who are directly
self-relevant) than a small number who are dissimilar (and
thus are less self-relevant). Therefore, we anticipate that a
consumer viewing a social media site with heterogeneous
MVP will be particularly sensitive to the presence of the
similar individual(s) in the array. In tum, brand evaluations
will be equivalent to those formed when consumers are
exposed to similar MVP. Notably, given that we hypothesize that ambiguous MVP will create brand evaluations like
those created by similar MVP, heterogeneous and ambiguous MVP should also produce equivalent brand evaluations.
A question that remains, however, is how much similarity must be present in a heterogeneous MVP array for it to
generate inferences and evaluations like homogenous similar MVP. Note that in Asch's classic work on conformity
(Asch 1955, 1956), social influence effects can be generated by even a small number of individuals in a larger
group. That is, homogeneity among confederates was not
necessary to prompt study participants to alter their judgments of stimuli. Thus, while there is no theory to directly
guide predictions about heterogeneity in the social media
context, we propose that even a small proportion of similar
individuals in a heterogeneous MVP array may produce
evaluations like those produced by homogeneous similar
MVP. To test this, we empirically manipulate number of
similar individuals in an MVP array to range from zero (dissimilar MVP) to 100% (similar MVP).
Thus, we suggest that MVP will influence brand evaluations as follows:
H: Ambiguous MVP produces (a) equivalent brand evaluations to homogeneous similar MVP, (b) equivalent brand
evaluations to heterogeneous MVP, and (c) significantly
more positive brand evaluations than homogeneous dissimilar MVP.
H2: The relationship proposed between MVP composition and
brand evaluations in Hi^ is mediated by inferences of
commonality with the brand's user base.2
We flrst test these hypotheses across three studies
employing different operationalizations of ambiguous MVP
and similarity. Study la tests all parts of H| using age to
manipulate similarity. Study lb tests the parts of H] pertaining to ambiguity, similarity, and dissimilarity using gender
to manipulate similarity. Then, given that Study 1 leaves
unanswered questions about heterogeneity. Study 2 focuses
primarily on heterogeneity, providing a direct test of Hib.
Studies lb and 2 both include tests of H2 (the mediation
2Note that because H|a and H|b predict equivalence, a mediation test would not be able to explain variance in the dependent
measure for these hypotheses. Thus, H2 predicts that ambiguous
MVP leads to a higher level of inferred commonality than does
dissimilar MVP, which explains the difference in brand evaluations between these types of MVP predicted in Hj^.
Study 1a
Study la compares participants' liking for an unfamiliar
brand when they observe different types of MVP. We use
age to manipulate similarity given previous work by practitioners and academics highlighting the influence of age
similarity on product preferences. For example, the
Yankelovich report on generational marketing argues that
determinants of product value are strongly influenced by
age cohort, shared experiences, media icons, and life stage
(Smith and Clurman 2009). Previous academic studies have
argued that other demographic characteristics drive attraction between consumers, but note that age is likely to be
correlated with many of these characteristics. For example,
Byme (1971) finds that shared job classification and marital
status make television media attractive to consumers. Similarly, Shachar and Emerson (2000) flnd that individuals
who have families prefer watching shows about families.
These characteristics are likely to be shared within at least
broad age ranges, such that college students will differ from
people 30-40 years of age, who will again differ from
people older than 65 years. Thus, perceived age of the individuals in an MVP array may act as a proxy for numerous
other demographic characteristics that have been shown to
influence similarity-based attraction.
Stimuii and Procedure
A total of 128 undergraduate students participating in this
study in exchange for extra credit were told that they would
be viewing an excerpt from the Facebook fan page created
by Roots, a Canadian clothing company. Participants read
the following information:
In this section of today's study, we'd like to you to look at
excerpts from an actual Facebook page for a real brand.
This is the type of page where you can "become a fan" of
a company or brand .3 You will see excerpts from a page
maintained by Roots, a real Canadian company interested
in expanding to the United States. Please look at the information featured on their Facebook page and respond to
the questions as honestly as possible.
Participants then viewed an excerpt from a simulated
Roots Facebook page (see the Web Appendix at www.
marketingpower.com/jm_webappendix) and answered questions about Roots clothing. As discussed previously, we
operationalized similarity using perceived age, holding gender constant. Participants indicated their gender before the
study began so that all participants viewed fans matched to
their gender. All participants were told that there were the
same number of total fans regardless of MVP condition.
