Skip to main content

The IQ Halo effect

On desirable correlates of intelligence

Increased IQ correlates with consequentially morally desirable actions, attitudes, and beliefs.

“All good things tend to go together, as do all bad ones.” –Edward Lee Thorndike

military: AFQT IQ test (subtest of the ASVAB) excellent predictor of training costs, success

the consensus reply to Bell Curve:

  1. IQ is strongly related, probably more so than any other single measurable human trait, to many important educational, occupational, economic, and social outcomes. Its relation to the welfare and performance of individuals is very strong in some arenas in life (education, military training), moderate but robust in others (social competence), and modest but consistent in others (law-abidingness). Whatever IQ tests measure, it is of great practical and social importance.

  2. A high IQ is an advantage in life because virtually all activities require some reasoning and decision-making. Conversely, a low IQ is often a disadvantage, especially in disorganized environments. Of course, a high IQ no more guarantees success than a low IQ guarantees failure in life. There are many exceptions, but the odds for success in our society greatly favor individuals with higher IQs.

  3. That IQ may be highly heritable does not mean that it is not affected by the environment. Individuals are not born with fixed, unchangeable levels of intelligence (no one claims they are). IQs do gradually stabilize during childhood, however, and generally change little thereafter.

  4. Although the environment is important in creating IQ differences, we do not know yet how to manipulate it to raise low IQs permanently. Whether recent attempts show promise is still a matter of considerable scientific debate.

The controversy over The Bell Curve (Herrnstein & Murray, 1994) was at its height in the fall of 1994. Many critics attacked the book for supposedly relying on outdated, pseudoscientific notions of intelligence. In criticizing the book, many critics promoted false and highly misleading views about the scientific study of intelligence. Public miseducation on the topic is hardly new (Snyderman & Rothman, 1987, 1988), but never before had it been so angry and extreme…It is obviously not the case that there is no disagreement about these important issues or that scientific truth is a matter of majority rule. A significant minority of the experts who were contacted disagreed in part or in whole with the statement, and many of the signers would have written the statement somewhat differently. Rather, the lesson here is that what have often been caricatured in the public press as discredited, fringe ideas actually represent the solid scientific center in the serious study of intelligence. As Snyderman and Rothman’s (1988) survey of IQ experts and journalists revealed, the media, among others, have been turning the truth on its head.

“General Mental Ability in the World of Work: Occupational Attainment and Job Performance”, Schmidt & Hunter2004

The accumulated evidence has become very strong that GMA is correlated with a wide variety of life outcomes, ranging from risky health-related behaviors, to criminal offenses, to the ability to use a bus or subway system (Gottfredson, 1997; Lubinski & Humphreys, 1997). In addition, the more highly a given GMA measure loads on the general factor in mental ability (the g factor), the larger are these correlations. The relative standing of individuals on GMA has been found to be stable over periods of more than 65 years (Deary, Whalley, Lemmon, Crawford, & Starr, 2000). Findings in behavior genetics, including studies of identical twins reared apart and together (eg. Bouchard, Lykken, McGue, Segal, & Tellegen, 1990), have shown beyond doubt that GMA has a strong genetic basis (eg. Bouchard, 1998; McGue & Bouchard, 1998).

  • Lubinski, D., & Humphreys, L. G. (1997). Incorporating general intelligence into epidemiology and the social sciences. Intelligence, 24, 159 -201

  • Deary, I. J., Whalley, L. J., Lemmon, H., Crawford, J. R., & Starr, J. M. (2000). The stability of individual differences in mental ability from childhood to old age: Follow-up of the 1932 Scottish mental survey. Intelligence, 28, 49 -55

People’s rankings or ratings of the occupational level or prestige of different occupations are very reliable; correlations between mean ratings across studies are in the .95 to .98 range, regardless of the social class, occupation, age, or country of the raters (Dawis, 1994; Jensen, 1980, pp. 339 -347). These occupational level ratings correlate between .90 and .95 with average GMA scores of people in the occupations (Jensen1998, p. 293). Individual level correlations are of course not this large. In the U.S. Employment Service’s large database on the General Aptitude Test Battery (GATB; Hunter, 1980), the individual level correlation between the GMA measure derived from that battery and occupational level is .65 (0.72 corrected for measurement error; Jensen1998). Much military data exist from both world wars (when samples of draftees were very representative of the U.S. male population) showing an increase in mean GMA scores as occupational level (as determined by ratings of the sort discussed here) increases (Harrell & Harrell, 1945; Stewart, 1947; Yerkes, 1921). Table 1, showing findings for 18,782 White enlisted men in the Army Air Force Command (Harrell & Harrell, 1945), presents typical findings. The GMA measure used was the Army General Classification Test (Schmidt, Hunter, & Pearlman, 1981). Mean GMA scores clearly increase with occupational level.

