Content uploaded by David Hardesty
Author content
All content in this area was uploaded by David Hardesty on Mar 05, 2014
Content may be subject to copyright.
Color and shopping intentions:
The intervening effect of price fairness and perceived affect
$
Barry J. Babin
a,
*, David M. Hardesty
b
, Tracy A. Suter
c
a
College of Business Administration, 306 JGH, University of Southern Mississippi, Hattiesburg, MS 39406-5091, USA
b
School of Business Administration, 523D Jenkins Building, University of Miami, Coral Gables, FL 33124-6554, USA
c
College of Business Administration, 370 North Hall, Oklahoma State University, Tulsa, OK 74106-0700, USA
Abstract
How do consumers react to various color, lighting, and price point combinations? The results described in this article depict varying
consumer reactions with the three-way congruence between a store’s environmental cues, consumers’ cognitive categories representing
known store types, and salient situational shopping motivations. For fashion-oriented stores, blue interiors are associated with more favorable
evaluations, marginally greater excitement, higher store patronage intentions, and higher purchase intentions than are orange interiors.
However, the results change substantially when the effect of lighting in combination with color is considered. The use of soft lights with an
orange interior generally nullifies the ill effects of orange and produces the highest level of perceived price fairness while controlling for
price. Additionally, the results suggest that the effects of environmental and price cues are mediated by consumers’ cognitive and affective
associations.
D2003 Elsevier Science Inc. All rights reserved.
Keywords: Atmosphere; Color; Retail patronage; Shopping emotion
1. Introduction
Consumer reactions to retail concepts are shaped largely
by atmospherics created by altering key in-store elements
(Kotler, 1974). Designers can manipulate cues such as
climate, music, scent, employee appearance, etc., in an
effort to stimulate positive consumer reactions (Baker
et al., 1994; Darden and Babin, 1994; Spangenberg et al.,
1996). Changes in physical store characteristics can alter
consumers’ mood, perceptions, shopping time, and satisfac-
tion with a retailer (Babin and Darden, 1996; Dawson et al.,
1990; Hui et al., 1997), among other things.
Surprisingly little research addresses how a store’s pre-
dominant color affects consumer reactions (Crowley, 1993).
This paper presents research addressing this topic. Follow-
ing from previous research describing how lighting and
background music alter a consumer’s perceptions and reac-
tions to a store design (Baker et al., 1992), this research
examines how color perceptions, alone and in combination
with store lighting, influence patronage intentions. Addi-
tionally, the role of price is considered since price helps
establish the congruence of a store environment with known
store types (Baker et al., 1994). A conceptual case is
presented and empirical results support that these effects
are experienced indirectly through affective and cognitive
consequences of environmental changes.
The hope is to shed light on several important contribu-
tions. First, the research addresses the importance of color
and lighting in retail atmospherics and shopping reactions.
Second, the research furthers evidence suggesting the
important role of environmental affect in mediating relation-
ships between store characteristics and consumer reactions.
Third, the study design furthers our knowledge of how
consumers cognitively store and respond to color, lights,
and retail pricing. Thus, the paper has both practical and
theoretical potential.
2. Conceptual background
Atmospherics research falls generally into environmental
psychology (Mehrabian and Russell, 1974). As such, atmo-
0148-2963/03/$ – see front matter D2003 Elsevier Science Inc. All rights reserved.
doi:10.1016/S0148-2963(01)00246-6
$
This manuscript was completed while the second and third authors
were on faculty at the University of Southern Mississippi.
* Corresponding author. Tel.: +1-601-266-4627; fax: +1-601-266-
4630.
E-mail address: barry.babin@usm.edu (B.J. Babin).
Journal of Business Research 56 (2003) 541 – 551
spheric models generally make stimulus–organism–res-
ponse (S–O–R) type predictions. Store cues cause specific
cognitive and affective reactions, and these reactions modify
shopping behavior (see also Bitner, 1992; Wakefield and
Baker, 1998). Likewise, the framework is consistent with
the consumption–emotion– value paradigm (Holbrook,
1986) in that it emphasizes the important, intervening role
played by affective reactions.
2.1. Color
Color’s effects on human performance and cognitive
interpretation provide important evidence suggesting poten-
tial consumer reactions (e.g., Jacobs and Suess, 1975;
Wexner, 1954). Crowley (1993) reviews this literature and
concludes that color influences both consumers’ evaluation-
related affect (affective tone) and activation-related affect
(arousal). A color’s hue or gradation is determined by its
wavelength. Short wavelengths are associated with ‘cool’
colors with violet being most extreme followed by blue.
Long wavelengths are associated with ‘warm’ colors with
red being most extreme followed by orange. A consistent
finding in this literature is that short (long) wavelength
colors are preferred (not preferred) leading to a linear
association between affective tone and wavelength. So, blue
environments generally evoke better feelings than do orange
environments (Valdez and Mehrabian, 1994).
Retailing research is generally consistent with these
results. Experimental research suggests that cool-colored
store environments are preferred over warm-colored store
environments (Bellizi et al., 1983; Crowley, 1993).In
addition, experimental research using a hypothetical tele-
vision purchase suggests that a blue background can
decrease the likelihood of postponing purchase compared
to a red background (Bellizi and Hite, 1992). Thus, it is
generally expected that violet/blue interiors will produce
higher levels of positive affective tone and increased pur-
chase intentions than will red/orange interiors.
Color’s influence on activation-related affect is not as
straightforward. Empirical results suggest that a U-shaped
relationship exists such that relatively extreme wavelengths
evoke greater arousal (Wilson, 1966).Red/orangeand
violet/blue, being at opposite ends of the U, evoke similar
levels of arousal. One explanation for this result is that
responses to color are more learned or instinctive rather than
physiological. More extreme wavelengths are speculated as
being more associated with danger and therefore, they evoke
greater activation (Wilson, 1966).
