You are on page 1of 19

Canadian Journal of Administrative Sciences

Revue canadienne des sciences de l’administration


27: 5–23 (2010)
Published online in Wiley Interscience (www.interscience.wiley.com). DOI: 10.1002/CJAS.129

e-WOM Scale: Word-of-Mouth


Measurement Scale for
e-Services Context*
Isabelle Goyette
CROP Inc.

Line Ricard**
Université du Québec à Montréal (ÉSG-UQAM)

Jasmin Bergeron
ÉSG-UQAM

François Marticotte
ÉSG-UQAM

Abstract Résumé
In this article, using data from a survey of 218 consumers Dans cet article, nous proposons, à partir d’une enquête
across two samples, we propose a measurement scale for réalisée auprès de 218 répondants, une échelle de mesure
word of mouth (e-WOM scale) in the context of elec- du concept de bouche-à-oreille (échelle BAO ou e-WOM
tronic service. A battery of statistical tests reveals that scale) dans le contexte de services électroniques. La bat-
the WOM construct encompasses four dimensions: WOM terie de tests statistiques réalisés révèle que le concept
intensity, positive valence WOM, negative valence WOM, de BAO comprend quatre dimensions, à savoir :
and WOM content. Our proposed e-WOM scale can be l’intensité du BAO, la polarité positive du BAO, la polar-
used as a strategic tool for business managers aiming to ité négative du BAO et le contenu du BAO. L’échelle de
improve their word-of-mouth marketing strategies. mesure proposée peut être utilisée comme un outil stra-
Copyright © 2010 ASAC. Published by John Wiley & tégique par les gestionnaires d’entreprises de services
Sons, Ltd. en ligne désireux d’améliorer leurs stratégies de market-
ing en matière de bouche-à-oreille. Copyright © 2010
JEL Classifications: M31, L81, C3 ASAC. Published by John Wiley & Sons, Ltd.

Mots clés : Bouche-à-oreille, e-WOM scale, marketing,


Keywords: word-of-mouth, e-WOM scale, marketing, Internet, équations structurelles
Internet, structural equations

The authors would like to offer their sincere gratitude to the guest editor, the anonymous referees, Jean-Mathieu Fallu (MBA, ÉSG-UQAM), and Marc-
Antoine Vachon (post-graduate student, ÉSG-UQAM) for their apt and constructive recommendations. The second database that was used to validate
the scale is the result of a survey undertaken by Mrs. Olfa Gmach in her master’s thesis. The authors would also like to highlight the logistical and
financial contribution from SSHRC and the Chair in Financial Services Management ÉSG-UQAM.
*Please note that this paper was originally submitted in French and translated into English. Both versions are available on Wiley Interscience.
**Please address correspondence to: Line Ricard, ÉSG, Université du Québec à Montréal, Case Postale 6192, succursale Centre-ville, Montréal (Québec),
Canada, H3C 4R2. Email: ricard.line@uqam.ca

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 5 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

Word-of-mouth (WOM) is probably the oldest of-mouth measurement scale in the area of e-services.
means of exchanging opinions on various goods and Given the enormous potential benefits of favourable
services offered by markets. At one time, word-of-mouth word-of-mouth (especially online), such a scale could
occurred mostly among neighbours exchanging news on serve as an important tool for managers wishing to
what was being offered by neighbourhood stores (Whyte, measure their companies’ performance and the effect of
1954). As early as 1955, Katz and Lazarsfeld believed their strategies on people’s propensity to talk favourably
that word-of-mouth was seven times more effective than about them. In addition, this scale could be used to
newspaper ads, four times more effective than direct predict customers’ purchasing intentions and their inclin-
sales, and twice as effective as radio advertising. Later, ation to speak well of the company (Arndt, 1968; Brown
Day (1971) estimated that word-of-mouth was nine times & Reingen, 1987; Maxham III, 2001; Ying & Chung,
more effective than advertising in changing consumer 2007).
attitudes, whereas Morin (1983) showed that “other This article is structured as follows. The first section
people’s recommendations” were three times more reviews prior research of the concept of WOM. Subse-
effective in terms of stimulating purchases of over 60 quent sections are dedicated to the methodology, presen-
different products than was advertising. According to tation of our WOM model, and an analysis of results,
Reicheld (1996), these effects are amplified by a higher respectively. The paper closes with a discussion of the
degree of customer loyalty and profitability. Today, intended contribution and the implications for theory and
many researchers continue to maintain that word-of- management.
mouth constitutes one of the most effective ways of
attracting and keeping customers (Duhan, Johnson, Literature Review
Wilcox, & Harrell, 1997).
Studies on word-of-mouth have demonstrated that Word-of-Mouth (WOM)
its effectiveness is based on the overwhelming influence
that it has on consumer behaviour. Researchers have Here we briefly explain what is meant by word-of-
shown that word-of-mouth was strongly and positively mouth by placing it alongside new concepts such as viral
associated with clients’ levels of trust (Bergeron, Ricard, marketing and buzz marketing, among others. Addition-
& Perrien, 2003), service quality (Parasuraman, Zeithaml, ally, relevant research on the subject, and especially
& Berry, 1988), satisfaction (Anderson, 1998), perceived studies that have developed a measure of WOM, are
value (Hartline & Jones, 1996), relationship quality reviewed.
(Boles, Barksdale, & Johnson, 1997), and with clients’ Over the past five years, WOM has been the object
intention to purchase (Crocker, 1986). of multiple studies in the field of marketing. Authors
In today’s virtual era, the power of word-of-mouth have sometime associated this concept with personal rec-
has grown exponentially. For example, the international ommendations (Arndt, 1967a), interpersonal communi-
bank HSBC announced in the summer of 2007 that it was cation (Godes & Mayzlin, 2004), interpersonal
introducing a charge of 9.9% interest for each student relationships (Arndt, 1967a), informal communication
account (previously free) with a balance of under 1,500 (Silverman, 2001), personal and interpersonal influence
pounds Sterling (approx. CAD 3000). The National (Arndt, 1967a, Brown & Reingen, 1987), and with
Union of Students (NUS) immediately created a group informal advertising (Arndt, 1967a).
on the Facebook website to bring together the largest WOM definitions by Westbrook (1987), Bone (1992,
possible number of students opposed to this change of 1995), Silverman (2001), and Anderson (1998) have all
policy at HSBC. In a few short weeks, the power of been inspired by that of Arndt (1967a), which focused
virtual word-of-mouth managed to mobilize 5,000 stu- on the informal aspect of WOM communication, the
dents on summer break, each threatening to boycott or communicator’s independence from a commercial
to change bank. Under pressure, HSBC reversed the source, and on the phenomenon of information diffusion
change, indicating it had “answered the needs of its (cf. Table 1).
customers.” Table 1 indicates that word-of-mouth is usually
Although many studies target WOM, very few have defined as an exchange, flow of information, communi-
focused on a measure of word-of-mouth, especially in cation, or conversation between two individuals. There
the context of e-services. The recent concept of viral is but a single author (Haywood, 1989) who considered
marketing (Godin, 2001), which represents a modern word-of-mouth as formal conversation. Other authors
version of word-of-mouth, also shows the relevance of agree that word-of-mouth is an informal and noncom-
word-of-mouth (WOM) in an online context. Thus, our mercial conversation. The term “informal” makes refer-
objective here was to develop a multidimensional word- ence to something that is not organized in an official

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 6 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

