IQ Metrics
IIF Certified Assessment Start IQ Test
IIF Certified Assessment
Start IQ Test
Data studyReasoning tests and puzzles · Version 2026.10

The world’s shortest IQ test: how many people really pass it?

The bat-and-ball puzzle is often called the world’s shortest IQ test, and headlines say only 17% pass. We found every published result we could read in the original and pooled them: how many people solve it, by country and by kind of sample, and whether it has got easier. Every sample in one open table.

By Leon Falkner, Data & Research Analyst · Reviewed by Aarav Vardhan, Editor-in-Chief · Published 26 October 2026 · 9 min read

Download data Cite this study

About this study

Report type
Data study: meta-analysis of published results
Data
91 samples from 46 papers, 2004 to 2025
Covers
117,989 people in 26 countries
Method
Random-effects meta-analysis of proportions (logit, REML)
Licence
CC BY 4.0 (our table)
Version
2026.10
Sources
50
Short answer

The “world’s shortest IQ test” is the 3-question Cognitive Reflection Test (2005), whose first item is the bat-and-ball problem. Across 91 published samples of 117,989 people in 26 countries, 35% solved the bat-and-ball problem and 21% got all three right. The famous “17%” comes from 3,428 mostly US students tested in 2003-2005.

91 samples26 countriesBat-and-ball problemCC BY 4.0
35%solve the bat-and-ball problem (pooled)
21%get all three questions right (pooled)
17%the famous figure: 3,428 mostly US students, 2003-2005

01 Key findings

  1. 35% of people solve the bat-and-ball problem (pooled across 28 samples, 87,225 people, 26 countries; 95% CI 27% to 44%).IQ Metrics Research meta-analysis
  2. 21% get all three questions right (61 samples). The famous “only 17%” is from Frederick’s 3,428 mostly US students, 2003-2005.IQ Metrics Research; Frederick (2005), Table 1
  3. Who is tested changes the answer: university students 51%, online panels 37% and other adults 20% on the bat-and-ball problem.IQ Metrics Research meta-analysis
  4. Over time: the share answering 5 cents rose by about 26 points per decade (95% CI 1 to 51).IQ Metrics Research meta-regression
  5. It tracks intelligence tests, loosely: correlation .43 with the Wonderlic (921 people), .46 with the ACT.Frederick (2005), Table 4

02 Statistics to quote

35%

Of people in 28 samples solved the bat-and-ball problem (pooled; 95% CI 27% to 44%).

IQ Metrics pooled estimate
21%

Got all three questions right (pooled across 61 samples).

IQ Metrics pooled estimate
17%

All three right in Frederick’s original 3,428 people, mostly US university students, 2003-2005.

Frederick, 2005
.43

Correlation between the 3 questions and the Wonderlic intelligence test (921 people).

Frederick, 2005
26

Countries with at least one published sample in our table (117,989 people in all).

IQ Metrics

The bat-and-ball problem is easy to solve and hard to notice. Most people who miss it could do the sum; they just never check the answer that pops into their head.

Leon Falkner, IQ Metrics Research

03 What is the "world's shortest IQ test"?

It is the Cognitive Reflection Test (CRT), three short maths questions published by Shane Frederick, then at MIT, in 2005 [15]. Each has an answer that jumps to mind and is wrong. The first is the most famous:

The bat-and-ball problem

“A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?” The quick answer is 10 cents. The right answer is 5 cents: then the bat costs $1.05, which is $1.00 more, and the two add up to $1.10.

The other two follow the same pattern: one about machines making widgets (answer 5 minutes, not 100) and one about a patch of lily pads that doubles every day (answer 47 days, not 24). Frederick wrote that “nearly everyone” who does not answer 10 cents gives the right answer, so the test mainly catches whether people stop to check [15].

Is it really an IQ test?

Not quite, but it is related. In Frederick’s data the three questions correlated .43 with the Wonderlic Personnel Test (921 people), .44 with the SAT and .46 with the ACT [15]. He called it a test of “cognitive reflection”: the ability or habit of resisting the first answer that comes to mind. Its newspaper nickname, the world’s shortest IQ test, overstates that.

04 Where does "only 17%" come from?

From Frederick’s own table. He gave the test to 3,428 people in 35 studies over 26 months from January 2003; most were undergraduates at universities in the US Midwest and Northeast, paid $8 for a 45-minute questionnaire. Overall, 17% got all three right and 33% got none; the average was 1.24 out of 3 [15]. The share with a perfect score ran from 48% at MIT to 5% at the University of Toledo.

Figure 1

Where "only 17%" comes from: Frederick's 2005 samples

Share who got all three questions right, by place tested, 2003-2005. Most were paid students at universities in the US Midwest and Northeast. The 17% is the total of all 3,428.

Source: Frederick (2005), Table 1, p. 29 [15]

The full spread is just as telling. At MIT almost half got all three right and 7% got none; at the University of Toledo it was the other way round, with 64% getting none right [15].

Figure 2

How many of the three each group got right, in Frederick's 2005 data

Share of each group scoring 0, 1, 2 or 3 out of 3. One bar per place tested; the last bar is all 3,428 people. Counts are worked back from Frederick's rounded percentages.

