Transparency News

Method

The research behind the temperature words

Each word on the temperature, softening and insult lists belongs to a group. For each group: what published studies found about readers, how far that applies to news wording, the choice the site made because of it, and whether the site has tested it locally. The strength of the evidence caps how many points a word can cost.

Compiled 2026-10-22. The number beside each word is its effective severity in points; "exp." means experimental, shown but not counted. Corrections are welcome on the corrections page.

Combat and impact verbs used for criticism or debate

Evidence: Indirect: related studies only

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Loftus and Palmer (1974) found that a more violent verb for the same car crash ("smashed" vs "hit") raised speed estimates and false memories of broken glass. Ecker et al. (2014) found that a headline that misstated the article left readers with less accurate memory of its facts.Hypothesis: a combat verb such as "slams" for a verbal criticism may lead readers to judge the exchange as more severe or hostile than the facts support.
Limit: Neither study tested combat verbs used for criticism or debate, and the Loftus and Palmer effect has mixed direct replications.		Severity capped at 1 point per word.Pending: the rating panel has not yet checked this group.

slams 1slammed 1blasts 1blasted 1rips 1ripped 1torches 1torched 1unloads 1lashes out 1lashed out 1smacks 1smacked 1hammers 1hammered 1destroys 1destroyed 1obliterates 1obliterated 1eviscerates 1eviscerated 1demolishes 1demolished 1shreds 1shredded 1

Anger and outrage words

Evidence: Category: studies tested the wider category

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Studies of social media posts and news articles found that moral-emotional words and anger were associated with more sharing (about 12% more shares per moral-emotional word; +34% odds of being emailed per standard deviation of anger), and that likes and shares reinforced users' outrage posts.Hypothesis: anger and outrage words in a news report may raise readers' arousal and push the story to spread for its heat rather than its facts.
Limit: These studies measured sharing and posting, not how readers judge a news report, and most of these exact words were not tested; anger words did not raise headline clicks in Robertson et al. (2023).		Severity capped at 2 points per word.Pending: the rating panel has not yet checked this group.

fury 2furious 2outrage 2outraged 2rage 2raging 2erupts 2erupted 2firestorm 2frenzy 2backlash 1fierce 1fiercely 1scathing 1blistering 1searing 1

Catastrophe and disorder words

Evidence: Category: studies tested the wider category

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Negative words in news headlines raised clicks by about 2.3% per word in a large field test, negative news raised physiological arousal on average, fear-arousing messages shifted attitudes (average d = 0.29), and status-quo protest coverage made viewers more critical of protesters.Hypothesis: catastrophe and disorder words such as "chaos" may make readers see an event as more threatening or a group as more disorderly than the facts support.
Limit: The studies tested negative words, fear messages and whole-story framing, not these specific terms, and clicks and arousal are not the same as misjudging the facts.		Severity capped at 2 points per word.Pending: the rating panel has not yet checked this group.

chaos 2chaotic 2nightmare 2catastrophe 1catastrophic 1devastating 1meltdown 2implodes 2imploded 2panic 1out of control 2brutal 1lawless 2lawlessness 2

War and conflict metaphors

Evidence: Category: studies tested the wider category

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Describing crime as a beast rather than a virus shifted readers toward enforcement in one set of studies, though a later replication found no effect; describing cancer as an enemy to fight lowered some prevention intentions.Hypothesis: war metaphors in news may push readers toward combative, us-versus-them reasoning about a policy issue.
Limit: The research tested extended metaphors about crime and illness, not single words such as "bombshell" in political news, and the core finding has a failed replication.		Severity capped at 2 points per word.Pending: the rating panel has not yet checked this group.

war on 2all-out war 2crusade 2onslaught 2bombshell 2explosive 1

Mockery and humiliation

Evidence: Indirect: related studies only

What the studies foundHow it applies to news wordingThe site's choiceLocal test
In social media posts from news outlets and members of Congress, words referring to the opposing party raised sharing (each out-group term raised the odds by 67%) and drew angry and mocking reactions.Hypothesis: describing a figure as mocking or humiliating an opponent may frame news as a contest between sides and invite the same out-group animosity.
Limit: The study measured mocking reactions from users, not the effect of a reporter using mockery verbs, and it measured sharing rather than reader judgment.		Severity capped at 1 point per word.Pending: the rating panel has not yet checked this group.

