Another Point of View: Scholarly Responses to the State of Third-Person Research

ABSTRACT

In a series of essays, scholars respond to Perloff and Shen’s article, “The Third-Person Effect 40 Years After Davison Penned It.” They offer further thoughts on how to measure the phenomenon, where future research is headed, and even whether the effect is real.

View correction statement:

Correction

This article is related to:

The Third-Person Effect 40 Years After Davison Penned It: What We Know and Where We Should Traverse

With this issue we mark another Scholarly Milestone. This time, Richard Perloff and Lijiang Shen bring us an essay reflecting on the 40 years of third-person research since Davison’s 1983 article appeared in Public Opinion Quarterly. Boldly, they don’t simply ask whether other scholars have been more influenced by this research than they themselves, but instead grapple with important lingering questions regarding the effect, its causes, and its future.

As a scholar with an enduring interest in this area and an editor who sees frequent submissions on the topic, I was happy to receive this essay and the favorable reviews that followed. The publication of this essay also afforded the opportunity to solicit the thoughts of many other prominent scholars whose work has helped shape the theory. They appear below, reacting to the essay and the broader evolution of the subject.

To add one more commentary to their own, I also think Perloff and Shen do a great service. In many ways, their essay celebrating the long-standing third-person effect (TPE) is also an essay questioning its very existence. Such questions about the core assumptions that drive many theories in our field are vital. Too often, we allow the accumulation of evidence of a specific component of a broader theory to stand in for evidence testing the theory as a whole, and it is certainly possible that at least portions of the third-person literature exhibit this shortcoming.

In the case of the third-person concept, we have ample evidence that for most kinds of media content, people anticipate that the influence on others is stronger than the influence on themselves. Furthermore, it is relatively easy to identify a form of corrective action—whether censorship, media literacy, or political counter-messaging—that is correlated with this perceptual gap. However, these two robust findings should not be confused with evidence that the perceptual gap between self and other is the force driving the tendency toward correction.

Note that this is not the same as suggesting we don’t yet know why the perceptual gap might trigger corrective action. Instead, we need to better consider whether the gap matters at all. Consider the following alternative possibilities: (1) That the apparent correlation is a mere artifact of measurement, (2) That while the perceptual gap matters, it is merely an indicator of general psychological tendencies that also drive corrective action, or (3) That specific factors shape how we see media influence and also account for the correlation, with no causal mechanism linking the perceptual gap and behavior outcomes directly.

1. Most approaches to measuring the third-person perception (TPP) involve measuring perceived influences on self and on others, and then computing a subtractive score to reflect their differences. Modifications such as the “diamond model,” which I once advocated for in the pages of this journal (Schmierbach et al., 2008) adjust for the possibility this difference score will grow as people assign higher numbers to both components, especially on a less restricted scale. But it is also possible to decompose the measure and separately model the influence on self and others. As S. Chung and Moon (2016) have pointed out, these two modeling approaches are simple arithmetic transformations; one will not produce a better “fit” than the other. As such, a result showing the perceptual gap is a significant predictor of some corrective outcome could be evidence that only one constituent element of the subtractive score matters. If people perceive a strong influence of media on others, this could account for their willingness to take action no matter how they think the content affects themselves. Conversely, when people see more of an effect on themselves, this could be correlated with more frequent use of the content and a resistance to restrictions. A valuable step would be accounting for the argument laid out by Baek et al. (2019). They suggest that we should not only model the effects on self and other, but also the interaction between these two. If high levels of perceived influence on others are influential only when the effects on self are low, this will show up with a significant interaction.

2. Even if we can better show that the perceptual gap is an important predictor, it doesn’t necessarily imply that it is causally responsible for supporting corrective actions. The TPP may be a mere example of the general tendency psychologists have observed of people to consider themselves better than average. Everyone from drivers to doctors to college professors assume that their peers are lower performing and that they fall within the upper range of performance. Media consumers appear no different. Yet presumably not everyone is equally prone to this bias. It could be that this is a chronic trait, and that individuals prone to such self-aggrandizement are also more prone to want to dictate the media use of others. Or perhaps this is a temporary state, activated by ego threat or issue-specific factors, but that also drives this protective tendency.

3. Alternatively, perhaps there is something different about the TPP compared with other better-than-average effects. In one recent study, a colleague and I found that individual differences in exemplar accessibility were predictive of the perceptual gap (Schmierbach & Boyle, 2020). Other studies by our team have found corresponding evidence that these accessibility differences predict support for corrective actions. Unlike some other examples of better-than-average effects, we have ample opportunities to encounter and remember exemplars related to media effects. Media influences are the subject of much public debate, and examples of supposed influences proliferate in the news and social media environments regularly in response to events such as gun violence (blaming video games). To be sure, such exemplars may be influential simply to the extent they remind us that others are supposedly vulnerable. But they could also be evidence of other cognitive biases, such as the fundamental attribution error. People may assume that the correlation between media exposure and negative outcomes is causal for others, due to the weak personalities of those others. Thus, a memory of media influence on others is formed. But when our own actions following media use could suggest an influence, we can point to other factors and dismiss the causal explanation. This argument is by no means unique to the exemplar evidence, but it does suggest an ongoing process shaping our schema of media influence, rather than a judgmental bias that only emerges when answering questions about media influence. Nonetheless, the tendency toward such biases could also be a predictor of the tendency to support corrective action.

