Impact of music on consumer behaviour

A

i m t m a e a ©

K

i v r s a w t

d g C t

l J

0 d

Journal of Retailing 83 (1, 2007) 115–130

The role of pleasant music in servicescapes: A test of the dual model of environmental perception�

Sylvie Morin a,1, Laurette Dubé b,∗, Jean-Charles Chebat c,2 a Département des Sciences de la Gestion, Université du Québec à Rimouski, 300, Allée des Ursulines, Rimouski, Que., Canada G5L 3A1

b Desautels Faculty of Management, McGill University, 1001, Sherbrooke St. West, Montréal, Que., Canada H3A 1G5 c Écoles des Hautes Études Commerciales, 3000, Chemin de la Côte-Sainte-Catherine, Montréal, Que., Canada H3T 2A7

bstract

This article explores the psychological processes by which background music in servicescapes influences service evaluation and purchase ntention. Drawing on the dual model of environmental perception, it is hypothesized that: (1) servicescape is perceived in an ambient holistic

anner and serves as the background for perceptions of the provider and thereby has direct and provider-mediated effects on outcomes; (2) he presence of music influences outcomes because it reinforces the holistic quality of the servicescape, makes the provider stand out, and

oderates the contribution of provider-mediated servicescape effects; (3) when pleasant music is present, a double-mediating process fully ccounts for the impact of variations in the music valence, such that music valence gets transferred into servicescape attitude, which then

xercises direct and provider-mediated effects on service outcomes. Two experimental studies in real and online environments support the bove research hypotheses.

2006 New York University. Published by Elsevier Inc. All rights reserved.

eywords: Background music; Music valence; Servicescapes; Transactions

R l t 1 2 a i

Introduction

Robust evidence indicates that the presence of music n retail stores and other service environments modulates arious facets of the transactions taking place in those envi- onments. However, most prior work examines music as a ingle environmental parameter and regards its effects as

result of the elicitation and transfer of affects, in line ith typical Stimulus-Organism-Response (S-O-R) models

hat dominate environmental psychology (Mehrabian and

� Authors’ note: The work presented herein was part of the first author’s octoral dissertation. Financial support for this research was provided by rants from the Quebec Fund for Research Support (FCAR) to Jean-Charles hebat and Laurette Dubé and grants from the Social Science and Humani-

ies Research Council of Canada (SSHRC) to Laurette Dubé. ∗ Corresponding author. Tel.: +1 514 398 4026; fax: +1 514 398 3876.

E-mail addresses: sylvie [email protected] (S. Morin), [email protected] (L. Dubé), [email protected] (J.-C. Chebat). 1 Tel.: +1 418 723 1986x1793; fax: +1 418 724 1840. 2 Tel.: +1 514 340 6846; fax: +1 514 340 6432.

t H t f t a H f e ( e ( 1 Y

022-4359/$ – see front matter © 2006 New York University. Published by Elsevie oi:10.1016/j.jretai.2006.10.006

ussell 1974). For example, pleasant music, compared with ess pleasant music, is associated with longer consumption imes (Caldwell and Hibbert 2002; Herrington and Capella 996; Holbrook and Anand 1990; Yalch and Spangenberg 000), shorter time perceptions (Cameron et al. 2003; Kellaris nd Kent 1992), less negative emotional reactions to wait- ng (Hui, Dubé, and Chebat 1997), more favorable attitudes oward the servicescape (Dubé and Morin 2001; North and argreaves 1996), an increased desire to affiliate, more posi-

ive attitudes toward the provider (Dubé et al. 1995), and more avorable service outcomes like evaluation, patronage inten- ions and behaviors (Caldwell and Hibbert 2002; Herrington nd Capella 1996; Hui, Dubé, and Chebat 1997; North and argreaves 1996). Although the bulk of previous research

ocuses on affective responses to music, studies also have xamined objective parameters, such as tempo and volume Milliman 1982, 1986; Smith and Curnow 1966), and the

ffect of the presence versus the absence of music per se Herrington and Capella 1996; Hui et al. 1997; Milliman 982; North and Hargreaves 1996, 1998; North et al. 2000; alch and Spangenberg 1990, 1993).

r Inc. All rights reserved.

1 f Retail

c w m t t t h m s s o a m v w

c c T d t i p T f m m f m t t t t t s t c d l e m c h I r s o

( c o w p t a

1 t m p h a t t

w h t e m q a v o m f m w s G a r o p e i 1

T

c e K p s I r o p a m a m e

16 S. Morin et al. / Journal o

Despite evidence that music influences various service omponents, which combine to shape service outcomes, as ell as these outcomes proper, the mechanisms by which usic operates remain underresearched. We know very lit-

le, for example, about how music as an ambient feature of he environment (Bitner 1992), integrates with space, func- ions, signs, symbols, and artifacts to shape people’s overall olistic perceptions of and responses to physical environ- ents, or servicescapes (Bitner 1990). Moreover, because a

ervice environment entails not only the servicescape but also ervice providers and other customers as core components f its social dimension (Baker and Cameron 1996; Baker et l. 2002), it is extremely unlikely that simple affect transfer echanisms – such as from music to service outcomes – pro-

ide satisfactory accounts of the psychological processes by hich music operates in servicescapes. Early evidence that the effects of music in servicescapes

onstitute more than mere affect transfers to service out- omes emerges from a field study by Dubé and Morin (2001). hese authors explore the possibility that music pleasantness oes not influence store evaluation through a direct affect ransfer but alternatively – or complementarily – by first alter- ng consumer attitudes toward the servicescape and service rovider, which then influence consumers’ store evaluation. heir study finds no support for either a direct affect transfer

rom music to store evaluation as a whole or the influence of usic on attitude toward the provider. Instead, the effect of usic on store evaluation is mediated fully by an affect trans-

er from music to servicescape attitude. In a complex pattern, ore pleasant music improves servicescape attitude, which

hen influences store evaluation both directly and indirectly hrough its beneficial impact on consumers’ attitude toward he provider (i.e., provider attitude), which itself contributes o store evaluation. In addition, follow-up analyses reveal hat this second-order mediation of the effects of music by ervicescape attitude occurs only among subgroups of par- icipants who report the most positive music valence. When onsumers’ music valence is low, servicescape attitude has a irect impact on store evaluation and provider attitude, but the atter does not significantly influence store evaluation. How- ver, this study’s lack of a control condition (no music) and ethodological concerns suggest insufficient experimental

ontrol to assess its contribution to a better understanding of ow music operates as an ambient feature of the servicescape. n addition, it fails to provide a theoretical explanation of its esults beyond asserting that it may be inappropriate to con- ider on the unique and direct impact of music on service utcomes.

We propose a dual model of environmental perception Ohno 1980, 1985, 2000; Ohno and Komuro 1984) as a omprehensive theoretical framework to explain the role f music in service environments and the mechanisms by

hich it affects consumer outcomes. The theoretical under- innings consist of the combination of two theories: Gestalt heory, which pertains to the perceptual processes of ambient nd focal features of the environment (Koffka 1935; Kohler

m w m h

ing 83 (1, 2007) 115–130

929), and Gibson’s theory of affordances, which pertains o the motivational characteristics or valence of an environ-

ent (Gibson 1966, 1979). According to this dual model, a erson’s response to an environment as a whole depends on is or her integration of multisensory information, in which mbient features serve as the background for the focal percep- ion of persons or objects, as well as on the valence attached o this integrated whole.

Drawing on this dual model of environmental perception, e suggest that the servicescape, perceived in an ambient olistic manner, serves as the background for perceptions of he provider and thereby has direct and provider-mediated ffects on outcomes. We also propose that the presence of usic influences outcomes because it reinforces the holistic

uality of the servicescape, makes the provider stand out, nd moderates the contribution of provider-mediated ser- icescape effects. Furthermore, instead of direct transfers f affects to service outcomes, we argue that when pleasant usic is present, a double-mediating process fully accounts

or the impact of variations in the music valence, such that usic valence gets transferred into servicescape attitude, hich then exercises direct and provider-mediated effects on

ervice outcomes. Finally, in line with research that shows estalt perceptions that distinguish between the background

nd foreground occur not only in perception of physical envi- onments but also in abstract domains, such as perceptions f persons based on list of traits (Asch 1946, 1952), we pro- ose that pleasant music influences service outcomes in an -environment in a similar manner. We test these hypotheses n two experimental studies conducted in traditional (Study ) and e-service (Study 2) environments.

Theoretical background and research hypotheses

he dual model of environmental perception

Originally developed through the study of visual per- eptions of buildings and landscapes, the dual model of nvironmental perception (Ohno 1980, 1985, 2000; Ohno and omuro 1984) accounts for people’s inability to process all erceptual information that simultaneously stimulates their ensory channels in an environment (Ittelson 1973, 1976; ttelson et al. 1974). Furthermore, it posits that a person’s esponse to the environment as a whole depends on his r her integration of multisensory information, which the erson processes according to two modes of perception, mbient and focal. Ambient perception occurs in a holistic anner and deals with broader areas. It can be considered preattentive system that warns observers of the key ele- ents that require their attention. Focal perception selectively

liminates unwanted information while enhancing those ele-

ents to which the person pays attention and increasing their eight for guiding the person’s responses to an environ- ent. Although the ambient and focal modes of perception

ave different functions, they interact to provide a perceptual

f Retaili

s m t i a w p

R

T

( G d t s h K o a t p s a f s i

r s p v d a t t a t i s i t a f g o t d 2 1 w c t a T

H a f v

M s

t a t i T e a c o i a o f g t e e p s

r p a p C 1 l N m a f S t v t t T

H e t c i

M

S. Morin et al. / Journal o

ynthesis that reflects a person’s response to the environ- ent. Borrowing from Gibson’s (1966, 1979) affordances

heory, the model further posits that the motivational qual- ty of the environment derives from its valence and other spects of its components. Our research hypotheses, which e present next, stem from this dual model of environmental erception.

esearch hypotheses

he mediating role of provider attitude According to the dual model of environmental perception

Ohno 1980, 1985, 2000; Ohno and Komuro 1984) and the estalt theory from which the perceptual part of the model is erived (Koffka 1935; Kohler 1929), perceiving and reacting o an environment requires a person to process certain of its ensory components in an ambient manner while focusing is or her attention on other objects. Both Koffka (1935) and ohler (1929) argue that the main tenet of Gestalt, regardless f the nature of the experience, is that people view scenes nd events as functional wholes. From a Gestalt perspective, he perceiver blends the diverse features of what he or she erceives and responds to into a coherent, unitary impres- ion that combines the meaning of individual components nd their interrelationships. Thus, service outcomes result rom the integration of the consumer’s response to both the ervicescape and the provider, in which context the latter is nstrumental to fulfill the function of the service transaction.

