 

’ﬂ'IESIS

)C‘E

 

LIBRARY
Michigan State
University

 

This is to certify that the
thesis entitled

THE EFFECT OF USER RATINGS ON PUBLIC OPINION
PERCEPTION AND WILLINGNESS TO EXPRESS OPINIONS: THE
THIRD-PERSON EFFECT AND SPIRAL OF SILENCE THEROY

presented by

Jaehyun Hong

has been accepted towards fulﬁllment
of the requirements for the

Master of Art degree in Communication

 

 

 

 

W

Major Professor's Signature

12/12/2007

 

Date

MS U is an affirmative-action, equal-opportunity employer

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

 

 

 

DATE DUE DATE DUE DATE DUE
5§Fi089a099 o $E9 a o 9009

MAR {I 8 ZUIU
011810

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:ICIRC/DateDue.indd—p.1

A~ 77—27 7, .——* ,, 7 _._._._.7 7‘7 —~ *7 -——___.

THE EFFECT OF USER RATINGS ON PUBLIC OPINION PERCEPTION
AND WILLINGNESS TO EXPRESS OPINIONS: THE THIRD-PERSON
EFFECT AND SPIRAL OF SILENCE THEORY
By

J aehyun Hong

A THESIS
Submitted to
Michigan State University

in partial fulﬁllment of the requirements
for the degree of

MASTER OF ART
Department of Communication

2007

ABSTRACT
THE EFFECT OF USER RATINGS ON PUBLIC OPINION PERCEPTION
AND WILLINGNESS TO EXPRESS OPINIONS: THE THIRD-PERSON
EFFECT AND SPIRAL OF SILENCE THEROY
By

J aehyun Hong

The current study examined the effects of volume and valence of user
ratings of a movie on individuals’ perceived effects of user ratings on self and on
others, and also on individuals’ willingness to express their opinions on-line. For
study 1, undergraduate participants (N = 403) were randomly assigned to one of 2
(volume: low versus high) X 2 (valence: positive versus negative) conditions. For
study 2, undergraduate participants (N = 243) were randomly assigned to one of 4
(volume: low, high, super high, and mega high) x 2 (valence: positive and
negative) conditions. The participants reported the extent to which they believed
the user ratings would inﬂuence themselves and others, indicated their ratings and
their estimated ratings Of others for motives before and after being exposed to the
user ratings, and reported their willingness to express their opinions. The ﬁndings
showed that people believe that others are more affected by user ratings than
themselves. As individuals predicted a greater effect of user ratings on others, they
were more willing to express opinions. Interpretations of these and others results

and implications for public opinion on the Internet are discussed.

ACKNOWLEDGEMENT

I would like to thank my advisor, Dr. Hee Sun Park for Showing me the
passion to study Communication and academic direction. Without her teaching and
encouragement, I could not have continued to study. This thesis would not have
been possible without her advice and help in inspiring the idea of the thesis and
advice during the correcting of the thesis. And I thank my committee members, Dr.
Tim Levine and Dr. Mira Lee, for their critical advice and encouragement, which

guided me to complete my thesis.

iii

TABLE OF CONTENTS

LIST OF TABLES ..................................................................................................... v
INTRODUCTION ..................................................................................................... 1'
LITERATURE REVIEW .......................................................................................... 4
The effect of User Ratings on Audience’s Movie Selection .................................. 4
User Ratings as Word of Mouth ............................................................................ 6
User Ratings as Public Opinion Polls .................................................................... 9
Public Opinion Perception and Third-Person Effect ............................................ 12
Willingness to Express Opinions, Spiral of Silence Theory, and TPE ................ 15
Spiral of silence theory. ................................................................................... 16

The third-person effect and willingness to express opinions ........................... 19
STUDY 1 ................................................................................................................. 21
Method ................................................................................................................. 21
Participants ....................................................................................................... 21
Design .............................................................................................................. 21
Stimulus ........................................................................................................... 22
Measures .......................................................................................................... 23
Results .................................................................................................................. 30
Preliminary Analysis ........................................................................................ 30
Main Analysis Overview ................................................................................. 30

The Third Person Effect of User Ratings on Self and Others .......................... 32
Third Person Effect on Willingness to Express One’s Opinion ....................... 38
STUDY 2 ................................................................................................................. 42
Study 2 Method .................................................................................................... 42
Participants ....................................................................................................... 42
Design .............................................................................................................. 42
Stimulus and Measurement .............................................................................. 43
Study 2 Results ..................................................................................................... 46
The Third Person Effect of User Ratings on Self and Others .......................... 46
Willingness to Express One’s Opinion ............................................................ 48
SUMMARY OF THE FINDINGS OF STUDY I AND STUDY 2 ........................ 49
DISCUSSION .......................................................................................................... 51
APPENDIX .............................................................................................................. 58
BIBLIOGRAPHY .................................................................................................... 62

iv

LIST OF TABLES
Table 1. Reliabilities, Means, Standard Deviations, and Correlations of Variables
of Study 1. .................................................................................................. 29
Table 2. Means and Standard Deviations of Study 1 Experimental Conditions. ..... 37
Table 3. Regression Results for Willingness to Express Opinions. ......................... 41

Table 4. Reliabilities, Means, Standard Deviations, and Correlations of Variables
of Study 2. .................................................................................................. 44

Table 5. Means and Standard Deviations of Study 2 Experimental Conditions. ..... 45

INTRODUCTION

People often seek out others’ experiences as information sources before
making decisions to purchase or consume products and service. Word of mouth
(WOM), which has traditionally referred to interpersonal communication among
consumers in order to exchange choice-related information, opinions, and
evaluations of products and services (Arndt, 1967; Anderson, 1998; Ozcan, 2004),
often serves as an important information source that affects people’s purchasing
decisions and behaviors. WOM also has been examined for its inﬂuence on
individuals’ movie selections, and WOM is not only a phenomenon that occurs in
interpersonal settings, such as face-to-face, but also and more increasingly occurs
on the Internet. One of the many functions of the Internet is providing handier
access to and easier participation in WOM for movies; people can easily see others’
Opinions and/or actively express their Opinions on the Internet. Many Internet sites
provide a place where users can post their ratings and comments about movies and
become a part of WOM. Online WOM is of interest to both scholars and marketing
practitioners regarding understanding why and how people participate in WOM.

Online WOM has several interesting features. First, WOM on the Internet
comes from anonymous others, not from one’s own social networks and
interpersonal relationships. Second, WOM on the Internet often provides

information on the number of others who like or dislike certain movies, and those

numbers tend to be much larger than those available in offline WOM (i.e., the

number of people in one’s social networks). Third, users can see the results of
ongoing polls about certain movies on a daily basis, if they choose to do so. Fourth,
participation in WOM on the Internet is completely voluntary and may reﬂect the
interactive aspects of individuals’ expressing their opinions affecting and being
affected by public opinions (Mutz, 1989). That is, individuals’ willingnesses to
express their Opinions can be affected by the existing public opinions on the
Internet site, and, at the same time, individuals may also want to affect public
Opinion by expressing their Opinions.

The current study focuses on the relationship between individuals’
perceptions of public opinions and their willingnesses to express their own opinions
to be a part of online WOM. User ratings on the Internet have both aspects of
WOM and public opinion polls on the Internet. Therefore, the purpose of this paper
is to explore the mechanism of user rating of movies by focusing on its role as
public Opinion poll. People have a tendency to overestimate media effect on others
and underestimate media effect on them (Davison, 1983). The main reason for
people’s behavior change is not necessarily due to the impact of persuasive media
messages but mostly due to their misperception of media impact on others. Based
on this, individuals may perceive that others, compared to themselves, are more
inﬂuenced by the user rating and that this discrepancy may motivate individuals to
express their opinions. According to Noelle-Neumann (1974), people are less
willing to express their opinions publicly when they perceive the opinion climate to

be against their own viewpoints. People’s perceptions of public opinion climate are

an essential element of individual opinion expression. Based on the Third-Person
Effect (TPE) and Spiral of Silence theory, this paper will advance hypotheses and
research questions concerning the mechanism of opinion expression as a part of

WOM on the Internet.

LITERATURE REVIEW

The effect of User Ratings on Audience ’3 Movie Selection

As motion pictures are experienced goods to be enjoyed by audiences rather
than to fulﬁll economic or basic needs of consumers, a traditional information
seeking framework does not ﬁt movie audience’s movie search behaviors
(Eliashberg & Shugan, 1997). Burzynski and Bayer (1977) demonstrated that
people who are exposed to positive information about movies Show higher
appreciation of movies than those who are exposed to negative prior informational
cues. In addition, information from other ordinary people who have watched the
movie can be perceived as trustworthy as, if not more than, information from mass
media (Austin, 1982).

The traditional form of WOM is from a type of interpersonal source and is
deﬁned as the evaluations and comments about a movie from other movie-goers
(Faber & O’Guinn, 1984). In general, WOM as interpersonal information source is
distinguished from information from experts and can be seen as credible as, if not
more credible than, advertisements and critics’ reviews because it could be a sign
of other peoples’ tastes and it is easily accessible through social networks (Faber &
O’Guinn, 1984; Banerjee, 1992; Murray, 1991; Holbrook, 1999; Liu, 2006).
Considering the rapidly expanding use of the Internet and diverse on-line
communities as popular information sources, scholarly attention has also been paid

to the effect of on—line WOM on people’s movie selections. The Internet allows

people to easily share information with others (Liu, 2006), and various on-line
communities make it possible for a single opinion to reach thousands of others
(Dellarocas et al., 2004). Therefore, the domain of WOM has extended from
interpersonal networks to the Internet, and the volume and valence of messages
posted on various Internet sites could replace off-line interpersonal WOM.
According to Dellarocas et a1. (2004), there is an average of 689 user reviews per
movie on the Internet, and these are as much a signiﬁcant information source for
movie selection as off-line interpersonal WOM.

WOM on the Internet takes various forms, ranging from users’ speciﬁc
comments to simple evaluations with numeric values. Many Internet sites allow
their users to indicate numerical evaluation of movies and also to offer various
lengths of review comments for each movie. Among them, user ratings as one form
of WOM on the Internet have been more widely researched (Dellarocas et al.,
2004; Dunn et al., 2005; Liu, 2006) because numerical user ratings motivate users
to become active contributors of WOM on the lntemet by submitting their
evaluation of movies with little effort (Dellarocas et al., 2004). Moreover, user
ratings are similar to public opinion polls in that they show the volume and valence
of public opinion about movies with numerical values or percentages at a glance,
how many people like/dislike or have positive/negative attitudes toward movies.

Although many scholars examine WOM on the Internet as an extended
form of off-line interpersonal WOM, the source of WOM on the Internet is

different. The information available ﬁ'om WOM on the Internet comes from the

anonymous public, not from someone in a close interpersonal network. The
inﬂuence power of WOM from interpersonal networks originates from closeness
and similarities in the interpersonal relationship (Gode & Mayzlin, 2004). WOM
from interpersonal works is perceived as credible information because it comes
from someone whom individuals know and/or trust and/or from ﬁiends who have
the similar tastes and preferences in movies. On the other hand, WOM on the
Internet is more likely to come from anonymous others who are not necessarily in
an individual’s interpersonal networks. However, WOM on the Internet, especially
numerical user rating, can still be perceived as credible because it may be seen as
the public opinion climate about movies. Therefore, user rating as one form of
WOM on the Internet has both the characteristics of WOM and of public opinion
polls.
User Ratings as Word of Mouth

Previous research on online WOM concerning motion pictures has focused
on the volume and valence of WOM and the predictive effects on box ofﬁce
revenue (Dellarocas et al., 2004; Liu, 2006). Liu (2006) claims that the volume and
valence of user ratings inﬂuence audience’s cognition and behaviors through
different routes. The volume which refers to the total quantity of interaction can
have an informative effect on awareness, while the valence, which is deﬁned as the
quality of WOM concerning positive or negative evaluation of movies, can have a
persuasive effect on audience’s attitudes toward movies (Liu, 2006). The volume of

WOM indicates the frequency with which people are exposed to WOM on the

Internet about a movie and also increased awareness about the movie. This
awareness results in purchasing movie tickets. The volume of user ratings is Often
the most signiﬁcant explanatory factor to predict future box ofﬁce revenue
(Dellarocas etal., 2004; Duan et al., 2005; Liu, 2006).

