Third Person
Effect
“It may persuade them.
It will not persuade me.”
Explore W. Phillips Davison’s account of perceived media influence, the self–other gap and the decisions people make when they expect an audience to be persuaded.
In this chapter · Open the contents
- The central idea
- Two components and a model
- Measuring the gap
- Why perceptions diverge
- Desirability and first-person perception
- Social distance and the imagined audience
- Source, involvement and knowledge
- Behavioural consequences
- Perception and actual influence
- Related concepts
- Research evidence and digital media
- Cases, strengths and criticisms
- Watch the lecture
- Check your understanding
- Activities and research design
- Glossary and readings
Explain the gap without assuming its accuracy.
- Distinguish perceptual and behavioural claims in third-person research.
- Calculate and interpret a matched self–other difference.
- Explain desirability, social distance and other conditions affecting the comparison.
- Separate perceived influence, actual influence and accuracy.
- Evaluate protective responses and design a focused research inquiry.
“I can see through it. Other people may believe it.”
A student watches an advertisement that promises instant success. She dismisses the claim, then worries that other viewers will accept it. A campaign worker doubts an opponent’s message but prepares a response because voters might be persuaded. A parent recognises a persuasive appeal and assumes that younger audiences will be less able to resist it.
These judgements turn on a comparison: how much a message seems likely to influence oneself and how much it seems likely to influence others.
W. Phillips Davison introduced the term in his 1983 article, The Third-Person Effect in Communication, published in Public Opinion Quarterly. His approach brought a different question into media research: what happens when people act on their beliefs about media influence? Read the foundational article ↓
The influence people imagine can itself have consequences
A person need not accept a persuasive claim to respond to it. Believing that an audience will accept it may prompt a warning, a competing message or a demand for restriction.
This makes perceived influence an object of study in its own right. It also creates a second task: checking whether the predicted influence actually occurred.
Why “third person”?
The phrase directs attention towards an imagined “they”: other viewers, other voters or people outside one’s immediate circle. It does not describe a third stage in a communication sequence or the grammatical voice used in a message.
A perception and its possible consequences are different claims.
Third-person research commonly distinguishes a perceptual component from a behavioural component. Separating them prevents a finding about judgement from silently becoming a finding about action.
| Component | Central question | Evidence needed |
|---|---|---|
| Perceptual | Does a person estimate greater influence on others than on self? | Comparable self and other estimates for a defined message and outcome. |
| Behavioural | Do perceived effects help explain a specified response? | A separately measured response and a design that examines its relationship with the perceptions. |
A specified message
A person sees a message or considers a clearly described communication.
“How much would it influence me?”
Estimated influence on oneself.
“How much would it influence them?”
Estimated influence on a defined comparison group.
Third-person perception: O > S
The person judges the specified influence to be greater on others.
Possible behavioural link: beliefs about harm, responsibility, effectiveness and the situation can shape the response.
Warning, counter-message, support for restriction—or no action
Measure the response rather than infer it from the comparison.
Actual influence requires another inquiry: compare appropriate observations of message effects with these perceived effects.
The gap does not compel one response
Two people may report the same self–other difference. One favours media education. Another supports restriction. A third takes no action because they doubt that intervention will help.
The behavioural claim therefore needs an account of how a judgement connects with a response. Concern, perceived responsibility, available choices and expected consequences may enter that connection.
A person can acknowledge some influence
Third-person perception does not require a claim of total immunity. Someone who estimates substantial influence on themselves can still estimate even greater influence on others.
Nor must the respondent consciously recognise a contradiction. The researcher observes the comparison; the explanation for it remains a further question.
Ask comparable questions before subtracting the answers.
Begin with a particular message, an identified comparison group and a specified outcome. “The media influence society” is too broad to reveal exactly what is being compared.
Suppose a class examines a misleading advertisement. Both items use a seven-point scale, from 1 = not at all to 7 = very much, and assume one viewing.
Estimate S
“After viewing this advertisement once, how much would it increase your willingness to buy the advertised product?”
Estimate O
“After viewing this advertisement once, how much would it increase the willingness of other students in your year to buy the advertised product?”