mocks 1mocked 1dunks 1dunked 1humiliating 1humiliation 1humiliates 1

Delegitimizing labels and accusations in reporter voice

Evidence: Indirect: related studies only

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Loftus (1975) found that wording that presupposed an object led more people to report seeing it; Gentzkow and Shapiro (2010) found that partisan phrases track a newspaper's slant; Recasens et al. (2013) found that Wikipedia editors remove one-sided and truth-presupposing words.Hypothesis: a label such as "witch hunt" or "hoax" in reporter voice may lead readers to accept a contested claim as settled.
Limit: None of these studies tested whether these labels in news change readers' beliefs; they show presupposition effects in memory, measure slant, and describe editing practice.		Since method 3.0.0 these words are scored under word choice as contentious labels, on the basis of the newsroom rule, not reader research.Pending: the rating panel has not yet checked this group.

witch hunt 2sham 2hoax 2cover-up 2coverup 2rigged 2weaponized 2weaponization 2tyranny 2tyrannical 2

Mental-illness insults

Evidence: Indirect: related studies only

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Utych (2018) found that dehumanizing descriptions of immigrants (animal and pest imagery) made readers' attitudes more negative, with anger and disgust carrying the effect.Hypothesis: mental-illness insults such as "unhinged" may, like dehumanizing labels, lower readers' regard for the person described.
Limit: No study located tested mental-illness insults in news; the link to dehumanizing imagery is by analogy only.		Since method 3.0.0 these words are part of the insults group (G12) and are experimental: shown, not scored.Pending: the rating panel has not yet checked this group.

insanity exp.lunacy exp.lunatic exp.unhinged exp.deranged exp.

Shock and surprise boosters

Evidence: Indirect: related studies only

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Molyneux and Coddington (2020) found that clickbait headlines lowered readers' ratings of a story's credibility and quality; Berger and Milkman (2012) found that articles evoking high-arousal emotions such as awe were more likely to be emailed.Hypothesis: shock adjectives such as "stunning" may make a report read as clickbait and lower its credibility, while raising arousal.
Limit: Neither study isolated these adjectives; clickbait was tested as a whole headline style and arousal was measured at the article level.		Severity capped at 1 point per word.Pending: the rating panel has not yet checked this group.

stunning 1shocking 1jaw-dropping 1staggering 1whopping 1epic 1dramatic 1dramatically 1

Degree and size intensifiers

Evidence: Category: studies tested the wider category

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Hosman (1989) found that hedges and hesitations lowered a speaker's perceived authority, while intensifiers had mixed effects depending on the other speech features; Recasens et al. (2013) found that Wikipedia editors remove such words as framing bias.Hypothesis: degree intensifiers such as "massive" may inflate how large or important readers judge a development to be.
Limit: Hosman tested spoken testimony and judgments of the speaker, not news readers' judgments of facts, and the intensifier results were mixed.		Severity capped at 2 points per word.Pending: the rating panel has not yet checked this group.

very 1extremely 1totally 1truly 1deeply 1hugely 1massive 1massively 1huge 1enormous 1sweeping 1

Rebuke and setback words

Evidence: None found

What the studies foundHow it applies to news wordingThe site's choiceLocal test
No published study was located that tests rebuke or setback words.Hypothesis: words such as "rebuke" or "major blow" may frame routine events as defeats and make a story read as a contest.
Limit: No research backs this group; it is on the list as a newsroom style rule.		Experimental: shown, not counted.Pending: the rating panel has not yet checked this group.

rebuke exp.rebuked exp.rebukes exp.major blow exp.blow to exp.scrambles exp.scrambling exp.

No published study found yet.