The persuasive role of valence of WOM is expected in such a way that
positive WOM enhances positive attitude and negative WOM reduces it; however,
this effect was not supported in previous studies in such a way that the valence of
WOM concerning a movie was not signiﬁcantly related to the ticket sales at box
ofﬁce for that movie (Dellarocas et al., 2004; Duan et al., 2005; Liu, 2006).
Dellarocas et a1. (2004) suggested that the valence (i.e., positive or negative value)
of user rating could be a factor which has greater inﬂuence on people’s attitudes
than on behaviors such purchasing movie tickets. Liu (2006) stated that the valence
of WOM may affect people differently according to overall WOM values of
movies. Further, Liu (2006) suggested that people may perceive WOM on the
Internet as a reﬂection of someone else’s taste, pOssibly causing WOM on the
Internet to lose its persuasive power. In sum, although the informative power (i.e.,
volume) of on-line WOM on ticket sales at box ofﬁce is supported, the persuasive
effect (i.e., valence) of on—line WOM on ticket sales at box ofﬁce is still
ambiguous.

These inconsistent results concerning the effects of volume and valence of
user ratings can be due to the misconception of WOM on the Internet as an

extended form of WOM in interpersonal relationships. First of all, the previous

 

studies considered the valence of WOM as an absolute value: a positive or negative
value. In order to demonstrate the predictive power of WOM for box ofﬁce
revenue, the researchers just assumed a simple relationship between the valence of
WOM and people’s movie selection, positive WOM increasing an audience’s
positive attitude and negative WOM reducing it. It should be considered that people
tend to see themselves as better than others (i.e., better than average effect, Alicke
et al. 1995), and accordingly people do not simply follow the majority’s opinion
(Alvaro & Crano, 1996). The effect of valence of WOM may not simply come
from the valence of WOM on the Internet, but may result from a discrepancy
between the valence of WOM and one’s own value about movies. Thus, it could be
helpful in clarifying the effects of valence if the valence of WOM is deﬁned as the
difference between others’ user ratings and an individual’s own ratings.

Moreover, the user rating is a dynamic process including interaction
between individuals and their perception of user ratings. WOM on the Internet
Often starts to form before movies are released. The early user ratings which are
posted on the Internet before the release reﬂect people’s expectations about movies,
and later user ratings (which are posted on the Internet after the release) indicate
people’s actual evaluations of the movies, which may be inﬂuenced by the early
user ratings. Thus, a Simple correlation between the valence of user ratings and
ﬁtture movie revenue may not fully explain the effect of user ratings. In previous
studies (Dellarocas et al., 2004; Duan et al., 2005; Liu, 2006), box Ofﬁce revenue,

the result of people’s choices, which are assumed to be inﬂuenced by WOM on the

Internet, was considered as a dependent variable. People’s immediate responses to
WOM or user ratings of a movie were not considered in those previous studies.
WOM on the Internet, especially numeric ratings of users or movie watchers, is
similar to opinion polls on the Internet in that people’s responses can be disclosed
immediately right in the WOM itself. Hence, by examining the characteristics of
public opinion polls on the Internet, it may be possible to better understand how
and why individuals respond or do not respond to the WOM on the Internet and
individuals are or are not inﬂuenced by WOM on the Internet. The next section will
review the characteristic of WOM on the Internet as an opinion poll.

User Ratings as Public Opinion Polls

As a collection of individual opinions, a public opinion is more than the
mere statistical aggregation of individuals’ opinions (Mutz, 1989; Park & Salmon,
2005) in that the formation process of public opinion involves interactions not only
among individuals but also between individuals and their perceptions of the
Opinions of others (Pan et al., 2005). Public opinion polls reported by the mass
media inﬂuence individuals’ perceptions of assumed Opinion climates (Fields &
Schuman, 1976; Glynn et al., 1995; Mutz, 1998).

WOM on the Internet, especially user ratings, takes on some characteristics
of public opinion polls on the Internet. First, WOM on the Internet comes from
diverse groups of people. Various Internet sites provide places for individuals to
directly access thousands of anonymous others’ opinions and post their own

opinions (Dellarocas et al., 2004). Dellarocas et a1. (2004) claims that WOM has

more persuasive powers when it comes ﬁom different social groups rather than
one’s own group. Especially, a user rating, which consists of numerical evaluation,
is a type of opinion poll about a certain move in that it shows at once how many
people like or dislike the movie.

The second characteristic of user rating on the Internet as online public
Opinion polls is interactivity. Interactivity is identiﬁed as the activity of
communication exchange in interpersonal and social communication (Rafaeli,
1988; Schultz, 1999). The main features of fully interactive communication are
identiﬁed as equal participation, symmetrical communicative power, and
interconnectedness (Schultz, 1999). In addition, interactivity connects not only
communication encounters but also between communication environments,
content, and users (Russell, 2006). The Internet is a public area where public
opinion is shaped through opportunities for each individual to express his or her
own Opinion and to participate in debates and polls (Savigrry, 2002). Interactivity
on the Internet is more than connectedness with each other, and it allows people to
express their opinions equally. While in traditional media the public has become a
passive audience which just reacts or responds to media content (Franklin, 1994),
on the Internet people are positioning themselves as the participants rather than
audience. Online discussion boards, chat rooms, and public opinion polls are
examples of interactive communication among users (Schultz, 2000).

Some scholars, however, have criticized online opinion polls in terms of

their nonscientiﬁc characteristics, such as the possibility of rating result

10

manipulation by multiple votes, sampling problems (e. g., self-selected participants)
(Kent et al., 2006), and biased results by a skewed sample with participants who
may already have interest in poll issues (Wu & Weaver, 1997). Kent et a1. (2006)
argued that the role of online opinion polls is not to measure public opinion
climate, but to allow people to express their opinions. Whether or not online
opinion polls provide the circumstances of public opinion, individuals can use the
results of online polls to investigate the circumstances of public opinion (Wu &
Weaver, 1997).

People might be more motivated to join online discussions or opinion polls
because of these non-scientiﬁc aspects of online opinion polls. The openness to
everyone and self-selected voluntary participation in polls on the Internet make
people react to the results of ongoing opinion polls, which people can see before
they participate. Although Schultz (1999) pointed out multiple voting as
unscientiﬁc aspect of online polls, people may use the chance of multiple voting in
an attempt to change the results of ongoing Opinion polls which are different from
their own opinions. Hence, the current study asks the question; why and when do
people choose to (or choose to not) participate in opinion polls in order to change
the direction of the opinion poll results to support their side? One hypothesized
explanation is that people participate in the polls not only to express their own
opinion but also to inﬂuence the opinion poll results. The decision to participate in
ongoing polls is inﬂuenced by people’s concerns for others who may be more

vulnerable to the effects of poll results than themselves. Especially, when the poll

11

results disagree with their opinions and when they perceive that others may be
more inﬂuenced by these poll results, people may be more willing to participate in
the polls. This phenomenon could be explained by the Third-person effect (TPE).
Public Opinion Perception and Third-Person Eﬂect

The TPE is based on peoples’ misperceptions of the effect of media
messages on themselves and others. Davison (1983) suggested the TPE hypothesis,
which says that people who are exposed to persuasive media messages will expect
that others will be greatly inﬂuenced by the messages when, in fact, they may not
be. According to the TPE, the media will not have a great inﬂuence on “me” or
“you,” but it will on “them” — the third person. According to Davison (1983), the
third person refers to an unspeciﬁed person, someone in the public who is
considered by ﬁrst persons when they perceive media messages. Perloff(1999)
indicated that the TPE is composed of perceptual and behavior components. The
perceptual component refers to the discrepancy between self and others, which can
lead to the misperception of media effects on others. The behavior component as a
key element of the TPE is the ﬁrst-persons’ reactions to their incorrectly-
anticipated media impact on the third-person. This overestimated media effect on
others by the ﬁrst person leads to the ﬁrst person’s attitude and behavior changes
about media messages. To sum up, although a persuasive media message does not
directly inﬂuence people’s attitudes or behaviors, it may be seen to indirectly
change people’s behavior by considering its inﬂuence on others.

Research has shown that the TPE is stronger for controversial news reports

12

(Salwen & Driscoll, 1997); undesirable messages, such as pomegraphy (Gunther,
1995); and persuasive media messages, such as commercials (Shah, Faber, &
Youn, 1999). News reports of Opinion polls also have a persuasive effect on
audiences because they present the public opinion climate (Mutz, 1989). Scholars
of public opinion assert that perceptual processes in public opinion should be
stressed because public opinion polls in the mass media systematically bias
people’s public opinion perception in favor of majority (Pan et al., 2005). PeOple
need to be sensitive to potential misperceptions of opinion climate because
sometimes individuals’ opinions reﬂect their personal values and reveal who they
are. Mutz (1989) claimed that there is a greater perceptual gap between self and
others in perceiving public opinion polls and the perception of public opinion
climate inﬂuences people’s willingnesses to express their own opinions.

As scholars have pointed out that there has been relatively little work done
to explore how exposure to the results of opinion polls inﬂuences the poll itself,
they have claimed that people’s perceptions of the effect of public opinion polls
also take part in the formation of public opinion itself (Lavrakas et al., 1991;
Morwitz & Pluzinski, 1996; Pan et al., 2005). In traditional opinion polls, people
express their opinions without consideration of other’s opinions, while in online
opinion polls, people can see the results of the ongoing polls before they participate
in it. The mere exposure to the poll results has an impact on people’s attitude
changes (Morwitz & Pluzinski, 1996). Morwitz and Pluzinski’s (1996) study

showed that people exposed to the poll results changed their attitudes toward

13

candidates in the direction of the poll results. Public opinion polls in news coverage
lead to a biased judgment of the assumed inﬂuence of opinion polls on others
(Gunther & Storey, 2003), and also it supports the restriction of opinion poll reports
(Milavsky et al., 1985); however, the effect of mere exposure to the poll results on
the poll itself is still undeveloped as of yet. Thus, the current study examines how
user ratings, as one type of public opinion poll, will inﬂuence people’s perceptions
of the opinion climate and how people’s perceptions of the results of ongoing
opinion polls also inﬂuence the opinion polls themselves. It is expected that people
will tend to minimize the effect of user ratings on themselves and to maximize the
effect of user ratings on others. The following hypotheses are thus advanced.

H1: User ratings on the Internet will be perceived by people to have a

greater effect on others than on themselves.

The volume of user ratings can affect the perceptual gap between self and
others. A great volume of user ratings, which may be perceived as the aggregation
of public Opinion at large, may have strong persuasive power. Thus, it is expected
that the higher the volume of users who have posted their ratings (i.e., the greater
the total number of user ratings), the greater the perceived effect of user ratings on
others will be than the perceived effect of user ratings on self.

H2: As the volume of user ratings increases, the self-other gap in perceived

effect of user ratings will increase.

The current study conceptualizes the valence of user ratings as the

differences between an average user rating and an individual’s own ratings. If the

14

results of the user ratings are in opposition to people’s own evaluations (a big
difference between the user ratings and one’s rating), it could be perceived as
undesirable message to force one’s opinion change in the manner of poll results and
people may be more likely to underestimate the effect of user ratings on themselves
and to overestimate the effect on others. Thus, the gap between the valences of user
ratings and people’s opinions will be negatively correlated with the self-other gap
in the perceived effect of user ratings.

H3: As the gap between the valences of user ratings and people’s own

ratings increases, the self-other gap in the perceived effect of user rating

will increase.
Willingness to Express Opinions, Spiral of Silence Theory, and T PE

One important aspect of online polling is that people voluntarily participate
in the polls. Traditional public opinion polls have used survey methods such as
telephone surveys, interviews, and visiting houses. Although the decision to answer
the survey questions fully depends on people’s own intentions, the chance to
participate in the traditional opinion poll is determined by a sampling method
chosen by the surveyors or polling agencies. For opinion polls on the Internet,
however, the decision to participate in the polls is up to individuals’ volition to
express their opinion publicly. Thus, motivations and behavioral patterns reﬂected
in participation in the Internet opinion polls may be different from those in
participation in the traditional Opinion polls.

Concern for others can be a reason for individuals’ willingness to engage in

15

WOM on the Internet. Dichter (1966) considers altruism and desire to warn others
as one reason why individuals become participants in WOM. Anderson (1998)
demonstrated that the relationship between consumer satisfaction and the rate of
participation in WOM follows a U-shape. Those who were very satisﬁed or
unsatisﬁed with the product are more likely to engage in either positive or negative
WOM than those who have middle levels of satisfaction. Especially, extremely
dissatisﬁed customers engage in greater WOM than highly satisﬁed customers
(Richins, 1983; Anderson, 1998). These results suggest that both highly satisﬁed
and highly dissatisﬁed people may believe the expression of their opinions could
inﬂuence other’s choices, wanting to persuade others to purchase or not purchase
the products.