Of course, these alternatives do not rule out the possibility that the perceptual gap itself does matter. Maybe people are specifically thinking about how their own experience compares with others when deciding whether restrictions on media are justified. The critical point, made on the pages that follow, is that we must not limit our research to finding reasons that link the perceptual and behavioral outcomes. Instead, we need to expand the boundaries of the theory by the counter-intuitive step of rejecting some fundamental assumptions until they are satisfactorily tested and supported.

Over the next pages, the following scholars help us begin this process with their own reflections on the piece by Perloff and Shen. In alphabetical order, they include Julie Andsager, Stephen Banning, Myojung Chung, Ben Lyons and Ye Sun, Douglas McLeod, Patrick Meirick, Zhongdang Pan, and Hernando Rojas.

Julie Andsager

Reflections on the third-person effect

This essay from Richard Perloff and Lijiang Shen ought to be included in graduate-level theory courses in the U.S. and around the world, as should Perloff’s previous essays on theoretical perspectives such as agenda-setting (2015a) cultivation (2015a), the hostile media effect (2015b), and uses and gratification (2015a), to name a few. As a field, we should appreciate scholars such as Perloff, who has often taken the time to philosophize, synthesize, and contextualize, situating the theoretical perspective into how and why a researcher came up with an idea. Shen and his colleagues have contributed a great deal to third-person research, focusing on its conceptual underpinnings, most notably on its still hazy mechanism (Shen et al., 2010, 2015, 2018). This brand of intrepid scholarship paves the way for the future.

Third-person research sometimes seem to treat the self-versus-others perception and the (behavioral) effect as similar at best, or as interchangeable at worst. (It occasionally includes presumed influence to cover all bases.) Davison’s original proposition does distinguish between the two, although he summarized them in terms of the perception as couched within the effect. First, Perloff and Shen (2023) aptly note that the model’s very inclusion of the word “‘effects’ … militates against people conceding they have been impacted by media.” Instead, they suggest we focus on perception: how did you feel about exposure, did you think about social issues differently, what did you take away from the message? The research directions that Perloff and Shen articulate should, if pursued, provide much clearer delineation between the TPP and TPE.

Of course, social media and emerging affordances present challenges to third-person research, but they also provide a means of more closely examining propositions that have been hypothesized in the past. I think that social media particularly offer an opportunity to focus on the TPE. As Davison focused on protecting society from the nefarious effects of propaganda as a primary motivation for the effect (e.g., individuals might take some sort of action to restrict propaganda from falling into the hands of the gullible), much study—though certainly not all—has asked hypothetical questions about behaviors related to speech and media messages. Whether one would contact a political representative to call for restrictions was a popular version of this in early days; it would be difficult for individuals to perform behaviors that would make a difference in mass media messages, however. I suspect that many confirmations of TPE in such scenarios were actually artifacts of social desirability or priming.

Social media offer a means of testing extant propositions related to the behavioral component of TPE by examining a close approximation to real-life situations. Depending on the platform, individuals have (a) some idea about who, how many, or at least what kinds of people, follow them; (b) evaluate the credibility of messages on topics that an individual is interested in (as opposed to the topics we’ve used as stimuli over the years because we researchers sometimes project our theoretical and methodological interest in those issues onto our participants); and (c) either consume the message and move on or share it with others. These abilities can advance TPE research in several ways, some of which Perloff and Shen have discussed.

While Davison and subsequent scholars focused on the protection aspect of the effect (that is, we must take action to protect society or segment thereof from certain kinds of messages), another way to consider how the TPE may function is as a special case of persuasion based on consideration of self as well as consideration of others (Andsager & White, 2007). The almost-inevitable thoughts of others’ responses when faced with a potentially problematic or beneficial message (in the case of first-person perceptions) may lead to perceptions that oneself has been thoughtful enough to be persuaded that the message is bad or good, perhaps through systematic processing. But others might be influenced or affected; they aren’t as self-reflexive. In other words, to understand how messages may operate requires recipients to define “self’s social reality, and social reality cannot occur without recognizing the existence of others” (Andsager & White, 2007, p. 134).

These concepts fit well within the future directions Perloff and Shen suggest, and they can easily be explored through social media based on the affordances summarized above. Current research on perceived effects of disinformation and misinformation is ripe for study through this lens, though one’s perceived social reality may not always reflect factual realities, as in the case of beliefs about COVID-19 vaccines. Despite the lack of a clear mechanism driving the third-person framework, there’s a reason we continue to pursue and understand how and why it appears to regularly: It’s so intuitive that it seems the explanations may be lurking around the next corner.

Stephen Banning

If research is a game in which the opponent is the unknown, the third-person effect has proven to be a worthy adversary, deflecting researchers over four decades. Despite this, research has continued to push the envelope of what we know and fill gaps in the literature. Indeed, the proliferation of TPE studies may rest with the persistence of the TPE itself, applicable to everything from stock market crashes to be susceptivity to anorexia. While underlying causes and interactions of the third-person effect often remain elusive, the TPE’s basic tenets are easily comprehended, and it is relatively easy to measure (with some room for improvement).

Third-person effect research continues to be seemingly endless as new media outlets emerge and lose prominence in quick succession. Researchers can certainly learn from studies that ascertain if the TPE is manifest in each new social media outlet, and there is no doubt future research will continue to test that this persistence remains. However, further momentum might be gained by exploiting three potential research opportunities, in addition to Perloff and Shen’s astute recommendations. These research paths include (a) seeking the ability to control third-person effect level, (b) adding qualitative insights, and (c) utilizing comparative studies.

