Third-Person Effect and Social Networking:

Implications for Online Marketing and

Word-of-Mouth Communication
Jie Zhang, University of Texas - Austin
Terry Daugherty, The University of Akron

Introduction
Social networking websites (SNWs) have emerged as an increasingly influential media platform.
As of 2007, there were 110 million active users on MySpace and 90 million active users on
Facebook all testifying to the booming popularity of SNWs (Boyd and Ellison 2007). Unlike
traditional media, SNWs are user-generated and user-centered. SNWs attract users by allowing
them to “construct a public or semi-public profile within a bounded system,” establish
connections and maintain interactions with others in the system (Boyd and Ellison 2007).

Davison’s (1983) third-person effect (TPE) theory proposes that individuals tend to expect mass
media to have a greater effect on others than on themselves. In other words, individuals may
anticipate that the media will make the least impact on “me” (the first person), a moderate impact
on “you” (the second person; i.e., close friends or immediate family), and the greatest impact on
“them” (the third person; i.e., distant relatives or general friends) (Wan, Faber, and Fung 2003).

Some researchers have begun to study how the use of SNWs is bringing changes to
communication patterns and interpersonal relationships (Byne 2007; Hargittai 2007; Humphreys
2007) but to the best of our knowledge, none have examined how users perceive the effect of
SNWs. Given the central role of users in the context of SNWs, the current research employs the
theoretical framework of the TPE (Davison 1983) to examine users’ perceptions of SNWs’ effect
on themselves and on other users. TPE hypothesizes that individuals perceive media to have
different effects on themselves versus on others (i.e., the perceptual component) and that such
perceptual discrepancy could lead to certain outcomes (i.e., the behavioral component): either
their own attitude or behavioral changes (Gunther and Storey 2003) or other communicative
actions (Sun, Shen, Pan 2008b). As word-of-mouth communication via SNWs holds appealing
marketing potential, studying the indirect effects of SNWs through users’ perceptions has
practical implications for marketers as well.

This study examines the TPE hypothesis with a SNW context. Based on the analysis of survey
data, this study explores the following key questions:

1. whether TPE exists in the SNW context and if it does, how the TPE in the SNW context
differs from traditional media context;
2. how such perceptual bias varies with different referent “others”;
3. and, finally, whether such perceptual differentials potentially lead to behavioral
outcomes.
Third-Person Effect and Social Networking Websites
TPE has been demonstrated in a wide range of media contexts, including pornography (Wu and
Koo 2001), public relations (Elder, Douglas, and Sutton 2006), advertising (DeLorme, Huh, and
Reid 2007), marketing (Grier and Brumbaugh 2007), and new media (Banning and Sweetser
2007). The TPE, together with other forms of perceptual phenomenon widely studied in social
psychology, such as “naïve realism” (Pronin, Gilovich, and Ross 2004), is a psychological
tendency to perceive oneself as more intelligent, objective and less prone to bias and error in
reaction to the surrounding world. Though no prior study has examined TPE in the context of
SNWs specifically, theoretical explanations for the occurrence of TPE proposed in previous
research, as reviewed below, suggest that such perceptual discrepancy can exist in the context of
SNWs as well.

One widely accepted explanation is that the perceptual bias results from a downward social
comparison process. Social comparison is a prevalent psychological process, as individuals tend
to use others as a yardstick to determine their position in society, derive self-conception, acquire
self-esteem, and make sense of themselves in their surroundings (Park and Salmon 2005).
According to the social comparison theory (Festinger 1954), such comparison is often
downward, as individuals are inclined to think that they are smarter, better off, and more
fortunate than most other people. When people evaluate the potential effect of media, they also
estimate themselves as much more resistant to media influence (Park and Salmon 2005). SNWs
provide an environment conducive to social comparison. The networking experience provides a
platform for users to compare against each other, and confirm or even enhance their self-identity.
For example, someone is depressed about not finding a date. When the person compares his or
her dating experience with peers in a downward fashion via SNWs, connecting to others with
similar anxiety about dating may alleviate the person’s depression or even enhance his or her
confidence in continuing to look for a date. Given the innate tendency for downward
comparison, it is plausible that SNW users will also regard themselves as more intelligent,
capable users than other people.

