You are on page 1of 20

Journal of Retailing 81 (3, 2005) 231–250

The effects of merchandise coordination and juxtaposition


on consumers’ product evaluation and purchase intention in
store-based retailing
Shun Yin Lam a,∗ , Avinandan Mukherjee b,1
a Marketing and International Business Division, Nanyang Business School, Nanyang Technological University, Singapore 639798, Singapore
b School of Business, Montclair State University, Montclair, NJ 07043, USA

Abstract

Many retailers attempt to juxtapose and coordinate complementary products in their stores on the basis of key attributes because good
coordination of complementary items can provide consumers with relevant information for their evaluation of individual products. In addition,
the aesthetic appeal and the social impression that good coordination offers can be transferred to the products themselves. On the basis of these
theories, the authors test the effects of merchandise coordination and juxtaposition (vs. separation) on consumer behavior using actual clothing
items as stimuli in a laboratory setting. The authors find that coordination affects consumers’ evaluation of and purchase intention toward a
target item only when the item is juxtaposed with a complementary item. This effect can be explained by the aesthetic response of consumers
to and their social impression of a grouping of the items. Furthermore, in comparison with displaying the two items separately, juxtaposing
them in a poorly coordinated manner appears to depress the target item’s evaluation, though juxtaposing them in a well-coordinated manner
does not seem to improve the evaluation.
© 2005 New York University. Published by Elsevier Inc. All rights reserved.

Keywords: Merchandise coordination; Juxtaposition; Retailing

Merchandise presentation is a key issue in store manage- items would look together in a home. Many apparel retail-
ment (Levy & Weitz 1998). Although retailers use various ers also juxtapose complementary items (such as shirts and
methods to present their merchandise, the most common pants) on mannequins and free-standing fixtures so that the
method is to place all items of a product category in the items appear to be complete outfits (Berman & Evans 1995).
same section to facilitate consumer choice. For example, Such juxtaposition may enhance shoppers’ awareness of a
some apparel stores put all shirts in a particular area and retailer’s merchandise variety and stimulate them to make
all pants in another area. A similar presentation method puts complementary purchases.
items of related categories in adjacent locations. For example, In addition, some retailers devote special attention to the
apparel retailers may put shirts and pants on adjacent sides of coordination of items, making sure that complementary prod-
a gondola to facilitate consumers’ search for complementary ucts are well matched or consistent on key attributes such as
items. In yet another method, retailers combine displays of color, style, and design. Merchandise displays that are well
complementary items in the same location; that is, they jux- coordinated on these key attributes may create or enhance a
tapose the items so that shoppers see them as an ensemble. store’s image, attract consumer attention, and influence feel-
For example, furniture retailers such as IKEA combine fur- ing, shopping experience, shopping behavior, and purchase
niture in room settings to give consumers an idea of how the outcomes. Although merchandise coordination and juxtapo-
sition can be treated as separate issues, retailers tend to con-

sider them together when presenting their merchandise. For
Corresponding author. Tel.: +65 6790 4776; fax: +65 6791 3697.
E-mail addresses: asylam@ntu.edu.sg (S.Y. Lam),
example, when juxtaposing complementary clothing items on
mukherjeeav@mail.montclair.edu (A. Mukherjee). mannequins, retailers generally attempt to coordinate these
1 Tel.: +1 973 655 5126; fax: +1 973 655 7673. items so that they appear consistent in color and style.

0022-4359/$ – see front matter © 2005 New York University. Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.jretai.2005.07.006
232 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

In this study, we examine how merchandise coordination mally testing the mediating roles of perceived unity, aesthetic
and juxtaposition affect consumers’ evaluation of and inten- response, and social impression, as well as comparing the
tion to purchase individual products. Our findings provide effect of coordination in juxtaposed and separate display con-
answers to some key retail management concerns: How do ditions.
coordination and juxtaposition affect product evaluation and To test our hypotheses, we conduct a laboratory exper-
purchase intention? Does juxtaposing products in a well- iment using undergraduate student participants with men’s
coordinated manner improve product evaluation and pur- casual wear and women’s formal wear as the products under
chase intention? Does placing complementary products on examination. Our results support the majority of our hypothe-
adjacent sides of a display unit have any effect on these two ses and offer important implications for managers involved
variables? We consider these and other retail concerns in this in retail merchandise presentation.
article.
The potential effects of merchandise coordination and
juxtaposition on consumer shopping behavior have been Conceptual framework and hypotheses
acknowledged by marketing researchers (Russell et al. 1999).
However, evidence about these effects is sparse, and previous In developing our conceptual framework (shown in Fig. 1),
investigation has been limited to the effect of coordination we reviewed relevant literature on aesthetics, consumer
on consumers’ evaluation of a juxtaposed product grouping information processing, and memory. In the following sec-
(Bell, Holbrook, & Solomon 1991). In our study, we pro- tions, we describe the theoretical grounds and existing
pose a conceptual framework to explain how coordination evidence that supports the relationships proposed in our
and juxtaposition (vs. separation) affect consumers’ evalu- framework.
ation of and intention to purchase individual products. We
postulate that the effects of coordination can be explained Coordination, juxtaposition, aesthetic response, and
by a consumer’s aesthetic response to and social impres- social impression
sion of this grouping. In addition, we argue that juxtapos-
ing complementary items makes the information conveyed Invoking the unity-in-variety principle developed by
by their coordination more vivid and accessible than sepa- philosophers of art (e.g., Beardsley 1981), Bell et al. (1991)
rating them does. Therefore, juxtaposition should enhance posit that an ensemble of complementary products may be
the coordination effect. Although some relationships in our evaluated in terms of not only the individual components, but
framework have been investigated in previous studies (e.g., also the aesthetic value and social impression communicated
Bell et al. 1991; Veryzer & Hutchinson 1998), we extend by the overall product grouping. Aesthetic value (e.g., dis-
previous research by examining several relationships jointly. pleasing/pleasing, unattractive/attractive) is created, among
In particular, we contribute to this stream of research by for- other things, by unity (Wagner 1999). Highly consistent

Fig. 1. Conceptual framework. Notes. Aesthetic response and social impression concern two complementary products as a whole. : Moderating effects
of juxtaposition for H1b , H2b , H3b and H4b . H5a , H5b , H7a and H7b : Hypotheses for the mediating roles of aesthetic response, social impression and purchase
evaluation.
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 233

complements of consumer goods, also known as “Diderot Coordination can affect the social impression conveyed by
unities,” share some commonality or unity such that they a product grouping, in that an uncoordinated grouping may
look as though they belong together or there is some visual convey an odd or abnormal image and leave a poor impression
connection beyond mere chance that has caused them to on other people. Previous research on social information pro-
come together (McCracken 1988; Veryzer 1993). Using fur- cessing also shows that models dressed in well-coordinated
niture ensembles consisting of chairs, tables, mirrors, and clothing give a better impression (more competent, more
other decorative objects, Bell et al. (1991) find that merchan- desirable as a coworker, more sociable) than do models
dise coordination positively affects consumers’ perceptions dressed in poorly coordinated clothing (Lennon 1990). Fur-
of unity, which in turn has a positive effect on their aes- thermore, because information conveyed by coordination is
thetic response to the ensemble. Similarly, in their study more accessible and vivid in the juxtaposed condition than in
of new product design, Veryzer and Hutchinson (1998) find the separated condition, we expect that the effect of coordina-
that consistency in shape and trim among different parts of tion on the social impression is stronger when two products
a product design has a positive effect on consumers’ per- are put together than when they are separated. Therefore, we
ceptions of unity of the design, which in turn affects their advance the following hypotheses:
aesthetic response to the design positively. On the basis of
these findings, we posit that the coordination of two com- H2a . Merchandise coordination has a positive effect on the
plementary clothing items by color and style will have a social impression of two complementary products as a whole.
positive effect on consumers’ aesthetic response to them as a H2b . The effect of merchandise coordination on social
whole. impression is moderated by merchandise juxtaposition.
However, the effect of coordination may not be limited Specifically, merchandise coordination has a stronger pos-
to situations in which the two products are juxtaposed. After itive effect on social impression when the complementary
seeing two products displayed separately, consumers retain products are put together than when they are separated.
visual information about the products in their short-term
memory (Ashcraft 1989). Various studies on visual imagery Aesthetic response, social impression, product
provide evidence that short-term memory contains visual evaluation, purchase intention, and juxtaposition
codes and that people use these codes to make judgments or
perform certain tasks (e.g., Cooper & Shepard 1973; Dahl, Using an elaboration likelihood framework, Bloch,
Chattopadhyay, & Gorn 1999; Posner 1978). Therefore, using Brunel, and Arnold (2003) posit that consumers process aes-
their memory, consumers can assess the unity of two com- thetic information in a peripheral or elaborate fashion. The
plementary products as a whole even though the products aesthetics of product groupings can provide consumers with
are placed separately. However, we expect that the effect of hedonic value that enhances their quality of life (Bell et al.
coordination on the aesthetic response is stronger when the 1991; Bloch et al. 2003). Through a process of affect transfer
two products are put together than when they are separated or association, a consumer’s aesthetic response to a product
because information conveyed by coordination is more acces- grouping can be used as a cue to evaluate the products that
sible and vivid in the juxtaposed condition than in the separate constitute that grouping (Petty & Cacioppo 1986). This aes-
display condition (hereafter, the separated condition). There- thetic response also may serve as an issue-relevant argument
fore, we advance the following hypotheses: in their evaluation of the attributes of individual products
(Petty & Cacioppo 1986). For example, consumers may have
H1a . Merchandise coordination has a positive effect on a higher opinion of a shirt’s color if they find the overall color
consumers’ aesthetic response toward two complementary tone of an ensemble that consists of the target shirt and a pair
products as a whole. of pants pleasing. Thus, the consumer’s aesthetic response
can have a positive effect on product evaluation by serving as
H1b . The effect of merchandise coordination on aesthetic a peripheral cue or issue-relevant argument. We expect this
response is moderated by merchandise juxtaposition. Specifi- effect to be stronger in the juxtaposed condition than in the
cally, merchandise coordination has a stronger positive effect separated condition because information about the aesthet-
on consumers’ aesthetic response when the complementary ics of a product grouping is more vivid and accessible in the
products are put together than when they are separated. juxtaposed condition. Also, in the separated condition, not
all people will attempt to visualize how the complementary
The perceived unity of a product grouping created by products would look if they were put together, leading to a
coordination also may affect the social impression of the weaker observed impact of the aesthetics on product evalua-
grouping (e.g., socially acceptable/unacceptable, fashion- tion. Bell et al. (1991) provide evidence that, in the juxtaposed
able/unfashionable). According to Reed (2002) and Bell et condition, aesthetic response to a product grouping has a pos-
al. (1991), to attain or maintain a particular social identity, itive effect on the evaluation of the grouping, but they do not
consumers may use the consumption of socially visible prod- examine this effect in the separated condition. Neither do
ucts (such as clothing) to make a desired impression on other they examine the effect of aesthetic response on consumers’
people. evaluation of individual products. Hence, we hypothesize
234 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