Controlling the effect

One of the highest levels of research experimentation is having the ability to control variables and by extension, predict the phenomenon’s effect in a particular situation. While this has not been achieved in TPE research, research in this direction could be used to suppress some of the TPE’s insidious effects that have been explored extensively in the last few years. If, as a great number of studies suggest, the TPE has deleterious effects, perhaps there is a way to reduce those antisocial results to the betterment of society. Thus, the direction would be to not only find a more causal understanding of the TPE but also finding how to reduce it.

The clues to controlling a pernicious TPE are probably in extant research, as we already know many factors that contribute to the TPE’s manifestation. The above proposed research would attempt a result different from previous research in that the researcher would seek to measurably lower the TPE. Conceivably, the research question for some future studies might be: How do we reduce injurious impacts of the TPE? To investigate hypotheses along these lines, researchers would need to distill methods of control from what we have learned from the literature, subsequently testing those methods.

Qualitative studies as a foundation for new quantitative research

Following through with Perloff and Shen’s perspicacious suggestion of contextualizing TPE research in the future, one avenue for this route might be to use a method almost untapped in TPE studies—qualitative research. Qualitative research might delve more deeply into why a person experienced a TPE, and the process of arriving at their conclusion.

The above research would require a mixed methods approach and would present challenges because the TPE is so closely aligned with the quantitative method. Therefore, it would require researchers familiar with TPE measurement as well as qualitative methods such as grounded theory, depth interviews and perhaps focus groups. Additionally, it would also require a journal open to this, and a special call. However, it has the potential to provide researchers with insights into why the TPE occurs from a case study level in addition to learning more about what the TPE cognitive process is.

Qualitative research is not generally used in building theory because it is exploratory, uses broader research questions rather than more focused hypotheses, and because theory building is not its immediate goal. However, it could provide an important foundation for future quantitative TPE studies, offering a deeper inductive understanding.

Comparative studies

Comparative studies in new media are vital and fall under the umbrella of Perloff and Shen’s call for contextualization. There could be much more research in comparing how the TPE manifests itself in different forms of new media. While there has been persistent evidence of the TPE in new media, the variables affecting the perceptions may exhibit differently because of divergent views of ingroups and outgroups. For instance, Snapchat uses smaller circles of individuals with fleeting messages when compared to Facebook, which uses more enduring messages (except with the “Stories” section) and often larger circles of “friends.” Testing these variables would assist researchers in learning factors such as whether differences in perceived ingroup size create faster, stronger reactions, and/or more likelihood of behavioral intention.

These are not the only avenues for future research, but they help illustrate the rich future TPE research has and the benefits it can create for researchers and society. The TPE is far from being fully understood, but the next theoretical breakthroughs may be accomplished in the next decade, perhaps by one of the young researchers reading this journal.

Myojung Chung

The central tenet of the TPE has been the lack of awareness of others’ responses to the presented media content. Traditional media consumers did not, at least at the time of exposure, have direct information about what other people thought of the news story or television show. Such ignorance drove people to assume the worst-case scenario that media influence others much more negatively than themselves, and hence to take actions to protect vulnerable others. However, the rapidly evolving media landscape has shaken this notion of pluralistic ignorance (Lee & Jang, 2010). On various media platforms, we now encounter not only media content but also others’ reactions to it in the form of comments, likes, or shares. This nexus of mass communication and interpersonal communication puts the third-person effect in a fresh context.

The interaction of media content and audience feedback on the TPE unfolds in complementary, competing, and convergent patterns (Chaffee, 1986; Lee & Tandoc, 2017). Audience feedback can provide additional information about the media content (complementary), guiding people in assessing media influence. For example, by presenting lay persons’ diverse thoughts and balanced responses to a news article, user comments may serve as a cue that the public is not overly swayed by the article, thereby reducing the self-other gap in the presumed media influence as compared to the news article alone (M. Chung et al., 2015). Furthermore, when other audiences contest the media message or present opposing viewpoints to the message (competitive), the conventional self-other gap in presumed media impact manifests differently. Politically balanced news stories accompanying partisan comments induced greater discrepancy in presumed media influence on self and others than those without comments (Houston et al., 2011). Compared to a news article alone, user comments that contradict the article’s position diminished the presumed influence of the news on the general public (Lee & Jang, 2010). Lastly, audience reactions that favor or support the media content (convergent) also alter the traditional directions of the perceptual bias. People take a large number of likes and shares as a cue to the popularity of the content and perceive reduced self-other discrepancy in media influence (M. Chung, 2019), demonstrating the bandwagon effect (e.g., “If many people think this content is good, I should think so, too,;” Simon, 1954). Here, perceived popularity is a separate concept from perceived likelihood of exposure that has been widely studied in classic TPE research. Perceived likelihood of exposure is predicated on the idea that many people watched or read the media content (Eveland et al., 1999), without taking into account others’ reactions to the content. Perceived popularity, on the other hand, indicates not only that a large number of people were exposed to the content, but also that those individuals endorsed or supported it.

