You are on page 1of 7

CYBERPSYCHOLOGY & BEHAVIOR

Volume 12, Number 2, 2009


Mary Ann Liebert, Inc.
DOI: 10.1089/cpb.2008.0109

Rapid Communication

How Consumers Evaluate eWOM


(Electronic Word-of-Mouth) Messages
Sun-Jae Doh, M.S. and Jang-Sun Hwang, Ph.D.

Abstract

This experiment explored how consumers evaluate electronic word-of-mouth (eWOM) messages about products. Each participant was exposed a 10-message set in a single board. Five groups were manipulated in terms
of their ratios of positive and negative messages. The result showed significant differences across various sets
of eWOM messages. Although more positive sets showed higher scores in many cases, this was not true in all
situations, especially for the case of credibility. Involvement and prior knowledge partially moderated the relationship between the ratio of messages and the eWOM effect. The credibility of Web sites and eWOM messages can be damaged in the long run if all of the eWOM messages are positive.

Introduction

INTERNET, eWOM (electronic word-of-mouth) has become an important influence on consumers product evaluation. Prospective customers visit Web sites and read reviews from other
customers (eWOM) to learn more about a product before
making a purchase. In South Korea, while traditional media have shown a drastic decline as information sources,
80% of consumers refer to the postings about products or
customer reviews on the Internet when they need product
information.1
Because of its anonymous nature and wide range of contents, the power of eWOM is expanding. There are several
critical antecedents of eWOM effects. First, the direction of
eWOM messages (positivenegative) affects the customers
(readers) response: customers are more likely to rely on
eWOM messages if the direction of the messages are all the
same. The consensus in eWOM represents the degree of
agreement between two or more users regarding a product
or its performance.2 Therefore, the eWOM messages with
higher consensus can be more persuasive and powerful than
messages with lower consensus.
ITH THE EMERGENCE OF THE

H1: The more positive sets of multiple eWOM messages


would yield higher eWOM effects than the less positive
sets.

Second, researchers widely agree that consumer-related factors, such as involvement with and prior knowledge about the
product, greatly influence word-of-mouth effects.3 As widely
known, consumers tend to elaborate their information processing as their involvement with the product is increased. It
is necessary to examine their relationships in the context of
eWOM communication. Thus, the following hypotheses were
proposed to investigate the interaction effects between these
factors and the direction of messages on eWOM effects.
H2: There would be interaction effects between the involvement with a product and the ratio of positive messages
on the eWOM effects.
H3: There would be interaction effects between prior
knowledge about a product and the ratio of positive messages on the eWOM effects.

The current study investigated the causal relationship between the ratio of eWOM messages (positivenegative) and
eWOM effects with the moderating roles of involvement and
prior knowledge. The research design in Figure 1 represents
the hypothetical relationships among these variables.
Methods
One hundred forty-three samples were collected in three
universities in South Korea with self-administered question-

Department of Advertising and Public Relations, Chung-Ang University, Seoul, Korea.

193

194

DOH AND HWANG

eWOM effect
Involvement
Attitude toward the Product

Purchase Intention

The Ratio of Messages


(PositiveNegative)

Attitude toward the Web site


Prior
Knowledge

FIG. 1.

Credibility of eWOM Messages

Research design.

naires employing two product categories: movies and digital


cameras. Participants were exposed to an experimental Web
site containing different types of eWOM messages. All messages were originally collected from real Web sites and modified to be appropriate for the study. Each eWOM message set
contained 10 individual messages, and the ratio of positive to

negative messages varied with each set. The number of positive messages ranged from 6 to 10 in five sets, and the rest
consisted of negative messages. Thus, the ratios of positive to
negative messages were 10:0, 9:1, 8:2, 7:3, and 6:4. With
pretests, all sets of individual eWOM messages were carefully
manipulated to control biases such as the intensity and length

Collecting the eWOM cases about products from the actual customer forum

Retrieval and revision of the eWOM message cases


(15 positive cases, 10 negative cases about movie & camera respectively)

Evaluating the intensity of the individual messages with a pretest


(Movie: positive mean  4.36; negative mean  4.63
Camera: positive mean  4.66; negative mean  4.53)

Establishing the all-positive set (positivenegative  10:0) for each product case
Exclude 5 cases of maximum and minimum scores
The mean score of final 10 messages the initial mean score of 15 messages

Changing the less positive set from 9:1 to 6:4


Replace one message case of the all-positive set (10:0) with one negative case

FIG. 2.

The procedure to set the experimental eWOM cases with a pretest.