To address questions regarding the respective and inter- elated contributions of the servicescape and the provider to ervice outcomes, we must theorize about whether each is erceived in an ambient or focal manner. Gestalt theory pro- ides further insights; it states that people organize sensory ata according to principles that optimize structural clarity nd comprehension (i.e., they read a figure against a ground o impose sensory order). In addition, Gestalt theory suggests hat if the environmental conditions can segregate into a larger nd a smaller unit, the smaller one, ceteris paribus, becomes he figure and the larger the ground. In a typical Gestalt exper- ment, two areas are laid out such that one encloses the other, o the enclosing one becomes the ground and is processed n an ambient, holistic manner, whereas the enclosed one is he figure and the object of focal processing. Therefore, in service environment, the physical features that combine to

orm the servicescape should represent the contextual back- round, and the service provider should represent the figure r foreground. That is, even though consumer responses to he servicescape and the service provider each may be imme- iate antecedents of service outcomes (Grace and O’Cass 004; Nguyen and LeBlanc 2002; Wakefield and Blodgett 999), the servicescape serves as the background against hich providers are perceived. If we assume that a figure’s

haracteristics depend on the ground on which it appears, he effect of the servicescape on outcomes must be medi- ted, at least in part, by attitude toward the service provider. herefore, we propose:

a

v f

ng 83 (1, 2007) 115–130 117

1. The servicescape and service provider are perceived ccording to ambient and focal modes, respectively, and the ormer has both direct and provider-mediated effects on ser- ice outcomes.

usic as a moderator of provider mediation of ervicescape effects

The dual model of environmental perception and Gestalt heory indicate that all experiences – smells, sounds, temper- ture, textures – occur in one perceptual space, which implies hat a person can experience an additional feature only after it s established as part of the person’s behavioral environment. hus, when music functions as an ambient feature in a service nvironment, it likely gets integrated into the servicescape nd is processed in an ambient mode. Because ambient pro- essing has a holistic nature, it does not concentrate or focus n any particular component of the background; thus, when ntegrated into the servicescape, music fosters connections mong and unites the elements in the perceptual organization f the background, which reinforces its ability to highlight the oreground. When perceived against such a holistically inte- rated background, the service provider stands out better in he foreground than it would were music not integrated. This ffect should increase the strength of the provider-mediated ffects that the servicescape exercises on service outcomes, rimarily by reinforcing the predictive power of the con- umer’s response to the provider for service outcomes.

Why should the integration of music into the servicescape einforce this predictive power and increase the strength of the rovider-mediated servicescape effects? Research on causal ttribution indicates that salience can determine whether a ossible cause is actually a cause (Choi and Nisbett 1998; hua et al. 2005a; Chua et al. 2005b; Morris and Peng 994). The more salient a possible causal agent is, the more ikely it will be perceived as a cause. For example, Jones and isbett (1972) rely on Gestalt theory to explain why observers ay attribute behavior to personal dispositions rather than to

ctors; in an observer’s perceptual field, the person becomes oregrounded against the background of the social situation. imilarly, a service provider becomes the foreground from

he perspective of an observant consumer. Because the ser- ice provider stands out even more in the presence of music, his increased salience should strengthen the causal attribu- ion of the role the provider plays in the service outcome. herefore, we hypothesize that:

2. The presence (vs. absence) of music has a moderating ffect on the mediating role played by the service provider for he impact of the servicescape on service outcomes. Specifi- ally, the power of the provider to influence service outcomes s stronger in a music-present (vs. absent) condition.

usic valence effects on service outcomes and their

ntecedents

As we mentioned previously, we question whether music alence effects may be accounted for merely by affect trans- ers from music to service outcomes. Instead, on the basis of

1 f Retail

t ( v s o a v c w b 1 G t e s d a p w m p w s w b p

H a o d e

O

u c p s t t m S o v e

M

P

A

p i w 6 p

E

b c a s m a a p v a a d A

v a p o o T t t p p e w r t s e

E

s T r s w i c p

i p

18 S. Morin et al. / Journal o

he dual model of environmental perceptions and Gibson’s 1966, 1979) theory of affordances, we propose that music alence, when it becomes integrated into the servicescape, ignals to the consumer that he or she should approach r distance him- or herself from the service environment, s well as from its physical (servicescape) and social (ser- ice provider) components. In this context, the affordances onveyed by objects, persons, and the environment as a hole consist of those actions made “affordable” or possi- le by the same objects, persons, and environment (Gibson 966, 1979). Although affordances may take different forms, ibson (1966, 1979) argues that the first level of response

o an environment always is affective, such that the direct motional impact of the situation governs the direction of ubsequent relations. That is, it sets the motivational tone and elimits the experiences the consumer expects and seeks. In precursor to affordances theory (Gibson and Crooks 1938, . 455), Gibson writes, “By valences, positive or negative, e refer to the meaning of objects by virtue of which we ove toward some of them and away from others”. We pro-

ose music valence in servicescapes operates in a similar ay. Specifically, variations in music valence make con-

umers perceive the servicescape as more or less favorable, hich, in accordance with the path of influence suggested y H2, influences service outcomes through both direct and rovider-mediated effects. Therefore, we hypothesize:

3. The effect of music valence on service outcomes is ccounted for by a double mediation, such that the effect f music valence on service outcomes is mediated by: (a) the irect effect of the servicescape and (b) the provider-mediated ffect of the servicescape on service outcomes.

Study 1

verview

Study 1 involves a laboratory experiment with video sim- lations of a retail banking service that were constant across onditions except for the background music. Participants role layed a protagonist from whose perspective the video was hot and reported their perceptions regarding their attitudes oward the servicescape and provider, as well as their evalua- ions of the service episode as a whole. When background

usic was present, participants also reported its valence. everal studies of service environments show this method- logical approach to be effective and have good ecological alidity (e.g., Baker et al. 2002; Bateson and Hui 1992; Dubé t al. 1995).

ethod

articipants We recruited students and employees of an eastern North

merican university with ads placed in the university’s news-

u α

a u

ing 83 (1, 2007) 115–130

aper inviting participation in a study on consumer behavior n the service industries. One hundred fifty-three adults (96 omen, 57 men; average age 26.4 years, ranging from 19 to 7 years) participated in the experiment and took part in a rize drawing.

xperimental design and stimuli The experimental design compares music presence to a

aseline or no-music control condition and ensures suffi- ient valence variability in the music-present condition with trilevel blocking factor of low, moderate, and high inten-

ity. The musical excerpts used as background music for each usic blocking level consist of classical pieces selected from pool originally developed by Dubé et al. (1995). Using population similar to that of the study participants, we

retested these musical excerpts in terms of the scope of alence intensity they elicited (means of 3.29, 4.71, and 6.16, ll ps < .05) while controlling for tempo, rhythm, harmony, nd mode as determined by a musical expert. We present a etailed description of the pretest procedures and results in ppendix A. Music played as background to the same baseline 3.5-min

ideo of a bank transaction between a service provider and customer at the cashier’s counter. The film was shot by

rofessionals in a real bank from the perspective of some- ne looking around while waiting in line to be served and bserving the provider–customer interaction at the counter. he video included images of the physical environment and

he interaction between the cashier and the current customer, hen ended as the protagonist, as role played by the partici- ant, was about to be served. The recorded background music layed at the same comfortable volume level throughout the xperiment. Subjects rated the scenarios as fairly consistent ith reality (mean of 5.08), similar to their previous life expe-

iences (mean of 4.96), realistic (mean of 5.09), and providing he impression of being in line (mean of 5.06) on seven-point cales (not at all/very much). We find no differences across xperimental conditions (all ps > .40).

xperimental procedure and measures Participants viewed the experimental video on a television

creen in a quiet room with basic daytime lighting conditions. hey participated in small groups of two to five unacquainted

espondents and were instructed not to interact during the ses- ion. In line with prior research using video simulations, they ere told to imagine themselves in a bank environment wait-

ng in line to be served. At the end of the video, participants ompleted a printed questionnaire about the servicescape and rovider attitudes, and service evaluation.