Although some research suggests red environments as
being more arousing than blue (Valdez and Mehrabian,
1994), results from retail studies of activation-related emo-
tions such as excitement are more mixed than for affective
tone. However, previous research may not have taken into
account the relative location of each color on the U-shaped
wavelength–activation curve. Crowley (1993) controlled for
this factor by manipulating the colors of a hypothetical
furniture store along different wavelengths at matching
points on the U (e.g., violet – red). Subjects in that experi-
ment did not display a linear trend in their ratings of
activation-related affect. Thus, store colors matching vertical
location on the U may show little or no difference of affect.
2.2. Congruent cues and consumer categorization
Retailing research has focused predominantly on color in
isolation of other factors. However, consumers do not pro-
cess environmental characteristics piecemeal (Ward et al.,
1992). Rather, combinations of ambient characteristics affect
how consumers react to a store concept. A store described as
having a combination of bright, fluorescent lights (soft,
incandescent lights) and popular (classical) background
music causes consumer reactions consistent with a discount
(prestige) image (Baker et al., 1994). Therefore, the possibil-
ity exists that color may interact with other ambient charac-
teristics and may alter consumer reactions to a store concept.
Like other categories, consumers’ shopping experiences
create cognitive categories representing store types (Ander-
son and Klatzky, 1987; Babin et al., 1995). Certain combi-
nations of lighting and color fit better (are assimilated more
easily) into specific categories. The active category makes
category consistent information salient — available for use
in decision making (Holland et al., 1989). Thus, a consumer
seldom considers requesting the sommelier (wine steward)
if seated in a brightly lit, red restaurant. However, the same
consumer might request a paper napkin in such an envir-
onment. Additionally, these cognitive processes help deter-
mine one’s feelings and evaluations (Babin et al., 1995).A
concept typified by specific environmental cues can cause a
favorable or unfavorable reaction depending on its congru-
ence with specific shopping motivations.
Thus, as a combination of classical music and soft lights
leads consumers to expect higher prices (Baker et al., 1994),
lights may moderate color’s effect. Taken separately,
research suggests that bright fluorescent (soft) lights and
warm (cool) colors are more consistent with a discount
(prestige) store concept (Baker et al., 1992; Bellizi and Hite,
1992; Schlosser, 1998).
The current study focuses on women’s fashion retailing
and thus involves quite different motivations. Given fash-
ion’s close relationship with self-image (Sproles, 1979),
clothing appeals strongly to one’s prestige sensitivity (Lich-
tenstein et al., 1993).Schlosser (1998) found socially
oriented products like clothing were evaluated more favor-
ably in a store described with a prestige atmosphere than
they were in a store described with a discount atmosphere.
In contrast, ratings of utilitarian products (e.g., vacuum
cleaner) were not influenced to the same degree by the
store categorization. In part, this congruence issue explains
why many consumers patronize discounters like Wal-mart
for many goods, but do not consider it a desirable place to
shop for their own clothes. Thus, to the extent that orange
and bright lights are most associated with a discount store
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551542
concept, fashion consumers (typical consumers involved in
clothing purchases) will express lower intentions to patron-
ize it for fashions. Whereas the orange – bright combination
is commonly experienced among discounters (cf. Kmart),
orange–soft is not; therefore it may ameliorate some of the
negative reactions associated with orange.
3. Specific predictions
We expect that changes in color will affect shopping
intentions through the associated cognitive and affective
reactions. More specifically, intentions will be operational-
ized two ways, both of keen interest to retailers: (1)
patronage intentions represent the likelihood of shopping
in a store and (2) purchase intentions represent the like-
lihood of purchasing an item. Consistent with previous
research (Crowley, 1993; Donovan and Rossiter, 1982),
two dimensions of affect are used: (1) evaluation or affective
tone which accounts for the degree of liking or how pleased
one is with the environment and (2) excitement or positive
arousal which is particularly relevant for fashion or mall
shopping contexts (Babin et al., 1998; Wakefield and Baker,
1998). Additionally, given the role that price plays in
establishing congruence with shopping motivations, effects
related to price are examined.
3.1. Hypothesis 1
Hypothesis 1 addresses how color, the interaction
between color and light, and price influence consumer affect
in the form of two key dependent variables: evaluation and
excitement. Following from Crowley (1993), and from
consumers’ learned responses to colors, a blue store con-
cept, blue having a longer wavelength, should be evaluated
more favorably than an orange store. However, orange and
blue should evoke similar levels of activation-related emo-
tion due to their location on the U-shaped arousal – wave-
length curve. Thus, to the extent that excitement represents
activation, no significant linear effect between color and
excitement is predicted.
Hypothesis 1a: A retail store described as having a blue
store interior is associated with a more positive evaluation
than is a store described as having an orange store interior.
Further, given the likelihood that lighting–color combi-
nations will vary consumer reactions based on their con-
gruency with known store types and shopping motivations,
significant Color Lights interactions are expected. In
particular, the more (less) a store’s lights and color are
consistent with discount (prestige) perceptions, fashion
consumers, motivated more by prestige and self-esteem,
will evaluate it less favorably and be less excited about
shopping for fashions in it. Among such motivated con-
sumers, a discount store may be appropriate for buying
detergent but it is less appropriate for buying a statement
about one’s self. An orange (blue) store with bright (soft)
lights being consistent with a discount (prestige) image, the
following hypothesis is offered:
Hypothesis 1b: A significant Lights Color interaction is
expected such that any positive (negative) effects of blue
(orange) are diminished by bright (soft) lights.
Low prices, controlling for other factors, have the poten-
tial to create positive affect through increased transaction
utility (Schindler, 1989; Thaler, 1985). Consumers are
believed to have increased motivation beyond just the
economic value of any money saved (Schindler, 1989).
