Facebook, the Third-Person Effect, and the Differential Impact Hypothesis


  • Accepted by previous editor Maria Bakardjieva

Correspondence To: Valarie Schweisberger, vnschwei@syr.edu


This study examined the effects of embedding and framing online news stories in social media contexts on perceived message influence and third-person effects (3PE). 88 undergraduates at a Northeastern U.S. university participated in an online experiment in which they evaluated news stories posted on Facebook. A 4 x 2 mixed experimental design was used with the between-subject variables of viewing condition (no Facebook frame, neutral Facebook, positive Facebook evaluation, and negative Facebook evaluation) and the within-subjects factor of story relevance (Low, High). Results indicate that perceptions of personal influence increase in social media contexts for more personally relevant stories. These results are consistent with the Differential Impact Hypothesis.

The old adage says that everyone's a critic, but new media have made everyone a gatekeeper as well. An increasing number of Americans—particularly younger adults—use social media networks to access and assess news (Purcell, Rainie, Mitchell, Rosensteil, & Olmstead, 2010). A recent study conducted by the Pew Research Center found that in 2010, 75% of online news consumers reported that they received news forwarded through e-mails or social networking sites and that approximately half of those users shared links to news stories using social media.

As Facebook and Twitter icons accompanying a “share this!” message become a staple of online articles, users can link back to these articles from their social networking pages with just the click of a mouse. It is possible that presenting stories in this manner on a Facebook page may suggest an implicit recommendation on the part of the sharer; further, sharers may choose to frame the story by adding a few editorial sentences of their own. This method of sharing information raises questions about perceived effects of media content on both the self and others, and it's possible that perceived effects on self could be influenced by framing of stories.

The current study examines third-person effects (3PE) in social media environments. Audiences tend to make third-person effect judgments concerning the effects of mass media messages—including news stories—and report that those messages have a greater effect on others than on themselves (Davison, 1983). Although research has found that people make 3PE judgments concerning the perceived effects of social media (Banning & Sweetser, 2007), it is less clear whether the social media context itself may limit or reduce 3PE.

The Impersonal Impact hypothesis (Tyler & Cook, 1984) predicts that mass media sources are less effective in influencing individuals than interpersonal sources. The Differential Impact hypothesis, however, predicts that media sources can influence individuals when they are personally involving (Basil & Brown, 1997; Snyder & Rouse, 1995).

Given the “social” nature of Facebook and the quasi-interpersonal nature of the medium, it is possible that simply reading stories in a Facebook environment could increase the “interpersonal” nature of communication and increase perceived relevance—contributing to increased effects on self, and serving to limit perceived differences between self and others.

However, the medium of delivery (Facebook) may not in and of itself increase perceived effects. The personal relevance of the story and story quality are also important.

In a world where it is increasingly possible for individuals to “personalize” news selections and limit their access to news stories of personal relevance and interest, it is important to examine the relationship between the self-relevance of stories and the potential influence (or lack of influence) of social media sites on the perceived effects of relevant stories.

This experimental study attempts to determine if social media contexts—Facebook pages—affect young adults' perceptions of the effects of online news stories. We will also attempt to determine if the effects of story context are influenced or outweighed by the personal relevance of those news stories. In addition, this study examines the Third Person Effects of “framing” news stories by the addition of positive or negative comments in Facebook sites.

Third-Person Effects

The Third-Person Effect hypothesis predicts that individuals tend to think that other people are influenced by mass media messages to a greater degree than themselves (Davison, 1983). This type of relative comparison has been found in a wide range of contexts and formats (Perloff, 1999). People, however, are more likely to report that messages influence themselves if those messages are positively evaluated (in terms of message quality and content), if those messages are viewed as personally self-relevant, and if influence is seen as being socially desirable (Duck, Terry, & Hogg, 1995; Gunther & Mundy, 1993; Innes & Zeitz, 1988). These factors can diminish perceived differences in effects on self and others. In some cases, they can even create a “first-person effect” (1PE) in which individuals rate messages as having a greater effect on themselves than on other people (Duck & Mullin, 1995; Duck et al., 1995).

Until recently, studies of 3PE have focused primarily on traditional mass media messages (DeLorme et al., 2006; Golan et al., 2008; Hoffner, 2001). Third-person effects, however, have been found in a range of online contexts. Banning and Sweetser (2007) examined 3PE among blogs, news websites, and newspapers, and concluded that 3PE exists across all platforms with relative consistency.