Depending on condition, participants saw one of the following: (1) total number of fans and pictures of six fans that
3At the time we began this research, Facebook called brand supporters "fans" and brand pages "fan pages." The term "fan" has
since been replaced by the "like" button; consumers who were
fans of a brand are now those that like the brand.
were the same age and gender as the participant^ (homogeneous similar MVP condition), (2) total number of fans and
pictures of six fans that were the same gender but a different
age than the participant (homogeneous dissimilar MVP condition), (3) total number of fans and three pictures of fans
that were the same gender and age and three pictures of fans
that were the same gender but a different age (heterogeneous
MVP condition) as the participant, or (4) no fan pictures,
only the total number of fans (ambiguous MVP condition).
No specific direction was given to attend to the fans, and
participants viewed the page as long as they liked. Other
information on the page was held constant across conditions.
We captured brand liking by asking, "Based on the
information you just saw from their Facebook page, how
much do you like Roots brand clothing?" on a nine-point
scale ranging from "do not like at all" to "like very much."
As manipulation checks, we asked participants in the similar, dissimilar, and heterogeneous MVP conditions to indicate how much they agreed that "Roots 'fans' are the same
age I am" and "Roots 'fans' are the same gender I am."
Finally, we asked all participants whether they had heard of
Roots before the study and, if so, how familiar they were
with the brand on a nine-point scale (1 = "not at all familiar," and 9 = "very familiar").
erogeneous MVP [HiJ, and [3] ambiguous MVP vs. dissimilar MVP [Hic]). This analysis thus indicates the effect
of the managerial decision to reveal consumers' similar or
different demographic information to one another or to
obscure it, maintaining ambiguity. Note that these contrast
codes partition the multivariate analysis of variance sums of
squares into interpretable subsets, obviating the need for
special alpha levels (Rosenthal, Rosnow, and Rubin 2000).
Consistent with our hypotheses, only the contrast code
comparing the ambiguous MVP condition with the dissimilar MVP condition was significant. There was no significant
difference in liking between the similar (M = 4.74) and
ambiguous (M = 4.48) MVP conditions (F(l, 107) = .86,/? =
.36) or between the heterogeneous (M = 4.89) and ambiguous MVP conditions (F(l, 107) =l.91,p= .17), in support
of Hja and H|b, respectively. We also note that there were
no differences in brand liking in the similar, heterogeneous,
and ambiguous conditions when these conditions were considered together in a separate analysis (F(2, 82) = .38,/? =
.68). As Hic predicts, the only one of the three contrast
codes that was significant was the one comparing ambiguous MVP with dissimilar MVP: Participants liked Roots
significantly less in the dissimilar (M - 3.81) than the
ambiguous (F(l, 107) = 5.04,p< .05) MVP condition.5
Results
Discussion
Sample and manipulation check. We first examined participants' familiarity with Roots. Of the 128 participants, 27
had heard of Roots before the study. Of these participants,
15 indicated a familiarity score of five or above on the ninepoint familiarity scale, and therefore we removed them
from the data set. Of the remaining participants, two indicated that Roots fans were not the same gender they were
(presumably because they did not follow instructions and
indicated a different gender than their own before the beginning of the study); they were also removed from the data
set. This left a final usable sample of 111 participants (59
men and 52 women) with an average age of 21 years. The
manipulation check revealed that participants in the similar
condition rated the fans as more similar to themselves in
age (M = 7.81) than did participants in the heterogeneous
(M - 5.11) and dissimilar conditions (M = 1.12; F(2, 81) =
147.76,/? < .0001). We also note that there were no differences in amount of time spent viewing the page across conditions (M = 26.7 seconds; F(3,107) = 1.13,/? = .34).
The result of Study la suggest that fans on a social networking site do not need to directly interact with a target
consumer or post comments about a brand to influence the
brand evaluations of a consumer new to the brand. Specifically, MVP evokes equivalent levels of liking when it is
composed of a homogeneous group of similar individuals,
when it is composed of a heterogeneous group of dissimilar
and similar individuals, and when brand supporters are left
demographically ambiguous. In contrast, a homogenous
group of dissimilar others produced significantly less brand
liking. Thus, H] is supported in this context.