None rightOne rightTwo rightAll three right

Source: Frederick (2005), Table 1, p. 29 [15]

So the 17% describes paid students at a handful of US universities two decades ago, not the world. Headlines that turn it into “only 17% of people can pass” drop all of that.

05 How many people get it right?

About 35% for the bat-and-ball problem, and about 21% for all three questions. Those are random-effects pooled estimates across 28 samples (bat and ball) and 61 samples (all three), 2004 to 2025. Samples differ a lot: the 95% prediction interval for the bat-and-ball problem runs from 6% to 81%, so a new study could land almost anywhere in that range.

Figure 3

Every sample we found, one by one

Share of each sample (square; area = sample size) with its 95% interval, and the pooled random-effects estimate (diamond). The dashed line is the 95% prediction interval: where a new sample would likely land. Pick a measure.

% of the sample

Source: IQ Metrics Research from 46 published papers (Sources); per-sample intervals are Wilson 95% intervals

Where papers give an average score, the middle sample averages 1.33 out of 3 (90 samples, 117,532 people; sample averages run from 0.37 to 2.77). About 32% of people get none of the three right.

Figure 4

Average scores of every sample, from 0 to 3

Number of samples by their average score out of 3 (90 samples). Dotted line: the middle sample. Frederick's 2005 average was 1.24.

Source: IQ Metrics Research from the papers listed under Sources

The trap works the way Frederick described. In the 22 samples that report both answers, the people who miss the problem nearly all write 10 cents; the middle sample has just 6% giving any other answer, blanks included.

Figure 5

Almost everyone who gets it wrong says "10 cents"

Answers to the bat-and-ball problem in the 22 samples that report both the right answer and the trap answer. One bar per sample, highest solve rate at the top.

10 cents (the trap)5 cents (right)Something else

Source: IQ Metrics Research from the papers listed under Sources (people = share x sample size; “something else” = the rest, incl. blanks)

06 Do students do better than everyone else?

It depends which students. University students solved the bat-and-ball problem 51% of the time (11 samples) and online panels 37% (7 samples). National or representative samples, the closest thing to a whole population, came in at 13% (one sample: a YouGov panel of 1,750 US adults, Kahan 2013). Students at selective universities score far higher than students elsewhere, as Frederick’s own table shows.

Figure 6

Who gets the bat-and-ball problem right, by kind of sample

Pooled share answering 5 cents, with the 95% confidence interval. Random-effects pooling of the samples in each group.

Source: IQ Metrics Research from the samples in Figure 2

The same holds for a perfect score and for the average. Within each kind of sample the results still range widely, from groups where hardly anyone gets all three to groups where most do.

Figure 7

Share getting all three right, sample by sample

Each dot is one sample; the shape shows where samples bunch; the navy tick is the median; the orange diamond is the pooled estimate.

Source: IQ Metrics Research from the samples in Figure 2

Figure 8

Average score out of 3, by kind of sample

Each dot is one sample's average. Box = middle half of samples; line = median; whiskers = lowest and highest sample.

Dotted line: Frederick 2005: 1.24.

Source: IQ Metrics Research from the papers listed under Sources

07 Which countries score highest?

Comparisons between countries are rough, because each country’s samples are different kinds of people. Among countries with at least 200 people tested, the highest pooled share with a perfect score is Australia (50%) and the lowest France (6%). Australia’s samples include mathematics students and mathematics academics, while France’s and Spain’s come from economics students and laboratory subject pools, so the gap says more about the samples than the countries. Treat the map as a list of what has been measured, not a ranking of nations.

Figure 9

Share getting all three questions right, by country

% of each country's samples with a perfect score (several samples in one country are pooled). 23 countries. Most samples are students or online panels, not a whole country, so read the colours as a rough guide.

40% or more30 to 40%20 to 30%10 to 20%Under 10%No sample found

Source: IQ Metrics Research from the samples in Figure 2

Figure 10

All three right: countries with at least 200 people tested

Pooled share with a perfect score, with the 95% confidence interval. Number of samples in brackets. Wide intervals mean few or very different samples.

Source: IQ Metrics Research from the samples in Figure 2

08 Has the puzzle got easier since it went viral?

Partly. The bat-and-ball problem has been in books, articles and job interviews for two decades, and many online panel members have seen it. Across our samples, the share answering 5 cents rose by about 26 points per decade (95% CI 1 to 51) (a mixed-effects meta-regression on the year of data, 20 samples, 2008 to 2023). Later years also bring different kinds of samples, so this describes the published record; it does not prove the puzzle got easier for the same kind of person.

Figure 11

Has the puzzle got easier since it went viral?

Each bubble is one sample: the year the data were collected (or the year before publication) and the share answering 5 cents. Bubble area = people tested. Colour = kind of sample. Samples pooled over many years are left out.

StudentsOnline panelsNational samplesOther adults
People tested1001k5k

Source: IQ Metrics Research from the samples in Figure 2

Having seen the questions clearly helps. In five studies that asked, people who had met the questions before scored 0.4 to 0.9 of a point higher out of 3.

Figure 12

People who have seen the questions before score higher

Average score out of 3 for people who said they had never seen the questions and for people who had. Meyer et al.: seen none of the three vs seen all three. Bialek & Pennycook print the share of items right; we multiplied by 3.