Research on 3PE in social media contexts, however, is limited. For example, Zhang and Daugherty (2009) explored the implications of 3PE for online marketing. The authors conclude that people make 3PE judgments concerning the effects of social networking. Adolescent gamers have also been found to make 3PE judgments concerning the effects of video games (Scharrer, 2008). It is possible that social media contexts themselves, however, may influence message effects and perceived differences between self and others.

The Impersonal and Differential Impact Hypotheses

A number of studies have found that there are differences in the influence of messages depending upon whether they are delivered through mass media or through interpersonal communication (Tyler & Cook, 1984; Basil & Brown, 1997). The Impersonal Impact Hypothesis (IIH) posits that cognitive assessments of risk stemming from media messages are perceived by individuals to occur less for the individuals themselves than for society in general (Tyler & Cook, 1984). In other words, media messages influence people's perceptions of risk for society as a whole, but have a limited impact on people's perceptions of their own risk. Tyler and Cook conclude that mass communication and interpersonal communication influence perceptions of risk differently, and that mass communication is more likely to influence individuals' perception of risk to society as a whole than to themselves. In contrast, interpersonal communication can influence individuals' personal risk assessments. Thus, if a friend or acquaintance informs an individual about a potential risk, that individual may be more likely to view that information as applying to himself or herself.

The Differential Impact Hypothesis (DIH), however, proposes that media messages can have an influence on perception of personal risk if the message is personally relevant (Basil & Brown, 1997; Snyder & Rouse, 1995), and if people view themselves as having a social or parasocial relationship with the media source (i.e. celebrities). Research on both IIH and DIH confirms that identification is key in determining personal relevance (Scarberry et al., 1997).

Most of the research on the DIH has examined the potential effects of televised messages. It is less clear how individuals will respond to messages in social media contexts. Although “news stories” may have originated from mass media sources, embedding these stories in a social media context—such as Facebook—may elicit perceptions that the stories are being delivered through a more personal, or interpersonal, channel. This perception may be heightened by the addition of comments or story evaluations by the Facebook “source.”

Social Media & Framing News Story Perceptions

A recent study conducted by the Pew Research Center found that young adults (18–32) are most likely to get their news from online sources. They tend to draw on a range of different news sites and sources and 22% report that they rely on their social networks for alerts or links to news stories (Purcell et al., 2010). The majority of audience members report that they access news through social media and that they have in turn used social media to disseminate news stories. Approximately 25% of online news users report commenting on news content (Purcell et al., 2010).

These personal comments usually appear with the news stories and can serve to provide a “frame” for the story that may influence story reception. In the simplest terms, framing deals with the way items are presented in the media, rather than what items are presented (McCombs et al., 1997; Scheufele, 1999). Studies concerning framing theory have tended to focus on specific kinds of new items and their presentation via broadcast media or more traditional news sources (Pajnik, 2010; Birkland, 2009). Facebook “frames” or comments, however, can also influence perceptions of story content through interpretation, praise, or ridicule.

The latter may be of particular importance. As noted earlier, perceived social desirability can influence third-person effects. When message effects are viewed as socially undesirable (pornography, violence, etc.), then third-person effects tend to be greater (see Perloff, 1999). Negatively framing news stories through the inclusion of negative or critical comments could contribute to perceptions of the social undesirability of acknowledging influence and limit people's willingness to acknowledge message effects. It could also increase third-person effects.

Study Goals and Hypotheses

Prior research on mass media persuasive messages (Duck & Mullin, 1995; Duck et al., 1995) found that higher evaluations of message self-relevance and quality were positively associated with perceived effects of those messages on self. These factors may also be related to perceived effects of online news stories. Therefore, the first set of hypotheses predicts that:

H1a: Perceived effects of news stories on self will increase as the perceived self-relevance of news stories increases.

H1b: Perceived effects of news stories on self will increase as the perceived quality of news stories increases.

News in Facebook

If social media sites such as Facebook increase the “interpersonal” nature of communication, then the Differential Impact Hypothesis would predict that reading online news stories in social media sites should increase perceived effects of those stories on self. This in turn should diminish 3PE. Therefore, this study predicts that:

H2: Perceived effects of news stories on self will be greater if stories are read as part of a Facebook page than if those stories are read as online news stories.

H3: Third person effects will be smaller if news stories are read as part of a Facebook page than if those stories are read as online news stories.