Another explanation for TPE is ego-involvement. Perloff (1989) proposes that greater ego-
involvement with an issue tilts audiences toward greater self-other perceptual bias. Twenge
(2006) found that young users of user-generated media are narcissistic and ego-involved in
general. She also found that the more someone blogs, the more that person becomes ego-
involved and perceives his or her blog to be important (Gale 2007). Banning and Sweester
(2007) also argued that ego involvement accounts for the TPE on blogs and found strong
support. In SNWs, many user-generated discussions and statements, consciously or
unconsciously, are driven by and serve to strengthen one’s ego involvement (Gale 2007).
Therefore, it is reasonable to believe that TPE also exists with SNW users. One other related
explanation is that people display stronger TPE when they believe they have more expertise on
certain issues, especially subjective matters such as personal opinions on the choice of color for
clothing (Salwen and Driscoll 1997). This “self-knowledge” explanation highlights individuals’
cognitive propensity to regard themselves as more knowledgeable and better informed and thus
deny their susceptibility to influence (Salwen and Driscoll 1997; Salwen et al. 2002). The
subjective matters are defined as subjective experiences, likes or dislikes, views, and ideas.
SNWs are, in large part, structured with subjective matters about users. Every user has a personal
opinion about those subjective matters and presumes that he or she knows better than others e.g.
“I think Jessica Alba’s nose is fake!” “I think Jessica Alba’s nose is real!” (Quotes from users’
comments on Jessica Alba in Facebook). TPE has been shown to be a robust phenomenon across
numerous media contexts (see the recent meta-analysis by Sun, Pan, and Shen 2008a). Based on
the above theoretical explanations and previous empirical findings, the current research proposes
that SNW users may also display self-other perceptual bias. Moreover, given the preceding
discussion that SNWs offer a more conducive setting for social comparison, with users mostly
being ego-involved youngsters and user-generated content mostly being subjective, the
magnitude of the TPE on SNWs may be stronger than traditional media. Thus, the following
hypotheses are posited:

H1A: Respondents tend to believe that SNWs have a greater impact on others than on
themselves.

H1B: Respondents tend to display a greater third-person effect on SNWs than on traditional
media.

Different Referent “Others” and Third-Person Effect

Social Distance Corollary

Self-other perceptual bias does not stay the same across different conditions. One important
influencing factor that has been discussed in previous literature is social distance between self
and the referent others, as summarized in social distance corollary (Cohen et al. 1988). The
greater the social distance, the larger the TPE ( Duck and Mullin 1995; Elder, Douglas and
Sutton 2006). The theoretical reason for this conclusion is that people usually perceive socially
close others as smart and as knowledgeable as themselves (i.e., husband) while socially distant
others are always more gullible and susceptible. A number of studies found that TPE increases as
“others” vary from socially close to socially distant (Banning and Sweetser 2007; Elder et al.
2006; Paek et al. 2005; Salwen et al. 2002; Wan Faber and Fung 2003). Cohen et al. (1988)
suggested a linear increase in the TPE magnitude as the hypothetical others were changed from a
relatively small, specific, and close circle of people to a large, vague, and distant group (i.e.,
other Stanford students versus other Californians versus public at large). Gibbon and Durkin
(1995) reported that the TPE increased as “others” grew more socially distant from family
members, to neighbors, to members of the state, and to others in general. Wan et al. (2003) also
reported that respondents tended to overestimate the influence of thin female models in
advertising on distant others as compared to friends and romantic partners. As the relationship
between social distance and TPE predicts, the following hypothesis is presented:

H2: Respondents tend to display a greater third-person effect on SNWs when “others” are
defined as distant from self (i.e., most college men; most college women) than when “others” are
defined as close to self (i.e., female friends; male friends; siblings or cousins, the important
other).

Gender
The magnitude of the TPE is also found to vary with the gender of perceived others. Duck, Terry
and Hogg (1995) found that respondents of both genders perceived women as more influenced
by ads than men; when comparing self to women, a stronger TPE emerged (cited in Wan et al.
2003). Lo and Wei (2002) indicated that females were perceived to be more daunted by the 9/11
tragedy than males and therefore respondents displayed a stronger TPE when others were
described as females.

Sociolinguistic theory suggests that men and women have different underlying communication
objectives. Women’s communication objectives are mainly cooperation, social support, and
networking (Kilbourne and Weeks 1997), while men’s chief communication goals are
characterized as increasing social standing (Tannen 1995). Gefen and Ridings (2005) found that
women generally give and seek social support online. Awad and Ragowsky (2008) found that
women prefer responsive online mechanisms more than men because the responsiveness from
others may increase a sense of social support. Anecdotes also showed that women like to reach a
circle of friends in order to gain social support. Hence, women may be perceived as ideal SNW
users and susceptible to the influence of SNWs. Therefore, the following hypothesis is proposed:

H3A: Respondents from both genders tend to display a greater third-person effect on SNWs
between self and perceived female “others” (i.e., female friends; most college women) than
between self and perceived male “others” (i.e., male friends; most college men).

