articol Despre Aglomeratie
-
Upload
carmina-elena -
Category
Documents
-
view
215 -
download
0
Transcript of articol Despre Aglomeratie
-
7/21/2019 articol Despre Aglomeratie
1/8
The differential effects of retail density: An investigation of goods versus service settings
Yue Pan a,,1, Jennifer Christie Siemens b,1,2
a 812 Miriam Hall, University of Dayton, 300 College Park, Dayton, OH 45469-2271, United Statesb 810 Miriam Hall, University of Dayton, 300 College Park, Dayton, OH 45469, United States
a b s t r a c ta r t i c l e i n f o
Article history:
Received 26 December 2008
Accepted 10 February 2010
Keywords:
Crowding
Density
Retail
Goods
Services
Two studies investigate the impact of retail density and time pressure on shoppers' store attitudes and
behavioral intentions. In a goods setting, experiment 1 results show a curvilinear pattern as the level of retail
density increases. That is, individuals prefer a medium level of crowding to a store with low or high
crowding. This inverted-U shaped crowding effect has not previously been tested in the retailing literature.
Experiment 2 employs a service setting, where the relationship between retail crowding and outcome
variables is found to be linear rather than curvilinear, except under conditions of time pressure. In contrast to
the goods setting, individuals have more favorable attitudes and expect to pay more for a service as the level
of crowding increases. Thus, our study ndings suggest that the optimal level of crowding depends on the
type of retail setting.
2010 Elsevier Inc. All rights reserved.
1. Background
Several studies have investigated consumption under conditions
of crowding, suggesting that a crowded environment can lead to
attributions of resource scarcity (Jain, 1987), stress (Stokols, 1978),and negative shopping behavior (Harrell et al., 1980). The crowding
literature seems to converge on an information processing explana-
tion, positing that people experience information overload under high
density conditions (Eroglu and Machleit, 1990). Confronted with
inadequate space, a person feels reduced control over the situation
and experiences stress and arousal that, in turn, eventuate in feelings
of being crowded (Rustemli, 1992).Rompay et al. (2008)suggest that
perceived control mediates the effects of spatial density on pleasure
and approach behaviors, and there is an overall negative relation
between spatial density and perceived control.
Manning Theory in ecological psychology (Barker, 1963), investi-
gates the behavioral and cognitive consequences ofundermanning,
a condition where store personnel are fewer than the number
typically required to maintain a setting at an expected, optimal
level. Extreme undermanning may produce a sense of hopelessness
and frustration, adversely affecting staff attitudes and behaviors
(Vecchio and Sussmann, 1981). Perception of crowding is often
associated with the scarcity of resources (e.g., spatial restrictions,
understaffed store) and its subsequent negative outcomes (e.g., poor
service), feeling cramped, and loss of control and privacy. Retail
crowding has been shown to reduce shopping satisfaction and lead to
avoidance behavior (Eroglu and Machleit, 1990; Machleit et al., 1994;
Pons et al., 2006; Pons and Laroche, 2007).
Conversely, low levels of crowding are also undesirable in shapingshopping expectations. Shoppers often rely on simple, easily acces-
sible social cues for forming inferences about product value when
they lack the time to examine complex information, or when such
information is lacking (Baker et al., 1994; Bloom, 1990). Similarly,
categorization theory states that consumers use available information
(e.g., crowding) to help them draw inferences about possibly missing
information (e.g., price, quality) by comparing the target stimulus
with their category knowledge (Baker, 1998). Thus, shoppers may
gauge the attractiveness of a store by the number of people shopping
there. Similarly, marketing semiotics proposes that the store
environment provides symbols that may characterize product
attributes. A retail outlet with few patrons may present an image of
being overpriced, having poor product quality, or simply unappealing
merchandise. Conceptual and empirical support for the symbolic
nature of store environmental cues abounds in the literature (e.g.,
Baker, 1998; Baker et al., 1988; Bitner, 1990; Grossbart et al., 1990).In
summary, low retail density may incur negative perceptions of the
store and hence less motivation to patronize the retailer. However,
too many shoppers could also be undesirable. Beyond a threshold,
crowding may be seriously dysfunctional, whereby shoppers may
experience high cognitive demands and frustration.
Surprisingly, few controlled studies have investigated medium
levelsof crowding. Instead, research hasfocused on conditions of high
crowding and the subsequent negative consequences of such. Are
there times when a moderately crowded setting can lead to more
favorable results? A medium level of crowding might enhance an
Journal of Business Research 64 (2011) 105112
Corresponding author. Tel.: +937 2291773; fax: +937 229 3788.
E-mail addresses:[email protected](Y. Pan),
[email protected](J.C. Siemens).1 Authors' names are listed alphabetically. Both authors contribute equally to this
study.2 Tel.: +937 229 1086; fax: +937 229 3788.
0148-2963/$ see front matter 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jbusres.2010.02.011
Contents lists available at ScienceDirect
Journal of Business Research
http://-/?-http://-/?-http://-/?-mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.jbusres.2010.02.011http://www.sciencedirect.com/science/journal/01482963http://www.sciencedirect.com/science/journal/01482963http://dx.doi.org/10.1016/j.jbusres.2010.02.011mailto:[email protected]:[email protected]://-/?-http://-/?-http://-/?- -
7/21/2019 articol Despre Aglomeratie
2/8
individual's attitude toward the store, increasing their behavioral
intentions. Although they do not empirically test it, Eroglu et al.
