Computers in Human Behavior 24 (2008) 2771–2791
Contents lists available at ScienceDirect
Computers in Human Behavior
journal homepage: www.elsevier.com/locate/comphumbeh
Interface design and emotions experienced on B2C
Web sites: Empirical testing of a research model
Jean Éthier a,*, Pierre Hadaya a,1, Jean Talbot b,2, Jean Cadieux a,3
a
Department of Information Systems and Quantitative Methods, Université de Sherbrooke, 2500 Boulevard de l’Université,
Sherbrooke, Québec, Canada JIK 2R1
b
Department of Information Technologies, HEC-Montréal, 3000 Côte Sainte-Catherine, Montréal, Québec, Canada H3T 2A7
a r t i c l e
i n f o
Article history:
Available online 21 May 2008
Keywords:
Interface design
Emotions
Cognitive appraisals
Web site usability
a b s t r a c t
This paper examines the impact of four Web site interface features on the cognitive process that trigger online shoppers’ emotions, operationalized as mental states of readiness that arise
from appraisal of events and considered as direct antecedents to
approach or avoidance behaviors. A research model was tested
with data collected from 215 Web shopping episodes for lowtouch merchandise. Results show that shoppers experienced all
six emotions posited in the model. The emotions of liking and
joy were experienced intensively by a substantial number of
shoppers. The results also demonstrate that interface features –
key components of the usability of a Web site – influenced the
three cognitive appraisals illustrated in the research model. Moreover, the cognitive appraisals of situational state and control
potential impacted the six emotions examined. This paper also
highlights several theoretical contributions and managerial implications that should help managers and Web site managers
improve the interface design of their Web sites in order to facilitate information gathering and better support online shopping
processes.
Ó 2008 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +1 819 821 8000x62308; fax: +1 819 821 7934.
E-mail addresses: [email protected] (J. Éthier), [email protected] (P. Hadaya), [email protected]
(J. Talbot), [email protected] (J. Cadieux).
1
Tel.: +1 819 821 8000x63670; fax: +1 819 821 7934.
2
Tel.: +1 514 340 6494; fax: +1 514 340 6880.
3
Tel.: +1 819 821 8000x61925; fax: +1 819 821 7934.
0747-5632/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chb.2008.04.004
2772
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
1. Introduction
In today’s Internet economy, few would dispute that a well-designed Business-to-Consumer (B2C)
Web site can be a strategic competitive tool for e-retailers selling both low-touch items – standard
goods such as books, CDs, and concert tickets and high-touch items – goods consumers prefer to
see and touch before purchase (The Economist, 2000). Indeed, a good design can increase Web site
success and enhance several consumer dimensions, including information gathering, intention to return to the Web site, trust, and performance improvement (Kumar, Smith, & Bannerjee, 2004; Nielsen,
2000; Palmer, 2002; Wang & Emurian, 2005; Zhang & von Dran, 2000). On the other hand, a poorly
designed Web site can negatively affect the online vendor’s corporate image and revenues and possibly lead to site failure (Buschke, 1997; Liu, Tucker, Koh, & Kappelman, 2003; Rayport & Jaworski, 2004;
Zona Research, 2001).
B2C Web sites are becoming more and more complex since the number of functionalities offered to
consumers is constantly increasing in order to improve both the information-gathering process and
the overall online shopping experience. For example, to support the shoppers’ decision process, Amazon.com offers more than 40 functionalities, including customized Web pages, one-click purchase
product previews, auctions and mobile access.
With these recent developments, Web site managers and developers are now overwhelmed with
recommendations on how to design effective Web sites. A search for books on the subject of ‘‘Web
design” on the book retailer Barnes & Noble’s Web site yields more than 5000 results. Articles on
the subject in trade and scientific journals are also abundant. However, two major shortcomings stand
out. First, the proposed sets of guidelines are frequently contradictory and few are based on theoretical
foundations (Geissler, Zinkhan, & Watson, 2001; Kim, Lee, Han, & Lee, 2002; Liu & Arnett, 2000; Song &
Zahedi, 2001; Zhang & von Dran, 2000). Second, Web site design recommendations for designers and
managers often ignore one key determinant of consumer behavior: their affective dispositions. Checklists and guidelines, even those that are user- or consumer-oriented, do not prioritize the feelings or
emotions that can arise during online shopping episodes despite important empirical findings regarding the crucial impact of shoppers’ affective states on their behaviors in both traditional and online
environments (Bagozzi, Gopinath, & Nyer, 1999; Kalbach, 2006; Turley & Milliman, 2000).
This study attempts to address these gaps in the literature as it focuses on emotions and proposes
and tests a conceptual model derived from empirical studies on information systems use, consumer
behavior and psychology. Specifically, within the traditional Stimulus-Organism-Response (S-O-R)
paradigm, this exploratory interdisciplinary research examines the relationships between several
Web site interface features (stimulus) and consumers’ cognitive processes leading to the emotions
elicited during online shopping episodes (organism) for low-touch goods. The relationships between
the consumers’ cognitive processes (organism) and their online behaviors (response) are not within
the scope of this study since several researchers in the field of marketing have already demonstrated
that affective states such as pleasure and arousal felt during traditional and online shopping episodes
for low-touch goods affect consumption behaviors (see the following section presenting the theoretical background).
The hypotheses presented and tested seek to provide answers to the following research questions:
(1) Do consumers experience emotions – defined as mental states of readiness arising from appraisals
of events – when shopping on the Web? (2) Can the cognitive perspective of emotions – identifying
cognitive appraisals as antecedents to emotions – be applied to online shopping and, if so, which cognitive appraisals and which emotions are involved? (3) Can Web site interface be considered as a
determinant of cognitive appraisals and, if so, which interface features affect which cognitive
appraisals?
The remainder of this paper is organized as follows. First, the relevant literature on Web site design
guidelines, on the relationships between environment, affective states and consumer behavior as well
as on emotions and their antecedents is examined. Then, the research model and hypotheses are presented, followed by an explanation of the research methodology applied. Research results are then
presented and discussed, after which the research limitations are highlighted. Finally, theoretical
and managerial contributions and avenues for future research are identified.
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
2773
2. Theoretical background
2.1. Web site design guidelines and recommendations
Professionals and scholars have proposed numerous sets of Web design principles and guidelines to
develop effective B2C Web sites. These guidelines can be classified into three general categories.
The first category of Web design principles and guidelines comprises opinions, analysis and recommendations made intuitively by professionals and specialists who have developed a practical expertise
in Web design (Cormier, 1999; Cotlier, 2001; Graham, 2001; Heck, 2000; Justice, 2001; Weinman, 2001).
This disparate body of literature, found in trade journals, specialized books and Web pages, is substantial
and covers various subjects, including architecture, graphics and presentation, content, navigation,
technical features, response time, advertisement, page structure, cohesion and consistency, measurement of effectiveness, and design strategy. It also presents tips and provides advice on common questions relating to the development of B2C Web sites (e.g., how to increase security, reduce abandoned
shopping carts, and increase traffic on the Web site).
The second category of Web design principles and guidelines comprises recommendations derived
from the traditional user interface design literature. These recommendations focus on reducing users’
tasks and cognitive load, and attempting to solve the disorientation problems many users face. The
authors in this research stream have used traditional interface models such as the Menu Lay-Out Model, the Hypertext Model, the Hypermedia Model and the Object–Action Interaction Model to provide specific guidelines for Web site designers (e.g., Carlson & Kacmar, 1999; Cockburn & McKenzie, 2001; Kim
& Yoo, 2000; McEneaney, 2001; Nilsson & Mayer, 2002; Shneiderman, 1998).
The third category encompasses recommendations derived from users’ own experiences in order to
ensure that Web site design focuses on usability (e.g., Brinck, Gergle, & Wood, 2001; Dempsey, 2002;
Dodson, 2001; Hassan & Li, 2005; Jagiello & Atkinson, 2004; Large, Beheshti, & Rahman, 2002; Ma,
2002; Nielsen, 2004; Thompson, 2003; Yu & Roh, 2002). For example, based on a usability approach
defined by five key components (learnability, efficiency, memorability, satisfaction and how easy it
is to recover from errors), Nielsen (2004) developed sets of recommendations that dealt with the
worst design mistakes found on the Web. In a recent study, Hassan and Li (2005) also identified 57
key criteria for Web site usability clustered into the following seven groups: screen appearance, content, accessibility, navigation, media use, interactivity, and consistency.
