Content uploaded by Eric R. Spangenberg
All content in this area was uploaded by Eric R. Spangenberg
Content may be subject to copyright.
Effects of Gender-Congruent Ambient Scent
on Approach and Avoidance Behaviors in a Retail Store
Eric R. Spangenberg
David E. Sprott
Washington State University
Daniel L. Tracy
The University of Tennessee at Martin
Contact information for correspondence:
Eric R. Spangenberg
Professor of Marketing
College of Business and Economics
Washington State University
P.O. Box 99164-4730
Pullman, WA 99164, USA
Phone: (509) 335-3596
Fax: (509) 335-6896
Submitted to the 7th Annual Retail Strategy and Consumer Decision Research Symposium 2004
Consumer Decision Track
Ambient scent in a retail environment can influence consumers, with such effects likely
moderated by congruity between the scent and the retailer’s product offering. Unfortunately,
minimal prior research has not documented such effects on a broader range of products and in
real-world settings. This research addresses these shortcomings by exploring the evaluative and
behavioral effects of congruity between the perceived gender of ambient scent and a store’s
gender-based products. A field experiment demonstrates scent congruity to influence
perceptions of the store, its merchandise, and actual sales. Supporting an S-O-R interpretation,
affective responses to the environment and arousal mediated these effects.
Keywords: Consumer psychology; retail environments; ambient scent; gender effects
Effects of Gender-Congruent Ambient Scent on Approach and Avoidance Behaviors
Retailers’ acquiescence to the importance of attending to environmental psychological
variables is fueled by research showing that environmental cues affect critical consumer
responses (e.g., Chebat and Michon, 2003; Mattila and Wirtz, 2001; Spangenberg, Crowley, and
Henderson, 1996). The burgeoning investigation of such effects has identified olfactory cues as
one of the many important components of the retail environment influencing people’s
perceptions of the store itself and products offered for sale therein (e.g., Bone and Ellen, 1999).
Although recognized as a moderator, the appropriateness (or congruity) of olfactory cues
has been given limited empirical attention in prior research with only two published studies
(Bone and Jantrania, 1992; Mitchell, Kahn, and Knasko, 1995). In these studies, congruity (as
compared to incongruity) between a scent and a focal product has been shown to lead to
favorable outcomes including improved information processing (Mitchell et al., 1995), enhanced
product evaluations (Bone and Jantrania, 1992), and altered choice behavior (Mitchell et al.,
1995). Existing research is limited, however, in that it has remained in the laboratory and has
only examined stimuli with inherent scents (i.e., hand lotion and cleaning agents, Bone and
Jantrania, 1992; chocolate, Mitchell et al., 1995). Of course in the real world, firms offer many
products that do not have an expected scent associated with them (e.g., clothing).
The current research extends the extant prior research by exploring customers’ responses
to an ambient scent in an actual store: rather than focusing on scents inherent to products (e.g.,
floral scents in a florist), we investigate the effectiveness of ambient scents that do not originate
from the product offering. More specifically, congruity between the gender-based product
offerings of a retailer and the perceived femininity or masculinity of ambient scents is explored.
Consistent with prior conceptualizations of product-scent congruity (i.e., the correspondence or
‘fit’ of a particular scent with a target object, or its appropriateness in certain contexts; Bone and
Ellen, 1999), gender–scent congruity is defined herein as correspondence of an ambient scent
with the gender-based products offered for sale by a retailer. Following a review of pertinent
literature and an overview of our theoretical expectations, we report the results of a field
experiment conducted in a clothing store, where we examine the effects of congruent versus
incongruent gender-based ambient scents on actual consumer response variables.
Ambient Scent and the Retail Environment
Specialty stores like bakeries, chocolate shops and florists often carry product lines with
inherent ambient scents (Mitchell et al., 1995) and have long relied on scents of their products to
attract and influence customers (Bone and Ellen, 1999). Contemporary service providers and
managers of stores carrying products not possessing an inherent (or ‘expected’) scent are also
adding ambient scents to their retail environments (e.g., an artificially diffused floral scent).
The use of ambient olfactory cues by business practitioners has led to increased attention
by scholars attempting to determine the psychological and behavioral effects of olfactory cues on
people. Extant research demonstrates that ambient scents impact a variety of consumer
perceptions and behaviors; a recent comprehensive review of the literature (Bone and Ellen,
1999) suggests that the presence of an ambient scent can elicit cognitive elaboration, affective
and evaluative responses (e.g., Spangenberg et al., 1996), influence purchase intentions (e.g.,
Mitchell et al., 1995; Spangenberg et al., 1996), and possibly alter actual customer behavior
(although this issue remains undemonstrated in the literature). These effects on consumers are
theoretically supported by research in environmental psychology including works by Mehrabian
and Russell (1974), Bitner (1992), and Gulas and Block (1995). The theoretical mechanism
underlying most of the published research in this area is the well-defined stimulus-organism-
response (S-O-R) paradigm (a detailed review is provided by Spangenberg et al., 1996).