(2005) suggest that there is a general positive effect of human
crowding on shopping satisfaction after certain mediating variables
are accounted for. They propose that there might be an inverted-U
relationship between crowding and satisfaction, whereby an ex-
tremely crowded or uncrowded condition will be undesirable. While
Michon et al. (2005)measure high, medium, and low crowding, they
focus on the effects of ambient odor as an atmospheric variable. Thepurpose of this research is to further investigate the possibility of a
desirable density level for retail settings, specically whether a
curvilinear relationship between crowding and shopping outcomes
exists.
2. Exploratory study
First, an exploratory study was conducted in foundational business
classes. Students were given class credit as an incentive for
participation (n =48). They received an email from the researcher,
randomly directing them to one of three websites where they read a
brief introduction to the study and were asked to imagine themselves
leisurely browsing the stores. Participants were then shown one
picture of a clothing store, which contained either low, medium, orhigh crowding. Participants were instructed to write down any
thoughts they had about the store and their potential to shop there.
2.1. Stimulus development and pretest
In order to present participants with realistic shopping environ-
ments to evaluate, three photos were developed with varying degrees
of crowding. Using photographs as a manipulation of perceived retail
crowding, a common method in the marketing literature, has been
shown to have ecological validity (Bateson and Hui, 1992; Eroglu and
Machleit, 1990). When choosing the retail establishment, careful
consideration was made regarding the merchandise selection, layout
and location of the store. In order to eliminate bias from prior
experience with a recognized store,the chosenstorewas in a differentstate and was not a chain retailer. The store contained both men's and
women's clothing so that merchandise type would not bias store
perception. Shopping volunteers were recruited for the pictures,
which enabled control over the picture content and layout. Thirteen
pictures of varying levels of crowdedness were used in a pretest.
Pretest subjects (n =89) were presented with all thirteen pictures
and were asked to evaluate them with three, seven-point likert items
with endpoints strongly disagree/strongly agree: The store seems
very crowded to me, The store is very busy,and There are a lot of
customers in the store (Machleit et al., 1994). Three color photos
were chosen based on their signicantly different results for the likert
items. These photos were shot from the same area of the store, with
the same type of merchandise shown in each picture. This allows
control over crowding perceptions due to differences in spatialdensity. These photos show varying degrees of human density (from
one shopper to ve shoppers).
2.2. Exploratory study analyses and results
Half of the respondents exposed to a low crowding picture
(n =8) indicated no interest in shopping in the clothing store. Of the
16 respondents who viewed a highly crowded store, the majority
(n =12) expressed no intention to enter the store. A review of the
responses revealed their negative feelings towards the store (e.g.,
messy, unkempt,and not well-maintained).
In contrast, of the 16 respondents assigned to the medium
crowding level, the majority indicated favorable attitudes (n =13) or
intentions (n =12) toward the store. Positive adjectives such as
sophisticated, classy, organized, and rened were use to
describe the store.
In summary, this exploratory study reveals differences in store
perceptions and purchase intentions across the three crowding levels.
While respondents were not asked to specically examine retail
density, their open-ended responses suggested that crowding level
was an important factor. The preliminary ndings provide the
foundation for a controlled experiment. Specically, the following
hypotheses are offered:H1. The effect of retail density on subjects' intentions to enter a store
will follow an inverted-U shaped trend, where intention to enter is
greatest when the store exhibits a medium level of crowding
compared to either low or high crowding levels.
H2. H2a) The effect of retail density on subjects' intentions to
browse in a store will follow an inverted-U shaped trend,
where intention to browse is greatest when the store
exhibits a medium level of crowding compared to either
low or high crowding levels.
H2b) Furthermore, subjects will intend to browse longer in the
retail store when there is a medium level of crowding
compared to low or high levels.
H3. The effect of retail density on subjects' attitudes toward the store
will follow an inverted-U shaped trend, where attitudes are more
positive when the store exhibits a medium level of crowding
compared to either low or high crowding levels.
H4. The effect of retail density on subjects' purchase intentions will
follow an inverted-U shaped trend, where intention to purchase is
greatest when the store exhibits a medium level of crowding
compared to either low or high crowding levels.
However, these proposed relationships are subject to an important
caveat. Reactionsto crowding arenot the same for all people, and may
depend on other environmental factors. Crowding is experienced
when the environment is judged as being dysfunctionally dense; thesejudgments will vary across individuals depending on factors such as
time pressure (Eroglu and Harrell, 1986). Time-constrained shoppers
may be more sensitive to crowded conditions (Eroglu and Machleit
1990), and may view crowding more negatively as it prevents
effective goal achievement. As such, we propose that as the amount of
time pressure increases, the negative impact of retail crowding on
shopping attitudes and intentions is likely to be enhanced.
H5. Time pressure will interact with retail density for all dependent
measures. Under the condition of time pressure, the effect of retail
density on subjects'
H5a) intentions to enter the store,
H5b) intentions to browse in the store,
H5c) attitudes toward the store andH5d) purchase intentions will follow a negative linear trend.
Generally, subjects will generally prefer a less crowded
store under conditions of time pressure.