EWOM (ELECTRONIC WORD-OF-MOUTH), ONLINE CONSUMER BEHAVIOR


TABLE 1.

MEAN SCORES

eWOM Effects
Attitude toward the product
Purchase intention
Attitude toward Web site
Credibility of eWOM

OF

EACH GROUP

FOR THE

195

EFFECTIVENESS

OF EWOM

10:0

9:1

8:2

7:3

6:4

Total

5.11
(0.71)
4.28
(0.95)
4.21
(0.90)
4.11
(0.87)

4.87
(1.20)
3.71
(1.48)
4.54
(0.99)
4.03
(1.17)

4.72
(0.87)
3.76
(1.39)
4.46
(0.62)
4.32
(0.73)

4.34
(0.76)
3.39
(0.99)
4.21
(0.54)
4.04
(0.72)

3.99
(0.77)
3.44
(1.11)
3.85
(0.86)
3.71
(0.81)

4.61
(0.95)
3.72
(1.23)
4.25
(0.83)
4.04
(0.88)

Standard deviation shown in parentheses.

customer reviews (m  4.94 in the 7-point scale). In order to


examine the proposed hypotheses, statistical analyses including analysis of variance (ANOVA) and general linear
model (GLM) were conducted. Following are the findings
for each hypothesis.

of message. Therefore, the mean scores of message directions


in five eWOM sets were almost equal. The intervals between
each group, from 10:0 to 6:4, were kept approximately equal,
and statistical differences among them were confirmed (F 
31.83, p  0.001). Figure 2 illustrated how experimental sets of
eWOM messages with five combinations of positive and negative directions based on a pretest were manipulated.
The dependent variables were attitude toward the product,
purchase intention, credibility of eWOM messages, and attitude
toward Web site. Each construct was measured with multiple
items that were developed and tested in previous studies.45
All variables measured with multiple items were confirmed
to have reliabilities with alphas of 0.74 or higher.

Influences of eWOM message directions (H1)


H1 examined how various combinations of eWOM messages could influence the eWOM effect in terms of attitude
toward the product, purchase intention, attitude toward the
Web site, and credibility of eWOM messages. Table 1 shows
the mean scores of each group for four individual measures
of eWOM effects.
As shown in Tables 1 and 2, eWOM effect measures, except for credibility of eWOM, were significantly different
among the groups. Therefore, H1 was strongly supported.
Attitude toward the product showed higher correlations
with the number of positive messages. The group of 10:0
(m  5.10, SD  0.71) showed the highest scores, followed
by groups of 9:1, 8:2, 7:3, and 6:4 respectively. As shown with
post hoc tests (Table 2), the group of 10:0 showed significantly higher attitudes toward the products than did groups

Results
Of the sample, 96.5% had made online purchases of one
or more items and did so, on average, nine times per year.
We found that participants frequently rely on eWOM messages with high credibility. Specifically, 97.9% of them usually referred to customer reviews (eWOM) prior to making
online purchases. They read an average of 13.9 reviews per
purchase. They also gave relatively high credibility to the

TABLE 2.
Sources
Attitude toward the product
Model
Error
Total
Purchase intention
Model
Error
Total
Attitude toward Web site
Model
Error
Total
Credibility of eWOM
Model
Error
Total
*p  0.05; ***p  0.001.

