When Corrections Fail: The Persistence of Political Misperceptions
Political Behavior, 2010 · paper · Political science
Reports four experiments in which subjects read mock news articles with a misleading politician's claim, with or without a correction. Corrections often failed to reduce misperceptions among the targeted ideological group, and sometimes produced a backfire effect increasing them.
Checked: PDF from bpb-us-e1.wpmucdn.com, fetched 5 Oct 2026. Page numbers are those of that PDF.
2 findings
Reference or label "While Lebo and Cassino interpret backfire effect-type results ... (2007: 723)" has no corresponding entry in Works cited.
- p. 9 Theoretical expectations, body text
While Lebo and Cassino
- p. 9 Theoretical expectations, backfire paragraph
While Lebo and Cassino interpret backfire effect-type results
- p. 36 Works cited
Lau, Richard R. and David Redlawsk (2001)
Judge's note: The p36 reference image goes straight from Lau, Lee and Linzer to Lodge with no Lebo entry. A text search finds 'Lebo' only in the in-text citation (p9).
Second check: another model agreed
Report a problem with this finding
Reference or label Text cites "Bartels 2002: 133-137" (p. 31) but Works cited lists "Bartels, Larry M. (2004). “Beyond the Running Tally: Partisan Bias in Political Perceptions.” Political Behavior 24(2): 117-150." (p. 33).
- p. 31 Conclusion, last line
Bartels 2002: 133-137
- p. 33 Works cited
Bartels, Larry M. (2004). “Beyond the Running Tally
- p. 33 Works cited
Bartels, Larry M. (2004). “Beyond the Running Tally: Partisan Bias in Political Perceptions.”
- p. 33 Works cited
Bartels, Larry M. (2004). “Beyond the Running Tally: Partisan Bias in Political Perceptions.” Political Behavior 24(2): 117-150.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that Political Behavior 24(2) was published in 2002. Both crops checked: the in-text year is 2002 and the reference-list year is 2004. There is no other Bartels entry, so the two years for the one work disagree. Which year is right cannot be settled from the files alone.
Second check: another model agreed
Report a problem with this finding
21 minor (typos and the like; most checked by the judge only)
Wording Abstract and Research design say "We conducted four experiments", but footnote 23 says "We also conducted an experiment correcting a claim made by Michael Moore" (a fifth experiment, not reported).
- p. 1 Abstract
conducted four experiments in which subjects read mock news articles
- p. 13 Research design
we conducted four experiments in
- p. 22 Footnote 23
We also conducted an experiment correcting a claim made by Michael Moore
Judge's note: Abstract (p1) and Research design (p13) both say 'conducted four experiments'. Footnote 23 (p22) describes a further experiment (Unocal/Afghanistan) whose results were not reported. The introduction (p3) says 'we report the results', so 'four' clearly means the four reported experiments. That makes this a wording imprecision, not a substantive error.
Second check: another model agreed
Report a problem with this finding
Wording Introduction: "In each of the four experiments ... ideological subgroups failed to update their beliefs", yet the Study 2 WMD results say "the correction made conservatives more likely to believe that Iraq did not have WMD".
- p. 3 Introduction, final paragraph
In each of the four
- p. 23 Results – Iraq WMD
correction made conservatives more likely to believe that Iraq did not have WMD
- p. 3 Introduction, final paragraph
In each of the four experiments, which were conducted in fall 2005 and spring 2006, ideological subgroups failed to update their beliefs
- p. 23 Results – Iraq WMD (Study 2)
the correction made conservatives more likely to believe that Iraq did not have WMD
Only the authors can say what is right here.
Judge's note: p23 says the correction was effective for conservatives in Study 2 WMD. p24 says 'a backfire effect did not take place among conservatives as a group'. The failure to update appears only in a post hoc issue-importance subgroup of 34 respondents (8 right of centre), and there the effect is null. The blanket 'in each of the four' overstates this experiment. Only the authors can say whether they meant the post hoc subgroup.
Second check: another model agreed
Report a problem with this finding
Wording The Study 1 stimulus is datelined "Wilkes-Barre, PA, October 7, 2004 (AP)" and says the speech was given "here today", yet also says "Bush maintained Wednesday". October 7, 2004 was a Thursday.
- p. 39 Appendix, Study 1 news text
Wilkes-Barre, PA, October 7, 2004 (AP) -- President Bush delivered a hard-hitting
- p. 39 Appendix, Study 1 news text
Bush maintained Wednesday
- p. 39 Appendix, Study 1 news text
Wilkes-Barre, PA, October 7, 2004 (AP)
- p. 39 Appendix, Study 1 news text
Bush maintained Wednesday that the war in Iraq
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that 1 January 2004 was a Thursday, which makes 7 October 2004 a Thursday (2004 is a leap year; 280 days later, exactly 40 weeks), and that the Duelfer Report came out on Wednesday 6 October. The page 39 image confirms both the dateline/'today' and 'Wednesday'. The stimulus is a mock AP article (p. 17), so it may reproduce the source's mixed dating. Only the authors can say what was intended.