3. Experiment 1
3.1. Subjects, design and procedure
Subjects were 270 undergraduates at a Midwestern university(see
Table 1 for demographics). Subjects were randomly assigned to one of
six conditions in a 2 (no time pressure vs. time pressure)3 (high,
medium, and low crowding) between-subjects design. Written
instructions were read by the experimenter to ensure equality of
106 Y. Pan, J.C. Siemens / Journal of Business Research 64 (2011) 105112
-
7/21/2019 articol Despre Aglomeratie
3/8
experience across conditions. Subjects read the following shopping
scenario: It is Saturday and you are browsing the stores. You are
thinking of buying something. However, since this day is not the only
stretch of time that you can devote to shopping, you are in no hurry
(no time pressure conditions). For subjects assigned to the timepressure condition, they instead read, Since this day is the only
stretch of time that you can devote to shopping, you are pressed for
time and are in a hurry.They were presented with one of the three
pretested retail photos, and then answered the dependent measures.
3.2. Dependent measures
Attitude towardsthe store wasassessedwith a six-item, seven-point
semantic differential scale. Intentions to enter the store, browse, and
make a purchase were assessed with separate three-item, seven-point
scales (seeTable 2). Participants were also asked to list the length of
time in minutes that they would spend shopping in the store.
The measurement model was assessed for convergent and
discriminant validity. The joint con
rmatory factor analysis (CFA)lends strong support to the convergent validity of all measures (see
Table 2). CFA results show an acceptable model t (2(125)=242.71,
pb0.001; CFI =0.988; RMSEA=0.062; SRMR= 0.067). Thus, the
proposed model (i.e., joint CFA with all constructs included simulta-
neously) provides a reasonable explanation of the observed covari-
ance among the constructs. All estimated item loadings are signicant
(pb0.05) and greater than the recommended 0.6 cutoff (Bagozzi and
Yi, 1988). Furthermore, all constructs show satisfactory levels of
average variance extracted (with all AVE's above the recommended
0.5 level,Fornell and Larcker, 1981) and composite reliability (all CR's
higher than the 0.7 cutoff recommended byNunnally and Bernstein,
1994). The internal validity of the measurement model seems
adequate. We then conduct Fornell and Larcker's (1981) test for
discriminant validity by comparing the AVE estimates for each
construct with the squared correlation between any two constructs.
All AVEs are larger than any squared correlation between any two
constructs. Therefore, discriminant validity is established. Intolerance
for crowding was identied as a potential covariate (Machleit et al.,
2000, 3-item, e.g., I avoid crowded stores whenever possible).
Table 1
Demographic characteristics of respondents.
Experiment 1 (n =270) Experiment 2 (n =403)
Demographic characteristics Frequency/Mean Frequency/Mean
Gender
Female 125 209
Male 144 194
Age 21.2 27.6
Household income
0$20,000 (%) 4 93
$20,001$40,000 (%) 17 14
$40,001$60,000 (%) 28 18
$60,001$80,000 (%) 36 41
$80,001$100,000 (%) 43 67
Over $100,000 (%) 132 155
Table 2
Construct measures and CFA results.
Experiment 1 Experiment 2
Construct, scale items, and source Factor loadings CR AVE Factor loadings CR AVE
Store attitude (Golden et al., 1987) 0.915 0.63 0.922 0.67
The retail store/salon shown in the picture is:
Unfavorablefavorable 0.844 0.869Badgood 0.860 0.883
Not likablelikable 0.866 0.883
Uninterestinginteresting 0.622 0.723
Unpleasantpleasant 0.806 0.824
Low-qualityhigh-quality 0.718 0.698
Intention to enter (based onSpears and Singh, 2004)
How likely would you be to enter the store/salon shown in the picture? 0.960 0.83 0.968 0.91
Denitely notdenitely 0.888 0.931
Not very likelyvery likely 0.931 0.983
Probably notprobably 0.915 0.946
Intention to browse (based onSpears and Singh, 2004)
How likely would you be to browse in the store shown in the picture? 0.971 0.86
Denitely notdenitely 0.913
Not very likelyvery likely 0.942
Probably not
probably 0.923
Purchase intention (based onSpears and Singh, 2004) 0.964 0.84
How likely would you be to buy something from the store shown in the picture?
Denitely notdenitely 0.892
Not very likelyvery likely 0.962
Probably notprobably 0.901
Crowding perception (Machleit et al., 1994) 0.905 0.73 0.910 0.77
The store/salon seems very crowded to me. 0.831 0.814
The store/salon is very busy. 0.919 0.959
There are a lot of customers in the store/salon. 0.808 0.854
Intention to recommend (based on Spears and Singh, 2004) 0.951 0.87
How likely would you be to recommend this salon to a friend?
Denitely notdenitely 0.913
Not very likelyvery likely 0.984
Probably notprobably 0.894
107Y. Pan, J.C. Siemens / Journal of Business Research 64 (2011) 105112
-
7/21/2019 articol Despre Aglomeratie
4/8
Finally, a manipulation check was conducted to ensure the treatment
of store crowding was effective. A one-way ANOVA was signicant,
and LSD post-hoc tests showed signicant differences across all three
levels of crowding (F (2, 266)=114.68, pb0.001; Mlow crowd =2.2,
Mmid crowd =2.6 and Mhigh crowd =4.8). This indicates that the
crowding manipulation was successful.