However, empirical research into B2C Web site usability is still limited. Based on a two-factor (hygiene and motivator factors) research model developed from Herzberg’s (1966) theory, Zhang and von
Dran (2000) identified a list of 44 features that affect user satisfaction in the context of a Web site providing news service. These authors claim that in addition to providing hygiene features – such as indications for navigating the Web site, accurate information, structure of information presentation,
attractive screen background and pattern – Web designers need to constantly identify and build motivational features, such as the presence of novel information and multimedia, attractive overall color
use, and screen layout. Within his longitudinal study, Palmer (2002) also found that Web site success,
measured by consumer satisfaction, likelihood of return, and frequency of use, was positively related
to the following usability features: download delay, navigation, content, interactivity and responsiveness (feedback options and FAQs). Based on these results, Palmer made three recommendations to
Web site managers: appropriate sequencing, layout and arrangements of Web sites in order to increase navigability, and the possibility for users to customize their experience and interact with the
Web site. After asking 216 participants to complete a specific set of tasks using the Web sites of
two compact disc vendors, Kumar et al. (2004) developed a list of usability interface features (e.g.,
information displayed on the screen, information for correcting wrong inputs, and shopping carts)
and features for which personalization is likely to be beneficial (e.g., arrangement of information on
the screen, steps to buy a product, and instructions for correcting wrong inputs) that designers should
consider when developing their Web site.
Overall, three broad conclusions can be drawn from the literature offering Web site design guidelines and recommendations. First, as pointed out by several authors, advice on Web site design
abounds and is often conflicting (Rosen & Purinton, 2004; Udo & Marquis, 2002; Zhang & von Dran,
2774
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
2000). Second, the number of scientific studies that are theory-oriented is small and few metrics on
Web site design have been developed (Palmer, 2002). Third, despite the abundance of recommendations offered to Web designers in trade and scientific journals alike, very few take into account the fact
that the use of B2C Web sites can trigger affective states such as moods, feelings, and emotions. Indeed, the vast majority of Web usability guidelines proposed in the trade and scientific literature
(e.g., an efficient information structure, ease of navigation, minimization of memory load and support
in handling errors) consider users exclusively as information processors. Also, there has not been
much scientific research on emotions per se in online shopping situations. Little is known concerning
which emotions are likely to be experienced during online shopping episodes, the conditions leading
to their emergence, and their impact on behaviors. Researchers have also been called upon to further
investigate the relations between design, affect and behavior (Eroglu, Machleit, & Davis, 2001, 2003;
Sautter, Hyman, & Lukosius, 2004).
2.2. Environment, affective states and behavior
Since Donovan and Rossiter’s (1982) pioneering work on store atmospherics, numerous researchers
in the field of consumer behavior have explored the relationships between environment, affect (an
umbrella term for a set of specific mental processes including feelings, moods and emotions) and
behavior. The evidence shows that a store’s atmospheric cues, including colors, scent and music, have
an impact on shoppers’ internal responses. Prior research has also demonstrated that shoppers’ affective states, operationalized as pleasure, arousal and dominance, influence several approach behaviors
such as willingness to buy (Baker, Levy, & Grewal, 1992), time and money spent (Yalch & Spangenberg,
1993; Sherman, Mathur, & Smith, 1997), number of items purchased (Gulas & Schewe, 1994), desire to
affiliate with store personnel (Dubé, Chebat, & Morin, 1995) and acceptance of a salesperson’s arguments (Chebat, Vaillant, & Gelinas-Chebat, 2000). Readers can consult Bitner (1992) and Turley and
Milliman (2000) for literature reviews on the subject.
Recently, some researchers have argued that the relationship between environment, affective state
and behavior may also apply to online retailing, even though the Web lacks some of the dimensions
found in bricks and mortar stores, such as temperature, odor, texture, and social interaction. For instance, Menon and Kahn (2002) have shown that the level of arousal and pleasure consumers experience on the Web influences their subsequent shopping behavior. Eroglu et al. (2001, 2003) have
proposed and empirically validated a research model that extends the S-O-R paradigm to e-tailing
(electronic retailing). In their view, the atmospheric cues of the online store, represented by content
related to shopping goals as well as design elements that do not directly support the purchasing tasks
(color, background patterns, typestyles and fonts), influence the outcomes of online retail shopping.
This influence takes the form of approach/avoidance behaviors through the intervening effects of
affective and cognitive states. Recently Sautter et al. (2004) extended the Eroglu, Machleit, and Davis’
model by proposing additional atmospheric elements such as vividness of information, interactivity,
symbolism and social elements as determinants of shoppers’ cognitive and affective responses. Semeijn, van Riel, van Birgelen, and Streukens (2005) discovered that customers of e-retailers experienced
joy both when shopping online and when receiving the products they purchased. This emotion was
positively related to customer loyalty through customers’ perceptions of overall satisfaction.
In the information systems literature, several studies have demonstrated that there is a relationship between system utilization, users’ affective states and users’ behaviors. Even though the concept
of affect has been operationalized by only a limited number of dimensions, researchers have already
shown that affective states such as anxiety and enjoyment influence the use of an information system
(Agarwal & Karahanna, 2000; Davis, Bagozzi, & Warshaw, 1992; Johnson & Marakas, 2000; Paré &
Elam, 1995; Venkatesh, 2000). Based on theoretical approaches derived from the consumer behavior
literature (Mehrabian & Russell, 1974) and the concept of flow (Csikszentmihalyi & Csikszentmihalyi,
1988), other IS researchers have shown that this relationship can also be found in online environments. They discovered that enjoyment was an antecedent to a range of behaviors such as the use
of e-mail technologies (Trevino & Webster, 1992), loyalty (Jarvenpaa & Todd, 1997), Web use (Novak,
Hoffman, & Yung, 2000) and the intention to return to a Web site (Koufaris, 2002). In a study among
users of online banking services, Bhattacherjee (2001) has shown that satisfaction – measured with
2775
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
four semantic differential adjective pairs related to affect: satisfied/dissatisfied, pleased/displeased,
frustrated/contented and terrible/delighted – is the strongest predictor of intention to continue to
use online banking.
To conclude this section, it is important to note that within both the consumer behavior and information systems literatures, researchers have used the concepts of affect, mood, attitude and emotion
inconsistently; they have also used operationalizations that correspond to more than one concept
simultaneously (Bagozzi et al., 1999). The following section presents the key concepts related to emotions, as they are defined and discussed by researchers in the field of psychology.
2.3. Emotions and their antecedents
The literature reviewed in this section is restricted to the cognitive perspective, one of the major
perspectives developed to understand how emotions are generated. According to the study by Bagozzi
et al. (1999), the value of the cognitive perspective is based on the fact that not only it allows for many
discrete emotions but also it specifies conditions for their occurrence. Nyer (1997) considers that the
cognitive perspective facilitates the development of research models since emotions are said to have
specific referent antecedents. Indeed, the cognitive perspective postulates that emotions are always
elicited by an intervening process of evaluation of an event (physical or mental) rather than the event
per se. This conscious or unconscious evaluation process is referred as a ‘‘cognitive appraisal”. Thus,
the claim is that a change in the way a situation is appraised will lead to a change in the emotion(s)
elicited.
Positive emotions
Negative emotions
Motive-Consistent
CircumstanceCaused
Appetitive
Hope
Uncertain
Joy
Aversive
Fear
Relief
Sadness
Distress
Frustration
Disgust
Hope
Uncertain
Certain
Appetitive
Surprise
Unexpected
Certain
Motive-Inconsistent
Aversive
Joy
Relief
Low
Control
Potential
High
Control
Potential
Other-Caused
Uncertain
Low
Control
Potential
Dislike
Certain
Liking
Uncertain
Anger
Contempt
Certain
High
Control
Potential
Self-Caused
Uncertain
Low
Control
Potential
Regret
Certain
Pride
Uncertain
Certain
Guilt
Non-Characterological
Shame
High
Control
Potential
Characterological
Fig. 1. Structure of the emotion system – Roseman et al. (1996). Note: Motive consistent and motive inconsistent are the
dimensions of the appraisal of situational state. Appetitive and aversive are the dimensions of the appraisal of motivational
state. Certain and uncertain are dimensions of the appraisal of probability. Circumstance-caused, other-caused and self-caused
are dimensions of the appraisal of agency. Low-control and high control are dimensions of the appraisal of control potential.
Characterological and non-characterological are dimensions of the appraisal of problem source.
2776
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Within the cognitive perspective, an emotion is defined as ‘‘a mental state of readiness that arises
from appraisals of events or of one’s thoughts” (Bagozzi et al., 1999, p. 184). This definition distinguishes emotions from other manifestations of affect, such as moods and feelings. Researchers have
proposed several theoretical frameworks to identify the antecedents of emotions and their influences
on a wide set of emotions (e.g., Frijda, 1986; Scherer, 1984; Smith & Ellsworth, 1987). All of the proposed frameworks are based on the belief ‘‘that assessments of situations are compiled to form appraisal configurations, and that configurations of appraisals lead to complete packages of emotional
responses” (Omdahl, 1995, p. 42).
