Deposition
Backfire Effect
In the matter on the record
The backfire effect is the proposed phenomenon in which correcting a false belief paradoxically strengthens it, especially when the belief is tied to a person's ideology or identity.
Q. The claim on the record
Coined by Nyhan and Reifler (2010), the backfire effect describes corrections that increase rather than reduce the targeted misperception, originally observed when conservatives told no WMD were found in Iraq became more convinced WMD existed. Large preregistered replications (Wood and Porter, 2019; ~10,100 subjects, 52 issues) and a 2020 methodological review (Swire-Thompson, DeGutis, and Lazer) found the effect to be rare and not a robust phenomenon, with most corrections moving beliefs toward accuracy. The current scholarly consensus is that worldview and familiarity backfire effects occur only in narrow, hard-to-reproduce circumstances, and fact-checkers should not avoid corrections out of fear of backfire.
A. For the defense
The effect has largely failed to replicate. Wood & Porter (2019), with ~10,100 subjects across 52 issues, found no corrections capable of triggering backfire. Swire-Thompson, DeGutis & Lazer (2020) reviewed the literature and concluded backfire effects are 'not a robust empirical phenomenon,' noting failures even under theoretically favorable conditions and that measurement error plus regression to the mean can manufacture spurious backfires. The familiarity backfire effect rests largely on an unpublished Skurnik, Yoon & Schwarz (2007) manuscript and has repeatedly failed to replicate. Some narrow, attitude- or behavior-level effects (e.g., Nyhan & Reifler 2015 vaccine intentions in high-concern subgroups) persist but are inconsistent. The current consensus is that backfire, if it exists, is rare and context-specific, not the norm.
Exhibit A — Nyhan & Reifler (2010) — WMD correction
Among conservatives, the correction increased rather than decreased belief that Iraq had WMD — the canonical 'backfire' result. Corrections also broadly failed to reduce misperceptions in the targeted group. — Brendan Nyhan, Jason Reifler, 2010
not reported in standardized form (paper reports OLS interaction coefficients on a 5-point belief scale; no Cohen's d given) · approx. 130 undergraduate participants in the WMD experiment (student samples, fall 2005 / spring 2006)Exhibit B — Wood & Porter (2019) — 'The Elusive Backfire Effect'
No corrections triggered backfire across the issues tested. The only prior backfire-prone item (WMD in Iraq) backfired only with an elaborate survey item and disappeared with a simpler item. By and large, citizens updated toward facts. — Thomas Wood, Ethan Porter, 2019
not reported as a single effect size; finding is a near-null/absence of backfire across 52 issues · approx. 10,100 subjects across the five published experiments (the August 2016 working paper reported 8,100 subjects / 36 issues)Exhibit C — Nyhan & Reifler (2015) — flu-vaccine correction
The correction reduced belief in the myth overall, but among respondents with high concern about vaccine side effects it REDUCED intent to vaccinate — a backfire on the behavioral outcome rather than the factual belief. — Brendan Nyhan, Jason Reifler, 2015
not reported as standardized effect size (subgroup interaction reported) · 822 US adultsA. The mechanism in dispute
The proposed mechanism is motivated reasoning: when a correction threatens a belief central to identity, ideology, or status, people counterargue against it, generating supportive thoughts that strengthen the original belief (the 'worldview backfire effect'). A separate proposed mechanism is the 'familiarity backfire effect,' where repeating the myth inside a correction boosts its later fluency so the myth feels true. Both are contested: methodological critics argue many apparent backfires are artifacts of measurement error and regression to the mean rather than genuine belief strengthening.
Q. On redirect
Swire-Thompson, DeGutis & Lazer (2020) argue backfire is largely a measurement artifact and not robust. Wood & Porter (2019) attribute the appearance of backfire to specific survey-item wording. The familiarity backfire account is also frequently conflated with the well-established illusory truth effect (repetition increasing belief in the ABSENCE of a correction), which is a distinct and robust phenomenon.
Matter of record — politics / public opinion
Iraq WMD belief among conservatives (Nyhan & Reifler 2010). In the original experiments, conservative subjects shown a correction that no weapons of mass destruction were found in Iraq became MORE likely to believe Iraq had had WMD before the 2003 invasion — the canonical documented backfire instance.
Matter of record — medicine / public health
Flu-vaccine correction reduced vaccination intent (Nyhan & Reifler 2015). Correcting the myth that the flu vaccine can cause the flu successfully reduced the false belief overall, but among US adults most concerned about side effects it lowered their stated intention to get vaccinated — a documented behavioral backfire in a 822-person experiment.
The ruling — replication failed
Not supported at scale
Correct misinformation anyway — the evidence says corrections mostly work. Use a clear, fact-focused message that leads with the truth, repeats the myth minimally, and explains why the myth is wrong; avoid framing that attacks the audience's identity. Measure belief change with reliable, multi-item scales and account for regression to the mean before concluding a backfire occurred.
Certified by Brendan Nyhan, Jason Reifler, 2010 — When Corrections Fail: The Persistence of Political Misperceptions. Reporter's note: The original Nyhan & Reifler (2010) paper reports interaction coefficients on Likert belief scales rather than standardized effect sizes, so no Cohen's d is available for the canonical WMD finding; the 'effect size' fields reflect this honestly. The published Wood & Porter (2019) sample figures (~10,100 subjects / 52 issues) come from the Springer landing page and secondary summaries; the primary PDF I extracted was the August 2016 working-paper version reporting 8,100 subjects / 36 issues, so the exact published totals carry slightly lower confidence than the qualitative null result, which is well established. The familiarity backfire literature leans heavily on an unpublished 2007 Skurnik et al. manuscript I could not retrieve directly (cited only via secondary reviews), so its specifics are reported as secondary. The 2021 Nyhan PNAS article was confirmed via search metadata but its full text was not retrieved (HTTP 403).
Cited authorities
- Belief perseveranceparentThe broader tendency to cling to a belief after its evidentiary basis is discredited; backfire is the stronger (and weaker-supported) claim that beliefs actually intensify.
- Motivated reasoningmechanistically-linkedThe proposed engine of worldview backfire: defending identity-relevant conclusions by counterarguing disconfirming evidence.
- Continued influence effectsiblingMisinformation keeps influencing reasoning even after a clear retraction — robustly replicated, unlike backfire; people don't necessarily believe MORE, they just fail to fully discard.
- Illusory truth effecteasily-confusedRepetition increases perceived truth in the ABSENCE of a correction; robust and well-replicated, and often conflated with the much weaker familiarity backfire effect.
- Confirmation biassiblingPreferential seeking and weighting of belief-consistent evidence; related but does not require beliefs to strengthen after disconfirmation.
- Familiarity backfire effectchildA specific sub-type attributing backfire to repeating the myth within the correction; weakly supported, largely failed to replicate.