Table 1
Definitions of Word-of-Mouth

Definitions1 Dimensions2

I F N C E

Arndt (1967a, p. 3) “. . . is defined as oral, person-to-person communication between a receiver and a X X X


communicator whom the receiver perceives as non-commercial, concerning a
brand, a product, or a service.”
Richins (1983, p. 17) “The WOM communication was defined as the act of telling at least one friend or X
acquaintance about the dissatisfaction”
Brown and Reingen “The WOM exists at the macro level of inquiry (e.g., flows of communication across X
(1987, p. 350) groups), as well as the micro level (e.g., flows within dyads or small groups)”
Higie, Feick, and “Conversations motivated by salient experiences are likely to be an important part of X X
Prince (1987, information diffusion”
p. 263)
Westbrook (1987, “In a postpurchase context, consumer word-of-mouth transmissions consist of X X X
p. 261) informal communications directed at other consumers about the ownership, usage,
or characteristics of particular goods and services and/or their sellers.”
Haywood (1989, “WOM is a process that is often generated by a company’s formal communications X X
p.58) and the behavior of its representatives.”
Swan and Oliver “Postpurchase communications included positive versus negative word-of-mouth and X X
(1989, p. 523) complaints and praising directed at the three entities in the exchange (i.e., the
salesperson, dealer, and manufacturer)”
Singh (1990, p. 1) “(c) telling others about the unsatisfactory experience (that is, negative X
word-of-mouth).”
File, Jude, and “Positive and negative word-of-mouth are examples of exit behaviors exhibited by X
Prince (1992, p. 7) consumers at the conclusion of a service encounter.”
File, Cermark, and “Word-of-mouth, both Input and Output, is the means by which buyers of services X
Prince (1994, exchange information about those services, thus diffusing information about a
p. 302) product throughout a market.”
Bone (1992, p. 579) “WOM communication is conceptualized herein as a group phenomenon—an X X
exchange of comments, thoughts, and ideas among two or more individuals in
which none of the individuals represent a marketing source.”
Bone (1995, p. 213) “Word-of-mouth communications (WOM), interpersonal communications in which X X
none of the participants are marketing sources, . . .”
Silverman (2001, “1) Word-of-mouth is communication about products and services between people X X X
p. 4) who are perceived to be independent of the company providing the product or
service, in a medium perceived to be independent of the company.”
2) Word-of-mouth is originated by a third party and transmitted spontaneously in a
way that is independent of the producer or seller.”
Anderson (1998, p. “Word of mouth refers to information communications between private parties X X
6) concerning evaluations of goods and services.”
Mangold & al. (1999, “WOM was far more likely to be initiated by receivers’ need for information than X X
p. 83) by communicators’ satisfaction level.”
Kim, Han, and Lee “Word of mouth is the interpersonal communication between two or more X
(2001, p. 276) individuals, such as members of a reference group or a customer and a
salesperson.”
Salzman, Matathia, Buzz is a “WOM effect, a transfer of information through social networks. It X X
and O’Reilly frequently occurs in a spontaneous manner, without so much as a raised finger on
(2004) the part of a marketing specialist or any other person.”
WOMMA (2006) WOM is “an act by consumers providing information to other consumers.” X
Total 5 1 5 7 13
1
Loose translation.
2
I = Informal, F = Formal, N = Noncommercial, C = Post-purchase behavior, E = Exchange/Flow of information/Communication/
Conversation.

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 7 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

manner (Rey-Debove & Rey, 2007). In addition, WOM contain anything that may be perceived as a commercial
communications are occasionally defined as post- or marketing intention to persuade or to inform one of
purchase behaviours. the communicating parties. Generally, it is this informal
In addition, according to previous work, for a con- and independent side of WOM that makes it unique.
sumer to be considered involved in a WOM-type conver-
sation, the message being transmitted and the medium New Terms Emerge
used for the transmission must be perceived as independ-
ent from influence by the company (Silverman, 2001). Since the advent of information technologies and the
For instance, can the context of e-services where many Internet, word-of-mouth has acquired several new names.
companies have discussion forums on their websites be Thus, mention is made of viral marketing, email market-
regarded as independent? In fact, in the current study ing, Internet word-of-mouth, word-of-mouth marketing,
such forums are considered to be a source of WOM and electronic WOM (e-WOM). In addition, the concept
insofar as consumers perceive these communications to of “buzz marketing” has made itself known as a new
be informal and not sponsored or subsidized by the marketing strategy derived from conventional word-of-
company. According to Silverman, any other type of mouth and bearing a strong resemblance to the concept
communication would be perceived as commercial and of viral marketing.
formal because advertising, public, and media relations Viral marketing is associated with word-of-mouth
communicate a message specifically selected, conceived, through electronic media. Internet is the central compon-
and expressed by the product or service vendor through ent of viral marketing and it is this necessary connection
an owned or leased medium. Including media relations with the Internet that makes it distinct from general
could be debatable because, although these are well word-of-mouth. The word viral refers, according to
planned, managers do not necessarily have any direct Godin (2001), to a virus, epidemic, or rather to an idea
control over the end result. virus, which he defined as follows: A big idea that runs
WOM communications can occur face to face, by amok in the target audience, a fashionable idea that
phone, email, mailing list, or any other means of com- propagates through a segment of the population, teaching
munication (Silverman, 2001). In addition, there are per- and changing and influencing everyone it touches (p. 17).
sonal and impersonal sources of recommendations that The consumer takes an active part in the advertising
have to be considered. Friends, family, and acquaint- process of a company by becoming its supporter, adver-
ances are personal sources of recommendations (Brown tising propagator, and, on occasion, advertising concept
& Reingen, 1987, Duhan, et al., 1997) recognized as developer (Stanbouli, 2003) or salesperson (Phelps,
WOM vehicles. Columns, articles, and commentary by Lewis, Mobilio, Perry, & Raman, 2004). One of the
journalists, columnists, consumers, and experts to be consumer’s strengths is that he or she is perceived as
found in newspapers, magazines, specialized publica- independent from the company that “hires” him or her.
tions, online discussion forums, and expert systems are Whether it is viral marketing or electronic word-of-
regarded as impersonal sources of WOM recommenda- mouth, these strategies serve as a way for companies to
tions. Expert systems and discussion forums are included empower the consumer (Hennig-Thurau, Gwinner,
as impersonal recommendation sources (Sénécal, Kalc- Walsh, & Gremier, 2004). Godin suggested that helping
zynski & Nantel, 2005) because consumers are influ- consumers communicate amongst themselves would be
enced in their choice of products online by preferable to attempting to address them directly. This
recommendations posted online (Sénécal & Nantel, could be done voluntarily by a group of consumers or
2004). Consumers should not perceive any commercial encouraged by financial incentives made available by a
or marketing intent behind the statements in these sources company. The latter option is further from the WOM
of recommendations. If that is not the case, these com- definition because customer independence is not as
munications cannot be considered WOM. Thus, a WOM obvious in this case.
communication can be based both on personal and Compared to viral marketing, the concept of buzz
impersonal sources. marketing is even less clearly defined. This situation is
In short, WOM could be affected by a marketing probably the result of the fact that these two notions are
effort such as advertising, media relations, and public frequently confused with each other (ABC-Netmarketing.
relations as well as by spontaneous conversations com, 2009). However, there is a difference. In fact,
between two individuals and by accounts of satisfactory unlike viral marketing, buzz marketing is not at all asso-
or unsatisfactory buying experiences. WOM sources can ciated with the Internet or any other electronic communi-
be both personal and impersonal. On the other hand, a cation medium. It does sometime happen that the word
conversation that follows a series of triggers may not “Internet” crops up in explanations of buzz marketing

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 8 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