This study’s theoretical underpinnings rest on Roseman, Antoniou, and Jose’s (1996) appraisal theory of emotions. Fig. 1 presents the theory’s 17 emotions as well as the cognitive appraisals’ sets of
dimensions on the borders of the chart. Fig. 1 also shows how emotions emerge by specifying which
combination of appraisals is solicited. For example, hope is solicited altogether by one of the situational state appraisal dimension (motive consistent), by one dimension of the agency appraisal dimension (circumstance-caused) and by one of the probability appraisal dimension (uncertain).
Roseman et al.’s (1996) appraisal theory of emotions was deemed most appropriate for the following three reasons: First, it is one of the most complete appraisal theory as it includes an elaborate set
of emotions and antecedents. Second, according to Bagozzi et al.’s (1999) literature review on the role
of emotions in marketing, Roseman et al.’s (1996) theory specifies conditions for the occurrence of
emotions during consumption behaviors. Third, since the Web is now an important consumption
medium, it seems pertinent to test whether the combinations of appraisals proposed by Roseman
et al.’s (1996) theory also yield discrete emotional responses during online shopping episodes. Not
to mention that the cognitive appraisal approach has already been used to explain how the use of a
computer system can elicit emotions such as joy/satisfaction, anger and sadness (Nyer, 1997).
3. Research model and hypotheses
The research model (see Fig. 2) proposes a set of specific relationships between four Web site interface features (key elements of the usability of a Web site), three cognitive appraisals related to the
consumer’s mental processes, and six emotions (three positive and three negative) the consumer
might experience during his/her online shopping episode.
This study is confined to the six emotions that were experienced by consumers during structured
interviews to pretest the model, namely liking, joy, pride, dislike, frustration and fear, as well as to the
three cognitive appraisals (situational state, probability and control potential) that are antecedents to
these emotions according to Roseman et al.’s (1996) framework. The other four cognitive appraisals
comprised in the framework are not included in the research model for two reasons. First, three of
them are not considered as antecedents to the set of emotions studied (problem source, motivational
state – used to distinguish frustration from disgust – and unexpectedness – sole factor for the emotion
of surprise). Second, the agency appraisal was not considered as it yields one of three states (circumstances-caused, other-caused, self-caused) and thus could not be conceptualized and operationalized
as the other appraisals comprised in the research model which yield one of only two states.
Structure of information presentation, navigation/orientation, text and visual aspects are the four
Web site interface constructs included in the research model. These four components were identified
in the literature as key interface features for the development of high-usability Web sites (IBM, 2000;
Hong and Moriai’s evaluation criteria for the design of commercial Websites, 1997). Structured interviews conducted to pretest the proposed model also confirmed that all the required Web site design
features were already taken into account in the research. Structure of information presentation measures to what extent the information available on the Web site is consistently organized in a way that
makes sense to the users (IBM, 2000). Navigation/orientation measures to what extent users can move
around easily in the Web site and to what extent the Web site provides information indicating where
visitors are (Hong & Moriai, 1997; IBM, 2000). Text measures to what extent the text on the Web site –
comprising the body text, headings, links, fonts and colors – is easy to scan and read (IBM, 2000).
Visual aspects measure to what extent the Web site provides page layouts that meet users’ needs
(IBM, 2000).
2777
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Liking
Structure
of information
presentation
H1
Situational State
H7a
(Overall evalution of the
online shopping episode)
H2a
Joy
H7b
H7c
Navigation /
Orientation
H2b
H7d
H3
H8a
Pride
Probability
H8b
(Abilityto predict
what will happen next
during the online
shopping episode)
H4a
Dislike
Text (appearance
and arrangement)
H4b
H9a
H5a
H9b
Control Potential
(Ability to control the
situation during the online
shopping episode)
Frustration
H5b
H6b
Visual aspects
H6a
H6c
Positive relationship:
Negative relationship:
Fear
Fig. 2. Research model.
The research model posits that each emotion is directly related to one, two or three cognitive
appraisals (following Roseman, Antoniou, and Jose’s theoretical assumptions) and postulates that
Web site atmospherics, through the four Web site interface features, have an impact on the three cognitive processes that lead to the emergence of emotions.
3.1. Cognitive appraisals and emotions
The appraisal of situational state accounts for the distinction between positive and negative emotions (Roseman et al., 1996; Roseman, Spindel, & Jose, 1990; Scherer, 1993; Wallbott & Scherer, 1988).
This appraisal is defined as the perception of the event as being either consistent or inconsistent with
one’s motives (Roseman, 1991). Motives are defined as goals and objects of one’s actions. Roseman
et al. (1996) claim that events that are appraised as consistent with one’s motives elicit the emotions
of joy, hope, relief, liking and pride, while events that are appraised as inconsistent with one’s motives
elicit the emotions of fear, sadness, distress, frustration, disgust, dislike, anger, contempt, regret, guilt
and shame (see Fig. 1).
When contextualized in a situation where a person is seeking information and shopping for goods
on a Web site, the appraisal of situational state refers to the person’s overall evaluation of the online
shopping episode in relation to his/her motives (goals pursued).
The appraisal of probability differentiates reactive emotions such as joy, relief, sadness and distress from preparatory emotions such as hope and fear (Frijda, 1986; Roseman, 1991; Roseman
et al., 1990; Smith & Ellsworth, 1987). This appraisal is defined as the certainty or uncertainty that
motive-relevant aspects of the event will occur (Roseman, 1991). Certainty is associated with sadness, distress, relief, and joy. Uncertainty is associated with hope and fear. Consequently, when one
is certain that the motive-relevant aspects of an event will occur, sadness, distress, relief, or joy are
present. On the other hand, when one is uncertain about their occurrence, hope or fear is elicited
(see Fig. 1).
2778
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
In a situation where a person is seeking information and shopping for goods on a Web site, the appraisal of probability refers to the person’s ability to predict what will happen next during the online
shopping episode (degree of certainty).
Control potential differentiates contending emotions such as frustration, anger, guilt, disgust, contempt and shame from accommodating emotions such as fear, sadness, distress, dislike and regret
(Roseman et al., 1996). This appraisal is defined as perceiving whether or not there is something
one can do (high control potential vs. low control potential) about the motive-relevant aspects of
an event. An event that is perceived as not being controlled by the self is considered as an antecedent
to the emotions of fear, sadness, distress, dislike, or regret, whereas an event perceived as being controlled by the self is considered as an antecedent to the emotions of frustration, disgust, contempt, anger, guilt, and shame (see Fig. 1).
In a context where a person is seeking information and shopping for goods on a Web site, the appraisal of control potential refers to the person’s ability to control the situation during the online shopping episode (degree of control).
According to Roseman et al. (1996), two appraisals – situational state, assuming the form of ‘‘motive-consistent”, and agency, understood as being influenced by others rather than by circumstances
or by the self – can elicit the emotion of liking. Hence, based on the specific dimensions of the proposed research model, we posit that:
H1: An online shopping episode that is favorably evaluated (the situational state is appraised as
motive-consistent) will positively influence the intensity of liking.
Three cognitive appraisals elicit the emotion of joy: situational state (motive-consistent), agency
(circumstance-caused), and probability (certain) (Roseman et al., 1996). Hence, it can be expected
that:
H2a: An online shopping episode that is favorably evaluated (situational state is appraised as
motive-consistent) will positively influence the intensity of joy.
H2b: An online shopping episode where it is possible to predict what will happen next (probability
is appraised as certain) will positively influence the intensity of joy.
According to Roseman et al. (1996), two cognitive appraisals elicit the emotion of pride: situational state (motive-consistent) and agency (self-caused). Consequently, it is hypothesized
that:
H3: An online shopping episode that is favorably evaluated (situational state is appraised as
motive-consistent) will positively influence the intensity of pride.
Three cognitive appraisals evoke the emotion of dislike: situational state (motive-inconsistent),
control potential (low) and agency (other-caused). Hence, based on the specific dimensions of the proposed model it is posited that:
H4a: An online shopping episode that is favorably evaluated (situational state is appraised as
motive-consistent) will negatively influence the intensity of dislike.
H4b: An online shopping episode where it is possible to control the situation (control potential is
appraised as high) will negatively influence the intensity of dislike.
Five cognitive appraisals evoke the emotion of frustration: situational state (motive-inconsistent),
control potential (high), agency (circumstance-caused), motivational state (appetitive), and problem
source (non-characterological factors). Hence, it is hypothesized that:
H5a: An online shopping episode that is favorably evaluated (situational state is appraised as
motive-consistent) will negatively influence the intensity of frustration.
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
2779
H5b: An online shopping episode where it is possible to control the situation (control potential is
appraised as high) will positively influence the intensity of frustration.
Finally, four cognitive appraisals evoke the emotion of fear: situational state (motive-inconsistent),
probability (uncertain), control potential (low), and agency (circumstance-caused). Hence, it can be
expected that:
H6a: An online shopping episode that is favorably evaluated (situational state is appraised as
motive-consistent) will negatively influence the intensity of fear.