The masspersonal communication context influences the behavioral component of the TPE as well. While earlier TPE literature primarily focused on restrictive actions (e.g., censorship), a growing body of research highlights other behavioral consequences such as corrective or promotional actions. The asymmetry in presumed media influence on self and others predicted intentions to leave comments to warn about the potential bias in media content, click dislikes to express disapproval of media messages, share information outlining the problems in media messages, or share a socially desirable news story on social media (Barnidge & Rojas, 2014; M. Chung, 2019; M. Chung & Kim, 2021; Lim, 2017). Notably, these actions are individual-level engagement in the form of audience feedback. Traditional media consumers had little ability to interfere with the presented media content and thus relied on restrictions by authorities to counter media influence. However, technology has enabled individuals to take actions to directly engage with hypothetical others to address the asymmetrical media effect.

The on-going presence and evolution of audience feedback compels scholars to revisit long-held assumptions about the TPE. Exploring how the TPE continues to adapt to the ever-shifting media environment will be a great source of scholarly joy for the next 40 years.

Benjamin Lyons and Ye Sun

Perloff and Shen offer a sweeping review of the sprawling TPE research literature, situating it in historical context from the 1980s to today. Following this overview and contextualization, the authors look forward, offering thoughts on how to study the effect in a new media ecology and in the midst of increasing polarization in society. But in our view, considering the TPE in the context of a changing media environment is ultimately a cosmetic fix until we fully understand the nature of the effect, particularly the behavioral component, through better design—which Perloff and Shen allude to themselves.

In their review, Perloff and Shen detail some of the shortcomings in research dedicated to the behavioral component of the TPE. For instance, the authors note the size of the correlation between the perceptual gap and behaviors in meta-analyses (between r = .11 and r = .15 depending on behavior), which suggests its potential contribution to behavior is small. Even more concerning, though—and again, as Perloff and Shen rightfully note—is the dearth of causal evidence for this perception-behavior link. We think these issues deserve even more emphasis in future work. In this response, we outline the case for increased scrutiny of the behavioral component first on methodological grounds, and then on theoretical terms.

As TPE researchers, can we claim what we want to claim? Perloff and Shen cover some design issues in extant TPE research. Many of the designs in prior TPE work may not allow for the causal claims that are commonly seen. Beyond demonstrating the existence of TPP itself, much TPE research has inherently causal claims to make. For example, message social desirability and perceived similarity of others or their exposure to media messages drive TPP, which in turn drives behavioral response. Overreliance on correlational designs, however, means that most theoretical models are not adequately tested. These designs could produce spurious outcomes, which are not salvageable by merely resorting to structural equation modeling (Sun, 2013). Of special concern is the path from TPP to behavior, which has not been causally identified.

A related modeling issue is that these models leave out likely confounders identified when using corrected measurement approaches (Lyons, 2022). Take debates about TPP’s mobilizing effects, for example (Xu & Gonzenbach, 2008). If traditional TPP measures not only perceptual gaps but also real differences driven by domain interest, the latter may drive spurious associations with corrective action. At a minimum, statistical control is called for.

Contemporary methodological advances have also identified notable statistical issues in mediational analysis that are fully ignored in TPE research (Glynn, 2021; Rohrer et al., 2022). In particular, mediation models assume sequential ignorability, with no confounding between the treatment and the outcome, and no confounding between the mediator and the outcome. In TPE models, the latter is not a given; for instance, political interest might underlie greater perceptual gaps as well as behavioral “effects” such as corrective action. Reverse causality is also a potential issue. As shown in Perloff and Shen’s review, directed acyclic graphs can help pinpoint where experimental designs can improve our understanding of the theory.

Beyond these methodological concerns, we think there is lack of precision in the theory itself. TPE is a two-part theory wherein the postulated perception bias leads to social actions. The behavioral hypothesis, while imbuing the theory with practical significance, was more a vague conjecture than a clear theoretical articulation in Davison’s (1983), accompanied by only anecdotes for illustration rather than empirical evidence. Those anecdotes then became the foundation of multiple behavioral outcomes in later studies, making the behavioral hypothesis a piecemeal development (Perloff & Shen, 2023) based on our re-imagination of Davison’s vague descriptions.

While some suggestions for better theorizing the behavioral component have been made (Sun, 2013; Sun et al., 2008), the fluidity of this research terrain calls for more systematic questions about its nature and state as a theory. So far, existing research seems to be mostly driven by phenomenological interests—for example, to empirically show that individuals support censorship as they have some layperson theories about media. Theoretical mechanisms explaining why perceptual bias leads to behaviors are seriously understudied. As the number of behavioral outcomes labeled as TPE expands rapidly, it is time that TPE researchers (re)articulate the contours of the theory and formulate predictions of theoretical clarity and coherence.

Douglas M. McLeod

Third-person perceptions and political polarization

In the 40 years since W. Phillips Davison published his article on the TPE, copious research has demonstrated that the TPP is one of the most robust findings in communication research. Citizens consistently report the belief that potentially harmful media effects are significantly greater on other people than they are on themselves. This observation has held across different forms of media content and for different types of content effects (libelous content, misogynistic rap lyrics, violent video games and many more). The TPP findings are impressive given the variability and unpredictability of human beings, which makes such consistent findings few and far between.

The challenge for third-person research has always been the “so what” question—what are the consequences of such perceptions, often referred to as the “behavioral corollary” or “third-person effects”? Following Davison’s directive, research initially focused on support for media censorship as a natural consequence of perceived effects differences based on the notion that the sophisticated censor could resist harmful media effects while making decisions about whether other people needed protection. Despite studies exploring a linkage to censorship, the lack of accumulated research on other consequences of such perceptions eventually caused the perception research to run out of steam.