Softening words (euphemism and agent removal)

Evidence: Direct: a study tested this kind of wording on readers

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Walker et al. (2021) found that actions described with euphemisms such as "enhanced interrogation" were judged more acceptable without costing the speaker credibility, and Moreno-Medina et al. (2025) found that "officer-involved" and passive wording in police-killing stories lowered readers' assignment of responsibility; Fausey and Boroditsky (2010) found that agent-free descriptions lowered blame.Hypothesis: softening words and agent-removing phrases in news may lead readers to judge actions as less serious and assign less responsibility.
Limit: Only a few terms were tested directly; others such as "gaffe", "pushback" and "sparked debate" have no reader-effect study, and the agent-removal effect replicated in English with mixed results in Croatian.		Severity capped at 3 points per word.Pending: the rating panel has not yet checked this group.

officer-involved 2collateral damage 2misspoke 2enhanced interrogation 2kinetic action 2mistakes were made 2misstep 2gaffe 2downsizing 2right-sizing 2revenue enhancement 2neutralized 2friendly fire 2raised eyebrows 2sparked debate 2pushback 2misspeaks 2controversial remarks 2missteps 2gaffes 2raises eyebrows 2sparks debate 2

Insults and name-calling

Evidence: Category: studies tested the wider category

What the studies foundHow it applies to news wordingThe site's choiceLocal test
Experiments found that incivility in political messages, including name-calling such as "you're an idiot" in reader comments, lowered political trust (meta-analysis g = -0.19, a small effect), polarized risk views by prior attitudes, raised anger, and lowered engagement and news source credibility in some studies; one study found an uncivil blog made an adjacent news story look more credible.Hypothesis: a reporter's own insults may lower readers' trust in the people described and in the outlet, and may polarize readers' views along existing lines.
Limit: The studies tested incivility by politicians, commenters, bloggers and conflict in news stories, not a reporter insulting people in their own voice, and none tested these single words in isolation.		Experimental: shown on report pages, not counted, until the rating panel has checked it.Pending: the rating panel has not yet checked this group.

moron exp.morons exp.idiot exp.idiots exp.liar exp.liars exp.loser exp.losers exp.clown exp.clowns exp.buffoon exp.buffoons exp.crook exp.crooks exp.traitor exp.traitors exp.scum exp.insanity exp.lunacy exp.lunatic exp.lunatics exp.unhinged exp.deranged exp.

Studies

  1. Moral-emotional words (replication and meta-analysis). Each additional moral-emotional word was associated with more sharing, with a smaller effect than first reported. Incidence rate ratio 1.12 (95% confidence interval 1.06–1.19): about 12% more shares per moral-emotional word.

    Evidence: Strong. This is the replication. Listed as a successful replication of the original study in the FORRT FLoRA Replication Atlas.

    Brady, W. J., Rathje, S., Globig, L., & Van Bavel, J. J. (2025). Estimating the effect size of moral contagion in online networks: A pre-registered replication and meta-analysis. PNAS Nexus. Source

  2. Outrage expression is reinforced by likes and shares. Positive social feedback for outrage made users more likely to express outrage later; users also matched the outrage norms of their networks.

    Evidence: Moderate. Preregistered; no independent replication found in this review.

    Brady, W. J., McLoughlin, K., Doan, T. N., & Crockett, M. J. (2021). How social learning amplifies moral outrage expression in online social networks. Science Advances, eabe5641. Source

  3. Words referring to the political out-group. Posts about the opposing party were shared about twice as often as posts about one's own party; out-group references drew angry and mocking reactions. Each out-group term raised the odds of sharing by 67%. Out-group language had about 4.8 times the effect of negative-affect words and 6.7 times the effect of moral-emotional words.

    Evidence: Strong. Large dataset across two platforms and two sources (media, Congress). Observational, so causation is not established.