H3a . An aesthetic response toward two complementary According to the preceding discussion, juxtaposition mod-
products as a whole has a positive effect on the evaluation erates the effect of coordination on product evaluation. As
of the individual target product. Fig. 1 indicates, this originates from the moderating effects
of juxtaposition on the relationships between: (1) coordina-
H3b . The effect of the aesthetic response on the evaluation of tion and aesthetic response (H1b ), (2) coordination and social
the target product is moderated by merchandise juxtaposition. impression (H2b ), (3) aesthetic response and product evalua-
Specifically, the aesthetic response has a stronger positive tion (H3b ), and (4) social impression and product evaluation
effect on the evaluation when the complementary products (H4b ). On the basis of the notion of “mediated moderation”
are put together than when they are separated. advanced by Baron and Kenny (1986), we suggest that aes-
thetic response and social impression mediate the moderating
effect of juxtaposition on the relationship between coordina-
Similarly, we can use the elaboration likelihood frame- tion and product evaluation. Aesthetic response and social
work to hypothesize the effect of social impression on prod- impression are mediators because the moderating effect of
uct evaluation. Because of the social-enhancing value that juxtaposition can be explained partly by aesthetic response
a product grouping brings, consumers may use its social and social impression, in that coordination affects these two
impression as a peripheral cue to evaluate individual prod- variables differently depending on whether the items are jux-
ucts. This social impression also may provide consumers with taposed. Therefore,
an issue-relevant argument to judge the social acceptability
of individual products, thereby affecting product evaluation H5b . Aesthetic response and social impression totally or
positively. Similar to the effect of aesthetic response, we partially mediate the moderating effect of merchandise jux-
expect that the effect of social impression on product eval- taposition on the relationship between merchandise coordi-
uation is stronger in the juxtaposed condition than in the nation and the evaluation of the target product.
separated condition because of the vividness and accessi-
bility of visual information in the former. We therefore posit The mediating role of product evaluation and its effect on
that purchase intention

H4a . The social impression of two complementary prod- Product evaluation, defined as an attitude toward a prod-
ucts as a whole has a positive effect on the evaluation of the uct, may positively affect the perceived value of acquiring
individual target product. the product. Perceived acquisition value has been found to
have a positive effect on consumers’ willingness to buy the
H4b . The effect of the social impression on the evaluation product (Grewal, Monroe, & Krishnan 1998). Furthermore,
of the target product is moderated by merchandise juxtaposi- the attitude-intention linkage is well supported by attitudinal
tion. Specifically, the social impression has a stronger positive literature (Bagozzi 1982; Dabholkar 1994). Therefore,
effect on the evaluation when the complementary products
are put together than when they are separated. H6 . The evaluation of the target product has a positive effect
on the consumers’ intention to purchase the product.
The mediating roles of aesthetic response and social Fig. 1 shows that aesthetic response and social impression
impression affect purchase intention positively because they have posi-
tive effects on product evaluation (H3a and H4a ), and product
Following the recommendation of Baron and Kenny evaluation in turn has a positive effect on purchase intention
(1986), we distinguish the mediating role of variables in a (H6 ). Therefore, we posit that
main effect from their mediating role in a moderating effect,
and formulate hypotheses about these roles separately. Our H7a . The evaluation of the target product totally or partially
discussion and conceptual framework in Fig. 1 suggest that mediates the main effects of the aesthetic response toward
coordination affects product evaluation positively because (1) and the social impression of two complementary products as
coordination has a positive effect on aesthetic response and a whole on the purchase intention of the target product.
social impression (H1a and H2a ) and (2) aesthetic response
and social impression have a positive effect on product eval- We also deduce that juxtaposition moderates the rela-
uation (H3a and H4a ). This discussion leads us to posit tionships between the aesthetic response and purchase
that intention, and between the social impression and purchase
intention. This is because (1) juxtaposition moderates the
H5a . The aesthetic response toward and social impression relationships between the aesthetic response and product
of two complementary products as a whole totally or partially evaluation, and between the social impression and prod-
mediate the main effect of merchandise coordination on the uct evaluation (H3b and H4b ), and (2) product evaluation
evaluation of the target product. affects purchase intention (H6 ). Consequently, based on the
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 235

notion of mediated moderation (Baron & Kenny 1986), we diagonal stripes. To create the poorly and well-coordinated
advance groupings, we selected complementary items that were either
consistent or inconsistent with the target item in color and
H7b . The evaluation of the target product totally or partially style (as shown in Fig. 2). For the poorly coordinated group-
mediates the moderating effects of merchandise juxtaposition ing, we chose a pair of red floral shorts for the men’s casual
on the relationships between aesthetic response and purchase wear and a skirt with a green, blue, yellow, and red floral pat-
intention and between social impression and purchase inten- tern for the women’s formal wear. For the well-coordinated
tion. grouping, we selected a pair of green Bermuda shorts for
the men and brown, straight-cut executive pants for the
women.
Method The juxtaposition factor was represented by two types of
separate display conditions and one juxtaposed display condi-
Design tion. The “separated by four dummies” condition simulates a
shopping situation in which complementary products are not
In our experiment, each participant was asked to inspect placed close to each other. The “separated but with no inter-
a set of men’s casual wear and women’s formal wear, which vening dummies” condition represents a display arrangement
represent two replications for testing our hypotheses. In in which the products are placed close to each other but not
this study, “men’s casual wear” is defined as clothes worn juxtaposed. As Fig. 3 indicates, for each trial, participants
by men in casual occasions and “women’s formal wear” were asked to inspect the target item, a complementary item,
as clothes worn by women in an office environment or at and four dummy items in a sequence that varied according to
social gatherings. Formal wear is relevant to our undergrad- the juxtaposed/separated condition. The dummy items were
uate student participants because most worked either part- products that did not complement the target item in daily
or full-time or as interns in organizations before taking part use, such as swimwear, bedsheets, towels, and children’s
in our experiment. The experiment followed a 2 × 3 × 2 wear.
between-subjects design, with two levels of coordination Specifically, in the “separated by four dummies” condi-
(well- and poorly coordinated), three levels of juxtaposi- tion, the complementary item was displayed flat on a table to
tion/separation (separated by four dummies, separated with the participant first. When the participant finished inspect-
no intervening dummies, and juxtaposed), and two lev- ing the product, it was hidden from sight. Four dummies
els of order (men’s wear before women’s wear and vice then were displayed successively in a manner similar to that
versa). of the complementary item. After the participant completed
inspecting the dummies, he or she was shown the target
Stimuli and manipulation item placed on a particular fixture. For the men’s casual
wear, the shirt was placed on a hanger, and for the women’s
Clothing was selected as the product for study for two rea- formal wear, it was displayed on a mannequin. In the “sep-
sons. First, coordinating and juxtaposing clothing items on arated with no intervening dummies” condition, the four
display is a common practice in apparel retailing (Berman & dummy items were shown first, followed by the complemen-
Evans 1995; Levy & Weitz 1998). Second, previous research tary item, and then the target item. The juxtaposed condition
suggests that in the purchase of clothing (particularly fash- was identical, except that the target and complementary items
ion), consumers commonly consider both the product’s aes- were juxtaposed and shown to the participant simultaneously
thetic value and the social impression it conveys (Bell et (Figs. 2 and 3).
al. 1991; Holbrook 1986). By replicating our study across To avoid the potential confounding effect of product famil-
items of clothing worn by the different genders, we avoid iarity, we purchased all the items from stores that our par-
any possible attribution of our findings to the gender-specific ticipants were unlikely to visit frequently. To control for
nature of the items. The replication also enables us to examine the effects of brand names and price, we chose items that
the applicability of our framework to two different product did not display brand names and were neither very expen-
classes. sive nor very cheap. For the same reason, we removed
Because the coordination effect may be strongest for those name labels from all items. However, we retained price and
items that are rated neutrally,2 we chose target items that we material labels on the items to make the shopping situation
rated neither good nor bad. The target item for men’s casual realistic.
wear was a blue T-shirt with horizontal stripes. The target item According to adaptation level theory (Helson 1964), a per-
for women’s formal wear was a yellow-green satin shirt with son’s prior exposure to related stimuli in the same modality
may affect his or her response to current stimuli. Therefore,
2 We thank an anonymous JR reviewer for this suggestion. The effects of
the order in which each set of clothing was shown could have
merchandise presentation may be stronger for ambiguous stimuli (clothing
affected the endogenous variables in our framework. To con-
with a neutral rating) than for clear-cut stimuli (clothing with an extreme trol for this potential bias, the order was randomized across
rating). participants. Accordingly, participants randomly were asked
236 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

