Notes
Epigraphs
The only thing I ever learned in school: Parker, 2009, 15n.
Fourteen percent believe Goebbels said it: The statistic is from my multiple-choice survey, conducted in the US. The survey offered six options (Goebbels, Oscar Wilde, Voltaire, Adolf Hitler, Winston Churchill, and Mark Twain). The quote does not appear in any of Goebbels’s writings. It was attributed to Goebbels in a 1946 report of the House Un-American Activities Committee and has since been widely cited. The most popular guess in my survey was Mark Twain, chosen by thirty-eight percent.
Introduction: Facts Are Obsolete
I have to show you my first edition Torah: Amy Miller, “New J-Lo Movie Makes ‘First Edition “Iliad”’ a Thing,” Legal Insurrection, February 4, 2015: bit.ly/1aKNK0j.
was not something I wrote: Hillin, 2015.
Sold at Sotheby’s for £25,000: Sotheby’s, Auction Results, Music, Continental Books, and Manuscripts, June 5, 2013: bit.ly/1GLJ06l.
We were just batting around ideas: Itzkoff, 2011.
1. “I Wore the Juice”
I wore the juice: Morris, 2010.
WOW: David Dunning, interview with author, June 12, 2015.
Gun hobbyists study: Ehrlinger, Johnson, Banner, et al., 2003.
One need not look far: Dunning and Kruger, 1999, 1132.
Unskilled and Unaware of It: Dunning and Kruger, 1999.
is someone we’ve all met: David Dunning, interview with author, June 12, 2015.
John Cleese’s YouTube video on Dunning–Kruger: bit.ly/1os8vxN.
To achieve the required standard to be licensed: Rosen, 2014.
Modern research indicates: Wagstaff, 2013.
Seven states have put cursive writing back in curriculum: Wagstaff, 2013.
Should schoolchildren be taught: Cathcart, 2009.
Now, what I want is, Facts: Dickens, 1859. The quote is from the first paragraph of the first chapter.
the network of information: Hirsch, 1987, 2.
How do you get to Central Square?: Krauss and Glucksberg, 1977.
Yes, well you go down on the subway: Hirsch, 1987, 4.
42 states and DC: Boylan, 2014.
We don’t ever want to educate: Ibid.
Bizarre homework assignments: For examples, see the Facebook page “Common Core Crazy Homework,” on.fb.me/1EetjGk.
Survey on Common Core first-grade curriculum: This was multiple choice, most questions having four options plus “don’t know.”
2010 Pew survey: Pew Research Center, “Who Knows What About Religion,” September 28, 2010, pewrsr.ch/1pD7bxq.
Kent State and Colorado State study: Tauber, Dunlosky, Rawson, et al., 2013.
What is the last name: Ibid.
ETS figures: Educational Testing Service, 2015, 17.
despite having the highest levels: Coley, Goodman, and Sands, 2015, 2.
Eighty-six percent of Americans under thirty: Nielsen, 2014. Figures for China, Russia, and Brazil: Pew Research Center, 2014.
Most Millennials don’t know: Most are based on my own surveys with US participants. For others, see Tauber, Dunlosky, Rawson, et al., 2013.
Rational ignorance: Downs, 1957, 244–46 and 266–71.
8 Degrees with the Worst Return on Investment: An actual Salary.com feature: bit.ly/1e7g5cz.
Study of museum photos: Henkel, 2013.
Research shows source memory is fallible: Zachs, 2015.
Stroop experiment with hungry subjects: Wegner and Ward, 2013, 80.
Trivia experiment with brand names: Wegner and Ward, 2013.
Third-most-popular search containing “my penis”: Stephens-Davidowitz, 2015.
Professional memorizer of Greco-Roman times: Nestojko, Finley, and Roediger, 2013, 321.
As part of a plea agreement: Quoted in Kaczynski, 2013.
The graduate students still see literary theft: Fisher, 2015.
Study at Harvard: Wegner and Ward, 2013, 61.
Annie Hall bit: The McLuhan clip is on YouTube at bit.ly/1Dlq4dD.
84 percent of Americans use Internet: Google Public Data: bit.ly/152hP8f.
Age differences: The correlation between age and correct answers had a p value of <.001 in all three cases. (Don’t know what a p value is? I explain it on page 108.) Sample size was 207 for senators; 445 for Brazil.
2. A Map of Ignorance
Susan Sherman barred from school: Bittenbender, 2014.