Yet there may be consequences of the TPP that are hiding in plain sight. To see them, we may need to broaden our scope in terms of how we think about (a) the nature of perceived media effects and (b) the way we think about consequences of effects perceptions.

The nature of perceived effects

The perception that media have powerful negative effects has many potential consequences beyond censorship. The simple perception that a particular type of media has powerful negative effects (known as “presumed influence”) on others may also have consequences regardless of perceived effects on self (“the influence of presumed influence”). Psychologically, it may be relatively easy to concede and even inflate the influence of media on others, which can have a variety of perceptual, attitudinal and behavioral consequences. Such effects may be stem linked directly to the perceived power of media effects, the other-self effects differential or both.

The process of examining linkages between effects perceptions and consequences should begin with logical reasoning to articulate the nature of the theoretical linkage between perceived effects (i.e., whether or not the other-self differential is relevant) and the potential outcome in question. Then, the linkage can be tested empirically using an appropriate analysis strategy.

As Perloff and Shen note, demonstrating the effects of the TPP goes beyond demonstrating that simple perceived media effects on others is linked to potential consequences—it requires using the “diamond model” technique to isolate presumed influence and the impact of the differential between perceived effects on others and self that has consequences.

Effects perceptions consequences

One approach to identifying potential avenues for examining perceived effects consequences is by listening to public discourse as presented by traditional media, social media or in interpersonal conversations. Politicians, pundits, activists and citizens frequently express beliefs about powerful effects of media on others, rarely if ever admitting to effects on themselves. These claims about media effects are used to rally public support, justify policies, and rationalize subsequent behaviors.

In locating the future of TPP research firmly in the 21st century, Perloff and Shen discuss the third-person phenomenon in the context of political and media polarization, particularly as claims related to the TPP get caught up in the dynamics of ingroup/outgroup. From this viewpoint, we might see TPP in terms of the first-person plural (e.g., your side is more affected than my side), the consequences of which might be expressed in terms of negative perceptions of the motivations, objectives, and viewpoints of others furthering partisan divides.

Concerns about the negative effects of ideologically hostile media have led partisans to select media diets that are conducive to their own political viewpoints. Here, such choices may be less motivated by concern that oppositional media might have an effect on oneself, but more out of a desire to support media that won’t be adversely affecting others. Ideologically selective media diets related to hostile media perceptions may further polarize citizens on the basis of subsequent differences in perceptions of facts and realities.

Perloff and Shen suggest that people will underestimate the impact of media effects on themselves in the polarization process, underestimating the impact of outgroup media (and the impact of ingroup media for that matter) and fueling hostility toward outgroups. On the other hand, people might overestimate such effects of media on outgroup members. Here, researchers could not only test these TPP hypotheses, but also examine the potential linkage to outgroup animosity and polarization as examples of TPE. Moreover, as Perloff and Shen also suggest, more research is also needed to examine whether perceptions and effects are moderated by political dispositions such that they are more influential among strong partisans or one side of the political spectrum.

Seen from this perspective, examples of perceived effects consequences are plentiful. For example, criticisms of the liberal (or conservative) bias of the media are often accompanied by claims about their effects on others. President Trump provided frequent examples demonizing journalists and news media for actively spreading fake news to undermine his administration. After he lost the 2020 presidential election, he blamed the media for suppressing public reaction to what he claimed was massive election fraud, which ultimately provided motivation for the January 6th insurrection.

In calling for policy change, conservatives have claimed that social media platforms suppress conservative viewpoints in order to sway public opinion on a variety of issues. Similarly, concerns in the opposite direction were raised by liberals when Elon Musk assumed control over Twitter. In such cases, claims about the effects of social media were being wielded publicly, largely without evidence of actual systematic content suppression.

Concerns about content effects on others extend to other forms of communication as well. For example, claims that critical race theory has permeated educational curriculum have included charges that such content brainwashes students from grade school through college. Such claims have been used to justify curriculum restrictions and to indict educational institutions, even in the absence of evidence about the nature of educational content and its effects. With a similar lack of supportive evidence, charges have been made about the effects of school curriculum dealing with gender and identity. When claims about content effects go beyond policy, results can be tragic. For instance, claims that drag shows are being used for child abuse grooming may have motivated a mass shooting at Club Q in Colorado Springs.

Globally, ideological polarization seems to be on the rise as many countries are experiencing a surge in political violence and terrorism, while fascism seems to be on the rise. And while it may be a stretch to blame such problems on the media and media perceptions, communication researchers should be attentive to public claims about content effects and the effects of those effects claims, particularly as they play a role in political polarization, especially when third-person effects go beyond antipathy and cynicism to behavioral consequences such as violence, hate crimes and terrorism.

Certainly, the principles laid out by Perloff and Shen provide both guidance and encouragement for reinvigorating third-person research by exploring linkages between TPP and the important questions concerning the consequences of TPP that are pervasive in the domestic and global political world.

Patrick Meirick

I have long appreciated Richard Perloff’s role in our discipline as a theoretical synthesist (e.g., Perloff, 1976, 1984, 2014) especially with regard to the third-person effect (Perloff, 1993, 2015b). In this latest Milestones essay, Perloff and coauthor Lijiang Shen propose a number of potentially productive paths and propositions for future research. I am honored to share my observations on a few of them.