A number of studies have found that female respondents perceived males as more influenced by
controversial media content such as violence and pornography (Andsager 1992; Youn, Wan, and
Faber 2001; Lo and Wei 2002). These findings reflect that the gender of a respondent may be
related to the size of the TPE as well. Men enjoy the “in control” feeling during their online
activities (Awad and Ragowsky 2008). As compared to female respondents, male respondents
may perceive themselves to have more control over their online activities such as SNW use and
be less “controlled” by SNWs. Male respondents may also perceive females to be more
susceptible to the influence of SNWs because females may enjoy the networking experience
from SNWs and let the networking experience affect their own attitudes or behaviors. Hence,
male respondents may perceive a larger TPE between self and perceived female “others” than
female respondents. Hence, the following hypothesis is raised:

H3B: Male respondents tend to display a greater third-person effect on SNWs between self and
perceived female “others” (i.e., female friends; most college women) than female respondents.

On the contrary, female respondents may not overestimate perceived male others. Female
respondents may even display a reversed TPE between self and perceived male others, with self
being more affected by SNWs than male others. Therefore, the following hypothesis is
presented:

H3C: Female respondents tend to display a smaller third-person effect on SNWs between self
and perceived male “others” (i.e., male friends; most college men) than male respondents.
Female respondents may even perceive self to be more affected by SNWs than perceived male
“others”.
Behavioral Consequences of Third Person Perception
TPE, although an intriguing perceptual phenomenon in itself, has practical importance when
such a psychological bias leads to actions. A recent meta-analysis by Xu and Gonzenbach (2008)
calls for an expansion of the behavioral consequence spectrum of the TPE. Previous literature
centered on the behavioral consequence of the TPE in negative media or mediated messages such
as pornography, violence, and gambling advertising, and linked TPE to the behavioral intention
to support censorship (i.e., Chia, Lu, and McLeod 2004; Lo and Wei 2002; Xu and Gonzenbach
2008; Yang 2005). The motives for respondents to indicate pro-censorship intentions were that
respondents perceived others to be vulnerable and respondents should take actions to protect
others from harmful media content (Yang 2005).

Little is known, however, about the perceptual bias-behavior linkage for non-harmful message
topics (Park and Salmon 2005). Sun et al. (2008b) provided a theoretical explanation to the
linkage between TPE and behavioral intentions in different contexts of social influence. The
perceived social influence of messages can be organized into three abstract levels: socially
undesirable, socially desirable, and mixed or ambiguous. In respect to socially undesirable
messages, a typical TPE should emerge in that others are perceived to be more gullible to those
messages. The presumed TPE will lead to restrictive behaviors such as regulating Internet
pornography. For media messages with desirable social influence, a reversed TPE may emerge
as individuals tend to show smartness by appearing to be vigilant of desirable media messages.
Promotional behaviors such as advocating for blood donation are assumed to come forward in
reaction to the reversed TPE. Media messages with vague or mixed social influence will also
first trigger a TPE as a reflection of perceptual disparity tendency and then lead to corrective
behaviors such as removing some erotic scenes from a reality TV show but not necessarily
boycotting the show. Anecdotes indicate that most media or mediated messages fall into the
category with vague or mixed social influence. The perceived overall social influence of SNWs
is assumed to be mixed or vague.

Two studies which examined the behavioral consequence of the TPE on media or mediated
messages with mixed or vague social influence found that the TPE negatively predicted
behavioral intentions such as supporting the company, preferring the company’s products, or
voting in a presidential election (Park and Salmon 2005; Banning 2006). Their findings may be
related to Sun et al.’s (2008b) proposition that media or mediated messages with mixed or vague
social influence should lead to corrective behaviors. For example, when respondents regard the
presidential election news as neither socially desirable nor undesirable, the TPE still emerges and
other people are assumed to turn out to vote. Such a perception leads to respondents’ indication
of corrective behaviors such as not turning out to vote in order to counterbalance others’ turning
out. The corrective behavior intention is usually embodied in respondents’ indication of not
performing the given task (e.g., I will not turn out to vote) rather than restricting a certain
incident (e.g., I will vote for the candidate who is not favored by news reports).

Banning (2006) provided another explanation. He inferred that when the election news was
perceived to be non-harmful and mixed or vague in social influence, respondents might think
that susceptible other people were likely to do the right thing. In this situation, instead of taking
actions to protect others, respondents may indicate behavioral intentions that are different from
presumed others. For example, when massive others are presumed to vote, I stay home, showing
some individuality or uniqueness. Such a corollary may be plausible for the dominant
individualistic culture in the U.S. (Perloff 2002).

Researchers note that the behavioral consequence of the TPE should be related to others. In other
words, when respondents consider how to act, the perception of others should constitute a serious
motivational factor (i.e., Tewksbury, Moy, and Weis 2004; Jensen and Hurley 2005). Word-of-
mouth (WOM) communication usually invites thoughts on others. Some people engage in WOM
communication to seek other people’s opinions on products or companies. Other people first
consider the characteristics of potential WOM message receivers and then deliver WOM
messages to the relevant receiver (Allsop, Bassett, and Hoskins 2007). This study employs two
behavioral intentions related to WOM communication via SNWs.