Never seen themSeen them before

The gap is 0.4 to 0.9 of a point, out of 3.

Source: Haigh (2016), Thomson & Oppenheimer (2016), Stieger & Reips (2016) Table 2, Meyer, Zhou & Frederick (2018) Table 2, Bialek & Pennycook (2018) Table 1 [19] [47] [43] [28] [3]

Studies that tracked the same pool of people tell a more careful story:

  • In a 2019 US MTurk sample (n about 1,003 who answered the exposure question), 65.6% said they had seen the bat-and-ball problem before, and 93.7% of those said they met it on MTurk; only 6.9% named a lecture/class. [49]
  • In a French-speaking online volunteer sample (France), 95 of 147 participants (64.6%) reported having come across the original bat-and-ball problem before. [14]
  • Across six earlier datasets (2,494 MTurk workers and undergraduates), about a third said they had seen a CRT item before (921 of 2,494 = 36.9%, computed by us); they scored higher: mean proportion correct .55 vs .34 for the unexperienced (d = .57). Scores of experienced MTurk workers .58 / .52 vs unexperienced .40 / .39. [3]
  • Tracking 14,053 MTurk workers who took the CRT 1 to 25 times (Nov 2013 to Apr 2015), each extra exposure added only about 0.024 items correct on average, versus 0.829 items implied by self-reported prior exposure. [28]
  • Across 23,226 CRT scores from 18,852 unique MTurk workers (11 studies), each earlier exposure raised the number correct by only 0.14 points on average; mean score of workers who appear once (first-and-only appearance) was 1.19. [42]
  • Time trend on first appearances (computed from the posted dataset): bat-and-ball correct 42.9% (Nov 2013, n=982), 37.8% (Mar-Jun 2014, n=4,864), 36.6% (Sep 2014-Jan 2015, n=5,518); mean CRT 1.43, 1.37, 1.32. No rise over that 14-month window among new MTurk workers. [28]
  • In a very large US web-volunteer sample collected ‘from 2009 to 2018’ in eight batches (n = 63,648), the share answering the bat-and-ball correctly stayed between 8.9% and 12.5% in every batch (batch A 9.4%, B 11.1%, C 10.6%, F 8.9%, H 10.1%, I 10.1%, J 10.4%, K 12.5%), i.e. no visible rise over the decade in this convenience sample. Batch dates are not given in the paper, so batches cannot be placed on a calendar. [6]
  • Same item, different panels: bat-and-ball 43.5% correct on US MTurk (2016, n=457) versus 11.7% in a commercial Qualtrics online panel (2018, n=1,238); mean CRT 1.93 (MTurk, Aug 2018, n=700) versus 0.45 (Qualtrics panel). The author concludes the MTurk improvement is not a general trend of media coverage or teaching. [49]
  • The 69-country COVID-19 project (ICSMP) used re-skinned CRT items on purpose: the authors say the original CRT is too widely known and answers are easy to find online. [31]

09 Every sample in one table

Sortable. Click a column heading to sort. The CSV adds the year, language, how the test was given, the exact sentence or table cell each number came from, and a link to it.