Wilk, Desmarais, and Sackett (1995), using the 3,887 young adults in the National Longitudinal Survey-Youth Cohort (NLSY; Center for Human Resource Research, 1989) for whom the required data were available, showed that over the 5-year period 1982–1987, GMA measured in 1980 predicted movement in the job hierarchy. Those with higher GMA scores in 1980 moved up the hierarchy, whereas those with lower GMA scores moved down in the hierarchy. In a larger follow-up study that was based on somewhat different methodology, Wilk & Sackett1996 examined two large government databases: the National Longitudinal Study of the Class of 1972 (NLS-72) and the National Longitudinal Survey of Labor Market Experience-Youth Cohort (NLSY). In both databases, Wilk and Sackett found that job mobility was predicted by the congruence between individuals’ GMA scores (measured several years earlier) and the objectively measured complexity of their jobs. If their GMA exceeded the complexity level of their job, they were likely to move into a higher complexity job. And if the complexity level of their job exceeded their GMA level, they were likely to move down into a less complex job.

In another study drawn from this same large database, Murray (1998) found that GMA predicted later income even with unusually thorough control for socioeconomic status (SES) and other background variables. This control took advantage of the large variability of GMA within families and was achieved by use of a sample of male full biological siblings, hence controlling for home background and many other variables (eg. schools, neighborhoods). Murray found that the siblings with higher GMA scores received more education, entered more prestigious occupations, had higher income, and were employed more regularly. When the siblings were in their late 20s (in 1993), a person with average GMA was earning on average almost $18,000 less per year than his brighter sibling who had an IQ of 120 or higher and was earning more than $9,000 more than his duller sibling who had an IQ of less than 80. This pattern of findings held up even in a sub-sample of persons who were all from “advantaged” homes (his “utopian” sample).

Judge, Higgins, Thoresen, and Barrick (1999) related GMA measures taken at around 12 years of age to occupational outcomes in the age range of 41 to 50 years. They found that childhood GMA scores predicted adult occupational level with a correlation of .51 and predicted adult income with a correlation of .53. Ball (1938) found that GMA measured in childhood correlated .47 with occupational level 14 years later and .71 with occupational level 19 years later. Other such studies include Brown & Reynolds1975, Dreher & Bretz1991, Gottfredson & Crouse1986, Howard & Bray1990, Siegel & Ghiselli1971, and Thorndike & Hagen1959.

Results for GMA are typified by the findings of the large study conducted by Hunter (1980; Hunter & Hunter, 1984) for the U.S. Employment Service using the database on the General Aptitude Test Battery (GATB). On the basis of 425 validity studies (N ϭ 32,124) conducted on civilian jobs spanning the occupational spectrum, Hunter & Hunter1984 and Hunter (1980) reported the results shown in Table 2. Hunter assigned each job to one of five job families based on complexity (ie. the information processing requirements of the job, measured using U.S. Department of Labor job analysis data for each job). This is the largest database available using a measure of performance on the job (measured using supervisory ratings of job performance). As can be seen, validity for predicting performance on the job ranges from .58 for the highest complexity jobs (professional, scientific, and upper management jobs) to .23 at the lowest complexity level (feeding/ off-bearing jobs). Job Family 2 (2.5% of all jobs in the economy) consists of complex technical jobs such as computer-systems trouble shooting or complex manufacturing set-up jobs. Job Family 3, with almost 63% of all jobs in the economy, includes skilled workers, technicians, mid-level administrators, paraprofessionals, and similar jobs. Job Family 4 is semiskilled work. Clearly, GMA predicts performance on higher level jobs better that it does for lower level jobs. However, there is substantial validity for all job levels. In particular, GMA predicts performance even for the simplest 2.4% of jobs (Job Family 5). Other findings are reported in Table 3. On the basis of 194 studies (N ϭ 17,539) of performance in clerical jobs, Pearlman, Schmidt, and Hunter (1980) reported a mean GMA validity for job performance of .52. For law enforcement jobs, Hirsh, Northrup, and Schmidt (1986) reported a mean validity for job performance of .38. In a large scale, multi-year military study on enlisted Army personnel (called “Project A”), McHenry, Hough, Toquam, Hanson, and Ashworth (1990) reported that GMA predicted “Core Technical Proficiency” with a validity of .63 and “General Soldiering Performance” with a validity of .65. Both job performance measures were based on hands-on work-sample measures. (Validities were not as high for ratings of “Effort and Leadership” [.31], “Personal Discipline” [.16], and “Physical Fitness and Military Bearing” [.20], which are secondary criterion measures with fewer cognitive demands.) Using similar job sample measures of job performance, Ree, Earles, and Teachout (1994) reported a mean value of .45 across seven Air Force jobs. Validities for the prediction of performance in training programs are even larger. As can be seen in Table 2, in the GATB training database (90 studies; N ϭ 6,496) used by Hunter & Hunter 1984, GMA predicted performance in job training programs for all job families for which data existed with a correlation above .50. The database for training performance is larger for military jobs. Hunter (1986) meta-analyzed military databases totaling over 82,000 trainees and reported an average validity of .63 for GMA. On the basis of 77,958 Air Force trainees, Ree & Earles 1991 reported a very similar value of .60. On the basis of 65 studies with an N of 32,157, Pearlman et al 1980 reported a mean validity of .71 for GMA in predicting training performance of clerical workers, whereas Hirsh et al 1986 found a mean value of .76 for predicting performance in police and other training academies for law enforcement trainees. These findings and others are shown in Table 3. Additional data of this sort are presented in Schmidt (2002).