Regarding ambient scents in retail settings, the S-O-R model posits that environmental olfactory
stimuli (S) (combined with other cues) affect consumers’ internal evaluations (O) (e.g., affective
responses), in turn eliciting approach or avoidance responses (R).
In particular, an environment’s characteristics combine to create degrees of affective
response and arousal in people. Affect is defined as the general positive or negative state of
emotion or feeling and affective response in the context of this work is the emotional reaction to
the environment that a person has come in contact with (Bower, 1981). According to the S-O-R
model, pleasant scents should lead to pleasant affective (or mood) states while unpleasant odors
result in unpleasant affective states (see Ehrlichman and Bastone, 1992). The term arousal refers
to the psychological feeling state elicited by the environment (Mehrabian and Russell, 1974).
The S-O-R model suggests that an arousing, affectively pleasant environment should produce
approach behaviors while an arousing, affectively unpleasant environment should produce
avoidance behaviors. The literature supports the S-O-R–based notion that pleasantly scented
environments encourage approach behaviors while unpleasantly scented environments elicit
avoidance behaviors (Bone and Ellen, 1999). Approach behaviors are positive responses
directed at the environment or items within the environment; for example, intentions to remain
in, or revisit a store, or actually spending money in a store. Avoidance behaviors reflect opposite
responses; that is, a desire to leave a store, no intention to revisit, or failing to spend money.
Research on the effects of olfactory cues is relatively sparse and published work typically
reports laboratory-based (as opposed to field) experiments. It has been shown, however, that an
important moderator of the demonstrated effects of olfactory cues is congruity between the scent
and the product offering (Bone and Jantrania, 1992) or the environment in which the product is
offered (Mitchell et al., 1995). Findings of two known studies suggest that scents congruent with
product offerings may lead to favorable outcomes for business practitioners (also see Bone and
Ellen, 1999). For example, Bone and Jantrania (1992) found that quality ratings of sunscreen
and household cleaner were more favorable when scented in a manner congruent with the
product category (i.e., coconut sunscreen and lemon cleaner) than when scented in an
incongruent manner (i.e., lemon sunscreen and coconut cleaner). It is noteworthy that scent
congruity had a significant and strong influence on product evaluations in this study whereas
scent pleasantness did not.
While Bone and Jantrania’s (1992) findings focus on scents emanating from the product,
a paper by Mitchell et al. (1995) suggests that such congruity is important for ambient scents as
well. In particular, Mitchell et al. demonstrated that ambient olfactory cues (i.e., chocolate or
floral scents) impacted people’s information processing and choice behavior regarding products
either related or unrelated to the scents (i.e., candy assortments or floral arrangements). More
specifically, this research indicates that congruent scents enhanced consumer judgments (e.g.,
research participants spent more time processing information in the presence of a congruent
scent) and choice behavior (e.g., participants in the congruent scent condition made choices that
were more evenly distributed across all options).
On the whole, extant research suggests that ambient scents in retail environments that are
congruent with products offered should result in positive evaluative outcomes. Further, those
shopkeepers manipulating their environments should avoid the use of product-incongruent
ambient scents. When a cue doesn’t “fit” the context, consumer cognition is potentially taxed to
the point of inhibiting attitude formation (Pomerantz, 1981). Thus, although an odor may be
objectively judged as pleasant, if it is not contextually congruent, evaluation of stimuli associated
with the odor may be counterproductive from the standpoint of a retailer.
The positive effect of congruence between ambient environmental scent and product
offerings is interpretable within the S-O-R paradigm. Social psychology has a long-standing
tradition of demonstrating that people generally seek out and embrace consistency (or
congruency) in their lives and avoid inconsistency when they can (e.g., Cialdini, 1993; Festinger,
1957; Heider, 1958). In the context of the current work, an arousing environment (i.e., a store
containing perceptible olfactory cues) will evoke positive affective responses if the cues are
consistent (or congruent) with people’s expectations. Arousal and positive affective response (as
a result of cue congruence) should elicit approach behavior while arousal coupled with negative
affective response (as a result of cue incongruence) should lead to a lower degree of approach
behavior or a higher degree of avoidance behavior.