H6b: An online shopping episode where it is possible to predict what will happen next (probability
is appraised as certain) will negatively influence the intensity of fear.
H6c: An online shopping episode where it is possible to control the situation (control potential is
appraised as high) will negatively influence the intensity of fear.
Web site interface features and the cognitive process giving rise to emotions.
The S-O-R paradigm states that store environment impacts shoppers’ organismic states or internal
responses (emotional, cognitive, and physiological). The assumptions of this paradigm have been used
to develop research models to explain online shopping behaviors (Eroglu et al., 2001; Sautter et al.,
2004). Indeed, Eroglu et al. (2001) claim that ‘‘just as the physical environment in a traditional retail
store impacts the various psychological and behavioral shopping outcomes, certain atmospheric qualities of online shopping are likely to affect use and results (satisfaction, repatronage, amount purchased, and time spent online in the virtual store) of online shopping” (p. 177). These authors
identified three computer-mediated atmospheric cues that could have a positive influence on Web
shoppers’ organismic states: the Web site’s information richness, its visual aspects (color, background
patterns, type styles and fonts), and its social elements such as shopping agents, Webcounters and online communities.
Kumar et al. (2004) also demonstrated that Web site interface design features such as colors,
amount of information, arrangement of the information on the screen, and the steps completed in
the buying process, are important determinants of Web site ease of use. Liu et al. (2003) found a strong
relationship between Web site interface elements (e.g., company information, search engine, and privacy statement) and Technology Acceptance Model (TAM) ease of use and user perceptions of usefulness constructs. Udo and Marquis (2002) found that ease of navigation and use of graphics positively
influenced users’ perception of a Web site’s effectiveness, as measured by repeat visits. Finally, Semeijn et al. (2005) found a positive relationship between the quality of navigation on the Web site and
the perception of its value.
All the studies identified above demonstrate that Web site interface design components similar to
those underlying this study’s conceptual model positively influence variables such as ease of use, usefulness, value and effectiveness. Determining whether an information system is easy to use, is useful,
and has value requires a cognitive process similar to the overall evaluation of an online shopping episode, as is the case in this study. Hence, it is hypothesized that:
H7: Structure of information presentation (H7a), navigation/orientation (H7b), text (H7c), and
visual aspects (H7d) will positively influence the overall evaluation of an online shopping episode
(appraisal of situational state).
To achieve usability and effectiveness, a Web site must provide adequate support in the form of a
strong sense of structure and place in order to let consumers know where they are and were they can
go (Nielsen, 2000). Consumers develop expectations concerning how to find different types of information and how to accomplish different tasks when shopping online. In particular, they trust that
the Web site will present information in a way that will make it easy for them to find what they need,
will indicate the function of links and will give them feedback as to where they are on the site (Ceri,
Fraternali, & Bongio, 2000; IBM, 2000; Taylor & England, 2006). These features allow them to predict
what will happen next and help them to control the situation during their shopping.
2780
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Hence, we can make the last two hypotheses:
H8: Structure of information presentation (H8a) and navigation/orientation (H8b) will positively
influence the ability to predict what will happen next during an online shopping episode (probability appraisal).
H9: Structure of information presentation (H8a) and navigation/orientation (H8b) will positively
influence the ability to control the situation during an online shopping episode (control potential
appraisal).
4. Research methodology
4.1. Data collection
Although several methods of inquiry have been used to study emotions (e.g., recall specific types of
emotional experiences and the features that elicited them, reaction to emotion words, reaction to
vignettes in an experiment), the data collected from this research comes from public-domain emotional experiences (immediate real-life situation methodology). According to Omdahl (1995), this approach ‘‘eliminates the confounds associated with memory storage and retrieval” and ‘‘assumes that
the appraisals the subjects are making of the situation are leading to reported emotional states”
(p. 46).
After an in-depth review of the literature on emotions and their antecedents, structured interviews
were conducted with 34 B.A.A. and M.B.A. students from two universities in the province of Québec,
Canada: HEC-Montreal and the Université de Sherbrooke. The objective was to identify what emotions
are experienced by consumers who shop online for low-touch goods. Each participant was asked to
describe his/her last shopping episode and to recall which of the 17 emotions included in the Roseman
et al.’s (1996) framework it had elicited. Liking, joy, pride, dislike, frustration and fear were the six
emotions experienced to some degree by participants.
Field survey participants were B.A.A. students at the Université de Sherbrooke. A research assistant
visited numerous classrooms at the Faculté d’administration to randomly recruit students. No participants were enrolled in classes given by the authors of this manuscript. Since students’ participation in
the survey was voluntary, their contribution was not rewarded in any way by their professors. Each
participant recruited was asked to shop on a predetermined Web site chosen from among the following four online retail sites: Archambault.ca, Amazon.ca, Renaud-Bray.com and Futureshop.ca. These
four French-language Canadian sites are the most popular online retail stores selling both music
CDs and movies in DVD format. The participants’ shopping episode was divided into two consecutive
tasks in which they had to gather specific product information. First, they were instructed to shop for a
specific retail item, a music CD (La Vallée des Réputations by Jean Leloup, a popular local singer) or a
movie DVD (Le Fabuleux Destin d’Amélie Poulin, a popular French movie released in local cinemas in
2000), in order to become familiar with their pre-assigned Web site. Next, they were invited to shop
for a product they would give to a family member or a friend. Immediately after their shopping episode, participants were asked to answer a questionnaire. Each student received $10 as compensation
for their collaboration. A total of 226 questionnaires were collected, 11 of which were eliminated due
to incomplete data.
4.2. Research variables
The measures for each Web site interface design feature were adapted from IBM’s design guidelines
proposed for the development of high-usability sites and from the design principles identified by Hong
and Moriai (1997) (Table 1). The measures for cognitive appraisals were adapted from the items proposed by Roseman et al. (1996) to reflect the specific context of this research. As for emotions, each of
them is measured by three items in accordance with previous studies that have used latent variables to
measure various affect constructs (Agarwal & Karahanna, 2000; Nyer, 1997; Semeijn et al., 2005; Venkatesh, 2000). The instrument used to measure emotions comes from Shaver, Schwartz, Kirson, and
2781
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Table 1
Constructs, items and sources
Construct
Measures
Web site interface design
How do you rate the:
Structure of
information
j Logical organization of the information?
presentation
j Consistent organization of the information
throughout the Web site?
j Relevant organization of the information?
Navigation/
orientation
How do you rate the:
j Design elements helping you to move easily
through the Web site?
j Design elements giving indications of where
you are on the Web site?
j Design elements helping you find the information or product on the Web site?
Text (appearance
and
arrangement)
How do you rate the:
j Readability of online text particularly type,
color and size of fonts?
j Layout characteristics of the text enabling a
quick scan of the Web site’s content and of
its main sections?
j Average amount of text per page?
Visual aspects
How do you rate the:
j Colors?
j Images?
j Background?
j Layout?
Cognitive appraisals
Situational state
Overall, the shopping episode on the Web site:
j Gave you the opportunity to accomplish the
tasks required successfully.
j Was a good example of what is expected
when you shop on the Web.
j Was satisfactory.
Probability
During the shopping episode, were you:
j Able to predict the end result?
j Certain of the consequences of your actions?
j Able to imagine what will happen next?
Control potential
During the shopping episode, were you:
j Able to control the situation?
j Able to modify the situation based on your
preferences?
j Able to change things as you wished?
Emotions
Liking
Joy
During the shopping experience, did you feel:
j Appreciation?
j Liking?
j Preferences?
Sources
Adapted from IBM’s design guidelines proposed for
the development of high-usability sites and from the
design principles identified by Hong and Moriai
(1997)
Adapted from Roseman et al., 1996
Each emotion is assessed by a set of three items from
Shaver et al. (1987) cluster analysis of 135 emotions
During the shopping experience, did you feel:
j Pleasure?
j Enjoyment?
j Enthusiasm?
(continued on next page)
2782
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Table 1 (continued)
Construct
Measures
Pride
During the shopping experience, did you feel:
j Self-confidence?
j Pride?
j Self-praise?
Dislike
During the shopping experience, did you feel:
j Antipathy?
j Dislike?
j Aversion?
Frustration
During the shopping experience, did you feel:
j Frustrated?
j Prevented from getting what you wanted?
j Blocked from certain actions?
Fear
During the shopping experience, did you feel:
j Afraid?
j Frightened?
j Dismayed?