    Rathje, S., Van Bavel, J. J., & van der Linden, S. (2021). Out-group animosity drives engagement on social media. PNAS, e2024292118. Source

  4. Negative words in headlines. Headlines with more negative words were clicked more; positive words lowered clicks. Sad words raised clicks; joy and fear words lowered them; anger words had no significant effect. Each negative word raised click-through rate by about 2.3% (relative) for an average-length headline; each positive word lowered it by about 1.0%.

    Evidence: Strong. Registered Report format (analysis plan reviewed before full data). Effects declined across 2013–2015.

    Robertson, C. E., Pröllochs, N., Schwarzenegger, K., Pärnamets, P., Van Bavel, J. J., & Feuerriegel, S. (2023). Negativity drives online news consumption. Nature Human Behaviour, 7, 812–822. Source

  5. Negative news stories and body arousal. Negative stories produced larger skin-conductance and heart-rate responses than positive stories, across countries; individuals varied widely and many showed no change. Reported as average differences in normalized skin conductance and heart rate variability; large individual variation.

    Evidence: Strong. Partly replicated: Dubèl et al. (2024), Media and Communication, found suggestive support that negative news raises skin conductance (https://doi.org/10.17645/mac.7807).

    Soroka, S., Fournier, P., & Nir, L. (2019). Cross-national evidence of a negativity bias in psychophysiological reactions to news. PNAS, 116(38), 18888–18892. Source

  6. High-arousal emotion (anger, awe, anxiety) in articles. Articles evoking high-arousal emotions (awe, anger, anxiety) were more likely to be emailed; sadness (low arousal) made sharing less likely. Lab studies showed arousal explained the effect. One standard deviation more anger: +34% odds of making the most-emailed list; awe: +30%; sadness: about -16%.

    Evidence: Moderate. Widely cited; no direct replication located in this review. Robertson et al. (2023) found anger words in headlines did not raise clicks, so the anger effect may apply to sharing more than clicking.

    Berger, J., & Milkman, K. L. (2012). What makes online content viral? Journal of Marketing Research. Source

  7. Violent verb for the same event. A more violent verb raised estimates of speed and led people a week later to falsely remember broken glass. Mean speed estimates: smashed 40.8 mph, collided 39.3, bumped 38.1, hit 34.0, contacted 31.8. False memory of broken glass: smashed 32%, hit 14%, control 12%.

    Evidence: Moderate. Mixed for direct replications. Goldschmied et al. (2017) failed in a sport-commentary version (https://doi.org/10.1080/17470218.2016.1237980). The broader misinformation effect (later wording changes memory) is among the most replicated findings in memory research. Small samples in the original.

    Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behavior, 13(5), 585–589. Source

  8. Presupposition in a question or clause. Wording that assumed an object existed led more people to remember seeing it a week later. Asking about "the" item instead of "a" item raised reports of seeing it. 17.3% reported seeing a nonexistent barn vs 2.7% of controls.

    Evidence: Moderate. Part of the misinformation-effect literature, widely replicated in concept. DOI not confirmed in this review (author PDF linked).

    Loftus, E. F. (1975). Leading questions and the eyewitness report. Cognitive Psychology, 7, 560–572. Source

  9. Crime described as a beast vs a virus. Readers given the beast metaphor favored enforcement and punishment more; readers given the virus metaphor favored reform and prevention more. 71% favored more enforcement after the beast frame vs 54% after the virus frame in one study.

    Evidence: Mixed. Mixed. Steen, Reijnierse & Burgers (2014, PLoS ONE, https://doi.org/10.1371/journal.pone.0113536) found no metaphor framing effect. Christmann & Göhring (2016, Scientific Data, https://doi.org/10.1038/sdata.2016.98) replicated the main effect in German.

    Thibodeau, P. H., & Boroditsky, L. (2011). Metaphors we think with: The role of metaphor in reasoning. PLoS ONE, 6(2), e16782. Source

  10. War and enemy metaphors. Describing cancer as an enemy to fight lowered intentions to adopt self-limiting prevention behaviors and did not raise intentions for screening or treatment.

    Evidence: Moderate. No direct replication located.