Forty-six percent of Britons can find Austria: Sample size of 111. The error for the Austria figure is plus or minus 9.1 percent.
Cartograms: I used the proportion of incorrect responses as a density in the Gastner-Newman diffusion-based algorithm. The software was ScapeToad.
Only five percent: The sample size for the US map surveys was just over one hundred. For the Republic of the Congo, the error was plus or minus 4.3 percent—nearly as big as the value itself.
Eighteen percent think Amazon is in Africa: National Geographic Education Foundation, 2006.
Americans are far from alone in the world: National Geographic Education Foundation, 2006, 6.
Harvard Crimson survey: Lexi M. Del Toro and Bessie X. Zhang, “Roving Reporter: Canada,” November 18, 2013: bit.ly/1DK92YL.
1.9 percent: Tauber, Dunlosky, Rawson, et al., 2013, 1129.
Nine percent don’t know what country New Mexico is in: The question was multiple choice with five options (United States, Mexico, Texas, Guatemala, and South America). The United States was chosen by 91.5 percent with error of plus or minus 3.3 percent. Sample size was 282.
Which of the cities: National Geographic Education Foundation, 2006, 56.
Impressively strong: p<.001 for correlation between knowledge and border-fence support. When education and age are included in a linear regression, knowledge is still highly significant in predicting border-fence support: p=.002.
2.25 out of 10: These figures are from a linear regression in which age and education are held constant at thirty-five years old and four years of university.
Those who couldn’t find Ukraine wanted a border fence: p<.001. Though there was a similar pattern with North Carolina, the data was not statistically significant (p=.277). This survey had 228 subjects.
Dinosaur question and border-fence support: p<.001.
Those who said true wanted border fence: The average border-fence rating was 7.31 out of 10 for true and 4.51 for false.
Survey on UK debt: sample size of 110.
2006 survey on US population: National Geographic Education Foundation, 2006.
net worth…defined as the total value: Norton and Ariely, 2011, 9.
Seniors have eighteen times the wealth of young adults: Schrager, 2014.
top twenty percent of earners makes just over half of total income: Statista, “Shares of Household Income of Quintiles in the United States from 1970 to 2014”: bit.ly/1PkkUng.
Survey of CEO and worker income: Kiatpongsan and Norton, 2014.
Given the consensus among disparate groups: Norton and Ariely, 2011, 12.
Just over half support gun rights: Kohut, 2015.
Decline in crime rate; Pew poll: Cohn, Taylor, Lopez, et al., 2013.
Desvousges and his colleagues: Desvousges, Johnson, Dunford, et al., 1992.
Fifteen percent: Ipsos MORI, 2014.
Fifty-six percent: Ibid.
Fifteen percent unemployed: Ibid.
People come into the country illegally: Ibid.
[P]oll after poll shows that voters: Romano, 2011.
Those who knew federal budget earned more: p=.005 with a sample size of 462.
3. Dumb History
“Life in the UK” survey: Sample size of 162.
Newsweek poll: Newsweek, 2011.
Arizona law requiring students to pass citizenship test: Wilson, 2014.
Contacted by Ford library: Henry Roediger III, interview with author, March 16, 2015.
have never seen an airline “ticket”: Beloit College, Mindset List, 2016 list: bit.ly/1CpeE3G.
Memories of world events influenced by age at the time: Koppel and Berntsen, 2014.
rank historical figures just as Google ranks web pages: Skiena and Ward, Time, 2013.
Error bars: Sample sizes were around 160. Error bars for figures recognized by about half the sample were plus or minus eight percent.
Thirty percent knew who proposed theory of relativity: Tauber, Dunlosky, Rawson, et al., 2013, 1123.
a disease of white people: Gewertz, 2007.
Memory experiment on three wars: Zaromb, Butler, Agarwal, and Roediger, 2013.
“Einstein” insanity quote in The Basic Text of Narcotics Anonymous: See bit.ly/1e0891p; Matt Novak, “9 Albert Einstein Quotes That Are Totally Fake,” Paleofuture, March 14, 2014: bit.ly/17hHKK7; Brown, 1983, 68.
Churchillian drift: The term was coined by British writer and broadcaster Nigel Rees. See Peters, 2009.
radically revisionist view of American history: Deam, 2014.
I’ve had kids tell me: Ibid.