Differential or specific aptitude theory hypothesizes that performance on different jobs requires different cognitive aptitudes and, therefore, regression equations computed for each job and incorporating measures of several specific aptitudes will optimize the prediction of performance on the job and in training. In the last 10 years, research has strongly disconfirmed this theory. Differentially weighting specific aptitude tests produces little or no increase in validity over the use of a measure of GMA. An explanation for this finding has been discovered. It has been found that specific aptitude tests measure GMA; in addition to GMA, each measures something specific to that aptitude (eg. specifically numerical aptitude, over and above GMA). The GMA component appears to be responsible for the prediction of job and training performance, whereas the factors specific to the aptitudes appear to contribute little or nothing to prediction. The research showing this is presented and reviewed in Hunter (1986); Jensen (1986); Thorndike (1986); Olea & Ree 1994; Ree & Earles 1992; Ree et al 1994; Schmidt, Ones, and Hunter (1992); and Sackett & Wilk1994, among other sources. A particularly dramatic refutation of specific aptitude theory comes from the large sample military research conducted by Hunter (1983b) for the Department of Defense on the performance of military personnel in job training programs. Four large samples were analyzed separately: 21,032 Air Force personnel, 20,256 Marines, and two Army samples of 16,618 and 79,926, respectively. In all samples, test data were obtained some months prior to measurement of performance in job training programs. In all samples, causal analysis modeling (with corrections for measurement error and range restriction) was used to pit specific aptitude theory against GMA in the prediction of performance. In the case of all four samples, models with causal arrows from specific aptitudes to training performance failed to fit the data. However, in all the samples, a hierarchical model showing a single causal path from GMA to performance - and no paths from specific aptitudes to performance - fit the data quite well. …Training performance is determined only by GMA, with the standardized path coefficient from GMA to performance being very large (.62). The findings for the other three samples were essentially identical (Hunter, 1983b). It is well known that analysis of causal models with correlational data cannot prove a theory. However, such analyses - especially when samples are very large, as here - can disconfirm theories. Theories that do not fit the data are disconfirmed. In these studies, specific aptitude theory is strongly disconfirmed.

McDaniel (1985) analyzed United States Employment Services (USES) data for groups whose level of job experience extended beyond 5 years. Controlling for differences in variability of GMA across groups, McDaniel correlated GMA with performance ratings for each level of experience to 12 years and up. The results are summarized in Table 4. As the level of experience increases, the predictive validity does not decrease. Validity goes from .36 for 0 - 6 years, up to .44 for 6 -12 years, up to .59 for more than 12 years (although the last value is based on a very small sample). If anything, McDaniel’s data suggest an increase in the validity of GMA for predicting performance ratings as level of worker experience increases.