The wording holds the outcome and exposure assumption constant. Only the target changes. These are example items for understanding the comparison; a research instrument also needs piloting and evaluation.
| Pattern | Self: S | Others: O | Difference: D | Interpretation |
|---|---|---|---|---|
| A | 2 | 5 | +3 | Greater perceived influence on others. |
| B | 5 | 2 | −3 | Greater perceived influence on self. |
| C | 4 | 4 | 0 | Equal perceived influence. |
| D | 4 | 7 | +3 | Greater perceived influence on others, at higher absolute estimates than A. |
Positive D: third-person perception. Negative D: first-person perception. Zero D: equal estimates for this comparison.
Patterns A and D share a three-point gap. They do not share the same level of perceived influence. Retain the two component estimates when interpreting the result.
What the number does—and does not—say
A three-point difference on this response scale is not “three times the influence” or “three per cent more influence”. It describes the distance between two reported ratings.
One respondent’s pair illustrates an individual judgement. A claim about a population requires suitable sampling, analysis and uncertainty estimates. A claim about actual persuasion requires a measure beyond these predicted effects.
People know their own resistance better than other people’s reasoning.
Several explanations address the self–other gap. They can overlap, but none should be treated as the single cause in every setting. A useful explanation specifies what changes the judgement and how that change can be tested.
Self-enhancement
Resisting manipulation can support a favourable self-image. A viewer may present themselves as alert and independent while imagining other viewers as less watchful.
Consider the boast, “I spotted the sales trick immediately.” It may reflect real scrutiny, a flattering judgement, or both. The statement alone cannot tell us which.
Optimistic bias
People may judge themselves less likely than comparable others to suffer an undesirable outcome. Applied to persuasion, this can become confidence that one will avoid being deceived or pushed into a poor decision.
This is a comparative belief. Whether the confidence is warranted remains an empirical question.
Attribution and unequal information
A person has access to their own doubts, prior knowledge and reasons for refusing an appeal. They may see far less of another viewer’s reasoning and fill that gap with assumptions about gullibility.
Attribution-based accounts ask how people explain their own responses and other people’s responses. They should not be reduced to an inflexible rule about how everyone assigns causes.
Everyday beliefs about powerful media
Someone may imagine advertising or propaganda as an almost irresistible force for an undifferentiated audience, while granting themselves an exception. That combination produces a ready-made route to a self–other gap.
The belief resembles a simple powerful-media account. Observing that belief does not establish that media actually operate in that manner.
Being influenced can sometimes be something people welcome.
The meaning of “influence” matters. Being deceived by a false promise is undesirable. Learning a useful practice may be welcome. People need not protect their self-image by denying both.
Distinguish the attractiveness of a message from the desirability of its presumed consequences. A polished advertisement may be enjoyable while the purchase it encourages is judged harmful.
| Presumed influence | Possible judgement | Interpretation |
|---|---|---|
| Undesirable | “Others would be more likely than I am to accept that deceptive claim.” | A third-person pattern may appear. |
| Desirable | “This reminder would improve my own checking habits more than other people’s.” | A first-person or reversed pattern may appear. |
| Mixed or disputed | “It helps people notice an issue, but it may also mislead them.” | Separate the outcomes before comparing estimates. |
First-person perception
A first-person perception occurs when estimated influence on oneself exceeds estimated influence on others. It is also called a reversed third-person pattern.
Suppose a student welcomes a public-service message encouraging source verification and expects to learn more from it than classmates will. The direction reverses: S exceeds O.
A possibility, not a mechanical reversal
Desirable messages do not always produce first-person perception. Respondents can disagree about the value of the intended effect, their need for the information and the likely response of the comparison group.
Moser and colleagues’ 2024 reconsideration of two meta-analyses found support for first-person effects in the desirable-message comparisons they reanalysed. Their discussion also shows why study inclusion and the definition of desirable influence deserve scrutiny. Read the reanalysis ↓
Ask what influence the respondent values. A general label such as “positive campaign” cannot replace that measurement.
“Other people” is a moving target.
A close friend, a classmate, an unfamiliar audience and the general public are different comparison targets. Each description supplies information—or invites assumptions—about knowledge, similarity and exposure.
The social-distance corollary proposes that self–other differences can grow as the comparison group becomes more distant from the self. Distance may concern social identity, perceived similarity or location.