Based on Sun et al.’s (2008b) proposition of corrective behavior and Banning’s (2006) corollary
of individuality maintenance, it is predicted that there is a negative relationship between the TPE
and behavioral intentions related to WOM communication via SNWs. Therefore, the following
hypotheses are developed:

H4A: Respondents with a higher degree of third-person effect are less likely to visit SNWs
recommended by friends.

H4B: Respondents with a higher degree of third-person effect are less likely to search out and
discuss gift ideas through social networking Websites.

Method
Participants

Survey questionnaires were administered via an online participant pool to a convenient sample of
245 undergraduate students enrolled at a major southwestern university. In return for their
participation, respondents received course credit. The use of a student sample was deemed
acceptable given that SNWs largely attract college age adults with as much as 85 percent of
college students logging into SNWs weekly (Arrington 2005). Of the 245 respondents, 58
percent are female and 42 percent are male, with an average age of 20.6.

On average, respondents spent 2.86 hours on SNWs on a typical day. Respondents were required
to identify which of the top ten SNWs they have visited in the last six months or for which they
have a user account. Ninety-seven percent of the respondents visited Facebook in the last six
months and 93 percent of the respondents had a Facebook user account. Eighty-seven percent of
the respondents visited MySpace in the last six months and 53 percent of the respondents had a
MySpace user account. More than one third of the respondents also went to Xanga in the last six
months and about 11 percent of the respondents had a Xanga user account. Less than 10 percent
of the respondents indicated having visited or having a user account with other top SNWs. The
results are very consistent with Hitwise’s 2007 survey results of general SNW users. According
to Hitwise (2007), which is an online tracking research company that monitors trends, about 91
percent of general American SNW users go to MySpace and Facebook.
Measures
To assess respondents’ TPE towards SNWs, parallel questions about media effects on self vs.
others were asked. The first question asked how much a respondent thought SNWs in general
have affected self, and the following questions asked how much the respondent thought SNWs in
general have affected different referent others. For each question, respondents indicated the
degree of being affected on an 11-point scale bounded by “none=0” and “greatest possible
effect=10”. Questions were asked about six specific others: their female friends, male friends,
current/past important other, sibling(s), most college women, and most college men (Wan et al.
2003). For the purpose of testing hypothesis H2, scores on four specific others (female friends,
male friends, the important other, siblings) were averaged to create an indicator of the perceived
impact of SNWs on “close others” (Cronbach’s alpha=.81). Similarly, the estimates on “most
college women” and “most college men” was averaged to obtain the perceived effect on “distant
others” (Cronbach’s alpha=.87).

Respondents were asked to estimate the impact of TV, print media, and radio on one’s self and
others. The self scores for all traditional media (TV, print media, and radio) were averaged to
create the estimate of the effect of traditional media on self (Cronbach’s alpha=.74). A set of
averaged others scores for traditional media was obtained through the same procedure
(Cronbach’s alpha=.75).

To examine the behavioral consequences related to word-of-mouth communication via SNWs,

respondents were asked to indicate their agreement with the following statements: “Even if my
friends start to visit a newly invented social networking website, I won’t begin to visit that
website” and “I search out and discuss gift ideas through social networking websites.” Both
questions are on 11-point scales (strongly disagree=0 to strongly agree=10). The first measure
was reversely coded in statistical analyses.

Results
H1A stated that respondents tend to believe SNWs have a greater impact on others than on
themselves. To test this hypothesis, a paired t-test was computed. A strong TPE (greater impact
on others than self) emerged (t=10.37, df=244, p <.001; Impact on others: M=7.55, N=245, SD
=2.18; Impact on self: M=6.11, N=245, SD=2.64). Thus, H1A was supported.

The differences between self and different referent others were compared by using a one-way
repeated measure ANOVA. The overall F value was significant [F (6, 184) =50.48, p <.001]. The
post hoc pairwise comparisons of the one-way repeated measure ANOVA indicated that the TPE
emerged when others were described as female friends, college women, and college men, while
the differences between the impact of SNWs on self and on male friends, sibling(s), or the
important other were not significant (Table 1A).

H1B predicted SNWs display a greater TPE than traditional media and was supported by a
repeated measure ANOVA [F (3, 242) =2.88, p <.02] (Table 1B). Furthermore, the post hoc
pairwise comparisons of the repeated measure ANOVA indicated that SNWs displayed a greater
TPE than print media and radio but not TV (Table 1B).

H2 posited that when others are socially closer to respondents (i.e., female friends, male friends,
the important other, siblings), the TPE index will be smaller than that when “others” are phrased
in a more socially distant term (i.e., most college men and most college women). Self, close
others, and distant others were compared using repeated measure ANOVA. The result supported
H2. As shown in Table 2, the respondents perceived socially distant others to be more
susceptible to the influence of SNWs than close others [F (2, 188) =49.30, p <.001]. Moreover,
the polynomial analysis of the repeated measure ANOVA suggested a significant linear increase
in the TPE magnitude as the perceived others were changed from close others to distant others [F
(1, 189) =58.20, p <.001]. However, the polynomial analysis also suggested a weak possibility of
quadratic pattern [F (1, 189) =4.58, p <.05]. Figure 1 presents the pattern of social distance
corollary.

