Content uploaded by Joan Meyers-Levy
Author content
All content in this area was uploaded by Joan Meyers-Levy on Jun 04, 2014
Content may be subject to copyright.
Journal of Marketing Research
Vol. XLVI (February 2009), 37–4537
© 2009, American Marketing Association
ISSN: 0022-2437 (print), 1547-7193 (electronic)
*Rui (Juliet) Zhu is Assistant Professor of Marketing, Sauder School of
Business, University of British Columbia (e-mail: juliet.zhu@sauder.ubc.
ca). Joan Meyers-Levy is Holden-Werlich Professor of Marketing, Carlson
School of Management, University of Minnesota (e-mail: jmeyers@umn.
edu). Financial support from the Social Sciences and Humanities Research
Council of Canada is gratefully acknowledged. Chris Janiszewski served
as associate editor for this article.
RUI (JULIET) ZHU and JOAN MEYERS-LEVY*
Can the surface material of a display table prompt context effects on
shoppers’ product evaluations? If so, how might the direction of such
effects be influenced by people’s use of different modes of cognition—
namely, holistic versus analytic cognition? The authors theorize and find
that people’s use of holistic cognition, as prompted by an interdependent
self-view, produces an assimilation effect. Product evaluations are
assimilated with associations with the table surface. However, people
who rely on analytic processing, as prompted by an independent self-
view, elicit a contrast effect in which evaluations are negatively related to
such associations.
Keywords
: context effects, self-view, display fixture
The Influence of Self-View on Context
Effects: How Display Fixtures Can Affect
Product Evaluations
In most stores, consumers view products on display fix-
tures that are presumed to be of little consequence. How-
ever, suppose that the consumer were shopping for a set of
trendy new coffee mugs and noticed some on a nearby table
or shelf. Might his or her evaluation of the trendiness of the
mugs be subject to context effects arising from the display
fixture’s surface material, such as the sheath of glass or
wood beneath the product? There is reason to believe so.
Several accounts have been offered to explain context
effects (e.g., the feature-matching model [Herr 1986], the
comparison relevance model [Stapel, Koomen, and Velthui-
jsen 1998], the inclusion–exclusion model [Schwarz and
Bless 1992]), but the prevailing view is that context effects
can occur at two stages of the evaluation process (Schwarz
and Bless 2006). In some instances, a concept that is acti-
vated by contextual data (e.g., perhaps the concept of
trendiness elicited by a glass display fixture) is used at the
encoding stage to help a person interpret and construct a
mental representation of the target (e.g., the mugs). In such
cases, this concept is typically included in the person’s rep-
resentation of the target, and an assimilation effect occurs
as target product evaluations are positively related to the
implications of the contextual data. At other times, how-
ever, the same concept may be evoked and used at a subse-
quent stage, when the person forms a judgment of the target
product. Here, the concept may serve as a comparison stan-
dard against which the target is judged. Because the con-
cept that embodies the standard is often fairly extreme, in
general, this comparison process produces a contrast effect,
and target product evaluations are negatively related to the
implications of the contextual data.
What determines how people use the contextually acti-
vated concept? Most research exploring this question has
focused on how characteristics of the contextual data can
influence the direction of the context effects (e.g., Herr
1986; Stapel, Koomen, and Velthuijsen 1998). For example,
research has found that when the concept activated by the
contextual data possesses moderate implications (e.g., a
concept, such as unruly) or is nondistinct (e.g., a general
trait, such as unkind), its lack of well-defined boundaries
causes it to be included in the representation of the target,
producing an assimilation effect during encoding. However,
when the concept is extreme (e.g., evil) or represents a dis-
tinct exemplar (e.g., Hitler), its clear-cut boundaries not
only deter such inclusion but often promote its use as a
comparison standard during the judgment stage, which
typically results in a contrast effect (Herr 1986; Stapel,
Koomen, and Velthuijsen 1998).
The preceding findings are important because they sug-
gest that factors that blur versus delineate the boundaries of
contextually activated concepts promote assimilation versus
contrast effects, respectively. Nevertheless, extant research
38 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2009
has not considered whether far more fundamental factors,
such as the nature of the cognition that people use, also
shape the direction of context effects. Moreover, although
most previous work has bolstered its internal validity by
employing sterile, carefully worded verbal representations
of the contextual data and targets, such procedures have
come at a cost. Their artificiality raises questions about the
applicability of the context effect findings in real marketing
settings, in which physically present and decidedly tangible
contextual and target goods typically dominate.
With these issues in mind, this research focuses on an
important theoretical and practical issue that has been
ignored in the context effects literature: whether and how
the mode of cognition that people use can affect the likeli-
hood and direction of context effects that may emerge when
a product is viewed on a display table with a surface com-
posed of, for example, glass, which seemingly implies
trendiness and manufacturing, versus wood, which con-
notes tradition and naturalness. Specifically, we examine
how holistic versus analytic cognition, as induced by a per-
son’s interdependent versus independent self-view mind-
set, can affect context effects. We explore this issue using a
setting and procedures that are both fairly realistic and
actionable (e.g., people’s mode of cognition is varied
through a mind-set task or depictions on nearby signage).
We begin by discussing these two modes of cognition
and the kinds of context effects they are likely to foster
(i.e., assimilation or contrast effects). Then, we apply such
theorizing to the product display table issue and test our
predictions in one lab and one field experiment.
THEORETICAL BACKGROUND
Self-construal theory holds that people possess multiple
views about how the self relates to others and the environ-
ment (Markus and Kitayama 1991). Furthermore, even if
one self-view is chronically accessible, people can be
induced to activate a different self-view temporarily (Singe-
lis 1994). People known as interdependents activate pre-
dominately interdependent self-knowledge such that the
self is viewed as fundamentally connected to and a continu-
ous part of both others and the context. In contrast, those
known as independents activate an independent self-view
and perceive the self as a unique and autonomous entity
with distinct boundaries that distinguish it from the social
and natural background (Markus and Kitayama 1991).