given by authors, but it is not done in a restrictive manner. Appendix A lists authors who have conducted an empir-
Propagation rates inherent in an epidemic are not men- ical study of WOM. This summary shows for each author
tioned. Rather, buzz marketing plays the part of a catalyst the type of measurement scale used (unidimensional or
of a WOM conversation. For example, a classic buzz multidimensional), Cronbach’s alpha or a reliability
catalyst is to place interesting and/or nice-to-look-at index, the recommended methodology, the WOM meas-
people who know what they are talking about, in loca- urements and context (the way the researchers conceived
tions on the Internet or in the real world where they are WOM), and the viewpoint from which WOM was meas-
able to share reflections and their personal ideas on one ured (receiver or communicator).
product or another (Salzman, Matathia, & O’Reilly, Among the authors presented, Harrison-Walker
2004, p. 30). (2001) and Godes and Mayzlin (2004) are the only ones
The objective of buzz marketing is to create a rela- to have dedicated their research primarily to the study of
tionship between brands and people as a means to influ- WOM measurement. For others, a WOM measure mostly
ence purchase choices and create loyalty to a single follows from the need to include this dependent or
brand (Salzman, et al., 2004). Unlike WOM, buzz mar- independent variable into their research. On the other
keting is structured to provide people with incentives to hand, there are only six papers that explicitly present a
speak favourably of a good or a service among them- Cronbach’s alpha. In these cases, the internal consistency
selves. The spontaneity of exchanges is in evidence to a level is high with coefficients ranging from 0.78 to 0.80.
lesser degree as demonstrated in the previous example. Appendix A shows that most WOM measurement
In short, the illusion of spontaneity should be well scales presented in published research are unidimen-
orchestrated by buzz marketing specialists such that con- sional, that is, they measure a single dimension of WOM
sumers do not at all suspect that a company is behind the using in most cases a single statement or a single ques-
entire communication process. tion. Moreover, those authors who use unidimensional
The three concepts could be summarized as follows: scales do not specify which dimension they were
attempting to measure. Only through an in depth analysis
• Word-of-Mouth is defined as a verbal informal
of statements and a comparison of dimensions measured
communication occurring in person, by telephone,
by other authors using multidimensional scales was it
email, mailing list, or any other communication
possible to identify the WOM dimension measured by
method regarding a service or a good. A recommenda-
these authors. For example, Burzynski and Bayer (1977)
tion source may be personal or impersonal.
focused on the valence of a WOM conversation, Higie,
• Viral Marketing is defined as a rapidly spreading
Feick, and Price (1987) analyzed the volume of WOM,
informal online communication between individuals
while Bone (1992) concentrated on WOM content.
regarding a service or a good.
Appendix A also shows that authors mostly utilized
• Buzz Marketing is defined as a catalyst for a WOM
self-administered questionnaires, and, to a lesser degree,
conversation to occur in person or online derived from
telephone questionnaires or interviews. In other words,
a formal corporate strategy with a view to creating an
personal interviews and telephone surveys are the second
illusion of spontaneity.
data collection method used by these authors. Finally,
Given that the Internet has changed the dynamic of some authors used experiments to test the impact of
word-of-mouth, this technological evolution will be WOM on other variables rather than to measure WOM
taken into account. In fact, in the current study we seek itself.
to define and develop a reliable and valid measurement Since the WOM construct is not the primary object
scale in an e-service context. of the majority of studies, specific items or scales for
measuring WOM are rarely found. Rather, they are found
Word-of-Mouth (WOM) Measure in the context that enabled the construct to be measured.
Appendix A takes this feature into account by providing
A review of studies on WOM reveals the small a two-column representation: for the statements them-
number of works dedicated to WOM measurement. selves and for the context.
Although WOM has been in existence for many years, Higie et al. (1987), Bone (1992), and Mangold,
researchers and managers are still interested in it because Miller, and Brockay (1999) studied the “WOM content
it is an important driver of consumer behaviour. Yet, few dimension.” Harrison-Walker (2001) focused on four
researchers have focused explicitly on measuring it. aspects of WOM: (1) frequency, (2) number of contacts,
Harrison-Walker (2001, p.62) stated: “. . . WOM was not (3) detail, and (4) word-of-mouth praise. Further refine-
treated as a construct to be measured but rather as a ment of this measurement scale led Harrison-Walker to
category to be assigned based on responding to a survey.” retain only two primary WOM dimensions: word-of-

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 9 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

mouth praise with two items and word-of-mouth activity e-services. The objective set for these exploratory inter-
with four items. WOM activity includes all items associ- views was to gain better insight into the WOM phenom-
ated with the action of engaging in WOM. As for Godes enon in the context of e-services. Respondents were also
and Mayzlin (2004), they analyzed two WOM dimen- asked about online companies pertinent to their word-of-
sions: volume and dispersion. mouth activity. As direction for these meetings, one of
To identify other potential WOM dimensions, we the authors used an interview guide with four primary
conducted a detailed analysis of the published works on open questions regarding: (a) their reasons for engaging
this subject. Following this analysis we discovered in WOM online; (b) the credibility of these conversa-
another dimension measured by several authors: valence tions; (c) their propensity to initiate a positive or a nega-
(Black, Mitra, & Webster, 1998; Bone, 1995; Burzynski, tive conversation; and (d) the content of their electronic
& Bayer, 1977; Singh, 1990; Swan & Oliver, 1989). messages. An analysis and an interview report were pro-
Researchers verified whether conversations were posi- duced by one of the researchers that developed a clas-
tive or negative, or whether they reflected satisfaction or sification matrix. In addition, there was an informal
dissatisfaction with respect to a good or a service. This meeting with two consultants specializing in buzz mar-
dimension, when positive, reflects praise. Negative keting, which helped assess the scope of the word-of-
valence remains to be explored. mouth phenomenon in the electronic context and better
Finally, the WOM viewpoint refers to the perspec- differentiate between buzz, viral marketing, and WOM.
tive from which the communication process is examined. This first phase also enabled us to explore the different
The examination could be based around the communica- dimensions of WOM (WOM intensity, negative and
tor (i.e., researchers wish to interview the individuals that positive valance, and content).
start the conversation) of information or the receiver (i.e., A self-administered questionnaire was selected for
the person on the receiving end of the comments). An the survey. This method enabled information to be col-
analysis of research statements and contexts helped show lected on multiple dimensions. Based on previous work
that in measuring WOM, the communicator’s viewpoint and exploratory meetings, one of the researchers wrote
had mostly been taken into consideration while the the first draft of the questionnaire and another researcher
receiver’s viewpoint had been ignored. critiqued it. In the end, researchers arrived at a four-page
In conclusion, four major dimensions are identified questionnaire intended to be concise, clear, and in
to measure online word-of-mouth (WOM): (1) WOM keeping with the objectives of the present study. To
intensity (activity, volume, dispersion), (2) positive ensure this objective was met, the questionnaire was
valence (praise), (3) negative valence, and (4) content. presented to two other researchers who made minor sug-
gestions mostly with respect to the sociodemographic
variables (for example, adding income) and the wordings
Method of some of the statements. Finally, the questionnaire was
pretested on ten e-service users with an inclination
This paper is an extension of the train of thought towards WOM. The respondents independently com-
started by Harrison-Walker (2001). This author developed pleted the questionnaire under the watchful eye of one
a measurement scale comprised of six items and two of the researchers. The time required to answer the ques-
WOM dimensions. She tested her scale using a sample tions was calculated and any remaining points of confu-
made up of veterinary clinic and hair salon customers. sion cleared up as required. Following updates based on
The author observed that although interesting, it is several minor comments primarily with respect to
important that her scale be refined using different con- rewording the first question, the questionnaire was final-
texts and that the possibility of adding other dimensions ized. Again, not a single statement was eliminated.
be examined. As such, we proceeded in the context of The questionnaire consists of three sections: (1)
e-service companies because it is considered higher risk WOM activities; (2) assessment of the various dimen-
(thus favouring WOM activity) and because this context sions of word-of-mouth; and (3) the general propensity
promotes the use of personal sources of information on the part of respondents to engage in WOM, their
(Murray, 1991). online buying activities, and their social and demo-
Before proceeding to the development of the ques- graphic characteristics. The respondents were expected
tionnaire and subsequent data collection, a few steps to evaluate their WOM activities with respect to the last
were required to define a reliable and valid measurement e-services company they had discussed (six such com-
scale. After the primary WOM dimensions and measures panies were proposed, namely: amazon.com or .ca, eBay.
derived from a detailed review of existing works were com, admission.com, chapters.indigo.ca, expedia.com or
identified, meetings were held with ten consumers of expedia.ca, and archambault.ca). The companies were

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 10 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

selected based on a Statistics Canada (2004) study in sented in the in-class sample (65.8%) but not in the email
which the above-mentioned sites were cited as the most group. Overall, however, men accounted for 41.3% of
popular in Canada. the total (combined) sample.
Those that had not recommended any e-service Although there were differences in the sample pro-
company had to answer the questions in Section 3 of the files, this was not the case in terms of the assessment of
questionnaire nonetheless. Table 2 presents the 19 state- the measurement scale statements. Of the 19 items in the
ments used to measure the four dimensions of WOM. measurement scale, only 3 have significantly different
Since this study is of a rather theoretical nature, the means (t-test) for the two samples, namely: I speak more
sample had to have a certain degree of homogeneity often of this company than about any other type of
(Calder, Philips, & Tybout, 1981). The first data collec- company (3.26 in class vs. 2.43 by email); I speak favour-
tion using a suitable sample was performed in four ably of this company to others (4.69 in class vs. 5.21),
university classrooms. It produced 116 completed ques- and I discuss variety (5.57 in class vs. 4.80 by email).
tionnaires. To complement data collection that was initi- Since there were few significant differences between the
ated in class and to obtain respondents of a different two groups in the assessment of WOM statements, the
profile, “snowball” sampling was subsequently per- model was tested on the combined samples.
formed using electronic mail. Some 475 questionnaires
were sent out by email and 107 were completed and Analysis and Results
returned. Following an analysis of outliers, the size of
the final sample was 218 respondents. Assessment of Measurement Scale Quality
The respondents’ profile is shown in Table 3. Certain
differences between the samples are apparent: the sample The psychometric properties of the scale were ana-
collected in-class had a lower age (mean of 27 years as lyzed in multiple ways. This section presents the analysis
compared to 32 years for email), lower income (84.6% of unidimensionality and reliability as well as of conver-
have incomes below $70,000 vs. 52% for email respond- gent and discriminant validity. The results are summar-
ents), and was mostly made up of junior college students ized in Table 4.
(51.4%) vs. four-year university programs (88.6% email Unidimensionality. A scale is regarded as uni-
respondents). On the other hand, women were overrepre- dimensional whenever a group of statements measures