H3A proposed that respondents from both genders tend to display a greater TPE on SNWs
between self and perceived female “others” (i.e., female friends; most college women) than
between self and perceived male “others” (i.e., male friends; most college men). H3B stated that
male respondents tend to display a greater TPE on SNWs between self and perceived female
“others” than female respondents. H3C contended that female respondents tend to display a
smaller TPE on SNWs between self and perceived male “others” than male respondents. Female
respondents may even perceive self to be more affected by SNWs than perceived male others. A
two-way repeated measure ANOVA was computed, with the gender of perceived others (i.e.,
female friends; most college women, male friends; most college men) as a within-subject
variable and the gender of a respondent as a between-subject factor. The overall F value was
significant [F (1, 241) =230.68, p <.001]. The TPE between self and perceived female “others”
was significantly larger than that between self and perceived male “others”. H3A was supported.
The interaction (the gender of a respondent and perceived gender of others) was insignificant [F
(1, 241) =1.77, p =.185]. The between-subjects effect (the gender of a respondent) of the two-
way repeated measure ANOVA was significant [F (1, 241) =15.23, p <.001]. Male respondents
displayed a greater TPE towards SNWs between self and perceived female others and between
self and perceived male others than female respondents. H3B was supported. H3C was partially
supported. Although female respondents perceived a very small gap between self and perceived
male others (M of the TPE=.10), the result did not indicate a reversed TPE, with self being more
affected by SNWs than perceived male others (Table 3).

H4A and H4B suggested that people with a greater level of the TPE are less likely to visit a new
SNW recommended by friends and to search out and discuss gift ideas through social networking
Websites. Two multiple regressions were run with respondents’ gender and average hours spent
on SNWs on a typical day as control variables. Tables 4A and 4B present the results. Both
predictions were supported: TPE was a significant predictor of the behavioral outcome in each
case [H4A: F (3, 241) =5.86, p <.01; H4B: F (3, 241) =4.63, p <.01]. Neither respondents’
gender nor average hours spent on SNWs in a typical day was related to respondents’ behavioral
intentions. The TPE did negatively affect respondents’ behavioral intentions related to word-of-
mouth communication via SNWs. H4A and H4B were both supported.
Implications, Future Research Suggestions, and Limitations
H1A, which stated that respondents tend to believe that SNWs have a greater impact on others
than on themselves, was supported. The psychological discrepancy of perceiving self as less
susceptible and others as more susceptible to the media influence was confirmed in the SNW
context for the first time. Future research needs to validate whether the psychological
discrepancy is caused by SNW users’ ego involvement, self-knowledge, and/or downward social
comparisons with others.

It is important for researchers and marketers to understand TPE since the construct is firmly
grounded in psychology and investigating the phenomenon may contribute to consumer
psychology. Moreover, though analyzing the direct influence of SNWs is important, this study
provides another research avenue aimed at examining the indirect influence of SNWs through
users’ perceptions on how much others are impacted by the media. Such an indirect influence is
noteworthy and this study addresses the literature gap.

The TPE is assumed to be salient when the target consumers are young people who are often
ego-defensive or even narcissistic. There are many such audiences in the US market, according
to the book “Generation Me” (Twenge 2007), which profiles American young people born in the
70s, 80s, or 90s. The phenomenon of the TPE deserves attention from marketers, particularly
those who market to young people via SNWs.

The TPE may create an illusion of being in control, because as compared to others, a user may
feel smarter and less susceptible. Although word-of-mouth communication occurs between SNW
users, marketers should build a well-planned campaign to stimulate word-of-mouth
communication among users. Marketers may consider using certain words, phrases, and/or
pictures acting alone or in combination in their campaigns to augment the TPE. It may prompt an
SNW users’ in control feeling, which is conducive to stimulating WOM communication. For
example, Sears stimulates female Facebook users to generate word-of-mouth about the prom
dresses offered by Sears. Instead of pursuing each individual female Facebook user, Sears invites
all female Facebook users to “help others on their prom dress choice from Sears”. An individual
user may think “I know exactly what kind of prom dress I want, but maybe I can help others pick
the right dress since they may be more influenced by marketers such as Sears.”

H1B, which predicted SNWs display a greater TPE than traditional media, was supported. The
irony is that as compared to traditional media, though user-generated media involve many user
interactions, users are even less sure of how much others are really affected by SNWs and tend to
overestimate others’ responsiveness to the media at a great level. The explanation could be that
although there is a high degree of interactions among SNW users, the interaction topics are too
diversified to accurately estimate the influence.