Accordingly, research has shown that when asked to
describe themselves, interdependents generate more
responses that refer to other people and/or fluidly mutate
across situations, whereas independents report more stable
personal traits (Ip and Bond 1995; Markus and Kitayama
1991).
Of particular relevance to our inquiry, however, is the
contention that these alternative ways of thinking about the
self affect people’s cognition more generally (Nisbett et al.
2001; Van Baaren et al. 2003). Interdependents’ emphasis
on the inseparability of objects (e.g., the self, others) and
their settings appears to broaden, prompting general holistic
processing (Nisbett 2003). Conversely, independents, who
view the self as an isolated entity that is distinct from the
social and natural background, generalize their context-
independent manner of thinking. Thus, they engage in ana-
lytic cognition, treating objects or items of any type as
well-bounded, self-contained units of data (Nisbett 2003).
It seems plausible that each of these two modes of cogni-
tion outlined by self-construal theory could influence the
context effects that may occur on people’s evaluations of a
target product when it is viewed on a display table made of
different surface materials. In support of this possibility,
extant research suggests that people often learn or develop
shared associations with objects (e.g., surface materials) as
a result of their experience with or common observations of
them (Peracchio and Meyers-Levy 2005; Scott 1994). Thus,
it would seem that most people are likely to associate glass
with the concepts of modernity and artificiality, given that
glass is a dominant material in modern buildings or decor,
and it is known to be commonly manufactured using mech-
anization. In contrast, wood is likely to be associated with
the concepts of tradition and naturalness because of its
time-honored use in construction (e.g., log cabins,
clapboard-sided colonials) and its natural origins. Thus, to
the extent that these concepts are associated with and acti-
vated by the glass or wood surface of a display table and
product viewers use a mode of cognition that either blurs or
accentuates the boundaries between these activated con-
cepts and the target product, alternative context effects may
arise.
Holistic cognition, as used by people with an inter-
dependent self-view, involves blurring or minimizing the
boundaries among separate pieces of data, so that such data
are perceived as continuous parts of a larger integrated unit.
This implies that interdependents, who are more prone to
employ holistic cognition, are likely to exhibit an assimila-
tion effect on their product evaluations. Accordingly, the
concepts elicited by the contextual table’s wood or glass
surface are likely to be assimilated with perceptions of the
product. In contrast, analytic cognition, as used by people
with an independent self-view, entails treating individual
pieces of data as separate and fairly autonomous elements.
Although analytic processors might be expected to use the
contextual data in a different way than holistic processors,
two viable approaches seem possible. First, if each piece of
data is regarded as a separate entity such that any and all
relationships between the pieces of data are minimized or
ignored, context effects might be entirely absent. That is,
people may encode the table surface and the target product
as completely independent elements and thus assess the
product in isolation of the table. This should result in no
context effects. Second, although analytic processors may
perceive all pieces of data as separate entities, they may be
inclined to attend to the distinctiveness of each self-
contained unit’s properties. This would require comparing
these properties with those of another or with a convenient
standard (Malaviya, Kisielius, and Sternthal 1996). If so,
information that is highly accessible might be used as a
plausible standard with which the target data are compared,
resulting in a contrast effect. If we apply this logic to the
focal issue, concepts activated by the table’s surface may be
used as a comparison standard against which the target
product is judged, prompting a contrast effect.
To shed light on which of these processes is likely to
occur, we examined self-construal theory and research on
independents’ use of analytic cognition. According to this
research, such people treat each element of data as a
The Influence of Self-View on Context Effects 39
1
A few social psychology studies have considered a topic seemingly
related to such context effects—namely, the influence of people’s self-
view on social comparison (Kühnen and Haberstroh 2004; Kühnen and
Hannover 2000). These studies have activated a person’s self-view and
then probed his or her perceived similarity to another known person.
Therein lies the disparity: Given the obvious semantic linkage between the
content of a person’s self-view (i.e., the representation of the self in rela-
tion to others) and the assessed object (i.e., the perceived similarity of the
self and another person), any effects observed could be attributed to the
nearly complete overlap between the semantic content of these two con-
structs (i.e., both are related to the perceived linkage between the self and
another person). Thus, these studies are silent about the broader issue that
interests us: whether the mode of cognition fostered by a person’s self-
view (i.e., not merely the activation of a person’s self-view per se) can
shape the direction of context effects.
bounded, self-contained unit of analysis, but they also are
likely to seek “to understand what an object’s distinctive
properties” are in relation to others (Nisbett 2003, p. 19,
emphasis in original). As Markus and Kitayama (1991, p.
226) note, among independents, a bounded target entity
may be “set contrastively both against other such wholes
and against a ... natural background.” Thus, “others, or
the ... situation in general, are important, but primarily as
standards of reflected appraisal” or comparison (Markus
and Kitayama 1991, p. 226). This implies that independents
who use analytic processing should be likely to exhibit a
contrast effect. Although they may be immune to contextual
influence at encoding, during the judgment phase, associa-
tions with the context (i.e., the glass or wood table surface)
are likely to be employed as a contrastive standard of com-
parison, producing a contrast effect.
1
Still, although the preceding propositions seem reason-
able, might outcomes be qualified at times by the nature of
the target product assessed? The predictions we developed
are likely to ensue if the target product is ambiguous or
neutral on the concepts evoked by the table surface (i.e.,
modernity and naturalness). However, what if the product
possesses fairly extreme associations pertaining to these
concepts? Here, the context effects may not be observed,
because perceptions of the target product may not be suffi-
ciently flexible for the contextual associations to exert any
effect. Thus, if both neutral and extreme products are
assessed, a three-way interaction should occur involving
people’s mode of cognition, table surface, and product type.