Fig. 2. Juxtaposed display of the target and complementary items.

to inspect either the men’s or women’s wear during the first from marketing literature (e.g., Bell et al. 1991; Herr, Kardes,
trial. & Kim 1991; Yi 1990) (Table 1).

Variables and measures Participants and procedure

The exogenous variables under investigation are the coor- Five hundred seventy-seven undergraduate students from
dination and juxtaposition factors and the two control vari- various disciplines took part in this study for either par-
ables, namely, gender and order. All are categorical variables. tial fulfillment of course requirements or monetary com-
We include gender because culturally defined roles might pensation. Each participant was assigned randomly to a
cause differences in preferences (e.g., women may prefer pas- particular combination of the coordination, juxtaposition,
tel shades) between men and women (Holbrook 1986) and and order conditions and underwent two trials, one focus-
thereby affect the endogenous variables (purchase intention, ing on men’s wear and the other on women’s wear.
product evaluation, aesthetic response, social impression, and For the men’s wear trial, participants were given the
perceived unity). The endogenous variables are measured by following instructions before inspecting the experimental
multi-item, seven-point, semantic differential scales adopted items:
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 237

Fig. 3. Sequence of product displays in the separated and juxtaposed conditions.

Table 1
Scale items and measurement properties (men’s wear)
Separated by four dummies Separated with no intervening dummies Juxtaposed

IRa CRa VEa IR CR VE IR CR VE


Perceived unityb
Low in unity/high in unity .85 .94 .90 .88 .95 .89 .90 .96 .91
Poorly coordinated/well-coordinated .93 .92 .95
Inconsistent/consistent .74 .80 .82
Aesthetic response
Offensive/enjoyable .62 .96 .86 .68 .96 .81 .68 .97 .87
Poor-looking/nice-looking .89 .84 .90
Displeasing/pleasing .81 .85 .88
Unattractive/attractive .80 .76 .82
Bad appearance/good appearance .83 .85 .88
Ugly/beautiful .80 .74 .83
Social impression
Socially unacceptable/socially acceptable .48 .90 .79 .56 .92 .79 .56 .91 .83
Unfashionable/fashionable .65 .67 .65
Undesired impression/desired impression .78 .80 .77
Disapproved by others/approved by others .68 .70 .74
Unpopular/popular .69 .72 .67
Low in status/high in statusc
Product evaluation
Bad/good .85 .96 .66 .82 .94 .69 .85 .96 .68
Unfavorable/favourable .93 .91 .89
Undesirable/desirable .86 .78 .87
Dislike/like .80 .73 .85
Not at all useful/very usefulc
Purchase intention
Not likely at all/almost sure to buy .97 .94 .84 .93 .94 .87 .94 .95 .89
Impossible/possible .82 .84 .87
a For each construct, the item reliability (IR), composite reliability (SR), and average variance extracted (VE) are provided.
b The scales of all constructs follow a seven-point semantic differential format.
c These items were not included in the finalized scales.
238 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

“Please imagine that you want to buy a T-shirt as a birth- our belief that the dependent variables in our study are inter-
day present for one of your male friends or relatives who related (Fig. 1), the probability values of the ANOVAs are
is in his twenties, and you are looking for a T-shirt that he not independent of one another (Tabachnick & Fidell 2001).
can wear on casual occasions. Imagine that you are now Also, this MANOVA–ANOVA approach is not useful for
shopping in a department store for the T-shirt and encounter understanding the nature and process of the experimental
a number of products during your shopping. I am going effects on the interrelated variables (Bray & Maxwell 1985;
to show you these products. You can touch them if you Yi 1990).
want.” Second, we could conduct a step-down analysis, which
better meets our research objectives (Bray & Maxwell 1985;
The instructions for the women’s wear trial were similar,
Roy 1958; Tabachnick & Fidell 2001) because it examines
except that they described a somewhat different shopping
sequential relations among the original set of dependent
goal:
variables and indicates whether variation in a dependent
“A female friend or relative of yours in her twenties recently variable is due to the direct or indirect effects of an inde-
got a new job in a company. As she is going to work in an pendent variable. Together with ANOVA and/or ANCOVA
office environment, you are asked to imagine that you want (Baker, Levy, & Grewal 1992), a step-down analysis can
to buy a top as a gift for her that she can wear in the office or perform the mediating test recommended by Baron and
in social gatherings with friends or relatives.” Kenny (1986). In addition, unlike the univariate ANOVA
tests of the MANOVA–ANOVA approach, the probability
By specifying the shopping goals, we intended to make
values associated with a step-down analysis are independent
the participants, regardless of gender, feel that the shopping
(Roy 1958). Hence, a step-down analysis was used in this
task was reasonable and thus ensure that they would involve
research.
themselves actively in inspecting the products.
In a step-down analysis, a researcher assigns priorities
After participants had finished inspecting all items per-
to the dependent variables being tested according to the-
taining to the first trial, they were asked to complete a
oretical or practical considerations (Tabachnick & Fidell
questionnaire containing the endogenous variable measures.3
2001). The highest-priority variable is tested in univariate
Participants assigned to the separate display conditions were
ANOVA, and the remaining dependent variables are tested in
asked to imagine that the target and complementary items
a series of ANCOVAs. Each successive dependent variable is
had been put together when they answered the questions on
tested with higher-priority dependent variables as covariates
perceived unity, aesthetic response, and social impression.
to determine what, if anything, it adds to the combination
After participants had completed the questionnaire for the
of dependent variables already tested. Using our conceptual
first trial, they proceeded to the second trial with a different
framework in Fig. 1, we assign the highest priority to the aes-
set of products. The procedure was repeated, and after the
thetic response and social impression constructs, followed by
two trials, participants were asked to answer a final question
product evaluation and then purchase intention.
about how often they found poorly matched product displays
in retail stores ((1) never, (2) seldom, (3) sometimes, (4) often,
(5) very often).
Results
Analysis
Confirmatory factor analysis (CFA)
Following Anderson and Gerbing (1988), we conducted
The results of the multi-group CFAs, for which we allowed
confirmatory factor analysis (CFA) to assess the conver-
the parameters to vary across the three groups (juxtaposed
gent (reliability) and discriminant validity of the multi-
and separate display), indicate problems with one product
item measures before testing our hypothesized relationships
evaluation item (very useful/not at all useful) and one social
using LISREL 8.30. We considered two hypotheses-testing
impression item (high in status/low in status). Across the three
approaches. First, we could conduct a MANOVA to exam-
groups and the two replications, these items are consistently
ine whether any of the dependent variables were affected
lower than others measuring the same constructs in their fac-
by the independent variables. Subsequently, using separate
tor loadings by approximately .05–.15. They also differ from
univariate ANOVAs, we could test the effects of the indepen-
the other items in meaning. Specifically, this evaluation item
dent variables on each dependent variable. However, given
focuses on the utilitarian value of a product, unlike the other
evaluation items that focus on the overall evaluation of a prod-
3 The first page of the questionnaire included questions on product eval- uct. The social impression item, which pertains to status, does
uation and purchase intention. Because a participant’s response to these not necessarily have a close relationship with social accept-
questions could provide him or her with diagnostic information for answer- ability or approval, because a product that does not project
ing subsequent questions on perceived unity, aesthetic response, and social
impression (Feldman & Lynch 1988), we placed the latter questions on the
a high status can still be socially acceptable. Therefore, on
second page, and the participant was instructed not to go back to the first both empirical and substantive grounds, we excluded these
page while answering questions on the second page. two items from subsequent analyses.
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 239