Examples where increased motivation has an ability to
create positive affect are frequent flyer programs and
blue–light specials, among others. Therefore, based on the
above findings, we expect the following:
Hypothesis 1c: Price relates negatively to consumer evalu-
ation and excitement.
3.2. Hypothesis 2
All things being equal, consumers view lower prices as
more fair than higher prices (Kalapurakal et al., 1991).
However, price fairness perceptions are also affected by
congruence. Thus, a price perceived as fair in one type of
environment is perceived as unfair in another (Martins,
1995). Previous research suggests that a blue color and
other characteristics consistent with prestige can enhance
the desirability of goods (Baker et al., 1994; Bellizi and
Hite, 1992; Middlestadt, 1990). For example, subjects rating
furniture against a red background considered it signific-
antly more ‘‘out of style’’ than did those rating it against a
blue background. Therefore, to the extent that ambient
characteristics influence desirability, price fairness will vary.
Hypothesis 2a: Price fairness perceptions are higher in a
blue store than in an orange store.
Additionally, those combinations that frame a purchase
along a discount–prestige continuum elicit consumers’
learned price expectations leading to what some may say
are counterintuitive predictions. To illustrate this point, an
item priced at US$100 at Nieman Marcus may be perceived
as priced fairly, while the same item priced at US$100 at
Kmart may not be viewed as priced fairly. Holding price
constant, consumers will rate a price as less (more) fair when
the atmospheric cues are congruent with low (high) price
expectations. Thus, the following hypotheses are offered:
Hypothesis 2b: Color and lights interact to affect price
fairness perceptions. For example, an orange (blue) store
with bright (soft) lights produces low (high) price fairness
perceptions.
Hypothesis 2c: Price relates negatively to price fairness
perceptions.
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551 543
3.3. Hypothesis 3
The third set of hypotheses addresses direct effects on
patronage and purchase intentions. Research supports rela-
tionships between store-associated affect and desire to
remain, associate, spend money, and return to a store (Babin
and Darden, 1996; Baker et al., 1992; Donovan and
Rossiter, 1982; Wakefield and Baker, 1998). Factors like
color and lighting lower or increase patronage intentions
through their effect on positive affect. Therefore, consumers
have a primary reaction to colors in terms of their effect on
price fairness and affect. These relationships facilitate the
eventual effect on shopping and purchase intentions. Addi-
tionally, consumers are likely to prefer shopping and be
more willing to buy where prices are perceived as fairer.
Hypothesis 3a: Evaluation relates positively to patronage
intentions and purchase intentions.
Hypothesis 3b: Excitement relates positively to patronage
intentions and purchase intentions.
Hypothesis 3c: Price fairness perceptions relate positively
to patronage intentions and purchase intentions.
3.4. Hypothesis 4
To complete the empirical examination of mediation
(Baron and Kenny, 1986), significant main effects and
interactions on patronage and purchase intentions should
be significantly attenuated by including evaluation, excite-
ment, and price fairness perceptions as covariates.
Hypothesis 4: Evaluation, excitement, and price fairness
perceptions mediate relationships between price, color, and
lights and patronage intentions and purchase intentions.
No main effects are predicted for lighting as there is little
guidance from the literature. The work in organizational
behavior shows that employee performance and satisfaction
can be influenced with changes in lighting, but, that the
changes are usually experienced over fairly large light
changes (Baron, 1990). One study examined the effect of
retail lighting empirically (Areni and Kim, 1994).Two
levels of lighting were varied in an actual store setting.
Lighting brightness showed no effects on arousal, purchas-
ing behavior, or time spent shopping. The only affect
appeared to be based on the number of items handled. Thus,
lighting is hypothesized to influence reactions only in
combination with color.
4. Research methods
Results were obtained from a 2 22 between-subjects
design. Treatments included color (orange vs. blue), lights
(bright vs. soft), and item price (US$59.95 vs. US$149.95).
Following previous atmospherics research studying effects
of ambient cues such as music and lighting (cf. Baker et al.,
1992, 1994; Schlosser, 1998), a scenario approach was used
to execute the experimental design. While scenarios are
certainly open to criticism for a lack of realism, they are
useful from a number of perspectives. Here, they allow an
examination of how consumers respond to the cognitive
representation of cues and cue combinations. This allows a
comparison of the effects of actual color presentation results
with those found here. By comparing results, inferences can
be made with respect to whether color effects are solely
physiological or psychological. In addition, store designs
must be concept tested. At the very least, a scenario
approach allows a valid examination of consumer reactions
to store concepts. Third, given the large body of evidence
suggesting that people categorize things based on associated
characteristics (see Anderson and Klatzky, 1987 for a
review), it is quite likely that the same cognitive concepts
and schema-based affect is evoked. Fourth, it offers a
pragmatic way to control for factors that would be difficult
to control for in a real store environment. While scenarios
may not be perfect, they are useful.
The scenario described a hypothetical retail store con-
sidering a location in the local area. Subjects were asked to
read the description carefully and imagine visiting the store
for the first time. Color and lights were manipulated
between subjects within this description. Each scenario
concluded with a detailed description of a fashionable
sweater taken from a Land’s End catalog. The sweater
was also selected to have general appeal and was styled to
be ‘‘acceptable for work or play.’’ These price points —
US$59.95 or US$149.95 — were picked based on study
pretests where female respondents were asked to give the
lowest and highest prices for which they may find a sweater
like the one described above. Following exposure to this
scenario, subjects responded to a questionnaire containing
relevant measures.
4.1. Sample
Data were collected by female, student interviewers who
received course credit for their participation. Guidelines for
respondent eligibility were provided to insure a varied
sample and to exclude participation by family members
and other students (cf. Mick, 1996). Interviewers were
required to (1) complete one questionnaire themselves and
(2) obtain two responses one from a female aged between 30
and 40 years, and another from a female older than 40. The
study context lent itself best to female subjects based on
their familiarity with shopping for women’s fashions. Sub-
jects’ first names and telephone numbers were obtained at
the end of each survey to verify from a subsample of
respondents that the data were collected as reported.