Participants expressed their attitude toward the phys- cal environment (servicescape) on a three-item, seven- oint scale, anchored by like not at all/very much,

npleasant/pleasant, and disagreeable/agreeable (Cronbach’s = .94). We borrowed these items from attitude measures in

dvertising research (Gorn 1982). We compute the individ- al average scores for use in our subsequent analyses. For

S. Morin et al. / Journal of Retailing 83 (1, 2007) 115–130 119

Table 1 Correlations and means (standard deviations) of measured variables: Studies 1 and 2

1 2 3 4

1. Music valence .48a .29a .29a

3.76 (1.36) 3.56 (1.56) 3.44 (1.45)

2. Servicescape attitude .46a .49a .47a

5.41 (1.26)

3. Provider attitude .20 .42a .69a

5.61 (1.38)

4. Service evaluation/Purchase intention .35a .62a .47a

3.91 (1.78)

N e diago d

t s u P v s 1 t m a

R

o s e s m

S o

o a t s s v t

p s i ( a v a i w t u p T

T

h i B a w s i a e a

otes. Study 1 correlations and means (standard deviations) appear above th iagonal. a Correlation is significant at the .01 level.

heir attitude toward the service provider, participant used the ame scale (Cronbach’s α = .94). We again compute individ- al average scores and use them in our subsequent analyses. articipants also provided their overall assessment of the ser- ice quality provided by the bank (single-item, seven-point cale, very poor/very high service quality; Bolton and Drew 991). Finally, participants in each background music condi- ion reported the valence of their subjective experience tied to

usic on a two-item, seven-point scale: unpleasant/pleasant nd made me feel bad/good (Cronbach’s α = .97).

esults

We provide the descriptive statistics and correlation coefficients f the measured variables in Table 1. All measures demon- trate adequate variability, and none evidence floor or ceiling ffects. The valence intensity reported by participants in the main tudy covers the full distribution spectrum (mean = 3.83, SD = 1.94, edian = 3.50, min = 1.0 and max = 7.0), as we show in Fig. 1.

ervicescape direct and provider-mediated effects on service utcomes

To test H1, we examine whether provider attitude mediates some f the effect of servicescape attitude on service evaluation across ll participants. As recommended by Baron and Kenny (1986), o test for mediation, we conduct a three-step regression analy-

is. In the first step, we regress provider attitude (mediator) on ervicescape attitude (predictor). In the second step, we regress ser- ice evaluation (outcome) on servicescape attitude. Finally, in the hird step, we regress service evaluation on both servicescape and

Fig. 1. Histogram of music valence distribution: Study 1.

c d m g a e d b t

s p a t v i i m

nal; Study 2 correlations and means (standard deviations) appear below the

rovider attitudes. We present the results from this set of analy- es in Table 2, which shows that servicescape attitude significantly nfluences provider attitude (β = .46, p < .001) and service evaluation β = .47, p < .001). When we enter both servicescape and provider ttitudes as predictors, provider attitude significantly affects ser- ice evaluation (β = .59, p < .001), whereas the effect of servicescape ttitude decreases (Sobel test, t = 5.23, p < .001) but remains signif- cant (β = .20, p < .01), indicating a partial mediation. In addition, e observe an increased change in the R2 value when we regress

wo predictors on service evaluation rather than only one (R2 val- es of .22 vs. .50; F[1, 151] = 43.13, p < .001 vs. F[2, 144] = 75.50, < .001 for the models with one vs. two predictors, respectively). hus, we find support for H1.

he moderating role of music presence on provider mediation H2 proposes that the presence (compared with absence) of music

as a moderating effect on the mediating role played by the provider n the impact of the servicescape on service evaluation. Following aron and Kenny (1986), as well as more recent specifications of nalytical approaches to moderated mediation (Muller et al. 2005), e test H2 with a two-phase analytical strategy. We first test the

ignificance of the moderating effect of music presence by defin- ng an interaction between a dichotomous variable (presence vs. bsence of background music) and the significance of the differ- nce between the direct effects of service provider and servicescape ttitudes on service outcomes. Next, using mediational analyses onducted separately for the music-present and music-absent con- itions, we examine the directionality of the moderating effect of usic presence on the mediating processes observed in our investi-

ation of H1 by conducting these analyses separately for the music nd no-music conditions and comparing their relevant parameter stimates. We present the results of these two sets of analyses in etail in the lower panel of Table 2. We first review the results that ear directly on our hypothesis testing and then report the details of he mediational analyses for each condition.

In support of H2, the results from our first phase of analysis how that the presence of music moderates the immediate impact of rovider attitude on service evaluation (t = 1.90, p = .06). Separate nalyses for the music present and absent conditions confirm that

he presence of music strengthens this relationship (B = .72/β = .63 s. B = .42/β = .45 for music present vs. absent). We also exam- ne the relative power of servicescape and provider attitudes as mmediate antecedents of service evaluation in the music and no-

usic conditions. In the absence of music, both service components

120 S. Morin et al. / Journal of Retailing 83 (1, 2007) 115–130

Table 2 Mediational analysis of the effects of servicescape on service evaluation: Study 1

Unstandardized coefficients Standardized coefficients t p Model statistics

B SE β

Overall (n = 153) 1; M = X Constant .19 .27 – 6.82 .000 R2(1, 150) = .22

Servicescape .44 .07 .46 6.41 .000 F = 41.03, p < .001 2; Y = X Constant 1.64 .29 – 5.71 .000 R2(1, 151) = .22

Servicescape .47 .07 .47 6.57 .000 F = 43.13, p < .01 3; Y = XM Constant .42 .27 – 1.59 .114 R2(2, 144) = .50

Servicescape .20 .07 .20 3.11 .002 F = 75.50, p < .01 Provider .63 .07 .59 9.10 .000

No music (n = 43) 1; M = X Constant .55 – 1.49 .143 R2(1, 41) = .40

Servicescape .73 .14 .64 5.26 .000 F = 27.70, p < .001 2; Y = X Constant .91 .52 – 1.77 .084 R2(1, 41) = .40

Servicescape .67 .13 .63 5.22 .000 F = 27.27, p < .001 3; Y = XM Constant .57 .48 – 1.18 .244 R2(2, 40) = .52

Servicescape .37 .15 .35 2.45 .019 F = 21.63, p < .001 Provider .42 .13 .45 3.16 .003

Music (n = 110) 1; M = X Constant 2.20 .31 – 7.10 .000 R2(1, 107) = .16

Servicescape .35 .08 .40 4.45 .000 F = 19.80, p < .001 2; Y = X Constant 1.87 .34 – 5.45 .000 R2(1, 108) = .17

Servicescape .41 .09 .42 4.75 .000 F = 22.56, p < .001 3; Y = XM Constant .25 .33 – .77 .441 R2(2, 106) = .51

.17

.63

N

a ( t a ( i r

fi i ( a a e b i s F p

e ( c e o i fi c e 1 p

t e m m u t o p r t t f m s w

M

p u m t b g w

Servicescape .17 .07 Provider .72 .08

otes. X = servicescape, Y = service evaluation, and M = provider.

re equally powerful immediate antecedents of service evaluation β = .45, p < .02 and β = .35, p < .01; t = .23, p > .10); however, in he presence of background music, the power of provider attitude s an immediate antecedent surpasses that of servicescape attitude β = .63, p < .001 vs. β = .17, p < .05; t = 2.21, p < .02). This find- ng is consistent with the strengthening of the provider–outcome elationship we propose in H2.

Turning to the mediational analyses for each music condition, we nd that in the absence of music, servicescape attitude significantly

nfluences provider attitude (β = .64, p < .001) and service evaluation β = .63, p < .001). In turn, provider attitude represents a significant ntecedent of service evaluation (β = .45, p < .01), and when entered s a second predictor, the effect of servicescape attitude on service valuation diminishes significantly (β = .35, p < .02), as indicated y the results of a Sobel test (t = 2.71, p < .01). Also, we observe an ncreased change in the R2 value when we regress two predictors on ervice evaluation as opposed to only one (R2 values of .40 vs. .52; [1, 41] = 27.27, p < .001 vs. F[2, 40] = 21.63, p < .001, one vs. two redictors, respectively).

In the presence of music, servicescape attitude significantly influ- nces provider attitude (β = .40, p < .001) and service evaluation β = .42, p < .001). In turn, provider attitude represents a signifi- ant antecedent of service evaluation (β = .63, p < .001), and when ntered as a second predictor, the effect of servicescape attitude n service evaluation diminishes significantly (β = .17, p < .02), as ndicated by the results of a Sobel test (t = 3.92, p < .001). This

nding indicates partial mediation. We also observe an increased hange in the R2 value when we regress two predictors on service valuation as opposed to only one (R2 values of .17 vs. .51; F[1, 08] = 22.56, p < .001 vs. F[2, 106] = 55.97, p < .001, one vs. two redictors, respectively).

r t a m o

2.36 .020 F = 55.97, p < .001 8.54 .000

In addition, we conduct a post hoc analysis to examine the rela- ive proportion of the total effect of servicescape attitude on service valuation, as mediated by provider attitude, in the music and no- usic conditions. We calculate the relative contribution of provider ediation to the total effect of servicescape attitude on service eval-

ation, in line with Judge et al. (2006). Our results reveal that, in he absence of music, 45.1 percent of the standardized total effect f the servicescape on service evaluation (β = .63) is mediated by rovider attitude (β = .29 and β = .35 for indirect and direct effects, espectively). In contrast, in the presence of music, 59.8 percent of he standardized total effect of the servicescape on service evalua- ion (β = .42) is mediated by provider attitude (β = .25 and β = .17 or indirect and direct effects, respectively). In conclusion, provider ediation accounts for a greater proportion of the total effect of

ervicescape attitude on service evaluation in the music, compared ith the no-music, condition (z = −1.65, p < .05).

usic valence effects The results we have presented thus far indicate that the direct and

rovider-mediated effects of servicescape attitude on service eval- ation vary as a function of the presence or absence of background usic. H3 also states that, in the presence of background music,

he effect of variations in music valence on service outcomes can e accounted for fully by a double mediation centered on the inte- ration of music valence into servicescape attitude. In testing H2, e already have examined the direct and provider-mediated indi-

ect effects of servicescape on service evaluation for participants in he music condition; to test H3, we perform two additional sets of nalyses with these participants. First, we test whether the effect of usic valence on service evaluation is mediated by the direct effect

f servicescape attitude on the same outcome. Second, we exam-

S. Morin et al. / Journal of Retailing 83 (1, 2007) 115–130 121

Table 3 Regression analysis of mediating role of servicescape in the effects of music valence on service evaluation and provider attitude: Study 1

Unstandardized coefficients Standardized coefficients t p Model statistics

B SE β

Music; service evaluation (n = 110) 1; M = X Constant 2.30 .28 – 8.22 .000 R2(1, 108) = .23

Music .37 .07 .48 5.61 .000 F = 31.45, p < .001 2; Y = X Constant 2.54 .30 – 8.46 .000 R2(1, 108) = .09

Music .22 .07 .29 3.20 .002 F = 10.24, p < .01 3; Y = XM Constant 1.73 .36 – 4.76 .000 R2(2, 107) = .19

Music .10 .08 .13 1.26 .210 F = 12.14, p < .001 Servicescape .35 .10 .36 3.59 .000

Music; provider attitude (n = 110) 1; M = X Constant 2.30 .28 – 8.22 .000 R2(1, 108) = .23