Liking for Roots clothing. We next examined participants' liking for Roots clothing. Because MVP composition
was a four-level variable, we used three orthogonal contrast
codes to compare the ambiguous condition with the other
three conditions (i.e., these codes compared [1] ambiguous
MVP vs. similar MVP [H|a], [2] ambiguous MVP vs. het^Because all participants in the subject pool at the university
where the studies were conducted were in their late teens to early
20s, the fans used in the similar condition were also in this age
range. All fans in the dissimilar age condition were older, ranging
from their 30s to their 60s. All pictures used were actual Facebook
profile pictures selected from the Facebook pages of individuals
whose profile picture was public. We note that these pictures
manipulate perceived age, not objective age.
Study 1b
It is possible that part of the reason that ambiguity was
treated like similarity in Study la was because the numeric
representation of fans in the ambiguous condition made it
difficult for consumers to consider the possibility that these
fans are different from themselves. It is also possible that
the results we obtained could be unique to using age to
manipulate similarity. Therefore, in Study lb, we tested
whether our results hold using a different type of ambiguity
(i.e., generic Facebook profile picture silhouettes) and when
manipulating (dis)similarity using participants' gender
5We also analyzed these data using an alternate set of contrast
codes comparing (1) the similar, ambiguous, and heterogeneous
MVP conditions with the dissimilar MVP condition; (2) the similar and ambiguous MVP conditions with the heterogeneous MVP
condition; and (3) the ambiguous and similar MVP conditions
with each other. Consistent with our hypotheses, only the first of
these alternate contrast codes had a signiflcant effect on brand liking: Roots was liked significantly less in the dissimilar than in the
other conditions (F(l, 107) = 5.04, p < .05); the other two contrast
codes had a nonsignificant effect on liking (both ps >.4O).
Results
Sample and manipulation check. Of the 116 undergraduate students who participated in this study, 25 had heard of
Roots before the study. Of these participants, we removed 4
from the data set because they reported being highly familiar with the brand. This left a final usable sample of 112
participants (53 men, 59 women) with an average age of 21
years. All participants in the similar MVP condition indicated that the fans of the day were the same gender they
were, and all participants in the dissimilar MVP condition
indicated that the fans of the day were the opposite gender.
We also note that there were no differences in amount of
time spent viewing the page across conditions (M = 35.6
seconds; F(2,107) =l.Ol,p= .37).
Brand liking and willingness to interact with the brand
through social media. Given that the independent variable in
this study had three levels, we used two orthogonal contrast
codes (no special alpha levels required) to analyze the data
comparing (1) the ambiguous and similar MVP conditions
with each other (H|a) and (2) the ambiguous with the dissimilar MVP condition (H](,). We analyzed brand liking and
willingness to interact with the brand through social media
separately. Again, consistent with Hia, participants in the
ambiguous (M = 4.16) and similar (M = 3.91) MVP conditions expressed equivalent liking for Roots (F(l, 109) = .14,
p = .71). Furthermore, consistent with Hi^, participants in the
dissimilar MVP condition (M = 3.46) liked Roots marginally
less than participants in the ambiguous MVP condition (F(l,
109) = 3.15,p = .06). The results for willingness to interact
with Roots through social media are similar. Participants
reported that they were equally likely to interact with the
brand in the ambiguous (M - 2.63) and similar (M - 2.57)
MVP conditions (F(l, 108) = .90,/? = .35) and more likely to
join the Roots community in the ambiguous than the dissimilar (M = 1.97) MVP condition (F(l, 108) = 4.50,p < .05).?
Mediation. Using the same contrast codes, we examined
whether consumers' inferences that they had "a lot in common with the typical Roots shopper" followed the same pattem, as H2 predicted. As we expected, participants reported
that they had the same amount of commonality with the
typical Roots shopper in the ambiguous and similar MVP
conditions (M^^biguous = 4.29, M,-,^,^^ = 3.85; F(l, 109) =
'We also conducted an alternate analysis of these data using two
contrast codes that compare (1) the similar and ambiguous MVP
conditions with the dissimilar condition and (2) the similar and
ambiguous MVP conditions with each other. The results revealed
that the brand was liked marginally more in the similar and
ambiguous MVP conditions than in the dissimilar MVP condition
(F(l, 109) = 3.75, p = .06) and was liked equally well in the similar and ambiguous MVP conditions (F(l, 109) = 2.37, p = .13).
