The IQ of politicians: what conscript tests and college records show

0 The short answer

Politicians score above the people they represent on the few cognitive tests that were ever matched to elected officials, but those tests exist only for men tested at conscription in Sweden and Finland, and the gap changes with the country and the test. In Sweden, members of parliament averaged about 0.84 standard deviations above the male population on the military cognitive score, according to the working paper of Dal Bó and colleagues. In Finland, members of parliament scored about three-fourths of a standard deviation above on verbal reasoning and about one-fourth above on visuospatial reasoning. For the United States no study has IQ scores for legislators, only proxies such as the 41.0 percent of senators and 20.6 percent of House members who attended a school that Jonathan Wai treated as an indicator of the top 1 percent of ability. None of this ranks a party or says anything about one politician.

1 What Does the IQ of Politicians Mean, and What Would Count as Evidence?

A claim about the IQ of politicians is a claim about a group average, and it can only be as good as the measure that produced it. Politicians are not one population. A town councillor who serves a few evenings a month, a mayor with a full-time salary, a member of a national legislature and an elected head of government are chosen through different routes, and the studies below separate several of these levels. Each study also defines ability differently. The word IQ here means a score on a standardized test with a mean of 100 and a standard deviation of 15, as the page on how IQ scores are normed explains. Few of the studies report anything on that scale.

Five kinds of evidence compete for the label. The first is a measured test score, taken by the person before or independent of a political career and later linked to election records. This is the strongest kind, and it exists only where governments kept the scores of a whole generation. The second is a proxy built from records: years of education, the selectivity of the school attended, or the income a person earns beyond what age, schooling and job sector predict. The third is an inference from the text politicians produce, such as the reading level of speeches. The fourth is a retrospective estimate from a biography, which is the method behind the circulating lists of presidential IQs, and the page on IQ estimates for US presidents covers those. The fifth is opinion, such as surveys of what voters think about politicians.

Two terms need fixing. Positive selection means the people who end up in office score higher on a trait than the population they were drawn from. It describes who is there. It does not say that office raises ability, that high ability makes a person run, or that voters chose on ability. The second term is effect size in standard deviations, the distance between two group means measured in units of the population spread. The page on why the standard deviation is 15 shows how those units translate to IQ points and percentiles.

ACIS has not tested any politician, holds no record belonging to one, and cannot assess anyone retrospectively. This page names no person and gives no estimate for any person. It also does not compare parties. Whether voters on the left and the right differ in measured ability is a separate literature, covered on the page on IQ and political orientation, and the studies here do not support ranking the politicians of one party above another.

2 Why Is the Cognitive Ability of Politicians Almost Never Measured?

Almost no democracy holds cognitive test scores for its politicians, because a score becomes available for research only when a government tested a whole generation and kept the results. Nothing in the law of any country in these studies requires a candidate to take an intelligence test, and the scores that candidates may have taken in school or on a job application are private. A researcher who wants to compare politicians with the population needs three things at once: a test that nearly everyone in a cohort took, records of who later ran for office and who won, and a way to link the two. Military conscription registers supplied the first requirement in the Nordic countries, and population registers supplied the other two.

That explains the geography of the evidence. Swedish enlistment records are kept for men born from 1951 onward, and Dal Bó and colleagues linked them to every candidate in municipal and national elections from 1982 to 2010. Finnish researchers linked the cognitive tests that the Finnish Defense Forces gave to male conscripts to candidates in local elections from 1996 to 2017 and to candidates in the national elections of 2007, 2011 and 2015. The Danish study of politicians described below used education and income records instead of test scores. The United States has no equivalent register, which is why every American analysis in this page rests on a proxy.

Retrospective assessment cannot fill the gap. A test needs an examinee, a standardized setting and a recorded result. A speech, a voting record or a biography is evidence about what a person did, not about how that person performs on a set of tasks designed to rank people on one ability. The page on what IQ measures sets out why a score belongs to a measurement event and not to a career.

Text analysis has been tried as a stand-in. Benoit, Munger and Spirling, in the American Journal of Political Science, argue that political scientists lack measures of the sophistication of political communication designed for that domain. They propose a method based on crowdsourced pairwise comparisons of text snippets, and they reanalyze the State of the Union corpus to show how conclusions change. Reading-level formulas count features such as sentence length and word length. By our reasoning, which is not a finding of that paper, those features describe a text, and a text is shaped by speechwriters, editors, the occasion and the audience as much as by the speaker. A reading grade attached to a speech therefore does not estimate the speaker's ability, and this page does not treat it as evidence of a politician's IQ.