Many people may also believe that personality is more important than GMA in determining ultimate occupational level. However, research supports the conclusion that personality is less important than GMA in both areas. In recent years, most personality research has been organized around the Big Five model of personality (Goldberg, 1990) …As indicated earlier, Judge et al 1999 found that three of the Big Five personality traits measured in childhood predicted adult occupational level and income. For Conscientiousness, these longitudinal correlations were .49 and .41, respectively; these values are only slightly smaller than the corresponding correlations in this study for GMA (discussed in the Longitudinal Studies section, above) of .51 and .53, respectively. For Openness to Experience (which correlates positively with GMA), the correlations were .32 and .26. Finally, Neuroticism produced longitudinal correlations of -.26 and -.34, for occupational level and income, respectively. Because of the unique nature of Judge et al.’s (1999) study, we conducted additional analyses of the data from this study. Because occupational level and income were highly correlated (r ϭ .83) and loaded on the same factor, we combined them into one equally weighted measure of career success. After correcting for the biasing effects of measurement error, we found that the three Big Five personality traits predicted this index of career success with a (shrunken) multiple correlation of .56. It is interesting to examine the standardized regression weights (betas). For Neuroticism, ␤ ϭ -.05 (SE ϭ .096); for Openness, ␤ ϭ .16 (SE ϭ .10); and for Conscientiousness, ␤ ϭ .44 (SE ϭ .123). Hence, in the regression equation, Conscientiousness is by far the most important personality variable, and Neuroticism appears to have little impact after controlling for the other two personality traits. However, it is also important to control for the effects of GMA. When GMA is added to the regression equation, the (shrunken) multiple correlation rises to .63. Again, it is instructive to examine the beta weights: Neuroticism, ␤ ϭ -.05 (SE ϭ .096); Openness, ␤ ϭ -.03 (SE ϭ .113); Conscientiousness, ␤ ϭ .27 (SE ϭ .128); and GMA, ␤ ϭ .43 (SE ϭ .117). From these figures, it appears that the burden of prediction is borne almost entirely by GMA and Conscientiousness, with GMA being 59% more important than Conscientiousness (ie. .43/.27 ϭ 1.59). In fact, when only GMA and Conscientiousness are included in the regression equation, the (shrunken) multiple correlation remains the same, at .63. The standardized regression weights are then .29 for Conscientiousness (SE ϭ .102) and .41 for GMA (SE ϭ .096). These analyses suggest that Conscientiousness may be the only personality trait that contributes to career success. …The best meta-analytic estimate for the validity of Conscientiousness, measured with a reliable scale, for predicting job performance is .31 (Mount & Barrick, 1995). Hence, the validity of GMA is 60% to 80% larger (depending on the GMA validity estimate used) than that of Conscientiousness. However, Conscientiousness measures contribute to validity over and above the validity of GMA, because the two are uncorrelated (Schmidt & Hunter, 1998). As noted above, Hunter & Hunter1984 estimated the validity of GMA for medium complexity jobs (63% of all jobs) to be .51. The multiple correlation produced by use of measures of both GMA and Conscientiousness in a regression equation for such jobs is .60, an 18% increase in validity over that of GMA alone (Schmidt & Hunter, 1998). The best meta-analytic estimate of the validity of Conscientiousness for predicting performance in job training is .30 (Mount & Barrick, 1995). The multiple correlation produced by simultaneous use of GMA and Conscientiousness measures is .65 (vs. .56 for GMA alone; Schmidt & Hunter, 1998).

“Who Rises to the Top? Early Indicators”, Kell et al 2013

Youth identified before age 13 (n = 320) as having profound mathematical or verbal reasoning abilities (top 1 in 10,000) were tracked for nearly three decades. Their awards and creative accomplishments by age 38, in combination with specific details about their occupational responsibilities, illuminate the magnitude of their contribution and professional stature. Many have been entrusted with obligations and resources for making critical decisions about individual and organizational well-being. Their leadership positions in business, health care, law, the professoriate, and STEM (science, technology, engineering, and mathematics) suggest that many are outstanding creators of modern culture, constituting a precious human-capital resource. Identifying truly profound human potential, and forecasting differential development within such populations, requires assessing multiple cognitive abilities and using atypical measurement procedures. This study illustrates how ultimate criteria may be aggregated and longitudinally sequenced to validate such measures.

Table 1 reveals the richness and scope of participants’ activities. One indication of the caliber of their contributions is the prestige of the organizations that have awarded them grants. The data on creative accomplishments speak for themselves, but a few summary remarks are in order. In the arts and humanities, 24 individuals had produced 128 creative written works (eg. poems, novels, refereed publications), an average of 5.3 accomplishments per individual. In the same domain, 52 people had produced 1,069 achievements in the fine arts (eg. music, sculpture), an average of 20.6 accomplishments per person. STEM achievements are also noteworthy. Fifty-nine individuals had produced refereed STEM publications, in areas ranging from biochemistry to engineering; the total number of STEM publications produced was 392 (6.6 per person). In the case of software development and patents, 117 people had made 820 contributions, an average of 7 per individual. Thirty-one individuals had received more than $25 million in grants, an average of $825,635 per person. The tally of awards and significant accomplishments for these 320 individuals was 2,749, or an average of 8.6 per person.