Interestingly, respondents indicated that they were influenced by TV and SNWs and they
perceived others to be even far more affected by those two media types. Respondents reported
themselves not to be much influenced by print media and radio, and they assumed others to be
only slightly more affected by print media and radio than themselves. This pattern supports the
proposition of some researchers that the more popular the media or messages are, the greater the
TPE (i.e., Eichholz 1999), although the explanation for such a positive correlation is lacking in
the literature and in this study. It awaits future study to explore. Some Facebook campaigns
aimed at stimulating WOM communication provide popular products such as iPhones or iPods as
rewards. Ties into popular products/brands may create attention and build a strong third-person
effect as well in the SNW context. It is worthy of consideration for marketers.

H2, which posited that the TPE will be smaller when others are socially close than when others
are socially distant, was also supported. The TPE is stronger with a large group of perceived
“others”. Theoretically, the TPE research provides a new avenue for scholars to understand
perceived group influence on individuals. For marketers, electronic WOM communication no
longer occurs between a few people. A SNW user may post a word-of-mouth message on a
discussion board for many to see. Therefore, the size of the TPE in electronic word-of-mouth
communication can be leveraged by choosing the suitable communication mode ranging from
one-to-one (only take close others into consideration) to one-to-many (take distant others into
consideration as well).

The confirmation of H3A, H3B, and H3C indicates that the TPE does vary with the gender of
perceived others and the gender of a user in the SNW context. Specifically, perceived female
others amplify the TPE and the TPE is more prominent with male users. The results suggest that
the gender variable should be an important control variable for future TPE research on SNWs.
Although males exhibit a greater TPE toward females in an SNW, it is unclear if this persists
with particular consumer messages. For example, brands such as Victoria’s Secret stimulate
SNW users to generate WOM communication about Valentine’s Day gifts. In this case, it is still
unclear which gender will be perceived as more affected and whether there is an interaction
between the gender of the perceived others and the gender of a user. This study provides a
baseline for future research to consider placing the gender variable on TPE research in a specific
SNW context. Marketers and researchers should also attempt to relate the gender impact on the
TPE to consumption behavior (i.e., Will males perceive females to be more affected by a
Victoria’s Secret Valentine’s Day advertisement and be encouraged to purchase the advertised
item?)

The results of H4A and H4B were consistent with Sun et al.’s (2008b) and Banning’s (2006)
theoretical propositions and empirical findings of previous studies (i.e., Banning 2006; Park and
Salmon 2005). Researchers and marketers should bear a few things in mind to link the perceptual
bias to behavioral consequences in the SNW context. First, in order to trigger the TPE and relate
it to behavioral consequences, the behavioral intention should be other-regarding; WOM
communication behavior complies. Second, respondents’ average hours spent on SNWs on a
typical day were insignificant to predict their behavioral consequences. It indicates that for some
behaviors that invite the perception of others’ attitudes or behaviors, TPE, rather than individual
involvement or media use, is the driving force of those behaviors. Respondents did consider the
TPE before forming a behavioral intention related to WOM communication.

Finally, though respondents did not indicate the behavioral intentions in a positive direction, the
negative relationship between the TPE and the behavioral consequences, to some degree,
reflected Sun et al.’s (2008b) and Banning’s (2006) theoretical propositions. The general social
influence of SNWs is assumed to be ambiguous or mixed (i.e., a TV reality show. Respondents
indicated corrective behaviors of not engaging in WOM communication via SNWs in order to
rectify the perceived phenomenon that other users are more affected by SNWs and others will by
and large engage in WOM communication via SNWs. Future studies should first measure
whether the general social influence of SNWs is perceived as ambiguous or mixed, and then
experiment to build a causal relationship between the ambiguous or mixed social influence and
corrective behaviors in relation to WOM communication in the SNW context. Though this study
echoes Sun et al.’s (2008b) theoretical propositions and provides a possibility of corrective
behavior in relation to WOM communication, marketers should consider the findings with the
caution that a specific campaign in SNWs may have a positive or negative social influence and
therefore may elicit restrictive or promotional behaviors in relation to WOM communication.
Future studies should focus on the relationship between the positive social influence of a
campaign and promotional behaviors in relation to word-of-mouth communication, which would
directly benefit marketers.

Another theoretical explanation for the negative relationship between the TPE and the behavioral
consequences in relation to word-of-mouth communication is Banning’s (2006) individuality
maintenance. Future research should also attempt to build a causal relationship between his
corollary and the behavioral consequences. If his corollary is a significant causal factor of the
negative relationship, the research on TPE and its behavioral consequences may be more
valuable in a collectivistic cultural background, as people with that background are less sensitive
to individuality maintenance. In the U.S. market, marketers may either target SNW users with
less tendency to maintain individuality or emphasize the smartness or uniqueness of others
partaking in word-of-mouth communication in order to rectify the negative relationship between
the TPE and the behavioral consequences.