3 Every Study Side by Side

The studies differ so much in measure and design that their results cannot be averaged, and the table below keeps them apart. The column on what the study is not matters as much as the result, because the commonest error in coverage of this topic is to treat a proxy as a test score.

Only the first two rows contain a measured cognitive score linked to election records. Every claim about the cognitive ability of American legislators belongs in the fifth row or below, which means it rests on a proxy.

4 How Did the Swedish Military Measure Cognitive Ability and Leadership?

The Swedish cognitive score comes from a two day enlistment procedure that nearly every man in a cohort went through at 18 or 19, and it is a coarse nine point scale, not an IQ. Dal Bó, Finan, Folke, Persson and Rickne describe the procedure in the working paper version of Who Becomes a Politician?, which the Quarterly Journal of Economics published in 2017. Enlistment evaluated medical status, physical fitness and cognitive abilities. About 75 percent of the men in their sample took it at 18 and about 25 percent at 19. Because enlistment generally falls in the year of high school graduation, the authors note that scores are not influenced by university training. The tests were high stakes, since better results led to more rewarding military placements, and takers were not told their exact results. The paper uses the 1951 to 1980 birth cohorts, 30 cohorts in all, and puts enlistment rates at around 90 percent.

The cognitive score comes from a written test of problem solving, induction, numerical, verbal, spatial and technology comprehension. The authors cite an army expert who argues that the test is a good measure of general intelligence, and they write that the score can be thought of as an IQ score. That is the authors' framing, and it is a reasonable one for a battery that samples several abilities, as the page on the CHC model describes for broad abilities such as Gf, Gc, Gv and Gq. It does not make the number an IQ, because the scale is a stanine. A stanine of 5 is given to the middle 20 percent of test takers, 6, 7 and 8 to the next 17, 12 and 7 percent, and 9 to the top 4 percent, with the lower scores symmetric. In the authors' footnote, a 5 spans roughly minus 0.25 to plus 0.25 standard deviations and a 9 means above plus 1.75. A nine point scale cannot separate a person at plus 2 from a person at plus 3.

The leadership score is a different instrument. Men who scored 5 or higher on the cognitive test went through an in depth evaluation for possible leadership roles, in which trained psychologists gave a semi structured interview and assigned a 1 to 9 score. The psychologist saw the conscript's cognitive result, physical endurance, strength, school grades and answers to 70 to 80 questions about friends, family and hobbies before the interview. The manual is classified. The authors say the score summarizes four traits, social maturity, psychological energy, intensity and emotional stability, and that these relate to the Big Five personality traits. Leadership is therefore not a second intelligence score. It is an interviewer's rating partly informed by the cognitive score, which matters when the two are compared.

The politicians come from candidate lists. The authors identified more than 200,000 unique people who ran for national or municipal office from 1982 to 2010 and about 50,000 elected individuals, and they linked them to population registers. The design therefore compares four groups of men with the population: nominated but not elected municipal candidates, elected councillors, mayors and members of the national parliament. Women are in the registers, but only men have the enlistment scores. The page on IQ versus the ASVAB covers the American military aptitude battery, which is a different instrument from the Swedish test.

5 How Far Above the Population Were Swedish Politicians?

On the cognitive score the gap rose at every step up the political ladder, from about half a stanine point for nominated candidates to 1.6 points, about 0.84 standard deviations, for members of parliament. The working paper's summary table pools the election years 1991, 1994, 1998, 2002, 2006 and 2010 and counts a politician once for each election in which he served. The text states that the population average was 5.3 on leadership, that mayors were 1.3 points (0.68 standard deviations) above the average person on cognition, and that members of parliament were 1.6 points, around 0.84 standard deviations, above. For leadership it puts mayors a bit more than a full point above, or 70 percent of a population standard deviation. The table below gives the group means as the paper's table prints them for men in the 1951 to 1980 cohorts.