…enough information is provided to make clear that a number of participants are working for world-class organizations and hold important positions of impact and responsibility in Fortune 500 companies, technology, law, and medicine. For the professoriate in our sample, Table 3 lists universities that either awarded them tenure or attracted them with tenure, plus some of their refereed publication outlets. In total, 11.3% of participants had earned tenure at accredited institutions; 7.5% had tenure at research-intensive institutions (Carnegie Foundation, 2010). This latter percentage is many, many times the base-rate expectation, given the 2% base rate for doctorates in the United States and the fact that only a tiny fraction of the individuals with doctorates have tenure at research0intensive institutions.

Although it would be difficult to quantify participants’ collective accomplishments in a single number, by any standard, it appears that many individuals identifiable by age 13 as having profound mathematical and verbal reasoning ability develop into truly outstanding contributors in their respective fields. Not only did participants choose prestigious occupations by age 38 (Fig. 2 and Table 2), but the organizations employing them were impressive as well (Tables 2 and 3). Although a number of our data counts do not reflect the quality of participants’ contributions, the organizations employing participants (eg. Fortune 500 companies, major law firms, large medical facilities, and research universities) and bestowing awards on them (eg. the U.S. Departments of State and Justice, the National Science Foundation, Intel Corporation, NASA, and The Wall Street Journal) afford reasonable quality appraisals of their creative products as well as the responsibilities, resources, and trust that they have earned. More than 7% of participants held tenure at research-intensive universities (including many considered the best in the world) by the time they were age 38. The 14 attorneys were predominantly working in positions of significant responsibility for major firms or organizations. The 19 physicians were also highly accomplished: Seven were assistant professors, 2 were directors of major private practices, and 1 codirected a hospital organ-transplant center serving more than 3 million people. Rather than working for established organizations, 14 individuals founded companies of their own. Two individuals were vice presidents at Fortune 500 companies; 2 others were Fortune 500 senior hardware or software engineers. Several participants were active in government agencies at local and federal levels-one advised the president of the United States on national policy issues. Although participants’ accomplishments are impressive in variety and scope, it is important to note the magnitude of individual differences in output, even in this exceptionally talented sample. Within several accomplishment groupings, some individuals far outstripped their intellectual peers. For example, in the arts and humanities, one individual produced 500 musical productions, accounting for more than 57% of the musical productions reported here; three individuals produced 100 software contributions each, or nearly 44% of the total reported. Seven participants received more than $1 million in grant funding each; collectively, their funding amounted to nearly $20 million, more than 77% of the total sample’s grant funding; one individual alone received $9 million in grant funding. Finally, one person founded three companies, and another was responsible for raising more than $65 million in private equity investment to fund his company. These findings mirror those in Galton’s (1869/2006) investigation of the Cambridge University “wranglers,” the 40 top-scoring students out of the approximately 100 honors mathematics graduates each year (400-450 students graduated from Cambridge annually). Wranglers were rank-ordered according to their scores on their final mathematics exam (a 44-hr test spread over 8 days). Although being even a low-ranked wrangler was enough for a graduate to obtain a fellowship at a small college, Galton found that the highest-ranked wrangler tended to do more than twice as well on the final exam as the second-ranked wrangler and approximately 4 times better than the lowest-ranked wranglers. Examiners emphasized that the units of measurement they employed were designed to index equal intervals, such that twice the score range translated into approximately twice the knowledge. Such outlying individual differences in accomplishments, even among the most talented, are readily observed throughout history (Murray, 2003). This is one reason why O’Boyle & Aguinis2012 argued that, given the output of truly outstanding performers, performance in general is better modeled through Paretian (power law) distributions as opposed to Gaussian (normal-curve) distributions (Simonton, 1999a, 1999b).

Sipe & Curlette1996 have found in their meta-synthesis of educational research that on the individual level the effect of intelligence on educational attainment was .6 (r = .5). The effects of other variables (motivation, SES, teacher education, etc.) were smaller.

The inbreeding depression had been calculated by Schull & Neel1965 from 1854 cousin marriages in Japan on the WISC and showed an overall 7.5 point decrement (0.50 SD) in the offspring, with each subtest showing a greater or lesser amount. There is no non-genetic explanation for why Black-White differences in the US should be more pronounced on those subtests showing the most inbreeding depression among the Japanese in Japan (Jensen also demonstrated inbreeding depression effects on the Raven Matrices in India; Agrawal, Sinha, & Jensen, 1984).