Historically, the strategic use of product congruent ambient scents would prove beneficial
for retailers with only relatively narrow product offerings that have an associated scent (e.g., a
florist, a bakery). The importance of ambient scent and product congruity is likely reduced for
retailers who have product offerings without any, or with multiple, associated scents. Thus,
while theme- or product-based scents may prove effective environmental cues for certain
retailers, the pragmatic applicability of this finding is limited for retailers offering multi-line
products with no single inherent or “expected” scent (e.g., discount and department stores).
The current research takes an approach differing from published studies by focusing on
congruity between ambient scent and products with no naturally occurring scent. In the retail
environment, a variety of factors exist (outside of product offerings) that could be used as a basis
for selecting congruent ambient scents. In the current research, we explore the congruity
between ambient scents and a non-scent characteristic of the retail product offering—namely,
gender associated with the items for sale. As such, this work is the first to take into account the
congruity of gender-based associations pertaining to ambient scents and gender-based
associations with products. Our choice of gender as the basis for scent congruity is grounded in
the pragmatic consideration that many multi-line retailers organize products in their stores based
on gender. For example, many department stores separate product offerings by gender, such that
women’s clothing occupies one floor of the store while male clothing occupies another. If
proven beneficial, such a retailer could effectively use ambient scents to alter the retail
atmosphere by infusing gender-congruent scents into each of the gender-oriented departments.
Our general expectation is that ambient scent and product gender congruity should lead to
favorable outcomes for the retailer.
Thus, building upon the foundational work in the literature on environmental
psychology and extant findings on the effects of product and scent congruity, we expect ambient
scents that are gender-congruent with product offerings to lead to enhanced evaluations and
behavioral responses in a retail setting, as compared to gender-incongruent scents. Specifically,
we expect people shopping in a gender-scent-congruent environment (i.e., a store where the
merchandise and ambient scent are gender-consistent) will exhibit approach behaviors. In
contrast, people shopping in a gender-scent-incongruent environment (i.e., where merchandise
and scent are gender-inconsistent) will manifest avoidance behaviors. Furthermore, as suggested
by the S-O-R model detailed previously, we expect that these effects of ambient scent are
mediated through affect and arousal.
Scent Selection Pretest
While the marketplace seeks to segment fragrances on the basis of gender (e.g., perfumes
targeted toward men and women), empirical evidence is lacking regarding the perceived gender of
ambient scents that could be used in retail settings. Although not all research agrees, it should be
noted that documented gender differences exist regarding olfaction (Brand and Millot, 2001). For
example, research indicates that women are more sensitive to certain odors and have greater
abilities to identify scents than men (Wysocki and Gilbert, 1989). Research also demonstrates
that males and females respond differentially to olfactory cues (Gustavson, Dawson, and Bonett,
1987; Jacob and McClintock, 2000; Kirk-Smith and Booth, 1980; Wysocki and Gilbert, 1989).
For example, Kirk-Smith and Booth (1980) found that women were attracted to (and men
avoided) chairs in a doctor’s waiting room when saturated with an androstenone odorant (a musky
odor that has been identified in human sweat and urine). Given the lack of research regarding
gendered-aspects of scent, a pretest was required to identify scents that would serve as appropriate
stimuli in the field experiment. In particular, we sought to determine the gender orientation of
various scents and to select scents appropriately perceived as feminine and masculine.
Stimuli, Participants, and Procedures
A commercial aroma distributor donated scents for the pretest based on two criteria. The
first criterion was related to internal validity and required that the scents be gender-oriented.
While the aroma distributor had no knowledge of any research documenting the gender of
diffusible ambient scents, the firm suggested several scents likely to be gender-oriented based on
their tacit knowledge. The second criterion was related to external validity and required the
scents to be currently in use by retailers, such that managers would generally consider them to be
commercially viable alternatives. Based on these criteria, twelve candidate scents were selected
and pretested; all were natural essential oils (plant extracts) donated by the single supplier.
Participants included 300 students, faculty, and staff (48 percent female; M = 27.2 years)
intercepted in the student union of a university located in the town where the main field study
was conducted. Following Spangenberg et al. (1996), participants read and signed an informed
consent screening for allergies and were given a vial containing a single olfactory stimulus and a
one-page survey consisting of focal measures (detailed subsequently). Scents were placed in
opaque vials to eliminate influence of different oil colors in evaluations (Zellner and Kautz,
1990). The vials (labeled 1 through 12 with no other descriptors) were tightly sealed and
contained twelve drops of essential oil on a cotton ball. Participants were asked to open the vial
and sniff as much as they liked while completing the survey; most individuals sniffed their vial at
least twice. Each research participant evaluated only one randomly assigned scent.