If messages are perceived as being personally relevant, then individuals are more likely to report that those messages influenced them. This tends to diminish third-person effects. On the other hand, if messages are not viewed as relevant, than audiences are more likely to make third-person-effect judgments concerning message effects. Although social media may enhance perceived effects of messages, this is less likely to be apparent for stories that audiences view as irrelevant. The effects of relevance should outweigh the effects of message context. Thus,

H4a: Regardless of viewing condition, participants will rate Low-Relevant news stories as having a greater effect on others than on self (3PE).

H4b: Participants will be less likely to make consistent 3PE judgments for High-Relevance stories.


H5a: Negatively framing stories in a Facebook page will decrease perceived effects on self.

H5b: Negatively framing stories in a Facebook page will increase third-person effects.



This study used a 4 x 2 x 2 mixed design with between-subject variables of evaluation conditions (no Facebook frame, neutral Facebook frame, positively evaluated Facebook frame, and negatively evaluated Facebook frame) and the within-subjects factors of story relevance (Low, High) and story presentation order. In order to limit potential order effects, story presentation was varied in each of the experimental condition. In the “control” or no-Facebook condition, participants were immediately directed to two online news stories. In the “neutral” Facebook condition, participants read stories linked to a Facebook page without any evaluative comments. The “positive” Facebook condition included positive evaluations by the Facebook “person” and the “negative” Facebook condition included negative evaluations.

Stimulus Construction

Three Facebook pages were created to fulfill the condition requirements of the study. Beyond the differences described above, all the information on the Facebook pages was identical. The Facebook “person” was given a fake, androgynous name (Casey Jones), and the profile picture contained no identifying features—but depicted the university logo. Information on the site, identified the “person” as an undergraduate student with no other visible photographs besides the profile picture. Interests included school sports and a few other non gender-specific hobbies.

The articles were chosen based on a pretest conducted with 11 graduate students at a mid-size university in the Northeastern US. Our goal was to find innocuous news stories with little or no potential to elicit biases in the reader. No political articles were considered, nor were any articles about controversial topics like abortion, same-sex marriage, the health care plan, etc. The group of graduate students evaluated six articles for potential controversy and story relevance. The two least controversial articles were selected. The first story was considered to be the least relevant and focused on a bunny exhibit in a renowned museum. The second story—judged as somewhat more relevant—concerned new technologies and facilities on Disney cruise ships.


Perceived self-relevance for each story was assessed using a 3-item Likert-type scale ranging from 1 (Strongly Disagree) to 7 (Strongly Agree). Participants were asked to rate their agreement with statements assessing the perceived relevance of the news stories (i.e. “this article contains news I should know”). Scale reliability tests were conducted on combined scales for both stories. Standardized Cronbach alpha reliability scores were acceptable (α = .82). Averaged responses from participants in all conditions were calculated in order to confirm the designated level of relevance of the Low- and High-Relevance stories.

Message “quality” was assessed by asking participants to rate their agreement using a 7-point Likert-type scale to a 3-item measure (“This article is of high quality”). Standardized Cronbach alpha reliability scores were acceptable (α = .73).

Effects on self and others were assessed using indirect comparisons. Unlike direct comparisons—where individuals are asked to assess perceived effects relative to a specified “other”—effects on self were assessed separately from effects on others. This approach is commonly used in Third-Person Effect research (Conners, 2005). Participants were asked to rate their agreement with the statements “How much do you think that you are affected by the contents of this article?” and “How much do you think that the average college student would be affected by the contents of this article?” using a 7-point scale ranging from 1 = Not at All to 7 = Very Much. Prior studies of 3PE have used a variety of different comparative targets including “the average American” and the “average person” (see Perloff, 1999). As the student participants in this study could reasonably assume that there might be differences among other age groups in use and access to Facebook, we attempted to limit potential variance by selecting “the average college student” as a comparative target. 3PE scores were calculated by subtracting ratings for self from ratings for the average college student. Positive scores indicate a third-person effect (greater effects on other than self). A negative score indicates a first-person effect (greater effects on self than other).

Participants were also asked to report their use of online news and Facebook. Participants were asked to respond to a series of questions using (Yes, No) bivariate responses: “Do you typically get your news online?;” “Do you post news stories on Facebook? (This can mean any kind of news- hard news, features, tabloid news, etc.);” and “Do you click on news stories featured on your friends' Facebook pages?” Participants were also asked to use a 6-point scale measuring number of hours per week spent using Facebook ranging from 0 to 10+ hours. Participants were also asked to respond to a series of questions assessing demographic information.