Miller2012, Singularity Rising:

g - the letter used as shorthand for general mental ability - is, in the words of Linda Gottfredson, a prolific scholar of human intelligence, “probably the best measured and most studied human trait in all of psychology.” [Gottfredson, Linda S. 2002. “Where and Why g Matters: Not a Mystery.” Human Performance 15 (1/2): 25-46. ]

economist Garett Jones notes, “Across thousands of studies on the correlation across mental abilities across populations, no one has yet found a reliable negative correlation [between performances on two different complex mental tasks]”; [117. Jones, Garett. 2011b. “National IQ and National Productivity: The Hive Mind Across Asia.” Asian Development Review 28 (1): 51-71 ]

Some might challenge IQ’s importance by claiming that all children have about the same academic potential, but for social reasons we feel the need to grade and rank children even though the rankings arise from differences that are very small. But differences in IQ correlate with starkly dissimilar levels of real academic performance. (One striking piece of evidence is that “the ninetieth percentile of nine-year-olds . . . performs in reading, math, and science at the level of the twenty-fifth percentile of seventeen-year-olds.”[120. Gottfredson, Linda S. 2005. “Suppressing Intelligence Research: Hurting Those We Intend to Help,” In Rogers H. Wright and Nicholas A. Cummings (eds.), Destructive Trends in Mental Health: The Well-Intentioned Path to Harm. New York: Taylor and Francis, 155-86. . Footnote omitted.] Because schools sort by age rather than ability, we don’t find smart nine-year-olds in higher grades than not-so-smart seventeen-year-olds, even when the former are more capable than the latter.)

IQ tests taken by children have been found to go a long way toward predicting life span.[126. Gottfredson, Linda. S., and Ian J. Deary. 2004. “Intelligence Predicts Health and Longevity, But Why?”] Higher-IQ individuals also have better dental health, even when controlling for income and ethnicity.[Sabbah, Wael, and Aubrey Sheiham. 2010. “The Relationships Between Cognitive Ability and Dental Status in a National Sample of USA Adults”, Intelligence 38 (6): 605-10] We don’t understand the causes of the health-IQ relationship, but plausible explanations include childhood illnesses that may both reduce a child’s IQ and shorten his life span and the possibility that higher-IQ individuals make better health decisions, get into fewer accidents, choose to live in a healthier environment, follow doctors’ advice better, and more often follow directions when taking medication. Genetics would also explain part of the correlation if the same genes that conferred high intelligence also boosted longevity.[128. Gottfredson & Deary2004.] The positive relationship between a man’s semen quality and his IQ supports the theory that genes play a role in the correlation.[Arden, Rosalind, Linda S. Gottfredson, Geoffrey Miller, and Arand Pierce. 2009. “Intelligence and Semen Quality are Positively Correlated.” Intelligence 37 (3): 277-82] Compounding IQs impact on inequality, higher-IQ people tend to be more physically attractive.[Kanazawa, Satoshi2011. “Intelligence and Physical Attractiveness”. Intelligence 39 (1): 7-14 ] Furthermore, it’s possible to make a decent guess at people’s intelligence just by looking at them. From an article in the online magazine Slate : “In 1918, a researcher in Ohio showed a dozen photographic portraits of well-dressed children to a group of physicians and teachers, and asked the adults to rank the kids from smartest to dumbest. A couple of years later, a Pittsburgh psychologist ran a similar experiment using headshots of 69 employees from a department store. In both studies, seemingly naive guesses were compared to actual test scores and turned out to be accurate more often than not.” Stare at a computer screen until a big green ball appears, and then hit the space bar as quickly as you can. You have just taken a partially reliable IQ test, since a person’s IQ has a positive correlation with reaction time.[132. Gottfredson (2002).]

…magnetic resonance imaging showing a strong positive correlation between brain size and IQ[136. Jones (2011b).]