We also examined the invariance of factor loadings across on the values obtained for the constrained and unconstrained
the three groups by setting the loadings to be equal. The models. Following the recommendation of Anderson and
invariance test is based on the difference in the χ2 goodness- Gerbing (1988), we conduct this test for one pair of factors at a
of-fit statistic between unconstrained and constrained CFA time and find that all the χ2 statistics are significant at the .001
models. The test results show no difference in the factor load- level, in support of discriminant validity. We also conduct a
ings across the three groups for either men’s or women’s complementary assessment by determining whether the con-
wear (men’s wear, χ2 (30) = 21.64, p > .05; women’s wear, fidence interval (plus or minus two standard errors) around
χ2 (30) = 39.55, p > .05). Therefore, we only examined the the correlation estimate between two constructs includes 1
results of the constrained model. (Anderson & Gerbing 1988). We find that, for every pair of
The majority of the goodness-of-fit indices suggest an constructs, the interval does not include 1, thus providing
acceptable model fit for the men’s wear data. The χ2 further support for discriminant validity.
statistic’s significance at the .001 level (χ2 (510) = 1,049) Having established the convergent and discriminant valid-
may be related to the large sample size used (577 cases). ity of the measures, we derive the score of each construct by
The root mean square error of approximation (RMSEA) averaging the items of each scale. We use this score in our
is .075, and the standardized root mean square residual subsequent manipulation check and hypothesis testing.
(RMR) is .044. According to the benchmarks recommended
by structural equation modeling literature (Bagozzi & Yi
Manipulation check
1988; Diamantopoulos & Siguaw 2000), the RMSEA shows
a reasonable fit (below .08), and the standardized RMR
We performed a manipulation check on merchandise coor-
shows an acceptable fit (below .05). Similarly, the incre-
dination with perceived unity as a criterion variable. The
mental fit indices are greater than .9, suggesting a good fit
check confirms that our coordination manipulation is effec-
(non-normed fit index [NNFI] = .95, comparative fit index
tive in influencing unity perception in the expected direc-
[CFI] = .96). However, the low goodness-of-fit index (GFI)
tion. As Table 2 shows, the marginal means of perceived
(.83, less than the .9 threshold) suggests a relatively poor
unity are higher when the complementary products are well-
fit. To investigate the model’s viability further, we inspect
coordinated than when they are poorly coordinated. We find
the factor loadings and find that all standardized loadings
an interaction effect between coordination and juxtaposition
are high (greater than .6) and significant at the .001 level,
on perceived unity (men’s wear: F(2, 569) = 12.8, p < .001,
corroborating the model fit. Considering these results, we
η2 = .043; women’s wear: F(2, 569) = 3.4; p < .05, η2 = .012).
regard the fit of the constrained model acceptable for men’s
Table 2 and Fig. 4 show that the mean difference between
wear.
the poorly and well-coordinated conditions is greater when
The fit indices for women’s wear are similar: χ2 (510) =
the complementary products are juxtaposed (men’s wear:
1,160, p < .001; RMSEA = .093; standardized RMR = .051;
Mdiff = 2.26; women’s wear: Mdiff = 2.45) than when they are
GFI = .83; NNFI = .95, and CFI = .96. Also, the standardized
separated by four dummies or with no intervening dummies
factor loadings are high (greater than .6) and significant at the
(men’s wear: Mdiff = 1.11, 1.25; women’s wear: Mdiff = 1.73,
.001 level, supporting model fit. Furthermore, we apply the χ2
2.01).
difference test to examine the invariance of the error variance
In addition, we examined in depth the estimated marginal
and factor covariance across groups and find no support for
means in the juxtaposed condition. Table 2 shows that for
invariance.4
men’s wear, the mean of perceived unity is 4.24 for the
Subsequently, we examined the convergent validity (reli-
well-coordinated condition and 1.98 for the poorly coor-
ability) of the measures using our CFA model with the factor
dinated condition. T-tests comparing these means with the
loadings set equal across all groups. In Table 1, we show
middle (neutral) point of the perceived unity scale indi-
that for the men’s wear sample, item reliability is moderate
cate that the former mean is higher than neutral at the
to high. The composite reliability of all constructs is greater
.05 level (t(100) = 1.9, one-sided test), whereas the latter
than .6, and the average variance extracted by each construct
is lower than neutral at the .001 level (t(95) = −16.5, one-
is greater than .5, thus providing evidence for convergent
sided test). For women’s wear, the respective mean estimates
validity for men’s wear (Bagozzi & Yi 1988; Fornell & Lar-
(5.01 for the well-coordinated condition and 2.58 for the
cker 1981). The similar reliability results for the women’s
poorly coordinated condition) also are significantly differ-
wear sample also establish their convergent validity.5
ent from the neutral point at the .01 level in the expected
We assess the discriminant validity of the measures by
directions.
constraining the estimated correlation parameter between two
constructs to be equal and then performing a χ2 difference test
Hypothesis testing
4 Detailed results on the analysis of the invariance are available from the

first author. We report the results of our analyses in Tables 3 and 4 and
5 The reliability results of the women’s wear sample are available from describe the results for each dependent variable in the order
the first author. of its priority in the step-down analyses.
240 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

Fig. 4. Plots of estimated marginal means of perceived unity, aesthetic response, social impression, product evaluation, and purchase intention against mer-
chandise coordination and juxtaposition. Notes. The estimated marginal means were generated from the univariate ANOVAs with merchandise coordination,
juxtaposition, gender, and age as fixed factors. Sep. (4) = separated by four dummies condition; Sep. (0) = separated with no intervening dummies condition;
and Jux. = juxtaposed condition.
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 241

Fig. 4. (Continued ).

Aesthetic response and social impression resenting the differences in the effect of coordination between
Tables 3 and 4 show a main effect of coordination on the juxtaposed and separated conditions are also significant
aesthetic response (men’s wear: F(1, 569) = 100.1, p < .001, at the .01 level (b = .71, t(569) = 3.2; b = .65, t(569) = 2.9).6
η2 = .150; women’s wear: F(1, 569) = 237.7, p < .001, Therefore, H1b is supported for men’s wear. However, as
η2 = .295). As Table 2 indicates, the mean differences in aes- Table 4 shows, the interaction effect is not significant for
thetic response between the coordination conditions (Mdiff ) women’s wear (F(2, 569) = .7; p > .05). Therefore, H1b is not
are positive, which indicates that aesthetic response is higher supported for women’s wear.
when complementary products are well coordinated than Examining the ANOVA results for social impression, we
when they are poorly coordinated. Therefore, H1a is sup- find a main effect of coordination on social impression (men’s
ported. wear: F(1, 569) = 93.2, p < .001, η2 = .141; women’s wear:
As shown in Table 3 and Fig. 4, we also identify an inter- F(1, 569) = 201.9, p < .001, η2 = .262). Also, as Table 2 indi-
action effect between coordination and juxtaposition for the
aesthetic response toward the men’s outfit (F(2, 569) = 6.1,
6 We used the General Linear Model Procedure of SPSS 11.5 to run the
p < .001, η2 = .021). As Table 2 indicates, the mean difference
ANOVAs and ANCOVAs. In addition to the F-statistics, the procedure also
in aesthetic response is much greater when the complemen- provided parameter estimates, or regression coefficient estimates conveying
tary products are juxtaposed (Mdiff = 1.38) than when they information about the direction and absolute magnitude of the effects under
are separated (Mdiff = .67, .73). The parameter estimates rep- study.
242 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

cates, the mean differences in social impression between


Mean difference

the coordination conditions (Mdiff ) are positive, which indi-


cates that the impression is higher when the complementary
(Mdiff )

products are well coordinated than when they are poorly coor-
2.26

1.38
1.29
.84
.97

2.45

1.40
1.58
.51
.50

dinated. Therefore, H2a is supported.


We also identify in Table 3 and Fig. 4 an interaction effect
Well-coordinated

between coordination and juxtaposition on the social impres-


sion of the men’s outfit (F(2, 569) = 4.8, p < .01, η2 = .016).
4.24 (.13)

3.74 (.11)
4.17 (.12)
3.78 (.14)
4.16 (.18)

5.01 (.14)

4.27 (.12)
4.78 (.12)
4.09 (.15)
4.32 (.17)

As Table 2 indicates, the mean difference in social impression


is much greater when the complementary products are juxta-
posed (Mdiff = 1.29) than when they are separated (Mdiff = .58,
Poorly coordinated

.86). The parameter estimate representing the difference


between the juxtaposed and the “separated by four dummies”
Juxtaposed

conditions is significant at the .01 level (b = .70, t(569) = 3.1),


1.98 (.12)

2.36 (.11)
2.88 (.11)
2.94 (.14)
3.19 (.17)

2.56 (.13)

2.87 (.12)
3.20 (.11)
3.58 (.15)
3.82 (.17)

but that representing the difference between the juxtaposed


and the “separated with no intervening dummies” conditions
is nonsignificant at the .05 level (b = .43, t(569) = 1.9). Over-
Mean difference

all, H2b is supported by the ANOVA results for men’s wear,


but, as Table 4 shows, the interaction effect is not significant
for women’s wear (F(2, 569) = 1.6, p > .05). Therefore, H2b
(Mdiff )

1.25

.73
.86
.10
.12

2.01

1.41
1.27
−.03
−.08

is not supported for women’s wear.