Spiral of silence theory. Noelle-Neumann’s spiral of silence theory could be
useful to explain people’s willingness to express their own opinions in relation to
the perception of the opinions of others. Underlying this, Noelle-Neumann (1974)
claimed that the public opinion formation process arises not only ﬁ'om interaction
among individuals but also from their perceptions of the opinion climate.
According to Noelle-Neumann, some people have a fear of isolation ﬁ'om others.
When it comes to controversial issues, individuals observe the overall distribution
of others’ opinions through ‘quasi-statistical pictures’ drawn from various social
circumstances and assess which opinion is dominant and which others are
degraded. If they perceive their views to be among the minority, they may be less

conﬁdent about their opinions and decide to be silent, rather than speaking out with

16

their opinions publicly. As people are more likely to perceive their opinions are in
the minority, they are more likely to keep their opinions to themselves. As this
process continues, the seemingly dominant opinion becomes more widespread and
other opinions become less popular. Noelle-Neumann (1974) argued that
willingness to express one’s opinion differs from willingness to change one’s
opinion. She argued that there is a positive correlation between people’s assessment
of present and future opinion distributions in such a way that the overall
distributions of opinions are consistent overtime. People do not necessarily change
their Opinions to ad0pt the seemingly dominant opinion, but they become
increasingly silent about their opinions, indirectly helping the dominant opinion to
appear more dominant.

Among the many factors that inﬂuence willingness to express one’s
opinions, individuals’ knowledge about the topic (Salmon & Neuwirth, 1990;
Shamir 1997), interest in the issues (Baldassare & Katz, 1996; Lasorsa I991;
Noelle-Neumann, 1974; Willnat et al., 2002), conﬁdence in their correctness
(Lasorsa, 1991), and people’s perceptions of the public opinion climate (Mutz,
1989) affect willingness to express one’s opinion. The mass media is an important
system for providing information about the opinion climate (N oelle-Neumann,
1974). Rather than reﬂecting the public opinion, however, the mass media is seen I
as shaping the public opinion; the mass media play an important role in creating
public opinion by providing social pressure for people to respond to it. Thus,

opinion polls on the media have been seen to inﬂuence individuals’ willingness to

17

express themselves.

However, some scholars have criticized the assumption of spiral of silence
theory: the fear of isolation. As Noelle-Neumann assumed that people have some
degree of fear of isolation from society at large and some people have more fear of
isolation than others, the fear of isolation varies with individual differences (Hayes
et al., 2005) and situations (Hayes et al., 2001). Glynn and McLeod (1985) pointed
out that fear of isolation could be a factor affecting one’s willingness to express
themselves rather than a basic assumption. At this point, the current study questions
do people have the same fear of isolation from others in the virtual world as off-
line? Previous studies have asked the participants to imagine a conversational
situation with friends or strangers and then measured participant’s willingness to
express themselves during the imagined conversational situation in experimental
settings. In this case, an imagined conversational situation is restricted to face to
face conversation; thus the fear of isolation depends on hypothetical situations and
conversation partners. Therefore, the fear of isolation actually stems from people’s
imagined relationships, not ﬁ'om society at large. On the other hand, the main
feature of the Internet is the anonymity which hides identity. People may have less
fear of isolation ﬁom society, even though they realize that their opinions are not
supported by the majority. On the Internet, therefore, the fear of isolation from
others could explain why people avoid expressing their opinions when they are in
the minority but could not explain why people have more willingness to express

their Opinions in general. If people have less fear Of isolation from others in the

18

virtual world than offline, why would people have more or less willingness to
express their opinion? In relation to the main feature of online polls, the
interactivity between users and their perception of online poll results, people’s
concern for others who may be inﬂuenced the results of polls could be a motivator
for expressing one’s opinions publicly on Internet polls.

The third-person eﬂect and willingness to express opinions. The behavior
component of the TPE argues that people’s motivations for their attitude or
behavior changes are not due to the persuasive power of media messages, but to
their misperceptions of the effect of mass media on others (Davison, 1983). Mutz
(1989) examined the role of people’s public opinion perceptions on their
willingnesses to express their Opinions, and the results demonstrated that people
who perceived the Opinion distribution to be in the favor of their position tended to
be more willing to speak out about their opinions. Therefore, the current study
suggests that people’s concern for others is related to the effect of online polls as an
important motivator that inﬂuences their willingnesses to express their own
opinions on online polls rather than to a fear of isolation. In the cyber-world,
although people do not need to feel fear of isolation ﬁ'om anonymous others
because their identity also could be hidden, so they may voluntarily participate in
on-line polls due to their concern for the perceived effect of on-line poll results on
others. The more people believe that the results of ongoing on-line polls have a
great impact on others, the more willing they will be to participate in the polls. The

following hypotheses are offered.

19

H4: As the gap between the perceived effect of user ratings on self and on
others increases, people’s willingnesses to express their opinion will
increase.

A great volume of user ratings will reduce willingness to express one’s
opinion because adding one vote may have little effect on changing the results of
ongoing polls. Thus, the volume of user rating increases and people’s willingnesses
to express their opinions will be negatively correlated.

H5: As the volume of user ratings increases, people’s willingnesses to

express their opinions will be reduced.

The valence difference between the user ratings and one’s own rating will
have a positive effect on willingness to express one’s Opinion. Speciﬁcally, when
user ratings and people’s opinions are signiﬁcantly different, people have a strong
willingness to express their opinions in order to minimize the effect of user ratings
on others.

H6: As the difference between the valence of user ratings and one’s own

rating increases, people’s willingness to express their opinions will increase.

20

STUDY 1
Method

Participants

Undergraduate students in Communication and Advertising classes at
Michigan State University were asked to participate in the survey in exchange for
extra research credit in each class. In total, 403 students voluntarily participated in
the survey. The nationality of 96.5% (n = 389) students were U.S, while 3.5% (n =
14) students were international students. Among the international students, the nine
Korean international students who had the greatest possibility of having been
inﬂuenced by information about the stimulus (Korean movies), from Korean
websites or other sources, were excluded ﬁ'om data. Therefore, the responses of
394 students were used in the ﬁnal analysis.
Design

The current study used a 2 (volume: high and low) X 2 (valence: positive
and negative) x 2 (before and after exposure to user rating) mixed design with
volume and valence as between-subject factors. The volume of user rating was
manipulated as the number of hypothetical students who were said to have
participated in the rating. In the high volume conditions, the ratings Of 350 students
were provided, while, in the low volume conditions, the ratings of 15 students were
provided to participants. The valence of user rating was deﬁned in terms of positive
and negative ratings; four points (i.e., of ﬁve stars, four were yellow and one was

white) were provided as a positive rating and one point (i.e., Of ﬁve stars, one was

21

yellow and four were white) was presented as a negative rating. Research
participants were randomly assigned to one of the four conditions. There were 25.1
% participants (11 = 99) in condition 1 (high volume and positive valence), 26.6%
participants (n = 105) in condition 2 (high volume and negative valence), 22.3%
participants (n = 88) in condition 3 (low volume and positive valence), and 25.9%
participants (11 = 102) in condition 4 (low volume and negative valence).
Participants were asked to indicate their own rating and their estimated ratings of
other students about movies before and after exposure to user ratings.
Stimulus

In order to control the effect of information from other sources, it was
decided to use movies unfamiliar to American participants. The current study used
hypothetical students’ ratings of three Korean movies, which were released in
Korea between January and August, 2007 and ranked high in the Box Ofﬁce but
were not released in the US. In addition, to avoid the confounding effect of genre
preference, the three movies were selected ﬁom different genres; drama, comedy,
and horror. The three movies were Secret Sunshine (drama), 200 Pounds Beauty
(comedy), and Black House (horror). Of the participants, 30.5% of them (n = 120)
were randomly assigned to conditions using Secret Sunshine, 37.3% (n = 147) were
randomly assigned to conditions using to 200 Pounds Beauty, and 32.3% (n = 127)
were randomly assigned to conditions using to Black House. The English version of
the posters and a simple synopsis of each movie were presented as information

about the movies for participants.

22

The survey was conducted on the Internet in order to provide similar
circumstances to the real situation of online ratings for participants. With the
permission of the instructors of each class, one hyperlink of the questionnaire out
Of the hyperlinks for the four conditions was randomly provided to participants via
e-mail. When participants clicked the hyperlink of the questionnaire on the Internet,
they saw the website for the questionnaire, including the poster for and a simple
synopsis of one movie out of three movies. After participants had read the
information about the movies, they were asked to answer to items assessing the
third-person effect of user ratings on themselves and other students and their
willingnesses to express their own Opinions.

Measures

Measures used in the current study are listed in Appendix. Table 1 shows
Cronbach’s a reliabilities, correlations, means, and standard deviations of the main
variables in the study.

The third-person eﬂect of user ratings. As the dependent variable of
hypotheses 1, 2, and 3 and the independent variable of hypotheses 4, the third-
person effect is deﬁned as the difference in perceived effects of user ratings on self
and on other students. The gap of the perceived effect of user ratings on self and on
others was measured in two ways: 1) measuring participants’ perceptions of the
effects that user ratings have on self and on others by using parallel questions
(hereafter, it is called TPE l)’rand 2) measuring the extent to which participants’

self rating of a movie changed from before to after exposure to user ratings and the

23

extent to which they effected participants’ estimated ratings of other students
before and after exposure to the user rating (hereafter, it is called TPE 2).

Three parallel items were used to measure the perceived effect of user
ratings on self and others (TPE 1). Participants were asked to answer the following
questions: “how much do you think this rating has affected your / other students’
ratings of the movie,” “how much do you think this rating has affected your / other
students’ attitudes toward the movie,” and “how much do you think this rating has
affected your / other students’ expectations Of the movie.” The three questions were
identical except for the ﬁrst person (you) and third-person (other students)
designations. The reliability Of the three items which asked about the perceived
effect of user ratings on self was a = .91. The reliability of the three items that
asked about the perceived effect of user rating on others was a. = .94. Three items
were averaged to produce a single perceived user rating effect score on self (M =
2.98, SD = 1.35) and a single perceived user rating effect score on the other (M =
3.67, SD = 1.09).

Participants were asked to indicate their ratings of a movie and their
estimated ratings of other students of a movie before and after being exposed to the
manipulated user ratings to measure the observed effect of user ratings on self and
others (TPE 2). Each person’s rating of the movie was measured with eight items
(interesting, funny, boring [receded], appealing, impressive, exciting, like a waste
of time [receded], and ordinary [recoded]) on a six—point scale from “O-not at all”

to “5-very much” and one additional item (“If you were to rate this movie, your

24

rating of this movie would be _ out Of 5”). After analyzing the reliability of the
nine items, two items (funny and ordinary) were removed because removing these
two items improved the reliabilities from a = .87 to a = .92 for self ratings before
exposure; from or = .81 to a = .87 for the other rating before exposure; from a = .87
to a = .92 for the self rating after exposure; and from a = .90 to a = .93 for the other
rating after exposure.

Willingness to express one ’s opinion. Before measuring participants’
willingness to express their Opinions, a simple script was presented to participants.
The script said that “The Department of Communication at Michigan State
University is collecting students’ opinions about foreign movies in order to host an
International Film Festival next year. The user rating about a movie was produced
by ratings of other MSU students in the last semester and participants’ rating in this
survey also will be included in this rating for other MSU students of next
semester.” The purpose of this script was to make participants believe that the
manipulated user ratings were really other students’ ratings and that their
participation in this rating had a real inﬂuence on other students’ ratings.

Participants’ levels of willingness to express their Opinions were measured
on a six-point scale from “O-not at all” to “5-very much” in terms of their
willingness to participate in user ratings and to review movies in various
hypothetical situations: willingness to have their ratings reﬂected in later versions
of this survey, willingness to vote multiple times in a user rating, willingness to

participate in a user rating and to make user comments on other movie sites, and

25

willingness to express their Opinions in a hypothetical class discussion. The
reliability Of willingness, measured with six items, was or = .71; however, item
number 1, “willingness to have their ratings reﬂected in a later version Of this
survey,” and item number 2, “willingness to vote multiple times in a user rating,”
lowered the reliability (i.e., low item — total correlation). After deleting these two
items, reliability of willingness to express Opinions increased from a = .71 to a =
.74. The scores on the four items were averaged to create willingness to express
Opinions (M = 2.67, SD = 1.04).