Eighty-eight undergraduate students (77% Female, 23% Male) participated in this study. Ages ranged from 18 to 22 (M = 19.09, SD = .89). The majority of participants (69.5%) identified themselves as White, 15.9% as Asian, 7.3% as Latina, 2.4% as African-American, 2.4% as Native American, and 2.4% as Middle Eastern. Given the limited racial diversity of participants, race was not analyzed as a factor in this study.


Participants were recruited from several undergraduate Communications courses at a midsize university in the Northeastern US. They received extra course credit for participation. The survey was conducted online using the surveymonkey.comdata collection program. Participants were randomly assigned to the different experimental conditions and then emailed the link to the survey. After reading each story, the participants completed the evaluative measures and then were directed to the second story.


Pearson Product Moment Correlation analyses were conducted to examine the overall relationships between ratings of perceived relevance, message quality, and perceived effects on self and others. As the amount of time spent using Facebook increased, perceived differences between effects on self and the average person for the Low Relevance story decreased, r = −.34, p = .007. Therefore, in subsequent analyses of story effects on self and others, time spent using Facebook was treated as a covariate. Mixed multivariate analysis of covariance tests were conducted with the between subjects factor of viewing condition (control, neutral Facebook, positive-framed Facebook, and negative-framed Facebook) and the within subjects factors of story relevance (low, high) and rating target (self, average person). Similar analyses were conducted to examine effects of viewing condition and story relevance on 3PE difference scores. In order to test hypotheses 2 and 3, only data from the neutral Facebook and control conditions were analyzed.


Manipulation Check

This study compared responses to stories with different levels of perceived relevance. The Low-Relevance story was rated as significantly less self-relevant (M = 2.59, SD = 1.20) than the High-Relevance story (M = 3.60, SD = 1.20) by participants in all four reading conditions, F(1,78) = 48.51, p < .001, ηp2 = .38. See Table 1, Appendix A.

Table 1. News Story Relevance
 Story Relevance
Viewing ConditionLowHigh
Neutral Facebook2.503.54
Negative Facebook2.603.58
Positive Facebook2.723.85

Online News and Facebook Behaviors

All of the participants who responded to the question concerning typical source of news reported that they got their news online. Approximately 60% reported that they clicked on news stories featured on their friend's Facebook pages, although only 27% reported that they posted news stories themselves. Approximately 60% of participants reported spending between 4 to 8 hours a week on Facebook.

The first set of hypotheses assessed the relationships between perceived effects of news stories and the perceived self-relevance and quality of those stories. As the perceived self-relevance of stories increased, so did perceived effects of those stories on self as well as on others. For the High-Relevance story, although not for the Low-Relevance story, as self-relevance increased perceived differences between self and the average person diminished. See Table 2. After controlling for time spent using facebook, participants also reported greater effects on self for the High-Relevance story than for the Low-Relevance story, F(1,77) = 23.52, p < .001, ηp2 = .23. Hypothesis 1a was supported. See Table 3. Self relevance was also significantly and positively correlated with positive evaluations of the quality of those stories. Hypothesis 1b was supported. See Table 2.

Table 2. Correlations Between Story Relevance, Quality, and Effects on Self and Others
 RelevanceQualitySelfAv. Person3PE
  1. ap<.05;
  2. bp < .001
Low Relevance     
Quality 1.00.62b.41b−.03
Self  1.00.67b−.18
Average Person   1.00.61b
3PE (Av –Self)    1.00
High Relevance     
Quality 1.00.58b.38b−.25a
Self  1.00.71b−.18
Average Person   1.00.33b
3PE (Av –Self)    1.00
Table 3. Perceived Effects of News Stories on Self and Others
 Story Relevance
  1. a3PE calculated by subtracting ratings for self from ratings for the average person. Positive scores indicate a third person effect (greater effects on other than self). Negative numbers indicate a first person effect (greater effects on self than other).
Viewing Condition      
Neutral Facebook1.751.94.202.502.25−.33
Negative Facebook1.331.67−.101.671.92.30
Positive Facebook1.381.54.172.402.20−.25

News in Facebook

The next set of hypotheses predicted that perceived effects of news stories on self would be greater if stories were read as part of a Facebook page than if those stories were read as online news stories. Consistent with this hypothesis, after controlling for time spent using Facebook participants rated stories read on the neutral Facebook page as having greater effects on self than stories read in the control (non-Facebook) condition, F(1,34) = 5.55, p = .024, ηp2 = .14. Hypothesis 2 was supported.