The authors say the progression of means from nominated to elected to mayor to member of parliament suggests increasing positive selection, and that mayors and members of parliament are as positively selected as members of elite occupations in the private and public sectors. They report all differences between groups as statistically significant at p below 0.001 because the samples are large, which says nothing about whether a difference is large in practice.

A translation helps, with a warning attached. If the stanine behaves like a normal scale, a gap of 0.84 standard deviations places the average member of parliament near the 80th percentile of the male population, and 0.68 places the average mayor near the 75th, as the IQ percentile calculator would show for the same distance. On the IQ scale of the page on IQ score versus percentile, that is about 13 points and about 10 points above 100. These conversions are our arithmetic, and the authors publish no IQ point figure. The conversion also assumes a normal distribution, equal spread in the groups and a stanine that reflects the underlying ability despite the coarseness of its nine steps. If the spread of the politicians were similar to that of the population, about one in five members of parliament would still sit below the male population average. A group mean is not a floor.

Three further results in the working paper bear on how to read the gap. First, the authors compared politicians with their own siblings, who share family background. The gap between politicians and their siblings is 70 percent of the gap between politicians and the population for cognition, 78 percent for leadership and 74 percent for the earnings score. The authors read this as evidence that individual ability, not only family background, separates politicians from the population. On our arithmetic, it also leaves about 30 percent of the cognitive gap that siblings share with the politician. Second, 96 percent of municipalities select positively on the cognitive score and 86 percent on the leadership score, so the average is not driven by a few large cities. Third, years of schooling correlate 0.51 with the cognitive score among Swedish men in the paper's 2011 appendix table, the cognitive and leadership scores correlate 0.34, and schooling and the earnings score correlate only 0.076. A degree is a weak stand-in for a test score, a point that matters below when the proxy studies appear.

The authors also state a limit that readers often drop. They write that conclusions based on a single country cannot be extrapolated to the rest of the world. Sweden elects through proportional representation and party lists, and the study cannot say whether a different electoral system would select differently. The figures in this section come from the working paper dated March 2017, which is the version opened for this page. The journal version may round or report them differently, so a reader who needs a number for citation should check the published article.

6 How Do Politicians Compare With CEOs, Lawyers and Doctors?

Swedish mayors and members of parliament scored about as high on cognition as chief executives of small and medium firms, below doctors, lawyers and academics, and the ranking says more about what each job selects than about worth. The working paper's second table compares four groups of politicians with seven occupations for men in the 1951 to 1980 cohorts, using 2011 data for the full working age population. The values below are group means on the 1 to 9 scales. The politician rows differ slightly from the pooled table above because this table covers one year.

The authors describe the pattern in words. Elected politicians have cognitive and leadership scores similar to CEOs of firms with 10 to 25 employees. Mayors have exactly the scores of CEOs of firms with 25 to 250 employees, although mayors earn substantially less. Members of parliament fall between those CEOs and the CEOs of firms with more than 250 employees. Lawyers and academic social scientists outscore CEOs and mayors on cognition, and medical doctors show the highest cognitive scores of all. Economists and political scientists have the most schooling and among the lowest leadership scores. The authors read this as intuitive, because academics are smart but do not lead organizations, whereas mayors, members of parliament and CEOs are slightly lower on cognition, substantially less schooled and higher on leadership.

Two cautions apply. First, the numbers of people differ widely: the paper lists 247 mayors and 320 members of parliament, against 6,825 small firm CEOs and 29,514 doctors. Means for small groups carry more sampling error than the table shows, and the paper does not print intervals in this table. Second, the comparison describes the professions as they sat in Sweden in 2011. The authors call medicine a highly prestigious profession in Sweden associated with excellence, so selection into it is strong, and the same ranking would not be expected in a country where entry works differently. The page on average IQ by profession and the page on IQ and entrepreneurship discuss how occupational averages are built and why they differ across sources.

The comparison with CEOs is also the place to separate ability from hierarchy. A higher mean among CEOs of larger firms means that firm size selects on cognition and leadership, not that being a politician is a lesser kind of job. The paper's own reading is that political selection is as strong as selection into elite private and public roles. The page on IQ and success covers the wider evidence on how ability relates to occupational level and pay.