Participants were also more likely to want to connect with the
brand through social media in the similar and ambiguous MVP
conditions than in the dissimilar MVP condition (F(l, 109) = 4.50,
p < .05) and were equally likely to want to connect with the brand in
the similar and ambiguous MVP conditions (F(l, 109) = 1.40,p =
.24).
Study 2
Although the numbers-only presentation used in Study la
and the fans-of-the-day format used in Study lb are both
common ways to display MVP, firms also increasingly
allow consumers to upload pictures of themselves using or
wearing a product to social media sites. If such pictures are
displayed in ways that do not provide complete demographic information, they would also present the consumer
with a type of ambiguous MVP. Therefore, Study 2
explores whether photos that do not reveal all of a supporter's demographic characteristics create identical effects
to those created by revealing only the total number of fans
or showing profile pictures that reveal no demographic
information. Study 2 also explores the effect of heterogeneous MVP in greater depth, testing the level of heterogeneity required to create effects equivalent to those seen
with similar MVP. A secondary goal of this study was to
test whether effects observed in prior studies extend to other
downstream consequences of interest to managers beyond
brand liking and interacting with a brand through social
media.
Stimuli and Procedure
Study 2 asked participants to react to an online clothing
retailer called asos:
In this survey, we are interested in your opinions about a
real brand's social networking presence. This brand, asos,
is an online clothing retailer that sells both men's and
women's clothing mostly in the U.K. As part of its social
networking strategy, in addition to a Facebook page and a
Twitter account, asos also hosts "asos marketplace" on its
company-owned website. Visitors to the website are
invited to join asos marketplace and to post photos of
themselves wearing asos brand clothing.
233) = 23.71,/7<.0001).
Purchase intentions. Because the three purchase intention measures were highly correlated (a = .90), we averaged them to form an overall purchase intention index. Our
analysis uses seven contrast codes that compare purchase
intentions in the ambiguous MVP condition with every
other condition (with no need for special alpha levels
because the codes are orthogonal). Consistent with H a and
H]),, none of the contrasts comparing ambiguous MVP with
the homogeneous similar MVP condition or any of the heterogeneous conditions were significant (all ps > .24). The
only significant contrast (of the seven contrast codes comparing the ambiguous MVP condition with every other condition) was the contrast code comparing the ambiguous
MVP condition directly with the homogeneous dissimilar
MVP condition. Consistent with H^^, participants in the dissimilar MVP condition were less likely to buy asos clothing
(M - 3.60) than participants in the ambiguous MVP condition (M = 4.60; F(l, 249) = 6.84,p < .Ol).io
To learn more about the effect of different levels of
heterogeneity, we conducted follow-up analyses comparing
the heterogeneous MVP arrays that contained the fewest
number of dissimilar individuals with the homogeneous dissimilar MVP array. We found that one similar individual in
the MVP array was not enough to create purchase intentions
significantly different from those generated by a homogeneous dissimilar MVP array (F(l, 65) = 1.29, p - .26).
However, when two of the six displayed individuals were
'"In a separate analysis (excluding the dissimilar MVP condition), we also tested for differences across the homogeneous similar and all heterogeneous MVP conditions. This omnibus analysis
revealed that there were no differences in purchase likelihood
across these seven conditions (F(6, 228) = .ll,p = .60).
TABLE 1
Study 2: Means by MVP Composition
Condition
1
2
3
4
5
6
7
8
MVP Composition
6 ambiguous users
6 simiiar, 0 dissimilar users
5 simiiar, 1 dissimilar user
4 simiiar, 2 dissimilar users
3 simiiar, 3 similar users
2 similar, 4 dissimilar users
1 similar, 5 dissimilar users
0 similar, 6 dissimilar users
4.61
4.38
4.48
4.25
4.21
4.23
3.96
3.54
3.63
3.94
3.75
3.49
3.35
3.45
3.21
3.00
Study 3
Study 3 tests our hypothesis that the effect of ambiguity
depends on the evaluation context in which a brand is
viewed. In addition. Study 3 shows that the results from
Studies 1 and 2 replicate in a different product category and
in an additional social networking context.