This process resulted in data from 209 females from the
university community. The average age was 33.2 years
(S.D. = 12.8). Reported household income varied with
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551544
23% listing less than US$20,000 annually, 25% checked the
US$20,000 to US$40,000 category, 24% checked the
US$40,000 to US$60,000 category, 10% checked the
US$60,000 to US$75,000 category, and 18% listing over
US$75,000 in annual income. Neither age nor income
influenced any hypothesized relationships. Therefore,
results reported below do not include age or income as
covariates for the sake of simplicity. Subjects were assigned
randomly to one of eight experimental conditions. Cell sizes
ranged from 18 to 28 across the conditions.
4.2. Measures
After a few instructions and exposure to the scenario,
subjects responded to bipolar semantic differential and
Likert scales. The direction of the items was varied to
inhibit response bias. The measures are derived from those
used in related research and adjusted to fit this specific
context (e.g., Richins, 1997; Babin et al., 1998; Martins,
1995). Excitement was operationalized as the summation
of 3 nine-point scale (ranging from ‘‘would not feel at
all’’ to ‘‘would feel very much’’) items indicating how
exciting, thrilling, and encouraged subjects would feel
when shopping in the store described. Evaluation was
measured using the sum of four bipolar semantic differ-
entials (e.g., bad–good, favorable–unfavorable, not like-
able–likeable, not acceptable–acceptable). Price fairness
was assessed as the sum of 2 six-point scale statements.
The items included ‘‘How fair/unfair do you think the
price offered to consumers is?’’ and ‘‘Overall, how fair is
its [the items] price?’’ Each was anchored with 1
(extremely unfair) and 6 (extremely fair). Store patronage
was measured via the sum of three items. Specifically,
subjects responded to an item such as, ‘‘How likely would
it be that you would go into this store? ... ____%.’’
Finally, intention to purchase the sweater was operation-
alized as the sum of 4 six-point semantic differentials
assessing how likely she would be to purchase that
particular item given the need (e.g., very likely– very
unlikely, improbable– probable ...). Both intention meas-
ures were included given the practical implications asso-
ciated with each. Success comes not simply from winning
shoppers, but it comes from converting large numbers of
shoppers into buyers.
Coefficient alpha estimates of internal consistency for
each multi-item dependent measure were as follows:
excitement, .81; evaluation, .92; patronage intentions, .91;
purchase intentions, .91. Additionally, the bivariate correla-
tion for the two fairness items was .67. Bivariate correla-
tions among the five dependent measures range from .23
(excitement–price fairness) to .63 (purchase intentions –
price fairness). Therefore, more complex discriminant
validity tests were not deemed necessary due to the
relatively high degree of internal consistency among the
items and relatively low intercorrelations between the
dependent variables.
4.3. Manipulation checks
Experimental manipulations appear effective. Although
the color manipulation seems quite obvious, its effective-
ness was examined by results of an unaided recall test.
Subjects were asked to recall a description of the store after
exposure to dependent measures. Color appeared quite
salient with over 80% of respondents listing blue or orange
correctly. No misidentifications were made (i.e., some did
not mention a color). The lighting manipulation was
checked by responses to a single six-point Likert item
capturing agreement with the belief ‘‘the store had very
bright lights.’’ Results showed a significant variation in the
expected direction [ F(1,206) = 126.70; P< .0001; 4.90 in
the bright condition vs. 2.69 in the soft condition]. Like-
wise, the price manipulation was checked with a single item
expressing agreement with the belief that ‘‘the sweater had a
very low price.’’ Results showed a significant variation in
the expected direction [ F(1,206) = 41.80; P< .0001; 3.10 in
the low price condition vs. 1.97 in the high price condition].
Although not of direct relevance, the color by light com-
bination appeared to affect perceptions of the store. The
interaction of color and lights predicted a measure differ-
entiating prestige from discount stores significantly
[F(1,204) = 4.66; P=.032]. Interestingly, the effects of color
and lighting alone are not as strong [ F(1,204) = 1.83 and
F(1,204)=.35, respectively].
5. Results
Tab le 1 displays descriptive statistics. A multivariate
analysis was undertaken before testing specific relation-
ships. This model used all experimental variables to
predict each composite dependent variable (affective eval-
uation, excitement, store patronage intentions, and item
purchase intentions) within a full-factorial design. Its
results suggest significant multivariate F(based on Wilkes
7) statistics for price [ F(5,188) = 6.30, P< .001], color
[F(5,188) = 2.30, P< .05], and the Color Lights interac-
tion [ F(5,188) = 2.00, P< .10]. The results presented below
reflect the univariate, full-factorial ANOVA and ANCOVA
analyses that followed (all models contained all main
effects and interactions as predictors). Note that tests
linking independent variables to mediators necessary to
establish mediation are also implicit in the models testing
Hypothesis 1 and Hypothesis 2.
5.1. Tests of hypotheses
5.1.1. Hypothesis 1
Two separate full-factorial ANOVA models were used to
test hypothesized relationships between ambient cues and
price and both affective dependent variables, evaluation and
excitement. The first predicted affective evaluation and the
second predicted excitement. Both univariate model F
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551 545
statistics are significant [ F(7,183) = 4.00, P< .001 and
F(7,190) = 1.90, P< .07, respectively).