Second check: another model agreed
Report a problem with this finding
Wording Table 1, Model 2: the ideology coefficient "0.199***" and its SE "(0.082)" do not correspond under the declared test (*** p < .01, two-sided, OLS, N = 130). The t-ratio is about 2.43, which falls in the ** band.
- p. 47 Table 1, Ideology row, Model 2
0.199***
- p. 19 Results Study 1
conservatives were more likely to agree with the statement (p < .01)
Only the authors can say what is right here.
Judge's note: Checked on the page 47 image. A t of 2.43 on about 124 df gives a two-sided p of roughly .017, above .01. The other Model 2 coefficients match their stars: knowledge 1.081/0.362 ≈ 2.99 and the interaction 0.359/0.127 ≈ 2.83 both earn ***. The text's 'p < .01' for ideology (p. 19) refers to Model 1 (0.347/0.064). The interaction (headline) result is unaffected. The authors should say whether the stars or the SE are misprinted.
Second check: another model agreed
Report a problem with this finding
Wording Footnote 32 subgroups "(n=60)" and "(n=136)" sum to 196, whereas Study 2 had 197 respondents with two excluded ("excludes two subjects with missing data") and Table 3 reports N = 195.
- p. 27 Footnote 32
went from 36% to 67% (n=60). By contrast, for non-conservatives, agreement went from 31% to 28% (n=136)
- p. 23 Footnote 26
Each of the statistical models in Study 2 excludes two subjects with missing data
- p. 22 Text
197 respondents participated in
- p. 27 footnote 32
went from 36% to 67% (n=60). By contrast, for non-conservatives, agreement went from 31% to 28% (n=136).
Only the authors can say what is right here.
Judge's note: Table 3 (p49 image) shows N = 195 for both models. Footnote 26 says one subject answered no dependent-variable question and one gave no ideology. The footnote 32 split sums to 196. A possible explanation is that the subject missing ideology was counted as 'non-conservative' in the raw tabulation, but the paper does not say so; footnote 26 refers only to statistical models. Study 1 footnote 21 sums exactly (33+97=130). This is a small unexplained discrepancy for the authors to clarify.
Second check: another model agreed
Report a problem with this finding
Typo "(see, eg, Kuklinski et al. 2000: 798)" uses 'eg' without periods, whereas elsewhere the paper writes 'e.g.'
- p. 2 paragraph 2
(see, eg, Kuklinski et al. 2000: 798)
Judge's note: The page image (p2) shows 'eg,'. Every other instance in the text layer is 'e.g.' or 'e.g.,'. This is a cosmetic inconsistency.
Second check: none (judge only)
Report a problem with this finding
Typo "(Kuklinkski et al. 2000 [study 1], Sides and Citrin 2007)" misspells Kuklinski.
- p. 7 First line
Kuklinkski et al. 2000 [study 1]
- p. 9 footnote 5
discriminatory faction against blacks
- p. 22 footnote 23
build an natural gas pipeline
Judge's note: Seen in the page 7 image. The name is spelled Kuklinski elsewhere and in the reference list (p36). The 'faction' and 'an natural' typos are handled in C007 and C011.
Second check: none (judge only)
Report a problem with this finding
Typo Footnote 5: "applied in a discriminatory faction against blacks" — 'faction' should be 'fashion'.
- p. 9 Footnote 5
applied in a discriminatory faction against blacks
- p. 9
discriminatory faction against blacks
Judge's note: Seen in the page 9 footnote image.
Second check: none (judge only)
Report a problem with this finding
Typo The ideology scale endpoints are "very liberal"/"very conservative" on p11 but "strongly liberal (-3) to strongly conservative (3)" on p19.
- p. 19 Study 1 Results, first sentence
ranging from strongly liberal (-3) to strongly conservative (3)
- p. 11 Theoretical expectations
a seven-point Likert scale from “very liberal” to “very
Judge's note: Both crops checked. The meaning is clear, so this is a cosmetic labelling inconsistency.
Second check: none (judge only)
Report a problem with this finding
Typo "we designed the experiments to be realistic as possible" is missing an 'as' ('as realistic as possible').
- p. 13 Research design, second paragraph
designed the experiments to be realistic as possible
- p. 13 Research design
we designed the experiments to be realistic as possible
Judge's note: Seen in the page 13 image.
Second check: none (judge only)
Report a problem with this finding
Typo "that President Bush banned on stem cell research" contains a stray 'on'.
- p. 21 Study 2 introduction
that President Bush banned on stem cell
- p. 21 Study 2 design paragraph
that President Bush banned on stem cell research
- p. 21 Study 2 design
President Bush banned on stem cell research
Judge's note: Seen in the page 21 image. Elsewhere the paper writes 'banned stem cell research'.
Second check: none (judge only)
Report a problem with this finding
Typo Footnote 23: "to build an natural gas pipeline" should read 'a natural gas pipeline'.
- p. 22 Footnote 23
build an natural gas pipeline
- p. 22 footnote 23
to build an natural gas pipeline
Judge's note: Seen in the page 22 footnote image.