Sources
O’Connor’s (1987) cluster analysis of emotion terms. A 9-point Likert scale, where 1 = poor, 5 = good
and 9 = excellent, was used to measure each component of the Web site interface design. A 9-point Likert scale, where 1 = not at all, 5 = moderately and 9 = very much, was also used to measure the items
for the three cognitive appraisals (situational state, probability and control potential) and the six emotions. Thus, the higher (or lower) the values of the items related to the appraisals, the more positively
(or negatively) the appraisals were evaluated. Moreover, the higher (or lower) the values of the items
related to the six emotions, the more (or less) strongly the emotions were experienced.
5. Descriptive statistics, structural equation modeling (SEM) analyses and results
This section reports on the research findings. Summary descriptive statistics are presented first followed by the results of the structural equation analyses.
5.1. Descriptive statistics
Eighty-three percent of the participants were between 18 and 24 years old, and 58% of the subjects
were men. Descriptive statistics also show that, on a weekly basis, 52% of participants spent time
shopping online to actually purchase products and/or services (38% less than 30 min, 8% between
30 and 60 min, 2% between 1 and 2 h and 4% more than 2 h).
Out of 215 shopping episodes, 59 were conducted on Archambault.ca, 50 on Amazon.ca, 60 on Renaud-Bray.com, and 46 on FutureShop.ca. Ninety-six percent of the participants had a shopping episode (comprising the first two tasks) that lasted between 5 and 24 min. Forty-seven percent of the
students shopped for movies (DVDs) and 53% shopped for music CDs. One hundred and twelve participants had never visited the B2C Web site that was assigned to them whereas 13 had visited that Web
site more than six times during the previous year.
As shown in Table 2, the four Web site interface features of the B2C Web sites were evaluated favorably (with means between 6.25 and 6.76). Many of the participants attributed item values of between 6
and 9 to navigation/orientation (81%), structure of information presentation (73%), and text (72%).
The means of the situational state, probability and control potential appraisals were, respectively,
7.20, 6.84 and 7.24, respectively (Table 2). Thus, overall, participants evaluated their shopping experience positively (situational state). They were easily able to predict what will happen next during
their shopping experience (probability) and felt very much in control (control potential). Between
83% and 90% of the values attributed by participants to the cognitive appraisals ranged between 6
and 9.
2783
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Table 2
Frequency statistics for emotions, appraisals and Web site interface design
Web site interface design
Meana
% of values between 6 and 9
on the 9-point Likert scale
Standard
deviation
Structure of information presentation
Navigation/orientation
Text (appearance and arrangement)
Visual aspects
6.38
6.76
6.47
6.25
73
81
72
60
1.50
1.53
1.65
1.57
Cognitive appraisals
Situational state
Probability
Control potential
7.20
6.84
7.24
90
83
89
1.42
1.56
1.43
Emotions
Liking
Joy
Pride
Dislike
Frustration
Fear
5.39
4.14
2.30
2.06
2.96
3.07
44
25
4
6
13
6
1.84
1.77
1.52
1.60
1.97
1.20
Scale: 1 = not at all, 5 = moderately and 9 = very much.
a
Sum of the scores attributed to each item of liking (appreciation, liking, and preferences) divided by 3.
Table 2 also shows that participants did in fact experienced emotions while shopping for low-touch
retail goods (means ranging from 2.06 to 5.39). Liking was the emotion with the highest mean intensity (5.39), followed by joy (4.14), fear (3.07), frustration (2.96), pride (2.30), and dislike (2.06). Even
though the mean intensity was below the moderate level for all emotions except liking, the data reveals that a substantial portion of participants felt some emotions strongly. Thirty percent and 28% of
the respondents experienced the emotions of liking and joy, respectively, at a higher than moderate
intensity level (between 6 and 9 on the Likert scale).
5.2. SEM analyses
Mplus Version 3.11 was chosen for data analysis as it is more prediction-oriented than LISREL and
thus is more appropriate for exploratory research (Barclay, Higgins, & Thompson, 1995).
Table 3 presents the fit indices for the measurement model. The probability value associated with
the chi-square statistic (1036.412, d.f. = 662, p = .000) shows that the model failed to reproduce the
observed correlation matrix. However, as pointed out by Gefen, Straub, and Boudreau (2000), this criterion is satisfied only rarely because the chi-square is sensitive to larger sample sizes and the power
test. On the other hand, the normed chi-square value (the ratio of the chi-square to the degrees of freedom) is within the acceptable range. The other global fit index, RMSEA (root mean square error of
approximation), has a value of 0.051 and thus is below McCloy, Campbell, and Cudeck’s (1994)
Table 3
Fit indices for the measurement model
Statistic
Name
Threshold guidelines
Observed value
v2
Chi-square test of the model fit
Normed
v2
RMSEA
Normed chi-square test of the model fit
Insignificant with a p-value above .05
(Bagozzi & Yi, 1988)
Between 1.0 and 3.0 (Gefen et al., 2000)
1036.412 with d.f. = 662,
p-value = .000
1.57
SRMR
Root mean square error of
approximation
Standardized root mean square residual
CFI
TLI
Comparative fit index
Tucker–Lewis index
Between 0.05 and 0.08 (McCloy et al.,
1994)
Below 0.05 (Gefen et al., 2000) or closest
to zero (Hair et al., 1998)
0.90 or greater (Bentler, 1990)
0.90 or greater (Timothy et al., 1994)
0.051
0.062
0.93
0.92
2784
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
proposed threshold. The comparative fit index (CFI) and Tucker–Lewis index (TLI) are also above their
criterion levels. As for the Standardized Root Mean Square Residual, it is a little higher than Gefen et
al.’s (2000) proposed threshold but is still close to zero, as recommended by Hair, Anderson, Tatham,
and Black (1998). Hence, our results indicate that the measurement model fits the data reasonably
well.
The validity of the constructs was assessed in terms of convergent validity and discriminant validity. Convergent validity examines the magnitude of correlation between item measures of a construct
(Gefen, 2003). As recommended by Fornell and Larcker (1981), three indicators were used to assess
convergent validity: (1) the loading of each item on its respective construct, (2) the Rho-coefficient
for internal consistency, and (3) the average variance extracted of each construct. Table 4 shows that
Table 4
Operationalization of constructs (item loadings)
Construct
Web site interface features
Structure of information
presentation
Navigation/Orientation
Text
Visual aspects
Cognitive appraisals
Situational state
Probability
Control potential
Emotions
Liking
Joy
Pride
Dislike
Frustration
Fear
a
Item
Loadings/
CFAa
j Logical organization of the information
j Consistent organization of the information throughout the site
j Relevant organization of the information
j Design elements helping you to move easily through the site
j Design elements giving indications of where you are on the site
j Design elements helping you find the information or product on the site
j Readability of online text particularly type, color and size of fonts
j Layout characteristics of the text enabling a quick scan of the Web site’s content
and of its main sections
j Average amount of text per page
j Colors
j Images
j Background
j Layout
0.84
0.85
0.82
0.88
0.69
0.73
0.81
0.95
0.84
0.81
0.80
0.73
0.84
j
j
j
j
j
j
j
j
j
Opportunity to accomplish the tasks required successfully
Good example of what is expected when you shop on the Web
Satisfactory
Able to predict the end result
Certain of the consequences of your actions
Able to imagine what will happen next
Able to control the situation
Able to modify the situation based on your preferences
Able to change things as you wished
0.67
0.81
0.87
0.80
0.84
0.80
0.86
0.85
0.78
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
Appreciation
Liking
Preferences
Pleasure
Enjoyment
Enthusiasm
Self-confidence
Pride
Self-praise
Antipathy
Dislike
Aversion
Frustrated
Prevented from getting what you wanted
Blocked from certain actions.
Afraid
Frightened
Dismayed
0.84
0.91
0.86
0.90
0.84
0.86
0.95
0.90
0.58
0.83
0.90
0.93
0.61
0.85
0.91
0.61
0.91
0.76
All scale items significant at p < .01.
2785
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
standardized confirmatory factor analysis (CFA) loadings for all scale items are significant at p < .01
and are above the minimum threshold of 0.60 recommended by Chin (1998) except for one item related to the emotion of pride at 0.58. However, we decided to keep this item since its loading is very
close to the threshold. As indicated in Table 5, the Rho-coefficients for internal consistency are also
above the 0.70 threshold and the average variance extracted (AVE) of each construct exceeds the variance attributable to its measurement error (i.e., 0.50). Hence, these results indicate that all three conditions for convergent validity of the research model’s constructs were met.
Discriminant validity is defined as the degree of uniqueness achieved from item measures in defining a latent construct (Gefen, 2003). The evaluation of discriminant validity followed Fornell and Larcker’s (1981) recommendation that the square root of AVE for each construct should be greater than
the levels of correlations involving the construct. Results in Table 6 confirm the discriminant validity
of the instruments.
The fit indices of the structural model were very similar to those provided by the measurement
model. Mplus Version 3.11 was used to estimate the path effects and their degrees of significance,
as well the percentage of variance explained (R2) for the endogenous constructs. Fig. 3 summarizes
the results.