7 What Did the Finnish Conscript Study Find?

Finnish politicians also scored above the male population on conscript cognitive tests, but the gaps for local politicians were modest and varied by test, from about 0.12 standard deviations on visuospatial reasoning to about 0.35 on verbal reasoning. Jokela, Meriläinen, Tukiainen and von Schoultz published the study in 2025 in the Journal of the European Economic Association. They used the tests that the Finnish Defense Forces gave to male conscripts, with a cognitive test made of three subtests of 40 questions each, on visuospatial, verbal and arithmetic reasoning. A separate personality test of 218 statements covered eight traits, and the study reports on those as well, but this page deals with the cognitive tests only.

The sample is large. Cognitive scores cover conscripts from 1982 to 2014, drawn as a 90 percent random sample of test takers, which covers about 80 percent of the male population. The comparison group is a 10 percent random sample of adults who were not politicians in each municipality and election year, which gives about 350,691 men for the cognitive comparison. Candidates come from six local elections, 1996 to 2017, with about 230,000 candidate election observations, and from the national elections of 2007, 2011 and 2015, where the sample is slightly less than a thousand people. The authors report 24,946 nominated local candidates who were not elected and 8,694 elected local politicians with cognitive scores.

The z scores are in standard deviation units relative to the conscript sample, as the paper's local politician table prints them. Subtracting the population row from the elected row gives about 0.12 for visuospatial, 0.35 for verbal and 0.33 for arithmetic reasoning, which is our arithmetic. The text states the verbal step as 0.21 from the population to the nominated candidates and a further 0.14 from nominated to elected, and the arithmetic steps as 0.16 and 0.17. The pattern matches Sweden in direction: nominated candidates sit above the population and elected politicians sit above the nominated.

For members of parliament the paper compares raw means. On the visuospatial, verbal and arithmetic subtests, members of parliament scored 26.82, 27.76 and 25.27, against 25.42, 22.51 and 20.11 in the population, and the text puts the gaps at about one fourth of a standard deviation for visuospatial, about three fourths for verbal and two thirds for arithmetic reasoning. The authors compare local politicians with occupations and degrees. Local politicians resemble technicians in cognitive scores and fall between men with secondary education and men with an undergraduate degree. Members of parliament score slightly above undergraduate degree holders in verbal and arithmetic reasoning and are on par with graduates in education or arts and humanities. Natural sciences graduates score above members of parliament.

Two cautions matter before comparing Finland with Sweden. The measures are not the same: the Swedish figure is a composite stanine and the Finnish figures are three subtest z scores. Our arithmetic on the Swedish table gives elected councillors a gap of about 0.9 points, or about 0.47 standard deviations, on a population spread of 1.9, so the councillor gaps are not wildly apart, but the sampling, the cohorts and the tests differ enough that nobody should read the countries as ranked. The second caution is the authors' own. Women are excluded from the cognitive and personality data because the conscript tests cover men, and the authors call this the most obvious and serious weakness of the data. They argue that positive selection on ability probably generalizes to women, pointing to education, income and occupational class among female candidates, and they call for further research.

8 What Do the Danish Studies and Other Proxy Studies Show?

When test scores are missing, researchers substitute education and an earnings based score, and these proxies track cognitive scores only loosely, so the Danish results speak to ability in a different sense from the Nordic conscript studies. Dahlgaard and Pedersen used data on the whole Danish population and every candidate in local and national elections from 1990 to 2015, and they published the study in Political Science Research and Methods. They apply two measures: years of education and an ability measure based on income. The income measure takes what a person earns beyond what age, education and employment sector would predict. The authors describe both as far from perfect measures of ability that follow the standard in the literature.

Their results run in the same direction as the conscript studies. Candidates and elected politicians had higher ability on these measures than the voters they represent. Selection on ability partly held when politicians were compared with their non-politician siblings. Fathers' incomes were spread across the distribution, which the authors call diverse but not fully representative, and they found no substantial trade-off between ability and representation of economic background. They also used a 2007 structural reform that merged 239 of 271 Danish municipalities into 66 and left 32 unchanged, so that the number of elected councillors fell from between 4,647 and 4,690 before the reform to between 2,444 and 2,522 after it. More competition per seat did not change selection on ability, economic background or the trade-off between them.