Product evaluation
Well-coordinated
Separated with no intervening dummies

The ANOVA results in Tables 3 and 4 depict the inter-


3.56 (.13)

3.35 (.11)
4.00 (.12)
3.83 (.14)
4.22 (.18)

4.70 (.14)

4.16 (.12)
4.54 (.12)
4.07 (.15)
4.37 (.17)

action effect of coordination and juxtaposition on the eval-


uation of the target product (men’s wear: F(2, 569) = 7.5,
p < .01, η2 = .026; women’s wear: F(2, 569) = 4.8, p < .01,
η2 = .016). The interaction pattern is also evident in Table 2
Poorly coordinated

and Fig. 4. For men’s wear, the mean difference in prod-


uct evaluation between the coordination conditions is greater
2.31 (.13)

2.62 (.11)
3.14 (.12)
3.73 (.14)
4.10 (.18)

2.69 (.14)

2.75 (.12)
3.27 (.12)
4.10 (.15)
4.45 (.17)

when complementary products are juxtaposed (Mdiff = .84)


than when they are separated (Mdiff = −.18, .10). Planned
pairwise comparison confirms that in the juxtaposed con-
dition, the mean product evaluation is greater when the com-
Mean difference

plementary products are well coordinated than when they


are poorly coordinated (t(195) = 4.2, p < .001). In contrast,
(Mdiff )

there is no significant difference in the mean between the


The lowest value these variables can take is one; the highest is seven.
1.11

.67
.58
−.18
−.22

1.73

1.65
1.19
−.40
−.44
Estimated marginal means and standard deviations for major variables

coordination conditions when the products are separated by


four dummies or with no intervening dummies (four dum-
Well-coordinated

mies: t(185) = −.0, p > .05; no dummies: t(191) = .5, p > .05).
Similarly, for women’s wear, the mean difference between
3.32 (.13)

3.17 (.11)
3.65 (.12)
3.51 (.14)
3.87 (.18)

4.56 (.14)

4.41 (.12)
4.54 (.12)
3.95 (.15)
4.27 (.17)

the coordination conditions is greater when the products


Separated by four dummies

are juxtaposed (Mdiff = .51) than when they are separated


(Mdiff = −.40, −.03). Planned pairwise comparison confirms
Poorly coordinated

that in the juxtaposed condition, the mean is greater when


the products are well coordinated than when they are poorly
coordinated (t(195) = 2.5, p < .05). In contrast, there is no sig-
2.21 (.13)

2.50 (.12)
3.07 (.12)
3.73 (.15)
4.09 (.18)

2.83 (.14)

2.76 (.12)
3.35 (.12)
4.35 (.15)
4.71 (.18)

nificant difference in the mean between the two coordinated


conditions when the products are separated by four dum-
mies or with no intervening dummies (t(185) = −1.9, p > .05;
Aesthetic response

Aesthetic response
Product evaluation

Product evaluation
Purchase intention

Purchase intention

t(191) = −.1, p > .05, respectively).


Dependent variables

Dependent variables
Social impression

Social impression
Manipulation check

Manipulation check
Variables in the studya

Women’s formal wear


Perceived unity

Perceived unity
Men’s casual wear

In Fig. 4, we also note that the mean differences between


the juxtaposed and separated conditions are much smaller
when the complementary products are well coordinated than
Table 2

when they are poorly coordinated. Planned pairwise compar-


a

ison shows that in the well-coordinated condition, there is


S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 243

Table 3
F tests—men’s casual wear

Notes. Aesthetic: aesthetic response toward two complementary products as a whole; social: social impression of two complementary products as a whole;
coordination: the coordination factor; juxtaposition: the juxtaposition factor; evaluation: evaluation of the target item; intention: intention to purchase the target
item; gender: male or female; order: order in which the men’s and women’s wear appeared. The step-down analysis is highlighted with bold borders. The other
analyses shown complete the mediation tests and examine the total effects of coordination and juxtaposition on product evaluation and purchase intention.

no difference in product evaluation between the juxtaposed response, social impression, and their interaction with jux-
and “separated by four dummies” conditions (men’s wear: taposition as covariates. We find that aesthetic response
t(189) = 1.3, p > .05; women’s wear: t(189) = −.2, p > .05) or has a main effect on product evaluation (men’s wear:
between the juxtaposed and “separated with no interven- F(1, 563) = 41.4, p < .001, η2 = .068; women’s wear: F(1,
ing dummies” conditions (men’s wear: t(190) = .7, p > .05; 563) = 51.5, p < .001, η2 = .084). Furthermore, the parame-
women’s wear: t(190) = .1, p > .05). In contrast, planned pair- ter estimates representing the simple main effect of aesthetic
wise comparison in the poorly coordinated condition shows response in different juxtaposed/separated conditions are all
that product evaluation is lower when the complementary positive and significant at the .05 level. Therefore, H3a is
products are juxtaposed than when they are separated by supported.
four dummies (men’s wear: t(191) = −4.0, p < .001; women’s In addition, the interaction effect between aesthetic
wear: t(190) = −8.9, p < .001). Similarly, the evaluation is response and juxtaposition is significant for men’s wear
lower when the complementary products are juxtaposed than (F(2, 563) = 3.1, p < .05, η2 = .011) but not for women’s
when they are separated with no intervening dummies (men’s wear (F(2, 563) = 1.6, p > .05). Aesthetic response appears
wear: t(196) = −4.0, p < .001; women’s wear: t(196) = −7.8, to have a greater effect on the evaluation of the men’s T-
p < .001). shirt when the complementary products are juxtaposed than
Overall, we conclude that there is a negative bias in the when they are separated. The parameter estimate repre-
effects of coordination and juxtaposition. Poor coordination senting the difference between the juxtaposed and “sepa-
depresses product evaluation when the products are juxta- rated by four dummies” conditions is nonsignificant at the
posed, but good coordination does not enhance product eval- .05 level (b = .33, t(563) = 1.8), whereas that for the dif-
uation in general. ference between the juxtaposed and the “separated with
In the next step of the step-down analysis, we perform no intervening dummies” conditions is significant at the
an ANCOVA on product evaluation by including aesthetic .05 level (b = .39, t(563) = 2.4). Overall, the men’s wear
244 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

Table 4
F tests—women’s formal wear

Notes. Aesthetic: aesthetic response toward two complementary products as a whole; social: social impression of two complementary products as a whole;
coordination: the coordination factor; juxtaposition: the juxtaposition factor; evaluation: evaluation of the target item; intention: intention to purchase the target
item; gender: male or female; order: order in which the men’s and women’s wear appeared. The step-down analysis is highlighted with bold borders. The other
analyses shown complete the mediation tests and examine the total effects of coordination and juxtaposition on product evaluation and purchase intention.

results support H3b , but the women’s wear results do the ANCOVA. In addition, when these two covariates are
not. controlled for, the main effect of coordination remains sig-
The ANCOVA also shows that social impression nificant (F(1, 563) = 17.7, p < .001, η2 = .030). The parameter
has a main effect on product evaluation (men’s wear: estimate representing this effect not only becomes less pos-
F(1, 563) = 11.9, p < .01, η2 = .021; women’s wear: F(1, itive, but even turns negative when the two covariates are
563) = 10.3, p < .01, η2 = .018). Furthermore, the parameter included in the analysis.7 Therefore, in contrast to its posi-
estimates of the simple main effect of social impression in tive effect, which is transmitted through aesthetic response
different juxtaposed/separated conditions are all positive and and social impression, coordination appears to have an addi-
significant at the .05 level. Therefore, the ANCOVA sup- tional negative effect on evaluation. Taken together, these
ports H4a (the positive effect of aesthetic response on product results suggest that aesthetic response and social impression
evaluation). However, the analysis shows that the interaction partially mediate the main effect of coordination on product
effect of social impression and juxtaposition is nonsignificant evaluation. Consequently, H5a is supported for men’s wear.
at the .05 level (men’s wear: F(2, 563) = .1, p > .05; women’s For women’s wear, the ANOVA on product evaluation shows
wear: F(2, 563) = .6, p > .05), and therefore, H4b is not sup- that the main effect of coordination on product evaluation
ported. is nonsignificant (F(1, 569) = .8, p > .05), and consequently,
To test whether aesthetic response and social impression H5a is not supported.
mediate the relationship between coordination and product We also conduct similar tests to examine whether aesthetic
evaluation (H5a ), we follow a procedure recommended by response and social impression mediate the moderating effect
Baron and Kenny (1986). For men’s wear, we already have
found that coordination affects aesthetic response, social 7 Baron and Kenny (1986) recommend that, in testing the fourth condition,
impression, and product evaluation and that both aesthetic researchers should examine not only the significance of an effect, but also
response and social impression have significant effects on its absolute size. Therefore, we examine the parameter estimate representing
product evaluation when they are included as covariates in the effect.
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 245