The emotions of liking, joy, pride, dislike, frustration and fear have R2 values of 0.27, 0.13, 0.04, 0.14,
0.13, and 0.03, respectively. The data analyses also demonstrate that from a consumer perspective:
1. A favorably evaluated online shopping episode positively influences the intensity of liking. Hence,
H1 is supported.
2. A favorably evaluated online shopping episode positively influences the intensity of joy. However,
an online shopping episode where it is possible for the consumer to predict what will happen next
does not influence the intensity of joy. H2a is thus supported while H2b is not.
3. A favorably evaluated online shopping episode positively impacts the intensity of pride. Hence, H3
is supported.
4. A favorably evaluated online shopping episode negatively influences the intensity of dislike while
an online shopping episode where the consumer has the perception of being in control does not
influence negatively the intensity of dislike. H4a is thus supported while H4b is not.
5. A favorably evaluated online shopping episode negatively influences the intensity of frustration
while an online shopping episode where the consumer has the perception of being in control does
not influence negatively the intensity of frustration. Hence, H5a is supported and H5b is not.
Table 5
Assessment of internal consistency and convergent validity
Construct
Number of items
Composite reliabilitya
Variance-extractedb
Web site interface features
Structure of information presentation
Navigation/orientation
Text
Visual aspects
3
3
3
4
0.87
0.81
0.90
0.87
0.70
0.60
0.76
0.63
Cognitive appraisals
Situational state
Probability
Control potential
3
3
3
0.82
0.85
0.87
0.62
0.66
0.69
0.90
0.90
0.86
0.91
0.84
0.84
2 P þ ki¼1 1 k2i .
i¼1 ki
0.76
0.74
0.67
0.72
0.64
0.65
Emotions
Liking
Joy
Pride
Dislike
Frustration
Fear
a
b
P
k
Measured with the Jöreskog Rho=
i¼1 ki
Variance extracted =
Pk
2
i¼1 ki
.hP
k
2
i¼1 ki
þ
3
3
3
3
3
3
2 P
k
Pk
i¼1 ð1
i
k2i Þ .
2786
Liking
Liking
Joy
Pride
Dislike
Frustration
Fear
Situation
Probability
Control
Structure
Nav./Orien.
Text
Visual
0.871
0.644
0.258
***
0.307
0.077
0.013
***
0.498
***
0.187
***
0.268
***
0.370
***
0.409
***
0.330
***
0.309
***
***
Joy
0.863
0.465
0.203
0.096
0.044
***
0.365
***
0.200
**
0.173
**
0.181
***
0.245
***
0.222
***
0.316
***
***
Pride
Dislike
Frustration
Fear
0.888
0.035
*
0.107
*
0.121
**
0.204
*
0.111
0.063
0.075
**
0.147
0.058
0.074
0.898
0.338
***
0.193
***
0.357
*
0.100
***
0.218
***
0.203
***
0.187
***
0.209
***
0.314
0.800
0.390
***
0.357
***
0.243
***
0.242
**
0.160
***
0.271
***
0.209
***
0.233
0.806
0.041
***
0.164
**
0.174
0.077
*
0.089
0.036
0.001
***
Note: Square Root values of average variance extracted on the diagonal (in bold).
p < .10.
**
p < .05.
***
p < .01.
*
Situation
***
***
***
***
***
***
0.784
0.577
0.692
0.584
0.612
0.454
0.388
Probability
***
**
***
***
***
0.814
0.540
0.364
0.303
0.339
0.233
Control
***
***
***
***
0.831
0.524
0.540
0.394
0.242
Structure
***
***
***
0.836
0.772
0.614
0.579
Navigation/Orientation
***
***
0.776
0.626
0.576
Text
***
0.870
0.750
Visual
0.793
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Table 6
Correlation between constructs
2787
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
[.52***]
Structure
of information
[.17*]
Situational State
H7a
H7b
Navigation /
Orientation
[.34***]
(Overall evalution of the
online shopping episode)
[.54***]
H8a
[.26**]
H2a
Joy
(.13)
(.47)
H3
[.18*]
[.20***]
Pride
Probability
H8b
Text (appearance
and arrangement)
Liking
(.27)
H1
(Ability to predict
what will happen next
during the online
shopping episode)
(.04)
H4a
[.-.38***]
(.17)
Dislike
H9a
(.14)
[.16*]
H5a
[.-.37***]
H9b
Control Potential
[.47***]
Visual aspects
Frustration
(Ability to control the
situation during the online
shopping episode)
(.13)
(.37)
H6c
[-.15**]
Fear
(.03)
Fig. 3. Structural model Note: Coefficient of estimation are in brackets, where
parentheses.
***
p < .01, **p < .05, and *p < .10. R2 values are in
6. An online shopping episode where the consumer has the perception of being in control negatively
influences the intensity of fear. However, contrary to our predictions, the intensity of fear is not
influenced by a favorably evaluated online shopping nor by an online episode where it is possible
for the consumer to predict what will happen next. H6c is supported while H6a and H6b are not.
The cognitive appraisals of situational state, probability, and control potential have R2 values of
0.47, 0.17, and 0.37, respectively. The data analyses also demonstrate that:
1. Structure of information presentation and navigation/orientation positively influence the consumer’s overall evaluation of the online shopping episode. Thus, H7a and H7b are supported. However, a favorable evaluation of text and visual aspects do not significantly influence the overall
evaluation of the online shopping episode. H7c and H7d are thus not supported.
2. Structure of information presentation and navigation/orientation have a positive impact on the
consumer’s ability to predict what will happen next during the online shopping episode. Hence,
H8a and H8b are supported.
3. Structure of information presentation and navigation positively influence the consumer’s perception of being in control during the online shopping episode. H9a and H9b are thus supported.
6. Discussion
This research shows that when consumers gather information and shop online for low-touch merchandise such as music CDs and films on DVD, they experience affective states; more specifically, they
experience the following emotions: liking, joy, pride, dislike, frustration, and fear. While the mean
intensity of these emotions was low to moderate, a substantial number of shoppers experienced
strong emotions of liking and joy. Only a small percentage of participants reported experiencing no
emotions at all.
2788
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
This research also demonstrates that the cognitive perspective of emotions paradigm can be applied to online shopping. Indeed, 6 out of 11 hypotheses predicting that cognitive appraisals would
impact shoppers’ emotions were supported. The key cognitive appraisal of situational state (the consumer’s overall evaluation of the online shopping episode), which distinguishes positive from negative
emotions, was a determinant of all emotions except fear. Participants’ Web experiences may explain
why some cognitive appraisals do not significantly impact certain emotions, as we had predicted. First,
experienced Web users may not necessarily consider that an unfavorable shopping experience for a
low-touch item leads to fear; moreover, participants were not required to actually purchase any products during their monitored shopping episodes. Second, to experience joy during online shopping episodes, experienced users do not necessarily have to feel certain of what lies ahead since their desire to
explore the Web site may in itself incite joy and thus outweigh their need to be confined within a secure but rigid process. At the other extreme, the ability to predict what will happen next may be taken
for granted by experienced users. Third, experienced users will not necessarily experience dislike
when they are unable to control their shopping episodes, since the need to control the situation
may decrease the quality of the online shopping episode and prevent the gathering of useful information. Finally, experienced users in control of their shopping episode will not easily experience frustration since they may be used to coping with flaws inherent in the Web.
This study also shows that Web site interface features act as determinants of the cognitive processes that trigger emotions during online shopping episodes. As predicted, structure of information
presentation and navigation/orientation positively and significantly influence all three cognitive
appraisals, while text (appearance and arrangement) and visual aspects do not significantly influence
the cognitive appraisal of the shopping episode. This last finding is not surprising since, compared to
the navigation/orientation and the structure of information presentation components, text and visual
aspects are not important determinants of the overall evaluation of the online shopping episode.
6.1. Limitations
As with all research, there are several limitations on this study. First, the selection of Web sites
offering music CDs and movies on DVD was limited to four, which prevents generalization to other
similar Web sites. Second, even though the researchers made every effort to simulate a real-life shopping experience, participants were required to follow a set of guidelines in a particular context. Third,
since the data collected in this study is cross-sectional survey data, the attribution of causation is only
hypothetical. Fourth, the impact of involvement with the product, in this case the participants’ level of
interest in music CDs or movie DVDs, was not explored. Several studies have shown that involvement
with the product affects internal responses and behaviors (Koufaris, 2002; Novak et al., 2000). Fifth,
the study did not take into account a number of individual and environmental factors, such as personality traits, motivation, culture, normative beliefs, and moods, that may impact emotional and cognitive responses. Sixth, the research model did not attempt to assess the agency appraisal even though it
is possible that participants treated the Web site they visited as a social actor. Seventh, Web site interface design was operationalized by only four features, which cannot fully capture its complexity and
multidimensionality. Eighth, the study did not require participants to engage in an open-ended
description of their thoughts, and thus it could be argued that the cognitive appraisals might not have
occurred before the creation of emotions.