Fear of isolation and self-censorship. In addition, the fear of isolation scale
(N euwirth, 2000) and the self-censor scale (Hayes et al., 2005) were presented to
participants in the last part of the questionnaire. The reliability of the fear of
isolation scale was a = .32. Among the six items, only two items, which showed
relatively high correlations with one another, were used to produce an average
score of fear of isolation. The reliability of fear of isolation with these two items
was 0. = .76. The self-censor scale consisted of eight items, and its reliability was a
= .80. After removing one item which showed low item — total correlation, the
reliability was a = .81. The scores on the seven items were averaged to produce an
average score of self-censorship.

Demographic information and movie selection behavior. Participants were
asked to indicate their age, gender, ethnicity, academic standing, and national
status. The range of participants’ ages was from 18 to 27. The proportion of gender

among the participants was 35.3% male (n = 139) and 64.7% female (n = 255).

26

Regarding the ethnicity of participants, 84.5% (n = 333) of the students were
Caucasian, 7.6% (n = 30) of the students were African American, and 3.0% (n =
12) Of the students were Asian American. In college standing, 17.8% (n = 70) were
freshmen, 8.6% (n = 34) were sophomores, 37.1% (n = 146) were juniors, and
30.5% (n = 120) were seniors.

In addition, participants were asked to answer questions about their movie
selection behaviors. In preference for watching movies, 52.3% (n = 206) of the
participants preferred to watch movies on DVD, 42.1% (n = 166) of the participants
preferred to watch movies in theaters, and 4.6% (n = 18) of the participants
preferred to watch movies on TV. The ﬁequency for watching movies in theaters
was 1.69 times per month, while the frequency of watching movies on DVD was
6.8 times per month. For genre preference, comedy (41.9%, n = 165) and romantic
comedy (27.4 %, n = 108) were the most highly preferred by participants, and the
rest of the order of genre preference was as follows: drama (11.9%, n = 47), action
(9.9%, n = 39), horror (6.1%, n = 24), fantasy (0.5%, n = 2), animation (0.3%, n =
1), and others (2.0%, n = 8). When asked with whom they watch movies, 74.6% (n
= 294) of the participants indicated friends, 18.5% (n = 73) indicated with dates,
3.3% (n = 13) indicated that they watched movies alone, and 2.5% (n = 10)
indicated family. When asked to indicate the trustworthiness Of each movie
selection source on a 7-point scale (1 = not trustworthy at all, 7 = most
trustworthy), the order of trustworthy sources was: information ﬁ'om ﬁ'iends (M =

6.05, SD = 1.07), television ads (M = 5.32 , SD = 1.42), movie Internet sites (M =

27

4.41, SD = 1.63), critics’ reviews (M = 4.23, SD = 1.66), user reviews on the
Internet (M = 3.89, SD = 1.68), magazine ads (M = 3.73, SD = 1.55), and
newspaper ads (M = 3.53, SD = 1.55). Interests in foreign movies and Asian movies
were measured on a 7-point scale from “1 - not at all” to “7 -very much.” The
participants’ interest in foreign movies was M = 2.80, SD = 1.56, and their interest
in Asian movies was M = 2.22, SD = 1.29. Only 29.2% (n = 115) of the participants
had had the experience of watching Asian movies, and 70.6% (n = 278) of the

participants had never watched an Asian movie.

28

:25me 2.: :0 voted»: 8m Co 9:23:88 mama—SEEM
gm 3 pm ES evacuate .5. var. .2? a...

 

 

 

ac: an: 65 Go. a 5.8 so: a.
$6 me Ed 26 and mad 3 ~88
3. * *5 ﬂ. vo. mo. o _ .- mo. comcouﬁom
:o. m 8 0 So
...:. :8. mo. 3. S. S. .2 z m
22:30 3295
..:.&N ‘23:. .L _. **Nm. 1.: . 3mm. 3 mmocmazzg
.050 so
.3. 1N0. mo. .ILN 3:. SEEN. Hoohm EZOOLOAH
. how so
. a. 3.- 1.2. *2. 1mm. 80am Bioeom
8:8me note
a. :3. 1mm. :8. mean .aoso
Eamoaxm coca
Na. :3. tax. SEER
Eamoaxm 2&3
a. :3. mass .223
2:89am Bowen
8. were tom
£050 0 enamomxm 8:89am enamemxm enamomxm
co .2 m Loam Lots 283 283
:o Locum
.6 and wears seam a mesa were wane wane
82088
262$ .aoao cam .8an arm

 

A beam no 33323 he 28:38.80 was .2333ch wﬁncﬁm £532 .mogﬁmzom .ﬁ 2an

29

Results

Preliminary Analysis

Before conducting main analyses, the effects of movie on the perceived effect of
user ratings and on the observed self-other rating changes were analyzed. An ANOVA
did not reveal a signiﬁcant effect of movie type on differences in perceived effects of
user ratings on self and others, F (2, 389) = 0.03, p = 0.97, 112 = .00, or on differences in
Observed self rating change and other rating change, F (2, 389) = 0.50, p = .61, n2 = .00.
For differences in the perceived effects of user ratings on self and others, Secret Sunshine
(M = 0.94, SD = 0.94), 200 Pounds Beauty (M = 0.97, SD = 0.97), and Black House (M =
0.96, SD = 0.89) were not different from one another. For differences in Observed self
rating change and other rating change, Secret Sunshine (M = 0.66, SD = 0.75), 200
Pounds Beauty (M = 0.59, SD = 0.62), and Black House (M = 0.58, SD = 0.66) were not
different from each other.
Main Analysis Overview

In the current study, the effects of user ratings on self and others were
conceptualized in two ways: 1) the perceived effects of user ratings on self and others
after exposure to the user ratings and 2) participants’ own rating change and estimated
others’ rating change before and after exposure to the user ratings. The rating change was
calculated by taking the absolute values of difference between self ratings before and
after exposure to user ratings and difference between the participants’ estimated other
ratings before and after exposure to user ratings. Thus, two scores of the third person
effect were produced as repeated measures: TPE 1) the perceived effect of user ratings on

self and the perceived effect of user ratings on others and TPE 2) the absolute difference

30

in self ratings before and after exposure to user ratings and the absolute difference in
other ratings (i.e., participants’ estimated ratings of others) before and after exposure to
user ratings. TPE 1 and TPE 2 were used in analyses testing hypotheses 1, 2, and 3,
which predicted the effects of the target (self versus others), volume, and valence,
respectively. Hypotheses 4, 5, and 6 predicted the effects of the target (self versus others),
volume, and valence on willingness to express opinions. Multiple regression analysis
testing hypotheses 4, 5, and 6, however, used only TPE 1 after calculating the difference
between the perceived effect of user ratings on self and the perceived effect of user
ratings on others (hereafter, it is called D-TPE 1), for the purpose of the clarity and
simplicity of the analysis.

In addition, the effect of user rating valence was tested in two ways: 1) Valence 1:
the valence of the manipulated user rating as a positive valence (i.e., four yellow stars out
of ﬁve stars) and negative valence (i.e., one yellow star out of ﬁve stars) and 2) Valence
2: the absolute difference between user ratings (one out of ﬁve stars in the negative
valence condition and four out of ﬁve stars in the positive valence condition) and
participants' own ratings before exposure to user ratings (e. g., when a participant’s self-
rating score before exposure to the user ratings was three, his or her valence 2 score was
two if he or she was in the negative valence). Hypotheses 3 and 6 were concerned with
the effect of the gap between the valence of user ratings and people’s own ratings on
perceived effect of user ratings (H3) and on willingness to express opinions (H6).
Although valence l was not the gap between valence of user ratings and people’s own
ratings but rather the manipulation of positive and negative valence, Valence 1 was

included in the mixed ANOVA analyses to test hypotheses 1 and 2. To directly test

31

hypothesis 3, valence 2 was included in a multiple regression analysis with the perceived
effect of user ratings on self and others as the dependent variable. Valence 2 was also
used as a predictor in multiple regression analyses testing for hypothesis 6, with
willingness to express opinions as the dependent variable.

The Third Person Eﬂect of User Ratings on Self and Others

It was predicted that user ratings would have greater effect on others than on self
(H1), especially if user ratings indicated a high volume rather than a low volume (H2),
and if the more likely user ratings would differ from participants’ own ratings (H3). TO
test hypotheses 1 and 2, mixed ANOVAS were conducted on TPE I and TPE 2 separately
as dependent variables. Table 2 reports means and standard deviations of the TPE 1,
TPE2, and willingness to express opinions in each experimental condition.

Overall Third Person Eﬂect (HI). Hypothesis I predicted that the effect of user
ratings would be larger on others than on self. In analysis 1 (TPE 1), a 2 (target: self
versus other) X 2 (volume: high versus low) X 2 (valence 1: positive versus negative)
mixed ANOVA was conducted to compare the perceived effect of user ratings on the
target (self and other). The analysis revealed a signiﬁcant main effect for the target, F (1,
388) = 156.19, p < .001, n2 = .07. Consistent with H1, the perceived effect of user ratings
on others (M = 3.68, SD = 1.25) was signiﬁcantly higher than on self (M = 2.97, SD =
1.36).

In analysis 2 (TPE 2), a 2 (target: self versus other) X 2 (volume: high versus low)
X 2 (valence 1: positive versus negative) mixed ANOVA was conducted to compare self
rating change and other rating change before and after exposure to user ratings. The

analysis revealed a signiﬁcant main effect for the target, F (l , 388) = 98.96, p = .00, n2 =

32

.09. Consistent with H1, other rating change before and after exposure to user ratings (M
= 0.84, SD = 0.85) was signiﬁcantly larger than the self rating change (M = 0.42, SD =
0.45).

Volume (H2) and valence 1. H2 predicted that the differential effects of user
ratings on self and others would be greater in the high volume than the low volume
condition. In analysis 1 (TPE 1), the analysis did not reveal a signiﬁcant main effect for
the volume of user ratings, F (1, 388) = 0.78, p = .38, n2 = .00, and did not reveal a
signiﬁcant main effect for the valence of user ratings, F (1, 388) = 2.20, p = .14, n2 =
.004. The two way interaction between volume and target was not signiﬁcant, F (1, 388)
= 2.13, p = .15, 112 = .001. The two way interaction between valence 1 and target was not
signiﬁcant, F (1, 388) = 3.08, p = .08, n2 = .001. The three way interaction among
volume, valence 1, and target was not signiﬁcant, F (1, 388) = 0.02, p = .89, n2 = .00.
Thus, inconsistent with H2, peoples’ perceived effects of user ratings on self versus
others were not affected by volume. Additionally, peoples’ perceived effects of user
ratings on self and others were not affected by the positive and negative valence of user
ratings.

In analysis 2 (TPE 2), the analysis revealed a signiﬁcant main effect for the
volume of user ratings, F (1, 388) = 3.99, p < .05, n2 = .006. The overall change (the
average of self change and other change) was larger in the low volume condition (M =
0.68, SD = 0.55) than in the high volume condition (M = 0.57, SD = 0.51). In addition,
the two-way interaction of the target (self versus other) and volume was signiﬁcant, F (1,
388) = 8.78, p = .003, n2 = .008. Self rating changes in the high volume condition (M =

0.43, SD = 0.41) and in the low volume condition (M = 0.41, SD = 0.50) did not differ, t

33

(390) = 0.32, p = .75, but the other rating change in the low volume condition (M = 0.96,
SD = 0.85) was greater than in the high volume condition (M = 0.72, SD = 0.83), t (390)
= 2.83, p = .005. Inconsistent with H2, the low volume condition produced more change
in participants’ estimated ratings of others before and after exposure to user ratings, while
participants’ self rating remained same before and after exposure to user ratings.

The analysis revealed a signiﬁcant main effect for the valence of user rating, F (1,
388) = 6.06, p = .01, n2 = .008. The overall change (the average of self change and other
change) was larger in the negative valence condition (M = 0.69, SD = 0.55) than in the
positive valence condition (M = 0.56, SD = 0.51). The two-way interaction of the target
(self versus other) and valence was not signiﬁcant, F (1, 388) = 0.03, p = .86, n2 = .00.
The three way of interaction among the target, valence, and volume was not signiﬁcant, F
(1, 388) = 0.20, p = .65, n2 = .00.