Colorado school board policy: Townes, 2014.
early American history to be less about the Pilgrims: Kurtz, 2014.
Oklahoma committee banned AP history course: Legum, 2015.
Entitled to land of “less advanced” people: Cave, 1996, 35–36.
Roediger “cringed” at American estimates: Henry Roediger III, interview with author, March 16, 2015.
Positive and factual textbook about Hitler: See Philipp Bouhler’s Kampf um Deutschland: Ein Lesebuch für die Deutsche Jugend [The Battle for Germany: A Textbook for the German Youth] (Munich: Zentralverlag der NSDAP, Franz Eher Nachfolger, 1938). A partial translation is online at bit.ly/1fG9qfC.
4. The One-in-Five Rule
Huffington Post article: Luippold, 2010.
2014 survey: Anti-Defamation League, 2014.
Results by nation: I have presented the ADL data a little differently from the way it’s presented on the organization’s website. The site reports that ninety-three percent of Germans had heard of the Holocaust and that, of that group, eleven percent thought it had been exaggerated. I think it’s more useful to be able to compare those who’ve never heard of the Holocaust to those who deny it or think it’s exaggerated. Thus the given values for multiple-choice questions have been multiplied by the number who had heard of the Holocaust. Percentages don’t always add up to 100 percent because of rounding.
Debunkings of the “ten-percent myth”: See, for instance, Boyd, 2008.
Princess Diana study: Wood, Douglas, and Sutton, 2012.
The Apollo moon landings never happened: Lewandowsky, Gignac, and Oberauer, 2014, 2–3.
5. The Low-Information Electorate
It’s hard to think of a major policy dispute: Krugman, 2015.
The Democratic party’s electoral majority: Carl, 2013.
a big chunk of America’s body politic: Krugman, 2015.
Many liberals have a deeply ideological view: Carl, 2013.
Judge Janavs lost to bagel store owner: Garvey and Garrison, 2006.
You know the most frightening thing: Garrison, 2006.
Vice president result better than some surveys: Pew Research Center, “Who Knows What About Religion,” September 28, 2010, pewrsr.ch/1pD7bxq.
No meaningful difference between politics and knowing elected officials: p=.742. My sample was small, however, at 110.
2014 survey: Annenberg Public Policy Center, 2014.
Our Constitution declares that “all men”: Dooley, 2014.
Thomas Paine wrote that “the duty of a patriot”: Gorman, 2015.
You’re the state: Ramer, 2011.
The last time I checked: For an audio recording, go to bit.ly/1GpEhZu.
Twelve percent, four percent chance: Vavreck, 2014.
It is tempting to think that something: Ibid.
Twenty-three percent of Los Angeles voters: Meyerson, 2014.
Eight percent turnout for school board election: Lopez, 2014.
Survey on voting: Ipsos MORI, 2014.
Opinium/ Observer survey: Coman and Helm, 2014.
6. Putting a Price Tag on Facts
Post and his lottery winnings: Sullivan, 2006.
Visitors to his crumbling mansion: Ibid.
Spurious correlations website: tylervigen.com.
As has been endlessly chronicled: See, for instance, the college salary report on Payscale.com.
Score remained significant when education added to model: p=.002 for quiz score.
Knowledge still a highly significant predictor: p=.004.
$75,000 income: Kahneman and Deaton, 2010; Luscombe, 2010.
No significant correlation between knowledge and happiness: p=.371.
Questions from quiz: The full quiz also asked for the capital of Brazil, where a shortstop plays, the name of the Speaker of the House, the third digit of pi, and about how long it takes to double an investment at a seven percent annual return.
7. Elevator-Pitch Science
It almost kind of looks like what the earth looks like: youtu.be/s_5j1mVE8Sk
Science surveys: National Science Foundation, 2006.
God created human beings pretty much: Newport, 2014.
Disgraceful photo of recreational hunter: Luke Lewis, “A Whole Bunch of People on Facebook Thought Steven Spielberg Killed a Real Dinosaur,” BuzzFeed, July 11, 2014: bzfd.it/1dzn4k0.
15 percent believed that humans and dinosaurs coexisted: The margin of error was plus or minus 4.9 percent. Sample size was 204.
Planets survey: Sample size was 121.
We are made of star-stuff: Sagan wrote or mentioned this a number of times, the first apparently being in Sagan, 1973, 189–90.
In many districts, they may have a different perspective: Candisky and Siegel, 2014.
focus on academic and scientific knowledge: Timmer, 2014.