Formally, we predict the following:
H
1
: Because interdependents should rely on holistic processing,
an assimilation effect should emerge, such that they will
evaluate a neutral product as more modern and/or less natu-
ral when it is displayed on a glass table than when it is dis-
played on a wood table. However, because independents
employ analytic cognition, a contrast effect should emerge,
such that they will view the product as more modern and/or
less natural when it is displayed on a wood table than when
it is displayed on a glass table.
H
2
: These effects will be absent if the product itself is unam-
biguously extreme (i.e., either high or low) on the evalua-
tive dimensions in question (i.e., modernity and/or
naturalness).
Experiment 1 tests H
1
with a target product that is neu-
tral on the two focal dimensions elicited by a display table
(i.e., modernity and naturalness). Furthermore, it examines
not only participants’ evaluations but also their product
thoughts and choice-related input about issues that are
rather distally related to the target product. Experiment 2 is
a field study designed to extend this. It conceptually repli-
cates the findings in a real retail setting in which people’s
holistic versus analytic processing is induced through
decidedly actionable means. Experiment 2 also tests H
2
by
investigating target products that are neutral versus extreme
on the focal dimensions, and it assesses the influence of
two tabletop materials.
PRETESTS FOR EXPERIMENT 1
Experiment 1 varied the mode of cognition participants
would use and assessed their evaluations of a target product
that was neutral on the contextually activated dimensions
(i.e., modernity and naturalness) and viewed on a glass or
wood table. To accomplish this, two pretests were neces-
sary. The first pretest examined whether people’s associa-
tions with glass and wood would vary as we expected on
the proposed dimensions.
In line with extant theory about how associations with
items are learned (Peracchio and Meyers-Levy 2005; Scott
1994), we reasoned that glass would be associated with
both modernity (e.g., trendiness) and artificiality but that
wood would be associated with opposite concepts, namely,
tradition and naturalness. Thus, in the first pretest, 34 par-
ticipants visually examined one of two comparable tables
(e.g., similar size, shape) that featured either a glass or a
wood tabletop. Then, they evaluated the table on six items
(1 = “not at all,” and 7 = “extremely”). We averaged three
items—trendy, modern, and urban—to produce a trendiness
index (α = .77). We averaged the others—natural, organic,
and artificial (last item was reverse coded)—to form a natu-
ralness index (α = .80).
A one-way analysis of variance (ANOVA) (table surface:
wood versus glass) revealed a main effect on both the
trendiness (F(1, 32) = 31.34, p < .001) and the naturalness
(F(1, 32) = 48.39, p< .001) indexes. Participants perceived
the glass table as more trendy than the wood table (M
glass
=
4.67, M
wood
= 3.00), but they regarded the wood table as
more natural that the glass table (M
wood
= 4.93, M
glass
=
2.98). Thus, the two tables appeared to evoke the antici-
pated opposing types of associations.
The second pretest attempted to identify a product that
was neutral rather than extreme on the modernity and natu-
ralness dimensions. We tested three products. One was a
glass candleholder that, because of its composition mate-
rial, appeared to be both trendy and manufactured. A sec-
ond product, a bamboo picture frame, appeared to be tradi-
tional and natural for the same reason. The third product, a
celadon green mug with gold trim and a subtle cracked
ceramic design, appeared to be neutral on both dimensions.
To confirm these product characterizations, 25 participants
visually examined and evaluated the products on the same
six items used in the first pretest (alphas ranged from .69 to
.86 for the products). Each product was viewed by itself on
a table covered with a white cotton tablecloth, and product
order was rotated.
Evaluations of the products supported our expectations.
A one-way ANOVA (type of product: candleholder, mug,
picture frame) revealed a main effect of product type on
both the trendiness (M
candleholder
= 4.84, M
mug
= 4.05,
M
picture frame
= 2.59; F(2, 23) = 25.70, p < .001) and the nat-
uralness (M
candleholder
= 2.48, M
mug
= 4.00, M
picture frame
=
4.85; F(2, 23) = 60.73, p < .001) indexes. The mug was
40 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2009
viewed as less trendy than the candleholder (F(1, 24) =
7.57, p < .05) but more trendy than the picture frame
(F(1, 24) = 15.45, p < .001). In addition, it was regarded as
less natural than the picture frame (F(1, 24) = 10.40, p <
.01) but more natural than the candleholder (F(1, 24) =
29.98, p < .001). Thus, whereas the candleholder and the
picture frame were extreme in trendiness and naturalness
(i.e., low and high, and high and low, respectively), the mug
was neutral on both dimensions.
EXPERIMENT 1
Stimuli and Procedure
A total of 47 participants took part in this study in small
groups. Each was given a clipboard with attached survey
materials and stationed near a small display table. Partici-
pants were informed that the study would examine their
views about a product and a new home decor store that
would be opening soon locally.
To begin, we administered a warm-up exercise. This was
actually a self-view mind-set task that prompted holistic or
analytic processing (see Trafimow, Triandis, and Goto
1991). To induce holistic processing, half the participants
completed an interdependent mind-set task. Specifically,
they were asked to write a paragraph describing a particu-
larly meaningful event, occasion, or activity they took part
in with their family and/or friends and were instructed to
explain why doing so with these parties made the episode
especially meaningful. The remaining participants per-
formed an independent mind-set task that evoked analytic
processing. They were asked to write a paragraph about a
comparable episode, but one that they engaged in entirely
by themselves, and to explain why doing so alone was
meaningful.