Valence drﬁerence (H3). With regard to the effect of valence difference (i.e.,
difference between user ratings and self rating scores), H3 predicted that the difference
between the user ratings and people’s self ratings would positively relate to differences in
the effects of user ratings on self and others. To test H3, two regression analyses were
used: one with D-TPE 1, the difference score in TPE 1 (i.e., the difference between
perceived effects of user ratings on self and perceived effects of user ratings on others),
as the dependent variable and another with D-TPE 2, the difference score in TPE 2 (i.e.,
the absolute difference in self ratings before and after exposure to user ratings and the
absolute difference in other ratings before and after exposure to user ratings), as the
dependent variable. Before conducting the analyses to test the hypotheses, the two

volume conditions were coded with low volume as 0 and high volume as 1 to create a

34

dummy variable. The continuous predictor variable, valence 2, was mean-centered. A
hierarchical multiple regression analysis was conducted with the continuous variables and
the dummy variable in the ﬁrst block and the product term of the predictor variables in
the second block.

In analysis 1 (D-TPE 1), the overall model was signiﬁcant, F (3,380) = 3.23, p =
.02, aasz = .016. The two predictors in the ﬁrst block, however, did not contribute
signiﬁcantly to the self—other gap of perceived'user rating effect, F (2,381) = 1.79, p =
.17, adj.R2 = .004. H2 and H3 predicted that the volume of user ratings (H2) and the
valence difference between user ratings and self rating (H3) would predict the self-other
gap of perceived user rating effect (i.e., difference between perceived effect of user
ratings on self and perceived effect of user ratings on others). Inconsistent with H2 and
H3, the results did not show signiﬁcant effects ofvolume (unstandardized coefﬁcient B =
-0.09, sE = 0.10, B = -.05, t = -0.94, p = .35) and valence 2 (unstandardized coefﬁcient B
= 0.07, SE = 0.04, B = .08, t = 1.59, p = .11). On the other hand, the product term
predictor in the second block accounted for a signiﬁcant amount of variance in the
difference in perceived effects of user ratings on self and others, F change (1 , 380) = 6.15, p
= .01, aaliﬂzchange = .017. The interaction term for volume and valence 2 was signiﬁcant,
unstandardized coefﬁcient B = 0.21, SE = 0.09, [3 = .17, t= 2.48, p = .01. To examine the
pattern of interaction, two separate regression analyses were conducted for the high
volume and low volume conditions. The analyses showed that valence 2 was a signiﬁcant
and positive predictor in the high volume condition, B = 0.18, SE = 0.07, B = .19, t =
2.76, p = .006, whereas valence 2 was not in the low volume condition, B = -0.03, SE =

0.06, B = -.04, t = -0.57, p = .57. Only in the high volume condition was the extent to

35

which user ratings differed from self ratings before exposure to user ratings positively
associated with the extent to which perceived effects of user ratings was greater on others
than on self.

In analysis 2 (D-TPE 2), the overall model was signiﬁcant, F (3, 380) = 9.06, p <
.001, adj.R2 = .06. The predictors in the ﬁrst block contributed signiﬁcantly to the
difference between self rating change and other rating change, F (2, 381) = 13.33, p <
.001, adj.R2 = .06. The results revealed that valence 2 (B = 0.14, SE = 0.03, B = .23, t =
4.62, p < .001) and volume (B = -0.14, SE = 0.07, [3 = -.10, t = -2.05, p = .04) were
signiﬁcant. Thus, the data were consistent with H2 and H3, in that other rating change
was larger than self rating change in the high volume condition than in the low volume
condition. As the difference people would perceive between the user ratings and their
own ratings increased, the more likely it was that other rating change would be larger
than self rating change before and after exposure to user ratings. On the other hand, the
interaction between volume and valence 2 volume was not signiﬁcant, B = 0.05, SE =
0.06, B = .05, t = 0.75, p = .45, and did not contribute signiﬁcantly to self-other change

before and after exposure, Fchange (1, 280) = 0.56, p = .45, chhange = .001.

36

Table 2. Means and Standard Deviations of Study 1 Experimental Conditions.

 

Perceived effect on self

Perceived effect on others

total

Self rating change before
and after exposure

Other rating change
before and after
exposure

total

Willingness to express

opinions

 

 

 

 

 

Low Volume High Volume
Negative Positive Negative Positive
Valence Valence Valence Valence

3.19 (1.44) 2.97 (0.28) 3.00 (1.40) 2.69 (1.19)
3.71 (1.29) 3.67 (0.17) 3.71 (1.37) 3.59 (1.13)
3.45 (1.19) 3.32 (0.12) 3.36 (1.28) 3.14 (1.05)
0.47 (0.41) 0.37 (0.42) 0.49 (0.58) 0.32 (0.36)
0.75 (0.80) 0.71 (0.87) 1.06 (0.89) 0.84 (0.75)
0.61 (0.48) 0.54 (0.56) 0.78 (0.60) 0.58 (0.44)
2.75 (1.02) 2.67 (0.98) 2.59 (1.04) 2.50 (0.88)
n=104 n=95 n=101 n=84

 

Note. Standard deviations are reported in the parentheses.

37

Third Person Eﬂect on Willingness to Express One ’3 Opinion

It was predicted that the perceived effects of user ratings on self and others (H4),
volume (H5), and the difference between the user ratings and individuals’ self ratings
(H6) would be predictors for individuals’ levels of willingness to express their opinions.
Five variables (volume, valence 1, valence 2, perceived effect of user ratings on self, and
perceived effect of user ratings on others) were used to predict willingness to express
opinions. Before conducting the analysis, the two volume conditions were coded with
high volume as land low volume as 0 to create a dummy variable. Continuous predictor
variables were mean-centered. A hierarchical multiple regression analysis was conducted
with the continuous variables and the dummy variable in the ﬁrst block, the second-order
product terms of the predictor variables (e. g., volume X valence 2, perceived effect on
self >< perceived effect on others) in the second block, and the third-order product terms in
the third block. No other higher-order or non-linear relationships were observed. An
examination of residuals and various indexes showed no violation Of statistical
assumptions.

The overall model was signiﬁcant, F (25, 360) = 2.88, p < .001, adj.R2 = .11. As
shown in Table 3, the predictors in the ﬁrst block and the second block contributed
signiﬁcantly to variances in willingness to express opinions, but none of the predictors in
the third block did. Among the ﬁrst-order predictors, only perceived effect on others was
signiﬁcantly positive, indicating that as individuals perceived greater effects of user
ratings on others, they were more willing to express their opinions. Among the second-
order predictors, the interaction term of valence 1 and valence 2 and the interaction term

of perceived effect on self and perceived effect on others were signiﬁcant. No other

38

interaction terms were statistically signiﬁcant.

To probe the pattern of the interaction between valence 1 and valence 2, a
multiple regression analysis was done separately for the positive and negative valence
conditions. The results showed that valence 2 was a positive predictor of willingness to
express opinions in the negative valence condition, with an unstandardized coefﬁcient, B
= 0.21, SE = 0.07, B = .22, t = 3.07, p = .002. That is, when the user ratings indicated one
point out of ﬁve possible points (i.e., only one star out of ﬁve stars were yellow), the
greater the difference between individuals’ self rating and the user ratings, the more
willing individuals were to express their opinions. In the positive valence condition,
however, the results showed that valence 2 was a negative predictor. of willingness to
express opinions, B = —0.29, SE = 0.09, [3 = —.23, t = —3.12, p = .002. That is, when the
user ratings indicated four points out of ﬁve possible points (i.e., four stars out of ﬁve
stars were yellow), meaning that a greater difference existed between individuals’ self
rating and the user ratings, individuals were less willing individuals to express their
opinions.

For the interaction between perceived effect on self and perceived effect on a
others, simple regression analyses were conducted at various points of perceived effect
on self. When perceived effect on self was at its mean, there was no relationship between
perceived effect on others and willingness to express opinion, B = 0.003, SE = 0.09, t =
0.03, p = .97. When perceived effect on self was at 1 SD below, there was a positive
relationship between perceived effect on others and willingness to express opinion, B =
0.11, SE = 0.09, t = 1.15, p = .25. The negative relationship between perceived effect on

others and willingness to express opinions increased in strength as perceived effect on

39

self increased in strength; for 2 SD below, B = 0.21, SE = 0.11, t = 1.92, p = .06. On the
other hand, when perceived effect on self was at 1 SD above, there was a negative
relationship between perceived effect on others and willingness to express opinions, B =
—0.10, SE = 0.10, t = —1.00, p = .32. The negative relationship between perceived effect
on others and willingness to express opinion became stronger as perceived effect on self

became stronger; for 2 SD above, B = —0.20, SE = 0.12, t = —1.67, p = .10.

40

Table 3. Regression Results for Willingness to Express Opinions.

 

 

 

 

B SE EB t sr

First Block
Volume] 0.12 0.10 .06 1.21 .12
Valence 12 0.02 0.12 .01 0.18 .02
Valence 23 0.04 0.05 .05 0.81 .04
Perceived Effect on Self 0.07 0.05 .10 1.49 , .07
Perceived Effect on Others 0.12 0.05 .15 236* .12

F(5, 380) = 4.77, p < .001, adj.R2 = .05

Second Block
Volume X Valence l —0.23 0.23 —.10 —0.98 —.23
Volume X Valence 2 —0.16 0.11 —.13 —l.56 —.16
Volume x PE on Self4 0.03 0.09 .03 0.34 .03
Volume x PE on Others’ 0.15 0.10 .14 1.52 .15
Valence l X Valence 2 —0.48,, 0.12 —.30 44.15,"? —.48
Valence l X PE on Self 0.09 0.11 .07 0.80 .09
Valence 1 X PE on Others —0.06 0.12 —.04 —0.48 —.06
Valence 2 X PE on Self -0.01 0.04 —.01 -0.18 —.01
Valence 2 X PE on Others 0.03 0.05 .05 0.75 .03
PE on Self X PE on Others —0.08 ,. . 0.03 —.14 32.76“ —.08

Fermge (10, 370) = 3.51, p < .001 , chhang, = .08 \‘TM

Third Block
Volume X Valence 1 X Valence 2 0.32 0.25 .15 1.31 .06
Volume X Valence 1 X PE on Self —0.03 0.22 —.02 —0.15 —.01
Volume X Valence 1 X PE on Others —0.36 0.23 —.21 —1.53 —.07
Volume X Valence 2 X PE on Self —0.02 0.09 —.02 —0.16 —.01
Volume X Valence 2 X PE on Others —0.15 0.10 —.15 —1.49 —.07
Volume X PE on Self X PE on Others 0.08 0.06 .12 1.34 .07
Valence 1 X PE on Self X PE on Others —0.04 0.07 -.05 —0.57 —.03
Valence 2 X PE on Self X PE on Others —0.03 0.03 —.07 —0.88 —.04
Valence l X Valence 2 X PE on Self —0.16 0.12 -.13 —1.30 -.06

Valence 1 X Valence 2 X PE on Others 0.11 0.12 .09 0.93 .05

chng. (10, 360) = 1.12, p = .35, Ram, = .03

*p < .05, **p < .01, ***p < .001

sr: semipartial correlation

Regarding multicollinearity, the ﬁrst-order and second-order predictors had variance

inﬂation factor (VIF) ranging from 1.01 to 5.49. Cohen, Cohen, West, and Aiken (2003)

mentioned 10 as the traditional rule of thumb threshold value, although there is “no good

statistical rationale for the choice of any of the traditional rule of thumb threshold values”

(p. 424).

ldummy-coded with low volume = 0 and high volume = 1

2 dummy-coded with negative valence = 0 and positive valence == 1

3 Valence 2 was calculated as the difference between the user ratings and participants’

self rating before being exposed to the user ratings.

" Perceived Effect on Self

5 Perceived Effect on Others

 

41

STUDY 2

Study 1 did not reveal a signiﬁcant effect for volume of user ratings. In the high
volume condition, ‘350 voters’ was presented to participants in order to make them
believe that the manipulated user ratings came from real students. Considering the results
for volume in study 1, it is possible that ‘350 voters’ may not have been sufﬁciently large
enough to represent a high volume. Thus, study 2 was conducted in order to test the effect
of volume further by adding more high volume conditions. Study 2 used the same
experiment procedure and questionnaire as in study 1, with the only difference between
the two being the volume conditions.