Correlation to gender: p<.001; sample size, 204.
Experiment by Ehrlinger and Dunning: Ehrlinger and Dunning, 2003.
No significant correlation between science knowledge and income: p=.129 with sample size of 204.
4.8 percent of workforce: John F. Sargent Jr, The U.S. Science and Engineering Workforce: Recent, Current, and Projected Employment, Wages, and Unemployment, Congressional Research Service, February 19, 2014: bit.ly/1amEn69.
Pi question had correlation to income: p=.016, with sample size of 124.
Turing-test question: Sample size was 204.
8. Grammar Police, Grammar Hippies
47,000 Wikipedia edits correcting “comprised of”: McMillen, 2015.
Restaurant people are not writers: Black, 2008.
I don’t expect chefs to be writers: Ibid.
GrubHub study: GrubHub, 2013; Satran, 2013.
Misspellings of “pizza Margherita,” other Italian foods: Paolo Rigiroli, “Top Misspelled,” Quattro Formaggi, n.d.: bit.ly/1ThRAOP.
mescaline on menu: Jane Black spotted this at the Yorktown Bistro, Arlington, Virginia.
Fieri mispronunciation of mascarpone: Madison, 2010.
Responses to correct and incorrect the same: The sample size was 222 for the correct menu and 215 for the incorrect one.
Trust is something I know: Memoli, 2013.
No correlations: Grammar quizzes had p=.605 for income and p=.086 for educational level with a sample size of 117; for age, p=.125 with a sample size of 226. The spelling quiz had p=.677 for age with a sample size of 103.
Income failed to correlate: For spelling, p=.079 with a sample size of 103. For grammar, p=.246 with a sample size of 226.
Earned $23,000 more a year: p=.023 with a sample size of 226. Note that the p value is the probability that the income is larger in the population, not that the $23,000-a-year figure is exactly right.
FBI list of Twitter slang: Kleinman, 2014.
Strong and negative correlation with age: It was p<.001 for both a set of eight slang questions and a set of eight acronyms and abbreviations.
Slang and acronyms didn’t correlate with income: p=.878 for acronym questions and p=.579 for slang questions. Both surveys had sample sizes of 107; for mansplain, it was 207.
The Oxford English Dictionary accepts both pronunciations: O’Leary, 2013.
Pronunciations correlate with income: p=.022 with a sample size of 183.
9. Nanofame
One way to describe Mr Jarre’s life: Bilton, 2015.
Turned down $1 million contract to promote “unhealthy food”: Ibid.
PewDiePie makes $4 million in ad sales: Grundberg and Hansegard, 2014.
Californication has been on: January 7, 2014, post on @AnthonyDeVito Twitter feed.
Vine has two hundred stars with a million followers: Williams, 2015.
Hi, Kanye!: New York Post, 2013.
Hip-hop survey: The sample was 261 Americans. The difference in recognition between Kanye West and Pitbull was not statistically significant. The margin of error was plus or minus 5.9 percent for West and plus or minus 5.8 percent for Pitbull.
Hip-hop knowledge correlated inversely with age: p<.001. The average quiz score was about seventy percent for Millennials versus thirty percent for those over sixty.
Correlation to income: p=.033. In a linear regression with age, hip-hop knowledge was no longer significant as a predictor of income; p=.371.
Spotify data analysis: Kalia, 2015.
taste freeze; music relevance: Ibid.
Celebrity endorsements: Said, 2013.
10. Is Shrimp Kosher?
Jews and Mormons outscored Christians: Pew Research Center, 2010.
Which of the following best describes Catholic teaching: Ibid., 70.
4.6 of the Ten Commandments: Stephen Prothero, “Religious Literacy Quiz,” Pew Research Center, 2007: pewrsr.ch/1aLl9bd.
Americans are both deeply religious: Prothero, 2007.
Bush didn’t mean, and was not understood: Sontag, 2007.
Believers an overwhelming majority: Pew Research Center, 2010, 2015.
No matter what the results: Pew Research Center, 2010, 4.
Religious knowledge had no correlations: p=.680 for income, p=.580 for happiness, and p=.839 for being married. The sample size was 118.
Religious knowledge is not necessary: Oppenheimer, 2007.
Last Supper and Gautama Buddha questions: The sample size was 118.