Some concern may arise over the use of a pretest sample that appears to be largely
student-based when the field experiment was conducted at a retail establishment open to the
general population. This potential concern is alleviated in several ways. First, the pretest was
conducted on an alumni weekend; the sample’s average age (27.2 years) was significantly higher
than the student population’s average. Secondly, the raison d’être for the community in which
the retail establishment is located is support of the university where the scent selection pretest
was conducted (students make up approximately 70 percent of the town’s population).
Perceived gender orientation of the scents was measured with the sum of three semantic
differential items: masculine/feminine, masculine/unmasculine, and feminine/unfeminine (cf.
Friedman and Dipple, 1978). Pleasantness of the scents was measured with three summed items:
bad/good, favorable/unfavorable, and positive/negative. Also measured was the perceived
intensity of the scents (very weak/very strong; Spangenberg et al., 1996). While perceived
intensity in the pretest may not be representative of the intensity experienced in the main
experiment, this measure was included to insure similar levels of intensity among those scents
selected for inclusion in the main experiment. All items were measured on 7-point scales. Age
and gender of research participants also were recorded.
Results and Discussion
The primary goal of the pretest was to find one scent judged to be masculine and one
judged to be feminine by both genders; that is, we sought strong manipulations for the main
study that were unambiguously feminine or masculine. Further, we sought scents that businesses
would actually consider using, based on criteria like pleasantness, expense, and intensity
(Spangenberg et al., 1996). Summary of pretest results are provided in Table 1.
Of the twelve scents tested, 7 showed a significant gender orientation for the overall
sample (α = .83 for three masculine-feminine items). When the scents were analyzed separately
for male and female research participants, however, only three scents were perceived to have
clear gender orientations. Of those, two scents were most clearly associated with masculinity
and femininity respectively. As detailed in Table 1, rose maroc was rated significantly
masculine and vanilla was rated significantly feminine by both genders, as compared to gender
neutrality, all p’s < .05. In addition, a measure of the scents’ pleasantness revealed that both
rose maroc and vanilla were moderately pleasant. Pleasantness scores (neutral = 12) for rose
maroc were 14.92 and 17.62 as rated by men and women respectively, t(23) = 2.49, p = .02, and
intensity (on 7-point scale) was rated moderate by both genders (4.92 for men and 5.38 for
women), p > .20. Pleasantness and intensity ratings of the vanilla scent did not differ across men
and women: pleasantness was 14.77 and 12.73 as rated by men and women respectively, p > .10,
with intensity ratings of 5.36 and 6.18 as rated by men and women respectively, p > .10. Rose
maroc was associated with a higher level of pleasantness than vanilla, F (1,46) = 37.84, p = .001,
yet an interaction effect of scent and gender on pleasantness scores did not emerge, p = .75.
Based on clear results for gender orientation and equivalent level of scent intensity, rose
maroc was selected as the masculine scent and vanilla was selected as the feminine scent for use
in the main field experiment. The moderately positive pleasantness scores for both of these
scents suggested that they would also be realistic choices for retailers.
The effects of congruent versus incongruent gender-based ambient scents on consumer
response variables were explored in a real clothing store. Ambient scents were manipulated in
the store environment, actual shoppers served as experimental participants, and actual sales
served as one of several dependent variables.
Design, Participants and Procedure. The field experiment employed a between-participants
design with two conditions: feminine ambient scent (i.e., vanilla) versus masculine ambient scent
(i.e., rose maroc).1 The experiment was conducted in a retail clothier located in a small university-
centered community. The retailer sold both men’s and women’s clothing within equivalent
quadrate floor spaces; product mix was equally divided into areas targeted toward each gender.
The experiment occurred during a two-week period when a storewide sale was held. The sale
began two weeks before the beginning of the experiment and ran for several weeks after the
1 A control condition (i.e., no scent) was not included as part of the design due to concerns of the retailer where the
study was conducted. In particular, the proprietor felt that the elimination of scent altogether would be
incommensurate with his best interests since he normally diffuses a scent (not used in this study) in his store. Given
experiment was completed; such a time frame ensured consistent advertising, pricing, and product
Participants were 181 shoppers (82 men and 99 women). The majority of customers
were shopping for themselves; thus, male customers predominantly purchased men’s clothing
and female customers most often bought women’s clothing. Of the 202 people visiting the store
over the course of the study, only 10 exhibited cross-gender buying; this number did not allow
for meaningful analyses with such a small cell size and thus these persons were dropped from
further experimental consideration. Of persons entering the store over the course of the study, 11
refused to complete the dependent measure questionnaire (4 women and 7 men with no pattern
associated with either scent). There were 92 participants visiting the store in the first week and
89 in the second. Gender of participants was consistently distributed across days and weeks
during the experiment; 90 customers were exposed to the masculine scent, and 91 to the feminine
scent. Thus, distribution of participants (across both genders) included 93 customers who
shopped in the presence of a congruent ambient scent and 88 in the incongruent ambient scent.