of juxtaposition on the relationship between coordination and dition shows that purchase intention is lower when the
product evaluation (H5b ). For men’s wear, our testing for H1b complementary products are juxtaposed than when they are
and H2b shows that juxtaposition moderates the relationships either separated by four dummies (men’s wear: t(191) = −3.7,
between coordination and aesthetic response and between p < .001; women’s wear: t(191) = −3.7, p < .001) or separated
coordination and social impression. We also find support with no intervening dummies (men’s wear: t(196) = −3.7,
for the moderating effect of juxtaposition on the relationship p < .001; women’s wear: t(196) = −2.7, p < .01). Therefore,
between coordination and evaluation. In addition, in testing the effects of coordination and juxtaposition on purchase
H3a and H4a , we provide evidence that aesthetic response intention indicate a negative bias, similar to that found for
and social impression affect evaluation.8 Finally, when aes- product evaluation.
thetic response and social impression are included in the To test H6 (the effect of product evaluation on purchase
ANCOVA as covariates, the interaction effect between coor- intention), we performed an ANCOVA performed on pur-
dination and juxtaposition on product evaluation becomes chase intention by including product evaluation and the
nonsignificant at the .05 level (F(2, 563) = .7), and the param- other dependent variables as covariates. The ANCOVA shows
eter estimates diminish. As such, H5b is supported for men’s that product evaluation has a significant effect on purchase
wear. In contrast, in testing H1b and H2b , we show that the intention (men’s wear: F(1, 562) = 328.7, p < .001, η2 = .369;
interaction effects between coordination and juxtaposition on women’s wear: F(1, 562) = 530.5, p < .001, η2 = .486). Fur-
aesthetic response and social impression are not significant thermore, the parameter estimate representing this effect is
for women’s wear. Therefore, the first condition laid down by positive. Therefore, H6 is supported.
Baron and Kenny (1986) is not met, and H5b is not supported To test H7a (the mediating role of evaluation in the main
for women’s wear. effects of aesthetic response and social impression on pur-
chase intention), we again followed Baron and Kenny’s
Purchase intention (1986) procedure. Aesthetic response and social impression
The univariate ANOVA results on purchase intention are affect product evaluation, therefore meeting the first condi-
similar to those on product evaluation. As Tables 3 and 4 indi- tion. To test the second condition, we conducted an addi-
cate, there is an interaction effect of coordination and juxtapo- tional ANCOVA in which we excluded product evaluation
sition on the evaluation of the target product (men’s wear: F(2, from the previous step-down analysis. As the last columns in
569) = 6.2, p < .01, η2 = .021; women’s wear: F(2, 569) = 3.9, Tables 3 and 4 indicate, aesthetic response has a significant
p < .05, η2 = .014). As Table 2 shows, the mean difference effect on purchase intention (men’s wear: F(1, 563) = 23.0,
in purchase intention between the coordination conditions p < .001, η2 = .039; women’s wear: F(1, 563) = 31.5, p < .001,
is greater when the complementary products are juxtaposed η2 = .053). Social impression also has a significant effect on
(Mdiff = .97) than when they are separated (Mdiff = −.22, .12). purchase intention for men’s wear (F(1, 563) = 3.9, p < .05,
Planned pairwise comparison confirms that the mean dif- η2 = .007) but not for women’s wear (F(1, 563) = 2.5, p > .05).
ference is significant if the products are juxtaposed (men’s Taken together, the additional ANCOVA suggests that aes-
wear: t(195) = 4.0, p < .001; women’s wear: t(195), p < .05). thetic response and social impression as a whole affect pur-
In contrast, the mean difference is not significant when the chase intention. The effect of product evaluation on purchase
products are separated by four dummies (for men’s wear: intention is also confirmed in the testing of H6 . Finally,
t(185) = −.9, p > .05; women’s wear: t(185) = −1.8, p > .05) comparing the ANCOVAs presented in Tables 3 and 4, we
or separated with no intervening dummies (men’s wear: observe that the main effects of aesthetic response and social
t(191) = .5, p > .05; women’s wear: t(191) = −.3, p > .05). impression, which are significant in the additional ANCOVA,
In Fig. 4, we note that the mean differences between become nonsignificant at the .05 level when product eval-
the juxtaposed and separated conditions are much smaller uation is included as a covariate. The parameter estimates
when the complementary products are well coordinated than pertaining to these effects also diminish. Therefore, we con-
when they are poorly coordinated. Planned pairwise compar- clude that H7a is supported.
ison shows that in the well-coordinated condition, there is Finally, we test H7b , which pertains to the mediating
no difference in purchase intention between the juxtaposed role of evaluation in the moderating effect of juxtaposition
and “separated with no intervening dummies” conditions on the relationships between aesthetic response and pur-
(men’s wear: t(189) = 1.2, p > .05; women’s wear: t(189) = .2, chase intention and between social impression and purchase
p > .05), and between the juxtaposed and “separated by four intention. We find that in the additional ANCOVA for pur-
dummies conditions” (for men’s wear: t(190) = .2, p > .05; chase intention, the moderating effect of juxtaposition on the
women’s wear: t(190) = −.2, p > .05). In contrast, planned relationship between aesthetic response and purchase inten-
pairwise comparison applied to the poorly coordinated con- tion is not significant (men’s wear: F(2, 563) = .4, p > .05;
women’s wear: F(2, 563) = .4, p > .05). Similarly, we find
8 According to Baron and Kenny (1986), whether juxtaposition moder-
that the moderating effect of juxtaposition on the relationship
ates the relationships between aesthetic response and product evaluation and
between social impression and purchase intention is not sig-
between social impression and product evaluation is irrelevant for establish- nificant (men’s wear: F(2, 563) = .3, p > .05; women’s wear:
ing the mediating roles of aesthetic response and social impression. F(2, 563) = .5, p > .05). Therefore, Baron and Kenny’s (1986)
246 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

second mediating condition is not fulfilled, and H7b is not on consumers’ willingness to buy durable goods (including
supported. apparel), Sweeney and Soutar (2001) also find that the effect
of the social value dimension is less than that of the other
dimensions (quality, emotion, and price) in terms of impor-
Discussion tance.
Our results also indicate a negative bias in the effects of
Contributions to knowledge and explanation of findings coordination and juxtaposition on product evaluation, which
translates into a negative bias in their effects on purchase
We summarize the results of our analysis in Table 5. Our intention. One possible explanation for this finding is that,
findings support most of our hypotheses and, by extension, in the separate display conditions, exposing consumers to a
the conceptual framework that we use to explain the effects of target item may evoke memories of complementary items
merchandise coordination and juxtaposition on product eval- they saw previously, including those that consumers already
uation and purchase intention. The consistency of our results possess or saw recently. For example, when evaluating a T-
for both men’s casual wear and women’s formal wear also shirt, consumers may attempt to recall products that match
enhances the validity of our proposed framework. In partic- the T-shirt because their assessment of the T-shirt’s utility
ular, we find that the aesthetic response toward and social depends partly on its ability to form a nice-looking outfit
impression of a product grouping play a central role in these with other products. As a result of this visualization, the rat-
effects and that some hypothesized relationships are stronger ing of the T-shirt in the separated conditions differs little
in the juxtaposed condition than in the separated conditions. from its rating in the juxtaposed, well-coordinated condi-
Our results are in line with the view that consumers’ prefer- tion. However, when the T-shirt is juxtaposed with a pair
ences can be influenced by situational or contextual factors, of pants in a poorly coordinated manner, consumers’ nega-
one of which is merchandise presentation in shopping envi- tive response toward the product grouping may suppress their
ronments (Bettman, Luce, & Payne 1998; Hsee & Leclerc recollection of pants that match the T-shirt. Consequently,
1998). the rating is reduced in the juxtaposed, poorly coordinated
Previous studies on merchandise coordination and the condition. This negative bias is consistent with the loss aver-
aesthetics of product design show that a good match of prod- sion assumption stated in prospect theory (i.e., losses loom
ucts in an ensemble enhances the evaluation of the ensem- larger than gains; Kahneman & Tversky 1979) and in line
ble and coordination of features in a product design affects with evidence that negative information is more perceptu-
the aesthetic response toward the product (Bell et al. 1991; ally salient and given more weight than positive informa-
Holbrook 1986; Veryzer & Hutchinson 1998). We extend tion in the evaluation of an object (Peeters & Czapinski
these observations by demonstrating that (1) the coordi- 1990).
nation of a product grouping affects the evaluation of the Furthermore, the interaction effects between coordination
individual product, (2) this coordination effect only takes and juxtaposition on product evaluation and purchase inten-
place in the juxtaposed condition, (3) this effect can be tion help clarify whether a complementary item affects the
explained in part by aesthetic response and social impres- evaluation of the target item when they are displayed sep-
sion, and (4) there is a negative bias in the coordination arately. Our results suggest that a complementary item that
and juxtaposition effects on product evaluation and purchase does not match the target item will not affect the evaluation
intention. of the latter. When both items are not juxtaposed, the evalu-
Our results also show that the effect of aesthetic response ation of the target item is similar in both the poorly and the
on product evaluation is stronger than that of social impres- well-coordinated conditions. In addition, the similar scores
sion. This difference may be related to the process that under- imply that consumers do not recall a poorly matched comple-
lies the formation of aesthetic response and social impression. mentary item they saw previously when evaluating a target
Aesthetic or hedonic value provided by a product is an end item. However, with our results, we are unsure whether con-
in itself (Bell et al. 1991), and a consumer’s response to a sumers think about the complementary item if it matches
product’s aesthetics is generally spontaneous, requiring little the target item. As we noted previously, our participants
cognitive or coding effort. Therefore, consumers are likely may recall complementary products that match the target
to take into account the aesthetics of a product when evalu- item when they inspect the target item in isolation, but these
ating it. In contrast, the social value provided by a product complementary products might include not only the comple-
is extrinsic in nature (i.e., a means to achieve a higher-order mentary item that they saw during the experimental trial, but
goal), and deciding whether a product conveys a desirable also products they possess or had seen before taking part
social impression requires some cognitive effort. Unless con- in the experiment. Therefore, it does not necessarily fol-
sumers are highly involved in their evaluation, they may low that the complementary item influences the target item’s
not deliberate actively on a product’s social acceptability. evaluation.
Thus, social impression may bear less weight than aesthetic Finally, whereas previous studies on aesthetic response
response for many product evaluation and usage situations. to merchandise coordination and product design use pic-
In their study of the effects of perceived value dimensions tures of products as experimental stimuli (Bell et al. 1991;
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 247