    Hauser, D. J., & Schwarz, N. (2015). The war on prevention: Bellicose cancer metaphors hurt (some) prevention intentions. Personality and Social Psychology Bulletin, 41(1), 66–77. Source

  11. Agentive vs non-agentive description. Readers assigned more blame and higher financial penalties when the same accident was described with a person as the doer.

    Evidence: Moderate. Mixed. Tonković et al. (2022), Legal and Criminological Psychology, preregistered direct replication, reproduced the effect in English with mixed results in Croatian (https://doi.org/10.1111/lcrp.12219).

    Fausey, C. M., & Boroditsky, L. (2010). Subtle linguistic cues influence perceived blame and financial liability. Psychonomic Bulletin & Review, 17(5), 644–650. Source

  12. Obscuring language in police-killing coverage. News coverage of police killings often used passive and "officer-involved" phrasing. In an experiment, readers who saw this obfuscating language were less likely to hold the officer morally responsible or to support penalties, especially when the victim was unarmed.

    Evidence: Moderate. Single paper; peer reviewed in a top journal. DOI taken from the RePEc handle.

    Moreno-Medina, J., Ouss, A., Bayer, P., & Ba, B. A. (2025). Officer-involved: The media language of police killings. Quarterly Journal of Economics, 140(2), 1525–1580. Source

  13. Attribute framing of a single figure. The positively labeled product was rated better; the effect was smaller after tasting.

    Evidence: Strong. Attribute framing is well replicated in later work.

    Levin, I. P., & Gaeth, G. J. (1988). How consumers are affected by the framing of attribute information before and after consuming the product. Journal of Consumer Research, 15(3), 374–378. Source

  14. Dehumanizing language about immigrants. Dehumanizing descriptions made attitudes toward immigrants more negative, with anger and disgust carrying the effect.

    Evidence: Moderate. Consistent with the broader dehumanization literature; no direct replication located.

    Utych, S. M. (2018). How dehumanization influences attitudes toward immigrants. Political Research Quarterly, 71(2), 440–452. Source

  15. Fear appeals. Fear-arousing messages changed attitudes, intentions and behavior; effects were larger when the message also said what to do about the threat. Average d = 0.29 (a small-to-moderate effect).

    Evidence: Strong. Meta-analysis; consistent with earlier meta-analyses.

    Tannenbaum, M. B., Hepler, J., Zimmerman, R. S., Saul, L., Jacobs, S., Wilson, K., & Albarracín, D. (2015). Appealing to fear: A meta-analysis of fear appeal effectiveness and theories. Psychological Bulletin, 141(6), 1178–1204. Source

  16. Partisan phrase choice ("death tax" vs "estate tax"). Members of Congress from each party use distinct phrases for the same policy; newspapers' phrase choices track their readers' politics. This measures slant; it does not measure emotional effect on readers. 2005: Republicans said "death tax" 365 times and "estate tax" 46 times; Democrats 35 and 195.

    Evidence: Strong. Method widely reused.

    Gentzkow, M., & Shapiro, J. M. (2010). What drives media slant? Evidence from U.S. daily newspapers. Econometrica, 78(1), 35–71. Source

  17. Euphemism vs plain or harsh term. Actions described with euphemisms were judged more acceptable, and speakers who used them were seen as honest, so euphemism shifts judgment without costing credibility. Knowing the facts reduced the effect.

    Evidence: Moderate. Four internal studies; no external replication located.

    Walker, A. C., Turpin, M. H., Meyers, E. A., Stolz, J. A., Fugelsang, J. A., & Koehler, D. J. (2021). Controlling the narrative: Euphemistic language affects judgments of actions while avoiding perceptions of dishonesty. Cognition, 104633. Source

  18. Hedges, hesitations and intensifiers. Hedges and hesitations lowered perceived authoritativeness; intensifiers interacted in complex ways with the other markers.

    Evidence: Mixed. Part of a larger "powerless language" literature with mixed results for intensifiers.