“Differentiate distorted from accurate perceptions”

As Perloff (1993) noted, the effort to identify the nature of the perceptual bias—is it overestimation of others (Gunther (1991, CR) Perloff (1989, CR) Price et al. (1998, JoC)) or underestimation of effects on self (Cohen et al., 1988)—includes some of the earliest TPP work. The conceptual approach Perloff and Shen discuss here, of partitioning perceptual gaps (Shen et al., 2018), presents one way to address this: In cases where there is a known effect on a subgroup, the “accurate perception” would be the perceptual gap between generalized others and the subgroup, which would be subtracted from the gap between the self and the general others gap to determine the “distorted perception.” Another recent approach comes from S. Chung et al. (2018), who coupled their TPP experiment with a panel survey, allowing them to compare actual effects of polling information (change from Time 1 to Time 2) with perceived effects on oneself and others. They found both overestimation of effects on others and underestimation of effects on self in the direction the self was actually influenced, regardless of what direction that was. Those who were persuaded in the direction of the poll results (a bandwagon effect) thought others were more persuaded in favor of the poll results than they themselves were, when in fact the opposite was the case. Similarly, those who were persuaded contrary to the poll results (an underdog effect) thought others were more persuaded against the poll results than they were themselves, when the reverse was true. The results of S. Chung et al. (2018) appear to reflect false consensus, a broadly observed tendency of people to generalize their own attitudes and reactions to those of others (Marks & Miller, 1987). Their results also echo what Perloff and Shen surmised: “Individuals might project their ‘true’ reactions onto individuals who are similar to themselves in the particular domain under investigation.”

“Bring third-person effect research into the polarized age”

I heartily agree with the suggestion that, to account for reactance or biased assimilation, “researchers would need to measure perceived effects, not with an item that assessed perceived absolute value of effects, but one that looked at directionality, i.e., with one pole stating that exposure to Fox leads viewers to become more liberal, and the other pole that it leads them to be more conservative.” As a young scholar eager to find backlash effects from political advertising and thus not wanting to presume a direction of (presumed) influence, I used bipolar scales that allowed for perceived effects contrary to the intended direction. The inclusion of the negative pole was not in vain: Upon seeing an ad from an out-group candidate, people on average indicated feeling less favorable toward that candidate and less likely to vote for them (Meirick, 2004).

Proposition 1 returns to the question of perception accuracy we discussed at the beginning, but from a partisan standpoint: “Partisan ingroup members will underestimate the degree to which exposure to outgroup media causes them to feel more positively toward the ingroup and more negatively toward the outgroup.” However, the oft-cited backfire effect found in Lord et al. (1979) is not often replicated (Swire-Thompson et al., 2020). In secondary analysis of survey data, I have repeatedly found that Fox News and conservative media use is associated with persuasion in a conservative direction across the board, and that this effect tends to be stronger for Democrats than for Republicans (Meirick, 2022; Meirick & Bessarabova, 2016; Meirick & Franklyn, 2022). In light of these results, I might revise Proposition 1 to its converse to suggest that ingroup members will underestimate the degree to which outgroup media will make them feel less positively toward the ingroup and less negatively toward the outgroup. That said, I have found these counterintuitive results more consistently for liberals than conservatives, which supports Proposition 2.

But what of perceived effects of outgroup media on others? Perloff and Shen suggest partisans will perceive that such outgroup messages would “push the public away from their perspective and toward the ‘dangerous’ viewpoint of the outgroup.” The hostile media perception looms large in the current media landscape, as does the early finding that the greater the perceived bias of a message, the greater the third-person perception (Cohen et al., 1988). In my study of political advertising, the out-group ad yielded a greater TPP than the in-group ad, but the in-group ad was perceived to be more influential on the public than the out-group ad was (Meirick, 2004). Indeed, the smaller the ideological distance between the self and the sponsoring candidate, the greater the perceived effect of the candidate’s ad, regardless of the target group’s leanings or that of the candidate. This again is indicative of a false consensus effect.

Proposition 3 suggests affective polarization and cynicism as potential consequences of TPP for partisan media: “In a politically polarized media landscape, third-person effects of partisan media content will predict antipathy toward the outgroup and more cynicism about democracy.” The question of whether affective polarization would be caused by TPP or causes TPP is worth examining. On one hand, if outgroup partisan messages are perceived as moving people or even exemplifying the kinds of ideas that the outgroup holds, then TPP might well lead to affective polarization. On the other hand, it might make more sense to think of affective polarization preceding TPP and predicting greater TPP for the outgroup, such that when one thinks poorly of the other side, they are seen as the kinds of people who would “fall for” a harmful outgroup message. Either way, the suggestion that TPP researchers consider affective polarization is a timely one: In a Google Scholar search, more than half of all research containing the exact phrase “affective polarization” has appeared since the beginning of 2021. As Perloff and Shen’s analysis suggests, TPP research volume has passed its peak, so it would do well to hitch its wagon to a topic still in its ascendency.

Zhongdang Pan

Tilting our conceptual orientation to move forward

Perloff and Shen have done a great service to the field with this carefully reasoned review and reflection of the TPE literature. I can’t help but notice that they talk about “the third-person effect” as an idea, a hypothesis, or a concept, but not a theory. They make a strong case that while much has been learned with 40 years of research, the area is conceptually “stalled” and “(n)ew directions are needed.” They propose some implementable steps along three lines to take us there. These are valuable insights, especially the ones directed at strengthening the efficacy in assessing behavioral outcomes of the TPP. This is the point that I wish to pick up and expand in my comments.