Table 5
Summary of hypotheses and results
Hypotheses Relationship Results
H1a Merchandise coordination has a positive effect on consumers’ aesthetic response Supported
toward two complementary products as a whole.
H1b The effect of merchandise coordination on aesthetic response is moderated by Supported for men’s wear but not for women’s wear
merchandise juxtaposition. Specifically, merchandise coordination has a stronger
positive effect on consumers’ aesthetic response when the complementary
products are put together than when they are separated.
H2a Merchandise coordination has a positive effect on the social impression of two Supported
complementary products as a whole.
H2b The effect of merchandise coordination on social impression is moderated by Supported for men’s wear but not for women’s wear
merchandise juxtaposition. Specifically, merchandise coordination has a stronger
positive effect on social impression when the complementary products are put
together than when they are separated.
H3a An aesthetic response toward two complementary products as a whole has a Supported
positive effect on the evaluation of the individual target product.
H3b The effect of the aesthetic response on the evaluation of the target product is Supported for men’s wear but not for women’s wear
moderated by merchandise juxtaposition. Specifically, the aesthetic response has a
stronger positive effect on the evaluation when the complementary products are
put together than when they are separated.
H4a The social impression of two complementary products as a whole has a positive Supported
effect on the evaluation of the individual target product.
H4b The effect of the social impression on the evaluation of the target product is Not supported
moderated by merchandise juxtaposition. Specifically, the social impression has a
stronger positive effect on the evaluation when the complementary products are
put together than when they are separated.
H5a The aesthetic response toward and social impression of two complementary Supported for men’s wear—partial mediation; not
products as a whole totally or partially mediate the main effect of merchandise supported for women’s wear
coordination on the evaluation of the target product.
H5b Aesthetic response and social impression totally or partially mediate the Supported for men’s wear—total mediation; not
moderating effect of merchandise juxtaposition on the relationship between supported for women’s wear
merchandise coordination and the evaluation of the target product.
H6 The evaluation of the target product has a positive effect on the consumers’ Supported
intention to purchase the product.
H7a The evaluation of the target product totally or partially mediates the main effects Supported
of the aesthetic response toward and the social impression of two complementary
products as a whole on the purchase intention of the target product.
H7b The evaluation of the target product totally or partially mediates the moderating Not supported
effects of merchandise juxtaposition on the relationships between aesthetic
response and purchase intention and between social impression and purchase
intention.

Holbrook 1986; Veryzer & Hutchinson 1998), we use ent product categories at retail outlets can improve their
actual products and allow participants to touch them in our products’ evaluation by merely juxtaposing them in a well-
study. Therefore, our findings have higher external validity coordinated manner. Instead, our results demonstrate that
than those of previous studies with regard to store-based even when a target item is juxtaposed with a complemen-
retailing. tary item in a well-coordinated manner, the evaluation of
the target item does not necessarily improve compared with
Managerial implications its evaluation in the two separated conditions. Our results
indicate that poor coordination of an ensemble in a juxta-
Our results show that for the product categories under posed condition depresses product evaluation and purchase
study, merchandise coordination has an effect on product intention. Therefore, retailers and manufacturers must be
evaluation and purchase intention only when complemen- concerned about whether the items juxtaposed as an ensem-
tary products are juxtaposed. The results also suggest that ble are poorly coordinated. This concern cannot be over-
even if two products are put on adjacent sides of a display stated, in that a significant proportion (37 percent) of our
stand (as in the case of the “separated with no interven- participants said that product displays in retail stores were
ing dummies” condition), no coordination effect likely will sometimes poorly coordinated, and only four percent indi-
arise out of product aesthetics. However, this finding does cated they had never seen poorly coordinated displays in retail
not suggest that retailers and manufacturers selling differ- stores.
248 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

Although good coordination does not seem to improve from other objects placed close to it through various semi-
evaluation or purchase intention, our results indicate that otic operations, such as metaphor and metonymy (Thwaites
it improves the aesthetics and social impression of prod- et al. 1994). For example, the “Jungle” line of perfume could
uct displays. The enhancement in aesthetic appeal may be strategically placed among objects or products associ-
lead to some outcomes beneficial to retail management, ated with a tropical jungle or before a photographic jungle
such as an improvement of store image or an increase in background. Such display arrangement may give the per-
store traffic due to the aesthetic attraction. Furthermore, fume connotations of adventure and exoticism, which could
because our results indicate that the aesthetic response and enrich the brand concept and enhance product evaluation.
social impression have positive effects on product evalua- Therefore, additional research should look into the coordi-
tion, apparel retailers and manufacturers should emphasize nation effect via the symbolic meanings that an ensemble
the former two aspects in their displays. Whereas coordi- conveys.
nating the color and style of the product components of an Our study investigates only one moderator of the coor-
ensemble is a means to enhance the aesthetics and social dination effect: merchandise juxtaposition. Further research
impression of a display, other visual elements may also help. might consider other moderators, including personality traits
For example, the color and style of individual products in and other marketing mix variables. Bloch et al. (2003) find
an ensemble and the coordination between the ensemble that people differ in the value they place on product aesthet-
and its background (e.g., wallpaper or posters that form ics as a means of enhancing the quality of life, as well as
the backdrop of a display) may affect the aesthetics and in their acumen for appreciating aesthetics. Considered to be
social impression and, consequently, individual products’ personality traits by Bloch et al. (2003), consumer value and
evaluation. acumen may moderate the effect of an aesthetic response
and, hence, the effect of coordination on product evalua-
Limitations and further research tion. In addition, merchandise coordination and juxtaposition
may interact with price promotion. During a price promotion,
Our explanation of the findings is based partly on the consumers may use the savings from the price discounts to
assumption that consumers tend to recollect complementary purchase other items, particularly those that are complemen-
items that match an item being evaluated. Further research tary to the promoted items (Soman & Cheema 2001). This
should verify the validity of this assumption by using thought- promotional effect on the sales of complementary items will
listing techniques to examine what consumers think or recall be enhanced if the items are placed close to the promoted
when evaluating a product. We find that social impression items in a well-coordinated fashion. Thus, we expect an inter-
plays a relatively minor role in explaining the effect of coordi- action effect among coordination, juxtaposition factors, and
nation on product evaluation, but its role could be investigated price promotion on the purchase intention of complementary
further, in that it might vary with consumers’ shopping goals items.
and usage situations. If consumers become highly involved Finally, future research should study other consequences
in a shopping task, they may put more effort into think- of merchandise coordination and juxtaposition. Merchan-
ing through the consequences of using the target item, in dise juxtaposition may stimulate need recognition or retrieval
which case the relationship between social impression and (Levy & Weitz 1998). The aesthetic appeal of well-
product evaluation may become more salient. To increase coordinated juxtaposed displays could affect consumers’
involvement, researchers might adopt products with higher emotional state (such as pleasure or arousal), which in turn
social risks (e.g., suits) and usage goals with more serious could affect their approach or avoidance behavior (Bitner
consequences (e.g., asking participants to imagine they are 1992; Donovan & Rossiter 1982). Approach or avoidance
considering buying suits that they will wear for job inter- behavior may be manifested as willingness to stay inside a
views). store longer, explore a store’s products, spend more money,
Our mediation analysis shows that coordination has an and so forth. Thus, researchers should examine whether good
“unintended” negative effect on product evaluation, in addi- coordination of products in complementary categories makes
tion to positive effects through aesthetic response and social consumers’ shopping experiences more positive, as well as
impression. This negative effect is counterintuitive, and its whether the pleasure or arousal induced by the coordination
cause is worth investigating. Perhaps a highly consistent affects their shopping behavior.
product grouping is not as arousing as a less consistent one,
and when other things are equal, consumers may like an excit-
ing grouping more than a less exciting one. Similarly, apart Acknowledgments
from aesthetic response and social impression, an ensem-
ble may affect product evaluation in other ways, depending The authors are grateful to the participating retailer for
on the type of display involved. An ensemble could give lending to them the mannequins used in the study. The
its individual constituent items certain symbolic meanings authors also express their appreciation to Teo Ai Ling, Fiona
(McCracken 1988; Thwaites, Davis, & Mules 1994), because Yap, Chen Yanling, and Lee Wen Ting for their assistance
a product on display can derive meaning or connotations in carrying out the study, as well as to Sally Burrows for
S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250 249