3.3. Results
MANOVA with LSDpost-hoccomparisonswas used for all analyses
unless otherwise noted. The overall MANOVA was signicant for both
crowding level (F(10, 468)= 3.32,pb0.001) and time pressure (F(5,
234)=11.48,pb0.001, seeTables 3 and 4). In support ofH1, subjects
exhibited a greater intention to enter the store under the medium
density condition compared to the low or high crowding conditions
(pb0.001). Furthermore, intentions to enter the store increased
between the low and medium density conditions but then decreased
as density reached a high level. Post-hoc comparisons revealed
signicant differences in intentions to enter thestorebetween thelow
and medium crowding conditions, and the medium and highcrowding conditions (p'sb0.001). Intention to enter the store did
not signicantly differ between the low and high crowding condi-
tions. This demonstrates support for the inverted-U relationship
proposed inH1.
Supporting H2a, intentions to browse were greatest under
conditions of medium crowding. Intentions to browse increased
between the low and medium density conditions, but then decreased
in the high crowding condition. Post-hoc comparisons showed a
signicant difference in intentions to browse between the low and
medium crowding conditions and between the medium and high
crowding conditions (p'sb0.001, Fig. 1). Furthermore, individuals
were equally likely to browse in the store under conditions of low and
high crowding (ns).
Intolerance for crowding was signicantly correlated with thedependent measure of intentionto browse, and thus wasassessedas a
covariate. ANCOVA results showed a signicant effect of crowding
level on intention to browse (F (2, 255)=9.31, pb0.001). Respon-
dents intended to browsesignicantly longerwhen crowding wasat a
medium level compared to either low or high crowding levels
(p'sb0.001,Mlow =9 min,Mmed =14 min,Mhigh =11 min). This sup-
portsH2b.
A signicant effect of retail crowding on attitudes toward the store
was also found (pb0.01). Despite an inverted-U shaped pattern,
attitudeswere equally positive in both the low and medium crowding
conditions. Post-hoc comparisons demonstrated that store attitudes
were signicantly different between the medium and high (pb0.001)
and the low and high crowding conditions (pb0.05), but were not
signicantly different between the low and medium crowding
conditions (seeFig. 1).
Finally, the results indicated a signicant effect of retail crowding
on purchase intentions (pb0.001). Post-hoc comparisons showed a
signicant difference between low and medium crowding conditions,
and between the medium and high crowding conditions (p'sb0.001,
Fig. 1), whereas individuals were equally likely to make a purchase
under conditions of low and high crowding (ns). Therefore,H4is also
supported. The overall MANOVA shows no interaction between
crowding and time pressure (p =0.13), refuting H5. Regardless ofwhether subjects were in a hurry, they always preferred the medium-
crowded store.
3.4. Discussion
Although they only examine two levels of human density
crowding,Machleit et al. (2000) lead to speculation about a oor
and ceiling effectof retail crowding. Eroglu et al. (2005) also propose
that some level of human crowding might positively affect shopping
satisfaction. However, neither of these studies assesses the boundaries
of this effect or empirically substantiates the claim. Most prior
research on crowding compares two crowding dimensions: low
versus high (e.g., Eroglu and Machleit, 1990; Rustemli, 1992; Machleit
et al., 2000). These studies have led to the general conclusion that asretail density increases, consumers encounter conditions that nega-
tively affect their shopping experience.
The results of study 1 show that a medium level of crowding may
actually be optimum for retailers. In fact, the effect of retail crowding
on subjects' intentions to enter the store, intentions to browse in the
store, length they intended to browse, and purchase intentions each
exhibited a curvilinear trend, where shopping outcomes became most
positive when retail density was moderate. As retail density reaches a
high level, consumers start to exhibit more negative attitudes and
shopping intentions. These ndings have implications for retailers
who attempt to maximize consumers' shopping experience and their
perceptions of the store. Factors such as store layout and other
atmospherics can affect the level of perceived crowding, and can be
manipulated to create an optimal shopping environment.
Table 4
Experiment 1 mean values.
No time pressure Time pressur e
Dependent measures Low
crowd
Medium
crowd
High
crowd
Low
crowd
Medium
crowd
High
crowd
Store attitude 5.1 5.1 4.3 3.6 4.3 3.8
Purchase intention 3.1 3.7 2.8 3.4 4.1 3.1
Intention to enter 3.7 4.5 3.4 3.6 4.3 3.4
Intention to browse 3.6 4.4 3.6 3.3 4.3 3.3
Length of browse 11.1 14.0 8.7 11.4 14.8 8.5
Note. Higher means indicate more favorable ratings. Length of browse means are
reported in minutes.
Fig. 1.Experiment 1 results.
Table 3
Experiment 1 ANOVA/ANCOVA results.
F-values anddffor dependent variables
Independent variables Store
attitude
Intention
to enter
Purchase
intention
Intention
to browse
Length of
browse
Covariate:
Crowd tolerance 5.95a
(1, 255)
Crowding level 6.04
b
9.14
c
10.86
c
9.52
c
9.31
c
(2, 255) (2, 245) (2, 245) (2, 244) (2, 255)
Time pressure 43.51c 0.46 3.84 1.21 0.04
(1, 255) (1, 245) (1, 245) (1, 244) (1, 255)
Time pressure
crowding
4.51a 0.01 0.25 0.11 0.02
(2, 255) (2, 245) (2, 245) (2, 244) (2, 255)
a pb0.05.b pb0.01.c pb0.001.