Table 1

Every sample in the analysis

Frederick 2005 [15]United StatesStudents61–48%2.18
Frederick 2005 [15]United StatesStudents121–26%1.63
Frederick 2005 [15]United StatesOther adults195–26%1.53
Frederick 2005 [15]United StatesStudents746–25%1.51
Frederick 2005 [15]United StatesStudents51–20%1.43
Frederick 2005 [15]United StatesStudents1,267–14%1.18
Frederick 2005 [15]United StatesOnline panels525–13%1.10
Frederick 2005 [15]United StatesStudents52–12%0.87
Frederick 2005 [15]United StatesStudents154–6%0.83
Frederick 2005 [15]United StatesStudents118–6%0.79
Frederick 2005 [15]United StatesStudents138–5%0.57
Oechssler et al. 2009 [32]GermanyStudents56455%42%2.05
Campitelli & Labollita 2010 [8]ArgentinaOther adults157–6%0.66
Koehler & James 2010, Experiment 1 [22]CanadaStudents84–27%1.67
De Neys et al. 2011 [11]BelgiumStudents403––1.12
Iyer et al. 2012 [21]United StatesOther adults9,721––1.75
Liberali et al. 2012, Study 1 [27]BrazilStudents25931%19%1.10
Liberali et al. 2012, Study 2 [27]United StatesStudents19048%26%1.50
Nieuwenstein & van Rijn 2012, Experiment 1 [29]NetherlandsStudents48––1.27
Baldi et al. 2013 [2]ItalyStudents168––1.29
Fehr & Huck 2013/2016 [12]GermanyStudents24032%–1.49
Kahan 2013 [24]United StatesNational samples1,75013%–0.65
Stupple et al. 2013, Experiment 1 [44]United KingdomStudents65––1.32
Stupple et al. 2013, Experiment 2 [44]United KingdomStudents49––1.12
Weller et al. 2013, Study 1 [48]United StatesOther1,970––0.83
Albaity et al. 2014 [1]MalaysiaStudents88030%–0.88
Bosch-Domenech et al. 2014 [4]SpainStudents623–8%0.74
Pennycook et al. 2014, Experiment 2 [33]CanadaStudents200––1.08
Royzman et al. 2014, Study 1 [39]United StatesOnline panels570–25%1.34
Browne et al. 2015 [5]AustraliaOther1,256––0.45
Corgnet et al. 2015 [9]United StatesStudents264––1.22
Finley et al. 2015 [13]United StatesStudents410––0.74
Graffeo et al. 2015 [17]ItalyStudents35––1.40
Grehl & Tutic 2015 [18]GermanyOther adults269––1.45
Pennycook et al. 2015 [34]CanadaStudents37330%12%1.04
Ponti & Rodriguez-Lara 2015 [36]ItalyStudents96–23%1.28
Thoma et al. 2015, Study 1 [46]United KingdomProfessionals44––2.73
Thoma et al. 2015, Study 2 [46]Not statedProfessionals53––1.62
Thoma et al. 2015, Study 2 [46]Not statedOther adults57––1.33
Barcellos et al. 2016 [37]BrazilProfessionals4,902–22%1.55
Haigh 2016 [19]Not statedOther adults142–46%1.93
Hanaki et al. 2016 [20]FranceStudents200–6%0.48
Juanchich et al. 2016 [23]United KingdomOnline panels384––0.97
Noori 2016 [30]IranStudents134–60%2.42
Noori 2016 [30]IranStudents115–46%2.17
Noori 2016 [30]IranStudents146–20%1.18
Pineros et al. 2016 [35]United StatesProfessionals187–22%1.41
Ring et al. 2016 [38]GermanyStudents131––1.76
Sartarelli 2016 [40]SpainStudents432–8%0.70
Stieger & Reips 2016 [43]several countriesOther adults2,27228%–1.41
Tay et al. 2016 [45]IrelandStudents12856%44%2.02
Thomson & Oppenheimer 2016 [47]Not statedOnline panels200––1.73
Thomson & Oppenheimer 2016, Study 2 [47]United StatesStudents14348%–1.39
Wood et al. 2016 [50]United KingdomStudents14874%57%2.31
Gervais et al. 2018 [16]AustraliaStudents143–25%1.28
Gervais et al. 2018 [16]ChinaOther adults207–33%1.80
Gervais et al. 2018 [16]FinlandOther adults992–33%1.76
Gervais et al. 2018 [16]Hong KongStudents128–21%1.37
Gervais et al. 2018 [16]IndiaOnline panels224–24%1.42
Gervais et al. 2018 [16]MauritiusOther adults161–11%1.13
Gervais et al. 2018 [16]NetherlandsStudents212–12%1.12
Gervais et al. 2018 [16]New ZealandStudents160–20%1.24
Gervais et al. 2018 [16]SingaporeStudents161–32%1.83
Gervais et al. 2018 [16]United Arab EmiratesStudents145–6%0.63
Gervais et al. 2018 [16]United KingdomStudents150–24%1.30
Gervais et al. 2018 [16]United StatesStudents591–11%0.77
Meyer et al. 2018 [28]Not statedOnline panels98243%29%1.43
Meyer et al. 2018 [28]Not statedOnline panels4,86438%26%1.37
Meyer et al. 2018 [28]Not statedOnline panels5,51837%23%1.32
Sirota & Juanchich 2018 [41]United KingdomOnline panels14740%28%1.50
Woike 2019 [49]United StatesOnline panels45744%––
Woike 2019 [49]United StatesOnline panels70058%47%1.93
Woike 2019 [49]Not statedOnline panels1,23812%5%0.45
Cresswell & Speelman 2020 [10]AustraliaStudents62–26%1.63
Cresswell & Speelman 2020 [10]AustraliaStudents27–70%2.40
Cresswell & Speelman 2020 [10]AustraliaStudents34–71%2.68
Cresswell & Speelman 2020 [10]AustraliaStudents44–70%2.56
Cresswell & Speelman 2020 [10]AustraliaProfessionals30–77%2.77
Byrd et al. 2024, preprint [6]United StatesOther adults63,64810%4%0.48
Byrd et al. 2024, preprint [6]United KingdomOther adults50831%20%1.15
Byrd et al. 2024, preprint [6]IndiaOther adults56039%30%1.53
Byrd et al. 2024, preprint [6]CanadaOther adults36627%18%1.14
Byrd et al. 2024, preprint [6]AustraliaOther adults24330%19%1.30
Byrd et al. 2024, preprint [6]PhilippinesOther adults1968%3%0.37
Byrd et al. 2024, preprint [6]South AfricaOther adults17611%2%0.41
Byrd et al. 2024, preprint [6]NigeriaOther adults1179%2%0.37
Li et al. 2024 [26]ChinaStudents43868%–1.71
Li et al. 2024 [26]ChinaStudents12087%–2.57
Campitelli et al. 2025 [7]ArgentinaStudents25––0.76
Campitelli et al. 2025 [7]ArgentinaProfessionals25––2.08
Kumari et al. 2026 [25]IndiaOther adults300––1.06

Source: As cited in the first column. Every row, with the sentence or table cell the number came from, is in the CSV.