Distance needs an explanation
Suppose a viewer expects an advertisement to affect unfamiliar audiences more than friends. The reason might be perceived vulnerability. It might instead be a belief that those audiences encounter the advertisement more often.
A distant group need not be imagined as more susceptible. A specialist community could be geographically distant yet regarded as especially knowledgeable about the product.
| Target description | Assumption it may invite | Research question |
|---|---|---|
| Close friends | Shared knowledge and similar habits. | Does the respondent actually know their viewing and buying practices? |
| Students at another institution | Difference or unfamiliarity. | Is the judgement driven by distance, exposure or an inferred trait? |
| People who often encounter this content | Greater opportunity for influence. | Does the comparison hold exposure constant? |
| “The general public” | A broad, poorly specified audience. | Which people does the respondent imagine? |
Perceived vulnerability is not a group fact
If a respondent imagines a group as easily misled, that judgement belongs to the data on perception. It should not be repeated as an established description of the group.
A strong study defines the target, measures relevant assumptions and examines alternative explanations. “More distant means more affected” is too blunt to capture these distinctions.
The same comparison can change with the message and the viewer.
Richard Perloff’s 1993 review stressed that third-person perception does not emerge for all people in all circumstances. It drew attention to perceived benefit, personal importance and judgements about the source. Read the review ↓
Persuasive intent and source judgements
A viewer who recognises an attempt to persuade may feel able to resist it while expecting others to miss the attempt. Perceived bias or low credibility can enter the same judgement.
Keep source credibility separate from estimated influence. A source may seem untrustworthy yet capable of persuading a particular audience. Conversely, a credible source may be judged influential on oneself as well as others.
Issue involvement
An issue tied to identity or a valued commitment can sharpen attention to possible influence on other people. Someone may scrutinise a message closely, defend their own position and worry about an undecided audience.
Involvement can also encourage careful learning. Its role depends on the message, the outcome and the comparison target; it is not a universal switch that enlarges every gap.
Knowledge and confidence
People who feel knowledgeable may expect their expertise to shield them. Test what they know as well as what they claim to know. Confidence and demonstrated knowledge can diverge.
Age or education may shape experience, exposure and confidence, but neither establishes immunity to persuasion. Avoid treating a demographic category as a ready-made explanation.
State a conditional claim
Instead of writing “experts are unaffected”, write: “For this message and outcome, respondents reporting greater expertise estimated less influence on themselves; performance and actual effects were examined separately.” That wording identifies what the study measured.
People may respond to the influence they expect in an audience.
Imagine a campaign team that doubts an opponent’s claim but expects voters to accept it. The team may release a rebuttal. Its decision can be driven by anticipated audience effects even if the original claim persuades few voters.
This is the behavioural question: how do perceptions of influence enter decisions about communication?
| Response | Possible reasoning | What to measure separately |
|---|---|---|
| Support for restriction | “Others may be harmed, so access should be limited.” | Support for a specific restriction, its rationale and any actual action. |
| Warning or education | “People need tools to assess this claim.” | The warning, intended audience and any change in checking practices. |
| Counter-message | “The audience should hear a competing account.” | Production, distribution and reception of the response. |
| Amplification | “This useful message should reach more people.” | Sharing or promotion and the perceived benefit. |
| No intervention | “The influence is possible, but this response would not help.” | Available options, expected effectiveness and reasons for restraint. |
Support is not the same as action
Agreeing with a survey statement about restrictions differs from signing a petition, changing a rule or removing a message. An intention to warn friends differs from sending the warning.
Label the measured outcome precisely. The theory can guide the inquiry without collapsing these steps.
Does the gap explain the response?
A person may favour intervention because they expect serious effects on others, regardless of how they rate themselves. Another may object because the message offends them personally.
Research must therefore examine whether the self–other disparity contributes information beyond its component estimates and other relevant judgements. A subtraction score alone does not settle the mechanism.
Xu and Gonzenbach’s 2008 meta-analysis examined the behavioural component, while Sun, Shen and Pan’s 2008 study considered different kinds of corrective response. These readings invite a more specific question than whether a gap always leads to censorship. Behavioural meta-analysis ↓ · Response study ↓
A difference in estimates is not yet evidence of an error.