11. Philosophers and Reality Stars
All art is quite useless: In Wilde’s preface to The Picture of Dorian Gray (1890). See e-book at bit.ly/1KfBR0J.
Contemporary art survey: The sample size was 164. The margin of error for the lowest-scoring artists was about four percent.
Auctioned for $26 million: Christie’s, Post-War and Contemporary Art sale results, November 12, 2013: bit.ly/1Ic0g77.
Shibboleth names…by which the privileged: The Tumblr blog 454 W 23rd St New York, NY 10011-2157 coined that term and supplies a list of correct pronunciations. See bit.ly/1yTivoP.
Correlated with income: p=.026 with a sample size of 183.
Estimate of percentages who saw movies: See box-office figures at TheNumbers.com. I’m guesstimating an average $10 ticket price.
Every few weeks, my wife mentions: Greenfeld, 2014.
Recognizing Nabokov correlated with education: p<.001 for education, p=.082 for income. The sample size was 119.
12. Sex and Absurdity
a girl is no longer clean: Semuels, 2014.
Almost $1 billion on abstinence-only programs: Audrey Tang with Matt Itelson, “Center Tests Americans’ Sexual Literacy,” SF State News, July 11, 2005: bit.ly/1OaylEz.
If you have sex, you will get AIDS: Semuels, 2014.
Two of the curricula: Vine, 2008.
The whole lesson here: Post on @AliceDreger Twitter feed, April 15, 2015. See also Nelson, 2015.
Circumcision study: Lilienfeld and Graham, 1958.
Seventy-nine percent of American men circumcised: Xu, Markowitz, Sternberg, and Aral, 2006.
the vast majority of my female students: Proctor and Schiebinger, 2008, 112.
San Francisco started programme for teens’ sex questions in 2006: Allday, 2006.
socioeconomic status, paranoia: Oliver and Weed, 2014.
Gluten-free diet survey: Consumer Reports, 2015, 37.
Jimmy Kimmel bit: The clip is on YouTube at bit.ly/1hxnfrr.
Perception is reality: Consumer Reports, 2015, 40.
Gluten survey: The sample size was 151.
Reality is that which: This quote comes from Dick’s 1978 speech, “How to Build a Universe That Doesn’t Fall Apart Two Days Later.” See bit. ly/1koJyFx.
the knowledgeable more likely to be married: p=.014 for the fifteen-question quiz that also asked about health.
Positive correlation between quiz scores and health: p=.029.
Correlations to health: These questions were on the fifteen-item quiz I discussed, and they were individually predictors of health as well. For Kant: p=.027. Sun: p=.008. The sample size was 445.
13. Moving the Goalposts
One in five Britons don’t know number of players: The UK sports survey had a sample size of 152. US sports survey: sample size of 154.
Sports knowledge correlated with income: For UK survey, p=.039; for US, p=.009.
Women make eighty percent as much: CONSAD Research Corporation, 2009.
Linear regression predicted incomes of £25,122 and £50,920: This survey did not have education data, so figures represent the household incomes of a 35-year-old (of unspecified educational level).
Gender not significant in model: p=.742. For sports knowledge in the model, p=.014. Both figures are from a US survey with sample size of 154.
Difficult quiz had no correlation to income: p=.359 with a sample size of 104.
Reran the “easy” sports quiz: In the second survey, p=.044, and the sample size was 110. The income gap was somewhat bigger: $29,511 per year (household) for those scoring zero percent versus $87,092 for those scoring 100 percent.
Happiness correlation to “easy” sports questions: p=.009 with a sample size of 110. The regression projected an average score of 3.89 out of 10 for someone who scored zero percent versus 6.45 for someone who scored 100 percent.
Wimbledon question: The sample size was 162.
Ted Williams question: p=.024 for correlation to income. The sample size was 117.
14. Marshmallow Test
Eighty-four-year-old woman won $590 million: New York Daily News, June 3, 2013.
What made me play…You pay $1: Piore, 2013.
Fifty-nine percent of British adults bought lottery ticket: British Gambling Prevalence Survey, 2010.
Those who answered correctly had higher incomes: p=.035 with a sample size of 322.
Twenty-four percent happier: The correlation between quiz score and self-reported happiness had p<.001.
2010 study of Federal Reserve Bank of Atlanta: Gerardi, Goette, and Meier, 2010.