Table 2

Samples we found but left out

StudySamplesCountriesPeopleWhy
Arechar et al. 20231616 countries32,920Reworded or re-numbered questions, not the original three
Azevedo et al. 20231313 countries9,980Reworded or re-numbered questions, not the original three
Arechar & Rand 20213United States9,626Computed from a data file with no stated licence, and the paper does not print the item wording
Sinayev & Peters 2015, Study 11United States1,413Reworded or re-numbered questions, not the original three
Alper et al. 20211Turkey1,088Computed from the authors’ data file, which has no stated licence; the paper does not print these figures
Knezevic et al. 20241Serbia1,003Computed from the authors’ data file, which has no stated licence; the paper does not print these figures
Morsanyi et al. 2014, Experiment 11Italy328Reworded or re-numbered questions, not the original three
Morsanyi et al. 2014, Experiment 21Italy184Reworded or re-numbered questions, not the original three
Akiyama et al. 20131Japan173Reworded or re-numbered questions, not the original three

Source: IQ Metrics Research

10 What these numbers cannot tell you

Read every number with these in mind

  • Almost no sample is a random draw of a country. Most are university students or paid online panels.
  • Papers report different measures. A sample counts toward a measure only if the paper printed it (or its open data allow it).
  • Translations, paper vs screen, time limits and payment for right answers all vary between studies.
  • People who have seen the questions before score higher, and familiarity has grown since 2005.
  • Three questions are a narrow test. A low score does not mean low intelligence, and a high score is not an IQ.
  • We searched widely but not every paper is open. Samples we could not read in the original are not in the table.

Try a full test instead

Three questions measure one habit. Our IQ test measures reasoning across many kinds of questions and places your score against adults, with a percentile.

Take the IQ test

11 How was this study built?

  1. 01

    Find the papers

    We searched Europe PMC, OpenAlex and Crossref for studies that gave the Cognitive Reflection Test, used the study list of a 2019 meta-analysis as leads only, and added multi-country studies with open data.

  2. 02

    Read every number in the original

    Each figure was copied from the paper’s own text or table (or computed from its open data), with the sentence or table cell saved next to it. Numbers from reviews or meta-analyses were not used.

  3. 03

    Clean

    One row per independent sample. Duplicates, retracted papers and versions with changed numbers or wording were left out (translations kept). Each row was checked by script: sample size, shares that add up, years.

  4. 04

    Pool

    For each measure, a random-effects meta-analysis of proportions: logit-transformed shares, between-sample variance by restricted maximum likelihood, 95% confidence and prediction intervals. Countries pooled the same way. Trend: mixed-effects meta-regression on the year of data.

  5. 05

    Publish

    Every row is in one CSV under CC BY 4.0, with the source of each number. Our standards are on the methods page.