Table 2
Word-of-Mouth Dimensions and Statements

Word-of-Mouth (WOM) dimensions Statements

WOM intensity • I spoke of this company much more frequently than about any other e-services company.
• I spoke of this company much more frequently than about companies of any other type.
• I spoke of this company to many individuals.
Positive valence WOM • I recommended this company
• I speak of this company’s good sides.
• I am proud to say to others that I am this company’s customer.
• I strongly recommend people buy products online from this company.
• I mostly say positive things to others.
• I have spoken favourably of this company to others.
Negative valence WOM • I mostly say negative things to others.
• I have spoken unflatteringly of this company to others.
WOM content • I discuss the user-friendliness of its website.
• I discuss security of transactions and its Internet site.
• I discuss the prices of products offered.
• I discuss the variety of the products offered.
• I discuss the quality of the products offered
• I discuss ease of transactions.
• I speak of the rapid delivery.
• I speak of the company’s notoriety.

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 11 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

Table 3
Respondent Profiles

Variable* Classification In class By email Total


111 respondents 107 respondents N (%)
(%) (%)

218 (100 )
Gender Men 34.2 48.6 90 (41.3)
Women 65.8 51.4 128 (58.7)
Education Primary/Secondary 8.1 3.8 13 (6.0)
College 51.4 7.6 65 (30.1)
University 40.5 88.6 138 (63.9)
Household revenue Under $ 29,999 41.3 17.5 61 (29.5)
$ 30,000 to $ 69,999 43.3 30.1 76 (36.7)
$ 70,000 and more 15.4 52.4 70 (33.8)
Data collection method In class 111 (50.9)
By email 107 (49.1)

Mean Mean Mean (Standard


Deviation)

Age 27 32 29.5 (8.1)

* At 95% confidence, the four sociodemographic variables are significantly different.

the same thing. Many authors agree that unidimensional- the ratio χ2/degrees of freedom of 1.96 (43.14/22), a CFI
ity is fundamental to measurement theory (Hattie, 1985) of 0.95, an AGFI of 0.88, an NFI of 0.91, an NNFI of
and is crucial in the development of a quality scale 0.92, and a GFI of 0.94 (Bagozzi & Yi, 1988; Bentler,
(Anderson & Gerbing, 1988). Cox and Cox (2002) tested 2005).
the unidimensionality of their measurement scale using Reliability. Baumgartner and Homburg (1996) rec-
principal component analysis. This technique was ommended that the reliability of measurement scales be
employed, and the results of Table 4 (first column of evaluated from different angles. Therefore, three differ-
results) indicate that each dimension of the scale is ent types of analysis were performed. Suggestions by
unidimensional. Churchill (1979) were used to calculate first the adjusted
A confirmatory factor analysis using the EQS 6.1 item-total correlations (Table 4). The results show that
structural equation modelling software (Bentler, 2005) all dimensions have adjusted item-total correlations
was also performed for a more in depth validation of the greater than 0.35, which is satisfactory (McKelvey,
scale’s psychometric properties. Baumgartner and 1976). Subsequently, component Cronbach’s alphas and
Homburg (1996) emphasized that this statistical method reliability indices for each dimension were generated.
helps validate the quality of the measurement tool. As The results shown in Table 4 indicate that the alphas
mentioned by Anderson and Gerbing (1988), structural range between 0.69 (WOM intensity) and 0.89 (WOM
equation modelling holds great potential in measurement positive valence), which is satisfactory (Peterson, 1994).
scale validation. Table 4 (first column of results—CFA) As for component reliability indices (CFI), they range
also indicates that the coefficients are all associated with between 0.64 (WOM intensity) and 0.85 (positive
their respective dimensions, that they are all greater than valence), exceeding the minimum of 0.60 suggested by
0.50, and that they are statistically significant. Bagozzi and Yi (1988), as well as by Fornell and Larcker
Multiple statistics were used to assess the model’s (1981).
goodness of fit to the data. Firstly, the average off- Convergent and discriminant validity. To assess
diagonal standardized residual (AOSR) is 0.067 and the convergent validity, many authors have suggested that it
RMSEA index is 0.08, which is somewhat high but is important to determine if each statement is associated
nonetheless acceptable. However, we can reconfirm a with the factor that it measures (Anderson & Gerbing,
satisfactory validation of goodness of fit to the data with 1988; Bagozzi & Yi, 1988). Estimated coefficients as

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 12 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

Table 4
Results of Reliability and Validity Analyses

DIMENSIONS Principal component Adjusted item-total Alpha


analysis/CFAa correlations (α)/CFIb

Word-of-mouth intensity 0.69/0.64


I spoke of this company much more frequently than about any other 0.76/0.58 0.47
e-services company.
I spoke of this company much more frequently than about companies of 0.84/0.80 0.58
any other type.
I spoke of this company to many individuals. 0.77/0.60 0.49
Positive valence word-of-mouth 0.89/0.85
I recommended this company 0.86/0.84 0,78
I speak of this company’s good sides. 0.83/0.79 0,73
I am proud to say to others that I am this company’s customer. 0.70/0.61 0.60
I strongly recommend people buy products online from this company. 0.81/0.80 0.73
I mostly say positive things to others. 0.81/0.82 0.70
I have spoken favourably of this company to others. 0.84/0.73 0.74
Negative valence word-of-mouth 0.82/0.78
I mostly say negative things to others. 0.92/0.93 0.69
I have spoken unflatteringly of this company to others. 0.92/0.74 0.69
Word-of-mouth content 0.80/0.77
I discuss the user-friendliness of its website. 0.58/0.51 0.46
I discuss security of transactions and its Internet site. 0.65/0.56 0.51
I discuss the prices of products offered. 0.61/0.58 0.45
I discuss the variety of the products offered. 0.70/0.67 0.55
I discuss the quality of the products offered. 0.70/0.66 0.55
I discuss ease of transactions. 0.62/0.52 0.50
I speak of the rapid delivery. 0.69/0.60 0.56
I speak of the company’s notoriety. 0.62/0.54 0.48
a
The first number represents the factor coefficient generated by principal component analysis. The second number represents the factor
coefficient generated by confirmatory factor analysis (CFA), performed using EQS 6.1 software (Bentler, 2005).
b
The first number represents Cronbach’s alpha. The second number represents the reliability index of the CFI components (cf. Fornell
and Larcker, 1981).

computed by the EQS software are all high (i.e., ≥0.50) fidelity indices of word-of-mouth dimensions were still
and statistically significant (p < 0.01). higher than the correlation between the respective
To evaluate discriminant validity, the model pre- dimensions.
sented in Figure 1 was assessed several times by combin-
ing two different dimensions together every time Assessment of WOM Measurement Scale in an
(Anderson & Gerbing, 1988). The Chi-square of the ori- E-services Context
ginal model was a great deal lower than in the models
where two dimensions were combined, which is an For reasons of statistical parsimony, the method
excellent indication of the discriminant validity of the of partial aggregation of statements was used for the
model’s dimensions (Bagozzi & Philips, 1982). In other dimensions of positive valence and WOM content
words, for each pair of measures forcing two dimensions (Bagozzi & Heatherton, 1994). For example, we ran-
into one, we saw a reduced goodness of fit to the data as domly selected two items from the “positive valence
compared to the original model (Figure 1). We also word-of-mouth” construct as a sample of the six-item
evaluated discriminant validity by following recommen- scale. Hence, this strategy helped us reduce the number
dations by Gaski (1984), who state that the correlation of items from 19 to 9 with each factor measured by two
between two dimensions should not exceed the reliability or three items. Thus, random error is reduced, a complex
of the respective dimensions. Results indicate that the model is simplified, and the items aggregated at random