The phrase “third-person bias” is common, but a self–other gap by itself does not reveal which estimate is inaccurate. Someone may underestimate their own susceptibility, overestimate others’ susceptibility, do both, or accurately recognise a difference.
Imagine two runners predicting their times. Different predictions are not evidence of overconfidence until an appropriate performance comparison is available. The same distinction applies when people predict media effects.
Build a suitable benchmark
If the perceived effect concerns accepting a false claim, measure acceptance under defined conditions. If it concerns purchasing, a test of whether people remember the advertisement is insufficient.
Actual-effect research also needs a credible account of what would have happened without the message. A post-viewing response alone cannot isolate the message’s causal contribution.
Confidence can sometimes be earned
Benjamin Lyons’s 2022 research compared perceptions of false-news susceptibility with behavioural performance. It showed why some comparative confidence can reflect real differences rather than presumed invulnerability alone. Read the accuracy study ↓
The implication is methodological: distinguish the observed perception pattern from a demonstrated calibration error. It does not establish that all self-confidence is justified.
| Claim | Evidence it needs | Common overreach |
|---|---|---|
| A perceptual difference exists | Comparable estimates for self and others. | Calling the difference proof of unequal actual influence. |
| An estimate is inaccurate | A defensible benchmark matched to the predicted outcome. | Calling confidence a bias without checking performance or effects. |
| Perceptions influence a response | A measured response and a design testing the proposed link. | Assuming that a reported gap necessarily causes action. |
Perceptions can matter even when they are mistaken. They can also matter when they are accurate. Their consequences and their accuracy are related questions, but they require different tests.
Compare the question each approach asks.
Third-person perception sits beside several accounts of media, judgement and public opinion. Similar language can conceal different dependent variables.
| Concept | Main question | Distinction from third-person perception |
|---|---|---|
| Direct media effects | Does exposure change knowledge, attitudes or behaviour? | Investigates observed effects rather than only estimates of effects. |
| Hostile media perception | Do involved audiences perceive coverage as biased against their own position? | Concerns perceived direction or fairness of coverage, rather than the self–other influence gap. |
| Spiral of Silence | Does perceived opinion support affect willingness to express a view? | Concerns opinion climate and expression; it does not require an influence comparison. |
| False consensus | Does a person overestimate how widely others share their own view or conduct? | Concerns similarity or prevalence, rather than relative susceptibility to a message. |
| Influence of presumed influence | Do beliefs about media effects on others shape one’s responses? | The broader mechanism can operate without a measured self–other gap. |
| First-person perception | Is influence estimated as greater on self than on others? | Reverses the direction of the perceptual comparison. |
Hostile media perception can coexist with the gap
A committed supporter might judge a report as unfair to their side and expect it to persuade undecided viewers more than themselves. The first judgement concerns the report’s perceived bias. The second concerns comparative influence.
Measure them separately. A respondent’s accusation of bias is neither proof that the report is biased nor proof that it is neutral.
Keep the outcome in view
When comparing theories, write down the outcome being explained: estimated susceptibility, perceived fairness, willingness to speak or an observed change in belief. This simple step prevents one result from being recruited to support a different claim.
A recurring pattern still leaves mechanisms to investigate.
What the research synthesis supports
Sun, Pan and Shen’s 2008 meta-analysis combined 60 papers, 106 studies and 372 effect sizes. It found robust evidence of third-person perception and identified conditions involving presumed message desirability and the characteristics of comparison audiences. Read the perceptual meta-analysis ↓
This supports taking the pattern seriously. It does not mean every respondent, message or comparison yields the same result. An aggregate finding also cannot establish the accuracy of any particular self or other estimate.
Perloff and Shen’s 2023 review revisits both the perceptual disparity and its behavioural corollary, bringing measurement problems and contemporary media conditions into the discussion. Read the later review ↓
Personalised feeds complicate the comparison
People may encounter different messages, sources and frequencies of exposure. “How much does this feed influence others?” can therefore compare different experiences as well as different people.
A study can hold exposure constant, measure natural exposure, or examine beliefs about exposure. Each choice answers a different question. State the choice before interpreting the gap.