editing support. The financial assistance of the Nanyang Donovan, Robert J., & Rossiter, John R. (1982). Store atmosphere:
Technological University’s Nanyang Business School is also An environmental psychology approach. Journal of Retailing, 58(1),
34–57.
gratefully acknowledged. The authors thank the editors,
Feldman, Jack M., & Lynch, John G. (1988). Self-generated valid-
three anonymous reviewers, and Professor Avijit Ghosh for ity and other effects of measurement on belief, attitude, intention,
their useful comments on previous drafts of this manuscript and behavior. Journal of Applied Psychology, 73(August), 421–
and are indebted to Professors Michael Hui, Sharan Jag- 435.
pal, Albert Chau, and Soh Star for their advice on data Fornell, Claes, & Larcker, David F. (1981). Evaluating structural equation
models with unobservable variables and measurement error. Journal
analysis.
of Marketing Research, 18(February), 39–50.
Grewal, Dhruv, Kent, B., & Krishnan, Monroe R. (1998). The effects
of price-comparison advertising on buyers’ perceptions of acquisition
value, transaction value, and behavioral intentions. Journal of Mar-
keting, 62(2), 46–59.
References Helson, H. (1964). Adaptation level theory. New York: Harper and Row.
Herr, Paul M., Kardes, Frank R., & Kim, John. (1991). Effects of
Anderson, James C., & Gerbing, David W. (1988). Structural equation word-of-mouth and product-attribute information on persuasion: An
modeling in practice: A review and recommended two-step approach. accessibility-diagnosticity perspective. Journal of Consumer Research,
Psychological Bulletin, 103(3), 411–423. 17(March), 454–462.
Ashcraft, Mark H. (1989). Human memory and cognition. New York: Holbrook, Morris B. (1986). Aims, concepts, and methods for the
HarperCollins Publishers. representation of individual differences in esthetic responses to
Baker, Julie, Levy, Michael, & Grewal, Dhruv. (1992). An experimental design features. Journal of Consumer Research, 13(December), 337–
approach to making retail store environmental decisions. Journal of 347.
Retailing, 68(4), 445–460. Hsee, Christopher, & Leclerc, France. (1998). Will products look more
Bagozzi, Richard P. (1982). A field investigation of causal relations attractive when presented separately or together? Journal of Consumer
among cognitions, affect, intentions, and behavior. Journal of Mar- Research, 25(September), 175–186.
keting Research, 19(November), 562–584. Kahneman, Daniel, & Tversky, Amos. (1979). Prospect theory: An
Bagozzi, Richard P., & Yi, Youjae. (1988). On the evaluation of structural analysis of decisions under risk. Econometrica, 43(March), 263–
equation models. Journal of the Academy of Marketing Science, 16(1), 291.
74–94. Lennon, Sharron J. (1990). Effects of clothing attractiveness on percep-
Baron, Reuben M., & Kenny, David A. (1986). The moderator-mediator tions. Home Economics Research Journal, 18(3), 303–310.
variable distinction in social psychological research: Conceptual, Levy, Michael, & Weitz, Barton. (1998). Retailing management (3rd ed.).
strategic, and statistical considerations. Journal of Personality and Boston, MA: Irwin McGraw-Hill.
Social Psychology, 51(6), 1173–1182. McCracken, G. (1988). Diderot unities and the diderot effect. In Culture
Beardsley, M. C. (1981). Aesthetics: Problems in the philosophy of crit- and consumption. Bloomington and Indianapolis: Indiana University
icism (2nd ed.). Indianapolis, IN: Haskett. Press, pp. 118–129.
Bell, Stephen S., Holbrook, Morris B., & Solomon, Michael R. (1991). Peeters, G., & Czapinski, J. (1990). Positive-negative asymmetry in evalu-
Combining esthetic and social value to explain preferences for ations: The distinction between affective and informational negativity
product styles with the incorporation of personality and ensem- effect. European Review of Social Psychology, 1, 33–60.
ble effect. Journal of Social Behavior and Personality, 6, 243– Petty, Richard, & Cacioppo, John T. (1986). The elaboration likelihood
273. model of persuasion. Advances in Experimental Social Psychology,
Berman, Barry, & Evans, Joel R. (1995). Retail management: A strategic 19, 123–205.
approach. Englewood Cliffs, NJ: Prentice-Hall. Posner, M. I. (1978). Chronometric explorations of mind. Hillsdale, NJ:
Bettman, James, Luce, May Frances, & Payne, John W. (1998). Con- Lawrence Erlbaum Associated.
structive consumer choice processes. Journal of Consumer Research, Reed, Americus, II. (2002). Social identity as a useful perspective for self-
25(3), 187–217. concept-based consumer research. Psychology & Marketing, 19(3),
Bitner, Mary J. (1992). Servicescapes: The impact of physical surround- 235–266.
ings on customer and employees. Journal of Marketing, 56(April), Roy, J. (1958). Step down procedure in multivariate analysis. Annals of
57–71. Mathematical Statistics, 29(December), 1177–1187.
Bloch, Peter H., Brunel, Frederic F., & Arnold, Todd J. (2003). Individ- Russell, Gary J., Ratneshwar, S., Shocker, Allan D., Bell, David, Boda-
ual differences in the centrality of visual product aesthetics: Concept pati, Anand, Degeratu, Alex, Hildebrandt, Lutz, Kim, Namwoon,
and measurement. Journal of Consumer Research, 29(March), 551– Ramaswami, S., & Shankar, Venkatash H. (1999). Multiple-category
565. decision-making: Review and synthesis. Marketing Letters, 10(4),
Bray, James H., & Maxwell, Scott E. (1985). Multivariate analysis of 319–332.
variance. Beverly Hills, CA: SAGE Publications. Soman, Dilip, & Cheema, Amar. (2001). The effect of windfall gains on
Cooper, L. A., & Shepard, R. N. (1973). Chronometric studies of the the sunk cost effect. Marketing Letters, 12(1), 49–60.
rotation of mental images. In W. G. Chase (Ed.), Visual information Sweeney, Jillian C., & Soutar, Geoffrey N. (2001). Consumer perceived
processing (pp. 75–176). New York: Academic Press. value: The development of a multiple item scale. Journal of Retailing,
Dabholkar, Pratibha. (1994). Incorporating choice into an attitudinal 77(2), 203–220.
framework: Analyzing models of mental comparison processes. Jour- Tabachnick, Barbara G., & Fidell, Linda S. (2001). Using multivariate
nal of Consumer Research, 21(June), 100. statistics (4th ed.). Needham Heights, MA: Allyn & Bacon.
Dahl, Darren W., Chattopadhyay, Amitava, & Gorn, Gerald J. (1999). Thwaites, Tony, Davis, Lloyd, & Mules, Warwick. (1994). Tools for
The use of visual mental imagery in new product design. Journal of cultural studies: An introduction. Melbourne: MacMillan Education
Marketing Research, 36(February), 18–28. Australia Pty Ltd.
Diamantopoulos, Adamantios, & Siguaw, Judy A. (2000). Intro- Veryzer, Robert W. (1993). Aesthetic response and the influence of
ducing LISREL: A guide for the uninitiated. London: SAGE design principles on product preferences. In X. X. Leigh McAl-
Publications. ister & Michael L. Rothschild (Eds.), Advances in consumer
250 S.Y. Lam, A. Mukherjee / Journal of Retailing 81 (3, 2005) 231–250

research (pp. 224–228). Provo, UT: Association of Consumer Wagner, Janet. (1999). Aesthetic value. In Morris B. Holbrook (Ed.), Con-
Research. sumer value: A framework for analysis and research (pp. 126–146).
Veryzer, Robert W., & Hutchinson, Wesley J. (1998). The influence London: Routledge.
of unity and prototypicality on aesthetic responses to new product Yi, Youjae. (1990). Cognitive and affective priming effects of the context
designs. Journal of Consumer Research, 24(March), 374–394. for print advertisements. Journal of Advertising, 19(2), 40–48.

You might also like