We also predicted that 3PE self/other differences would be smaller if news stories were read as part of a Facebook page than if those stories were read as online news stories. After controlling for time spent using facebook, the effects of viewing condition alone were non-significant—possibly due to an interaction between story relevance and viewing condition, F(1,31) = 3.81, p = .06, ηp2 = .11. Although not significant at the traditional p < .05 level, the nature of self/other distinctions for High Relevance stories differed depending upon viewing condition. Participants who read the High Relevance online news story in the control condition (non-Facebook) were more likely to make third-person effect judgments (others more affected then self) whereas participants in the neutral Facebook condition were more likely to make first-person effect judgments (self more affected than others). See Table 3.


The next set of hypotheses predicted that participants would rate Low Relevant news stories as having a greater effect on others than on self—regardless of viewing condition. After controlling for time spent using Facebook, participants consistently made 3PE judgments rating effects on others as greater than effects on self, F(1,56) = 10.22, p = .002, ηp2 = .15. As Hypothesis 4a predicted, there were no significant interactions between judgment type and viewing condition. See Table 3.

We also predicted that participants would be less likely to make consistent 3PE judgments for High Relevance stories. Participants did not make significant consistent 3PE judgments for High Relevance stories. Hypothesis 4b was supported.


The last set of hypotheses addressed the effects of negatively framing Facebook stories. Participants who read negatively framed High Relevance news stories reported lower effects of those stories than did participants in any other condition and also made 3PE judgments concerning perceived effects. These differences, however, were not significant. Hypotheses 5a and 5b were not supported. See Table 3.

Discussion and Conclusion

The results of this study suggest that social media contexts can influence the perceived personal influence of news stories. Participants reportedly perceived that personally relevant stories have a greater impact on themselves than do nonpersonally relevant stories. In addition, the third-person effect is also supported by the results of our study, as participants perceived that Low-Relevance stories would have a greater impact on others than on themselves.

Limited investigation has been conducted in either DIH, IIH, or 3PE in a social media environment. We feel that these results illuminate relevant and useful trends which address a sizeable gap in the literature. The unique quality of social media to provide a space in which interpersonal interaction exists within the larger structure of a mass media interface creates a dynamic new area of inquiry using DIH, IIH, and 3EP theoretical frameworks. As the results of this study indicate, there is much to investigate regarding these theoretical groundings in social media.

This research might be said to have the greatest implications for news readers, who could use the information to more effectively evaluate news texts (both in and outside of social media environments). Readers might also think more critically about their news-sharing tendencies and behaviors. News networks, more specifically the social media directors at news networks, might also consider the results of this study in an effort to more effectively frame and share their own articles.

Limitations and Future Research

As in any study, this was not without limitations. The lack of significance for some hypotheses was likely due to a relatively limited number of participants. As this was an exclusively online experiment, we also had no control over the environment in which participants completed the survey. It is possible that their respective environments influenced the evaluations of the articles. In addition, the use of a university logo as a profile photo signified the profile holder's in-group inclusion, which may have resulted in a greater perceived social credibility and, in turn, skewed results. Finally, the authors' need to find unbiased and neutral stimulus articles might well have contributed to the participants' evaluations of each article's relevance, as these qualities largely apply to stories of little consequence (e.g. human interest stories).

Future research might limit its stimuli to solely High-Relevance or Low-Relevance stories in efforts to predict more accurately the effects of either condition. Researchers might also consider broadening the sample to include not only more participants, but a more socioeconomically and racially diverse group. There are many social networking sites besides Facebook (e.g. Twitter) that might be of interest to researchers in the area of social media. Due to the quickly and constantly evolving nature of Facebook and other social networking sites, research on the medium must be continually updated. The slightest change in format may have significant ramifications for social media research.


  • Valarie Schweisberger is a Ph.D. Candidate at the S.I. Newhouse School of Public Communications at Syracuse University. Her work investigates the relationship between new digital technologies and queer-identified populations.

  • Jennifer Billinson is a Ph.D. Candidate at the S.I. Newhouse School of Public Communication at Syracuse University. She does research regarding the use of popular music after tragic events.

  • T. Makana Chock (Ph.D, Cornell) is an associate professor in the Dept. of Communications of the S. I. Newhouse School at Syracuse University. Her research area is in media processes and effects with a focus on investigating the ways in which these influence perceptions of self and others.