Influencers and algorithmic persuasion
A viewer may believe that other users are unusually vulnerable to an influencer’s endorsement while regarding their own viewing as entertainment. Another may assume that a recommendation system identifies everyone else’s weaknesses more effectively than their own.
These are applications to investigate. To test them, specify the endorsement or recommendation, the audience, the predicted change and the response. Avoid treating the word “algorithm” as an explanation by itself.
AI-generated messages
For a study of AI-generated content, compare judgements about a defined message and its source label. Separate the content’s quality from beliefs about how it was produced.
A possible research question is whether recognising an AI label changes self–other estimates, verification choices or support for warnings. The direction should be tested rather than assumed.
Visible reactions are incomplete evidence
A like, share or comment can reflect endorsement, criticism, humour or an effort to alert others. Counts of visible activity cannot, by themselves, establish persuasion or the reasoning behind a protective response.
Follow the comparison into a concrete decision.
These cases connect a particular message with a particular audience and response. Begin with the measured judgement, then ask which explanation survives the evidence.
Case 1 · A deceptive advertisement on a student feed
Suppose students encounter an advertisement that promises an unrealistic academic shortcut. Several say they would ignore it but worry that other students will pay for the service. Some favour a warning; others want the advertisement removed.
Apply: define the predicted effect—believing the promise, clicking the link or purchasing. Ask matched questions about self and a specified student group.
Investigate the response: measure support for warnings and removal separately. Ask about perceived harm, personal offence and expected effectiveness.
Check the inference: the difference in predicted purchases is a perception measure. Claims about actual purchases require separate evidence. Students may also support removal because the claim is deceptive, regardless of comparative susceptibility.
Case 2 · A campaign prepares a rebuttal
Suppose a local campaign team sees an opponent’s misleading video. Team members doubt that it will change their own views but expect it to influence undecided residents. They prepare a rebuttal.
Apply: identify the self estimate, the defined audience estimate and the decision to respond. The comparison group differs in prior commitment, which may make some difference in susceptibility plausible.
Investigate the response: examine meeting records or interviews alongside the released rebuttal. Consider strategic visibility, fundraising and pressure from supporters as alternative motives.
Check the inference: releasing a rebuttal shows a communication response. It does not establish that the original video persuaded residents or that the perceptual gap caused the decision.
Case 3 · A useful verification campaign
Suppose a college circulates a short message about checking the source of a striking claim before forwarding it. A student expects the message to improve their own checking more than that of other students.
Apply: this is a possible first-person pattern for a desirable outcome. Ask whether the student values the proposed practice and how much room they see for improvement.
Investigate the response: compare stated intentions with subsequent checking choices in an appropriate task. Measure any sharing of the campaign separately.
Check the inference: welcoming the message does not guarantee learning. A first-person estimate is still an estimate, and favourable self-presentation may coexist with real improvement.
What the approach contributes
- It makes beliefs about media effects visible as a source of social action.
- It joins individual self–other comparisons with debates over messages and audiences.
- It explains why rejecting a message can coexist with responding vigorously to it.
- It prompts researchers to distinguish expected influence from observed influence.
Where criticism improves the inquiry
A broad label can hide different effects. Attention, learning, agreement, purchase and harm should not be packed into one unspecified measure of “influence”.
A difference score can hide its components. Equal gaps can arise at different levels of perceived influence. Examine the self and other estimates as well as the comparison.
Self-reports have limits. A report can reflect self-presentation, interpretation of the question or uncertainty about the audience. Use additional evidence when testing mechanisms or accuracy.
Behavioural links are conditional. A cross-sectional association does not establish temporal order or rule out common causes. More precise designs can test competing accounts.
Protection can shade into paternalism. Concern for an audience can prompt useful support, but it can also justify speaking over that audience. Investigate people’s capacities and experiences rather than assuming incapacity.
Listen for the move from “they may be influenced” to “we should respond”.
Watch Dr. USP’s Media Influence: The Third-person Effect. Keep three columns in your notes: perceived influence on self, perceived influence on others and the response being discussed.