Study 2 Method
Participants

Undergraduates in Communication and Advertising classes at Michigan State
University and at the University of Hawaii were asked to participate in the survey in
exchange for extra research credit. In total, 133 students voluntarily participated in the
study. They were randomly assigned to one of four additional conditions: super high and
positive valence, super high and negative valence, mega high and positive valence, and
mega high and negative valence. Of the participants, ten Korean international students
were excluded ﬁom the main analyses. Therefore, the responses of 123 students were
used in the analysis. In addition, the study 1 data of low and high volume conditions were
included in the study 2 data analysis. Thus, 243 participants was the total number for the
study 2 data analysis.

Design

The current study used a 4 (volume: low, high, super high, and mega high) X 2

42

(valence: positive and negative) x 2 (before and after exposure to user ratings) mixed
design with volume and valence as between-subject factors. The volume of user ratings
was manipulated through low (15 voters), high (350 voters), super high (1050 voters),
and mega high (3150 voters) volume conditions. The valence Of user rating was
manipulated in the same way as in study 1 (i.e., positive valence = one yellow star;
negative valence = four yellow stars).

There were 11.5% participants (11 = 28) in condition 1 (low volume and positive
valence), 10.3% participants (n = 25) in condition 2 (low volume and negative valence),
15.2% participants (11 = 37) in condition 3 (high volume and positive valence), 12.3%
participants (n = 30) in condition 4 (high volume and negative valence), 23.9% (n = 58)
in condition 5 (super high volume and positive valence), 13.2% (n = 32) in condition 6
(super high volume and negative valence), 6.2% (n = 15) in condition 7 (mega high
volume and positive valence), and 7.4% (n = 18) in condition 8 (mega high volume and
negative valence).

Stimulus and Measurement

Study 2 used only one movie, Secret Sunshine (drama), as a stimulus material,
unlike study 1, which used three movies (comedy, drama, and horror). A larger font size
and different colors indicating volume (i.e.1050 voters and 3150 voters) were used to
make the volume of user ratings more noticeable for participants in the super high and
mega high volume conditions.

The procedure and measurements of study 1 were the same as in study 2. Table 4

reports the reliabilities, means, standard deviations, and correlations of all variables.

43

.Bcowae 05 :o 00:88 0.8 cc Enema—:88 83:30:0m
:a 3 ”mm see 805% .3. var. .3v a...

 

 

 

320T Gm. : so. a com. 3 am. 0 A8. 3 A8. a 5.8 Go. s cm
wed 3d ﬂaw gum mod mm.m aﬁm nod Worm 3 BER
E. .33. mo: *m _. *3. mo. mo. 5.- No. h850.2%
3.. mo. in _. em ﬂ. *3. m _. 3. *3. seam—8:0 000m
:oEEO mm0axm
3.. **MN. **2. *2. **mm. **NN. :90. 3 mm0cw:==3
. h050
co. **3. 1.3. 1mm. *1. _. 1.3. so swam eo>_00._0m
:0m
co. No. .I. _ N. N— . **NN. co “00.0.5 “003080;
osmoaxm 00%
mo. 1.5x **VM. .1. _ N. mega .8050
ocsmomxm 00cm
3. 1%. tax. ease :om
Eamoaxm 0.8.003
S. :3. Seem .aoeo
Esmogxm 00809
a. wane :om
0.58930 ocsmomxm oesmomxm
E050 20m 038qu
8300 cote—oi 30: 00% 283 0.8002
:0 80km :0 80km Ste
03 co so“. 05:; mean megs. mesa
002023 003080; meted 20m
.2050 .3050 :0m

 

.N 163m 00 m0333> 00 20:00:00 new .mcocﬂgm Eeecmum .2002 .m0Ezpmz0m .v 0331

44

00005500: 05 :_ 50:30.. 000 0:05:30: 5:05:05 .08 Z

 

 

 

 

 

 

m~H= w~ﬂ= hm": NM": hm”: Om”: wNH: mm":
825::
30.: SM 2:: E0 2 20 m2 $28 03 20.: EN 80.: 3.0 A8: $0 0:: $0 $293
8 Au..m0:w:==>9
$8 $0 $8 $0 :98 $0 5.8 03 $8 $0 308 Ed $8 $0 5.8 $0 :38
0.5898
.050 5:0
$8 :0 $8 $0 :38 2: 30.8 6.0 A8,: 3.0 $08 So A88 2: 8:8 33 283 00520
$50: :050
05890 .050
5.8 as 398 $2 608 ”no :98 $0 3.8 33 $8 $2 8:8 38 5.8 :5 0%.. 8&2.
.. 0w:0:0 wéﬁ 20m
Gm: $.m $8 03. E: can 60.: N2 :08 PM 80.: $0 6:: 03 8:: Rm :38
£050 :0
$0.: $5. $8 03 2 :8 3m 5.: $.m 60.8 $3 $0: $5 :20 $0 an: $.m 08:0 8383:
20m :0
80.: 5.. A8,: Ba and 8.0 $0.: $0 $20 mom 6:: own so: 30 cm: a.“ 0850 8383:
8:053 00:0_0> 00:0_0> 00:0_0> 00:0_0> 00:0_0> 00:0_0> 00:0_0>
03:8; 0>50w0z 03:08 0>50m0z 0>Emom 0>50w0z 03:8; 0>50w0z
0E:_o> :33 $02 05375 if .096 0E=_o> :32 08:55 >64

 

0:055:00 _S:0E_:0mxm N .355 mo 0550300 5.820% 5:: 0:002 .m 030,—.

45

Study 2 Results

The analyses here focused on the effect of the four volume conditions. Table 4
reports means and standard deviations of the TPE l, TPE2, and willingness to express
opinions in each experimental condition.

The Third Person Eﬂect of User Ratings on Self and Others

It was predicted that user ratings would be perceived to have a greater effect on
others than on self (H1), especially if user ratings showed high volume rather than low
volume conditions (H2), and if the more likely user ratings would be to differ from
participants’ own rating (H3). To test hypothesis 1, 2, and 3, three sets of ANOVA
analyses were conducted on TPE 1 and TPE 2 as dependent variables.

Overall Third Person Eﬂect (HI). In analysis 1 (TPE 1), a 2 (target: self and
others) X 4 (volume: low, high, super high, and mega high) x 2 (valence 1: positive and
negative) mixed ANOVA was conducted to compare the perceived effect of user ratings
on the target (self and others). The analysis revealed a signiﬁcant main effect for target, F
(1, 233) = 48.76, p < .001, 112 = .04. Consistent with H1, the perceived effect of user
ratings on others (M = 3.69, SD = 1.34) was signiﬁcantly higher than on self (M = 3.08,
SD = 1.38).

In analysis 2 (TPE 2), a 2 (target: self and others) X 4 (volume: low, high, super
high, and mega high) x 2 (valence 1: positive and negative) mixed ANOVA was
conducted to compare self rating change and other rating change before and after
exposure to user ratings. The analysis revealed a signiﬁcant main effect for the target, F
(1, 234) = 39.49, p < .001, n2 = .06. Other rating change before and after exposure to user
ratings (M = 0.89, SD = 0.88) was signiﬁcantly greater than self rating change (M = 0.46,

46

SD = 0.43).

Volume (HZ) and valence I. In analysis 1 (TPE 1), the analysis did not reveal a
signiﬁcant main effect for volume, F (3, 233) = 2.54, p = .06, n2 = .02, and did not show a
signiﬁcant main effect for valence, F (1, 233) = 0.09, p = .77, n2 = .00. Thus, people’s
different perceptions of user ratings on self and others were not affected by the volume
and valence of user ratings. Volume and valence 1 did not interact signiﬁcantly, F (3,
233) = 0.16, p = .92, n2 = .00. Volume did not interact with the target (perceived effect on
self and others), F (1, 233) = 1.86, p = .14, n2 = .004, and valence did not interact with
the target, F (1, 233) = 0.01, p = .99, n2 = .00. The three-way interaction among volume,
valence, and the target was not signiﬁcant, F (1, 233) = 1.30, p = .28, n2 = .003.

In analysis 2 (TPE 2), the analysis did not reveal a signiﬁcant main effect for
volume, F (3, 234) = 0.64, p = .59, n2 = .00, and did not show a signiﬁcant main effect for
valence, F (l, 234) = 0.48, p = .49, n2 = .00. Therefore, inconsistent with H2 and H3, the
before and after rating change was not affected by the volume and valence of user ratings.
Volume and valence did not interact signiﬁcantly, F (3, 234) = 0.45, p = .72, n2 = .00.
Volume did not interact with target (self rating change versus other rating change), F (1 ,
234) = 0.84, p = .47, n2 = .00. The three-way interaction among volume, valence, and the
target was not signiﬁcant, F (1, 234) = 0.39, p = .76, n2 = .00. The two-way interaction
between valence and the target was signiﬁcant, F (1, 234) = 5.31, p = .02, n2 = .009. The
extent to which other rating change was larger than self rating change was greater in the
positive valence condition (other rating change M = 0.99, SD = 0.95 versus self rating
change M= 0.42, SD = 0.39, t [130] = 6.85,p < .001, n2 = .26) than in the negative

valence condition (other rating change M = 0.77, SD = 0.76 versus self rating change M =

47

0.52, SD = 0.49, t [130] = 3.35,p = 001,112 = .10).

Valence diﬁ’erence (H3 ). With regard to the effect of valence difference between
user ratings and self rating (valence 2), H3 predicted that, as the difference between the
user ratings and people’s own ratings increased, the self-other gap of user rating effect
would increase. Overall, valence 2 was not signiﬁcantly related to the difference between
the perceived effect of user ratings on self and others (D-TPE 1), r (239) = .01, p = .90,
and it was not signiﬁcantly related to the difference between self rating change and other
rating change before and after exposure to user rating (D-TPE 2), r (240) = .05, p = .49.
When correlations were conducted for each volume condition, these statistically
insigniﬁcant correlations between valence 2 and D-TPE 1 and between valence 2 and D-
TPE 2 did not change across low, high, super high, and mega high volume conditions.
Willingness to Express One ’s Opinion

To further examine the effects of volume, a 4 (low, high, super high, and mega
high volumes) X 2 (positive and negative valences) between-subject ANOVA was
conducted on willingness to express opinion. The results showed no main effect for
volume, F (3, 242) = 0.70, p = .55, n2 = .00, no main effect for valence, F (3, 242) = 0.11,
p = .74, n2 = .00, and no interaction effects for volume and valence, F (3, 242) = 0.34, p =
.80, n2 = .00. The data were inconsistent with hypothesis 5, which predicted that, as the
volume of user ratings increased, people’s willingness to express their opinions would be
reduced.

When valence 2 and the difference between perceived effect on self and others
were examined for their relationships with willingness to express opinions across the four

volume conditions, none of the correlations were signiﬁcant.

48

SUMMARY OF THE FINDINGS OF STUDY 1 AND STUDY 2

The results of study 1 conﬁrmed the overall Third Person Effect of user ratings in
that the perceived user rating effect on others (TPE 1) was signiﬁcantly higher than on
self, and the other rating change was signiﬁcantly larger than the self rating change
before and after exposure to the user ratings (TPE 2).

Inconsistent with H2, however, both TPE l and TPE 2 were not affected by the
volume of user ratings. People’s perceived effects of user ratings on self versus others did
not differ in the high or in low volume conditions. For the observed rating change, the
overall change was larger in the low volume condition than in the high volume condition.
In addition, other rating changes in the low volume conditions were larger than in the
high volume conditions, while participants’ self rating remained the same before and
after exposure to user ratings.

The valence of user ratings did not affect the perceived effects on self and others
(TPE 1), but they did affect the observed changes of self ratings and other ratings (TPE
2). The overall change (the average of self change and other change) was larger in the
negative valence condition than in the positive valence condition.

The valence difference between user ratings and self ratings (valence 2) did not
predict the difference in the perceived user rating effects on self and others (D-TPE l),
but it was a signiﬁcant predictor of the difference in self rating change and other rating
change (D-TPE 2). For D-TPE 1, valence difference was a signiﬁcant and positive
predictor in the high volume condition but not in the low volume condition. Only in the
high voltune condition was the extent to which user ratings differed from self rating

positively associated with the extent to which the perceived effect of user ratings was

49

greater on other than on self. On the other hand, valence difference was a signiﬁcant
predictor for the difference in self rating change and other rating change (D-TPE 2). The
greater the difference people perceived between the user ratings and their own ratings, the
more likely they were to perceive that other ratings would change to a greater extent
before and aﬁer exposure to user ratings.