Stimuli and Procedure
A total of 312 undergraduate students (178 men, 134
women) with a mean age of 21 years participated in this
study for course credit. All participants read the following
introduction:
Now we'd like to you to look at some social networking
websites developed by restaurants with locations nationwide. To protect confldentiality, the names of the restaurants have been changed to Restaurants X, Y, and Z [in the
joint evaluation condition; only Restaurant X was men-
Joint evaluation. An omnibus repeated measures analysis of all three conditions revealed that there was not a significant between-subjects effect of order (in which the
restaurants were viewed) on brand liking, so we dropped
order from further analysis (F(l, 163) = .65,p = .42).i3 We
then used within-subject contrast codes analogous to the
ones used in the separate evaluation analysis to analyze participants' liking for the three restaurants. Consistent with H3
and diverging from Hig, participants liked the bar area in the
restaurant with similar MVP (M - 5.97) significantly more
than the bar area in the restaurant with ambiguous MVP (M =
5.34; F(l, 164) = 13.22,/? < .001), showing a cost of ambiguity relative to known similarity. However, even under
joint evaluation, ambiguity is stiU preferable to dissimilarity:
Participants anticipated liking the bar area in the restaurant
with ambiguous MVP more than the bar area in the restaurant with dissimilar MVP (M = 4.39; F(l, 164) = 30.01,p <
.0001).i4
Discussion
Study 3 examines participants' response to similar versus
ambiguous MVP in joint versus separate evaluation contexts. In separate evaluation, ambiguous MVP leads to an
almost identical response to that generated by similar MVP,
as in Studies 1 and 2. However, in joint evaluation, ambiguous MVP leads to a significantly less positive response than
does similar MVP. Still, ambiguous MVP generates liking
greater than that evoked by dissimilar MVP.
These findings shed light on the decision managers
must make about whether to reveal the identity of their
brand's supporters: The decision must be determined not
just by whether the brand supporters shown are likely to be
perceived as similar or dissimilar to a target consumer but
also by whether the consumer is likely to encounter the supporters in a joint or separate evaluation context, an issue we
retum to in the "General Discussion" section.
General Discussion
At the end of 2011, iMedia Connection published an article
titled "Why Facebook Fans Are Useless" (Lake 2011). In the
article, the author notes that "on their own, Facebook 'likes'
don't add any value." Yet research has shown that social
media can translate into increases in sales (Stephen and
i3We also analyzed the joint evaluation data using only each
participant's rating of the first restaurant they saw. Because participants viewed the restaurants sequentially, we expected these
results to be consistent with the separate evaluation results. As we
expected, participants liked the bar area equally well in the similar
(M = 5.57) and ambiguous MVP conditions (M = 5.88; F(l, 162) =
.90,p = .34) but more in the ambiguous MVP condition than in the
dissimilar (M = 4.66) MVP condition (F(l, 162)= 10.69, p< .01).
'''The results using the same alternate contrast codes used in the
separate evaluation analysis reveal that participants anticipated
liking the bar area in the restaurants with similar and ambiguous
MVP more so than the bar area in the restaurant with dissimilar
MVP (F(l, 164) = 63.47, p < .0001). However, in contrast to the
separate evaluation results, they liked the bar area in the restaurant
with similar MVP significantly more so than they did the bar area
in the restaurant with ambiguous MVP (F(l, 164) = 13.22, p <
.001).
Theoretical Contributions
This work provides several novel theoretical insights. First,
the concept of MVP offers a new framework for understanding social influence. Although spatial proximity is
absent, exposure is only passive, typically just a handful of
individuals are shown, and no future relationship is likely to
exist among the consumers, we show that MVP still has
substantial effects on consumers' brand evaluations and
purchase intentions. As such, the concept of MVP highlights the ways that online social influence may have an
effect despite its difference from the offline presence of others and despite its failure to conform to the parameters of
SIT (Latane 1981).
Importantly, in contrast to traditional advertising or
spokesperson contexts, we also note that MVP created by
social media exposure is provided by individuals who voluntarily affiliate with a brand, making it less likely that
their action will be discounted by consumers due to reactance against marketer-driven recommendations (Fitzsimons and Lehmann 2004) or persuasion knowledge (Friestad and Wright 1994). Further research could explore
whether the effects of MVP hold if consumers do not trust
that the brand supporters presented are truly other consumers (as opposed to, e.g., employees of the brand "posing" as supporters).