Sources

  1. Cambridge University Press (CUP). Albaity, Rahman et al. Cognitive reflection test and behavioral biases in Malaysia. Judgment and Decision Making.. 2014. doi.org/10.1017/s1930297500005507Facts cited (open access)
  2. Central Library of the Slovak Academy of Sciences. Baldi, Iannello et al. Cognitive reflection and socially biased decisions. Studia Psychologica.. 2013. doi.org/10.21909/sp.2013.04.641Facts cited (open access)
  3. Springer Science and Business Media LLC. Bialek, Pennycook. The cognitive reflection test is robust to multiple exposures. Behavior Research Methods.. 2017. doi.org/10.3758/s13428-017-0963-xFacts cited
  4. Elsevier BV. Bosch-Domènech, Brañas-Garza et al. Can exposure to prenatal sex hormones (2D:4D) predict cognitive reflection? Psychoneuroendocrinology.. 2014. doi.org/10.1016/j.psyneuen.2014.01.023Facts cited
  5. Public Library of Science (PLoS). Browne, Thomson et al. Going against the Herd: Psychological and Cultural Factors Underlying the ‘Vaccination Confidence Gap’. PLOS ONE.. 2015. doi.org/10.1371/journal.pone.0132562Facts cited (open access)
  6. Center for Open Science. Byrd, Stich et al. Analytic Atheism & Analytic Apostasy Across Cultures. Preprint.. 2024. doi.org/10.31234/osf.io/qrz9gOpen data, facts cited
  7. MDPI AG. Campitelli, Labollita et al. Planning, Cognitive Reflection, Inter-Temporal Choice, and Risky Choice in Chess Players: An Expertise Approach. Journal of Intelligence.. 2025. doi.org/10.3390/jintelligence13030040Facts cited (open access)
  8. Cambridge University Press (CUP). Campitelli, Labollita. Correlations of cognitive reflection with judgments and choices. Judgment and Decision Making.. 2010. doi.org/10.1017/s1930297500001066Facts cited (open access)
  9. Public Library of Science (PLoS). Corgnet, Hernán Gonzalez et al. Cognitive Reflection and the Diligent Worker: An Experimental Study of Millennials. PLOS ONE.. 2015. doi.org/10.1371/journal.pone.0141243Facts cited (open access)
  10. Public Library of Science (PLoS). Cresswell, Speelman. Does mathematics training lead to better logical thinking and reasoning? A cross-sectional assessment from students to professors. PLOS ONE.. 2020. doi.org/10.1371/journal.pone.0236153Facts cited (open access)
  11. Public Library of Science (PLoS). De Neys, Novitskiy et al. Cognitive Control and Individual Differences in Economic Ultimatum Decision-Making. PLoS ONE.. 2011. doi.org/10.1371/journal.pone.0027107Facts cited (open access)
  12. Cambridge University Press (CUP). Fehr, Huck. Who knows it is a game? On strategic awareness and cognitive ability. Experimental Economics.. 2016. doi.org/10.1007/s10683-015-9461-0Facts cited
  13. Public Library of Science (PLoS). Finley, Tang et al. Revisiting the Relationship between Individual Differences in Analytic Thinking and Religious Belief: Evidence That Measurement Order Moderates Their Inverse Correlation. PLOS ONE.. 2015. doi.org/10.1371/journal.pone.0138922Facts cited (open access)
  14. Ubiquity Press, Ltd.. Franiatte, Boissin et al. Adieu Bias: Debiasing Intuitions Among French Speakers. Psychologica Belgica.. 2024. doi.org/10.5334/pb.1260Facts cited
  15. American Economic Association. Frederick. Cognitive Reflection and Decision Making. Journal of Economic Perspectives.. 2005. doi.org/10.1257/089533005775196732Facts cited (open access)
  16. Cambridge University Press (CUP). Gervais, van Elk et al. Analytic atheism: A cross-culturally weak and fickle phenomenon? Judgment and Decision Making.. 2018. doi.org/10.1017/s1930297500007701Facts cited (open access)
  17. Frontiers Media SA. Graffeo, Polonio et al. Individual differences in competent consumer choice: the role of cognitive reflection and numeracy skills. Frontiers in Psychology.. 2015. doi.org/10.3389/fpsyg.2015.00844Facts cited (open access)
  18. Public Library of Science (PLoS). Grehl, Tutić. Experimental Evidence on Iterated Reasoning in Games. PLOS ONE.. 2015. doi.org/10.1371/journal.pone.0136524Facts cited (open access)
  19. University of Economics and Human Sciences in Warsaw. Haigh. Has the Standard Cognitive Reflection Test Become a Victim of Its Own Success? Advances in Cognitive Psychology.. 2016. doi.org/10.5709/acp-0193-5Facts cited (open access)
  20. Frontiers Media SA. Hanaki, Jacquemet et al. Fluid Intelligence and Cognitive Reflection in a Strategic Environment: Evidence from Dominance-Solvable Games. Frontiers in Psychology.. 2016. doi.org/10.3389/fpsyg.2016.01188Facts cited (open access)
  21. Public Library of Science (PLoS). Iyer, Koleva et al. Understanding Libertarian Morality: The Psychological Dispositions of Self-Identified Libertarians. PLoS ONE.. 2012. doi.org/10.1371/journal.pone.0042366Facts cited (open access)
  22. Springer Science and Business Media LLC. J. Koehler, James. Probability matching and strategy availability. Memory & Cognition.. 2010. doi.org/10.3758/mc.38.6.667Facts cited
  23. Wiley. Juanchich, Dewberry et al. Cognitive Reflection Predicts Real‐Life Decision Outcomes, but Not Over and Above Personality and Decision‐Making Styles. Journal of Behavioral Decision Making.. 2015. doi.org/10.1002/bdm.1875Facts cited
  24. Cambridge University Press (CUP). Kahan. Ideology, motivated reasoning, and cognitive reflection. Judgment and Decision Making.. 2013. doi.org/10.1017/s1930297500005271Facts cited (open access)
  25. SAGE Publications. Kumari, Hasan et al. Psychological Distress and Bias-relevant Thinking During the Jharkhand Elections Among Young Voters. Annals of Neurosciences.. 2026. doi.org/10.1177/09727531261417818Facts cited (open access)
  26. MDPI AG. Li, Yan et al. Does the Cognitive Reflection Test Work with Chinese College Students? Evidence from a Time-Limited Study. Behavioral Sciences.. 2024. doi.org/10.3390/bs14040348Facts cited (open access)