The results showed that participants’ perceived effect on others was a signiﬁcant
predictor of willingness to express their opinions. As individuals perceived others were
greatly inﬂuenced by the user ratings, they were more willing to express their opinions.
In addition, as the tendency to minimize the perceived effect on self and to maximize the
perceived effect on others was stronger, their willingness to express their own opinion
also was stronger. Finally, the valance difference between user ratings and individuals’
own ratings affected their willingness to express their opinions with relation to the
manipulated valence of user ratings. In the negative valence condition, as the more likely
individuals’ self ratings would be to different from the user ratings, the more willing they
were to express their opinion; however, in the positive valence condition, they were less
willing to express their opinions when greater difference existed between individuals’
self rating and the user ratings.

Study 2 was conducted to further test the effect of the volume of user ratings by
adding 2 more high volume conditions. TPE 1 and TPE 2 were not affected by the
volume and valence of the user ratings. Finally, the volume of user rating was not a

signiﬁcant predictor of willingness to express one’s opinion.

50

DISCUSSION

The purpose of this study was to explore the formation process of online user
ratings of movies by focusing on the relationship between individuals’ perceptions of the
results of online user ratings as public opinions and willingness to express their opinions.
Based on the Third Person Effect theory, it was predicted that people may perceive that
the online user ratings have great effects on others but not on themselves. Especially
when the online user ratings were perceived as public opinions and when the results of
user ratings were not agreeable with individuals’ own opinions, people may tend to
minimize the user rating effects on self and to maximize the user rating effects on others.
In addition, peoples’ concerns for the perceived user rating effects on others could act as
a motivation to express their opinions on online user ratings, especially when they
perceived that the results of the online user ratings were different from their opinions. In
order to test the third person effect of online user ratings and their role as a moderator of
willingness to participate in user ratings, online survey about Korean movies with
manipulated user ratings was conducted.

The results indicated that people perceived the user ratings to have greater effects
on others than on themselves. Moreover, how observed peoples’ own ratings and others’
imagined ratings changed before and after exposure to user ratings supports this
tendency. People indicated that other rating changes would be larger than their rating
change before and after exposure to user ratings. The volume of user ratings did not
affect the self-other gap of perceived effects and self and others’ observed rating changes;
however, the valence difference between user ratings and individuals’ own ratings

inﬂuenced the gap of self and other rating changes. That is, as the more people perceived

51

the results of online user ratings to disagree with their own opinions, the more they
tended to minimize the effects of user ratings on themselves and to maximize the effect
of user ratings on others. Thus, the other rating change was greatly larger than the self
rating change.

The Spiral of Silence theory argues that people who perceived their opinion to be
in the minority may be less willing to express their opinion due to fear of isolation from
others; however, the current study assumed that there may be a difference between
peoples’ willingnesses to express their opinions in off-line situations and on—line
situations. On the Internet, people may have less fear of isolation than in off line
conversations due to the main characteristic of the virtual world, anonymity. At this
point, the current study questioned what the motivation is for people to have more or less
willingness to express their opinions on the Internet. With relation to the argument of the
Third Person Effect theory, it was predicted that people whose own ratings were greatly
different from the valence of user ratings would be more willing to participate in user
ratings in order to change the undesirable results of the user ratings due to concern for
great user rating effects on others.

Consistent with the prediction, a relationship between peoples’ perceptions of the
effect of user ratings on others and their willingness to express their opinions was
observed. People’s concern for the perceived effect of user ratings on others was a
signiﬁcant predictor of their willingness to express their opinions. As more people tended
to minimize the perceived effect of user ratings on self and tended to maximize the
perceived effect of user rating on others, the more they were willing to express their

opinions. Finally, only in negative valence conditions the valence difference between

52

user ratings and individuals’ self ratings positively predicted willingness to express their
opinions. As the gap grew between user ratings and self ratings, people were more
willing to express their opinions in negative valence user ratings, while they were less
willing to express their opinions in positive valence user ratings.
Implications for Public Opinion on the Internet

In research of online user ratings about movies, previous ﬁndings have not
revealed a direct persuasive effect of the valence of user ratings on people’s attitudes
toward movies, which should be reﬂected in box ofﬁce revenue, and have revealed a
positive predictive effect of the volume of user ratings on box ofﬁce revenues (Dellarocas
et al., 2004; Duan et al., 2005; Liu, 2006); however, the current study revealed the
positive effects of the valence and did not reveal a predictive effect of the volume user
ratings. These contradictory ﬁndings came from the different dependent variables
between previous research of user ratings and the current study; the dependent variable of
previous research was the box office revenue of movies (i.e. people’s real behavior to
purchase a movie ticket) and the dependent variable of the current study was peoples’
willingness to participate in user ratings. Although positive user ratings do not make
people purchase movie tickets and negative user ratings do not change people’s attitudes
toward movies according to the ﬁndings of previous studies, they may affect the
formation process of user ratings based on the current study. As the current study
indicated, online user ratings may affect people’s behavior rather than peoples’ attitudes.
When people perceive user ratings to disagree with their own opinions, they tend to deny
the user rating effects on self instead of changing their attitudes about movies; however,

people are more willing to participate in user ratings which disagree with their opinions

53

due to their concerns for others who may be greatly inﬂuenced by the results of user
ratings.

For the effect of volume, previous research used real user ratings about released
movies on real Internet sites from before to after releasing, while this study used
manipulated user ratings about unreleased movies. Thus, the volume of real user ratings
which were used in previous research included both people’s expectations about movies
before their release and their evaluations of movies aﬁer their release, while the volume
of user rating in the current study involved only people’s expectations about released
movies. In addition, the number of participants of real user ratings could be related to the
number of audience members who purchased movie tickets and watched the movies.
Thus, the volume of real user ratings could be both an independent variable and a
dependent variable, and there might be a strong positive relationship between the volume
of user ratings and box office revenues of movies; however, the volume of hypothetical
user ratings of the current study was used as an independent variable to affect people’s
perceptions of user ratings.

The results of study] and study 2 did not reveal a signiﬁcant effect of volume of
user ratings. It may suggest that people do not pay attention to the volume of public
opinion polls when they view the results of public opinion polls. In relation to this
phenomenon, Tversky and Kahneman (1974) explained that people use some heuristic
principles when they have to judge uncertain events and probabilities. Intuitive judgnents
depending on heuristic principles are not affected by sample size (Tversky & Kahneman,
1974). That is, when people need to investigate public opinion climate about issues, they

easily use the results of online opinion polls, regardless of sample size and whether or not

54

it provides real public opinion. Similar to opinions polls reported by the mass media,
online opinion polls are perceived as the ‘public’ opinions by people by just being posted
on the Internet site (Wu & Weaver, 1997).

While the Spiral of Silence theory explains why people choose to be silent based
on their perception of the opinion climate, the current study explained why pe0ple choose
to speak out due to the interaction between individuals’ own opirnions and their
perceptions of opinion circumstance. As the current study revealed, concerns for others
who may be vulnerable to the poll results act as a signiﬁcant motivation of people’s
willingness to participate in online polls. In addition, individuals’ willingness would be
inﬂuenced by their perceptions of public opirnion climates. When opinion polls showed
negative opirnion climates, individuals who have positive opinions were more willing to
express their opinions; however, when opinion polls showed positive opinion climates,
individuals who have negative opinions avoided expressing their opinions. This result
provides more speciﬁc explanations about why and when people have more or less
willingness to express their opinions, which did not fully identify with the Spiral of
Silence Theory.

Limitations

The current study may have had some limitations. The ﬁrst limitation pertains to
the stimulus materials. The manipulated user ratings about three Korean movies were
used as stimulus for American participants. Although ‘350 voters’ of user ratings was
presented as the high volume condition, it could be not enough to represent public at
large for study 1. In addition, American participants had little interest in foreign moves

(M = 2.80, SD = 1.56 out of 6 points scales) and in Asian movies (M = 2.22, SD = 1.29

55

out of 6 points scales). Only 29.2% (n = 115) of the participants had the experience of
watching Asian movies and 70.6% (n = 278) of the participants had never watched Asian
movies. Thus, the lower willingness to participate in the user rating about an Asian movie
may be due to their lower interests in Asian movies regardless of their perceptions of user
rating results. The roles of issue importance, personal involvement, and issue interest
need to be investigated as moderators in further research. Finally, although all of the
participants were randomly assigned to one of the three movies, their genre preference
also might still have inﬂuenced the user rating effects on themselves and on others and
their willingness to participate in user ratings.

A second limitation is the problem of measurement of willingness to express
one’s opinion. The current study asked participants to rate about a movie before and after
exposure to the user ratings and then measured participants’ willingness to participate in
user ratings about the movie in imagined on-line situations by using direct questions. This
caused two problems. The ﬁrst problem was that participants had already participated in
user ratings, regardless of their willingness. Thus, they may have perceived the questions
asking about their willingness to be redundant. The second problem is that there may be a
gap between hypothetical on-line situations and real on-line rating situations. In addition,
the items asking about willingness to express one’s opinion were created for this study,
and their reliabilities were only slightly above .70. Thus, they need to be further validated
through rigorous testing.

Conclusion
This study addressed an aspect of on-line user ratings as a form of public opinion

poll by focusing on people’s perceptions of others and its effect on willingness to

56

participate in user ratings. Further research in this area is needed to investigate how
people perceive online opinion polls and why people are more or less willing to
participate in them. With increases in the importance of the Internet and the number of
Internet users, it is crucial to provide a new paradign to understand the mechanism of

public opinions on the Internet both for scholars and active audiences.

57

APPENDIX

Questionnaire: The Effect of User Ratings

03° Directions: The following picture shows information about a movie. Please read the
following information about the movie and answer the questions.

ONE’S OWN RATING BEFORE BEING EXPOSED TO THE USER RATING
Q1. To me, this movie seems to be

Interesting

Funny"‘l

Boring (r)2

Appealing

Impressive

Exciting

Like a waste of time (r)

Ordinary (r)*

Q2: If you were to rate this movie, your rating of this movie would be _ out of 5.

9°NP‘P‘PP‘PI"

IMAGINED OTHERS’ RATINGS BEFORE BEING EXPOSED TO THE USER
RATING

Q3. To other students, this movie seems to be

Interesting

Funny*

Boring (r)

Appealing

Impressive

Exciting

Like a waste of time (I)

. Ordinary (r)*

Q4. If other students were to rate this movie, their rating of this movie would be _ out
of 5.

WSQMPPN?

THE EFFECT OF USER RATING ON SELF AND OTHERS
Directions: The College of Communication Arts & Sciences at MSU is planning on
hosting an International Film Festival next year. Various movies from various countries
are being considered for the International Film Festival.

For future International Film Festival progarns, The College of Communication Arts
& Sciences has been collecting students' opinions about foreign movies. These student
opinions about foreign movies provide useful information for the movie selection.

The following rating of a movie shows the rating of other MSU students who
participated in this survey in last spring and summer semester.

 

1 * indicates deleted items when creating composite variables.
2 (r) indicated recoded items.

58

Your rating about this movie will be included in an updated version of this survey for
next spring semester. Therefore, other MSU students who will participate in this survey
will also see average students' rating and your rating contributes to this average.

Please read students' rating about this movie and answer the questions.

ONE’S OWN RATING AFTER BEING EXPOSED TO THE USER RATING

Q5. To me, this movie seems to be

Interesting

Funny *

Boring (r)

Appealing

Impressive

Exciting

Like a waste of time (r)

Ordinary (r)*

Q6. If you were to rate this movie, your rating of this movie would be

Q7. How much do you think this rating has affected your rating of this movie?

Q8. How much do you think this rating has affected your attitude about this movie?
Q9. How much do you think this rating has affected your expectations of this movie?

?°>'.°‘E":“ 9’3"!"

IMAGINED OTHERS’ RATING AFTER BEING EXPOSED TO THE USER RATING
Q10. To other students, this movie seems to be

1. Interesting

2. Funny*

3. Boring (r)

4. Appealing

5. Impressive

6. Exciting

7. Like a waste of time (r)
8 Ordinary (r)*

Q1 1. If other students were to rate this movie, their rating of this movie would be

Q12. How much do you think this rating has affected other students’ ratings of this

movie?

Q13. How much do you think this rating has affected other students’ attitudes about this

movie?

Q14. How much do you think this rating has affected other students’ expectations of this

movie?

Q15. In your opinion, what percentage of students in this class will change their ratings
of this movie after being exposed this rating?