As predicted, a significant color main effect on evalu-
ation was found ( F= 5.94, P< .05). Subjects rated a blue
store with a mean evaluation of 18.79 compared to 17.09
for the orange condition. Following Crowley (1993),we
did not expect (or hypothesize) a strong, linear main effect
of color on excitement. The results are fairly consistent
with her findings in that the main effect of color on
excitement (x
¯
blue
= 10.65 vs. x
¯
orange
= 8.98) is less strong
(Z
2
= 0.027 for evaluation vs. Z
2
= 0.014 for excitement) but
significant at the .10 level ( F= 2.77, P< .10). These find-
ings are generally consistent with Hypothesis 1a. Hypo-
thesis 1c also predicts main effects and is partially
supported in that lower prices evoked significantly high-
er evaluations ( F= 5.71, P< .05) than did higher prices
(x
¯
US$59.95
= 18.82, x
¯
US$149.95
= 17.18), but the price main
effect on excitemen t is insignificant ( F= 0.37, P> .10).
Consistent with Hypothesis 1b, the Color Lights inter-
action affected evaluation significantly ( F= 10.51, P< .01).
The interaction is displayed in Fig. 1(A). A blue store
interior is associated with more favorable affect in the bright
lights condition (x
¯
blue – bright
= 19.80) relative to soft lights
(x
¯
blue – soft
= 17.72). In contrast, orange store interiors
resulted in greater positive affect in the soft lights condition
(x
¯
orange – soft
= 18.57) relative to bright lights (x
¯
orange –
bright
= 15.96). A similar pattern of results is observed in
the ANOVA model predicting excitement. The model pro-
duced a marginally significant Color Lights interaction
(F= 2.67, P< .10) for excitement. This interaction is shown
in Fig. 1(B). Whereas bright and soft lights resulted in
similar excitement levels for a blue store interior (x
¯
blue –
bright
= 10.65, x
¯
blue – soft
= 10.53), with an orange interior, soft
lights resulted in significantly greater excitement (x
¯
orange –
soft
= 10.59) relative to bright lights (x
¯
orange – bright
= 7.97).
Although not hypothesized, the ANOVA model predict-
ing evaluation is qualified by a significant Color Price
interaction ( F=4.16, P< .05). This effect suggests that
similar evaluation scores are evoked across price in the
orange condition (x
¯
US$59.95 – orange
= 17.10, x
¯
US$149.95 –
orange
= 17.00), but not in the blue condition (x
¯
US$59.95 –
blue
= 20.10, x
¯
US$149.95 – blue
= 17.30). Additionally, a margin-
ally significant three-way interaction e merged ( F=3.02,
P< .10) for excitement. A post hoc analysis suggested that
this is primarily attributable to ratings in both the orange,
soft, and high price condition and the blue, soft, and low
price condition. In these conditions, subjects reacted with
significantly (t
199
= 2.48; P< .01) greater excitement (11.70)
than did subjects in other conditions (9.30).
5.1.2. Hypothesis 2
A third full-factorial ANOVA tests relationships between
ambient cues and perceptions of price fairness [model
F(7,179) = 3.36, P< .01]. The results suggest a nonsignifi-
cant main effect of color on fairness ( F= 1.00, P>.10). The
means are available in Table 1. These results fail to sup-
port Hypothesis 2a. However, consistent with Hypothesis
2b, a significant Color Lights interaction is observed
(F= 13.86, P< .01). Fig. 1(C) suggests the nature of this
interaction. For a blue store interior, bright lights resulted in
higher price fairness perceptions (x
¯
blue – bright
= 7.07) than did
soft lights (x
¯
blue – soft
= 6.60). For the orange store interior,
soft lights resulted in higher price fairness perceptions
(x
¯
orange – soft
= 7.67) than did bright lights (x
¯
orange–
bright
= 5.76). These results support Hypothesis 2b.
Consistent with Hypothesis 2c, a significant main effect
of price on price fairness is observed ( F= 37.96, P< .001)
and is in the observed direction (x
¯
US$59.95
= 7.61,
x
¯
US$149.95
= 5.71). Although not hypothesized, results sug-
gest a significant main effect of lights on price fairness
(F= 4.89, P< .05). Cell means suggest greater perceived
price fairness in the soft condition (7.00) as opposed to the
bright condition (6.32).
5.1.3. Hypothesis 3
Hypothesis 3 was tested using two separate full-factorial
ANCOVA models. The first ANCOVA predicts store pat-
ronage intentions using each treatment as a main effect, all
four interaction terms, and subjects’ perceived evaluations,
excitement, and price fairness as covariates. The second is
Table 1
Means and standard deviations
Independent variables Means (S.D.)
Evaluation Excitement Fairness Store patronage Purchase intentions
Color
Orange 17.09 (4.97) 8.98 (5.50) 6.44 (2.80) 131.80 (55.20) 13.18 (5.72)
Blue 18.79 (5.13) 10.65 (6.08) 6.88 (2.48) 153.80 (55.14) 15.24 (6.46)
Lights
Bright 17.90 (5.38) 9.23 (6.18) 6.32 (2.62) 140.35 (56.56) 14.40 (6.28)
Soft 18.10 (4.81) 10.65 (5.37) 7.00 (2.61) 144.73 (56.17) 14.08 (6.11)
Price
US$59.95 18.82 (4.87) 10.00 (6.19) 7.61 (2.45) 152.17 (56.80) 16.12 (5.79)
US$149.95 17.18 (5.23) 9.78 (5.51) 5.71 (2.44) 132.58 (54.30) 12.41 (6.04)
Overall mean 17.99 (5.11) 9.82 (5.85) 6.71 (2.63) 142.32 (56.25) 14.25 (6.19)
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551546
the same with the exception that it predicts product purchase
intentions and includes store patronage as an additional
covariate.
The results from the lower portion of Table 2 show
significant effects of evaluation ( F=104.4, P< .001,
b= 0.65) and excitemen t ( F= 15.09, P< .001, b= 0.22) on
store patronage intentions. Conversely, price fairness
(F= 0.30, P>.10, b= 0.06) is not related significantly to
store patronage intentions. Therefore, Hypotheses 3a and 3b
are supported with respect to patronage intentions.