  27. Wiley. Liberali, Reyna et al. Individual Differences in Numeracy and Cognitive Reflection, with Implications for Biases and Fallacies in Probability Judgment. Journal of Behavioral Decision Making.. 2011. doi.org/10.1002/bdm.752Facts cited (open access)
  28. Cambridge University Press (CUP). Meyer, Zhou et al. The non-effects of repeated exposure to the Cognitive Reflection Test. Judgment and Decision Making.. 2018. doi.org/10.1017/s1930297500007683Open data, facts cited
  29. Cambridge University Press (CUP). Nieuwenstein, van Rijn. The unconscious thought advantage: Further replication failures from a search for confirmatory evidence. Judgment and Decision Making.. 2012. doi.org/10.1017/s1930297500003338Facts cited (open access)
  30. Cambridge University Press (CUP). Noori. Cognitive reflection as a predictor of susceptibility to behavioral anomalies. Judgment and Decision Making.. 2016. doi.org/10.1017/s1930297500007634Facts cited (open access)
  31. Springer Science and Business Media LLC. Nurse, Ross et al. Analytic thinking predicts accuracy ratings and willingness to share COVID-19 misinformation in Australia. Memory & Cognition.. 2021. doi.org/10.3758/s13421-021-01219-5Facts cited
  32. Elsevier BV. Oechssler, Roider et al. Cognitive abilities and behavioral biases. Journal of Economic Behavior & Organization.. 2009. doi.org/10.1016/j.jebo.2009.04.018Facts cited
  33. Springer Science and Business Media LLC. Pennycook, Cheyne et al. Cognitive style and religiosity: The role of conflict detection. Memory & Cognition.. 2013. doi.org/10.3758/s13421-013-0340-7Facts cited (open access)
  34. Springer Science and Business Media LLC. Pennycook, Cheyne et al. Is the cognitive reflection test a measure of both reflection and intuition? Behavior Research Methods.. 2015. doi.org/10.3758/s13428-015-0576-1Facts cited (open access)
  35. Oxford University Press (OUP). Pineros, Doctor et al. Cognitive reflection and antibiotic prescribing for acute respiratory infections. Family Practice.. 2016. doi.org/10.1093/fampra/cmw015Facts cited
  36. Frontiers Media SA. Ponti, Rodriguez-Lara. Social preferences and cognitive reflection: evidence from a dictator game experiment. Frontiers in Behavioral Neuroscience.. 2015. doi.org/10.3389/fnbeh.2015.00146Facts cited (open access)
  37. Associacao Nacional de Programas de Pos-Graduacao em Ciencias Contabeis (ANPCONT). Portugal Barcellos, Lopes Cardoso et al. An assessment of professional accountants' cognitive reflection ability. Advances in Scientific and Applied Accounting.. 2016. doi.org/10.14392/asaa.2016090206Facts cited (open access)
  38. Frontiers Media SA. Ring, Neyse et al. Gender Differences in Performance Predictions: Evidence from the Cognitive Reflection Test. Frontiers in Psychology.. 2016. doi.org/10.3389/fpsyg.2016.01680Facts cited (open access)
  39. Cambridge University Press (CUP). Royzman, Landy et al. Are good reasoners more incest-friendly? Trait cognitive reflection predicts selective moralization in a sample of American adults. Judgment and Decision Making.. 2014. doi.org/10.1017/s1930297500005738Facts cited (open access)
  40. Public Library of Science (PLoS). Sartarelli. Handedness, Earnings, Ability and Personality. Evidence from the Lab. PLOS ONE.. 2016. doi.org/10.1371/journal.pone.0164412Facts cited (open access)
  41. Springer Science and Business Media LLC. Sirota, Juanchich. Effect of response format on cognitive reflection: Validating a two- and four-option multiple choice question version of the Cognitive Reflection Test. Behavior Research Methods.. 2018. doi.org/10.3758/s13428-018-1029-4Open data, facts cited
  42. Cambridge University Press (CUP). Stagnaro, Pennycook et al. Performance on the Cognitive Reflection Test is stable across time. Judgment and Decision Making.. 2018. doi.org/10.1017/s1930297500007695Facts cited
  43. PeerJ. Stieger, Reips. A limitation of the Cognitive Reflection Test: familiarity. PeerJ.. 2016. doi.org/10.7717/peerj.2395Facts cited (open access)
  44. Preprint or report. Stupple et al. 2013, Experiment 1. 2013. escholarship.org/uc/item/36989187Facts cited (open access)
  45. The Association of Faculties of Medicine of Canada. Tay, Ryan et al. Systems 1 and 2 thinking processes and cognitive reflection testing in medical students. Canadian Medical Education Journal.. 2016. doi.org/10.36834/cmej.36777Facts cited (open access)
  46. Public Library of Science (PLoS). Thoma, White et al. Good Thinking or Gut Feeling? Cognitive Reflection and Intuition in Traders, Bankers and Financial Non-Experts. PLOS ONE.. 2015. doi.org/10.1371/journal.pone.0123202Facts cited (open access)
  47. Cambridge University Press (CUP). Thomson, Oppenheimer. Investigating an alternate form of the cognitive reflection test. Judgment and Decision Making.. 2016. doi.org/10.1017/s1930297500007622Facts cited (open access)
  48. Wiley. Weller, Dieckmann et al. Development and Testing of an Abbreviated Numeracy Scale: A Rasch Analysis Approach. Journal of Behavioral Decision Making.. 2012. doi.org/10.1002/bdm.1751Facts cited (open access)
  49. Frontiers Media SA. Woike. Upon Repeated Reflection: Consequences of Frequent Exposure to the Cognitive Reflection Test for Mechanical Turk Participants. Frontiers in Psychology.. 2019. doi.org/10.3389/fpsyg.2019.02646Facts cited (open access)
  50. American Physical Society (APS). Wood, Galloway et al. Can dual processing theory explain physics students’ performance on the Force Concept Inventory? Physical Review Physics Education Research.. 2016. doi.org/10.1103/physrevphyseducres.12.023101Facts cited (open access)