How far does an earnings based score stand in for a test score? The Swedish paper offers two numbers. Among Swedish men, the earnings score correlates 0.17 with the cognitive score and 0.20 with the leadership score, and years of schooling correlate 0.51 with the cognitive score. Dahlgaard and Pedersen add that earlier work by Besley and colleagues showed that both proxies correlate with political success, with cognitive and leadership tests from the military, and with policy outcomes in Sweden. Correlations of this size mean a proxy carries real information about a group and little certainty about an individual, a distinction the page on what IQ correlates with develops for many pairs of variables. They also mean a proxy studied in Denmark cannot be read as an IQ gap.

A second Swedish study sheds light on the process. Besley, Folke, Persson and Rickne studied the 1993 zipper quota of the Social Democrats in the American Economic Review, which required local parties to alternate men and women on ballots. They used national earnings register data for politicians in seven parties, 290 municipalities and ten elections from 1982 to 2014. The paper finds that the quota raised the competence of male politicians most where it raised female representation most, and attributes this mainly to the resignation of mediocre male leaders. The competence measure is built from earnings records, not from cognitive tests, so the study is evidence that party leaders shape who is on the ballot, not a statement about IQ.

The proxy studies also show how much the word ability is stretched. Dahlgaard and Pedersen note that studies disagree on whether politicians' education predicts performance in office: one study found a link and another failed to find a stable one. They add that Danish voters perceive candidates with a university education as more competent than candidates without one. That is a finding about perception. The page on average IQ by education explains why education and measured IQ overlap without being the same thing.

9 What Can Be Said About the United States Congress?

For the United States there is no study of measured cognitive ability among legislators, and the available evidence shows that members are highly educated and drawn from advantaged families, which is related to ability but is not a score. Wai, in Intelligence, examined five elite groups with a total of 2,254 people: Fortune 500 CEOs, federal judges, billionaires, the 100 senators and 441 House members, whose data came from the Biographical Directory of the United States Congress in 2012. His proxy was attendance at a school whose admission tests suggest top 1 percent ability. The undergraduate list had 29 schools, 21 national universities and 8 liberal arts colleges with a median combined SAT critical reading and mathematics score of at least 1400, plus the 12 highest ranked law schools by LSAT and 12 business schools by GMAT. Wai relies on Charles Murray's calculation that a score of 1400 sits near the 97th percentile of test takers and well within the top 1 percent of seventeen year olds, because test takers were only 36 percent of that age group in 2010. He calls attendance a reasonable indicator of top 1 percent ability.

The first four rows are Wai's, the last three come from the Congressional Research Service profile Membership of the 119th Congress, updated August 4, 2025 and read on October 13, 2026. For comparison, the Census Bureau reported that 37.7 percent of people age 25 and older had a bachelor's degree or higher in 2022, in a release dated February 16, 2023. The contrast in degrees is large, and it is a contrast in education. The page on average IQ by education explains how far degrees track measured ability and where they stop.

The proxy has a critic. Sternberg, in the Journal of Expertise, reviewed a later analysis by Wai and colleagues that used the same method on the Time 100 list. He called the inference from prestigious university attendance to top 1 percent cognitive ability quite a bit of a stretch, because admission also weighs athletic prowess, alumni connections and wealthy parents who have contributed or are expected to contribute. That critique targets the later paper, not the 2013 one, but by our reading the same logic applies to both, and the pages on average IQ by university and IQ versus SAT show how an admission test and an ability measure overlap without coinciding.

Family background is the second line of evidence. Feigenbaum, Hall, Thompson and Yoder, in the NBER working paper Fortunate Sons, dated August 2026, traced members of Congress born from 1830 to 1950 through complete count census records, together with their parents and brothers. By 1940, nearly 60 percent of members aged 18 to 40 had a college degree, against under 5 percent of comparable young men outside Congress. In nineteenth century censuses, adult members held about three times the wealth of matched controls. Brothers fell between the population and the members, and the authors estimate that about half of a member's adult advantage is shared with his brother. The paper concludes that Congress has been drawn from the most advantaged families in every cohort observed. It measures no ability. Set beside the Swedish finding that the parents of politicians approximate the population, it shows that the American record points the other way on family background, which is our comparison and not the authors.