Limitations
A few limitations of this study are noteworthy. This study is based on a convenient college
student sample. The results and implications for this study should be restrained to college student
SNW users. However, the most recent meta-analysis by Sun et al.’s (2008a) suggests that neither
a non-probabilistic sample nor the student population significantly affects the findings of the
TPE. This study only supported a strong TPE on SNWs in general (see Table 5). The TPE on
specific commercial content in SNWs, such as brand profile pages, may vary from the findings
in this study. Although the findings of the TPE magnitude differences between SNWs and
traditional media are interesting and worth further exploring in future studies, one shortcoming
of this study is that the comparison is on the general level of media types (e.g., SNWs versus
TV). Though previous studies also provide similar cross-media comparisons at the general level,
a comparison of specific media content (e.g., a SNW banner versus a TV commercial) may
generate more direct implications for marketers and researchers. Both Sun et al.’s (2008b)
theoretical propositions and Banning’s (2006) corollary of individuality maintenance are
theoretically assumed rather than measured in this study. Therefore, this study cannot provide
empirical evidence of the relationship between the theoretical assumptions and the negative
relationship between the TPE and the behavioral consequences related to word-of-mouth
communication. Future studies should directly measure these theoretical assumptions. Although
the survey method in this study helps identify variables worthy of consideration in research on
TPE in the SNW context and provides exploratory relationships between variables, future studies
should employ the experimental method to complement the findings of this survey study and
build causal relationships between variables.

References
Allsop, D.T., B.R. Bassett, and J.A. Hoskins. December 2007. Word-of-mouth research:
Principles and applications. Journal of Advertising Research: 398-411.

Andsager, J.L. 1992. Differentiation media practices: how demographic variables relate to
support for freedom of the press. Paper presented at the Association for Education in
Journalism and Mass Communication, Montreal.

Arrington, M. September 7, 2005. 85 percent of college students use Facebook.

http://www.techcrunch.com/2005/09/07/85-of-college-students-use-facebook/.

Awad, N.F. and A.Ragowsky. 2008. Establishing trust in electronic commerce through online
word of mouth: An examination across genders. Journal of Management Information Systems
24(4):101-121.

Banning, S.A. 2006. TPEs on political participation. Journalism & Mass Communication
Quarterly 83(4):785-800.

________ and K.D. Sweetser. 2007. How much do they think it affects them and whom do they
believe?: Comparing the TPE and credibility of blogs and traditional media. Communication
Quarterly 55(4):451-466.

Boyd, D.M. and N.B. Ellison. 2007. Social network sites: Definition, history, and scholarship.
Journal of Computer-Mediated Communication 13(1):article 11.
http://jcmc.indiana.edu/vol13/issue1/boyd.ellison.html.

Byrne, D.N. 2007. Public discourse, community concerns, and civic engagement: Exploring
black social networking traditions on BlackPlanet.com. Journal of Computer-Mediated
Communication 13(1):article 16. http://jcmc.indiana.edu/vol13/issue1/byrne.html.

Chia, S.C., K. Lu, and D.M. McLeod. 2004. Sex, lies, and video compact disc: A case study on
third-person effect and motivations for media censorship. Communication Research 31(1):109-
130.

Cohen, J., D. Mutz, V. Rice, and A. Gunther. 1988. Perceived impact of defamation: An
experiment on third-person effect. Public Opinion Quarterly 52:161-173.

Davison, W.P. (1983). The TPE in communication. Public Opinion Quarterly 40:1-15.

DeLorme, D.E., J. Huh, and L.N. Reid. 2007. Others are influenced, but not me: Older adults’
perceptions of DTC prescription drug advertising effects. Journal of Aging Studies 21(2):135-
151.
Duck, J.M., D.J. Terry and M.A. Hogg. 1995. The perceived influence of AIDS advertising:
Third-person effects in the context of positive media content. Basic and Applied Social
Psychology 17: 305-325.

Duck, J.M. and B. Mullin. 1995. The perceived impact of the mass media: reconsidering the
third-person effect. European Journal of Social Psychology 25: 77-93.

Eichholz, M. 1999. Hierarchy of communication effects: The perceived persuasive power of

public relations. Association for Education in Journalism and Mass Communication.

Elder, T.J., K.M. Douglas, and R.M. Sutton. 2006. Perceptions of social influence when
messages favour ‘us’ versus ‘them’: A closer look at the social distance effect. European
Journal of Social Psychology 36(3):353-365.

Festinger, L. 1954. A theory of social comparison processes. Human Relations 7: 117-140.

Gale Reference Team. 2007. Narcissistic, but in a good way. The Advocate. 983:6.

Gefen, D. and C. Ridings. 2005. If you spoke as she does, sir, instead of the way you do: A
sociolinguistics perspective of gender differences in virtual communities. DATABASE for
Advances in Information Systems 36(2):78-92.