Next, participants were asked to view the target product,
a mug, that the new store would carry. It was displayed on
either a glass or a wood table. Participants evaluated the
mug on the same trendiness and naturalness items used in
the pretests to gauge the two dimensions. The items that
captured each dimension were interspersed, and they pro-
duced reliable indexes of the product’s trendiness and natu-
ralness (α = .70 and .83). In addition, participants were
asked to list their thoughts about the product. Two inde-
pendent, naive coders classified thought content reliably
(α = .81) for the quantities that implied trendiness (e.g.,
“modern design, but not durable”; “the mug was pretty cool
and innovative”) and naturalness (e.g., “the mug looks kind
of rustic and reminds me of an older person”; “old-
fashioned, looks like homemade”).
Then, all display materials were removed from view, and
participants focused on some other questions pertaining to
the soon-to-open home decor store. To help choose a name
for the store, participants assessed the appropriateness of
six potential names (1 = “not at all,” and 7 = “very much”).
According to the results of a focus group, three of the
names were suggestive of trendiness: Current, Vogue, and
Silhouette. The other three connoted naturalness: Elements,
Beginnings, and Simplicity. We averaged the responses to
these sets of trendy and natural names to create two sepa-
rate indexes (α = .70 and .72). Then, to assist in choosing a
gift basket that the store would give away to its first 50 cus-
tomers, participants viewed photos of two options and
assessed how appropriate they were (1 = “not at all,” and
7 = “very much”). The order of the photos was rotated. The
focus group results indicated that one possible gift, a large
silver bowl filled with shiny new kitchen utensils and wares
(e.g., a spatula, pepper mill, colorful pot holder), displayed
a trendy appearance. The other option, a wicker basket
filled with nuts, fruit, and muffins, exuded a natural appear-
ance. The study ended with some demographic questions.
Results
All ANOVAs included self-view (interdependent or inde-
pendent) as a between-subjects factor. For thoughts, the
only other factor was table surface (glass or wood), which
we varied between subjects. For all other measures, we
included two additional factors: table surface and a within-
subject trendiness or naturalness factor, which bore a label
that varied depending on the dependent measure. For prod-
uct evaluations, it was labeled as “evaluative dimension”;
for store name appraisals, it was labeled as “store name
connotations”; and for gift basket assessments, it was
labeled as “basket appearance.” Thus, except for thoughts,
we conducted all analyses on a 2 (self-view) × 2 (table sur-
face) × 2 (variably labeled; e.g., evaluative dimension) fac-
torial. Degrees of freedom were 1 and 43. Treatment means
appear in Table 1.
Product evaluations and thoughts. A three-way interac-
tion among self-view, table surface, and evaluative dimen-
sion emerged on participants’ product evaluations (F =
18.32, p < .001). The two-way interaction between self-
view and table surface was significant on both the trendi-
ness (F = 16.89, p < .001) and the naturalness (F = 7.07,
p < .02) evaluative dimensions. Furthermore, we obtained
Ta bl e 1
EXPERIMENT 1: TREATMENT MEANS AND STANDARD
DEVIATIONS
Interdependents Independents
Glass Wood Glass Wood
Product Evaluations
Trendiness
4.53
a
(1.37)
2.55
b
(.97)
3.36
b
(1.14)
4.47
a
(1.57)
Naturalness
3.33
a
(1.41)
4.42
b
(1.70)
4.08
a,b
(1.16)
3.11
a
(.96)
Thoughts
Trendiness related
1.58
a
(1.44)
.00
b
(.00)
.25
b
(.45)
1.17
a
(1.58)
Naturalness related
.75
a
(.75)
1.64
b
(1.36)
1.50
a,b
(.80)
.67
a
(1.23)
Store Name
Trendiness
connotations
3.61
a
(.91)
2.67
b,c
(1.24)
2.36
b
(.98)
3.47
a,c
(1.02)
Naturalness
connotations
3.94
a,c
(1.43)
5.03
b
(1.06)
4.83
a,b
(1.08)
3.83
c
(1.24)
Gift Basket Options
Trendiness
associations
3.33
a
(1.30)
2.18
b,c
(1.08)
2.42
a,c
(1.17)
3.42
a
(1.51)
Naturalness
associations
03.92
a,b
(1.31)
4.82
a
(1.25)
4.58
a,b
(1.51)
3.42
b
(1.83)
Number of Respondents 12 11 12 12
Notes: Means within the same row that do not share a common super-
script differ at p < .05.
The Influence of Self-View on Context Effects 41
two-way interactions between self-view and table surface
on the number of thoughts pertaining to trendiness (F =
14.92, p < .001) and on those related to naturalness (F =
7.70, p < .01).
Planned contrasts on each of these measures upheld our
theorizing: Interdependents, who engage in holistic cogni-
tion, exhibited assimilation effects, and independents, who
employ analytic cognition, displayed contrast effects.
Specifically, interdependents evaluated the mug as more
trendy (F = 13.57, p < .001) and generated more thoughts
about its trendiness (F = 11.71, p < .001) when it was
viewed on a glass table than when it was viewed on a wood
table. Similarly, they evaluated the mug as more natural
(F = 3.87, p < .05) and generated more thoughts about its
naturalness (F = 4.00, p < .05) when it was viewed on a
wood table than when it was viewed on a glass table. How-
ever, such outcomes reversed and transformed into contrast
effects for independents. That is, independents evaluated
the mug as more trendy (F = 4.46, p < .05) and generated
more thoughts about its trendiness (F = 4.10, p < .05) when
it appeared on a wood table than when it appeared on a
glass table. They also evaluated the mug as somewhat more
natural (F = 3.21, p < .08) and offered more thoughts about
its naturalness (F = 3.70, p < .06) when it was on a glass
table than when it was on a wood table.