108 Y. Pan, J.C. Siemens / Journal of Business Research 64 (2011) 105112
-
7/21/2019 articol Despre Aglomeratie
5/8
A potential limitation ofExperiment 1 istheuseof onlyonetype of
retail environment. While Experiment 1 investigated the effects of
retail density within a store selling a tangible product (clothing),
there may be conditions inherent to a goods setting that are different
from a service setting. For example, in a crowded retail store selling
tangible products, a consumer may make inferences about how
difcult it would be tond what they are looking for, get waited on by
the salesperson, or check out with their purchase. This may explain
why individuals view a high level of retail crowding as undesirable.But do the same inferences occur in service settings? Whether
consumers are purchasing a good or service may determine the
degree of informational value of the store environmental cues.
According to Baker (1998), store environment is more likely to have
a pivotal role in the purchase of services for two reasons. First, the
intangible nature of services makes it harder for consumers to assess
value obtained before purchase and consumption. Second, because a
service involves simultaneous production and consumption,consumers
often have to enter the service facility. Therefore, retail density not only
provides extrinsic cues to consumers, it also becomes an intrinsic cue
that forms parts of the total service experience.
Well-established differences between tangible and intangible
products lead to a general expectation that reasons for patronizing a
store in a service setting might differ from those in a goods setting. For
example, services are generally inseparable from the service provider,
do not have inventory, and are nonperishable. In service settings, one
could make an appointment for service, reducing the likelihoodthat the
level of crowding will affect the timeliness of service. Furthermore, in
goods settings, customers mayfeel like they are essentially competing
for products (i.e., if there are too many people in the store, it might
reduce the chance that they get the product they want). Previous
research suggests that products/stores desirability may decrease as a
resultof productscarcitybecausebuyers cannotachieve a high degreeof
personal uniqueness (e.g.,Gierl et al.,2008; Worchel, 1992). Consumers,
by buying a product, equate themselveswithmany other owners of the
same product. The need to compete with fellow shoppers for a product
may inhibit many from patronizing a crowded retail outlet. However,
the outlook for a crowdedservice provider canbe very different. Ponset
al. (2006) nd that conditions of high human density may enhance,rather than harm, consumer experience in leisure services.
Among the few studies that have mentioned an enhancement of
the service experience as a result of the presence of other consumers
in the service setting (e.g., Eastman and Land, 1997; Holt, 1995; Pons
et al., 2006; Price et al., 1995), all have essentially dealt with
experiential and hedonic consumption (e.g., sports spectatorship,
bars). We suspect that the positive inuence of crowds on consumers'
service experience can be extended beyond the context of hedonic
services, for the following reasons. First, services cannot be invento-
ried and demand may not be easily predicted. Waiting before
receiving a service is a common experience (Baker and Cameron,
1996), and customersmay be more tolerant of crowding in stores that
offer services than those that carry goods. Second, retail density in a
service setting signals the desirability of services provided, with anunderstanding that they do not necessarily need to face the
dysfunctional consequences of crowding. For example, shoppers can
avoid the adverse effects of crowding by setting an appointment and
visiting the store at a less busy time. Third, there is less likelihood that
consumers need to compete for a service. Unlike goods that typically
come in limited supply, services can last forever as long as the
provider continues its operation. In a goods setting, scarcity increases
with each sold unit. Shoppers may undergo pressure to strive for the
goods when they perceive a limitation of available units. Scarcity due
to supply is not a concern for services. Therefore, we propose that
shoppers' adaptation strategy will differ in a highly crowded service
context. Avoidance behavior is less likely to occur, and favorable
feelings towards a service provider are more likely to be evoked.
Hence, the relationship between crowding and satisfaction will be
linear in a service setting, where consumers' affective evaluation
becomes more positive as density increases. Specically, we propose:
H6. In the service setting, the effect of retail density on intentions to
enter the store will follow a linear trend. That is, intentions to enter
the store will increase as the level of retail density increases.
H7. In the service setting, the effect of retail density on intention to
recommend the store to a friend will follow a linear trend. That is,
intention to recommend will increase as the level of retail densityincreases.
H8. In the service setting, the effect of retail density on attitudes
toward the store will follow a linear trend, where attitudes become
more positive as the level of retail density increases.
We chose to again assess the time pressurecrowding interaction
in the service setting. As argued earlier, the negative effects of
crowding may be more pronounced when consumers are under time
constraints than when they are not. Therefore, we propose the
following hypotheses:
H9. In the service setting, time pressure will interact with retail
density. For shoppers under time pressure, the effect of retail densityon their
H9a) intention to enter the store,
H9b) intention to recommend the store, and
H9c) attitudes toward the store will exhibit an inverted-U
shaped trend. That is, subjects will prefer a medium level
of crowding when they are under time constraints.
While attitude toward and intention to enter the store remain
consistent with those tested in Experiment 1, here the variable
intention to recommend was included to assess shoppers' behav-
ioral consequences. This variable replaces two outcome variables (i.e.,
purchase intention and intention to browse in the store) used in
Experiment 1, due to the unique features of a service environment
(e.g., consumers aren't likely to browse in a service setting, andpurchase intention is assumed based on entrance to the store). This
outcome variable is more appropriate for use in a service setting.
4. Experiment 2
4.1. Pretest
A pretest was conducted to test the manipulation of crowding in
the service setting (n =59). As inExperiment 1, three photographs of
a hair salon were selected. Pretest subjects saw a photograph andread
a description of a service scenario ranging from one to four customers.