In a sale, a shop is selling: Ibid., 11.
financial knowledge gap: Maranjian, 2015.
rule of 72: This financial rule of thumb says you divide 72 by the annual rate of return, in percentage points, to get the number of years it will take to double an investment. In this case you’d divide 72 by 7 to get 10.29 years. The answer is not mathematically exact, but it’s close enough for investors who are usually guesstimating the rate of return anyway.
Reported $32,000 more income: The correlation between the correct answer and income had p<.001. The sample size was 427. For savings, the correlation had p=.012 with a sample size of 322. Happiness ratings, on a scale of 0 to 3, were 2.091 for those who answered the compounding question correctly and 1.815 for those who didn’t (p=.004).
Obesity, crack cocaine, divorce: Casey, Somerville, Gotlib, et al., 2011; Schlam, Wilson, Shoda, 2013.
An uneducated individual armed with a credit card: Pelletier, 2013, 2.
2013 report found that only seven states: Pelletier, 2013.
Financial literacy in college: Matthew Reed and Debbie Cochrane, “Student Debt and the Class of 2011,” Project on Student Debt of the Institute for College Access and Success, October 2012: bit.ly/21wksZ2.
Several recent studies have poured cold water: See S. H., 2014.
Follow-up study on Midwestern high school students: Mandell and Klein, 2009.
No correlation between financial instruction and investment income: Cole and Shastry, 2008.
Those who answered retirement-account question correctly earned more: p=.004. Were happier: p=.012. The sample size was 322.
15. The Value of Superficial Learning
We are what we do: Quoted in Sweet, 2010, 51.
Knowing how to spell prerogative correlates with consensus: p=.019. With supersede: p=.002. The p values are impressive despite the modest sample size of 103.
Asian map test a strong predictor: p=.014 with a sample size of 100.
Survey shown by hollow dot not statistically significant: p=.132 with a sample size of 207.
Both “easy” surveys highly significant: p<.001 with sample sizes of 228 and 207. Other “hard” survey: p=.003 with a sample size of 228.
Chess is an analogy: LoBrutto, 1997, 19.
Simon and chunks of knowledge: Chase and Simon, 1973.
Graham’s invention of Liquid Paper: See the entries for Bette Nesmith Graham at Famous Women Inventors (bit.ly/1Fn7Chu) and Wikipedia.
Imagination is more important than knowledge: Einstein, 1931, 97.
Other examples: For Einstein and Darwin-Wallace, see Asimov, 2014.
Posterior hippocampus is larger: Rosen, 2014.
Money can buy better education: Mohan, 2015.
16. When Dumbing Down Is Smart
mystery meat: Mobley, 2014.
Young people recognize icons: Correlation with age was steeply negative and had p<.001 with a sample size of 106.
2006 study in Annals of Internal Medicine: Davis, Wolf, Bass, et al., 2006.
Inserts are legal, not medical documents: Brown, 2014.
145,000 measles deaths a year: This is the World Health Organization figure for 2013. See bit.ly/1myPji2.
If it takes 5 machines 5 minutes: Frederick, 2005.
MIT requires two semesters of calculus: See the “Admissions” page at the MIT website: bit.ly/1Qfyhfo.
Britain replaced “avoid alcohol” warning: Paddock, 2011.
Seventy-eight percent comprehension for deductible: Loewenstein, Friedman, McGill, et al., 2013.
Consumer mistake of overpaying for low deductibles: Ibid., 860.
Hard-boiled egg, steak survey: The sample size was 268.
Ingredients in bread: The sample size was 121.
$24,000 income difference for oysters question: p=.048 with a sample size of 267.
Half gave the correct answer, three: The sample size was 268.
Tool questions: The sample size was 268.
Thirty-eight percent chose 3,000 miles: The sample size was 268.
7,500 or ten thousand miles, Mobil guarantee: Reed and Montoya, 2014.
Psst…this is your engine speaking: Prange, n.d.
a marketing tactic that dealers use: Reed and Montoya, 2013.
17. Curating Knowledge
Actor Adam Sandler is reported: See the “Celebrity Fake News Hoax Generator” story on FakeAWish.com at bit.ly/1n0YtS7.
seems to make it harder: Miller, 2014.
It might not be all that different: Henry Roediger III, interview with author, March 16, 2015.
cognitive load comment: Ibid.
I was in Paris: Brian Cathcart, interview with author, June 5, 2015.
Quiz of twelve questions: The sample size was 458, and the Fox News score was 56.6 percent plus or minus 4.4 percent at the 95 percent confidence level.