These two conditions served as the independent variable in subsequently reported analyses.2
Experimental scents were diffused throughout the entire store using a commercial scent
diffuser designed for use by retailers. Based on the results of the pretest, the ambient scents
included rose maroc as the masculine scent and vanilla as the feminine scent. Thus, scent was
congruent when its perceived gender orientation (as determined by the pretest) matched the
gender of the products offered (i.e., rose maroc for men’s clothing, and vanilla for women’s
clothing), and incongruent when the scent’s perceived gender did not correspond with the
that prior research has demonstrated the basic effect of ambient scent on consumer reactions, we felt that exclusion
of a control condition would not be unduly harmful to the study.
2 An alternate analysis for the study would be a 2 x 2 ANOVA with shopper gender and ambient scent serving as the
two independent variables. Results of those alternate analyses mirror those reported in this manuscript.
product offering (i.e., rose maroc for women’s clothing and vanilla for men’s clothing). In order
to minimize any potential confounding effects of scent intensity and pleasantness, the proprietor
and three researchers pretested and adjusted the intensity of the scent in the store environment to
a mild intensity level. The two focal scents were counterbalanced during the two-week
experiment. In particular, data were collected on the Tuesday, Wednesday, Friday, and Saturday
of each week and the scents were changed each preceding Sunday and Thursday. Such a
procedure allowed for the scents to be switched and manipulations to be counterbalanced
between days and weeks. Data were not collected for at least one day after changing scents in
order to allow the previous scent to dissipate and the new scent to diffuse fully throughout the
store. Intensity of the scents was constantly maintained across the various experimental
treatments, so as not to draw attention to the change in scents. Participants were unaware of the
scent manipulation; that is, scent was not mentioned by the retailer or research assistants at any
point in the procedure or by research participants in response to the open-ended hypothesis-
guessing question at the end of the questionnaire.
Measures. Dependent measures included evaluations of the store and its merchandise,
and customer approach/avoidance behaviors. Shopper’s overall evaluation of the store was
measured with sum of three items: bad/good, unfavorable/favorable, and negative/positive, >
.90. Evaluation of the merchandise (Belizzi, Crowley, and Hasty, 1983) was assessed with a
series of single items assessing: merchandise selection (adequate/inadequate), merchandise style
(outdated/up-to-date), merchandise quality (low/high), and merchandise prices (low/high). All
of the preceding items were measured on 7-point scales.
Actual time spent in the store was measured following prior research (Milliman, 1982,
1986; Spangenberg et al., 1996): a research assistant located at the store’s entrance/exit measured
the actual time spent in the store, in seconds, for each shopper (i.e., the time from entering the store
until beginning the survey). Intentions to visit the store (Spangenberg et al., 1996) were measured
by asking the question: “Assuming you were going to purchase this type of merchandise and had
the money, how likely would you be to visit this store?” (unlikely/likely; 7-point scale).
Because no known academic research has shown the effects of olfactory cues on actual
behavioral measures of approach/avoidance, the measures for the present experiment were based
on related research exploring the effects of music on consumers in actual retail settings
(Milliman, 1982; 1986). In particular, the retailer provided the number of individual clothing
items purchased and dollars spent by each individual customer; itemized sales records were
matched with questionnaire responses to confirm accuracy of participant self-report.
Mediators and Other Measures. To provide process evidence, two mediators were also
collected: (1) Affective reaction to the store’s environment was assessed using the sum of
appropriate items from Fisher’s (1974) environmental quality scale (negative/positive,
unattractive/attractive, tense/relaxed, uncomfortable/ comfortable, bad/good; = .83), and (2)
Arousal from the environment was measured using 5 additional items from the Fisher scale
(boring/stimulating, unlively/lively, dull/bright, unmotivating/motivating, uninteresting/
interesting; = .91). Gender and age of all customers were recorded. Participants were also
asked (for classification purposes):“What type of clothing did you buy today, men’s or
Analyses include a MANOVA model assessing the impact of scent congruity on outcome
variables (store evaluations, merchandise evaluations, and approach/avoidance behaviors). We
also conducted a series of regression model estimations to assess mediation.