7. Conclusions
This research makes some theoretical contributions to the field of human–computer interaction.
Although, as mentioned earlier, several previous empirical studies have shown that different affective
states such as anxiety and pleasure may be experienced while using information systems and Web
sites, this study is the first to discover that six specific emotions – considered from a psychological perspective – are experienced during interactions with a Web site that take the form of information transfer and communication processes. In addition, our research findings show that it is possible to apply the
Roseman et al. (1996) theoretical framework to non-dramatic events such as online shopping for
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
2789
low-touch goods. This study also demonstrates that several Web site interface elements can impact
emotions through the three cognitive appraisals studied. This finding corroborates the results of earlier
studies that have found that Web site interface features influence consumers’ cognitive responses.
Five research avenues emerge from these theoretical contributions. First, testing whether the relationships between the cognitive appraisals and the emotions differ according to consumers’ online
Web experience could empirically validate the arguments put forth to explain the hypotheses that
were not supported in this research. Second, extending this proposed model so we can simultaneously
explore the relationship between Web site interface features (Stimulus), consumers’ cognitive processes underlying the emotions experienced during online shopping episodes (Organism), and consumers’ online behaviors (Response) could further demonstrate the pertinence of the S-O-R
paradigm as the theoretical foundation for the study of consumer behavior during online shopping
episodes. Third, extending/adapting our research model to incorporate other Web site design variables
and capture the influence of information presentation, processing and dissemination on different
emotions and their antecedents appears to be another promising research avenue. Fourth, fear should
be further investigated since our findings indicate that, contrary to the Roseman, Antoniou, and Jose’s
framework, there is no relationship between a poor evaluation of the online shopping experience (cognitive appraisal of situational state) and fear, or between the ability to predict what will happen next
during the online shopping episode and this emotion. Fifth, the role of trust should be explored, either
as an emotion, or as a cognitive antecedent to online shopping behaviors.
These research results also have implications for managers. For one thing, this study reiterates the
importance of e-retailers’ developing usable Web site interfaces since these impact the cognitive processes leading to emotions, which in turn, as clearly demonstrated in several other empirical studies
cited in this paper, influence approach and avoidance behaviors. Indicators of the four Web site interface features used in this research can serve as fundamental guidelines for managers and developers
as they develop and implement Web sites. This study also shows that, consciously or unconsciously,
consumers make a cognitive evaluation of their online shopping episode. In this regard, their evaluation of the cognitive processes leading to their emotions should provide useful information that can
guide the development of e-retailers’ Web sites.
Three further research avenues emerge from these managerial implications. First, other types of
products, particularly those with a higher perceived value (e.g., travel tickets, cars, banking services)
could be studied. Researchers could then compare if effect sizes between the variables of the research
model differ according to the type of goods purchased. Second, researchers should attempt to identify
what emotions are experienced by Web shoppers during each phase of their buying process (e.g.,
information search, product selection, payment, and settlement). Third, measuring how the design
of other types of IS applications (Web-based or not) influences emotions may help us discover
whether the emotions experienced by IS users vary from one context to another, and thus determine
whether our research findings can be generalized.
References
Agarwal, R., & Karahanna, E. (2000). Time flies when you’re having fun: Cognitive absorption and beliefs about information
technology usage. MIS Quarterly, 24, 665–694.
Bagozzi, R. P., Gopinath, M., & Nyer, P. U. (1999). The role of emotions in marketing. Journal of the Academy of Marketing Science,
27, 184–206.
Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation model. Journal of Academy of Marketing Science, 16, 79–94.
Baker, J., Levy, M., & Grewal, D. (1992). An experimental approach to making retail store environmental decisions. Journal of
Retailing, 68, 445–458.
Barclay, D., Higgins, C., & Thompson, R. (1995). The partial least squares approach to causal modeling: Personal computer
adoption and use as an illustration. Technology Studies, 2, 285–309.
Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246.
Bhattacherjee, A. (2001). Understanding information systems continuance: An expectation confirmation model. MIS Quarterly,
25, 351–370.
Bitner, M. J. (1992). Servicescapes: The impact of physical surroundings on customers and employees. Journal of Marketing, 56,
57–71.
Brinck, T., Gergle, D., & Wood, S. (2001). Usability for the Web: Designing Web sites that work. New York: Morgan Kaufman.
Buschke, L. (1997). The basics of building a great Web site. Training Development, 51, 46–48.
Carlson, J. R., & Kacmar, C. J. (1999). Increasing link marker effectiveness for WWW and other hypermedia interfaces: An
examination of end-user preferences. Journal of the American Society for Information Science, 50, 386–398.
2790
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
Ceri, S., Fraternali, P., & Bongio, A. (2000). Web modeling language (WebML); a modeling language for designing Web sites.
Computer Networks, 33, 137–157.
Chebat, J.-C., Vaillant, D., & Gelinas-Chebat, C. (2000). Does background music in a store enhance salespersons’ persuasiveness?
Perceptual and Motor Skills, 91, 405–524.
Chin, W. W. (1998). Issues and opinion on structural equation modeling. MIS Quarterly, 22, vi–xvi.
Cockburn, A., & McKenzie, B. (2001). What do Web users do? An empirical analysis of Web use. International Journal of Human–
Computer Studies, 54, 903–922.
Cormier, K. A. (1999). Developing a Web site – Med resorts’ experience. TMA Journal, 19, 66–68.
Cotlier, M. (2001). Adieu to abandoned carts. Catalog Age, 18, 39–42.
Csikszentmihalyi, M., & Csikszentmihalyi, I. S. (1988). Optimal experience: Psychological studies of flow in consciousness.
Cambridge, UK: Cambridge University Press.
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1992). Extrinsic and intrinsic motivation to use computers in the workplace. Journal
of Applied Social Psychology, 22, 1111–1132.
Dempsey, K. (2002). Could your Web site be better? Information World Review, 176, 13.
Dodson, J. (2001). The search for success. Internet Week, 885, 19.
Donovan, R. J., & Rossiter, J. R. (1982). Store atmosphere: An environmental psychology approach. Journal of Retailing, 58, 34–57.
Dubé, L., Chebat, J-C. J.-C., & Morin, S. (1995). The effects of background music on consumers’ desire to affiliate in buyer–seller
interactions. Psychology & Marketing, 12, 305–319.
Eroglu, S. A., Machleit, K. A., & Davis, L. M. (2001). Atmospheric qualities of online retailing: A conceptual model and
implications. Journal of Business Research, 54, 177–184.
Eroglu, S. A., Machleit, K. A., & Davis, L. M. (2003). Empirical testing of a model of online store atmospherics and shopper
responses. Psychology & Marketing, 20, 139–150.
Fornell, C., & Larcker, D. F. (1981). Structural equation models with unobservable variables and measurement error: Algebra and
statistics. Journal of Marketing Research, 18, 39–50.
Frijda, N. H. (1986). Emotion: Cognitive structure and action tendency. Cognition & Emotion, 1, 115–143.
Gefen, D. (2003). Assessing unidimensionality through Lisrel: An explanation and example. Communications of the AIS, 12,
23–47.
Gefen, D., Straub, D., & Boudreau, M. (2000). Structural equation modeling and regression: Guidelines for research practice.
Communications of the AIS, 4, 41–79.
Geissler, G., Zinkhan, G., & Watson, R. T. (2001). Web home page complexity and communication effectiveness. Journal of the AIS,
2, 1–44.
Graham, J. R. (2001). How do we get more visitors to our Web site? Agency Sales, 31, 38–40.
Gulas, C. S., & Schewe, C. D. (1994). Atmospheric segmentation: Managing store image with background music. In R. Achrol & A.
Mitchel (Eds.). Enhancing knowledge development in marketing ( Vol. 5, pp. 325–330). Chicago: American Marketing
Association.
Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate data analysis (5th ed.). Englewood Cliffs, NJ: Prentice
Hall.
Hassan, S., & Li, F. (2005). Evaluating the usability and content usefulness of Web sites: A benchmarking approach. Journal of
Electronic Commerce in Organizations, 3, 46–67.
Heck, M. (2000). Add some bang, get some bucks. InfoWorld, 4, 51–52.
Herzberg, F. (1966). Work and the nature of man. New York: World Publishing.
Hong, S., & Moriai, M. (1997). Evaluation criteria for the design of commercial Web sites. Department of Computer Information
Systems, Atlanta, Georgia.
IBM (2000). Web design guidelines. <http://www.ibm.com/easeofuse/Webdesignguidelines.htm> Accessed date: 6/11/2001.
Jagiello, G., & Atkinson, K. (2004). Improving Web usability. Association Management, 16–19.