Music .37 .07 .48 5.61 .000 F = 31.45, p < .001 2; Y = X Constant 2.72 .27 – 10.12 .000 R2(1, 107) = .09

Music .20 .06 .29 3.18 .002 F = 10.08, p < .01 3; Y = XM Constant 2.05 .33 – 6.26 .000 R2(2, 106) = .17

Music .09 .07 .14 1.39 .167 F = 10.96, p < .001 .33

N cape.

i a a m ( W

a o o ( a c v i i a p t

i s ( p v t a o . v a t a

a ( t n

d ( ( n χ

l z i m c w z v a n z v t p

u v o w a i O a e ( p t (

Servicescape .29 .09

otes. X = music valence, Y = service evaluation/provider, and M = services

ne whether servicescape attitude plays a similar mediating role in ccounting for the effects that music valence may have on provider ttitude. According to Kenny (2006), such first- and second-order ediational analyses are typical in studies with limited sample sizes

e.g., Morgeson et al. 2005; Stuart and Holtzworth-Munroe 2005). e present our results from these analyses in Table 3. The results from the first set of analyses, examining the medi-

ting role of servicescape attitude in the impact of music valence n service evaluation, reveal that valence is a significant predictor f servicescape attitude (β = .48, p < .001) and service evaluation β = .29, p < .01). When both servicescape and music valence serve s predictors of service evaluation, servicescape attitude signifi- antly influences service evaluation (β = .36, p < .01), but the music alence effects diminish (Sobel test, t = 3.02, p < .01) and shift to nsignificance (p > .20). In addition, we observe an increased change n the R2 value when we regress two predictors on service evaluation s opposed to only one (R2 values of .09 vs. .19; F[1, 108] = 10.24, < .01 vs. F[2, 107] = 12.14, p < .001, one vs. two predictors, respec-

ively). From our analyses of the mediating role of servicescape attitude

n valence effects on provider attitude, we find that music valence ignificantly influences servicescape (β = .48, p < .001) and provider β = .29, p < .01) attitudes. Servicescape attitude in turn influences rovider attitude (β = .33, p < .001), and when we add it to music alence as a predictor, the pleasure coefficient diminishes (Sobel est; t = 2.85, p < .01) to insignificance (p > .15). We again observe n increased change in the R2 value when we regress two predictors n provider attitude as opposed to only one (R2 values of .09 vs. 17; F[1, 107] = 10.08, p < .01 vs. F[2, 106] = 10.96, p < .001, one s. two predictors, respectively). Thus, for both service evaluation nd provider attitude, music valence effects are fully mediated by heir integration into servicescape attitude, in support of both H3a nd H3b.

Although testing for mediation using a sequence of three sep- rate regression equations, in accordance with Baron and Kenny’s 1986) approach (Kenny 2006; Muller et al. 2005), enables us to est the significance of the different parameter estimates, it does ot allow us to examine the simultaneous effects of the indepen-

D

f

3.30 .001

ent variables. To test the model simultaneously, we use EQS version 6.1 for Windows; Bentler 1995) and find good fit indices Bentler–Bonett normed index [BBNI] = .97; Bentler–Bonett non- ormed index [BBNNI] = .99; comparative fit index [CFI] = 1.00, 2 = 25.27, with 22 df, p = .27; residuals distribution: average abso-

ute = .02, average off-diagonal = .03; distribution symmetric and ero centered). The results (see Fig. 2) show that music valence nfluences service evaluation according to the following path:

ore pleasurable music adds to servicescape attitude in a valence- ongruent manner (standard parameter = .51, z = 5.23, p < .001), hich in turn improves provider attitude (standard parameter = .35, = 3.11, p < .01), which then has a powerful influence on ser- ice evaluation (standard parameter = .66, z = 7.72, p < .001). In ddition, more positive servicescape attitude has a marginally sig- ificant direct effect on service evaluation (standard parameter = .14, = 1.62, p = .10). We do not observe any direct effect of music alence on provider attitude or service evaluation. In summary, he results of both our mediational and our structural analyses, as resented in Fig. 2, are almost identical.

We further examine whether the observed mediated processes nderlying music valence effects induce differences in the mean alues of the measures. To this end, we form a posteriori subgroups, n the basis of a median split, of background music participants ho varied in their reports of music valence intensity (means of 2.24

nd 5.73, t[108] = −21.60, p < .001 for low- and high-music valence ntensity subgroups, respectively), then compare the two groups. ur analyses reveal that participants for whom the music induces more pleasurable experience, compared with those who experi-

nce less intense pleasure, report more positive servicescape attitude means of 4.39 and 3.12 for high and low pleasure, t[108] = −4.89, < .001), more positive provider attitude (means of 3.87 and 3.17,

[108] = −2.95, p < .001), and more favorable service evaluation means of 3.78 and 3.07, t[108] = −2.60, p < .01).

iscussion

The results from Study 1 support the predictions derived rom the dual model of environmental perception. That is,

122 S. Morin et al. / Journal of Retailing 83 (1, 2007) 115–130

F nal ana i

t m r v i i a i t p i e p o t i w o n e n a r o e

p p i i a r a s c h e

1 i a ( c i m i g a t d

1 u t i b p T e i o s e a m c s t a i e c

O

ig. 2. Music valence effects on service evaluation—results from mediatio talics and structural modeling in bold. ap < .001; bp < .01; cp = .10.

he effect of servicescape on service evaluation is partly ediated by provider attitude, and the presence of music

einforces the contribution of these provider-mediated ser- icescape effects to service evaluation. The overall increase n provider-mediated servicescape effects is driven primar- ly by an increase in the strength of provider attitude as n antecedent of service evaluation. In fact, the increase n provider-mediated servicescape effects occurs in spite of he fact that the coloring effect of servicescape attitude on rovider attitude is diminished by the presence of music, as s indicated by the significant interaction between this param- ter estimate and the dichotomous variable defined by music resence (t = 2.43, p < .02). When music is present, the impact f its valence on service evaluation is fully accounted for by he proposed double-mediating process, in which music gets ntegrated into the consumer’s response to the servicescape, hich then exercises a direct and provider-mediated effect n service evaluation. Moreover, music valence, when it is ot integrated into servicescape attitude, fails to influence ither provider attitude or service evaluation directly. This ull result replicates Dubé and Morin’s (2001) and Baker nd colleagues’ (2002) null results for such provider attitude elationships, as well as Dubé and Morin’s (2001) finding f a lack of music effect on service outcomes (i.e. store valuation).

Study 2

Because the results from Study 1 generally support our redictions derived from the dual model of environmental erception for the effect of music as an environmental, phys- cally perceived parameter in a service environment, we argue t also is relevant to examine whether these results apply in n e-environment, in which other physical components are educed to the flat sensory field of computer space. Pleas- nt music may operate in the same manner to influence

ervice outcomes in an e-environment, because Gestalt per- eptions that contain background/foreground differentiation ave been observed not only in the perception of physical nvironments but also in abstract domains (Asch 1940, 1946,

a l s

lyses and structural modeling: Study 1. Notes. Mediational analyses are in

952). For example, subjects do not attribute the same mean- ng to the quality of being “intelligent” when they judge it gainst a background of the same person as warm-hearted i.e., wise) versus an alternative background of the person as old (i.e., calculating) (Asch 1946). Similarly, in addition to ts sensorial diversity and immediacy, a service environment ay integrate a broader context (whether physical or virtual)

n which the service provision is embedded. In this case, back- round/foreground distinctions between the e-servicescape nd the e-provider still may exist, which would mean that he same set of psychological mechanisms observed for tra- itional services would apply.

Study 2 also addresses some of the limitations of Study . First, in Study 1, the measure we use for service eval- ation is a single time measure of service quality. Second, he range of variation in valence intensity in Study 1, which ncludes negative valence, may not represent actual practice, ecause no manager would purposefully choose music that eople dislike as the background to a service environment. herefore, in Study 2, we examine whether the dual model of nvironmental perception supports a more realistic variation n valence intensity and use that to induce either moderate r intense pleasure. Third, even though the use of video imulation in research on consumer responses to service nvironments has been proven ecologically valid (Bateson nd Hui 1992), we want to examine whether our proposed odel might be stronger with an experimental design that

ombines the rigor of laboratory experiments with an actual ervice experience in which participants are both exposed o the experimental manipulation of background music nd experience the service provision process themselves nstead of through video simulation. In Study 2, participants ngage in an exchange with providers in different music onditions.

verview

With Study 2, we attempt to reproduce, as accurately s possible, an actual e-service experience in a controlled aboratory environment and create experimental conditions imilar to Study 1—namely, the presence versus absence of

f Retaili

m m fi v F t p i M r t P m

M

P

m c p W a l h 5 c t o i p b

E

t w o ( o S p C b i e i o t f v A

E

c

e w T f a b v a b c p t m t p t ( t c c

a v s w w “ w ( W a

R

t v s r S

M

p t T c i a a s n w

S. Morin et al. / Journal o

usic and a blocking factor to ensure the variability of the usic valence, albeit with a more limited scope. The e-service rm is an existing retail store with online catalog sales ser- ices that was relatively unknown to the study population. or the purpose of our study, the store added a chat room

o its Web catalog to offer the services of an online service rovider who interactively assisted the participant in purchas- ng a gift for his or her friend through real-time exchanges via

SN Messenger, version 6.2. As in Study 1, all participants eported their servicescape and provider attitudes, as well as heir purchase intention as a measure of service outcomes. articipants in the music conditions also provided ratings of usic valence.

ethod

articipants We recruited one hundred thirty adults (68 women, 62

en; average age 26.9 years) from a major North Ameri- an city with ads placed in a local newspaper that invited eople to participate in a study on consumer behavior on the orld Wide Web. Participants received CAN$15 as a token of

ppreciation for completing the experimental session, which asted an average of 45 min. Overall, the participants had a igh level of experience with the Web or Internet (mean of .85 on seven-point scale, not at all/very much) and online hatting (mean of 4.55). They also reported shopping online o a moderate degree (mean of 3.24). Most had never heard f the service firm used in this study (74.6 percent) or vis- ted its brick-and-mortar outlets (89.2 percent). None of the articipants reported ever visiting the experimental Web site efore.