Influence of Gender Scent Congruity on Outcome Variables. The MANOVA model
indicated that the nature of ambient scent (i.e., congruent or incongruent) had a significant
impact on store and merchandise evaluation, and approach/avoidance behaviors, Wilk’s Λ = .78,
p < .001. Summary statistics for all dependent measures are reported in Table 2.
As predicted, congruent ambient scent had a positive impact on all evaluative responses,
all ps ≤ .02. In particular, gender–congruent ambient scents led to positive significant effects on
the overall evaluation of the store, F (1, 179) = 22.56, p < .01. Further, gender-congruent
ambient scents led to more favorable evaluations of the store’s merchandise regarding: selection,
F (1, 179) = 5.52, p < .05; style, F (1, 179) = 17.34, p < .01; quality F (1, 179) = 14.96, p < .01;
and prices F (1, 179) = 9.87, p < .01.
Gender–congruent ambient scents also had significant effects on approach/avoidance
behaviors. Significant effects emerged for: actual time shopping in the store F (1, 179) = 19.99,
p < .01; intentions to visit the store F (1, 179) = 10.95, p < .01; the number of items purchased
F (1, 179) = 6.23, p < .01; and the amount of money spent in the store F (1, 179) = 10.59, p
< .001). In the presence of gender–congruent ambient scent, shoppers spent more time in the
store, bought more items, and spent more money on their purchases. In addition, they expressed
stronger intentions to visit the store in the future. Taken as a whole the results overwhelmingly
support our prediction of a positive effect of gender–congruent ambient scent on store
evaluation, merchandise evaluations, and the measured approach/avoidance behaviors.
Mediation Analysis. The mediating role of affect and arousal in the preceding effects
was tested via 4 sets of regression equations following Baron and Kenny (1986). To establish
mediation the following conditions must be met: (1) the mediator must be related to the
independent variable; (2) the dependent variable must be related to the mediator; (3) the
independent variable must influence the outcome variable; and (4) the once significant influence
of the independent variable on the dependent variable weakens in the presence of the mediator.
Table 3 shows results of our mediation analyses indicating that congruity of ambient
scent is positively related to affect, = .347, p < .01, and arousal, = .264, p < .01. In addition,
with one exception, affect and arousal are each significantly related to the dependent variables at
p < .01 (the sole non-significant relationship is between price evaluation and arousal, p = .11).
As expected, ambient scent congruity is related to each of the dependent variables, all ps < .05.
Importantly, the relationships between scent congruity and the dependent variables are reduced
in the presence of affect and arousal, such that beta coefficients are reduced in all instances (and
become non-significant in the majority of the cases). Overall, results of the mediation analysis
provide evidence (in addition to the significant MANOVA results presented earlier) that the
S-O-R model accounts for the findings of this field experiment.
The current research demonstrates that shoppers evaluated the store and its merchandise
more favorably, and were more likely to exhibit approach behaviors in the presence of an
ambient scent that is congruent with gender-based products, as compared to an incongruent
scent. These effects were obtained while other store factors like pricing and advertising were
held constant. Of considerable practical interest was the finding that the presence of a gender-
congruent scent increases the majority of approach behaviors for shoppers in an actual store,
including money spent. Of theoretical importance, we provide the first clear empirical evidence
that the S-O-R model of environmental psychology accounts for observed effects of
environmental cue congruity (cf. North et al. 1999).
Clear implications for managers of retail and service outlets can be derived from the
results of this experiment with regard to the benefits of including an appropriate ambient scent as
an element of a store’s environment. Our results advance the findings of Spangenberg et al.
(1996) by showing that the presence of an ambient scent congruent with gender-based products
enhances evaluations of the store, its merchandise and environment, as well as increases
approach behaviors. Perhaps the most important substantive finding from the present research is
the effect of congruent ambient scents on sales. Few controlled experimental studies of
atmospheric effects have access to actual sales figures—indeed, we are not aware of any
published research on olfaction demonstrating the effects of varying scents on actual sales.
The findings of the current experiment extend prior research by demonstrating that
beyond merely incorporating ambient scent, managers would be well served to identify
appropriately gendered scents to diffuse in their retail environments. Our findings also lend
support to prior research indicating that scent “appropriateness” or congruity is an important
evaluative determinant. A vital distinction between the current investigation and prior work is
that our research focused on congruity between an ambient scent and a product category without
an inherent scent. This distinction is of substantive importance in that not all retailers have
scents inherently congruent with their product offerings; for retailers such as these who desire to
use ambient scents, some other form of congruity, such as that between scent and product
gender, is necessary for effective implementation of this atmospheric cue. As many retail outlets
(or departments within outlets) physically separate gender-based product offerings (e.g., men’s
and women’s clothing), the identification and implementation of gender-congruent scents may be
an easily implemented, inexpensive, and effective way to enhance consumer reactions to store
environment and merchandise.