The low crowding condition showed and described only one salon
customer. The medium crowding condition showed and described
one customer waiting to get her nails done and one customer havinghis hair washed.The high crowding condition showed and described
onecustomerwaiting to get her nails done, one getting a haircut, one
getting his hair washed, and still another customer waiting to be
served. Perception of crowding was measured with a three-item,
seven-point likert scale (Cronbach alpha=0.79). Subjects perceived
signicant differences in crowdedness across the three treatment
levels (F(2, 56)=27.64,pb0.001).
4.2. Subjects, design and procedure
Experiment 2 employed a 3 (low, medium, and high crowding)2
(time-pressured vs. non-time-pressured) between-subjects design.
Participants were faculty, staff and students at two Midwestern
universities. Subjects were sent a mail survey to their campus address
109Y. Pan, J.C. Siemens / Journal of Business Research 64 (2011) 105112
-
7/21/2019 articol Despre Aglomeratie
6/8
and were instructed to return the completed questionnaire in a self-
addressed, postage-paid envelope. As an incentive to participate,
respondents were told that their names would be entered into a
randomdrawing for one of three $30 gift cards. Of the 750 mailed, 403
completed surveys were returned (53.7%).
Subjects were shown one of the three pretested photos of the hair
salon and were instructed to imagine themselves in the following
scenario:It is a Saturday and you are at a new salon to get a haircut.
You have plenty of time and are in no hurry
(for the no time pressurecondition). Subjects in the time pressure condition read, You're
pressed for time and are in a hurry. After reading the scenario and
viewing the picture, subjects were instructed to complete a survey
that included the dependent measures.
4.3. Dependent measures
The same scale was adapted from Experiment 1 to measure
attitude towards the hair salonand intentions to enter the salon.
Subjects were also asked to evaluate whether they would recommend
the salon to a friend using a 3-item, 7-point likert scale (see Table 2).
We conduct CFA to assess convergent and discriminant validity of
the measurement model. Fit indices provide evidence of a reasonably
good t (2(84)=266.25, pb0.001, CFI =0.982, SRMR= 0.051,
RMSEA=0.074). All items signicantly and substantially load onto
the expected latent construct. AVE and composite reliability are
calculated, with all above the recommended thresholds, conrming
convergent validity of the measurement model. Within the CFA
setting, discriminant validity is assessed by calculating the shared
variance between constructs and verifying that it is lower than the
variances extracted for the constructs involved. All AVEs are larger
than any squared correlation among constructs (largest squared
correlation=0.42), suggesting that discriminant validity is achieved.
4.4. Results
The overall MANOVA with LSD post-hoc comparisons indicates
signicant effects for crowding level (F(6, 786)=4.63, pb0.001),
time pressure (F(3, 393)=13.0,pb0.001), and their interaction (F(6,786)=7.99, p b0.001). This suggests that the two treatment factors,
separately and in combination, affect shoppers' evaluation of the store
and their behavioral intentions.
Gender is signicantly correlated with subjects' intentions to enter
the salon (i.e., females were more likely to enter the salon), but is not
signicantly correlated with the treatment conditions. As such,
gender is used as a covariate in the analyses for intentions to enter
the salon. Age is signicantly correlated with subjects' intentions to
enter, their intention to recommend, and their attitude toward the
salon. Therefore, age is used as a covariate in the analyses for these
variables.
ANCOVA results show that the level of crowding signicantly
affects intention to enter the salon (pb0.001,Table 5). However, this
main effect was qualied by the signicant interaction of time
pressure and crowding level. For the no time pressurecondition, no
signicant differences were found across the three levels of crowding,
despite a linear, upward trend (see Table 6 for means). Thus, H6 is not
supported.
Crowding level did emerge as an important factor for predicting
subjects' intentions to recommend the salon to a friend. Subjects
under no time constraints were more willing to recommend a salon
that was more crowded compared to less crowded (Mlow crowd =3.5,
Mmid crowd =3.9,Mhigh crowd =4.2). Means were signicantly different
between the low, medium and high conditions (p'sb0.05). Therefore,
H7is supported.
Furthermore, signicant differences were found in store attitudes
between the three levels of crowding (p'sb0.05). In contrast to the
curvilinear relationship portrayed in the goods setting, subjects under
no time constraints view the service provider more favorably when
the salon is more crowded, showing an upward, linear relationship.
This supportsH8.
In support ofH9, a signicant interaction of crowding and time
pressure was found across all three ANCOVA models (p'sb0.001). As
predicted, under conditions of time pressure, individuals generally
preferred a salon with medium levels of crowding. Post-hoc
comparisons revealed that under time pressure, intentions to enter
the salon were signicantly more favorable with medium crowding
compared to eitherlow (pb0.05) or high (pb0.001) crowding (Fig. 2).
This supports H9a. Similarly, under time pressure subjects had a more
positive intention to recommend the salon when crowding was at amedium level compared to either low (pb0.05) or high (pb0.001)
levels of crowding, supporting H9b (Fig. 3). Finally, under time
pressure, attitudes were most favorable when the store exhibited a
medium level of crowding compared to low or high (p'sb0.001,
Fig. 4). This supports H9c.
4.5. Discussion
Most studies on retail density associate crowding with negative
affective evaluation of a given dense situation. Only recently, a few
researchers suggest that in the context of some hedonic services (e.g.,
disco, bar, and sports game), crowding may favorably inuence
consumers' service experience (Arnould and Price, 1993; Holt, 1995;
Table 5
Experiment 2 ANCOVA results.