Jarvenpaa, S., & Todd, P. A. (1997). Consumer reactions to electronic shopping on the World Wide Web. International Journal of
Electronic Commerce, 1, 59–88.
Johnson, R. D., & Marakas, G. M. (2000). Research report: The role of behavioral modeling in computer skills acquisition –
Toward refinement of the model. Information Systems Research, 11, 402–417.
Justice, L. (2001). Predicting the success of your Web site. Design Management Journal, 12, 63–69.
Kalbach, J. (2006). I’m feeling lucky: The role of Emotions in seeking information on the Web. Journal of the American Society for
Information Science and Technology, 57(6), 813–818.
Kim, J., Lee, J., Han, K., & Lee, M. (2002). Businesses as buildings: Metrics for the architectural quality of Internet businesses.
Information Systems Research, 13, 239–254.
Kim, J., & Yoo, B. (2000). Toward the optimal link structure of the cyber shopping mall. International Journal of Human–Computer
Studies, 52, 531–551.
Koufaris, M. (2002). Applying the technological acceptance model and flow theory to online consumer behavior. Information
Systems Research, 13, 205–223.
Kumar, R. L., Smith, M. A., & Bannerjee, S. (2004). User interface features influencing overall ease of use and personalization.
Information & Management, 41, 289–302.
Large, A., Beheshti, J., & Rahman, T. (2002). Design criteria for children’s Web portals: The users speak out. Journal of the
American Society for Information Science and Technology, 53, 79–94.
Liu, C., & Arnett, K. P. (2000). Exploring the factors associated with Web site success in the context of electronic commerce.
Information and Management, 38, 23–33.
Liu, S.-P., Tucker, D., Koh, C. E., & Kappelman, L. (2003). Standard user interface in e-commerce sites. Industrial Management Data
Systems, 103, 600–610.
Ma, Y. (2002). A design analysis model for developing World Wide Web sites. Journal of the American Society for Information
Science and Technology, 53, 531–535.
McCloy, R. A., Campbell, J. P., & Cudeck, R. (1994). A confirmatory test of a model of performance determinants. Journal of Applied
Psychology, 79, 493–505.
J. Éthier et al. / Computers in Human Behavior 24 (2008) 2771–2791
2791
McEneaney, J. E. (2001). Graphic and numerical methods to assess navigation in hypertext. International Journal of Human–
Computer Studies, 55, 761–786.
Mehrabian, A., & Russell, J. A. (1974). An approach to environmental psychology. Cambridge, MA: MIT Press.
Menon, S., & Kahn, B. (2002). Cross-category effects of induced arousal and pleasure on the Internet shopping experience.
Journal of Retailing, 78, 31–40.
Nielsen, J. (2000). Designing Web usability: The practice of simplicity. Indianapolis, IN: New Riders Publishing.
Nielsen, J. (2004). Top ten mistakes of Web design. Jacob Nielsen’s alert box. <ttp://www.useit.com/alertbox/990530.html>
Access date: 6/04/2004.
Nilsson, R. M., & Mayer, R. E. (2002). The effects of graphics organizers giving cues to the structure of a hypertext document on
users’ navigation strategies and performance. International Journal of Human–Computer Studies, 57, 1–26.
Novak, T. P., Hoffman, D. L., & Yung, Y. F. (2000). Measuring the customer experience in online environments: A structural
modeling approach. Marketing Science, 19, 22–44.
Nyer, P. U. (1997). A study of the relationships between cognitive appraisals and consumption. Journal of the Academy of
Marketing Science, 25, 296–304.
Omdahl, B. L. (1995). Cognitive appraisal, emotion and empathy. Mahwah, NJ: Lawrence Erlbaum.
Palmer, J. W. (2002). Web site usability, design, and performance metrics. Information Systems Research, 13, 151–167.
Paré, G., & Elam, J. J. (1995). Discretionary use of personal computers by knowledge workers: Testing of a social psychology
theoretical model. Behavior and Information Technology, 14, 215–228.
Rayport, J. F., & Jaworski, B. J. (2004). Best face forward. Harvard Business Review, 1–10.
Roseman, I. J. (1991). Appraisal determinants of discrete emotions. Cognition & Emotion, 5, 161–200.
Roseman, I. J., Antoniou, A. A., & Jose, P. E. (1996). Appraisal determinants of emotions: Constructing a more accurate and
comprehensive theory. Cognition & Emotion, 10, 241–277.
Roseman, I. J., Spindel, M. S., & Jose, P. E. (1990). Appraisals of emotion-eliciting events: Testing a theory of discrete emotions.
Journal of Personality and Social Psychology, 59, 899–915.
Rosen, D. E., & Purinton, E. (2004). Website design: Viewing the Web as a cognitive landscape. Journal of Business Research, 57,
787–794.
Sautter, P., Hyman, M. R., & Lukosius, V. (2004). E-tail atmospherics: A critique of the literature and model extension. Journal of
Electronic Commerce Research, 5, 14–24.
Scherer, K. R. (1984). On the nature and function of emotion: A component process approach. In P. Ekman & K. R. Scherer (Eds.),
Approaches to emotion (pp. 293–317). Hillsdale, NJ: Lawrence Erlbaum.
Scherer, K. R. (1993). Studying the emotion-antecedent appraisal process: An expert system approach. Cognition & Emotion, 7,
325–355.
Semeijn, J., van Riel, A. C. R., van Birgelen, M. J. H., & Streukens, S. (2005). E-services and offline fulfilment: How loyalty is
created. Managing Service Quality, 15, 182–194.
Shaver, P., Schwartz, J., Kirson, D., & O’Connor, C. (1987). Emotion knowledge: Further exploration of a prototype approach.
Journal of Personality and Social Psychology, 52, 1061–1086.
Sherman, E., Mathur, A., & Smith, R. (1997). Store environment and consumer purchase behavior: Mediating role of consumer
emotions. Psychology & Marketing, 14, 361–377.
Shneiderman, B. (1998). Designing the user interface: Strategies for effective human–computer interaction (3rd ed.). Reading, MA:
Addison-Wesley.
Smith, C. A., & Ellsworth, P. C. (1987). Patterns of appraisal and emotions related to taking an exam. Journal of Personality and
Social Psychology, 52, 475–488.
Song, J., & Zahedi, F. M. (2001). Web design in e-commerce: A theory and empirical analysis. In Paper presented at the 22nd
International Conference on Information Systems, New Orleans.
Taylor, M. J., & England, D. (2006). Internet marketing: Web site navigational design issues. Internet Marketing, 24(1), 77–85.
The Economist (2000). Survey on E-Commerce shopping around the Web. February 26.
Thompson, S. (2003). Web site usability vs usefulness. B to B, 88, 9.
Timothy, A. J., Boudreau, J. W., & Bretz, R. D. (1994). Job and life attitudes of male executives. Journal of Applied Psychology, 79,
767–782.
Trevino, L. K., & Webster, J. (1992). Flow in computer-mediated communication: Electronic mail and voice mail evaluation and
impacts. Communication Research, 19, 539–573.
Turley, L. W., & Milliman, R. E. (2000). Atmospheric effect on shopping behavior: A review of the experimental evidence. Journal
of Business Research, 49, 193–211.
Udo, G. J., & Marquis, G. P. (2002). Factors affecting e-commerce Web site effectiveness. Journal of Computer Information Systems,
42, 10–16.
Venkatesh, V. (2000). Determinants of perceived ease of use: Integrating control, intrinsic motivation and emotion into the
technology acceptance model. Information Systems Research, 11, 342–365.
Wallbott, H. G., & Scherer, K. (1988). How universal and specific is emotional experience? Evidence from 27 countries and five
continents. In K. Scherer (Ed.), Facets of emotion: Recent research (pp. 31–56). Hillsdale, NJ: Lawrence Erlbaum.
Wang, Y. D., & Emurian, H. H. (2005). Trust in e-commerce: Consideration of interface design factors. Journal of Electronic
Commerce in Organizations, 3, 42–60.
Weinman, L. (2001). Biggest Web mistakes. SBS Digital Design, 13, 18–21.
Yalch, R. F., & Spangenberg, E. R. (1993). Using store music for retail zoning: A field experiment. In M. L. Rothschild & L. McAlister
(Eds.), Advances in consumer research (pp. 632–636). Provo, UT: Association for Consumer Research.
Yu, B.-M., & Roh, S.-Z. (2002). The effects of menu design on information-seeking performance and user’s attitude on the World
Wide Web. Journal of the American Society for Information Science and Technology, 53, 923–933.
Zhang, P., & von Dran, G. M. (2000). Satisfiers and dissatisfiers: A two-factor model for Website design and evaluation. Journal of
the American Society for Information Science, 51, 1253–1268.
Zona Research (2001). The need for speed II. Zona Market Bulletin, 5, 1–9.