DIFFERENCES

IN EWOM

EFFECTS

AMONG THE

GROUPS

SS

df

MS

LSD

22.492
106.162
128.654

4
138
142

5.623
0.769

7.309***

10:07:3, 6:4
9:17:3, 6:4
8:26:4

14.716
197.266
211.982

4
137
141

3.679
1.440

2.555*

10:07:3, 6:4

8.067
89.148
97.215

4
138
142

2.017
0.646

3.122*

9:16:4
8:26:4

5.361
104.422
109.783

4
137
141

1.340
0.762

1.758

8:26:4

196

DOH AND HWANG

of 7:3 and 6:4, while the groups with higher positive messages, including 10:0, 9:1, and 8:2, showed no significant differences from the others.
These results imply that a perfect set of eWOM messages,
such as 10:0, is not required to improve the customers attitude toward the product. In other words, although positive
messages should be helpful in promoting positive attitudes
toward the products, a few negative messages within the majority of positive messages are not critically harmful. Thus,
H1 was supported for the case of attitude toward the product.
Purchase intention showed similar patterns to those of attitude toward the product, and the result also supported H1.
The highest score was shown in the group of 10:0, followed
by 8:2, 9:1, 7:3, and 6:4respectively. The group of 10:0 showed
the highest mean score, with statistical differences from the
groups of 7:3 and 6:4. Other comparisons showed no significant differences in the post hoc test.
Attitude toward Web site showed somewhat interesting
results. In fact, the group of 9:1, with one negative message,
yielded the highest score (m  4.54), and the following
group, 8:2, showed the second-highest mean score of 4.46.
The all-positive group (10:0) ranked the third-highest group,
tied with the group of 7:3. With the post hoc test, the groups
of 9:1 and 8:2 showed significant differences from the lowest group of 6:4, while the group of 10:0 did not. This result
implies that a few negative messages should improve the
performance of Web sites, while a perfect set of eWOM messages is not helpful, as indicated by our findings regarding
credibility of eWOM.
Although the overall ANOVA results did not show significant differences among five groups for credibility of
eWOM, the group of 8:2 showed the highest score, followed
by 10:0, 9:1, and so forth. With a sophisticated post hoc test,
this could be confirmed. Specifically, the group of 8:2 yielded
higher credibility than the group of 6:4, while the group of
10:0 did not show any significant differences from other
groups. This result is likely due to consumers suspicion that
the set of perfect positive messages stems from corporate unethical behavior such as stealth marketing activities. Because
some companies reportedly try to manipulate and manage
the voice of general consumers with intentional interruptions
in online user-to-user communication, some participants did
not trust the sincerity of perfect sets with all positive messages. This suspicion might influence the results regarding
the attitude toward Web site, and credibility of eWOM.

TABLE 3.

DIFFERENCES

eWOM Effects
Attitude toward the product
Purchase Intention
Attitude toward Web site
Credibility of eWOM
*p  0.05; ***p  0.001.

IN EWOM

EFFECTS

Role of involvement (H2)


H2 examined the relationship between involvement and
the ratio of positive eWOM messages regarding their influence on consumers responses. At the initial stage of the analysis, however, the interaction effect of all five groups and
two levels of involvement (high and low) did not show any
significant results. Nevertheless, further analyses to compare
individual groups in terms of their interactions with involvement showed some significant results. Thus, H2 was
supported in part.
Specifically, regarding attitude toward the product,
groups 7:3 and 6:4 showed interaction effects with involvement (MS  1.673, F  4.121, p  0.05). At the low-involvement context, group 7:3 showed higher scores than group
6:4, while at the high-involvement condition, their differences were minimal; even group 6:4 showed a slightly higher
score. Therefore, it can be inferred that consumers with
lower-involvement situations are likely to prefer the product with 7:3 combination over the 6:4 set, while the reverse
should be true in a higher-involvement context. A higher involvement situation may cause consumers to become suspicious about the positive messages, and the effect of negative
messages would therefore be increased.
The main effect of involvement was also examined. Most
eWOM effect measures showed significant differences between high- and low-involvement groups. Specifically, for
eWOM effect measures except purchase intention, high-involvement groups showed significantly higher scores than
low-involvement groups (Table 3).
Role of prior knowledge
The last hypothesis examined the role of prior knowledge
about the product. The interaction effect between the ratio
of positive messages and prior knowledge was analyzed.
Similar to the case of involvement, no significant interaction
effect was found when examining the five groups with two
prior knowledge levels.
However, a further analysis based on this initial result
showed some significant interaction effects. Therefore, H3
was partially supported. Groups 7:3 and 6:4 showed significant interaction effect with the level of prior knowledge for
purchase intention (MS  5.663, F  5.356, p  0.05). For the
low prior knowledge group, group 7:3 showed higher scores
for purchase intention than did group 6:4. However, this
turned reversed at the context of higher prior knowledge.