Store name and gift basket options. The context effects
we observed for people’s evaluations of the target product
also seem to carry over and affect their views about the
store in which the product would be sold, even though such
input was rendered when the table and product were out of
sight. An ANOVA revealed a three-way interaction among
self-view, table surface, and store name connotations on
participants’ appraisals of the store names (F = 17.72, p <
.01), and we obtained a similar interaction among self-view,
table surface, and basket associations on their gift basket
assessments (F = 11.17, p < .01). Follow-up analysis indi-
cated that for store name appraisals, the two-way inter-
action of the first two factors was significant for both the
store names that connoted trendiness (F = 11.49, p < .01)
and those that connoted naturalness (F = 8.63, p < .01). For
gift basket assessments, the same interaction emerged for
the baskets that appeared trendy (F = 8.33, p < .01) and for
those that appeared natural (F = 5.60, p < .03).
Planned contrasts on each of these measures supported
our expectation that interdependents (independents), who
engage in holistic (analytic) cognition, would exhibit an
assimilation (contrast) effect. Specifically, interdependents
perceived the trendy-connoting store names (F = 4.75, p <
.05) and the trendy-looking gift basket (F = 4.67, p < .05)
as more befitting of the store when the mug was viewed on
a glass table than when it was viewed on a wood table. Also
exhibiting assimilation effects, interdependents perceived
the store names that connoted naturalness (F = 4.57, p <
.05) as more befitting of the store when the mug was
viewed on a wood table than when it was viewed on a glass
table. The natural-looking gift basket showed a direction-
ally similar effect, though it did not reach significance (F =
2.90, p < .10). Conversely, independents revealed contrast
effects. They perceived the trendy-connoting store names
(F = 6.87, p < .02) and, marginally, the trendy-looking gift
basket (F = 3.68, p < .07) as more befitting of the store
when the mug was viewed on a wood table than when it
was viewed on a glass table. Similarly, independents per-
ceived the store names that connoted naturalness (F = 4.06,
p < .05) and the gift basket that appeared natural (F = 3.63,
p < .06) as more befitting of the store when the mug
appeared on a glass table than when it appeared on a wood
table.
Discussion
Experiment 1 examined how people’s use of holistic ver-
sus analytic modes of cognition affected the direction of
context effects on their evaluations of a neutral product dis-
played on either a glass or a wood table. Specifically, when
people adopted an interdependent self-view, they processed
the displayed items (i.e., the product and table) holistically
as a single unit. Thus, they assimilated their associations
with the table with their mental representation of the prod-
uct. Conversely, those who assumed an independent self-
view processed the items analytically, perceiving each item
as a separate entity. Yet because these participants also
placed substantial value on an entity’s distinctiveness or
uniqueness, they used their associations with the context
(i.e., the table) as a relevant standard for assessing the tar-
get product’s properties, and this comparison resulted in a
contrast effect. Furthermore, such effects emerged on
people’s target product evaluations and their thoughts about
the item’s trendiness and naturalness. Likewise, these
effects occurred on people’s appraisals of the store name
and gift basket choices that were distally related to the
product and rendered in the absence of any display
materials.
In summary, the findings are important because they
shed light on whether these well-known modes of process-
ing evoke different context effects. Furthermore, our find-
ings offer implications for a crucial applied issue. They
indicate that the surface material of a store display structure
that holds a product can influence consumers’ perceptions
of that product, but the mode of cognition that consumers
use determines how such perceptions are altered.
Nevertheless, our findings could be augmented further.
Experiment 1 did not assess our prediction (e.g., H
2
) that
the observed context effects should be limited to situations
in which the target product possesses neutral rather than
extreme associations on the contextually elicited dimen-
sions (i.e., trendiness and naturalness). In addition, the
applied implications of our work would be bolstered further
if our predictions were upheld in a field study that used
even more actionable means to vary people’s mode of cog-
nition and if we examined the contextual influence of alter-
native table surface materials. Experiment 2 addresses both
points.
EXPERIMENT 2
Stimuli
Experiment 2 was similar to Experiment 1, but it was
conducted in an alcove of a retail area within a student
union. Shoppers were recruited by asking them to provide
input about some products. To manipulate participants’ self-
view, we contracted a graphic artist to create three large
posters that expressed either an interdependent or an inde-
pendent self-view or mind-set. To exemplify, one poster
that conveyed an interdependent self-view invited viewers
42 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2009
to “Treat Those Who Are Special to You with a Gift That
Creates Memories.” It displayed an open photo album filled
with pictures of family and friends at special events. The
counterpart poster that conveyed an independent self-view
encouraged viewers to “Treat Yourself to Something That
You Alone Have Been Longing for.” It showed a person
thinking about travel (e.g., depicted by an airplane), leisure
activities (e.g., musical notes near a hammock hung on
palm trees), and fine dining (e.g., a wine bottle and glass).
Sample posters appear in the Appendix.
A pretest assessed whether each group of three posters
activated the intended alternative self-views. Thirty-four
undergraduate students viewed one of the alternative groups
of posters and then responded to separate sets of three ques-
tions on a seven-point scale (1= “not at all,” and 7 = “a lot”)
that probed whether the posters prompted thoughts either
about oneself as well as family and friends or just about
oneself (Aaker and Lee 2001). We averaged the questions
about self-cognitions to form a self-thoughts index (α =
.87). We averaged those about both self and others to form
an others-thoughts index (α = .85). As we expected, partici-
pants who viewed the posters intended to prompt an inde-
pendent (versus interdependent) self-view produced higher
ratings on the self-thoughts index (M
independent
= 5.00,
M
interdependent
= 2.35; F(1, 32) = 87.61, p < .001) but lower
ratings on the others-thoughts index (M
independent
= 2.11,
M
interdependent
= 4.92; F(1, 32) = 190.12, p < .001).