Table 2 suggests that product purchase intentions are
significantly related to evaluation ( F= 4.30, P<.05,
Fig. 1. Means of interaction effects.
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551 547
b= 0.24), excitement ( F= 4.34, P< .05, b= 0.15), and price
fairness perceptions ( F= 35.49, P< .001, b= 0.45) pos-
itively. The relationship between store and purchase
intentions was insignificant. These results support Hypoth-
eses 3a, 3b, and 3c with respect to product intentions.
5.1.4. Hypothesis 4
Procedures suggested by Baron and Kenny (1986) were
used to test for mediation effects of evaluation, excitement,
and price fairness. Having established significant relation-
ships between predictors and mediators, and between the
mediators and the dependent variables above, the effect of
adding the three proposed mediating constructs on the
predictive power of hypothesized relationships is analyzed.
Table 2 displays ANOVA results showing the effects of the
predictors on the dependent variables in both the presence
and absence of the mediators.
The ‘‘nonmediated’’ models suggest a pattern of signifi-
cant main effects and interactions on patronage and pur-
chase intentions similar to that observed in the models
testing Hypothesis 1 and Hypothesis 2. The model F
statistics are both significant [ F(7,172) = 2.90, P< .01 and
F(7,172) = 4.20, P<.001, respectively). For example, color
and price displayed significant main effects on both inten-
tions measures (see Table 2 for the Fstatistics; P< .05), and
a significant Color Lights interaction is also observed.
These interactions (see Fig. 1) suggest that whereas a blue
and orange interior produce similar levels of store patronage
(x
¯
orange – soft
= 140.97, x
¯
blue – soft
= 147.59) and product pur-
chase intentions (x
¯
orange – soft
= 10.94, x
¯
blue– soft
= 10.79) in
the soft lights condition, these levels change considerably
in the bright lights condition for both store patronage
(x
¯
orange – bright
= 122.63, x
¯
blue – bright
= 159.14) and purchase
intentions (x
¯
orange – bright
= 9.60, x
¯
blue – bright
= 12.66). In addi-
tion, the Color Price interaction affected store patronage
primarily because of a far higher mean in the low price, blue
condition (x
¯
orange – US$149.95
= 127.70, x
¯
blue – US$149.95
=
136.70, x
¯
orange – US$59.95
= 132.50, x
¯
blue – US$59.95
= 170.90).
For both dependent variables, store patronage and pur-
chase intentions (see Table 2), all significant effects (main
effects and interactions) of the independent variables on
dependent variables were no longer evidenced when medi-
ators are introduced. For example, the main effect of color
on store patronage (see Table 2) is significant when no
mediators are present ( F= 8.70, P< .01) yet is nonsignifi-
cant when the mediators are included ( F= 1.97, P>.10).
Thus, support is found for Hypothesis 4 in that subjects’
affective and cognitive reactions mediate the relationships
between physical store characteristics and behavioral inten-
tions. This result is qualified only in that the price fairness–
patronage intentions link is insignificant.
6. Discussion
6.1. Summary
This research offers several potential practical and theor-
etical contributions. First, additional evidence is presented
suggesting ways store consumers may be affected by color.
In particular, the results suggest interesting findings with
respect to consumers’ reactions to various color and light
combinations. Second, the research suggests the efficacy
with which consumer reactions to these characteristics are
mediated by their subsequent cognitive and affective reac-
tions. Third, by comparing these with previous results, the
Table 2
Analysis of variance results for store patronage and purchase intentions
Store patronage Purchase intentions
Nonmediated results Mediated results Nonmediated results Mediated results
df F Significance
of F
FSignificance
of F
FSignificance
of F
FSignificance
of F
Main effects
Color 1 8.70 .00 1.97 .16 5.64 .02 0.39 .54
Lights 1 0.04 .87 0.21 .64 0.20 .66 1.19 .28
Price 1 4.70 .03 1.15 .29 14.86 .01 1.81 .18
Two-way interactions
Color Lights 1 3.96 .05 0.02 .96 5.02 .03 0.00 .97
Color Price 1 4.34 .04 1.31 .26 0.08 .78 1.09 .30
Lights Price 1 0.24 .62 0.61 .43 0.03 .87 0.07 .79
Three-way interaction
Color Lights Price 1 0.43 .51 0.15 .70 0.63 .43 1.37 .26
Covariates
Evaluation – – – 104.41 .00 – – 4.30 .04
Excitement – – – 15.09 .00 – – 4.34 .04
Fairness – – – 0.30 .58 – – 35.49 .01
Store patronage – – – N.A. N.A. – – .66 .42
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551548
research offers insight into the mechanism with which
consumers respond to lights. Fourth, the research suggests
further how physical store characteristics frame purchase
decisions and affect perceptions of price fairness. This
section summarizes and elaborates on the results with
respect to these contributions.
Consistent with previous research (e.g., Crowley, 1993;
Bellizi et al., 1983), consumers reacted more favorably to
cool store interiors. Subjects rated a blue interior as sig-
nificantly more likeable, and they expressed relatively
greater shopping and purchase intentions in the blue as
opposed to the orange condition. However, a persistent
interaction suggested that lighting and color combinations
affect consumers’ cognitive representation and affective
reaction. The interactions imply that the relatively poor
reaction of consumers to an orange store interior is mitigated
by combining it with soft lights. Consumers reported the
lowest evaluation, excitement, price fairness, patronage and
purchase intentions in the orange and bright condition.
However, combining orange with soft lights produced
reactions that were much more comparable to either blue
condition. The cell means suggest a less noticeable, reverse
trend for consumers to react more favorably to brightly lit,
blue stores as opposed to softly lit, blue stores.