A number of directions for future research emerge from the findings of the work reported
herein. Most obvious are inquiries aimed at extending the findings of the current research. For
example, future work should be done to determine additional scents exhibiting product gender
congruity and to examine the effects of such scents in a wider array of retail outlets. Another
practical direction for future research activity (motivated by retailers in multi-line stores) would
be to consider the interactive effects of multiple scents within the retail environment and the
potential for the “blending” of incongruent scents. Alternately, the findings of the present article
could be extended to other environmental cues, such as music or décor. More broadly, the
findings of the current field experiment may also be usefully applied in non-retail settings since
ambient scent should influence approach/avoidance behaviors for other situations where people
are surrounded by a controllable environment. One interesting direction for such research would
be the use of ambient scent to influence the behaviors of workers in an organization’s office
building or manufacturing facility (cf. Bitner 1992). Regardless of the direction this type of
work may take, the ideas discussed above provide interesting opportunities for collaboration
between academics and practitioners.
Research that is more theoretically oriented is also motivated. One area worthy of
inquiry is to determine dimensions of scent congruity beyond perceived product gender. As
noted previously, perhaps one of the most important findings of the current work is the notion
that product gender congruity matters. Given this finding, what other dimensions might
similarly be to the benefit of environmental psychologists? A variety of theoretically intriguing
dimensions could be explored (e.g., geographic region, consumer gender, ethnicity, and age),
however, it may not be easy to identify the “right smell” (as suggested by research on the effects
of music; Yalch and Spangenberg, 1990). This stream of research is challenged, therefore, to
identify effective olfactory stimuli that can be used to identify and thereby target viable market
Additional research could also investigate the use and coordination of multiple
environmental cues (scents, sounds, décor, etc.) in a retail setting. Existing research has not
addressed the interactive effects of multiple environmental cues. Similar to work in advertising
and promotions (e.g., Schulz, Tannenbaum, and Lauterborn, 1993), it may be necessary to
develop a theory of “integrated atmospherics” to help guide retailers choice and use of various
environmental psychological cues. One useful direction for such research would be to explore
the congruity of multiple environmental cues on consumer behavior.
In summary, the mere presence of an ambient scent clearly leads to important changes in
consumers’ evaluation of retail outlets, the associated environment, and offered merchandise, in
addition to changes of key behaviors. It is obvious that retailers should not use unpleasant
ambient scents in their retail environments. It seems, however, that there is considerable latitude
for managers when determining a “good smell” to implement and it is critical to consider
important moderating factors in order to optimize the effects of this atmospheric variable. Our
work suggests that retailers should consider the use of scents congruent with gender-based
products as a method for increasing positive evaluations of their store environment, the
merchandise contained therein, and ultimately increasing their bottom line.
Baron Reuben M., Kenny David A. The moderator-mediator distinction in social psychological
research: Conceptual, strategic, and statistical considerations. Journal of Personality and
Social Psychology 1986; 51 (6): 1173-1182.
Belizzi Joseph, Crowley Ayn, Hasty Ronald. The effects of color in store design. Journal of
Retailing 1983; 59 (Spring): 21-45.
Bitner Mary Jo. Servicescapes: The impact of physical surroundings on customers and employees.
Journal of Marketing 1992; 56 (April): 57-71.
Bone Paula, Ellen Pamela Scholder. Scents in the marketplace: Explaining a fraction of olfaction.
Journal of Retailing 1999; 75 (2): 243-262.
Bone Paula, Jantrania Swati. Olfaction as a cue for product quality. Marketing Letters 1992; 3
Bower Gordon H. Mood and memory. American Psychologist 1981; 36 (2): 129-148.
Brand Gérard, Millot Jean-Louis. Sex differences in human olfaction: Between evidence and
enigma. The Quarterly Journal of Experimental Psychology 2001; 54B (3): 259-270.
Chebat Jean-Charles, Michon Richard. Impact of ambient odors on mall shopppers’ emotions,
cognition, and spending: A test of competitive causal theories. Journal of Business Research
2003; 56 (7): 529-540.
Cialdini Robert B. Influence: The psychology of persuasion (rev.ed.). New York: Quill, 1993.
Ehrlichman Howard, Bastone Linda. Olfaction and emotion. In: M. Serby and K. Chobor, editors.
Science of Olfaction. New York: Springer-Verlag, 1992, pp. 410-438.
Festinger Leon. A Theory of Cognitive Dissonance. Evanston, Il: Row, Peterson, 1957.