A person’s IQ is largely, but not completely, determined by age eight.[Heckman, James, Jora Stixrud, and Sergio Urzua. 2006. “The Effects of Cognitive and Noncognitive Abilities on Labor Market Outcomes and Social Behavior”. Journal of Labor Economics 24 (3): 41 1-82] Tests given to infants measuring how much attention the infant pays to novel pictures have a positive correlation with the IQ the infant will have at age twenty-one.[Hunt, Earl. 2011 . Human Intelligence. Cambridge: Cambridge University Press ] The Scottish Mental Survey of 1932 has helped show the remarkable stability of a person’s IQ across his adult life.[Deary, Ian, Martha C. Whiteman, John M. Starr, Lawrence J. Whalley, and Helen C. Fox2004. “The Impact of Childhood Intelligence on Later Life: Following Up the Scottish Mental Surveys of 1932 and 1947”. Journal of Personality and Social Psychology 86 (1): 130-47 ] On June 1, 1932, almost every child in Scotland born in 1921 took the same mental test. Over sixty years later, researchers tracked down some of the test takers who lived in one particular part of Scotland and gave them the test they took in 1932. The researchers found a strong correlation between most people’s 1932 and recent test results.

IQ is the single best predictor of job performance. [Gottfredson, Linda S. 1997. “Why g Matters: The Complexity of Everyday Life.” Intelligence 24 (1): 79–132]

The effect of IQ on wages might be mitigated by the possibility that some high-IQ people are drawn to relatively low-paying professions. Consider, for example, Terence Tao, a reasonable candidate for the smartest person alive today. Terence works as a math professor, and according to Wikipedia, his greatest accomplishment to date is coauthoring a theorem on prime numbers. Prime number research doesn’t pay well, but you can’t do it, and wouldn’t find it interesting, unless you had a super-genius level IQ. Similarly, a poet with an extremely high IQ might have become a lawyer had her IQ been a bit lower because then she wouldn’t have understood the subtleties of poetry that drew her into a poorly remunerated profession. I suspect that many math professors and poets would have higher incomes if some brain injury lowered their IQs just enough to force them out of their professions. If you have an IQ of 135, then you’re already smarter than 99% of humanity. [I’m assuming an IQ standard deviation of 15 for this and all other IQ calculations used in this book.] Would you do better in life if your IQ went well above 135? Research by Heckman says yes. He found that among men with IQs in the top 1% of the population, having a higher IQ boosts wages throughout one’s entire working life, and this effect exists even after taking into account an individual’s level of education.[Gensowski, Miriam, James J. Heckman, and Peter Savelyev. 2011. “The Effects of Education, Personality, and IQ on Earnings of High-Ability Men” Working paper [See also for example SMPY results like Kell et al 2013 ]]

A researcher at the London School of Economics has even shown that one-fourth of the differences in wealth between different US states can be explained by differences in the average IQ of their population.[Kanazawa, Satoshi. 2006. “IQ and the Wealth of States.” Intelligence 34 (6): 593-600. ]

a British study showed that IQ tests given to a group of ten and eleven-year-olds strongly correlated with the level of trust these subjects expressed when they became adults.[Sturgis, Patrick, Sanna Read, and Nick Allum. 2010. “Does Intelligence Foster Generalized Trust? An Empirical Test Using the UK Birth Cohort Studies”. Intelligence 38 (1): 45-54 ]

Having a low IQ makes you, on average, more disposed to crime.[Beaver, Kevin M., and John Paul Wright. 2011. “The Association between County-Level IQ and County-Level Crime Rates”. Intelligence 39 (1): 22-26 ] Since crime reduces productive economic activity, this is another means by which high IQ contributes to economic growth. The higher a person’s IQ, the more likely he is to support economic policies that most economists consider to be healthy for a nation’s economy.[Caplan, Bryan, and Stephen C. Miller. 2010. “Intelligence Makes People Think Like Economists: Evidence from the General Social Survey”. Intelligence 38 (6): 636-47 ] Consequently, if you live in a democracy and trust economists’ judgments, you should want your fellow voters to be smart. Having a high IQ makes you more long-term oriented.[Warner, John T, and Saul Fleeter. 2001. “The Personal Discount Rate: Evidence from Military Downsizing Programs” ] Future-oriented people are more likely to make the kind of calculated long-term investments critical to economic growth.

“Intelligence makes people think like economists: Evidence from the General Social Survey”, Caplan & Miller2010:

Using data from the General Social Survey (GSS), we show that the estimated effect of education sharply falls after controlling for intelligence. In fact, education is driven down to second place, and intelligence replaces it at the top of the list of variables that make people “think like economists.” Thus, to a fair degree education is proxy for intelligence, though there are some areas-international economics in particular-where education still dominates. An important implication is that the political externalities of education may not be as large as they initially appear.