xperimental design and stimuli Similar to Study 1, we use an experimental design

hat compares a no-music to a music condition, which e designed to ensure sufficient variability in the valence f musical excerpts according to a bilevel blocking factor i.e., moderate and high). We selected the musical excerpts n the basis of a pretest similar to the one we used for tudy 1, but this time, the music, from Stéphane Pom- ougnac CDs (Hôtel Costes, Saks Fifth Avenue, and Hôtel ostes 5), was of a type more frequently encountered as ackground music in service firms. Using a population sim- lar to the participants in Study 2, we pretested musical xcerpts to obtain two CD-recorded music tracks that var- ed in the intensity of their valence (means of 4.29 and 5.55 n seven-point scale, t[18.45] = −2.40, p < .05) while con- rolling for tempo, as determined by objective measures, and amiliarity, as determined by subjects’ evaluations. We pro- ide the detailed procedures and results for this pretest in ppendix B.

xperimental procedure and measures Participants progressed through the experimental pro-

edure one step at a time. They first looked through the

t u p p

ng 83 (1, 2007) 115–130 123

xperimental Web site and browsed the Internet catalog ith the objective of finding a gift for their best friend. he Web site featured primarily decorative objects, items

or the kitchen, furniture, lighting, bathroom items, office ccessories, and other gadgets. After the participants finished rowsing the Web site by themselves, they communicated ia chat exchanges with the online service provider, who ssisted them on a real-time basis in finding a gift for their est friend. The online provider’s suggestions were based on ustomized information provided by the participants about articular characteristics of their best friend, the type of gift hey wanted or both. The online advisor had access to a

atrix of selective product offerings developed and pretested o assist her. The same experimenter, who was blind to the articipants’ experimental conditions, offered online advice o all participants. During the entire experimental session i.e., both individual exploration and provider exchanges), he Web site broadcast background music in the music onditions and remained silent in the baseline, no-music ondition.

Except for service evaluation, the measures in this study re the same as those in Study 1. However, we measure ser- ice outcomes in terms of purchase intention on a four-item cale adapted from Eroglu et al. (2001): “Given a choice, I ould probably not go back and visit this website again,” “I ould definitely recommend this website to other people,”

I would definitely add this website to my favorites”, and “I ould definitely go and visit one of their stores in the future”

seven-point scale; not at all/very much; Cronbach’s α = .86). e compute individual average scores and use them in the

nalyses.

esults

We present the means, standard deviations, and correlations of he measures from Study 2 in the lower diagonal of Table 1. The alence intensity reported by participants in the main study includes ufficient variability and covers the spectrum of the distribution, anging from moderate to high valence intensity (mean = 5.21, D = 1.51, median = 5.71).

ediating role of provider attitude in servicescape effects We use the same analyses as in Study 1 to examine whether

rovider attitude mediates some of the effect of servicescape atti- ude on purchase intention. As we show in the upper panel of able 4, these analyses reveal that servicescape attitude signifi- antly influences provider attitude (β = .42, p < .001) and purchase ntention (β = .62, p < .001). When we enter both servicescape nd provider attitudes as predictors, provider attitude significantly ffects purchase intention (β = .25, p < .01), whereas the effect of ervicescape lessens (Sobel test: t = 5.23, p < .001) but remains sig- ificant (β = .52, p < .001), indicating partial mediation. In addition, e observe an increased change in the R2 value when we regress

wo predictors on purchase intention as opposed to only one (R2 val- es of .39 vs. .44; F[1, 128] = 80.48, p < .001 vs. F[2, 127] = 49.45, < .001, one vs. two predictors, respectively). Thus, H1 is sup- orted.

124 S. Morin et al. / Journal of Retailing 83 (1, 2007) 115–130

Table 4 Mediational analyses of the effects of servicescape on purchase intention: Study 2

Unstandardized coefficients Standardized coefficients t p Model statistics

B SE β

Overall (n = 130) 1; M = X Constant 3.40 .43 − 7.82 000 R2(1, 128) = .17

Servicescape .42 .08 .42 5.17 000 F = 26.69, p < .01 2; Y = X Constant −.02 .46 − −.03 .974 R2(1, 128) = .39

Servicescape .77 .09 .62 8.97 000 F = 80.48, p < .001 3; Y = XM Constant −1.06 .54 – −1.97 .051 R2(2, 127) = .44

Servicescape .65 .09 .52 7.07 .000 F = 49.45, p < .001 Provider .31 .09 .25 3.42 .001

No music (n = 38) 1; M = X Constant 3.55 .94 – 3.76 .001 R2(1, 36) = .12

Servicescape .38 .17 .35 2.24 .032 F = 5.00, p < .05 2; Y = X Constant −1.66 .89 – −1.87 .070 R2(1, 36) = .53

Servicescape 1.03 .16 .73 6.42 .000 F = 41.19, p < .001 3; Y = XM Constant −2.33 1.05 – −2.23 .032 R2(2, 35) = .55

Servicescape .96 .17 .68 5.63 .000 F = 21.56, p < .001 Provider .19 .16 .14 1.20 .239

Music (n = 92) 1; M = X Constant 3.33 .49 – 6.77 .000 R2(1, 90) = .20

Servicescape .44 .09 .44 4.70 .000 F = 22.06, p < .001 2; Y = X Constant .47 .53 – .87 .386 R2(1, 90) = .35

Servicescape .70 .10 .59 6.91 .000 F = 47.78, p < .001 3; Y = XM Constant −.76 .62 – −1.22 .226 R2(2, 89) = .42

Servicescape .54 .11 .45 5.04 .000 F = 32.43, p < .001 .31

N

M

t f c l a m p p n r a d p i p p t i t p p

m m t H e w

i i a p

i t a e o S p v o p r

t p a t p i n 2

Provider .37 .11

otes. X = servicescape, Y = purchase intention, and M = provider.

oderating role of music presence on provider mediation As in Study 1, we conduct analyses with dummy variables to

est the moderating role of music on the preceding relationships, ollowed by analyses conducted separately according to the music ondition. We present the results in the lower panel of Table 4. In ine with H2, the parameter for the relationship between provider ttitude and purchase intention differs between the music and no- usic conditions (t = −1.97, p < .05). Also consistent with H2, the

resence of music strengthens the impact of provider attitude on urchase intention (B = .37/β = .31 vs. B = .19/β = .14 for music vs. o-music conditions, respectively). As in Study 1, we examine the elative power of servicescape and provider attitudes as immediate ntecedents of purchase intention in the music and no-music con- itions. This time, in both the presence and absence of music, the ower of provider attitude as an immediate antecedent of purchase ntention is inferior to that of servicescape (music absent, β = .14, > .20 vs. β = .68, p < .001; t = 3.34, p < .002; music present, β = .31, < .001 vs. β = .45, p < .01; t = 1.31, p = .09). However, the predic-

ive power differential between servicescape and provider attitudes s three times smaller when music is present versus absent. Thus, hough in a weaker manner than in Study 1, we again find sup- ort for H2 and argue that the presence of music strengthens the rovider–outcomes relationship.

With regard to the mediating processes observed for the usic and no-music conditions, we find that, in the absence of

usic, servicescape attitude significantly influences provider atti-

ude (β = .35, p < .05) and purchase intention (β = .73, p < .001). owever, provider attitude in the absence of music in the e-

nvironment of Study 2 fail to influence purchase intention (p > .20), hich suggests that the effects of servicescape attitude on purchase

o ( p v i

3.39 .001

ntention are mediated by provider attitude. We observe no change n the R2 value when we regress two predictors on purchase intention s opposed to only one (R2 values of .53 vs. .55; F[1, 36] = 41.19, < .001 vs. F[2, 35] = 21.56, p < .001).

In the presence of music, servicescape attitude significantly nfluences provider attitude (β = .44, p < .01) and purchase inten- ion (β = .59, p < .001). In turn, provider attitude offers a significant ntecedent of purchase intention (β = .31, p < .01), and when ntered as a second predictor, the effect of servicescape attitude n purchase intention diminishes according to the results of the obel test (t = 2.76, p < .01), but it remains significant (β = .45, < .001). Moreover, we observe an increased change in the R2

alue when we regress two predictors on purchase intention as pposed to only one (R2 values of .35 vs. .42; F[1, 90] = 47.78, < .001 vs. F[2, 89] = 32.43, p < .001 for one vs. two predictors,

espectively). As we did for Study 1, we conduct a post hoc analysis to examine

he relative proportion of the total effect of servicescape attitude on urchase intention that is mediated by provider attitude in the music nd no-music conditions. Then, we calculate the relative contribu- ion of provider mediation to the total effect of servicescape on urchase intention for the music condition only (we observe no ndication of provider mediation in the absence of music). While o mediation is found in the absence of music, when it is present, 3 percent of the standardized total effect of servicescape attitude

n purchase intention (β = .59) is mediated by provider attitude indirect β = .14; direct β = .45). This post hoc analysis implies that rovider mediation accounts for a larger proportion of the total ser- icescape effects on purchase intention in the music condition than n the no-music condition (z = −5.73, p < .001).

S. Morin et al. / Journal of Retailing 83 (1, 2007) 115–130 125

Table 5 Regression analysis of the mediating role of servicescape on the effects of music valence on purchase intention and provider attitude: Study 2

Unstandardized coefficients Standardized coefficients t p Model statistics

B SE β

Music; purchase intention (n = 92) 1; M = X Constant 2.99 .45 – 6.63 .000 R2(1, 90) = .21

Music .40 .08 .46 4.84 .000 F = 23.43, p < .001 2; Y = X Constant 2.13 .57 – 3.75 .000 R2(1, 90) = .12

Music .37 .10 .35 3.52 .001 F = 12.36, p < .01 3; Y = XM Constant .19 .60 – .32 .751 R2(2, 89) = .36

Music .11 .10 .10 1.05 .296 F = 24.47, p < .001 Servicescape .65 .11 .54 5.68 .000

Music; provider attitude (n = 92) 1; M = X Constant 2.99 .45 – 6.63 .000 R2(1, 90) = .21

Music .40 .08 .46 4.84 .000 F = 23.43, p < .001 2; Y = X Constant 4.65 .49 – 9.48 .000 R2(1, 90) = .04

Music .17 .09 .20 1.92 .058 F = 3.69, p < .10 3; Y = XM Constant 3.34 .55 – 6.06 .000 R2(2, 89) = .20

−.00 .45

N cape.