    Hosman, L. A. (1989). The evaluative consequences of hedges, hesitations, and intensifiers: Powerful and powerless speech styles. Human Communication Research, 15(3), 383–406. Source

  19. Epistemological bias: factive verbs, assertives, one-sided terms. Wikipedia editors removed words that presuppose truth (revealed, exposed), assert certainty (claimed, pointed out), or carry one-sided framing, in order to restore a neutral point of view. Readers and a model could locate the biased word about a third of the time on the first guess. Top-1 accuracy at finding the biased word: humans 37.4%, model 34.4%.

    Evidence: Moderate. Widely reused in media-bias detection work.

    Recasens, M., Danescu-Niculescu-Mizil, C., & Jurafsky, D. (2013). Linguistic models for analyzing and detecting biased language. Proceedings of ACL 2013, 1650–1659. Source

  20. Clickbait headlines and credibility. Clickbait headlines lowered readers' ratings of the story's credibility and quality.

    Evidence: Moderate. No direct replication located.

    Molyneux, L., & Coddington, M. (2020). Aggregation, clickbait and their effect on perceptions of journalistic credibility and quality. Journalism Practice, 14(4), 429–446. Source

  21. Headline that misstates the body. Readers given a misleading headline remembered the article's facts less accurately and judged people in the story differently, even after reading the full text. Fact-quiz scores 46% with a misleading headline vs 58% with a matching headline (factual articles).

    Evidence: Moderate. Small samples; consistent with later headline-correction research.

    Ecker, U. K. H., Lewandowsky, S., Chang, E. P., & Pillai, R. (2014). The effects of subtle misinformation in news headlines. Journal of Experimental Psychology: Applied, 20, 323–335. Source

  22. Status-quo framing of protest. Viewers of more status-quo-supporting coverage were more critical of protesters, more favorable to police, less supportive of protesters' rights, and judged the protest less effective and less newsworthy.

    Evidence: Moderate. Part of the "protest paradigm" literature.

    McLeod, D. M., & Detenber, B. H. (1999). Framing effects of television news coverage of social protest. Journal of Communication, 49(3), 3–23. Source

  23. Uncivil vs civil televised political debate. In three lab experiments, the same political disagreement shown in an uncivil manner lowered viewers' trust in politicians and the political system; disagreement shown civilly did not. Uncivil exchanges also raised viewer interest.

    Evidence: Moderate. Lab samples; the trust effect is consistent with the 2022 meta-analysis (Van't Riet & Van Stekelenburg), which found the effect stronger for video than for other media.

    Mutz, D. C., & Reeves, B. (2005). The new videomalaise: Effects of televised incivility on political trust. American Political Science Review, 99(1), 1–15. Source

  24. Name-calling in reader comments under a news blog post ("you're an idiot"). Uncivil comments had no direct effect on risk perceptions of nanotechnology, but they polarized readers: those with low support saw more risk and those with high support saw less. Interaction coefficients beta = .09 (support x incivility) and beta = -.07 (religiosity x incivility); the interaction block added 1.0% of explained variance. N = 1,183 in the nanotechnology condition of a national U.S. sample.

    Evidence: Moderate. No direct replication located in this review. The incivility was in fictional reader comments, not the article, and the effects were small interactions rather than a main effect.

    Anderson, A. A., Brossard, D., Scheufele, D. A., Xenos, M. A., & Ladwig, P. (2014). The "nasty effect:" Online incivility and risk perceptions of emerging technologies. Journal of Computer-Mediated Communication, 19(3), 373–387. Source

  25. Uncivil vs civil mediated political conflict in three European countries. Uncivil conflict lowered intention to participate politically and support for the policies discussed, with similar effects in the Netherlands, the United Kingdom and Spain. It did not raise cynicism toward the EU. People with low tolerance for disagreement were most affected.

    Evidence: Moderate. Cross-national design replicates the effect across three countries within one study; no independent replication located.

    Otto, L. P., Lecheler, S., & Schuck, A. R. T. (2020). Is context the key? The (non-)differential effects of mediated incivility in three European countries. Political Communication, 37(1), 88–107. Source

  26. Uncivil political posts on an online message board. Uncivil posts from the other side made participants angry and averse and less satisfied with the discussion; uncivil posts from their own side made participants more uncivil in their own comments. Effects appeared mainly for emotional, histrionic incivility.

