Retrospective warnings that flag news as misinformation can help readers discard false details, but they can also backfire: when the warning is false or overbroad, readers discard accurate information from the warned source — the tainted truth effect. The original political-communication demonstration left open which factors govern this correction-versus-tainting trade-off: the modality of the original event and the post-event description, the difficulty and centrality of the tested facts, the source characteristics and ideological congruence of the original-event source, the post-event article source and the warning issuer, the time intervals between event, description, warning, and test, and generalization across sample compositions and memory-test formats. The problem is to determine experimentally how these factors moderate both the corrective benefit and the collateral accuracy loss of misinformation warnings in political news.

Background

When people witness an event and later read a description containing false details (post-event misinformation), their memory for the original event is systematically distorted — the misinformation effect, studied since the 1970s in eyewitness-memory research. A common countermeasure is the retrospective warning: after people have read the misleading description, they are told that it contained errors and should be discounted. Echterhoff, Groll and Hirst (2007) showed that such warnings have a side cost: people overcorrect, becoming less accurate even about details that were described correctly or never mentioned. They called this the tainted truth effect (TTE): an erroneous warning about the presence of misinformation taints belief in, and memory for, authentic information from the warned source. Blank and Launay's (2014) meta-analysis of post-warning studies confirmed warnings reduce the misinformation effect, while Szpitalak (2017) showed the effect can arise even without an explicit warning. Freeze, Baumgartner, Bruno, Gunderson, Olin, Ross and Szafran (2021) extended the TTE to political communication: participants watched a video of a political event, read a news article about it (accurate or containing misinformation), received or did not receive a retrospective warning, then completed a recognition memory test (choosing the originally witnessed detail among alternatives). Valid warnings only weakly corrected false information, and warnings attached to a source that also carried misinformation reduced belief in that source's accurate statements. Because fact-checking and platform labeling are now pervasive, the accuracy-loss side of warning systems is a live policy concern: Hoes et al. (2024) documented a related spillover, where prominent anti-misinformation interventions increase skepticism toward accurate information, and Horne and Nevo (2025) showed people continue trusting warning labels even when the labels are wrong. Freeze et al. closed by identifying the unresolved design dimensions of their paradigm — the moderators listed above — which define when warnings help versus harm. The relevant conventions for studying them are standard experimental ones: preregistration (time-stamped specification of hypotheses and analyses before data collection), factorial designs crossing warning validity with the candidate moderator, and memory measures separating contaminated items (those the post-event description distorted) from uncontaminated items (those described accurately), so that correction and tainting can be measured simultaneously.

Problem Statement

For political news, determine how the trade-off between the corrective benefit and the collateral accuracy loss of retrospective misinformation warnings — the tainted truth effect — depends on the factors Freeze et al. (2021) left untested. Specifically, measure both misinformation acceptance (correction) and memory/belief for accurate, uncontaminated information (tainting) when experimentally varying: (1) the modality and format of the original event and the post-event description (video, audio, or written, and crossed combinations); (2) the difficulty and centrality of the tested factual items, targeted enough to test the hypothesized nonlinear relationship between memorability and warning potency; (3) the source characteristics and political/ideological congruence of the original-event source, the post-event article source, and the warning issuer (researcher-issued versus partisan or high-credibility fact-checker); (4) the delays between original event, post-event description, warning, and test, extended to ecologically realistic intervals beyond the minutes-long laboratory session; and (5) sample composition (nationally representative versus convenience samples, with designs that limit attrition) and memory-assessment format (recognition versus free recall or forced choice with confidence/uncertainty measures — the latter partially informed by the aggregate test-format moderator evidence in Belli and Blank 2026). The overarching objective is a boundary-condition map of the tainted truth effect: a characterization, family by family, of the configurations under which warnings' corrective gains do, or do not, outweigh their damage to belief in accurate information. The characterization covers all five factor families above, with none of them primary, and, because no single study can vary all five families at once, it is not confined to the scope of one experiment.

References

  1. Freeze, M., Baumgartner, M., Bruno, P., Gunderson, J. R., Olin, J., Ross, M. Q., & Szafran, J. (2021). Fake Claims of Fake News: Political Misinformation, Warnings, and the Tainted Truth Effect. Political Behavior, 43(4), 1433–1465. DOI: 10.1007/s11109-020-09597-3. https://doi.org/10.1007/s11109-020-09597-3
  2. Echterhoff, G., Groll, S., & Hirst, W. (2007). Tainted Truth: Overcorrection for Misinformation Influence on Eyewitness Memory. Social Cognition, 25(3), 367–409. DOI: 10.1521/soco.2007.25.3.367. https://doi.org/10.1521/soco.2007.25.3.367
  3. Blank, H., & Launay, C. (2014). How to protect eyewitness memory against the misinformation effect: A meta-analysis of post-warning studies. Journal of Applied Research in Memory and Cognition, 3(2), 77–88. DOI: 10.1037/h0101798. https://doi.org/10.1037/h0101798
  4. Szpitalak, M. (2017). The indirectly generated tainted truth effect: Warning is not necessary to worsen the testimony of non-misled persons. Psychology, Crime & Law, 23(4), 323–341. DOI: 10.1080/1068316X.2016.1258470. https://doi.org/10.1080/1068316X.2016.1258470 (online first November 2016)
  5. Martel, J., & Rand, D. G. (2023). Misinformation warning labels are widely effective: A review of warning effects and their moderating features. Current Opinion in Psychology, 54, 101710. DOI: 10.1016/j.copsyc.2023.101710. https://doi.org/10.1016/j.copsyc.2023.101710
  6. Martel, J., & Rand, D. G. (2024). Fact-checker warning labels are effective even for those who distrust fact-checkers. Nature Human Behaviour, 8(10), 1957–1967. DOI: 10.1038/s41562-024-01973-x. https://doi.org/10.1038/s41562-024-01973-x
  7. Hoes, E., Aitken, B., Zhang, J., Gackowski, T., & Wojcieszak, M. (2024). Prominent misinformation interventions reduce misperceptions but increase scepticism. Nature Human Behaviour, 8(8), 1545–1553. DOI: 10.1038/s41562-024-01884-x. https://doi.org/10.1038/s41562-024-01884-x
  8. Horne, B. D., & Nevo, D. (2025). People adhere to content warning labels even when they are wrong due to ecologically rational adaptations. Scientific Reports, 15, 13896. DOI: 10.1038/s41598-025-98221-7. https://doi.org/10.1038/s41598-025-98221-7
  9. Torrance, M. G., Karanian, J. M., Race, E., & Thomas, A. K. (2025). Examining the impact of warnings on eyewitness memory. Scientific Reports, 15, 33508. DOI: 10.1038/s41598-025-17377-4. https://doi.org/10.1038/s41598-025-17377-4
  10. Belli, R., & Blank, H. (2026). A Meta-Analysis of the Tainted Truth Effect [data and coding files]. OSF project, created March 24, 2026; no DOI assigned. https://osf.io/3a8dw