Overall, results also suggest that effects of color, lights,
and price on behavioral intentions are mediated by the
cognitive and affective reactions they create. Specifically,
all significant, direct, experimental effects on either patron-
age or purchase intentions disappeared when subjects’
perceived evaluation, excitement, and price fairness percep-
tions were introduced as predictors. This finding suggests
that although color, lighting, and price are linked to shop-
ping intentions, the effect is indirect and other factors that
also influence affect and price fairness must be considered
in understanding their effects more fully.
The findings also are useful in understanding the mech-
anism by which color works. With respect to color main
effects, results presented here showed a preference for blue
interiors. The effect on excitement was only marginally
significant. These results are fairly consistent with Crowley
(1993) who showed preferable evaluations for blue over red
or yellow, but activation means did not display a strong
linear trend (closer to a quadratic trend). Given that previous
research operationalized store interiors by the background
color of slides, and that we operationalized it through
subject imagery, the consistency in results supports further
that consumer reactions to color are the result of learning
more than being physiological.
Further, color and color–lighting effects on price fairness
perceptions suggest that combinations change the cognitive
representation of the store. Controlling for price, consumers
rated the store merchandise’s price as least fair in the orange
and bright condition. This would be consistent with a
classification of a bright, orange store interior with a
‘discount’ or ‘off-price’ type of store in which a high price
is unexpected. In contrast, orange and soft produced the
most fair price perceptions suggesting it is clearly not
classified as a ‘discount’ or ‘off-price’ store. Whereas this
research cannot address this issue fully, it raises interesting
points for further research. Additionally, we found that the
Color Price interaction affected store patronage percep-
tions. The nature of this interaction suggests that consumers
react positively to low prices when stores have a blue
interior compared to an orange interior. Implications of
these findings need to be explored further.
Several practical implications can be suggested. The
results suggest that consumers will react more favorably
and be more likely to shop and purchase socially oriented
products in a new store with a blue interior. However,
lighting must be considered, too. The use of orange with
bright lights can reduce shopping intentions through
reduced affect and excitement. In addition, this combination
appears to produce expectations of very low prices. In
contrast, the use of orange with soft lighting appears to
produce quite the opposite result. Clearly, the selection of
color and lighting is a complex question and it would be
wise to consider other factors as well. For example, due to
the mediation effects shown, any negative effects of color
and lighting may be overridden by introducing other factors
that enhance evaluation and excitement. Additionally, cer-
tain combinations are traditionally associated with certain
retailers and as such, a desire for consistency may dom-
inate. Interestingly, traditional Kmart stores use orange
liberally in their brightly lit stores. Future research might
examine the extent to which these color schemes affect their
performance relative to perceptually similar competitors
like Wal-mart.
6.2. Limitations and future research
The study reported here suffers from several limitations.
For example, our experimental design involved a clothing
context. Therefore, these results may not be replicated in a
different setting involving less involving or socially mean-
ingful goods (i.e., a grocery store). Additionally, we studied
only women and the potential exists that men may react
differently. Further, we chose to operationalize the store
through a scenario approach. While other options (experi-
mental and nonexperimental) exist, this approach has been
effective in previous atmospheric research as cited earlier.
Although having subjects rate actual clothing within actual
stores would maximize realism, this approach would make
control of extraneous factors difficult and previous research
suggests that consumers can create realistic, visual images
from verbal stimuli (MacInnis and Price, 1987). Addition-
ally, this approach allowed us to compare previous results
from studies using slide backgrounds to operationalize store
color schemes (cf. Bellizi and Hite, 1992; Bellizi et al.,
1983; Crowley, 1993) with those presented here and offers
insights into learning aspects of color. Also, consumers
generally do not decide which stores to patronize with
physical images of a store interior within their sight. Rather,
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551 549
they use the images they recall from memory (Darden,
1979). However, future researchers should consider these
limitations and might consider adopting methodologies
using computer-generated virtual store images in an effort
to reproduce these results. This type of experiment would
allow subjects to be exposed to actual colors (e.g., shades of
blue) and specific levels of light (e.g., degrees of brightness).
These results suggest several avenues for further re-
search. Perhaps other ambient factors (e.g., store temper-
ature) can be studied in combination with color and/or
lights. Whereas the affects of color appear learned, temper-
ature would represent a more physiological factor. Other
research may also address further the way consumer retail
categories are created and affect decision making. Addition-
ally, are category exemplars or prototypes most appropriate
for representing these stores (Holland et al., 1989)?For
example, it may be useful to get consumer reactions to
hypothetical store descriptions in terms of what stores they
seem most similar to? If a novel store shares some physical
characteristic(s) of a familiar store, to what extent are
category-based affective and cognitive reactions trans-
ferred? Such research may provide insight into situations
when a typical design is beneficial or not beneficial.
6.3. Conclusion
This study suggests that store color is important in
understanding patronage behavior. However, the relation-
ships between color and patronage and purchase intentions
are not simple and depend considerably on other moderating
and mediating factors. Whereas our results suggest that blue
is generally a ‘safe’ color scheme producing few negative
reactions, it also suggests that colors that appear detrimental
to retail performance, like orange, may produce positive
results if other atmospheric elements are altered in com-
bination with them. This view suggests that consumers do
process atmospheric characteristics holistically more than
piecemeal. Overall, the results further an appreciation for
the importance of merchandising and physical design in
shaping patron reactions.
References
Anderson SM, Klatzky RL. Traits and social stereotypes: levels of
categorization and person perceptions. J Pers Soc Psychol 1987;53:
235 – 46 (August).
Areni CS, Kim D. The influence of in-store lighting on consumers’
examination of merchandise in a wine store. Int J Res Mark 1994;11:
117– 25 (March).
Babin BJ, Darden WR. Good and bad shopping vibes: spending and patron-
age satisfaction. J Bus Res 1996;35:201 – 6 (March).
Babin BJ, Boles JS, Darden WR. Salesperson stereotypes, consumer emo-
tions, and their impact on information processing. J Acad Mark Sci
1995;23:94 – 105 (Spring).