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 13 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

Figure 1.
Standardized results for word-of-mouth measurement scale in an e-services context

Item 1 0.63
0.75 WOM Intensity
Item 2
0.62
Item 3 0.65

0.92
Item1 Positive Valence
0.71 0.81
Item 2 WOM

Word-of-
Mouth
Item 1 0.93 -0.22
Negative Valence
Item 2 0.74 WOM
0.57

Item 1 0,87
χ2 / DF = 43.14 / 22 = 1.96
0,72 WOM Content
Item 2 AOSR = 0.067 ; RMSEA =
0.08
NFI = 0.91 ; NNFI =0.92
AGFI = 0.88 ; GFI = 0.94
*
All the coefficients are statistically significant (p< 0.05).
**
A satisfactory goodness of fit of the model to the data is possible whenever the average off-diagonal
standardized residual (AOSR) and the RMSEA index approach zero. In addition, the model is deemed
satisfactory whenever the majority of the indices NFI, NNFI, AGFI, GFI, and CFI are greater than 0.9. Finally,
the goodness of fit of the data to the model is satisfactory whenever the ratio χ2 / DF is under 4 (Bentler, 2005).
***
For reasons of statistical parsimony, the method of partial aggregation of statements was used for the
dimensions of positive valence and WOM content (Bagozzi & Heatherton, 1994). Hence, the two following
items were randomly selected for each construct: “I recommended this company” and “I have spoken
favourably of this company to others” for the Positive Valence WOM construct and “I discuss the quality of the
product offer” and “I discuss the variety of the product offer” for the WOM Content construct.

are more reliable than dimensions measured using a sors, and students of a Québec university. Thus, a new
single statement (Bagozzi & Heatherton). database of 150 respondents was created. The percentage
The results presented in Figure 1 show that the of men in the second sample is 51.3%. The mean age is
dimension of positive valence (i.e., WOM praise) 31.3 with a standard deviation of 8.49. Just over 75% of
explains the greater part of the variance in the WOM the respondents make under $ 50,000 a year.
construct (λ = 0.81) followed by the dimensions of WOM Here again the psychometric properties of the scale
intensity (λ = 0.65), content (λ = 0.57), and negative were satisfactory. Cronbach’s alphas are 0.78, 0.83, 0.94,
valence (λ = −0.22). All these coefficients are statistic- and 0.84, respectively for the following dimensions:
ally significant (p < 0.01). content, WOM intensity, positive valence, and negative
To improve scale validity, a new round of data col- valence. A confirmatory factor analysis was performed
lection was undertaken. A second survey was carried out again with the same statements shown in Figure 1.
by email using the snowball sampling method. This Indices of model goodness of fit to the data are satisfac-
method involved asking the respondents to complete the tory (e.g., NFI = 0.92 and CFI = 0.94). As in the first
questionnaire and to forward it on. A Word.doc file with model, the dimension of positive valence explains
the questionnaire was sent out to acquaintances, profes- the greater part of the variance in the WOM construct

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 14 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

(λ = 0.87) followed by the dimensions of WOM intensity of Desormeaux and Labrecque (1999), which showed a
(λ = 0.83), content (λ = 0.74), and negative valence difference between “satisfaction” and “dissatisfaction.”
(λ = −0.16). Recognition of valence as a dual concept is theoretically
In addition to performing the same reliability and interesting; more so given that the effects of one seem
validity analyses, we empirically demonstrated the not to have the same impact as the other. According to a
nomological validity of the scale. We analyzed the cor- frequently cited statistic, a dissatisfied customer men-
relation coefficients for the four dimensions of our scale tions his or her dissatisfaction to nine people, while satis-
and three constructs strongly associated with WOM in faction is only expressed to five individuals. However,
previous research, namely, service quality, trust, and the results show that the principal constituent of WOM
satisfaction (Anderson, 1998; Bergeron et al. 2003; is positive valence. Does that mean that the propagation
Reicheld & Sasser, 1990). Absolute correlation coeffi- of valence is reversed in the specific context of
cients between each scale dimension and each construct e-commerce?
measuring service quality, respondent trust, and satisfac-
tion range between 0.22 and 0.50. They are all statistic- Applied Implications
ally significant, which indicates that the nomological
validity of our scale is acceptable. Thus, we can conclude The strategic importance of word-of-mouth as a
that the internal validity of our scale is acceptable because communication medium for an organization no longer
it has been tested using two different samples. has to be proven. Paradoxically, companies have very
little, if any control over its reach or content. A company
is dependent on the message that will be communicated
Discussion by its consumers and others with whom it has never had
and never will have any contact. The virtual environ-
Summary ment and its global scope only increase the power of
word of mouth and, consequently, the level of depend-
The primary objective of the study was to create a ence of the companies in question. The instant popular-
multidimensional measurement scale for WOM in the ity and notoriety of trading and noncommercial sites
context of electronic services—the e-WOM Scale. A such as Facebook, YouTube, eBay, or Wikipedia are
review of prior research highlighted the paucity of studies irrefutable evidence of the ubiquity of WOM in the
focusing on this set of problems, especially in the domain online world. Virtual communities such as DPReview
of e-services. Taking inspiration from the results of a owe their existence and their growth only to word-of-
study by Harrison-Walker (2001) and from many others, mouth. An integral part of our everyday life, some, like
the nine items of the e-Scale measure four dimensions of Google, have even become common names. It is diffi-
word-of-mouth: (1) intensity (activity); (2) positive cult to imagine how much these online companies would
valence/praise; (3) negative valence; (4) content. Statis- have had to invest in conventional communications
tical tests applied to two different samples helped confirm media to achieve the same result. It is understandable;
the validity and the reliability of this measurement tool. WOM and the essence of these companies are one on
the Web.
Contributions to Scholarship Although these companies rarely have the means to
control their members’ and nonmembers’ word-of-
The e-WOM Scale adapted to the e-services domain mouth, thanks to the e-WOM Scale, they now have the
helped double the number of measured dimensions with opportunity to measure its impact and to make the neces-
respect to the scale of Harrison-Walker (2001), while sary adjustments. This scale fills a void whereby an
slightly increasing the number of items (from six to organization is now able to assess how its brand and
nine). Thus, the e-WOM Scale enabled us to grasp in a products are communicated through WOM. It enables
more complete manner the concept of word-of-mouth the organization to measure what is being said about it
without complicating the scale by increasing the number (content), the scope of what is being said (intensity), and
of items. This increased precision provides a better the surfers’ attitude towards the organization (positive or
understanding of the components of word-of-mouth. For negative valence). The company will then have the
example, Burzynski and Bayer (1977) spoke of valence capability to see if, for instance, the valence being
in a global way. The e-WOM Scale, on the other hand, expressed is compatible with the content or if the inten-
shows that valence is divisible into two distinct dimen- sity correlates with the valence or the content. In the
sions: positive valence—frequently described as praise— event that the results observed are not the results sought
and negative valence. This result is similar to the results by management, managers will be able to retarget the