M

o i p o b m p

p i v a e ( v t F

n p

e m ( I w . t e a H

E S p o i s p

F i

Music −.00 .09 Servicescape .44 .11

otes. X = music valence, Y = purchase intention/provider, and M = services

usic valence effects We examine the direct and provider-mediated indirect effects

f servicescape attitude on purchase intention among participants n the music conditions for our test of H2; to test H3, we again erform two additional sets of analyses to test whether the effect f music valence on purchase intention is mediated, in full or part, y servicescape attitude and whether servicescape attitude plays a ediating role in accounting for the effects of music valence on

rovider attitude. We present the results in Table 5. Our first set of analyses reveals that valence is a significant

redictor of servicescape attitude (β = .46, p < .001) and purchase ntention (β = .35, p < .001). When both servicescape and music alence serve as predictors, servicescape attitude significantly ffects purchase intention (β = .54, p < .001), whereas the valence ffects diminish (Sobel test: t = 3.69, p < .01) to insignificance p > .20). In addition, we observe an increased change in the R2

alue when we regress two predictors on purchase intention rather han only one (R2 values of .12 vs. .36; F[1, 90] = 12.36, p < .001 vs.

[2, 89] = 24.47, p < .001 for one vs. two predictors, respectively).

From our second set of analyses, we find that valence sig- ificantly influences servicescape attitude (β = .46, p < .001) and rovider attitude (β = .20, p < .06). Servicescape attitude also influ-

m m t p

ig. 3. Music valence effects on purchase intention—results from mediational ana talics and structural modeling in bold. ap < .001; bp < .01; cp = .10.

−.04 .971 F = 10.91, p < .001 4.18 .000

nces provider attitude (β = .45, p < .001), and when we add it to usic valence as a predictor, the valence coefficient diminishes

Sobel test: t = 3.17, p < .01) and shifts to insignificance (p > .95). n addition, we observe an increased change in the R2 value when e regress two predictors on provider attitude (R2 values of .12 vs.

36; F[1, 90] = 12.36, p < .001 vs. F[2, 89] = 24.47, p < .001, one vs. wo predictors, respectively). Thus, in both cases, the music valence ffects on provider attitude and purchase intention are fully medi- ted by their integration into servicescape attitude, in support of 3.

As we did in Study 1, to test the model simultaneously, we use QS. The fit indices are satisfactory but not as good as those from tudy 1 (BBNI = .92; BBNNI = .96; CFI = .97, χ2 = 73.85, 49 df, < .05; residuals distribution: average absolute SD = .07, average ff-diagonal = .07, distribution slightly asymmetric, right). Again as n Study 1, our results (see Fig. 3) suggest that elicited music valence ets the stage for provider attitude and maximizes its impact on urchase intention along the following primary path: more intense

usic valence adds to servicescape attitude in a valence-congruent anner (standard parameter = .42, z = 3.88, p < .001), which in

urn improves provider attitude (standard parameter = .42, z = 3.57, < .01), which then has a significant influence on purchase intention

lyses and structural modeling: Study 2. Notes. Mediational analyses are in

1 f Retail

( i o W a

s a 4 v m i v m p p

D

o A c t m c m p e 1 d a w f t t d F i i t p

e o a g s e w a o a

t t a w m t g t t w e S l t v s m s s f a f p

e b n s t b a a o t a o t 2

h l s v d s m d v t p e r

26 S. Morin et al. / Journal o

standard parameter = .33, z = 3.17, p < .01). In addition, more pos- tive servicescape attitude has a marginally significant direct effect n purchase intention (standard parameter = .45, z = 3.50, p < .001). e do not observe any direct effect of music valence on provider

ttitude or purchase intention. We again examine whether the observed mediating processes are

uccessful in influencing means. We find that in subgroups, formed posteriori on the basis of a median music valence split (means of .06 and 6.32, t[51.91] = −10.64, p < .001, for moderate- and high- alence intensity subgroups, respectively), respondents who report ore pleasurable, compared with less pleasurable, experiences

nduced by the background music also report more positive ser- icescape attitude (means of 5.44 and 4.73, t[90] = −2.64, p < .01), ore positive provider attitude (5.87 and 5.24, t[90] = −2.34, < .05), and stronger purchase intention (4.56 and 3.49, t[90] = 3.38, < .001).

iscussion

The results from Study 2 generally support the dual model f environmental perception in the context of an e-service. cross all participants, the effect of the servicescape on pur-

hase intention is partly mediated by provider attitude, and his mediating process is sensitive to the moderating effect of

usic. In the absence of music, servicescape effects on pur- hase intention are not mediated by provider attitude. When usic is present, the predictive power of provider attitude for

urchase intention grows stronger, and the provider-mediated ffect of servicescape reaches significance. In line with Study , the increase in provider-mediated servicescape effects is riven primarily by the increase in the strength of provider ttitude as an antecedent of service evaluation. However, hile in Study 1 this effect was obtained in spite of the

act that the coloring effect of servicescape on provider atti- ude was diminished by the presence of music, in Study 2, he coloring effect of the servicescape on provider attitude oes not vary between the music and no-music conditions. inally, the path by which music valence influences purchase

ntention is similar to that observed for traditional services n Study 1; that is, the effect is fully mediated by its integra- ion into servicescape attitude, which in turn has direct and rovider-mediated effects on purchase intention.

General discussion

The results of our two studies, conducted in traditional and -service environments, support the validity of the dual model f environmental perception as a comprehensive theoretical pproach to map the psychological processes by which back- round music in servicescapes influences service outcomes, uch as service evaluation and purchase intention. With a few xceptions that do not bear directly on our tested hypotheses,

e find similar results in both studies; furthermore, at the

ggregate level, when we include all participants regardless f the music background conditions, provider attitude serves s a partial mediator of servicescape effects. In both studies,

t r i e

ing 83 (1, 2007) 115–130

he moderating role of music presence on this relationship ranslates into a stronger provider–outcomes relationship nd more significant provider-mediated servicescape effects hen music is present. In Study 1, we observe provider- ediated servicescape effects even in the absence of music,

hough of weaker magnitude. In Study 2, without back- round music, the provider effect is not powerful enough o influence service outcomes; rather, the direct effects of he servicescape on outcomes appear to do the bulk of the ork in terms of influencing service outcomes. The param-

ter is higher in the no-music than the music conditions in tudy 2, and we observe no music presence effects at this

evel in Study 1. Across both studies, when music is present, he path by which valence intensity variations influence ser- ice outcomes necessarily involve a first-level integration into ervicescape attitude, which then has direct and provider- ediated impacts on service outcomes. Our results therefore

uggest that prior reports of direct effects of music valence on ervice evaluation or provider attitude (e.g., Dubé et al. 1995 or provider attitude; Caldwell and Hibbert 2002; Herrington nd Capella 1996 for service evaluation) may have been con- ounded by their lack of consideration of the mediating role layed by servicescape.

Can these results be accounted for by alternative mod- ls? We do not believe so. For example, they clearly cannot e accounted for by the traditional S-O-R models, which do ot provide any rationale for why music would have affected ervicescape and provider attitudes differently. According to hese models, if affect transfers occur, they should occur for oth the servicescape and the provider, indiscriminately. This bsence of discrimination in the music effects on servicescape nd provider attitudes might also have been expected from ther existing models of environmental perception that posit ransfers other than affect, such as inference from over- ll perceptions of the physical environment (Bitner 1990) r an overall image from the environment to provider atti- ude (Greenland and McGoldrick 1994; Sharma and Stafford 000).

Instead, our pattern of results appears consistent with the ighly differentiated psychological mechanisms that under- ie the dual model of environmental perception. Across both tudies, the increase in provider-mediated effects of ser- icescape on service outcomes in the presence of music is riven by the strengthening of the provider–outcome relation- hip, a hypothesis predicated on the idea that the presence of usic changes the background/foreground interplay, which

etermines how a person perceives and responds to a ser- ice environment. These results provide novel insights into he role of music in servicescapes and support an earlier roposition by Bitner (1992), who stated that for specific nvironment parameters such as music to affect consumers’ esponses, they first must contribute to Gestalt or holis-

ic perceptions of the physical environment. We provide a easonable account for such mechanisms for the first time n services literature. However, additional research should xamine whether the dual model of environmental perception

f Retaili

a a o h t i m s

I

n o l o p o b m t i t v T p i r o t c o a r l d p m T a l p fi

e q i S t c d p a l W t

s i s o

L

t m s a f d t m r b o 2

t s s m v s w c s t

s o p e s o c f a t e

v a n m i p T m

S. Morin et al. / Journal o

lso accounts for the integration of other ambient cues, such s lighting or smell. The dual model posits that the presence f music operates at the perceptual level by contributing to the olistic quality of the servicescape and creating a greater con- rast against which the provider stands out more powerfully; t therefore is reasonable to assume that other ambient factors

ight operate according to similar mechanisms. This suppo- ition remains an empirical question for further research.

mplications for retailing practice

In retailing and other service organizations, design, engi- eering, and management decisions are made at the level f single environmental parameters such as music, color, ighting while the effects of these on service outcomes rely n consumers’ holistic perceptions of the servicescape. The resent results mapping the processes underlying how music perates contribute a much needed theoretical and empirical asis for improving such decisions. They show that decisions ade at the level of specific environment parameters, such as

he presence of pleasant background music, once integrated nto consumer’s holistic perception of the servicescape, have he potential to set the stage for creating greater customer alue through improved attitude toward a given provider. hat is, decisions related to the physical environment and rovider performance are not independent; however, either nterference or synergies is equally possible. Our research ecommends more systematic and evidence-based planning f service and e-service environments, based on the effec- iveness of each parameter of the physical environment to reate value for consumers and contribute to operational r market outcomes. Yet, in most service firms, facilities nd human resources/operations management remain sepa- ate and receive attention at different points during the firm’s ifecycle. Therefore, our results also call for a dramatically ifferent approach that balances physical and human com- onents in both strategic planning and everyday decision aking, which will translate into higher service outcomes. his shift may lead to breakthrough developments in design nd management that enable firms to provide traditionally abor-intensive services in a more reliable, broader, and otentially less costly manner, which benefits both service rms and consumers.