I want to start by getting back to Davison’s original paper. Reading it again, it is hard not to notice that he spent the first two pages describing concrete action scenes. In each of them, individuals, whose perceptions of the potential message effects were assessed, faced the task of appraising a scenario where some media message was injected. At least some actors in the scene found it relevant or compelled to act because they felt that the message infusion might stir up something problematic. Davison reinforced this “What to do?” focus in the last section of the paper: “The Third-Person Effect in our Lives.”

My take is that understanding behavioral choices in an action scene was the focal point of Davison’s formulation. The TPP is a factor on the independent side of the equation predicting such choices that individuals may make based on their appraisal of a situation that involves media. But somehow, in the 40 years of empirical research, the self-other perceptual gap took the center stage, and behavioral choices are relegated to a “corollary.” With the cumulative evidence on the robustness of TPP solidly in place, I believe it is now time for us to center behavioral choices. In brief, we need to re-orient the scholarly endeavor, in which, instead of adding to the inventory of manifestations of TPP, we should seek to understand how individuals’ perceptions of communication, including but not limited TPP, play an explanatory role in their behavioral choices.

In our paper (Sun et al., 2008), my collaborators and I talk about a variety of possible behaviors that individuals on the reception side of the mass media flow may engage in. They may eschew a problematic situation or rectify such a situation. We characterize those in the former category (e.g., refraining from expressing one’s view, moving out a town that has been “tarnished” by negative media coverage) “accommodating behaviors” in that they involve individuals taking the presumed effects of some media message as part of the (anticipated) reality that they do not wish to participate in. We consider support for censorship, which Perloff and Shen describe as “the time-honored factor in the behavioral component literature,” a specific type of act to rectify a “problematic communication situation.” More broadly, we argue that perceptions of media (even more broadly, communication), including TPP, are part of individuals’ appraisals of a communication situation, and the behaviors are among the available options for individuals to take to remedy the situation. Media messages may be differentiated based on their presumed effects on different groups of people. On this basis, individuals, depending on their value-based appraisals of the presumed effects of a media message, may choose to take either restrictive (e.g., support for censorship), corrective (e.g., advocate for media literacy education), or amplifying (e.g., promote the message/content in question) behavior.

That was meant to be our first step to foreground behavioral choices, but we never followed it up. I wish we did. I can see the relevance of doing so in today’s media landscape. For example, would people’s perceptions of potential harm of fake news, especially on unsuspecting fellow citizens, entice people to excise more gatekeeping caution when sharing news on social media? Could some political operatives choose to spread certain messages based on their perceptions of vulnerability to such messages among a particular segment of the electorate? The list may go on. My point is that today’s digital media environment affords the prospect that various kinds of actors take up specific actions based on their appraisals of how such actions may help or thwart the potential impact of certain messages on specific groups of people.

TPP is a key category of such appraisals, which include various media-related perceptions and beliefs that have been examined in the extant literature. Their relationships with one another and the behavioral implications, however, remain to be explored more systematically. It is telling that Davison in his seminal piece went on to discuss the relationships between the TPE and “other phenomena,” including hostile media perceptions, pluralistic ignorance, and perceptions of opinion climate. Similarly telling is that Vallone et al. (1985) in their classic study on the hostile media phenomenon included the measure of “percentage of neutral viewers who would become more negative to Israel after viewing the (American networks’) news coverage” (p. 581) of the Beirut Massacre. To this list, we may add perceived differential susceptibility of various audience groups, and perceived credibility or trust ratings of the media or sources that are responsible for communication messages in question.

We need a more integrative theoretical framework to bring them together. This is part of the conceptual re-orientation that I am advocating. We need to open up a broader conceptual space to talk about how these perceptual and cognitive states concerning media (or more broadly, public communication) follow similar or overlapping venues to develop and get manifested in different domains. In Sun et al. (2008), we assert that our “conception of rectification behaviors in response to the self-other perceptual gap is also logically congruent with the cognitive reasoning underlying TPP” (p. 260). Some of the ideas in this line of reasoning have been variably tested, including self-categorization, likelihood of exposure, media effect schema, similarity vs. difference testing mind-set, etc.

All these ideas sound to me like building blocks of lay theories of media and society. Maybe this label better brackets the empirical phenomena for our inquiry. I don’t have space in this essay to further articulate this notion, suffice to say that such lay theories are individuals’ perceptions, beliefs, feelings, and explanations that are organized for them to experience their reality; they are anchored in self in part because they are inductively developed from people’s life experiences; they render complex and highly fluid social differentiations of various in- and out-groups; they objectivize media as entities, agents or tools, that are external to one’s self but intertwined with social differentiations. It is possible that when placed under this general umbrella, all those concepts or variables that have populated the TPE literature as explanations of TPP can get re-articulated and help us address the questions of how people form their lay theories of media in and related to the society that they experience and how they act out or are guided by such theories.

Hernando Rojas

Broader theoretical integration

Perloff and Shen carefully locate and systematize forty years of TPE research. Posing the, mostly rhetoric, argument that the third-person tradition might have little relevance in a media world enhanced by social interactions and immersion, they then move on to delineate possible new research paths and offer us a series of testable propositions.