F-values anddffor dependent variables
Independent variables Store
attitude
Intention to
enter
Intention to
recommend
Covariate:
Gender 17.58a(1, 393)
Age 45.12a (1, 393) 11.97a (1, 393) 21.14a (1, 393)
Crowding level 14.36a (2, 393) 6.87a (2, 393) 5. 64b(2, 393)
Time pressure 6.2c (1, 393) 2.56 (1, 393) 5.25c (1, 393)
Time pressurecrowding 24.78a (2, 393) 10.64a (2, 393) 6.93a (2, 393)
a pb0.001.b pb0.01.c
pb
0.05.
Table 6
Experiment 2 mean values.
No time pressur e Time pr essure
Dependent measures Low
crowd
Medium
crowd
High
crowd
Low
crowd
Medium
crowd
High
crowd
Store attitude 4.2 4.6 5.0 4.6 5.4 4.0
Intend to recommend 3.5 3.9 4.2 4.4 5.0 4.0
Intention to enter 4.1 4.2 4.4 4.1 4.8 3.4
Note. Higher means indicate more favorable ratings.
Fig. 2.Experiment 2 results for intention to enter.
110 Y. Pan, J.C. Siemens / Journal of Business Research 64 (2011) 105112
-
7/21/2019 articol Despre Aglomeratie
7/8
Ponset al, 2006). This study explores the consumption setting of a less
experiential, less hedonistic service (hair salon). Findings suggest the
presence of positive consumer responses and the potential contribu-
tion of fellow participants in creating heightened expectation for the
service experience.
Experiment 2suggests that retail density in a service context may
enhance product desirability, especially for those shoppers under no
time pressure. This study sheds light on several distinct differences
between goods and service settings that ultimately affect customers'
interpretation of the crowded retail environment. Unlike goods,
quantitative scarcity due to shortage of supply is not a problem for
services. Theoretically, the supply of a service is endless (whereas a
good can be sold out), as long as the service provider remains in
business. Furthermore, a service can be arranged to be delivered at a
time of a shopper's choice. Therefore, in a service encounter where
crowding exists, shoppers may not pursue the avoidance strategy as
they would do in a goods setting. Shoppers have more cognitive
control, knowing that they are not competing against others for a
commodity of limited availability. Thus, the negative psychological
consequences caused by retail density (e.g., emotional discomfort,
stress) can be largely avoided in a service setting. Furthermore, high
retail density may cultivate a perception of a high demand for the
service, thus increasing the service provider's attractiveness in theview of shoppers. A caveat to these ndings is when the consumer is
pressed for time. When consumers are in a hurry, they typically prefer
a hassle-free retail environment that does not work against
completing a shopping task in a timely manner. Conversely, they
are less favorable towards the store when it exhibits a high level of
crowding. Nevertheless, they still do not favor low levels of crowding,
even when under time pressure.
5. General discussion and future research
Based on the ndings of two experiments, it appears that intrinsic
attributes of goods versus service settings might dominate the result
of a patronage decision. Shoppers rely on heuristic information in
deriving judgments about the desirability of a retail store. Retail
density can be interpreted as a heuristic cue. The effect of retail
density on consumers' store evaluations and purchase intentions,
however, largely hinges upon the retail setting. When confronted
with crowding, shoppers may apply different heuristic decision rules
in a goods vs. a service setting. In a goods setting, a moderately
crowded store proves to generate the most favorable store attitudes
and patronage intentions. However, shoppers seem to be more
tolerant of retail density in a service context, as they, when notpressed for time, tend to favor a crowded store than a less crowded
one.
As there are many options relevant for retailers to generate store
trafc, it seems benecial to identify theconditions under which retail
density should be applied as an instrument for enhancing favorable
store attitudes and patronage behavior. The main result of our
empirical investigations is that retail settings can inuence the
direction of the effect of crowding on store desirability. Retail
executives may nd this information particularly relevant and
interesting, as it could help them determine an optimal crowding
strategy that is most likely to stimulate favorable valuation and
purchase intent.
Another interesting question concerns the effect of the three
crowding levels on store image formation. Perhaps environmental
designs can be devised that could enhance store image and consumer
satisfaction by aiding shoppers to adapt to human and spatial
crowding in situations where store trafc is higher than usual (e.g.,
at peak selling times). Specic decisions such as store layout could
directly affect one's perception of retail density, thereby inuencing
attitudes and behavioral intentions.
Future research may look into the psychological effects of retail
density. Our study suggests different patterns of crowding effects in a
goods and a service setting. A logical continuation of this research is to
investigate consumers' reasoning behind their patronage decisions.
For example, researchers can explore how consumers apply extrinsic
cues to form a value assessment and make a choice decision. This
might be accomplished through the use of open-ended surveys or
interviews with consumers.
Another area of further research is the interaction of productscarcity with retail density effects. Crowdedness in a store can
produce an image of strong demand for a product, which can act as a
double-edged sword. On one hand, consumers may foster a more
favorable attitude, such as high assumed quality, and thereby
increased value of the product. On the other hand, they may perceive
a quantitative scarcity due to demand, as they can view this as an
indication of a high number of other people who will own the same
product. Therefore, a high degree of uniqueness can no longer be
achieved by purchasing this product. An individual may lose their
sense of autonomy or distinctiveness (Vecchio and Sussmann, 1981).
In this case, the negative effect of product scarcity can be dominant.