Watch the lecture on YouTube ↗
| Time | Focus | Pause and ask |
|---|---|---|
| 00:00–02:20 | Davison’s proposal and the self–other gap. | What makes the comparison a perception rather than an actual-effect measure? |
| 02:20–03:30 | Measurement and responses to expected influence. | Which additional evidence would test the behavioural claim? |
| 03:30–05:35 | Attribution, self-image and beliefs about media power. | How could competing explanations be distinguished? |
| 05:35–07:30 | Undesirable influence and the reversed pattern. | What would make admitting influence desirable? |
| 07:30–09:20 | Source judgements, distance, involvement and knowledge. | Who is being imagined as the audience, and why? |
| 09:20–end | Question order, accuracy and digital research. | How would you separate confidence from demonstrated resistance? |
After viewing
Choose one example from the lecture. Rewrite it as three testable questions: one about the perceptual comparison, one about accuracy and one about a specified response.
Test the distinctions before applying the theory.
Choose one answer for each question. Explain your choice before opening the answer.
01Who introduced the Third Person Effect in the 1983 foundational article?
02A respondent rates influence on self as 3 and on others as 6. With D = O − S, what follows?
03Which observation concerns the behavioural component?
04What does a first-person perception describe?
05Two respondents have S/O ratings of 2/5 and 4/7. Which statement is correct?
06Why should a study define “others” carefully?
07What is needed to call a comparative estimate inaccurate?
08A supporter calls a report unfair to their side. Which concept does that judgement most directly concern?
09Which pair of survey questions offers the clearest comparison?
10Which conclusion is strongest after a one-time survey finds an association between a gap and restriction support?
Turn “people will be influenced” into a question you can investigate.
Activity 1 · Audit a claim about an audience
Choose an advertisement, a public-service message or a public statement about media influence. Record the claim without endorsing it.
- Identify the message: specify the content, source and context.
- Name the target: replace “people” with a defined audience.
- Specify the outcome: distinguish attention, belief, intention and action.
- Locate the comparison: is influence on the speaker’s own self estimated, or only influence on others?
- Trace the response: identify a warning, counter-message, restriction or other decision.
- Assess the evidence: separate what is perceived from what has been observed.
Activity 2 · Compare two meanings of influence
Develop matched self and other questions for one undesirable outcome and one desirable outcome. Keep the audience definition and response scale consistent within each pair.
Explain why the outcome is expected to be desirable or undesirable, and ask how that assumption could be checked. Do not classify a message solely from the label attached by its producer.
Predict a possible pattern for each pair. Then state an outcome that would challenge your explanation. This turns a plausible story into a testable proposal.
Research extension · Design a study in stages
| Stage | Decision | Reason |
|---|---|---|
| Message and exposure | Use a defined stimulus or a precise description; state the exposure assumption. | Direct exposure and imagined exposure answer different questions. |
| Perception measures | Match self and other items; record both component estimates. | The subtraction should compare the same outcome. |
| Question order | Randomise or counterbalance the self/other order when appropriate. | The measurement sequence may shape the comparison. |
| Audience assumptions | Measure relevant perceived similarity, exposure or vulnerability. | A target label can carry several explanations. |
| Response outcome | Separate policy support, behavioural intention and observed action. | These outcomes are related but not interchangeable. |
| Accuracy or causality | Add suitable performance benchmarks or a design identifying change. | A survey gap alone tests neither actual effects nor causal direction. |
Price and Tewksbury’s 1996 study provides a reading on question order, contrast and knowledge. Check measurement effects without assuming that the entire phenomenon is a questionnaire artefact. Read the measurement study ↓
For analysis, estimate the paired contrast appropriately, report uncertainty and examine the component scores in relation to the substantive question. A proposed causal pathway needs a design capable of addressing temporal order and plausible alternatives.
For a classroom demonstration, explain the purpose and allow students to decline participation. A convenience sample can illuminate the method; it cannot represent an entire population.
Examination and seminar prompts
- Explain Davison’s contribution to the study of media influence.
- Distinguish the perceptual and behavioural components with one example.
- Discuss desirability, social distance and source judgements as conditional influences.
- Evaluate the statement: “A third-person gap proves a self-serving bias.”
- Design a study that separates the gap from perceived influence on others alone.
- Assess a protective response without assuming either audience helplessness or audience immunity.