I concur with Perloff and Shen’s conclusions about the continued relevance of TPE, with the need for further methodological innovations that are more in sync with a media ecosystem that is less a delivery of information system (although it was never just that) and more of a conversational space (although it is not fully that).

While teenagers might not think in terms of Instagram effects (Manago et al., 2015), they undoubtedly are making social comparisons that involve concentric circles of comparison logics. These comparisons happen in referential groups that are more nuanced than what the original blunt measures of TPE allowed for.

Social influence and social comparisons antecede the 20th century media system and logically do not end as the influence of analog media wane. In Perloff and Shen’s essay this is addressed as a need to “Bring Third-Person Effect Research into the Polarized Age,” but I would argue the task goes beyond what is outlined. While partisan political polarization has received much attention in the United States, in other societies polarization along these lines is less central.

To address the challenges of polarization we need to consider forms beyond issue polarization, including perceived polarization (Yang et al., 2016), and affective polarization (Iyengar et al., 2012). Initial evidence on affective polarization, that is, a tendency to view outgroups as less intelligent, uninformed, or dangerous, suggests that in certain countries (e.g., the U.S.) TPPs contribute to affective polarization above and beyond hostile perceptions of media (that in turn conflate a third person phenomenon), while in others (e.g., Mexico and Brazil) this is not the case (Tong et al., 2021).

Perceived polarization, that is, overestimating the extremity of outgroup preferences, appears as a particular fertile ground to explore perceptions of media effects, as those perceiving strong effects on others and hostile messages would be more likely to perceive polarization which could be a gateway into affective and issue polarization.

Furthermore, the notions of perceived media biases and differential effects is necessary to understand intergroup relations in increasingly balkanized yet globalized societies and social movements, and this goes well beyond party polarization to include many kinds of social animosities. So beyond partisan polarization, perceptions of media effects can serve us well in theorizing social integration mechanisms based on communication.

Finally, I would suggest that broader and more integrative models of perceived effects need to be developed. These at a minimum need to consider perceptions about oneself, relevant others (both ingroups and outgroups), and perceived position of the message/source of a given communication. Using these perceived positions, we could imagine that people make mental triangulations that result in third-person, hostile media, projections, or spiral of silence

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Mike Schmierbach

Mike Schmierbach (Ph.D., University of Wisconsin-Madison) is a professor in the Bellisario College of Communications at Penn State, and editor of Mass Communication and Society. His research focuses on the factors that shape how individuals assess and react to media, with particular interests in news credibility, presumed influence, and enjoyment of video games.

Julie Andsager

Julie Andsager (Ph.D., 1993, University of Tennessee) is a professor in the School of Journalism & Electronic Media at The University of Tennessee, Knoxville. She and H. Allen White (Ph.D., University of Tennessee) wrote the book Self vs. others: Media, messages, and the third-person effect (2007, NJ: Lawrence Erlbaum Associates).

Stephen Banning

Stephen Banning (Ph.D., Southern Illinois University at Carbondale) is a professor at Bradley University’s Slane College of Communications and Fine Arts. His research interests include media effects, particularly involving the Third-Person Effect, Agenda Setting and the Theory of Reasoned Action.

Myojung Chung

Myojung Chung (Ph.D., Syracuse University) is an assistant professor in the School of Journalism at Northeastern University. Her research focuses on message processing in digital media contexts, particularly through the lens of the third-person effect.

Ben Lyons

Ben Lyons is an Assistant Professor of Communication at the University of Utah. He received his Ph.D. from Southern Illinois University and completed postdoctoral work at the University of Pennsylvania and the University of Exeter.

Douglas M. McLeod

Douglas M. McLeod is the Evjue Centennial Professor in the School of Journalism & Mass Communication at the University of Wisconsin–Madison where he teaches courses in political and strategic communication, cognitive media effects and public opinion. McLeod has more than 100 research publications in academic journals and edited books. He also wrote the book, News Frames and National Security: Covering Big Brother, with Professor Dhavan V. Shah. McLeod’s research focuses media framing and framing effects, particularly on how news media construct messages regarding social movements and how news story framing of social protests influences public opinion.

Patrick C. Meirick

Patrick C. Meirick (Ph.D., University of Minnesota) is an associate professor of communication at the University of Oklahoma and an associate editor of Mass Communication and Society. His research examines misinformation and misperceptions, political advertising, framing, agenda-setting, and perceptions of media effects. He is the co-author (with Jill Edy) of A Nation Fragmented: The Public Agenda in the Information Age.

Zhongdang Pan

Zhongdang Pan is a Professor of Communication Science in the Department of Communication Arts at University of Wisconsin-Madison. His research focuses on mediated communication in public life. He has published research on news framing and its effects, civic implications of political talk, perceptions of media effects, Chinese journalism, and mediated civic engagement in China.

Hernando Rojas

Hernando Rojas (Ph.D. University of Wisconsin–Madison) is the Helen Firstbrook Franklin Professor at the University of Wisconsin‐Madison. His scholarship focuses on political communication, examining: (a) the deployment of new communication technologies for social mobilization; (b) the influence of audience perceptions of media (and audience perceptions of media effects) on both public opinion and the structure of the public sphere; and (c) the conditions under which media support democratic governance.

Ye Sun

Ye Sun (Ph.D., University of Wisconsin-Madison, 2008) is an Associate Professor in the Department of Media and Communication at the City University of Hong Kong.