    Evidence: Moderate. Single experiment; no direct replication located.

    Gervais, B. T. (2015). Incivility online: Affective and behavioral reactions to uncivil political posts in a web-based experiment. Journal of Information Technology & Politics, 12(2), 167–185. Source

  27. Political incivility (meta-analysis of experiments). Across 35 experiments, incivility lowered political trust by a small amount and had almost no effect on political participation. The trust effect was stronger for video and in U.S. studies. Trust: Hedges' g = -0.19 (95% CI -0.30 to -0.09). Participation: g = -0.02 (95% CI -0.11 to 0.08).

    Evidence: Strong. This is the meta-analysis. The authors report some sign of publication bias, so the true trust effect may be smaller.

    Van't Riet, J., & Van Stekelenburg, A. (2022). The effects of political incivility on political trust and political participation: A meta-analysis of experimental research. Human Communication Research, 48(2), 203–229. Source

  28. Name-calling and insults as perceived incivility. Readers rated name-calling, insults and rude language as more uncivil than failures of deliberation such as refusing to compromise; these grouped together as personal-level incivility. Politicians from the reader's own party were rated as more civil.

    Evidence: Moderate. Two studies (N = 258 and N = 1,024) found the same structure. Measures what readers perceive as uncivil, not the effect on their attitudes.

    Muddiman, A. (2017). Personal and public levels of political incivility. International Journal of Communication, 11, 3182–3202. Source

  29. Civil vs uncivil online political news content. Civil news drew more engagement than news with the most incivility; articles with several types of incivility, or that showed the out-group party acting uncivilly, discouraged engagement.

    Evidence: Moderate. Two experiments within the paper; no independent replication located.

    Muddiman, A., Pond-Cobb, J., & Matson, J. E. (2020). Negativity bias or backlash: Interaction with civil and uncivil online political news content. Communication Research, 47(6), 815–837. Source

  30. Uncivil vs disagreeable conflict news. Uncivil conflict news raised perceived incivility, which raised engagement through surveillance emotions but also lowered engagement directly and by lowering the news source's credibility.

    Evidence: Moderate. Single experiment; no independent replication located.

    Muddiman, A., & Scacco, J. M. (2024). The influence of conflict news on audience digital engagement. Journalism Studies, 25(3), 278–298. Source

  31. News story placed inside an uncivil partisan blog post. A news story embedded in an uncivil, partisan blog post was rated as more credible, apparently by contrast with the rude blog around it.

    Evidence: Moderate. Single experiment; no independent replication located. The incivility was in the surrounding blog, not the news story.

    Thorson, K., Vraga, E., & Ekdale, B. (2010). Credibility in context: How uncivil online commentary affects news credibility. Mass Communication and Society, 13(3), 289–313. Source

  32. Uncivil elite exchanges vs issue polarization. Uncivil exchanges between politicians lowered political trust and raised affective polarization (dislike of the other party) but did not polarize policy attitudes. Issue polarization polarized attitudes and raised affective polarization but did not lower trust.

    Evidence: Moderate. Separates incivility from disagreement, in line with Mutz and Reeves (2005); no direct replication located.

    Skytte, R. (2021). Dimensions of elite partisan polarization: Disentangling the effects of incivility and issue polarization. British Journal of Political Science, 51(4), 1457–1475. Source

  33. Perceived incivility among politicians (panel survey). In a Danish panel, citizens who perceived more incivility among politicians lost trust in politicians and satisfaction with democracy; the effects came from incivility, not from issue conflict.

    Evidence: Moderate. Panel survey of over 6,000 respondents with a natural experiment; observational, not a wording experiment.

    Bøggild, T., & Jensen, C. (2024). When politicians behave badly: Political, democratic, and social consequences of political incivility. American Journal of Political Science. Advance online publication. Source