Gibbon, P. and K. Durkin. 1995. The TPE: Social distance and perceived media bias. European
Journal of Social Psychology 25:597-602.

Grier, S.A. and A.M. Brumbaugh. 2007. Compared to whom? The impact of status on third
person effects in advertising persuasion in a South African context. Journal of Consumer
Behaviour 6:5-18.

Gunther, A.C. and J.D. Storey. 2003. The influence of presumed influence. Journal of
Communication June: 199-215.

Hargittai, E. 2007. Whose space? Differences among users and non-users of social network sites.
Journal of Computer-Mediated Communication 13(1):article 14.
http://jcmc.indiana.edu/vol13/issue1/hargittai.html

Hitwise.com. March 14, 2007. iMeem and Bebo are the fastest movers.
http://www.hitwise.com/press-center/hitwiseHS2004/socialnetworkingmarch07.php .

Humphreys, L. 2007. Mobile social networks and social practice: A case study of dodgeball.
Journal of Computer-Mediated Communication 13(1):article 17.
http://jcmc.indiana.edu/vol13/issue1/humphreys.html.

Kilbourne, W. and S. Weeks. 1997. A socio-economic perspective on gender bias in technology.

Journal of Socio-Economics 26(1):243-260.
Jensen, J.D. and R.J. Hurley. 2005. Third-person effects and the environment: Social distance,
social desirability and presumed behavior. Journal of Communication 55:242-256.

Lo, V.H. and R. Wei. 2002. Third-person effect, gender, and pornography on the internet.
Journal of Broadcasting & Electronic Media 46(1):13-33.

Paek, H., Z. Pan, Y. Sun, J. Abisaid, and D. Houden. 2005. The third-person effect as social
judgment: An exploration of social distance and uncertainty in perceived effects of political
attack ads. Communication Research 32(2):143-170.

Park, H.S. and C.T. Salmon. 2005. A test of the third-person effect in public relations:
Application of social comparison theory. Journalism & Mass Communication Quarterly
82(1):25-43.

Perloff, R.M. 1989. Ego-involvement and the third-person effect of televised news coverage.
Communication Research 18:236–262.

______ 2002. The third person effect. In Jennings Bryant and Dolf Zillmann, Eds., Media
effects: Advances in theory and research, Mahwah, NJ: Lawrence Erlbaum Associates.

Pronin, E., T.D. Gilovich, and L. Ross. 2004. Objectivity in the eye of the beholder: Divergent
perceptions of bias in self versus others. Psychological Review 111: 781-799.

Salwen, M.B. and P.D. Driscoll. 1997. Consequences of third-person effect in support of press
restrictions in the O.J. Simpson trial. Journal of Communication 47:60-77.

Salwen, M.B., P.D. Driscoll, B. Garrison, R. Abdulla, K. Campbell, D. Casey. 2002. Third-
person effects of fear during the war on terrorism: Perceptions of online news users. Association
for Education in Journalism and Mass Communication.

Sun, Y., Z. Pan, and L. Shen. 2008a. Understanding the third-person perception: Evidence from a
meta-analysis. Journal of Communication 58:280-300.

Sun, Y., L. Shen, and Z. Pan. 2008b. On the behavioral component of the third-person effect.
Communication Research 35:257-278.

Tannen, D. 1995. The power of talk: Who gets heard and why. Harvard Business Review
73(5):138-148.

Tewksbury, D., P. Moy, and D.S. Weis. 2004. Preparations for Y2K: Revisiting the behavioral
component of the third-person effect. Journal of Communication 54:138-155.

Twenge, J.M. 2007. Generation me: Why today’s … New York. Free Press.
Wan, F., R.J. Faber, and A. Fung. 2003. Perceived impact of thin female models in advertising:
A cross-cultural examination of third person perception and its impact on behaviors. Asia Pacific
Journal of Marketing and Logistics 15(1/2):51-74.

Wu, W. and S.H. Koo 2001. Perceived effects of sexually explicit Internet content: The third-
person effect in Singapore. Journalism & Mass Communication Quarterly 78(2):260-274.

Xu, J. and W.J. Gonzenbach. 2008. Does a perceptual discrepancy lead to action? A meta-
analysis of the behavioral component of the third-person effect. International Journal of Public
Opinion Research 20(3): 375-385.

Yang, K.C. 2005. Consumers’ attitudes toward regulation of Internet auction sites: A third-
person effect perspective. Internet Research 15(4):359-377.

Youn, S., F. Wan, and R.J. Faber. 2001. We should censor because they are vulnerable:
Censorship of controversial web sites and the third-person effect. In Taylor, C. R. (Ed.), The
Proceedings of the 2001 Conference of The American Academy of Advertising, The American
Academy of Advertising: 72-81.

http://www.bsu.edu/mcobwin/majb/?p=594 6 dec 2010