Furthermore, our investigation into consumers' responses
to ambiguous others may prompt deeper explorations of
interpretations of interpersonal ambiguity. In the present
research, we show the equivalence of three types of ambiguity: that created when (1) MVP is represented only
numerically, (2) MVP is displayed using silhouette pictures
that suggest real individuals but obscure all their demographic characteristics, and (3) MVP is displayed using pictures of real supporters that obscure some of their demographic characteristics. Although these operationalizations
of ambiguity appear to have similar effects, further research
may find additional nuances in the concept of ambiguity
and may actually find that different types of ambiguity have
variant impacts on consumers.
Finally, by directly comparing consumer response to
ambiguous others in separate and joint evaluation, we
explain Nay lor, Lamberton, and Norton's (2011) findings.
We suggest that the difference in ambiguity's effects across
evaluation modes stems from the finding that the intemally
derived egocentric anchor is the determinant of similarity
when ambiguity is encountered in a separate evaluation
mode but that the importance of this intemal anchor is
diminished in joint evaluation. Therefore, in joint evaluation, a brand with ambiguous MVP will be less attractive
when compared with a brand with similar MVP. Thus, in
Study 3, consumers evaluated a brand with identified similar
supporters more positively than a brand whose supporters
were not identified. This finding is consistent with work by
Forman, Ghose, and Wiesenfeld (2008), who find that consumers rate product reviews containing identity-descriptive
FIGURE 1
Framework for Decisions About Revealing the Identity of a Brand's Fan Base and Selection of a Social
Media Platform
Fan base is
homogeneous and
similar to target
audience
Fan base is
heterogeneous but
includes fans similar
to target audience
Fan base is
homogeneous and
different from
target audience
Fan base is
heterogeneous and
includes no fans
similar to target
audience
chase intentions. This is true in both more and less competitive product categories. If the brand is being evaluated in
isolation, revealing a fan base similar to a target consumer
(or one that contains at least some similar supporters) is
likely to prompt relatively positive evaluations. In joint
evaluation contexts, revealing the brand's similar supporter
base may give the brand an edge over brands that provide
no supporter information on their social media sites.
When a brand's supporters are better left ambiguous. If
a brand's current supporters are likely to be perceived as
dissimilar by new consumers, our results suggest that
revealing the identity of a brand's existing supporters will
undermine brand liking in both separate and joint evaluations. Thus, revealing supporters that are dissimilar to the
target consumer is an inferior alternative to leaving supporters ambiguous. One situation in which a brand's current
supporters are likely to be dissimilar to targeted consumers
is when a brand extends into new demographic segments. In
these cases, it will be important to recruit new supporters
first (who are perceived to be similar to the new target market). Managers should then ensure that these new supporters are displayed either as a homogenous group when the
new target market visits the brand's social networking site
or mixed in with the old supporters to create heterogeneous
MVP in which at least some supporters are shown that the
new target market perceives as similar.
magnified. An exploration of consumers' goals when interacting with a brand through social media, the effect of
MVP, and the interaction between the two on the influence
of downstream variables (e.g., consumer desire to stay affiliated with a brand over time) is an important avenue for further research.
REFERENCES
Aaker, Jennifer, Anne M. Brumbaugh, and Sonya Grier (2000),
"Non-Target Markets and Viewer Distinetiveness: The Impact
of Target Marketing on Advertising Attitudes," Journal of Consumer Psychology, 9 (3), 127-40.
Algesheimer, Ren, Utpal M. Dholakia, and Andreas Hermann
(2005), "The Social Influence of Brand Community: Evidence
from European Car Clubs," Journal of Marketing, 69 (July),
19-34.
Argo, Jennifer J., Darren W. Dahl, and Rajesh V. Manchanda
(2005), "The Influence of a Mere Social Presence in a Retail
Context," Journal of Consumer Research, 32 (September),
207-212.
Asch, Solomon E. (1955), "Opinions and Social Pressure," Scientific American, 193 (5), 31-35.
(1956), "Studies of Independence and Conformity: A
Minority of One Against a Unanimous Majority," Psychological Monographs, 70 (416).
Baird, Carolyn Heller and Gautam Parasnis (2011), "From Social
Media to Social Customer Relationship Management," Strategy & Leadership, 39 (5), 30-37.
Bearden, William O. and Michael J. Etzel (1982), "Reference
Group Influence on Product and Brand Purchase Decisions,"
Journal of Consumer Research, 9 (September), 183-94.
, Richard G. Netemeyer, and Jesse E. Teel (1989), "Measurement of Consumer Susceptibility to Interpersonal Influence," Jourrml of Consumer Research, 15 (March), 473-81.