Wai's paper also reports a breakdown of school attendance by party affiliation. This page does not reproduce it. A school selectivity proxy is a weak instrument for comparing two groups, because the schools members attended depend on region, family wealth and the era in which each member was recruited, and the paper is a single snapshot of 2012. A ranking of parties by a proxy would turn a caveat-heavy measure into a headline, and nothing in the Nordic evidence supports treating parties as having different cognitive averages.

10 What Produces Positive Selection, and What Do the Studies Not Show?

Politicians end up above average through three filters at once, who runs, who the party nominates and ranks, and who voters elect, and the registers describe the result without separating the filters. The Swedish and Finnish data both show the same staircase: nominated candidates sit above the population, elected politicians above the nominated, and in Sweden mayors and members of parliament above councillors. Each step is a different filter. The authors of the Swedish study offer suggestive evidence on two of them. Parties appear to screen positively, promoting more able people to higher ranks regardless of social class, and individual incentives matter, since pay counts in the decision to run, although strong intrinsic motivation and well paid full time posts keep able people in the pool.

The Swedish result also cuts against a standard prediction. Economic models predict adverse selection, because less able people have lower opportunity costs and so a comparative advantage at entering politics. Dal Bó and colleagues found the opposite on every ability measure, even though politicians had substantially higher pre-office incomes than the population, which means higher opportunity costs. The Swedish quota study adds that party rules change who is on the ballot, while the Danish merger study found that more competition per seat did not change selection. Together they support a modest claim: in these Nordic systems, the people who reach office are on average more able on the measures available than those who do not.

Four things remain unshown. The studies do not show that office raises ability, because every score comes from before a political career began. They do not show that ability causes political success, since a third factor such as ambition, schooling or social networks could drive both. They do not show that higher ability produces better policy; Dahlgaard and Pedersen note that studies disagree on whether education predicts performance in office. And they do not cover older politicians, because Swedish enlistment records are kept only for men born from 1951, so politicians born earlier have no score. The Swedish paper also notes that voters ranked politician competence among the top three reasons for choosing a party. That describes what voters say they want, not what they get.

The cognitive and leadership scores correlate only 0.34 among Swedish men, so they are different traits. A politician can be high on one and ordinary on the other. The page on critical thinking versus intelligence covers why judgment, a quality voters care about, is not what an IQ test measures, and the page on IQ and job performance summarizes the relation between ability and work outcomes in the occupations where it has been measured.

11 What Does This Not Say About Any Single Politician, or About You?

A group average says nothing about where one politician stands, because individuals inside a group with a higher mean still range widely. A mean gap of 0.84 standard deviations leaves large overlap between members of parliament and the population, and the Swedish and Finnish studies describe groups of men tested at conscription, not the person on a news broadcast. No politician can be placed on the table in this page, and no number that circulates for a named politician is supported by these studies. The numbers that circulate for named officials, living or dead, are not drawn from these studies, and the studies contain no individual scores. The page on IQ estimates for US presidents shows what the historical record of such estimates is for one office.

The studies also say nothing about parties. The Swedish and Finnish analyses pool politicians across parties, and the American party breakdown rests on a school selectivity proxy in a single year. Whether the voters of one side score differently from the voters of another is the subject of the page on IQ and political orientation, which concerns voters and not officeholders. What politicians or voters believe about the world is a separate empirical question, and the page on conspiracy beliefs and IQ treats it on its own terms. Nor do these studies say that high ability makes a good politician, an honest one or a popular one. The authors of the Finnish study also measured personality traits, because ability is only one of several things that voters and parties may select on.

For the reader the lesson is smaller still. If you work in politics, vote or follow it, nothing here predicts your own score or anyone else's from a job, a degree or an office. The studies also do not say that an IQ test should be used to choose among candidates, and the page on why IQ matters explains the limits of using one number to choose among people. Your own cognitive profile is a different question, and it is answered by a test you take, not by a comparison with a group. The page on reliability and validity explains what it takes for a score to carry that kind of meaning. ACIS is an online, unsupervised test, not a clinical instrument, and it is not designed for hiring, school admission or entry to high IQ societies.