Fisher Jeffrey. Situation-specific variables as determinants of perceived environmental aesthetic
quality and perceived crowdedness. Journal of Research in Personality 1974; 8 (August):
Friedman Hershey H., Dipple William S. The effect of masculine and feminine brand names on
the perceived taste of a cigarette. Decision Sciences 1978; 9: 467-471.
Gulas Charles, Bloch Peter. Right under our noses: Ambient scent and consumer responses.
Journal of Business and Psychology 1995; 10 (Fall): 87-98.
Gustavson Andrew R., Dawson Michael E., Bonett Douglas G. Androstenol, a putative human
pheromone, affects human male choice performance. Journal of Comparative Psychology
1987; 101 (2): 210-212.
Heider Fritz. The Psychology of Interpersonal Relations. New York: John Wiley and Sons, 1958.
Jacob Suma, McClintock Martha K. Psychological state and mood effects of steroidal
chemosignals in women and men. Hormones and Behavior 2000; 37: 57-78.
Kirk-Smith M.D., Booth David A. Effects of androstenone on choice of location in other’s
presence. In: Ed H. Van der Starre, editor. Olfaction and Taste VIII. London: IRL Press,
1980, pp. 397-400.
Mattila Anna s, Wirtz Jochen. Congruency of scent and music as a driver of in-store evaluations
and behavior. Journal of Retailing; 77 (2) 273-289.
Mehrabian Albert, Russell James. An Approach to Environmental Psychology. Cambridge, MA:
The MIT Press, 1974.
Milliman Ronald. Using background music to affect the behavior of supermarket shoppers.
Journal of Marketing 1982; 46 (April): 86-91.
Milliman Ronald. The influence of background music on the behavior of restaurant patrons.
Journal of Consumer Research 1986; 13 (September): 286-289.
Mitchell Deborah, Kahn Barbara, Knasko Susan. There’s something in the air: Effects of ambient
odor on consumer decision making. Journal of Consumer Research 1995; 22 (September):
North Adrian C., Hargreaves David J., McKendrick Jennifer. The influence of in-store music on
wine selections. Journal of Applied Psychology 1999; 84 (2): 271-276.
Pomerantz James. Perceptual organization in information processing. In: M. Kubovy J.
Pomerantz, editors. Perceptual Organization. Hillsdale, NJ: Erlbaum, 1981, pp. 141-180.
Schultz Don E., Tannenbaum Stanley I., Lauterborn Robert F. Integrated Marketing
Communications. Lincolnwood, Ill: NTC Business Books, 1993.
Spangenberg Eric R., Crowley Ayn, Henderson Pamela. Improving the store environment: Do
olfactory cues affect evaluations and behaviors? Journal of Marketing 1996; 60 (2): 67-80.
Wysocki Charles J., Gilbert Avery N. National geographic smell survey: Effects of age are
heterogeneous. Annals of the New York Academy of Sciences 1989; 561: 12-28.
Yalch Richard, Spangenberg Eric R. Effects of store music on shopping behavior. Journal of
Consumer Marketing 1990; 7 (Spring): 55-63.
Zellner Deborah, Kautz Mary. Color affects perceived odor intensity. Journal of Experimental
Psychology: Human Perception and Performance 1990; 16 (2): 391-397.
Results of Ambient Scent Selection Pretest
Gender Orientation of Scent
NOTE: Gender neutrality is equal to 12, values greater than 12 indicate masculinity. Cell
entries include means and associated t-tests comparing mean against scale midpoint. Scent
pleasantness is reported for both genders with neutrality equal to 12. Significance levels p
< .05 (2-tail) indicated by “*” and p < .01 (2-tail) indicated by “**”.
Results of Field Experiment: Effects of Ambient Gender Scent on Store Evaluations,
Merchandise Evaluations and Approach/Avoidance Behaviors
Means for Ambient Scent
(n = 88)
(n = 93)
Evaluations of the Store and Merchandise
Overall Store Impression
Shopping Time Actual
Intent to Visit Store
Number of Items Purchased
Money Spent (sales)
NOTE: All F-tests associated with (1,179) d.f. Significance levels p < .05 (2-tail) indicated by
“*” and p < .01 (2-tail) indicated by “**.”
Tests of Mediation
NOTE: Cell entries include standardized beta coefficients associated with appropriate
independent and dependent variables. There is a positive relationship between ambient scent
congruity and affect, = .347, p < .01, and arousal, = .264, p < .01. Significance levels p <
.05 (2-tail) indicated by “*” and p < .01 (2-tail) indicated by “**.”