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 15 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

message to their advantage by propping up those aspects strength of the association between WOM and future
that are seen as weak in the e-WOM Scale. For example, sales would have to be assessed in further research.
should those surfers that are favourably disposed towards The multidimensional WOM measurement scale
the company (positive valence) be encouraged to speak developed in the context of electronic services is an
of it more (i.e., increased intensity through specials, for anchor point in the pursuit of a reliable and valid scale.
instance), or should the message target those that trans- Since no measurement scale has been developed to date
mit negative valence to convince them to change their in the context of e-services, this study contributes to the
minds? It is obvious that the measurement tool helps advancement of research in the domains of WOM and
target and adapt the steps to be undertaken to use WOM electronic services.
to its full potential.
The quality of the scale (validity and reliability) and
its relative length (nine items) make its application References
simple and accessible to every manager.
ABC-Netmarketing.com (July 7, 2009). Buzz marketing ou
marketing viral? Retrieved from http://www.abc-
netmarketing.com/article.php3?id_article = 1765
Limitations and Future Research Directions Anderson, E.W. (1998). Customer satisfaction and Word-of-
Mouth. Journal of Service Research, 1(1), 5–17.
The newness of the WOM measure makes the Anderson, J.C., & Gerbing, D.W. (1988). Structural equation
development of a multidimensional measurement scale modeling in practice: A review and recommended two
more urgent as this important topic should develop on step approach. Psychological Bulletin, 103(May),
sound methodological grounds. With its four distinct 411–423.
dimensions, the e-WOM Scale appears to be the most Arndt, J. (1967a). Word of Mouth Advertising: A review of the
complete WOM measurement tool to date. However, literature. New York: The Advertising Research Founda-
this does not suggest that there are no new or second- tion Inc.
Arndt, J. (1967b). Role of product-related conversations in the
order factors underlying the four e-WOM Scale
diffusion of a new product. The Journal of Marketing
components. Research, 4(2), 291–295.
Another avenue of research is to apply the e-WOM Arndt, J. (1968). Selective processes in Word of Mouth.
Scale to word of mouth related to brick-and-mortar com- Journal of Advertising Research, 8(3), 19–22.
panies. The tool would help compare WOM components Bagozzi, R.P., & Heatherton, T.F. (1994). A general approach
showing similarities or differences depending on what to representing multifaceted personality constructs. Struc-
respondents say about the same company based on their tural Equation Modeling, 1(1), 35–67.
in-person or online experience. Bagozzi, R.P., & Philips, L.W. (1982). Representing and
This research has demonstrated the need for com- testing organizational theories. Administrative Science
panies to measure WOM. It is also imperative proves Quarterly, 27(3), 459–489.
imperative to identify the antecedents and consequences Bagozzi, R.P., & Yi, Y. (1988). On the evaluation of structural
equation models. Journal of the Academy of Marketing
of WOM in the context of e-services. As for WOM ante-
Science, 16(1), 74–79.
cedents that were investigated in conventional markets, Baumgartner, H., & Homburg, C. (1996). Applications of
we note client trust (Bergeron et al. 2003), service quality structural equation modeling in marketing and consumer
(Parasuraman et al., 1988), and satisfaction (Anderson, research: A review. International Journal of Research in
1998). As far as the consequences of WOM are con- Marketing, 13(2), 139–161.
cerned, many researchers have found impacts on rela- Bentler, P.M. (2005). EQS 6.1: Structural Equation Program
tionship quality (Boles et al., 1997) and customer Manual. Encino, CA: Multivariate software.
purchasing intention (Crocker, 1986; Maxham III, 2001; Bergeron, J., Ricard, L., & Perrien, J. (2003). Les déterminants
Ying & Chung, 2007). Given the power of WOM online, de la fidélité des clients commerciaux dans l’industrie
it would be appropriate to study these relationships in the bancaire canadienne. Canadian Journal of Administrative
context of e-services. The e-WOM Scale would help Sciences, 20(2), 107–120.
Black, D.S.S., Mitra, K., & Webster, C. (1998). Word-of-
measure WOM both against its antecedents and its
Mouth communications: A motivational analysis.
consequences. Advances in Consumer Research, 25, 527–531.
Since many researchers have established a signifi- Boles, J.S., Barksdale, H.C. Jr., & Johnson, J.T. (1997). Busi-
cant association between word of mouth and intention to ness relationships: An examination of the effects of buyer-
purchase (Crocker, 1986; Maxham III, 2001; Ying & salesperson relationships on customer retention and
Chung, 2007), the e-WOM Scale could also be useful in willingness to refer and recommend. Journal of Business
predicting future demand for e-services companies. The & Industrial Marketing, 12(3/4), 248–258.

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 16 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

Bone, P.F. (1992). Determinants of Word-of-Mouth communi- Harrison-Walker, L.J. (2001). The measurement of Word-of-
cation during product consumption. Advances in Con- Mouth communication and an investigation of service
sumer Research, 19, 579–583. quality and customer commitment as potential anteced-
Bone, P.F. (1995). Word of Mouth effects on short-term and ents. Journal of Service Research, 4(1), 60–75.
long-term product judgments. Journal of Business Hartline, M.D., & Jones, K.C. (1996). Employee performance
Research, 32(3), 213–223. cues in a hotel service environment: Influence on per-
Brown, J.J., & Reingen, P.H. (1987). Social ties and Word-of- ceived service quality, value and word-of-mouth inten-
Mouth referral behavior. Journal of Consumer Research, tions. Journal of Business Research, 35(March),
14(3), 350–362. 207–215.
Burzynski, M.H., & Bayer, D.J. (1977). The effect of positive Hattie, J.R. (1985). Methodological review: Assessing uni-
and negative prior information on motion picture appre- dimensionality of tests and items. Applied Psychological
ciation. The Journal of Social Psychology, 101(1), Measurement, 9(June), 139–164.
215–218. Haywood, K.M. (1989). Managing Word of Mouth communi-
Calder, B.J., Philips, L.W., & Tybout, A.M. (1981). Design cations. The Journal of Services Marketing, 3(2), 55–67.
research for application. Journal of Consumer Research, Hennig-Thurau, T., Gwinner, K.P., & Gremler, D.D. (2002).
8(1), 197–206. Understanding relationship marketing outcomes. Journal
Churchill, G.A. (1979). A paradigm for developing better mea- of Service Research, 4(3), 230–247.
sures of marketing constructs. Journal of Marketing Hennig-Thurau, T., Gwinner, K.P., Walsh, G., & Gremier,
Research, 16(1), 64–73. D.D. (2004). Electronic Word-of-Mouth via consumer-
Cox, D., & Cox, A.D. (2002). Beyond first impressions: The opinion platforms: what motivates consumers to articulate
effects of repeated exposure on consumer liking of vis- themselves on the Internet?. Journal of Interactive Mar-
ually complex and simple product designs. Journal of the keting, 18(1), 38–52.
Academy of Marketing Science, 30(2), 119–130. Herr, P.M., Kardes, F.R., & Kim, J. (1991). Effects of Word-
Crocker, K.E. (1986). The influence of the amount and type of of-Mouth and product-attribute information on persua-
information on individuals’ perception of legal services. sion: An accessibility-diagnostic perspective. Journal of
Journal of the Academy of Marketing Science, 14(4), Consumer Research, 17(4), 454–462.
18–27. Higie, R.A., Feick, L.F., & Price, L.L. (1987). Types and
Day, G.S. (1971), Attitude change, media, and word of mouth. amount of Word-of-Mouth communications about retail-
Journal of Advertising Research, 11(6), 31–40. ers. Journal of Retailing, 63(3), 260–279.
Desormeaux, R., & Labrecque, J. (1999). La mesure de la Katz, E., & Lazarsfeld, P.F. (1955). Personal influence.
satisfaction de la clientèle, Gestion, 24(2), 74–81. Glencoe, IL: Free Press.
Duhan, D.F., Johnson, S.D., Wilcox, J.B., & Harrell, G.D. Kim, W.G., Han, J.S., & Lee, E. (2001). Effects of relationship
(1997). Influences on consumer use of Word-of-Mouth marketing on repeat purchase and Word of Mouth. Journal
recommendation sources. Journal of the Academy of Mar- of Hospitality & Tourism Research, 25(3), 272–288.
keting Science, 25(4), 283–295. Mangold, W.G., Miller, F., & Brockay, G.R. (1999). Word-of-
File, K.M., Cermark, D.S.P., & Prince, R.A. (1994). Word-of- Mouth communication in the service marketplace. Journal
Mouth effects in professional services buyer behavior. The of Services Marketing, 13(1), 73–89.
Service Industries Journal, 14(3), 301–314. Maxham III, J.G. (2001). Service recovery’s influence on con-
File, K.M., Judd, B.B., & Prince, R.A. (1992). Interactive mar- sumer satisfaction, positive Word-of-Mouth, and purchase
keting: The influence of participation. Journal of Services intentions. Journal of Business Research, 54(1), 11–29.
Marketing, 6(4), 5. McKelvey, W.W. (1976). An approach for developing shorter
Fornell, C., & Larcker, D.F. (1981). Evaluating structural equa- and better measuring instruments. Working Paper 76–6.
tion models with unobservable variables and measurement Human Systems Development Study Center, Graduate
error. Journal of Marketing Research, 18(February), School of Management, UCLA.
39–50. Morin, S.P. (1983). Influentials advising their friends to sell
Gaski, J. (1984). The theory of power and conflict in channels lots of high-tech gadgetry. Wall Street Journal, February
of distribution. Journal of Marketing, 48(Summer), 9–29. 28, 30.
Godes, D., & Mayzlin, D. (2004). Se servir des conversations Murray, K.B. (1991). A Test of Services Marketing Theory :
en ligne pour étudier le bouche-à-oreille. Recherche et Consumer Information Acquisition Activities. Journal of
Applications en marketing, 19(4), 89–111. Marketing, 55(1), 10–25.
Godin, S., (2001). Les secrets du marketing viral : le bouche- Parasuraman, A., Zeithaml, V.A., & Berry, L.L. (1988).
à-oreille à la puissance 10!. Paris : Maxima Laurent Du SERVQUAL: A multiple-item scale for measuring cus-
Mesnil Éditeur, 197p. tomer perceptions of service quality. Journal of Retailing,
Goyette, I. (2007). Élaboration d’une échelle de mesure mul- 64(1). 12–40.
tidimensionnelle du bouche-à-oreille dans le secteur des Peterson, R.A. (1994). A meta-analysis of Cronbach’s coeffi-
services électroniques. Mémoire de maîtrise, École des cient alpha. Journal of Consumer Research, 21(2),
Sciences de la Gestion, UQAM, 149p. 381–391.