Keep the estimate, the comparison and the consequence visible.
- Third-person perception
- Greater estimated influence of a specified message on others than on oneself.
- Perceptual component
- The self–other comparison in estimated media influence.
- Behavioural component
- The proposition that perceptions of influence contribute to specified responses.
- First-person perception
- Greater estimated influence on oneself than on others.
- Referent others
- The comparison group a respondent considers when estimating influence.
- Social-distance corollary
- The proposal that a larger perceived distance from a comparison group can be associated with a larger self–other gap.
- Self-enhancement
- Maintaining or promoting a favourable view of oneself.
- Optimistic bias
- Comparative optimism about experiencing favourable outcomes or avoiding unfavourable ones.
- Presumed influence
- The media effect a person expects in an audience.
- Difference score
- A quantity formed by subtracting one component estimate from another; here, O − S.
- Calibration
- The correspondence between an estimate and a suitable benchmark.
- Moderator
- A condition under which the size or direction of a relationship differs.
- Hostile media perception
- Perceiving coverage as biased against one’s own position.
- Actual media effect
- A change attributable to media exposure under a design capable of supporting that inference.
Core and further readings
- Davison, W. P. (1983). The third-person effect in communication. Public Opinion Quarterly, 47(1), 1–15. Article DOI ↗
The foundational formulation of perceived influence and its possible social consequences.
- Perloff, R. M. (1993). Third-person effect research 1983–1992: A review and synthesis. International Journal of Public Opinion Research, 5(2), 167–184. Article DOI ↗
An early synthesis of evidence, explanations and conditions affecting the pattern.
- Price, V., & Tewksbury, D. (1996). Measuring the third-person effect of news: The impact of question order, contrast and knowledge. International Journal of Public Opinion Research, 8(2), 120–141. Article DOI ↗
A methodological reading on how the comparison is elicited.
- Sun, Y., Pan, Z., & Shen, L. (2008). Understanding the third-person perception: Evidence from a meta-analysis. Journal of Communication, 58(2), 280–300. Article DOI ↗
A quantitative synthesis of the perceptual component and its moderators.
- Sun, Y., Shen, L., & Pan, Z. (2008). On the behavioral component of the third-person effect. Communication Research, 35(2), 257–278. Article DOI ↗
Examines different responses and the modelling of perceptual disparities.
- Xu, J., & Gonzenbach, W. J. (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. Article DOI ↗
A synthesis focused on the link between perceptions and responses.
- Lyons, B. A. (2022). Why we should rethink the third-person effect: Disentangling bias and earned confidence using behavioral data. Journal of Communication, 72(5), 565–577. Article DOI ↗
Uses behavioural evidence to examine the distinction between a perceptual gap and inaccurate confidence.
- Perloff, R. M., & Shen, L. (2023). The third-person effect 40 years after Davison penned it: What we know and where we should traverse. Mass Communication and Society, 26(3), 384–413. Article DOI ↗
A later review of the perceptual disparity, behavioural corollary and research agenda.
- Moser, K., Paul, K. I., Soucek, R., Eskofier, A., & Galais, N. (2024). The first-person effect. A reconsideration of two meta-analyses. PLOS ONE, 19(12), e0311155. Open-access article ↗
Reanalyses evidence concerning desirable messages and reversed self–other patterns.
- Dr. USP. Media influence: The third-person effect [Video]. YouTube. Watch the lecture ↗
The chapter in six statements
- Third-person perception compares estimated influence on self and a defined group of others.
- The perceptual gap and a behavioural response require separate evidence.
- Desirability, comparison audiences and message conditions can change the pattern.
- A positive difference score does not establish unequal actual influence or inaccurate confidence.
- Warnings, counter-messages and restrictions are possible responses, rather than automatic consequences.
- Careful research matches outcomes, retains component estimates and tests the proposed mechanism.
Follow a claim from the message to the audience.
Social Cognitive Theory examines learning through observation and the conditions that shape it. Gerbner’s Model follows perception and representation within communication. The Propaganda Model shifts attention towards structural influences on media production.
The next time a message seems dangerous only for “other people”, ask: what am I estimating, what do I know about that audience, and how could the judgement be checked?
Back to the chapter opening ↑