In addition, to assess whether display materials other
than glass and wood would produce context effects, we dis-
played all the products on a table covered with a cloth made
of either a silver metallic fabric or a coarse brown burlap.
Using the same procedure and measures employed previ-
ously to assess associations with glass and wood, we asked
22 participants to view one of the two tablecloths and to
evaluate it on the six trendiness and naturalness items. We
averaged these items to form separate trendiness (α = .80)
and naturalness (α = .78) indexes. One-way ANOVAs on
the two types of tablecloths (metallic and burlap) revealed a
main effect on both indexes (trendiness: F(1, 20) = 14.64,
p < .001; naturalness: F(1, 20) = 8.00, p < .01). As we
expected, the metallic tablecloth was perceived as more
trendy than the burlap one (M
metallic
= 4.33, M
burlap
= 2.70),
but the burlap tablecloth was viewed as more natural than
the metallic one (M
burlap
= 4.57, M
metallic
= 3.28). Thus, the
two tablecloths evoked the intended opposing types of
associations.
The products used in this study included the same neutral
mug from Experiment 1 plus two products that the pretest
for Experiment 1 identified as extreme on the focal dimen-
sions. The latter items were a glass candleholder that was
trendy but low in naturalness and a bamboo picture frame
that was natural but low in trendiness. Thus, in accordance
with H
1
, we expected to observe an interaction among self-
view, tablecloth type, and evaluative dimension on
responses pertaining to the neutral mug, but not on those
pertaining to the two extreme products.
Procedure
A total of 120 shoppers participated in the study. As
many as 5 participants completed the study at the same
time, and each group was assigned randomly to treatments.
Participants were positioned around three sides of a display
table that was covered with either a metallic or a burlap
tablecloth. The fourth side of the table was against a wall,
on which hung three salient posters that fostered either an
interdependent or an independent self-view. The alternative
sets of posters were changed at the beginning of each day,
and the tablecloths were varied in the middle of each day.
Furthermore, because the study was conducted in an alcove
in the student union, the preceding materials were largely
hidden from the view of passersby.
To begin, participants were shown three target products
separately. Each was set on the cloth-covered table in a
counterbalanced order. For each product, participants com-
pleted two tasks in an order that was rotated. One was an
evaluation task in which participants were given the same
six items used previously; for each, they were asked to cir-
cle the number that best reflected their feelings about the
product on display. We averaged the two sets of three items
to create a trendiness and a naturalness index for each prod-
uct (alphas ranged from .62 to .81). The other task involved
making actual choices. Participants were told that the prod-
uct on display was a private-label good that would be sold
only in a new local gift store. They were asked to choose
which of two possible brand names for the good seemed
preferable. In accordance with focus group results, one pro-
posed name was always associated with the concept of
trendiness, and the other implied naturalness. The choices
(name order was rotated) were Elegance and Earthborn for
the mug, Eloquence and EcoSystem for the candleholder,
and Tuned In and Terrain for the picture frame.
Results
Product evaluations. We analyzed product evaluations as
a 2 (self-view: interdependence versus independence) × 2
(tablecloth type: metallic versus burlap) × 2 (evaluative
dimension: trendiness versus naturalness) × 3 (type of prod-
uct: extremely low versus neutral versus extremely high)
factorial. The first two factors were varied between sub-
jects, and the last two were varied within subjects. An inter-
action involving all four factors emerged on participants’
product evaluations (F(2, 115) = 7.80, p< .001). Thus, we
examined the evaluations of each product separately. Treat-
ment means appear in Table 2.
Evaluations of the neutral mug revealed an interaction
among self-view, tablecloth type, and evaluative dimension
(F(1, 116) = 18.47, p < .001). The interaction between self-
view and tablecloth type was significant on both the trendi-
ness (F(1,116) = 13.40, p < .001) and the naturalness
(F(1,116) = 10.73, p < .001) dimensions. Furthermore,
planned contrasts conceptually replicated our previous find-
ings. Participants who viewed the interdependent self-view
signage exhibited an assimilation effect, whereas those who
saw the independent self-view signage displayed a contrast
effect. Specifically, participants exposed to the interdepend-
ent posters evaluated the mug as more trendy when it was
displayed on a metallic tablecloth than when it was dis-
played on a burlap tablecloth (F(1, 116) = 9.53, p < .01),
but they evaluated the mug as more natural when it was
presented on a burlap tablecloth than when it was presented
on a metallic tablecloth (F(1, 116) = 5.89, p < .02). In con-
trast, participants who saw the independent self-view
posters revealed the opposite pattern. They evaluated the
mug as more trendy when it was on a burlap tablecloth than
The Influence of Self-View on Context Effects 43
Ta bl e 2
EXPERIMENT 2: TREATMENT MEANS AND STANDARD
DEVIATIONS
Interdependents Independents
Burlap Metallic Burlap Metallic
Product Evaluations
Mug: trendiness
3.54
a
(1.67)
4.59
b
(1.08)
4.46
b
(1.25)
3.74
a
(1.22)
Mug: naturalness
4.32
a
(1.28)
3.44
b
(1.31)
3.38
b
(1.54)
4.18
a
(1.48)
Glass candleholder:
trendiness
4.99
a
(.86)
4.95
a
(1.12)
5.03
a
(.93)
5.00
a
(1.01)
Glass candleholder:
naturalness
2.86
a
(1.03)
2.82
a
(1.13)
2.61
a
(.87)
2.62
a
(.93)
Bamboo picture frame:
trendiness
3.21
a
(1.26)
3.45
a
(1.37)
3.08
a
(1.30)
2.98
a
(1.24)
Bamboo picture frame:
naturalness
5.34
a
(1.17)
5.12
a
(1.06)
5.27
a
(1.02)
5.07
a
(.94)
Choice of Brand Name
Mug: % of people
choosing natural
(versus trendy) name
83
(17)
32
(68)
33
(67)
90
(10)
Glass candleholder: %
of people choosing
natural (versus
trendy) name
27
(72)
21
(79)
13
(87)
30
(70)
Bamboo picture frame:
% of people
choosing natural
(versus trendy) name
93
(7)
90
(10)
80
(20)
83
(17)
Number of Respondents 29 31 30 30
Notes: Means within the same row that do not share a common super-
script differ at p < .05.
when it was presented on a metallic tablecloth (F(1, 116) =
4.36, p < .04), and they evaluated it as more natural when it
was presented on a metallic tablecloth than when it was
presented on a burlap tablecloth (F(1, 116) = 4.87, p < .03).