BETWEEN

LOW-

AND

HIGH-INVOLVEMENT GROUPS

Involvement

Mean

SD

Low
High
Low
High
Low
High
Low
High

69
73
68
73
69
73
69
72

4.44
4.79
3.56
3.86
3.93
4.56
3.78
4.30

0.95
0.92
1.31
1.14
0.82
0.71
0.82
0.87

df

140

2.207*

139

1.476

140

4.930***

139

3.687***

EWOM (ELECTRONIC WORD-OF-MOUTH), ONLINE CONSUMER BEHAVIOR


Another eWOM effect measure showed significant interaction effect between the ratio of positive messages and prior
knowledge. In attitude toward Web site, the interaction effect between two sets (8:2 and 9:1) and prior knowledge was
significant (MS  3.111, F  5.639, p  0.05). The group of
lower prior knowledge showed similar scores for attitude toward Web site regardless of ratios of positive messages; consumers with higher prior knowledge showed significantly
different attitude toward Web site for 9:1 (m  5.19) and 8:2
(m  4.33). This finding shows that consumers with higher
prior knowledge can be more sensitive to the negative messages than consumers without prior knowledge. A further
investigation was performed to examine the mean differences between low and high prior knowledge groups for
eWOM effect measures. However, eWOM measures did not
show any significant differences between low and high prior
knowledge groups.
Discussion
A few negative messages can be helpful in promoting positive attitude toward Web site and credibility of eWOM messages. This might be fairly reasonable, because some consumers may suspect the credibility of the Web site or the set
of multiple eWOM messages if they find hardly any negative messages. A single negative message itself can be harmful for product evaluation; however, one negative message
in a 10-message set is not much harmful and even can be
beneficial in the eWOM context. The moderating roles of involvement and prior knowledge were also supported in
some situations.
This study can contribute to the eWOM research field in
some respects. Previous studies simply confirmed that negative messages were more influential than positive ones.
However, this finding should be further investigated with
multiple messages. Moreover, the credibility of individual
messages should be considered in future research. Evaluation of eWOM source credibility can differ from evaluation
of traditional word-of-mouth contexts. Because this study

197

limits its scope to within the online shopping mall, future research needs to explore somewhat different contexts of
eWOM. Although the intimacy and source credibility are not
critical in the context of online shopping malls, they should
play important roles in other types of Web sites, such as online brand communities.
Disclosure Statement
The authors have no conflict of interest.
References
1. National Internet Development Agency of Korea. (2006) The
Internet usage of netizens in the Web 2.0 erathe Internet of participation and sharing. Seoul: Author.
2. Chiou J, Cheng C. Should a company have message boards
on its Web sites? Journal of Interactive Marketing 2003;
17:5061.
3. Richins ML, Root-Shaffer T. The role of involvement and
opinion leadership in consumer word-of-mouth: an implicit
model made explicit. Advances in Consumer Research 1988;
15:326.
4. Bettman JR, Park CW. Effect of prior knowledge and experience and phase of the choice process on consumer decision
process. Journal of Consumer Research 1980; 7:23448.5.
Zaichkowsky JL. Measuring the involvement construct. Journal of Consumer Research 1985; 12:34152.

Address reprint requests to:


Dr. Jang-Sun Hwang
Department of Advertising & PR
College of Political Science & Economics
Chung-Ang University
Heukseok-Dong 221, Dongjak-Gu
Seoul
South Korea 153756
E-mail: seralpha@cau.ac.kr

This article has been cited by:


1. Qiuju Luo, Dixi Zhong. 2015. Using social network analysis to explain communication characteristics of travel-related electronic
word-of-mouth on social networking sites. Tourism Management 46, 274-282. [CrossRef]
2. Jannine Nieto, Rosa M. Hernndez-Maestro, Pablo A. Muoz-Gallego. 2014. Marketing decisions, customer reviews, and business
performance: The use of the Toprural website by Spanish rural lodging establishments. Tourism Management 45, 115-123.
[CrossRef]
3. Christy M.K. Cheung, Bo Sophia Xiao, Ivy L.B. Liu. 2014. Do actions speak louder than voices? The signaling role of social
information cues in influencing consumer purchase decisions. Decision Support Systems 65, 50-58. [CrossRef]
4. Sidharth Muralidharan, Hye Jin Yoon, Yongjun Sung, Jessica Miller, Arturo Lee. 2014. Following the breadcrumbs: An analysis
of online product review characteristics by online shoppers. Journal of Marketing Communications 1-22. [CrossRef]
5. Sue-Ting Chang, Tom M. Y. Lin, Pin Luarn. 2014. The effects of word-of-mouth consistency on persuasiveness. Canadian
Journal of Administrative Sciences / Revue Canadienne des Sciences de l'Administration 31:10.1002/cjas.v31.2, 128-141. [CrossRef]
6. Qian Xu. 2014. Should I trust him? The effects of reviewer profile characteristics on eWOM credibility. Computers in Human
Behavior 33, 136-144. [CrossRef]
7. Andrew J. Flanagin, Miriam J. Metzger, Rebekah Pure, Alex Markov, Ethan Hartsell. 2014. Mitigating risk in ecommerce
transactions: perceptions of information credibility and the role of user-generated ratings in product quality and purchase intention.
Electronic Commerce Research 14, 1-23. [CrossRef]
8. Dongyoung Sohn. 2014. Coping with information in social media: The effects of network structure and knowledge on perception
of information value. Computers in Human Behavior 32, 145-151. [CrossRef]
9. Eun-Ju Lee, Soo Yun Shin. 2014. When do consumers buy online product reviews? Effects of review quality, product type, and
reviewers photo. Computers in Human Behavior 31, 356-366. [CrossRef]
10. Vincent T. P. Cheng, Mei Kun Loi. 2014. Handling Negative Online Customer Reviews: The Effects of Elaboration Likelihood
Model and Distributive Justice. Journal of Travel & Tourism Marketing 31, 1-15. [CrossRef]
11. Luke Greenacre, Lynne Freeman, Karen Cong, Tom Chapman. 2014. Understanding and predicting student Word of Mouth.
International Journal of Educational Research 64, 40-48. [CrossRef]
12. David Eberle, Guido Berens, Ting Li. 2013. The Impact of Interactive Corporate Social Responsibility Communication on
Corporate Reputation. Journal of Business Ethics 118, 731-746. [CrossRef]
13. Jacques Bulchand-Gidumal, Santiago Melin-Gonzlez, Beatriz Gonzlez Lopez-Valcarcel. 2013. A social media analysis of the
contribution of destinations to client satisfaction with hotels. International Journal of Hospitality Management 35, 44-47. [CrossRef]
14. Wen-Hai Chih, Kai-Yu Wang, Li-Chun Hsu, Su-Chen Huang. 2013. Investigating Electronic Word-of-Mouth Effects on Online
Discussion Forums: The Role of Perceived Positive Electronic Word-of-Mouth Review Credibility. Cyberpsychology, Behavior, and
Social Networking 16:9, 658-668. [Abstract] [Full Text HTML] [Full Text PDF] [Full Text PDF with Links]
15. Aurelio G. Mauri, Roberta Minazzi. 2013. Web reviews influence on expectations and purchasing intentions of hotel potential
customers. International Journal of Hospitality Management 34, 99-107. [CrossRef]
16. Hui-Fei Lin, Chi-Hua Chen. 2013. An Intelligent Embedded Marketing Service System based on TV apps: Design and
implementation through product placement in idol dramas. Expert Systems with Applications 40, 4127-4136. [CrossRef]
17. Katherine B. Hartman, James B. Hunt. 2013. What Ratemyprofessors.com Reveals About How and Why Students Evaluate Their
Professors: A Glimpse Into the Student Mind-Set. Marketing Education Review 23, 151-162. [CrossRef]
18. Aaron Tham, Glen Croy, Judith Mair. 2013. Social Media in Destination Choice: Distinctive Electronic Word-of-Mouth
Dimensions. Journal of Travel & Tourism Marketing 30, 144-155. [CrossRef]
19. Yoon Jin Ma, Kim Hahn, Hyun-Hwa Lee. 2012. Effect of Individual Differences on Online Review Perception and Usage
Behavior: The Need for Cognitive Closure and Demographics. Journal of the Korean Society of Clothing and Textiles 36, 1270-1284.
[CrossRef]
20. Christy M.K. Cheung, Dimple R. Thadani. 2012. The impact of electronic word-of-mouth communication: A literature analysis
and integrative model. Decision Support Systems 54, 461-470. [CrossRef]
21. Kyung-Tag Lee, Dong-Mo Koo. 2012. Effects of attribute and valence of e-WOM on message adoption: Moderating roles of
subjective knowledge and regulatory focus. Computers in Human Behavior 28, 1974-1984. [CrossRef]
22. Wen-Chin Tsao, Ming-Tsang Hsieh. 2012. Exploring how relationship quality influences positive eWOM: the importance of
customer commitment. Total Quality Management & Business Excellence 1-15. [CrossRef]

23. Nathalia Purnawirawan, Nathalie Dens, Patrick De Pelsmacker. 2012. Balance and Sequence in Online Reviews: The Wrap Effect.
International Journal of Electronic Commerce 17, 71-98. [CrossRef]
24. Kelley O'Reilly, Sherry Marx. 2011. How young, technical consumers assess online WOM credibility. Qualitative Market Research:
An International Journal 14, 330-359. [CrossRef]
25. Chih-Chien Wang, Yi-Ting Wang. 2010. Persuasion Effect of e-WOM: The Impact of Involvement and Ambiguity Tolerance.
Journal of Global Academy of Marketing Science 20, 281-293. [CrossRef]