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 17 27(1), 5–23 (2010)
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT GOYETTE ET AL.

Phelps, J.E., Lewis, R., Mobilio, L., Perry, D., & Raman, N. ior: a clickstream analysis. Journal of Business Research,
(2004). Viral marketing or electronic Word-of-Mouth 58, 1599–1608.
advertising: Examining consumers responses and motiva- Sheth. J.N. (1971). Word-of-Mouth in low-risk innovations.
tions to pass along email. Journal of Advertising Research, Journal of Advertising Research, 11(3), 15–18.
44(4), 333–348. Silverman, G. (2001). The Power of Word of Mouth. Direct
Ranaweera, C., & Prabhu, J. (2003). On the relative importance Marketing, 64(5), 47.
of customer satisfaction and trust as determinants of cus- Singh, J. (1990). Voice, exit, and negative Word-of-Mouth
tomer retention and positive word of mouth. Journal of behaviors: An investigation across three service catego-
Targeting, Measurement and Analysis of Marketing, ries. Journal of the Academy of Marketing Science, 18(1),
12(1), 82–90. 1–15.
Reicheld, F.F. (1996). The loyalty effect. Boston, MA: Harvard Stanbouli, K. (2003). Marketing viral et publicité. Revue Fran-
Business School Press. çaise du Marketing, 192/193(May) , 97–106.
Reichheld, F.F., & Sasser, W.E. (1990). Zero defections: Statistics Canada (September 23, 2004). Achats en ligne:
Quality comes to services. Harvard Business Review, magasinage des ménages sur Internet. Retrieved from
5(September/October), 105–111. http://www.statcan.ca/Daily/Francais/040923/q040923a.
Rey-Debove, F. & Rey, A. (2007). Le Nouveau Petit Robert : htm
Dictionnaire alphabétique et analogique de la langue Swan, J.E., & Oliver, R.L. (1989). Postpurchase communica-
française, Paris, Éditions Le Robert. tions by consumers. Journal of Retailing, 65(4),
Richins, M.L. (1983). Negative Word-of-Mouth by dissatisfied 516–533.
consumers: A pilot study. Journal of Marketing, 47, Westbrook, R.A. (1987). Product/consumption-based affective
68–78. responses and postpurchase processes. Journal of Market-
Salzman, M., Matathia, I., & O’Reilly, A. (2004). Buzz . . . Le ing Research, 24(3), 258–270.
marketing du bouche-à-oreille. Paris: Village Mondial. Whyte, W.H. Jr. (1954). The web of Word of Mouth. Fortune,
Sénécal, S., & Nantel, J. (2004). The Influence of online 50(5), 140–143.
product recommendations on consumer’s online choices. Ying H.L., & Chung C.M. (2007). The effects of single-mes-
Journal of Retailing, 80(2), 159–169. sage single-source mixed Word-of-Mouth on product
Sénécal, S., Kalczynski, P.J., & Nantel, J. (2005). Consumers’ attitude and purchase intention. Asia Pacific Journal of
decision making process and their online shopping behav- Marketing and Logistics, 19(1), 75–93.

Can J Adm Sci


Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd. 18 27(1), 5–23 (2010)
Appendix A

Word-of-Mouth Measurement Scales

Authors Type of Cronbach’s Methodology WOM measurement context WOM Viewpoint for
measurement Alpha statements WOM
scale measurement

Arndt (1967b) Information not Information Personal interview Since each respondent was questioned about comments received Information Receiver (Harrison-
available (NA) not Respondents: married and given, the conversations could be cross-checked by not Walker, 2001)
available female students comparing the questionnaires of both parties in the available and
(NA) Sector: food products conversation. (p.291) (NA) communicator
(Goyette, 2007)
Sheth (1971) NA NA Personal interview (1) Respondents were asked to recollect the time they became NA Receiver and
Respondents: men aware of the new blades, what source informed them for the communicator
Sector: razor blades first time. (Goyette, 2007)
(2) Respondents were asked whether they adopted them

Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd.


immediately after becoming aware or some time later, and
whether friends and other personal informal sources were
influential in their adoption decision.
(3) Respondents were also asked if they had influenced someone

19
else after their own adoption. (p.16)
Burzynski and Unidimensional NA Field experiment and Respondents were exposed to either positive, negative, or no NA Receiver (Harrison-
Bayer (1977) self-administered comments about the film they were about to see. (p.216) Walker, 2001)
questionnaire (1) positive prior information: “I wouldn’t mind seeing this one
Respondents: adults again,” and “The acting was fantastic, but the plot was even
going to see a movie better”. (p.216)
Sector: cinema (2) negative prior information: “You couldn’t pay me to see that
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT

thing again” and “Well, another two bucks shot,” (p.216)


Richins (1983) Unidimensional NA Self-administered The act of telling at least one friend or acquaintance about the NA Receiver (Harrison-
questionnaire dissatisfaction. (p. 71) Walker, 2001)
Respondents: adult and
consumers communicator
Sector: clothing and (Harrison-
apparel (electronic) Walker, 2001)
Brown and Unidimensional NA Telephone interview NA “Who-told- Receiver (Harrison-
Reingen Respondents: whom- Walker, 2001)
(1987) professors, adults, about-the- and
piano students. service” (p. communicator
Sector: music 351) (Goyette, 2007)

Goyette (2007), p.28–35


GOYETTE ET AL.

27(1), 5–23 (2010)


Can J Adm Sci
Authors Type of Cronbach’s Alpha Methodology WOM measurement context WOM Viewpoint for
measurement measurement WOM
scale items measurement

Higie, Feick, and NA NA Telephone questionnaire The respondents were given the opportunity to recall NA Communicator
Price (1987) Respondents: residents of a and report “any other attributes of retail outlets that (Harrison-
metropolitan region in they discussed with other people.” Dimensions which Walker, 2001)
north-eastern United pilot respondents did not discuss (e.g., location and
States. store hours) and topics that were too broad or general
Sector: retail sales (e.g, service) were dropped (p.266)
Dimensions included in the final instrument were:
(1) quality of merchandise, (2) special sales,
(3) usual or everyday prices, (4) helpfulness and
friendliness of employees, (5) variety of products
available, (6) availability of particular brands,
(7) return policy.
The response categories were: never, a few times a
year, about once a month, a few times a month, and
once a week or more.

Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd.


Westbrook NA Reliability index: Personal interview and self- (1) Reported frequency of discussions with others about NA Communicator
(1987) 0.86 (measured administered CATV and local cable operator. (Goyette,
by frequency, questionnaire (2) Number of persons involved. 2007)
number of Respondents: adult men and (3) Number of topics discussed. (p.263)

20
persons and women
number of Sector: automotive and
topics) cable TV
Swan and Oliver Unidimensional NA Self-administered NA (1) “Did you say Communicator
(1989) seven-point questionnaire mostly positive (Goyette,
Likert. Respondents: new car or mostly 2007)
buyers negative things
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT

Sector: automotive about the car.”