Babin BJ, Darden WR, Babin LA. Negative emotions in marketing re-
search: affect or artifact? J Bus Res 1998;42:271 – 86 (July).
Baker J, Grewal D, Levy M. An experimental approach to making retail
store decisions. J Retailing 1992;68:445– 60 (Winter).
Baker J, Grewal D, Parasuraman A. The influence of store environ-
ment on quality inferences and store image. J Acad Mark 1994;22:
328 – 39 (Fall).
Baron RA. The physical environment of work settings: effects on task
performance, interpersonal relations, and job satisfaction. Research in
organizational behavior. New York: JAI Press, 1990. p. 1 – 46.
Baron RM, Kenny DA. The moderator – mediator variable distinction in
social psychological research: conceptual, strategic, and statistical con-
siderations. J Pers Soc Psychol 1986;51(6):1173– 82.
Bellizi JA, Hite RE. Environmental color, consumer feelings and purchase
likelihood. Psychol Mark 1992;9(5):347 – 63.
Bellizi JA, Crowley AE, Hasty RW. The effects of color in store design. J
Retailing 1983;59:21 – 45 (Spring).
Bitner MJ. Servicescapes: the impact of physical surroundings on custom-
ers and employees. J Mark 1992;56:57 – 71 (April).
Crowley AE. The two-dimensional impact of color on shopping. Mark Lett
1993;4:59 – 69.
Darden WR. A patronage model of consumer behavior. In: Stampfl R,
Hirschman E, editors. Competitive structure in retail markets: the de-
partment store perspective. Chicago, IL: American Marketing Associa-
tion, 1979. p. 51 – 74.
Darden WR, Babin BJ. Exploring the concept of affective quality:
expanding the concept of retail personality. J Bus Res 1994;29:
101 – 10 (February).
Dawson S, Bloch PH, Ridgway NM. Shopping motives, emotional states
and retail outcomes. J Retailing 1990;22:408 – 27 (Winter).
Donovan RJ, Rossiter JR. Store atmosphere: an environmental psychology
approach. J Retailing 1982;58:34 – 57 (Spring).
Holbrook MB. Emotion in the consumption experience: toward a new
model of the human consumer. In: Peterson R, Hoyer W, Wilson W,
editors. The role of affect in consumer behavior. Lexington, MA: Lex-
ington Books, 1986. p. 17 – 52.
Holland JH, Holyoak KJ, Nisbett RE, Thagard PR. Induction Cambridge,
MA: MIT Press, 1989.
Hui MK, Dube L, Chebat J-C. The impact of music on consumers’ reac-
tions to waiting for services. J Retailing 1997;72:87 – 104 (Spring).
Jacobs KW, Suess JF. Effects of four psychological primary colors on
anxiety state. Percept Mot Skills 1975;49:143–61.
Kalapurakal R, Dickson PR, Urbany JE. Price fairness and dual entitlement.
In: Advances in consumer research, vol. 18. Provo, UT: Association for
Consumer Research, 1991. p. 788 – 93.
Kotler P. Atmospherics as a marketing tool. J Retailing 1974;49:48 – 64
(Winter).
Lichtenstein DR, Ridgway NM, Netemeyer RP. Price perceptions and
consumer shopping behavior: a field study. J Mark Res 1993;30:
234 – 45 (May).
MacInnis DJ, Price LL. The role of imagery in information processing:
review and extensions. J Consum Res 1987;13:473 – 91 (March).
Martins MOB. An experimental investigation of the effects of perceived
price fairness on perceptions of sacrifice and value. Unpublished Dis-
sertation, College of Business Administration, University of Illinois at
Urbana-Champaign, 1995.
Mehrabian A, Russell JA. An approach to environment psychology Cam-
bridge, MA: MIT Press, 1974.
Mick DG. Are studies of dark side variables confounded by socially desir-
able responding? The case of materialism. J Consum Res 1996;23:106 –
19 (September).
Middlestadt SE. The effect of background and ambient color on product
attitudes and beliefs. Adv Consum Res 1990;17:244 – 9.
Richins ML. Measuring emotions in the consumption experience. J Con-
sum Res 1997;24:127 – 46 (September).
Schindler RM. The excitement of getting a bargain: some hypotheses con-
cerning the origins and effects of smart-shopper feelings. In: Srull TS,
editor. Advances in consumer research, vol. 16. Provo, UT: Association
for Consumer Research, 1989. p. 447 – 53.
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551550
Schlosser AE. Applying the functional theory of attitudes to understanding
the influence of store atmosphere on store inferences. J Consum Psy-
chol 1998;7:345 – 69 (Fall).
Spangenberg ER, Crowley AE, Henderson PW. Improving the store envi-
ronment: do olfactory cues affect evaluations and behaviors? J Mark
1996;60:67 – 80 (April).
Sproles GB. Fashion: consumer behavior toward dress Minneapolis, MN:
Burgess, 1979.
Thaler R. Mental accounting and consumer choice. Mark Sci 1985;4:199 –
214 (Summer).
Valdez P, Mehrabian A. Effects of color on emotion. J Exp Psychol: Gen
1994;123(4):394 – 409.
Wakefield K, Baker J. Excitement at the mall: determinants and effects on
shopping response. J Retailing 1998;74:515 – 40 (Fall).
Ward JC, Bitner MJ, Barnes J. Measuring the prototypicality and meaning
of retail environments. J Retailing 1992;68:194 – 220 (Summer).
Wexner LB. The degree to which colors (hues) are associated with mood-
tones. J Appl Psychol 1954;38(3):432 – 5.
Wilson GD. Arousal properties of red versus green. Percept Mot Skills
1966;23:947 – 9.
B.J. Babin et al. / Journal of Business Research 56 (2003) 541–551 551