Our results also provide insights into the design of e- nvironments. Prior research had shown that the atmospheric ualities of online retailing, despite possible differences n their nature and intensity, conform to well-established -O-R models (Eroglu et al. 2001, 2003). Our results show

hat the more differentiated dual model of environmental per- eption also applies to both real and e-environments. The ifferentiation that occurs between the servicescape and the rovider on the basis of their ground-figure qualities and

mbient-focal processing at the perceptual and evaluative evels has important functional significance. As observed by

atson and colleagues (2004, p. 40), “people are using elec- ronic networks to become and stay informed, to perform

b o s

ng 83 (1, 2007) 115–130 127

ervices (banking, redeeming air miles, trading stocks), to nteract with private and public institutions, to entertain them- elves.” Therefore, our results can help improve the design f e-environments and their market potential.

imitations and future research

Our results must be interpreted in light of the limi- ations of this research. First, we acknowledge common

easurement bias concerns; in both studies, we measure ervicescape and provider attitudes with identical pairs of djectives. Additional research therefore should examine dif- erent operationalizations of these constructs. However, we o not believe that this potential bias presents a significant hreat to the validity of our results, because similar measure-

ent approaches have been used extensively in advertising esearch on music (e.g., Gorn 1982), and the correlations etween the measures in both studies are comparable to those bserved in similar environmental studies (e.g., Baker et al. 002).

Second, we undertake a double objective of considering he effects of both the presence of music and its plea- ure intensity. Therefore, we suffer an unbalanced sample ize to compare the baseline and music conditions, which ay influence the lesser impact of provider attitude on ser-

ice outcomes in the baseline conditions. Further research hould replicate our results with a balanced design; however, e again believe this limitation cannot explain the theory-

onsistent differences observed, because not every link is less ignificant in the baseline compared with the music condi- ions.

Third, though our results across the two studies are con- istent with our hypotheses derived from the dual model f environmental perception, we do not examine the pro- osed interplay of ambient and focal processes using formal xperimental manipulations, nor do we perform precise mea- urements of these processes. This limitation should be the bject of additional studies designed to measure more pre- isely or manipulate the two core concepts of ambient and ocal processing, such as by using tracking devices of visual ttention. Further work also should manipulate the conditions o favor one or another mode of processing and identify mod- rators of their respective contributions to service outcomes.

Fourth, across both studies, both the provider and the ser- icescape attributes were constant across music conditions nd were perceived by participants as all being relatively eutral. It is not clear whether the mechanisms by which usic operates would remain the same if it were integrated

nto more extreme instances, such as if either (or both) the rovider and the servicescape were very poor or very good. his question should be examined in further research that anipulates these factors.

Fifth, we conducted our studies within a single cultural

ackground. Some of the mechanisms we propose for the peration of music in the servicescape rely on the increased alience of the provider in the presence of music, which

1 f Retail

i t s t i N I m o A

i d t r l m

A S

D c o m e o i O i r t t w t a a a s v b e a n 4 s c o t m a

B S

S H b c i h c v b i t e a o o v f h 1 t p (

m o D f d t t

A

A

A

B

B

28 S. Morin et al. / Journal o

ncreases the causal attribution of this service component and herefore its impact on service outcomes. All subjects in our tudies belong to Western cultures, which tend to be biased oward focal processing, unlike the predisposition observed n Eastern cultures toward contextual information (Choi and isbett 1998; Chua et al. 2005a,b; Morris and Peng 1994).

t would be interesting to provide a stronger test of the dual odel of environmental perception by determining whether

ur observed effects can be replicated with participants from sian cultures. Despite these limitations, the insights we provide regard-

ng how music operates in servicescapes should help esigners and managers develop a more integrative approach o the decisions they make with respect to their service envi- onment, their production, and their consumers, which should ead to superior operational, market, and financial perfor-

ance.

Appendix A

.1. Selection and validation of experimental stimuli in tudy 1

We selected musical excerpts from a pool developed by ubé et al. (1995) to vary in valence but still enable us to

ontrol for arousal and familiarity, as well as for a series f objective parameters, including tempo, rhythm, harmony, ode, pitch, and musical volume, as determined by a musical

xpert. The three pretested excerpts were the first movements f the following pieces: (1) Brahms, Piano and Cello Sonata n F Major, (2) Haydn, String Quartets no. 3 in C Major p. 76 (“Emperor”), and (3) Mozart, Horn Concerto no. 3

n E Flat Major K. 447. We hypothesized they would yield, espectively, a negative, neutral, and positive valence. Four- een participants completed the pretest by listening to the hree musical excerpts in different listening sessions in which e counterbalanced the order of the pieces. Participants rated

heir pleasure intensity using the Affect Grid (Russell et l. 1989), a two-dimensional grid that captures the valence nd arousal elicited by different environments (Holbrook nd Anand 1990). We measured familiarity using three-item, even-point scale: not at all/very familiar, never heard/heard ery often, and never listened to/listened to regularly (Cron- ach’s α = .88). Pleasure intensity differed according to our xpectations; the respective means for the Brahms, Haydn, nd Mozart excerpts were 3.29, 4.71, and 6.16, and they sig- ificantly differed (t = 1.88, 45 df, p < .05 for 1 vs. 2; t = 1.99, 5 df, p < .05 for 2 vs. 3). The three musical excerpts all had low tempos (66 beats/min [BPM] or less), and their per- eived level of arousal remained relatively constant (means

f 2.07, 2.64, and 2.58, p > .10) and strongly correlated with he excerpt’s musical tempo. Overall, familiarity with each

usical excerpt was weak to moderate (means of 2.54, 2.44, nd 2.55, p > .10).

B

ing 83 (1, 2007) 115–130

Appendix B

.1. Selection and validation of experimental stimuli in tudy 2

We selected 45 musical excerpts from three albums by téphane Pompougnac (Hôtel Costes, Saks Fifth Avenue, and ôtel Costes 5) that possessed constant music characteristics ut different levels of induced valence. His style of music an be classified as house, lounge, or downtempo music and s played in many retail outlets, especially those that offer ousehold design objects. We conducted two pretests to, first, hoose comparable musical excerpts that vary in terms of the alence induced and, second, validate the two musical CDs to e used as experimental stimuli. In the first pretest, ten partic- pants from the same population as the main study evaluated he 45 musical excerpts, and on the basis of their subjective valuation, we chose fifteen excerpts (seven for the moderate- nd eight for the high-valence intensity music). The excerpts f both selections offered different levels of valence (means f 3.30 and 5.82 on seven-point scale for moderate and high alence intensity; p < .05) and were comparable in terms of amiliarity and perceived stimulation. More specifically, they ad an intermediate musical tempo (means of 112.78 and 19.41 BPM; p > .10), the participants were not familiar with he excerpts (means of 1.69 and 2.00 on seven-point scale; > .10), and they induced a moderate level of stimulation

means of 4.07 and 5.23 on seven-point scale; p > .10). The second pretest, involving 24 participants, validated the

usic valence at moderate and high intensity levels for each f the 2 CDs, to which participants were randomly assigned. ifferent levels of pleasure emerged (means of 4.18 and 5.39

or moderate and high valence intensity; t = −2.47, 12.41 f, p < .05) when we control for familiarity (2.67 and 3.25; = −.85, 22 df, p > .40) and attention induced (5.33 and 4.75; = .63, 22.41 df, p > .60).

References

sch, Solomon E. (1940). “Studies in the Principles of Judgments and Atti- tudes. II. Determination of Judgments by Group and by Ego-Standards,” Journal of Social Psychology, 12 433–465.

sch, Solomon E. (1946). “Forming Impressions of Personality,” Journal of Abnormal and Social Psychology, 41 258–290.

sch, Solomon E. (1952). Social Psychology, Englewood Cliffs, NJ: Prentice-Hall.

aker, Julie and Michaelle Cameron (Fall 1996). “The Effects of the Service Environment on Affect and Consumer Perception of Waiting Time: An Integrative Review and Research Propositions,” Journal of the Academy of Marketing Science, 24 338–349.

aker, Julie, A. Parasuraman, Dhruv Grewal and Glenn B. Voss (April 2002). “The Influence of Multiple Store Environment Cues on Perceived Merchandise Value and Patronage Intentions,” Journal of Marketing, 66

120–141.

aron, Reuben M. and David A. Kenny (1986). “The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,” Journal of Personality and Social Psychology, 51 (6) 1173–1182.

f Retaili

B

B

B

B

B

C

C

C

C

C

D

D

E

E

G

G

G

G

G

G

H

H

H

I

I

I

J

J

K

K

K

K M

M

M

M

M

M

N

N

N

N

O

O

S. Morin et al. / Journal o

ateson, J.E.G. and Michael K. Hui (September 1992). “The Ecological Validity of Photographic Slides and Videotapes in Simulating the Service Setting,” Journal of Consumer Research, 19 271–281.

entler, P.M. (1995). EQS Structural Equation Program Manual, Encino, CA: Multivariate Software.

itner, Mary Jo. (April 1990). “Evaluating Service Encounters: The Effects of Physical Surroundings and Employee Responses,” Journal of Mar- keting, 54 69–82.

itner, Mary Jo (April 1992). “Servicescapes: The Impact of Physical Sur- roundings on Customers and Employees,” Journal of Marketing, 56 57–71.

olton, Ruth and James H. Drew (March 1991). “A Multistage Model of Customers’ Assessments of Service Quality and Value,” Journal of Con- sumer Research, 17 375–384.