Version history

  • 2026.1026 October 2026First release.
Leon Falkner

Leon Falkner, Data & Research Analyst. Leon owns the arithmetic on IQ Metrics: the percentile tables, the score converter and the figures in everyone else's drafts. On the research desk Leon pulls the data, builds each index and writes the method. More about Leon.

Questions

What is the world's shortest IQ test?

A nickname for the Cognitive Reflection Test, three short questions published by Shane Frederick in 2005. The best known asks the price of a ball when a bat and ball cost $1.10 and the bat costs $1.00 more.

What is the answer to the bat-and-ball problem?

5 cents. The ball costs 5 cents and the bat $1.05, which is $1.00 more, for a total of $1.10. The quick answer, 10 cents, is wrong.

How many people get the bat-and-ball problem right?

About 35%, pooled across 28 published samples in 26 countries. It varies a lot with who is tested.

Is it true that only 17% of people pass?

17% got all three questions right in Frederick's 2005 data: 3,428 people, mostly US university students tested in 2003-2005. Pooled across 61 samples, 21% got all three right.

Is the Cognitive Reflection Test an IQ test?

No, though it is related. Frederick found it correlates .43 with the Wonderlic intelligence test and .44 with the SAT. It mainly measures whether people check their first answer.

Use the data

Share it, cite it, download it

You may reuse everything on this page under CC BY 4.0: in articles, reports, lessons and posts. The one condition is credit to IQ Metrics with a link to this page.

Cite this study

Version 2026.10. Citing the version lets readers find the exact numbers.

Archived copy with a DOI: 10.5281/zenodo.23664398 (report PDF and data on Zenodo).

Falkner, L. (2026). The world's shortest IQ test: how many people really pass it? (Version 2026.10) [Data set]. IQ Metrics Research. https://iqmetrics.org/research/worlds-shortest-iq-test/ (DOI: 10.5281/zenodo.23664398)

Get the data

One row per published sample of the 3-item Cognitive Reflection Test: country, kind of sample, sample size, years, and the share solving the bat-and-ball problem, getting all three right or none, and the mean score, each with the sentence or table cell it came from.

Download CSV
Data dictionary
ColumnWhat it holds
id, study, doi_or_urlOur row id, the paper and its DOI or URL
country, sample_type, sample_desc, nWhere and who, and how many answered
data_year, pub_year, year_usedYear collected (if stated), year published, and the year used in the trend
language, items, mode, incentiveLanguage, item set (original 3 / plus more / bat-and-ball only), paper or screen, paid for right answers
bb_correct_pct, bb_intuitive_pct% answering 5 cents, and 10 cents, on the bat-and-ball problem
all3_pct, zero_pct, mean_score% with all three right, none right, and the mean (0 to 3)
where_in_source, read_from, quote, notesWhere in the paper, the URL read, the exact text holding the number, and checker notes

Share this finding

A finished image and a ready message. The image carries the source line, so the credit travels with it.

Each network gets a version sized for it unless you edit the message.

Press kit

For journalists, researchers and teachers

Everything on this page may be quoted, charted and republished with credit. No embargo, no permission needed, no sign-up. If you need a number that is not here, ask.

Summary you can paste

Plain news style. Edit freely; keep the credit.

The "world's shortest IQ test" is Shane Frederick's 3-question Cognitive Reflection Test (2005). An IQ Metrics Research meta-analysis of 91 published samples (117,989 people, 26 countries, 2004-2025) finds 35% solve its bat-and-ball problem and 21% get all three questions right. The widely repeated "only 17%" is Frederick's figure for 3,428 mostly US university students tested in 2003-2005. Source: IQ Metrics Research, https://iqmetrics.org/research/worlds-shortest-iq-test/

Angles for a story

Quotes you can use

Attributable as written, with the name and title. For a different angle or a quote on a specific figure, ask through the contact page.

  • The bat-and-ball problem is easy to solve and hard to notice. Most people who miss it could do the sum; they just never check the answer that pops into their head.

    Leon Falkner, Data & Research Analyst, IQ Metrics Research

  • The famous 17% is real, but it describes paid students at a few US universities in 2003. Across everything published since, the answer depends heavily on who is tested.

    Leon Falkner, Data & Research Analyst, IQ Metrics Research

  • Three questions measure one habit, checking your first answer. They track IQ tests loosely, and calling them an IQ test oversells them.

    Leon Falkner, Data & Research Analyst, IQ Metrics Research

Files

  • CSV of every sample, with the source of each number (CC BY 4.0)
  • Every figure as a print-ready PNG: use the Image button under each figure
  • Share images for social media in three sizes, with the credit line built in
  • Citations in APA, MLA, Chicago, BibTeX and plain text

Terms, contact and corrections

Credit: “IQ Metrics Research, World’s Shortest IQ Test” with a link to this page. On air or in print, “IQ Metrics” and the page address are enough.

Data credit: per-sample figures from the papers listed under Sources; the table, pooling and charts are ours.

Please say “Cognitive Reflection Test” at least once; it is not an IQ test in the clinical sense.

Contact: the research team, through the contact page. Corrections are made on this page with a dated note in the version history.

Keep reading

Related research

Data studyThe brain numbers everyone quotes, checked
Data studyWhich big employers use aptitude tests?
Data studyWhat teachers believe about the brain: neuromyths by country
Technical reportHow IQ Metrics Research works: methods and standards