WILLINGNESS TO PARTICIPATE IN RATING
1. Do you agee to have your rating of this movie reﬂected in later versions of
this questionnaire for other students in other classes?*
2. If multiple voting was possible, please indicate how many times you would
wish to vote in this user rating.*
3. I am willing to post my rating on a web site for this movie.

59

4. If you were to see information about this move on other website, how much

5.

6.

would you wish to participate in a user rating of this movie?

If you were to see information about this move on other website, how much
would you wish to write a user review of this movie?

If your class were to deal with this movie as a discussion topic, how much
would you wish to participate in the class discussion?

DEMOGRAPHIC INFORMATION

1.
2.

3.

NP‘

 

 

Age _

Gender _ Male _ F ernale

Ethnicity (Please check one of the following)

_ Caucasian __ Hispanic

_ African American __ Paciﬁc Islander

__ Native American __ Mixed (Please specify)

__ Asian American _ Other (Please pecify)

If you are an Asian American, which country in Asia do you associate

yourself with?
Current college standing

 

ﬁeshman _ senior
_ sophomore _ 5th year senior (or beyond)
_ junior __ MA student
__ PhD student _ Other (please specify):
International student yes no

If you are an international student, which country are you from?

MOVIE SELECTION BEHAVIOR

l.
2.
3.

>195”?

99°

How often do you watch movies in the theater?
How often do you watch movies on a DVD?
What is your favorite movie genre? Please indicate one.

Action Fantasy Comedy
Romantic comedy Drama Animation
Horror Others (write in):

 

 

How much interest do you have in foreign movies?
How much interest do you have in Asian movies?
Have you ever watched Asian movies?
If you answered ‘yes’ to number six, please indicate the names of movies
which you have watched recently.
How much did you enjoy those Asian movies which you listed in Question 7?
Which sources do you usually use to watch movies? Please indicate two
sources.

_ Television ads __ Newspaper ads
__ Magazine ads __ Movie websites
__ Information from friends _ Critics’ reviews
__ User reviews on the Internet _ Others (write in):

 

60

10. How important do you consider information about movies from these
sources when you decide to watch movies?
Television ads
Newspaper ads
Magazine ads
Movie sites
Information from ﬁiends
Critics’ reviews
User reviews on the Internet

Others

11. With whom do you usually watch movies?
_Alone _Friends _A date
_Family ___Others (write in):

 

12. Why do you watch movies?
13. How do you prefer to watch movies?
Theater DVD TV

Internet Others (write in):

 

THE FEAR OF ISOLATION SCALE (Neuwirth, 2000)

1. Those members who are in basic disageement with a group’s ideas ought to be
thrown out of the goup.*

2. It’s better to keep one’s friends than to keep one’s ideas.*

3. At times, I worry a lot that others might not like me.

4. At times, I worry about being alone.

5. I like to stick to my opinions even though everyone is against me (r).*

6. I’m not worried that, if I really say what I think about things and other people,
this will cause me to lose my ﬁ'iends.*

THE WILLINGNESS TO SELF-CENSOR SCALE (Hayes, A. F ., Glynn, C. J ., &
Shanahan, J. 2005).
1. It is difficult for me to express my opinion if I think others won’t agee with
what I say.
2. There have been many times when I have thought others around me were
wrong, but I didn’t let them know.
3. When I disagee with others, I’d rather go along with them than argue about it.
4. It is easy for me to express my opinion around others who I think will
disagee with me (r).*
5. I’d feel uncomfortable if someone asked my opinion and I knew that he or she
wouldn’t agee with me.
6. I tend speak my opinion only around ﬁiends or other people I trust.
7. It is safer to keep quiet than publicly speak an opinion that you know most
others don’t share.
8. If I disagee with others, I have no problem letting them know it (r).

61

BIBLIOGRAPHY

Alicke, M. D., Klotz, M. L., Breitenbecher, D. L., Yurak, T. J ., & Vredenburg, D. S.
(1995). Personal contact, individuation, and the better-than-average effect.
Journal of Personality and Social Psychology. 68, 804-825.

Alvaro, E. M., & Crano, W. D. (1996). Cognitive responses to minority or majority-based
communications: Factors that underlie minority inﬂuence. British Journal of
Social Psychology, 35, 105-121.

Anderson, E. W. (1998). Customer satisfaction and word of mouth. Journal of Service
Research, 1, 5-17.

Amdt, J. (1967). Word of mouth advertising and informal communication. In D. F. Cox
(Eds.), Risk Taking and Information Handling in Consumer Behavior (pp. 188-
239). Boston: Harvard University.

Austin, B. (1982). Film audiences: why college students chose to see their most recent
ﬁlms. Journal of Popular Film and Television, 9, 43-49.

Baldassare, M., & Katz, C. (1996). Measures of attitude strength as predictors of
willingness to speak to the media. Journalism and Mass Communication
Quarterly, 73, 147-158.

Banerjee, A. (1992). A simple model of herd behavior. Quarterly Journal of Economics,
107, 797-817.

Burzynski. M., & Bayer, D. (1977). The effect of positive and negative prior information
on motion picture appreciation. Journal of Social Psychology. 101, 215-218.

Davison, W. P. (1983). The third-person effect in communication. The Public Opinion
Quarterly, 47, 1-15.

Dellarocas, C. N., Awad, N., & Zhang, X. (2004).Using online ratings as a proxy of
word-of mouth in motion picture revenue forecasting. Working paper, MIT Sloan

School of Management. Available at SSRN: http://ssrn.com/abstract=620821

Dichter, E. (1966). How word-of mouth advertising works. Harvard Business Reviews,
16, 147-166.

Duan, W., Gu, B, & Whinston, A. (2005). Do online reviews matter? An empirical
investigation of panel data. Working paper, Department of Management Science
and Information System, University of Texas at Austin.

Eliashberg, J ., & Shugan, S. M. (1997). Film critics: inﬂuencers or predictors? Journal of

62

Marketing, 61, 68-78.

Faber, R., & O’Guinn, T. (1984). Effect of media advertising and other sources on movie
selection. Journalism Quarterly, 61, 371-377.

Fields, J. M., & Schuman, H. (1976). Public beliefs about the beliefs of the public. Public
Opinion Quarterly, 40, 427-448.

Franklin, B. (1994). Packaging politics, London: Edward Arnold.

Glynn, C. J ., & McLead, J. M. (1985). Implications of the spiral of silence theory for
communication and public opirnion research, In K. R. Sanders, L. L. Kaid, & D.
Nimmo, (Eds) Political Communication Yearbook, Carbondale and Edwardsville,
Southern Illinois University Press, 43-65.

Godes, D., & Mayzlin, D. (2004). Using online conversations to study word of mouth
communication. Marketing Science, 23, 545-560.

Gunther, A. C. (1995). Overrating the X-rating: The third-person perception and support
for censorship of pomogaphy. Journal of Communication, 45, 27-3 8.

Gunther, A. C., & Storey, J. D. (2003). The inﬂuence of presumed inﬂuence. Journal of
Communication, 53, 199-215.

Hayes, A. F ., Shanahan, J ., & Glynn, C. J. (2001). Willingness to express one’s opinion
in a realistic situation as a function of perceived support for that opinion.
International Journal of Public Opinion Research, 13, 45-58.

Hayes, A. F., Glynn, C. J ., & Shanahan, J. (2005). Willingness to self-censor: A construct
and measurement tool for public opinion research. International Journal of Public
Opinion Research, 17, 298-323.

Holbrook, M. (1999). Popular appeal versus expert judgnents of motion pictures.
Journal of Consumer Research, 26, 144-155.

Kent, M. L., Harrison, T. R., & Taylor, M. (2006). A critique of intemet polls as
symbolic representation and pseudo-event. Communication Studies, 5 7, 299-315.

Lasorsa, D. L. (1991). Political outspokenness: Factors working against the spiral of
silence, Journalism Quarterly, 68, 13 1-139.

Lavrakas, P. J ., Holley, J. K., & Miller, P. V. (1991). Public reactions to polling news
during the 1988 presidential election campaign. In P. J. Lavrakas & J. K. Holley
(Eds.), Polling and presidential election coverage (pp. 151-183). Newbury Park,
CA: Sage.

63

Liu, Y. (2006). Word of mouth for movies: Its dynamics and impact on box ofﬁce
revenue. Journal of Marketing, 70, 74-89.

Milavsky, J. R., Swift, A., Roper, B. W., Salant, R., & Abrams, F. (1985). Early calls on
election results and exit polls: Pros, Cons, and constitutional considerations.
Public Opinion Quarterly, 49, 1-15.

Morwitz, V.G., & Pluzinski, C. (1996). Do polls reﬂect opinions or do opinions reﬂect
polls? The impact of political polling on voters’ expectations, preferences, and
behaviors. The Journal of Consumer Research, 23, 53-67.

Mutz, D. C. (1989). The inﬂuence of perceptions of media inﬂuence: Third person effects
and the public expression of opinions. International Journal of Public Opinion
Research, 1, 3-23.

Mutz, D. C. (1998). Impersonal inﬂuence: How perceptions of mass collectives affect
political attitudes. Cambridge, UK: Cambridge University Press.

Murray, K. (1991). A test of services marketing theory: Consumer information
acquisitions activities. Journal of Marketing, 55, 10-25.

Neuwirth, K. J. (2000). Testing the spiral of silence model: The case of Mexico.
International Journal of Public Opinion Research, 12, 138-59.

Noelle-Neumann, E. (1974). The spiral of silence: A theory of public opinion. Journal of
Communication, 24, 43-51.

Ozcan, K. (2004). Consumer-to-consumer interactions in a networked society: word of
mouth theory, consumer experiences, and network dynamics. Dissertation for
PhD, Department of Business Administration, University of Michigan.

Park, H. S., & Salmon .C. T. (2005). A test of the tlnird-person effect in public relations:
Application of social comparison theory. Journal of Mass Communication
Quarterly, 82, 25-43.

Pan, 2., Abisaid, J. L., Paek, H. J ., Sun, Y., & Houden, D. (2005). Exploring the
perceptual gap in perceived effects of media reports of opinion polls.
International Journal of Public Opinion Research, 18, 340-350.

Perloff, R. M. (1999). The third-person effect: A critical review and synthesis. Media
Psychology, 1, 353-378.

Rafaeli, S. (1988). Interactivity: From new media to communication. In R. P. Hawkin, J.

M. Wiemann & S. Pingee (Eds.), Advancing communication science: Merging
Mass and interpersonal process (pp. 110-134). Newburry Park2Sage.

64

Richins, M. L. (1983). An analysis of consumer interaction styles in the marketplace.
Journal of Consumer Research, 10, 73-82.

Russell, R. (2006). Users, interactivity, and generation. New Media & Society, 8, 531-
550.

Salmon, C. T., & Neuwirth, K. (1990). Perception of Opinion climates and willingness to
discuss the issue of abortion. Journalism Quarterly, 67, 567-577.

Salwen, M. B., & Driscoll, P. D. (1997). Consequences of third-person perception in
support of press restrictions in the O. J. Simpson trial. Journal of Communication,
4 7, 60-75

Savigny, H. (2002). Public opinion, political commurnication and the internet. Politics,
22, 1-8.

Schultz, T. (1999). Interactive options in online journalism: a content analysis of 100
US. newspapers. Journal of Computer-Mediated Communication, 5, URL
(consulted January 2003): http://jcmc.indiana.edu/volS/issue1/schultz.htrnl

Schultz, T. (2000). Mass media and the concept of interactivity: an exploratory study of
online forums and reader email. Media, Culture & Society, 22, 205-221.

Shah, D., Faber, R. J ., & Youn, S. (1999). Susceptibility and severity: Perceptual
dimensions underlying the third-person effect. Communication Research, 26, 240-
267.

Shamir, J. (1997). Speaking up and silencing out in face of a changing climate of opinion.
Journalism and Mass Communication Quarterly, 74, 602-614.

Tversky, A., & Kahneman, D. (1974). Judgnent under uncertainty: Heuristics and biases.
Science, New Series, 185, 1 124-1 131.

Willnat, L., Lee, W., & Detenber, B. H. (2002). Individual-level predictors of public
outspokenness: A test of the spiral of silence theory in Singapore. International
Journal of Public Opinion Research, 8, 229-247.

Wu, W., & Weaver, D. (1997). On-line democracy or on-line demagoguery?: Public

opinion “polls” on the internet. The Harvard International Journal of
Press/Politics, 2, 71-86.

65

 

ulill1lljijlnuljjnjjjlumI