Next, we examined evaluations of the two extreme prod-
ucts: the glass candleholder and the bamboo picture frame.
As we predicted, no treatment effects emerged for either
product (Fs < 1). Specifically, the glass candleholder was
evaluated consistently as high in trendiness (Ms ranged
from 4.95 to 5.03) and low in naturalness (Ms ranged from
2.61 to 2.86). Similarly, the bamboo picture frame was
always evaluated as high in naturalness (Ms ranged from
5.07 to 5.34) but low in trendiness (Ms ranged from 2.98 to
3.45).
Choice of brand name. An initial ANOVA of brand name
choices for each product revealed a three-way interaction
among self-view, tablecloth type, and product type (F(2,
113) = 14.16, p< .001). Thus, we investigated the out-
comes further with binary logistic regression by entering
the preceding three factors and all interaction terms as cate-
gorical independent variables.
A regression on participants’ brand name choice for the
neutral mug revealed an interaction between self-view and
tablecloth type (β = –5.20, p < .001). Further analyses
showed that participants who saw the interdependent self-
view posters exhibited an assimilation effect. They chose
the trendy Elegance name more frequently than the natural
Earthborn name when the mug appeared on a metallic
tablecloth (68% versus 32%; χ
2
= 3.91, p < .05), but they
opted for the natural versus the trendy name when it was
viewed on a burlap tablecloth (83% versus 17%; χ
2
=
12.45, p < .001). Alternatively, participants exposed to the
independent self-view posters revealed a contrast effect.
They chose the natural name more frequently than the
trendy name when the mug appeared on a metallic table-
cloth (90% versus 10%; χ
2
= 19.20, p < .001), but they
chose the trendy name more frequently than the natural
name when it was viewed on a burlap tablecloth (67% ver-
sus 33%; χ
2
= 3.33, p < .07).
Conversely, regression analyses on participants’ brand
name choices for each of the two extreme products revealed
no treatment effects (ps > .12). For the trendy glass candle-
holder, participants consistently preferred the trendy Elo-
quence name over the natural EcoSystem name (ps < .03).
Similarly, for the natural bamboo picture frame, the natural
Terrain name was uniformly preferred over the trendy
Tuned In name (ps < .001).
To summarize, this field study conceptually replicated
and extended our previous findings. The results suggest that
various kinds of display materials (e.g., metal, burlap) can
prompt context effects on both shoppers’ product evalua-
tions and their choices. Yet this occurs only when associa-
tions with the product are neutral and not extreme on the
contextually primed dimensions. We also find that salient
content on nearby signage represents a highly actionable
means of altering people’s mode of cognition, thus control-
ling the direction of the context effects.
GENERAL DISCUSSION
This research indicates that the surface material of a
store display fixture on which products are shown can
affect consumers’ evaluations of those products. Impor-
tantly, however, the direction of this influence depends on
the particular mode of cognition that consumers use.
Specifically, when participants used holistic cognition that
was prompted by an interdependent self-view, an assimila-
tion effect occurred; their associations with the contextual
data—namely, a display table’s surface material—merged
with their evaluations of the product. However, when par-
ticipants employed analytic processing, as prompted by an
independent self-view, a contrast effect emerged; here,
associations with the contextual display table served as a
standard that was compared with those of the product. Fur-
thermore, this patterns of context effects emerged when the
target product possessed neutral or flexible associations on
the contextually implied concept. As Experiment 2 shows,
such context effects disappeared when the product pos-
sessed associations with the concept that were extreme.
This work makes an important theoretical contribution. It
adds to the context effect literature by clarifying for the first
time how and why the direction of context effects can
depend on people’s mode of cognition. We show that
whereas the use of holistic processing can elicit assimila-
tion effects, analytic processing produces contrast effects.
Notably, our research further introduces a new and action-
able way of inducing these alternative modes of cogni-
tion—namely, self-view-shaping signs that can elicit holis-
tic or analytic cognition. It also adds to extant work on
atmospherics by showing that overlooked, but omnipresent,
fixtures in retail milieus can influence product evaluations.
44 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2009
Appendix
POSTERS PRIMING SELF-VIEWS
A: Experiment 2: Posters Priming an Independent Self-View Paired with
Burlap Tablecloth
B: Experiment 2: Posters Priming an Interdependent Self-View Paired
with Metallic Tablecloth
Applications of our research also are evident. Many
brands are positioned on the dimensions we investigated in
our studies—namely, trendiness (e.g., Diesel) versus tradi-
tional (e.g., Bass) or all natural (e.g., Aveda) versus syn-
thetic (e.g., Lycra). Furthermore, retail fixtures and mer-
chandising materials often relay these same concepts (e.g.,
glass suggests trendiness), and our findings indicate that
such items can either bolster or damage product perceptions
on these dimensions. If marketers want to control which
outcome will occur, they will need to anticipate and/or
actively manage the kind of cognition that pertinent seg-
ments of shoppers use. To some degree, this could be
accomplished by considering shoppers’ demographic pro-
files. For example, extant research suggests that people of
American versus Eastern Asian descent typically employ
analytic versus holistic cognition (Gardner, Gabriel, and
Lee 1999). Yet marketers could also actively try to shape
shoppers’ cognition by, for example, using salient self-
view-altering signage of the type we used in Experiment 2.