Berger, Jonah and Chip Heath (2007), "Where Consumers Diverge
from Others: Identity Signaling and Product Domains," Journal of Consumer Research, 34 (August), 121-34.
and
(2008), "Who Drives Divergence? Identity Signaling, Outgroup Dissimilarity, and the Abandonment of Cultural Tastes," Journal of Personality and Social Psychology, 95
(September), 593-607.
Bumkrant, Robert E. and Alain Cousineau (1975), "Informational
and Normative Social Influence in Buyer Behavior," Journal of
Consumer Research, 2 (December), 206-215.
Byme, Donn Erwin (1971), The Attraction Paradigm. New York:
Acadmie Press.
Chatman, Jennifer A. and Francis Flynn (2001), "The Influence of
Demographic Heterogeneity on the Emergence and Consequences of Cooperative Norms in Work Terms," Academy of
Management Journal, 44 (5), 956-74.
Childers, Terry L. and Akshay R. Rao (1992), "The Influence of
Familial and Peer-Based Reference Groups on Consumer Decisions," Journal of Consumer Research, 19 (September),
198-211.
Cunningham, George B. (2007), "Perceptions as Reality: The
Influence of Actual and Perceived Demographic Similarity,"
Journal of Business and Psychology, 22 (1), 79-89.
Deshpand, Rohit and Douglas M. Stayman (1994), "A Tale of
Two Cities: Distinetiveness Theory and Advertising Effectiveness," Journal of Marketing Research, 31 (February), 57-64.
Dholakia, Utpaul M. and Silvia Vianello (2009), "The Fans Know
Best," Sloan Management Review/ Wall Street Journal Business Insights, (August 17), (accessed July 11, 2012), [available
at http://online.wsj.com/artiele/SB10001424052970204482304
574222062946162306 .html].
Latane, Bibb (1981), "The Psychology of Social Impact," American Psychologist, 36 (4), 343-56.
Lydon, John E., David W. Jamieson, and Mark P. Zanna (1988),
"Interpersonal Similarity and the Social and Intellectual
Dimensions of First Impressions," Social Cognition, 6 (4),
269-86.
McCracken, Grant (1988), Cuiture and Consumption: New
Approaches to the Symboiic Character of Consumer Goods
and Activities. Bloomington: Indiana University Press.
Morry, Marian M (2007), "Relationship Satisfaction as a Predictor
of Perceived Similarity Among Cross-Sex Friends: A Test of
the Attraction-Similarity Model," Journal of Social and Personal Relationships, 24 (1), 117-38.
Muiz, Albert M., Jr., and Thomas C. O'Guinn (2001), "Brand
Community," Journal of Consumer Research, 27 (March)
421-32.
Naylor, Rebecca Walker, Cait Poynor Lamberton, and David A.
Norton (2011), "Seeing Ourselves in Others: Reviewer Ambiguity, Egocentric Anchoring, and Persuasion," Journal of Marketing Research, 48 (June), 617-31.
Newman, Andrew Adam (2011), "Brands Now Direct Their Followers to Social Media," The New York Times, (August 4), B3.
Novak, Thomas P., Donna L. Hoffman, and Adam Duhachek
(2003), "The Influence of Goal-Directed and Experiential
Activities on Online Flow Experiences," Journal of Consumer
Psychology, 13 {112), 3-16.
Perkins, Andrew W., Mark R. Forehand, and Anthony G. Greenwald (2006), "Decomposing the Implicit Self-Concept: The
Relative Influence of Semantic Meaning and Valence on
Attribute Self-association." Social Cognition, 24 (4), 387-408.
Preacher, Kristopher J. and Andrew F. Hayes (2008), "Asymptotic
and Resampling Strategies for Assessing and Comparing Indirect Effects in Multiple Mediator Models," Behavior Research
Methods, 40 {3), 819-91.
Rosenthal, Robert, Ralph L. Rosnow, and Donald B. Rubin
(2000), Contrasts and Effect Sizes in Behavioral Research: A
Correiationai Approach. New York: Cambridge University
Press.
Sassenberg, Kai and Tom Postmes (2002), "Cognitive and Strategic Processes in Small Groups: Effects of Anonymity of the
Self and Anonymity of the Group on Social Influence," British
Journal of Social Psychology, 41 (3), 463-81.
197-206.
Copyright of Journal of Marketing is the property of American Marketing Association and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.