(2) “Did you
recommend
“buy the car”
or “not buy the
car”.” (p. 522)
Singh (1990) Unidimensional NA Self-administered NA “Told my friends Communicator
dichotomous, questionnaire by mail and relatives (Goyette,
(0) for no Respondents: Households about my bad 2007)
and (1) for that have had an experience.”
yes. unsatisfactory experience (p. 7)
with one of the services
being studied.
Sector: grocery trade,
vehicle repair, medical
services
GOYETTE ET AL.

27(1), 5–23 (2010)


Can J Adm Sci
Authors Type of Cronbach’s Methodology WOM measurement context WOM measurement items Viewpoint for
measurement Alpha WOM measurement
scale

Herr, NA NA Experiment The respondents are subjected to a NA Receiver (Harrison-


Kardes, Subjects: students negative OR positive conversation. Walker, 2001)
and Kim (undergraduate) (1) WOM (+): “It’s the best car he’s
(1991) Experimental unit: printed ever had. He hasn’t spent a dime
or verbal information on repairs since he bought it. He
about a product says if it ever wears out he’ll get
another just like it.” (2) WOM (−):
“It’s the worst car he’s ever had. It
seems like it’s always in the shop
being repaired. I think he’s spent
more to keep it running than it
originally cost him.” (p. 458)
Bone (1992) Multidimensional The three Individual self-administered NA (1) “We did not talk about the food Communicator
with three statements questionnaire at all” and “We talk about the (Goyette, 2007)
statements were Respondents: group of two food a lot.”
standardized or more after a meal. (2) Whether the food eaten was a

Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd.


and added for Sector: restaurants large part of the mealtime
αWOM = 0.79 conversation.
(3) How much of the table
conversation dealt with the food

21
being eaten. Responses ranged
from “nothing was said about the
food” to “the biggest topic of
conversation was our food.”
(p. 580)
File, Judd, Unidimensional NA Questionnaire WOM for this study refers to: NA Receiver (Harrison-
and Respondents: clients of an “recommending the firm and the Walker, 2001) &
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT

Prince attorney service to others as well as communicator


(1992) Sector: Trust and estate communications with the firm.” (Goyette, 2007)
planning (p. 6)
File, Unidimensional NA Structured personal NA “A personal endorsement of the Communicator
Cermark, interview (professional services firm) from a (Harrison-Walker,
and Respondents: company business associate in the decision 2001)
Prince executives to retain the professional service
(1994) Sector: professional provider.”
services (consultants) “Telling other business associates
what you thought of the (service
provider).” (p. 308).
Bone (1995) NA NA Experimental presentation. The respondents were exposed to NA Receiver (Harrison-
Complete a card on either a positive conversation or Walker, 2001)
product expectations and with a negative conversation
performance. between two students. (p. 217).
Respondents: students
Test product: biscuit
GOYETTE ET AL.

27(1), 5–23 (2010)


Can J Adm Sci
Authors Type of Cronbach’s Methodology WOM measurement context WOM measurement Viewpoint for
measurement scale Alpha items WOM measurement

Anderson (1998) Unidimensional NA Telephone Respondents report word-of-mouth activity in terms of “Willingness to Communicator
interview the number of individuals spoken to about recent recommend”, (Harrison-Walker,
Respondents: users experiences with quality. (p. 10) “recommendations to 2001)
Sector: various others” (p. 6)
Black, Mitra, Information NA Critical incident The respondents were asked to provide details on a “Exactly what did you Communicator
and Webster unavailable technique recent positive WOM and a negative WOM. The tell the other person (Goyette, 2007)
(1998) Respondents: respondents were probed about the product discussed, and what motivated
persons stopped approximate time period of conversation with the you to share this
on location other person, specifics of the WOM conversation, and particular
Sector: various motivations behind the WOM. (p.527) experience?” (p. 528)
businesses.

Mangold, Miller, Multidimensional NA Self-administered The respondents were then asked to think about the last NA Receiver (Goyette,

Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd.


and Brockway questionnaire time someone told them something positive about a 2007)
(1999) Respondents: service that they may have had an interest in
students purchasing. Then, they were asked to think about the
Sector: 77 different last time someone told them something negative about

22
services a service that they may have had an interest in
purchasing. Thus, each respondent reported on two
WOM communication incidents, one positive and the
other negative. In regard to each event, respondents
were asked: (1) what service they were thinking about,
(2) how long ago the communication had occurred, (3)
what was said, (4) respondents’ relationship with the
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT

person to whom they were speaking, (5) how the


particular WOM came about, (6) whether the WOM
was part of a broader conversation; and, if so, (7) how
that broader conversation came about.
Kim, Han, and Unidimensional αWOM: 0.80 Self-administered “Desire to recommend a hotel to other people and (1) “I am willing to tell Communicator
Lee (2001) seven-point questionnaire willingness to say good things about a hotel to other other people about (Goyette, 2007)
Likert Respondents: hotel people” (p. 279) the good aspects of
customers (Influences: Swan and Oliver (1989) this hotel”
Sector: hospitality (2) “I am willing to
recommend this
hotel to others”
(p. 281)
GOYETTE ET AL.

27(1), 5–23 (2010)


Can J Adm Sci
Authors Type of Cronbach’s Methodology WOM measurement context WOM measurement items Viewpoint for
measurement Alpha WOM
scale measurement

Harrison-Walker Multidimensional α WOMpraise = Self-administered Based on 13 statements (cf. right- Since I have been with this service organization, I have Communicator
(2001) 7-point Likert 0.80 questionnaire hand column) created to mentioned the name of this service organization very (Harrison-
with α WOMactivity = Respondents: measure 4 aspects of WOM. rarely. 2) I mention this service organization to Walker, 2001)
“completely 0.78 Consumers of one of Statements (1) to (3) : Frequency others quite frequently. 3) I rarely have occasion to
agree” and the two services Statements (4) to (6) : Number of mention the name of this organization to others. 4)
“completely under investigation. contacts I’ve told more people about this service organization
disagree” at Sector: veterinary clinic Statements (7) to (9): Details than I’ve spoken about most other service
either end. and hair salon Statements (10) to (13): Praise organizations. 5) I seldom miss an opportunity to tell
Following a factor analysis in others about this service organization 6) I’ve told
principal components, the author very few people about this service organization. 7)
cleaned up the measurement When I tell others about this service organization, I
scale retaining only the tend to talk about the organization in great detail. 8)
statements in bold in the right- I seldom do more than mention the name of this
hand column. service organization to others. 9) Once I get talking
Statements (2), (4), (5), and (7) : about this service organization, it’s hard for me to
WOM activity stop. 10) I have only good things to say about this

Copyright © 2010 ASAC. Published by John Wiley & Sons, Ltd.


Statements (10) and (13): service organization. 11) Although I use this service
Praise WOM organization, I tell others that I do not recommend it.
Statements (1), (3), (6), (8), (9), 12) In general, I do not speak favourably about this
(11), and (12): service organization. 13) I am proud to tell others

23
Rejected during scale that I use this service organization . . . (p. 72–73).
purification process
Hennig-Thurau, Unidimensional NA Self-administered Information unavailable “I often recommend this service provider to others” Communicator
Gwinner, questionnaire (p. 245) (Goyette, 2007)
Gremler Respondents: students
(2002) (undergraduate)
Sector: service
Ranaweera and Two-item αWOM: 0.79 Qualitative interview (1) Recommendation of the service NA Communicator
e-WOM SCALE: WORD-OF-MOUTH MEASUREMENT SCALE FOR e-SERVICES CONTEXT

Prabhu (2003) measurement (statements Respondents: consumers (2) Involuntary recommendation and receiver
scale unavailable) Sector: phone service based on the two-dimensional (Goyette, 2007)
typology, which identified two
key types of WOM – receiver
initiated and sender initiated (p.
85)
Godes and NA NA Direct observation of Measure WOM on the Internet (1) Volume: “what is the scope of word of mouth?” Communicator
Mayzlin (2004) interpersonal using online conversations. (p. 94) and receiver
conversations Study of two distinct WOM (2) Dispersion: “extent and diversity of virtual (Goyette, 2007)
(thousands of dimensions: volume and communities in which conversations on a given
discussion forums) dispersion (p. 94). product are found” (p. 90)
on the “Usenet” site.
Respondents: audience
of a new TV series
Sector: new TV series
(44)
GOYETTE ET AL.

27(1), 5–23 (2010)


Can J Adm Sci

You might also like