Second check: none (judge only)
Report a problem with this finding
Typo "they actually has the effect of increasing government revenue" has a subject–verb disagreement ('has' should be 'have').
- p. 25 Method – Tax cuts
they actually has the effect of
- p. 25 Method – Tax cuts
they actually has the effect of increasing government revenue
Judge's note: Seen in the page 25 image.
Second check: none (judge only)
Report a problem with this finding
Typo Missing word in the conclusion: "Many citizens seem or unwilling to revise their beliefs" (presumably "seem unable or unwilling").
- p. 32 Conclusion, final paragraph
Many citizens seem or unwilling to revise their beliefs
Judge's note: The page image shows the text exactly as quoted. The sentence is ungrammatical as printed.
Second check: none (judge only)
Report a problem with this finding
Typo The Cohen et al. (2005) title ends "on Support for George W. Bush in the 2004.”" and is missing a noun after "2004".
- p. 34 Works cited, Cohen et al. (2005)
on Support for George W. Bush in the 2004.”
- p. 34 Works cited
Support for George W. Bush in the 2004.
Judge's note: The page 34 image shows the title as truncated. The exact missing words (e.g. 'Presidential Election') would need checking against the source.
Second check: none (judge only)
Report a problem with this finding
Typo The year parenthesis is never closed in two reference entries: "Druckman, James N. and Kjersten R Nelson (2003. “Framing…" (p. 34) and "Vallone, Robert P., Lee Ross, and Mark R. Lepper (1985. “The Hostile Media…" (p. 38).
- p. 34 Works cited
Druckman, James N. and Kjersten R Nelson (2003. “Framing
- p. 34 Works cited
Druckman, James N. and Kjersten R Nelson (2003.
- p. 34 Druckman and Nelson
Kjersten R Nelson (2003.
- p. 38 Vallone et al.
Mark R. Lepper (1985.
Judge's note: Both page images show '(2003.' and '(1985.' with no closing parenthesis.
Second check: none (judge only)
Report a problem with this finding
Typo Three Works cited entries have no in-text citation: Harris Poll (2006), Wilkes and Leatherbarrow (1988), and Wilkes and Reynolds (1999).
- p. 35 Works cited
Harris Poll (2006). “Belief that Iraq Had Weapons of Mass Destruction Has Increased
- p. 38 Works cited
Wilkes, A.L., and M. Leatherbarrow (1988)
- p. 38 Works cited
Wilkes, A.L. and D.J. Reynolds (1999)
- p. 35 Works cited
Harris Poll (2006). “Belief that Iraq Had Weapons of Mass Destruction
Judge's note: I searched the full text for Harris, Wilkes, Leatherbarrow and Reynolds. Each appears only in the reference list. The only other 'Wilkes' hits are the place name Wilkes-Barre (p. 17, p. 39). This is a minor list/text mismatch.
Second check: none (judge only)
Report a problem with this finding
Typo Misspelling in the Ross and Lepper (1980) title: "The Perserverance of Beliefs" should be "Perseverance".
- p. 38 Works cited, Ross and Lepper (1980)
“The Perserverance of Beliefs
- p. 38 Works cited
The Perserverance of Beliefs
Judge's note: The page 38 image shows 'Perserverance'.
Second check: none (judge only)
Report a problem with this finding
Typo The first author's surname is printed twice: "Shapiro, Robert Y. Shapiro and Yaeli Bloch-Elkon (2008)".
- p. 38 Works cited
Shapiro, Robert Y. Shapiro and Yaeli Bloch-Elkon (2008)
- p. 38 Works cited
Shapiro, Robert Y. Shapiro and Yaeli Bloch-Elkon
Judge's note: The page 38 image shows the duplicated surname. The in-text citation is 'Shapiro and Bloch-Elkon 2008'.
Second check: none (judge only)
Report a problem with this finding
Typo The Sides and Citrin (2007) title has doubled quotation marks: "““How Large the Huddled Masses? … Immigrant Populations.””".
- p. 38 Works cited, Sides and Citrin (2007)
““How Large the Huddled Masses?
Judge's note: The page 38 image shows doubled opening and closing quotation marks, unlike the other entries.
Second check: none (judge only)
Report a problem with this finding
Typo In Table 1, Model 2, the standard-error cell under the constant 3.156*** reads "-0.323", with a minus sign and no parentheses. Every other standard error is positive and in parentheses, e.g. Model 1 "(0.331)".
- p. 47 Table 1, Constant row, Model 2 SE cell
-0.323
- p. 47 Table 1, Model 2, Constant rows
3.156*** -0.323
- p. 47 Table 1, Model 2, Constant
3.156*** / -0.323
Judge's note: Checked on the page 47 image. The cell sits in the SE row under the Model 2 constant. A standard error cannot be negative, so this is almost certainly a formatting slip for (0.323). The coefficient and stars are unaffected (3.156/0.323 is far above the *** cutoff).
Second check: none (judge only)