12 What the Evidence Supports, Stated Narrowly

The evidence supports one narrow claim, that elected politicians in Sweden and Finland scored above the male population on conscript cognitive tests taken years before, and it supports nothing about individual politicians, about parties or about the United States beyond education. Four statements follow from the sources. In Sweden, members of parliament averaged about 0.84 standard deviations above the male population on a nine point cognitive score and mayors about 0.68, in the working paper version of the study. In Finland, elected local politicians scored above the population on three cognitive tests by amounts between about 0.12 and 0.35 standard deviations, and members of parliament by about one fourth to three fourths of a standard deviation depending on the test. In Denmark, candidates and elected politicians had more education and a higher income based ability score than the voters. In the United States, members of Congress are far more educated than the adult population and, in historical records, come from advantaged families, while the one study that inferred ability from schools used a proxy that a critic of the same method calls a stretch.

The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014), published by the American Educational Research Association, the American Psychological Association and the National Council on Measurement in Education, frame validity as the degree to which evidence and theory support an interpretation of scores for a proposed use. Applied here, the conscript scores were used to place enlisted men in military roles, and the studies use them to describe groups, which is a use the authors argue for. Using a stanine as an IQ, a school list as an IQ, or the reading grade of a speech as an IQ is a different interpretation and needs its own evidence, which the sources do not supply. The page on how IQ scores are normed shows what a score needs before it is an IQ.

APA guidance on professional conduct points the same way. Standard 9.01 of the American Psychological Association's Ethical Principles of Psychologists and Code of Conduct concerns the bases for assessments. In summary, the standard asks that psychologists give opinions about a person's psychological characteristics only after an examination adequate to support their conclusions, and where an examination is not practical they explain the effect of the limited information and limit their conclusions. A newspaper clip, a speech or a biography does not meet that bar for an IQ. In every study here the unit of analysis is a group with a measured mean, and every statement beyond that unit is the reader's, not the data's.

13 Sources Behind This Page

Every figure above is traceable to one of the following, and each is linked at the point where it is used. Percentiles, IQ point conversions and subtraction of group means are our arithmetic on published figures and are labeled as such. Swedish figures are those of the March 2017 working paper opened for this page, and the journal version should be checked before a number is quoted. Web pages were read on October 13, 2026.

• Dal Bó E, Finan F, Folke O, Persson T and Rickne J. Who Becomes a Politician? The Quarterly Journal of Economics, 2017, volume 132, issue 4, pages 1877 to 1914. Working paper version, March 2017, read here.
• Jokela M, Meriläinen J, Tukiainen J and von Schoultz Å. Personality Traits and Cognitive Ability in Political Selection. Journal of the European Economic Association, 2025, volume 23, issue 5, pages 1908 to 1949.
• Dahlgaard JO and Pedersen RT. Inclusive meritocracy: ability and descriptive representation among Danish politicians. Political Science Research and Methods, 2025, volume 13, issue 3, pages 567 to 590.
• Besley T, Folke O, Persson T and Rickne J. Gender Quotas and the Crisis of the Mediocre Man: Theory and Evidence from Sweden. American Economic Review, 2017, volume 107, issue 8, pages 2204 to 2242.
• Wai J. Investigating America's elite: Cognitive ability, education, and sex differences. Intelligence, 2013, volume 41, issue 4, pages 203 to 211.
• Sternberg RJ. How and Why Wai and His Colleagues Reach Conclusions That Do Not Necessarily Follow from Their Data: Reflections on "The Role of Elite Education and Inferred Cognitive Ability in Eminent Creative Expertise: An Historical Analysis of the Time 100". Journal of Expertise, 2019, volume 2, issue 2, beginning on page 140.
• Feigenbaum JJ, Hall AB, Thompson DM and Yoder J. Fortunate Sons: Elite Political Selection in American History. NBER Working Paper 35569, August 2026.
• Benoit K, Munger K and Spirling A. Measuring and Explaining Political Sophistication through Textual Complexity. American Journal of Political Science, 2019, volume 63, issue 2, pages 491 to 508.
• Congressional Research Service. Membership of the 119th Congress: A Profile. Report R48535, updated August 4, 2025.
• U.S. Census Bureau. Census Bureau Releases New Educational Attainment Data, press release CB23-TPS.21, February 16, 2023.
• American Educational Research Association, American Psychological Association and National Council on Measurement in Education. Standards for Educational and Psychological Testing, 2014.
• American Psychological Association. Ethical Principles of Psychologists and Code of Conduct, Standard 9.01, Bases for Assessments.

14 Frequently Asked Questions