Tabloids are factually accurate: Should you doubt this claim, see Poundstone, 1990, 311–15. I tried to verify all the stories appearing in six weekly US tabloids. I was able to confirm all 196 National Enquirer stories as substantially factual. Even the Weekly World News, famed for made-up stories (love-starved orangutan goes ape for dwarf), was ninety-six percent factual.
Don’t watch the news: Marshall, 2013.
Differing habits of rolling and prime-time news viewers: Vavreck, 2015.
twenty-five pages viewed versus five pages: Ibid.
The average American: “The Facts: Why Digital Detox?” at bit. ly/1aLo1F9. No sources are given for the statistics.
Fox News viewers answered 1.04 questions: Farleigh Dickinson, 2012, 3.
18. The Ice Cap Riddle
Climate scientists believe: Kahan, 2014, 22.
14 percent got North Pole question right: Ibid., 42.
If you are one of us: Kahan, 2012.
Obviously, no one will answer: Kahan, 2014, 29.
Thirty percent say it’s a real threat: Ibid., 12.
this style of reasoning: Ibid., 14.
to refine and enlarge the public views: Quoted in Fishkin, 2006.
Today the New England town meeting: Quoted in Fishkin and Luskin, 2005, 286.
One way to make people recognize their incompetence: Dunning and Kruger, 1999, 1131.
The idea…is to find out what people would think: Aizenman, 2015.
52 to 84 percent shift, 43 to 73 percent: Fishkin, 2006.
More than seventy polls in twenty nations: See website of the Center for Deliberative Democracy, cdd.stanford.edu.
19. The Fox and the Hedgehog
The fox knows many things: Berlin, 1978, 22.
Berlin’s interpretation: Ibid., 23.
McDonald’s, the fox in this scenario: Gara, 2014.
journalists or attentive readers: Tetlock, 2005.
hate group: The Southern Poverty Law Center so terms it. See Wikipedia entry for Stop Islamization of America, bit.ly/1Q91MGd.
Strong correlation between knowledge and “Ground Zero mosque” attitudes: p=.02.
Correlation with US independence: p=.004. DC Comics villain: p=.004. The correct answers are Great Britain and the Joker.
About 2,100 mosques in the United States: See Wikipedia entry “List of Mosques in the United States,” bit.ly/1GBdPeT.
Masjid Manhattan is four blocks from Ground Zero: Barnard, 2010.
Prayer room in World Trade Center: Freedman, 2010.
Individual facts predicted opinions: I should warn readers about multiplicity, a statistical hazard too little appreciated by journalists and decision makers. The convention is that a result is deemed significant when the probability of its occurring purely by random-sampling error is five percent or less. This comes from a traditional conception of science in which collecting data (and forming hypotheses) is labour-intensive. Today it is easy and cheap to collect data online and let software search for any correlations that may exist. This may be tantamount to testing thousands of potential hypotheses. In such contexts p values are less able to separate wheat from chaff. The researcher must expect to find about one “significant” but bogus correlation for every twenty hypotheses checked.
Multiplicity is a potential concern here, as I have selected correlations that are more or less amusing and illustrative of general connections between knowledge and opinions or self-reported behaviours. I have therefore been cautious in favouring single-fact correlations with highly significant p values, representative of correlations between performance on general-knowledge quizzes and beliefs.
Not knowing which famous document: p=.009, sample of 206.
Not knowing what happened in the stock market: p=.038 with sample of 206.
Thinking Alice B. Toklas wrote: p=.03 with sample of 206.
People who didn’t know their MP were more likely to post fake reviews: p=.016
Those who said they’d vaccinate: p<.001 for all other cited opinions; the sample size was 207.
More knowledgeable, more likely to buy the light bulb: p=.020.
Screw question correlated strongly: p<.001 (would vaccinate child); p=.002 (approve reusable bags); p=.009 (would buy $100 light bulb). The sample size was 207.
Seven percent of the British public would throw pet: Sample size of 206. Error was plus or minus 3.6 percent. Correlation with not knowing largest ocean and MPs had p values of .015 and .006 respectively.
Shrimp; reusable bags: p=.005 with a sample size of 207 Americans.
One in five would push the button: The exact result was 18.9 percent plus or minus 4.8 percent. Correlation to knowing or not knowing year of the 9/11 attacks has p=.001. The sample was 254 Americans.