The point is that by using selected tools strategically to
shape the type of cognition shoppers use and by coordinat-
ing this with appropriate contextual materials (e.g., glass
versus wood display tables), marketers should be able to
use our findings to their advantage.
Still, many unanswered questions remain. For example,
will people be more likely to purchase a product when con-
gruity exists between associations implied by a brand’s
positioning and the composition of the display fixture (e.g.,
when a brand named Earthborn is displayed on a wood ver-
sus glass table)? Furthermore, our research attempted to
shed light on how one dichotomous mode of cognition (i.e.,
holistic versus analytic cognition) can affect the direction
of context effects. Yet many other dichotomous modes of
cognition have been proposed (e.g., relational versus item-
specific processing, top-down versus bottom-up process-
ing), and it remains unclear whether and how these differ-
ent modes of cognitions truly differ from each other and
whether they produce parallel or different context effects.
Further research is needed to investigate this issue and to
produce a coherent framework that elucidates how all such
modes of cognition may be interrelated but different. In
addition, it is not clear which concepts are conveyed by the
myriad other materials or structural elements commonly
found in stores. Could carpeting elicit concepts such as
softness and comfort, or could ceramic floor tile evoke con-
cepts such as durability and low maintenance? Further-
more, what associations might be engendered by, for exam-
ple, brick, papered, or merely painted walls? We hope that
further research will pursue these and other important
questions.
REFERENCES
Aaker, Jennifer L. and Angela Y. Lee (2001), “‘I’ Seek Pleasures
and ‘We’ Avoid Pains: The Role of Self-Regulatory Goals in
Information Processing and Persuasion,” Journal of Consumer
Research, 28 (June), 33–49.
Gardner, Wendi L., Shira Gabriel, and Angela Lee (1999), “‘I’
Value Freedom, but ‘We’ Value Relationships: Self-Construal
Priming Mirrors Cultural Differences in Judgment,” Psycho-
logical Science, 10 (July), 321–26.
The Influence of Self-View on Context Effects 45
Herr, Paul M. (1986), “Consequences of Priming: Judgment and
Behavior,” Journal of Personality and Social Psychology, 51
(December), 1106–1115.
Ip, Grace Wai-Man and Michael H. Bond (1995), “Culture, Val-
ues, and the Spontaneous Self-Concept,” Asian Journal of Psy-
chology, 1 (1), 29–35.
Kühnen, Ulrich and Susanne Haberstroh (2004), “Self-Construal
Activation and Focus of Comparison as Determinants of Assim-
ilation and Contrast in Social Comparisons,” Current Psychol-
ogy of Cognition, 22 (April), 289–310.
——— and Bettina Hannover (2000), “Assimilation and Contrast
in Social Comparisons as a Consequence of Self-Construal
Activation,” European Journal of Social Psychology, 30
(November), 799–811.
Malaviya, Prashant, Jolita Kisilieus, and Brian Sternthal (1996),
“The Effect of Type of Elaboration on Advertising Processing
and Judgment,” Journal of Marketing Research, 33 (Novem-
ber), 410–21.
Markus, Hazel R. and Shinobu Kitayama (1991), “Culture and the
Self: Implications for Cognition, Emotion, and Motivation,”
Psychological Review, 98 (April), 224–53.
Nisbett, Richard E. (2003), The Geography of Thought. New York:
The Free Press.
———, Kaiping Peng, Incheol Choi, and Ara Norenzayan (2001),
“Culture and Systems of Thought: Holistic and Analytic Cogni-
tion,” Psychological Review, 108 (April), 291–310.
Peracchio, Laura A. and Joan Meyers-Levy (2005), “Using Stylis-
tic Properties of Ad Pictures to Communicate with Consumers,”
Journal of Consumer Research, 32 (June), 29–40.
Schwarz, Norbert and Herbert Bless (1992), “Constructing Reality
and Its Alternatives: An Inclusion/Exclusion Model of Assimi-
lation and Contrast Effects in Social Judgment,” in The Con-
struction of Social Judgments, Leonard L. Martin and Abraham
Tesser, eds. Hillsdale, NJ: Lawrence Erlbaum Associates,
217–45.
——— and ——— (2006), “Mental Construal Processes: The
Inclusion/Exclusion Model,” in Assimilation and Contrast in
Social Psychology, Diederik A. Stapel and Jerry Suls, eds. New
York: Psychology Press, 119–41.
Scott, Linda M. (1994), “Images in Advertising: The Need for a
Theory of Visual Rhetoric,” Journal of Consumer Research, 21
(September), 252–73.
Singelis, Theodore M. (1994), “The Measurement of Independent
and Interdependent Self-Construals,” Personality and Social
Psychology Bulletin, 20 (October), 580–91.
Stapel, Diederik A., Willem Koomen, and Aart Velthuijsen (1998),
“Assimilation or Contrast? Comparison Relevance, Distinct-
ness, and the Impact of Accessible Information on Consumer
Judgments,” Journal of Consumer Psychology, 7 (1), 1–23.
Trafimow, David, Harry C. Triandis, and Sharon G. Goto (1991),
“Some Tests of the Distinction Between the Private Self and the
Collective Self,” Journal of Personality and Social Psychology,
60 (May), 649–55.
Van Baaren, Rick B., William W. Maddux, Tanya L. Chartrand,
Cris De Bouter, and Ad Van Knippenberg (2003), “It Takes Two
to Mimic: Behavioral Consequences of Self-Construals,”
Journal of Personality and Social Psychology, 84 (May),
1093–1102.