An empirical testing of how scarcity effects moderate the effect of
retail density on store patronage would provide many interesting
insights.
References
Arnould EJ, Price LL. River magic: extraordinary experience and extended serviceencounter. J Consum Res 1993;20:2445.
Bagozzi R, Yi Y. On the evaluation of structural equation models. J Acad Mark Sci1988;16:7494.
Baker J. Examining the informational value of store environments. In: Sherry JF, editor.ServiceScapes. Chicago: NTC/Contemporary Publishing; 1998. p. 5579.
Baker J, Berry L, Parasuraman A. The marketing impact of branch facility design. J RetailBank 1988:3342 July.
Baker J, Cameron M. The effects of the service environment on affect and consumerperception of waiting time: anintegrative review and research propositions. J AcadMark Sci 1996;24(4):33849.
Baker J, Grewal D, Parasuraman A. The inuence of store environment on qualityinferences and store image. J Acad Mark Sci 1994;22(4):328 59.
Barker R. Ecology and motivation. Nebr Symp Motiv 1963;8:1-50.
Fig. 3.Experiment 2 results for Intention to recommend salon.
Fig. 4.Experiment 2 results for Attitude toward salon.
111Y. Pan, J.C. Siemens / Journal of Business Research 64 (2011) 105112
-
7/21/2019 articol Despre Aglomeratie
8/8
Bateson J, Hui M. The ecological validity of photographic slides and videotapes insimulating the service setting. J Consum Res 1992;19(2):271 81.
Bitner MJ. Evaluating service encounters: the effects of physical surroundings andemployee responses. J Mark 1990;54:6982 April.
Bloom PN. Transmitting signals to consumers for competitive advantage. Bus Hor 1990:5866 July/August.
Eastman ST, Land AM. The best of both worlds: sports fans nd good seats at the bar. JSport Soc Issues 1997;21:15678.
Eroglu S, Harrell G. Retail crowding: theoretical and strategic implications. J Retail1986;62:34663.
Eroglu SA, Machleit KA. An empirical study of retail crowding: antecedents and
consequences. J Retail 1990;66(2):201
20.Eroglu SA, Machleit KA, Barr TF. Perceived retail crowding and shopping satisfaction:the role of shopping values. J Bus Res 2005;58(8):114653.
Fornell C, Larcker DF. Evaluating structural equation models with unobservablevariables and measurement error. J Mark Res 1981;28:39 50 February.
Gierl H, Plantsch M, Schweidler J. Scarcity effects on sales volume in retail. Intl RevRetail, Distrib Consum Res 2008;18(1):4561.
Golden LL, Albaum G, Zimmer M. The numerical comparative scale: an economicalformat for retail image measurement. J Retail 1987;63(4):393410.
Grossbart S, Hampton R, Rammohan B, Lapidus RS. Environmental dispositions andcustomer response to store atmospherics. J Bus Res 1990;21(3):225 41.
Harrell GD, Hutt MD, Anderson JC. Path analysis of buyer behavior under conditions ofcrowding. J Mark Res 1980;17:4551.
Holt D. How consumers consume: a typology of consumption practices. J Consum Res1995;22:1-21.
Jain U. Effects of population density and resources on the feeling of crowding andpersonal space. J Soc Psychol 1987;127(3):1318.
Machleit KA, Kellaris JJ, Eroglu SA. Human versus spatial dimensions of crowdingperceptions in retail environments: a note on their measurement and effect onshopper satisfaction. Mark Lett 1994;5(2):18394.
Machleit KA, Eroglu SA, Powell Mantel S. Perceived retail crowding and shoppingsatisfaction: what modies the relationship. J Consum Psychol 2000;9(1):2942.
Michon R, Chebat JC, Turley LW. Mall atmospherics: the interaction effects of the mallenvironment on shopping behavior. J Bus Res 2005;58:57683.
Nunnally JC, Bernstein IH. Psychometric theory. New York: McGraw-Hill; 1994.Pons F, Laroche M, Mourali M. Consumer reactions to crowded retail settings: cross-
cultural differences between North America and the Middle East. Psychol Mark2006;23(7):55572.
Pons F, Laroche M. Cross-cultural differences in crowd assessment. J Bus Res 2007;60(3):26976.Price LL, Arnould EJ, Deibler SL. Consumers' emotional responses to service encounters:
the inuence of the service provider. Intl J Serv Ind Mgmt 1995;6:3463.Rompay T, Van JL, Galetzka M, Pruyn A, Garcia JM. Human and spatial dimensions of
retail density: revisiting the role of perceived control. Psychol Mark 2008;25(4):31935.
Rustemli A. Crowding effects of density and interpersonal distance. J Soc Psychol1992;132(1):518.
Spears N, Singh SN. Measuring attitude toward the brand and purchase intentions. JCurr Iss Res Advert 2004;26(2):5366.
Stokols D. A typology of crowding experience. In: Baum A, Epstein YM, editors. Humanresponse to crowdingHillsdale, NJ: Earlbaum; 1978. p. 21955.
Vecchio R, Sussmann M. Stafngsufciency and jobenrichment:supportfor an optimallevel theory. J Occup Behav 1981;2(3):17787.
Worchel S. Beyond a commodity theory analysis of censorship: when abundance andpersonalism enhance scarcity effects. Basic Appl Soc Psychol 1992;13(1):7992.
112 Y. Pan, J.C. Siemens / Journal of Business Research 64 (2011) 105112