aldwell, Clare and Sally A. Hibbert (November 2002). “The Influence of Music Tempo and Musical Preference on Restaurant Patrons’ Behavior,” Psychology and Marketing, 19 895–917.

ameron, Michaelle Ann, Julie Baker, Mark Peterson and Karin Brauns- berger (June 2003). “The Effects of Music, Wait-Length Evaluation, and Mood on a Low-Cost Wait Experience,” Journal of Business Research, 56 421–430.

hoi, Incheol and Richard E. Nisbett (1998). “Situational Salience and Cul- tural Differences in the Correspondence Bias and Actor-Observer Bias,” Personality and Social Psychology Bulletin, 24 949–960.

hua, Hannah Faye, Julie E. Boland and Richard E. Nisbett (2005). “Cultural Variation in Eye Movements During Scene Perception,” Proceedings of the National Academy of Sciences, 102 (35) 12629–12633.

hua, Hannah F., Janxin Leu and Richard E. Nisbett (2005). “Culture and Diverging Views of Social Events,” Personality and Social Psychology Bulletin, 31 (7) 925–934.

ubé, Laurette and Sylvie Morin (November 2001). “Background Music Pleasure and Store Evaluation: Intensity Effects and Psychological Mechanisms,” Journal of Business Research, 54 107–113.

ubé, Laurette, Jean-Charles Chebat and Sylvie Morin (July 1995). “The Effects of Background Music on Consumers’ Desire to Affiliate in Buyer- Seller Interactions,” Psychology and Marketing, 12 305–319.

roglu, Sevgin A., Karen A. Machleit and Lenita M. Davis (November 2001). “Atmospheric Qualities of Online Retailing–A Conceptual Model and Implications,” Journal of Business Research, 54 177–184.

roglu, Sevgin A., Karen A. Machleit and Lenita M. Davis (2003). “Empir- ical Testing of a Model of Online Store Atmospherics and Shopper Responses,” Psychology and Marketing, 20 (2) 139–150.

ibson, J.J. (1966). “The Theory of Information Pickup,” in The Senses Considered as Perceptual Systems, Boston: Houghton, 267–321.

ibson, J.J. (1979). “The Theory of Affordances,” in The Ecological Approach to Visual Perception, Boston: Houghton, 127–143.

ibson, J.J. and L.E. Crooks (1938). “A Theoretical Field-Analysis of Automobile-Driving,” American Journal of Psychology, 51 453–471.

orn, Gerald J. (Winter 1982). “The Effects of Music in Advertising on Choice Behavior: A Classical Conditioning Approach,” Journal of Mar- keting, 46 94–101.

race, Debra and Aron O’Cass (2004). “Examining Service Experiences and Post-Consumption Evaluations,” Journal of Services Marketing, 18 (6) 450–461.

reenland, Steven J. and Peter J. McGoldrick (January 1994). “Atmo- spherics, Attitudes, and Behavior: Modeling the Impact of Designed Space,” The International Review of Retail, Distribution, and Consumer Research, 4 1–16.

errington, J. and Louis M. Capella (1996). “Effects of Music in Service Environments: A Field Study,” Journal of Services Marketing, 10 (2) 26–41.

olbrook, Morris B. and Punam Anand (1990). “Effects of Tempo and Sit-

uational Arousal on the Listeners’ Perceptual and Affective Responses to Music,” Psychology of Music, 18 (2) 150–162.

ui, Michael K., Laurette Dubé and Jean-Charles Chebat (Spring 1997). “The Impact of Music on Consumers’ Reactions to Waiting for Services,” Journal of Retailing, 73 87–104.

O

ng 83 (1, 2007) 115–130 129

ttelson, William H. (1973). “Environment Perception and Contemporary Perceptual Theory,” in Environment and Cognition, William H. Ittelson ed. New York: Seminar Press Inc., 1–19.

ttelson, William H. (1976). “Environment Perception and Contemporary Perceptual Theory,” in Environmental Psychology, People and Their Physical Settings, New York: Holt, Rinehart and Winston Inc., 141– 154.

ttelson, William H., Harold M. Proshansky, Leanne G. Rivlin and Gary H. Winkel (1974). “Environment Perception,” in An Introduction to Envi- ronmental Psychology, New York: Holt, Rinehart and Winston Inc., 102–125.

ones, E.E. and R.E. Nisbett (1972). “The Actor and the Observer: Divergent Perception of the Causes of the Behavior,” in Attribution: Perceiving the Cause of Behavior, E. E. Jones ed. Morristown, NJ: General Learning Press, 79–84.

udge, Timothy A., Brent A. Scott and Remus Ilies (2006). “Hostility, Job Attitudes, and Workplace Deviance: Test of a Multilevel Model,” Journal of Applied Psychology, 91 (1) 126–138.

ellaris, James J. and Robert J. Kent (1992). “The Influence of Music on Consumers’ Temporal Perceptions: Does Time Fly When You’re Having Fun?,” Journal of Consumer Psychology, 1 (4) 365–376.

enny, David A. (2006). “Mediation,” (accessed January 23) [available at http://www.davidakenny.net/cm/mediate.htm].

offka, Kurt (1935). Principles of Gestalt Psychology, London: Lund Humphries.

ohler, W. (1929). Gestalt Psychology, New York: Liveright. ehrabian, Albert and James A. Russell (1974). An Approach to Environ-

mental Psychology, Cambridge, MA: MIT Press. illiman, Ronald E. (Summer 1982). “Using Background Music to Affect

the Behavior of Supermarkets Shoppers,” Journal of Marketing, 46 86–91.

illiman, Ronald E. (September 1986). “The Influence of Background Music on the Behavior of Restaurant Patrons,” Journal of Consumer Research, 13 286–289.

orgeson, Frederick P., Kelly Delaney-Klinger and Monica Hemingway (2005). “The Importance of Job Autonomy, Cognitive Ability, and Job- Related Skill for Predicting Role Breadth and Job Performance,” Journal of Applied Psychology, 90 (2) 399–406.

orris, Michael W. and Kaiping Peng (1994). “Culture and Cause: American and Chinese Attributions for Social and Physical Events,” Journal of Personality and Social Psychology, 67 (6) 949–971.

uller, Dominique, Charles M. Judd and Vincent Y. Yzerbyt (2005). “When Moderation Is Mediated and Mediation Is Moderated,” Journal of Per- sonality and Social Psychology, 89 (6) 852–863.

guyen, Nha and Gaston LeBlanc (2002). “Contact Personnel, Physical Environment and the Perceived Corporate Image of Intangible Services by New Clients,” International Journal of Service Industry Management, 13 (3) 242–262.

orth, Adrian C. and David J. Hargreaves (March 1996). “The Effects of Music on Responses to a Dining Area,” Journal of Environmental Psychology, 16 55–64.

orth, Adrian C. and David J. Hargreaves (1998). “The Effects of Music on Atmosphere and Purchase Intentions in a Cafeteria,” Journal of Applied Social Psychology, 28 (24) 2254–2273.

orth, Adrian C., David J. Hargreaves and Jennifer McKendrick (2000). “The Effects of Music on Atmosphere in a Bank and a Bar,” Journal of Applied Social Psychology, 30 (7) 1504–1522.

hno, Ryuzo (1980). “Visual Perception of Texture: Development of a Scale of the Perceived Surface Roughness of Building Materials,” Proceeding of EDRA, 11 193–200.

hno, Ryuzo (1985). “Notion of Duality in the Visual System and its Impli- cation for Environmental Design,” in Proceeding of the Second US-Japan

Seminar on Cross Cultural Research in Environment and Behavior, 209–217.

hno, Ryuzo (2000). “A Hypothetical Model of Environmental Perception–Ambient Vision and Layout of Surfaces in the Environment,” in Theoretical Perspectives in Environmental–Behavior Research, Sey-

1 f Retail

O

R

S

S

S

W

W

Y

Y

schild (Eds.), Advances in Consumer Research, vol.20, Provo, UT: Association for Consumer Research, 632–636.

30 S. Morin et al. / Journal o

mour Wapner, et al. eds. New York: Kluwer Academic/Plenum Publishers, 149–156.

hno, Ryuzo and K. Komuro (1984). “Texture Perception versus Object Perception in the Environment,” Proceedings of the Eighth IAPS, 17–21.

ussell, James A., Anna Weiss and Gerald A. Mendelsohn (1989). “Affect Grid: A Single-Item Scale of Pleasure and Arousal,” Journal of Person- ality and Social Psychology, 57 (3) 493–502.

harma, Arun and Thomas F. Stafford (August 2000). “The Effect of Retail Atmospherics on Customers’ Perceptions of Salespeople and Customer Persuasion: An Empirical Investigation,” Journal of Business Research, 49 183–191.

mith, Patricia and Ross Curnow (1966). “Arousal Hypothesis and the Effects of Music on Purchasing Behavior,” Journal of Applied Psychol-

ogy, 50 (3) 255–256.

tuart, Gregory L. and Amy Holtzworth-Munroe (2005). “Testing a The- oretical Model of the Relationship Between Impulsivity, Mediating Variables, and Husband Violence,” Journal of Family Violence, 20 (5) 291–303.

Y

ing 83 (1, 2007) 115–130

akefield, Kirk L. and Jeffrey G. Blodgett (January 1999). “Customer Response to Intangible and Tangible Service Factors,” Psychology and Marketing, 16 51–68.

atson, Richard T., Pierre Berthon, Leyland F. Pitt and George M. Zinkhan (2004). “Marketing in the Age of the Network: From Marketspace to U-Space,” Business Horizons, 42 (6) 33–40.

alch, Richard and Eric Spangenberg (Spring 1990). “Effect of Store Music on Shopping Behavior,” Journal of Consumer Marketing, 7 55– 63.

alch, Richard and Eric Spangenberg (1993). “Using Store Music for Retail Zoning: A Field Experiment,” Leigh McAlister and Michael L. Roth-

alch, Richard F. and Eric R. Spangenberg (August 2000). “The Effects of Music in a Retail Setting on Real and Perceived Shopping Times,” Journal of Business Research, 49 139–147.