“Generalized Trust and Intelligence in the United States”, Carl & Billari2014:

An extensive empirical literature has established that generalized trust is an important aspect of civic culture. It has been linked to a variety of positive outcomes at the individual level, such as entrepreneurship, volunteering, self-rated health, and happiness. However, two recent studies have found that it is highly correlated with intelligence, which raises the possibility that the other relationships in which it has been implicated may be spurious. Here we replicate the association between intelligence and generalized trust in a large, nationally representative sample of U.S. adults. We also show that, after adjusting for intelligence, generalized trust continues to be strongly associated with both self-rated health and happiness. In the context of substantial variation across countries, these results bolster the view that generalized trust is a valuable social resource, not only for the individual but for the wider society as well.

…Two recent studies have documented a strong correlation between generalized trust and intelligence [16], [17]. Sturgis et al. [“Does intelligence foster generalized trust? An empirical test using the UK birth cohort studies”] analyse data from the U.K., and show that intelligence at age 10-11 predicts generalized trust at age 34, even after conditioning on a large number of socio-economic variables, including self-rated health and happiness. Similarly, Hooghe et al 2012 [“The cognitive basis of trust. The relation between education, cognitive ability, and generalized and political trust”] examine Dutch data, and find that a large part of the association between generalized trust and education is accounted for by cognitive ability.

Hooghe et al 2012:

Previous studies - mainly based on UK data - indeed show a positive relation between intelligence and generalized and political trust (Yamagishi, 2001; Yamagishi, Kikuchi & Kosugi, 1999; Sturgis, Read & Allum, 2010, p. 52; Schoon & Cheng, 2011; Schoon et al 2010). In this line of reasoning, Deary, Batty and Gale (2008, p. 1) stated: “bright children become enlightened adults”.

  • Yamagishi, T. (2001). “Trust as a form of social intelligence”, pp. 121-147 in K. Cook (ed.), Trust in society. New York: Russell Sage Foundation

  • Yamagishi, T., Kikuchi, M. & Kosugi, M. (1999). “Trust, gullibility and social intelligence”. Asian Journal of Social Psychology, 2(1), 145-161

  • Sturgis, P., Read, S. & N. Allum (2010). “Does intelligence foster generalized trust? An empirical test using the UK birth cohort studies”, Intelligence, 38(1), 45-54

  • Schoon, I. & H. Cheng (2011). “Determinants of political trust. A life-long learning model”. Developmental Psychology, 47(3), 619-631

  • Schoon, I., Cheng, H., Gale, C., Batty, D. & Deary, I. (2010). “Social status, cognitive ability, and educational attainment as predictors of liberal social attitudes and political trust”. Intelligence, 38(1), 144-150

“Suppressing Intelligence Research: Hurting Those We Intend to Help”, Gottfredson2005

The results of a 1984 survey (Snyderman & Rothman, 1988) of experts on intelligence and mental testing therefore surprised even Jensen. The experts’ modal response on every question that involved the “heretical” conclusions from Jensen’s1969 article was the same as his (Jensen1998, p. 198). (The experts’ mean response overestimated test bias, however, because there is none against blacks or lower social class individuals; Jensen, 1980; Neisser et al 1996; Snyderman & Rothman, 1988, p. 134; Wigdor & Garner, 1982). Here in abbreviated form are the survey’s major questions and the 600 experts’ responses.

  • Q: What are the important elements of intelligence?

  • A: “Near unanimity” (96-99%) for abstract thinking or reasoning, problem solving ability, and capacity to acquire knowledge (p. 56).

  • Q: Is intelligence best described as a single general factor with subsidiaries or as separate faculties?

  • A: A general factor (58%, or 67% of those responding; p. 71).

  • Q: What heritability would you estimate for IQ differences within the white population?

  • A: Average estimate of 57% (p. 95).

  • Q: What heritability would you estimate for IQ differences within the black population?

  • A: Average estimate of 57% (p. 95).

  • Q: Are intelligence tests biased against blacks?

  • A: On a scale of 1 (not at all or insignificantly) to 4 (extremely), mean response of 2 (somewhat, p. 117).

  • Q: Are intelligence tests biased against lower social class individuals?

  • A: On a scale of 1 (not at all or insignificantly) to 4 (extremely), mean response of 2 (somewhat, p. 118).

  • Q: What is the source of average social class differences in IQ?

  • A: Both genetic and environmental (55%, or 65% of those responding; p. 126).

  • Q: What is the source of the average black-white difference in IQ?

  • A: Both genetic and environmental (45%, or 52% of those responding; p. 128).

The supposedly fringe scientist, Jensen, was actually in the mainstream because the mainstream had silently come to him, where it remains today (Gottfredson, 1997a). Meanwhile, public opinion was still being pushed in the opposite direction, creating an ever greater gulf between received opinion and scientifically informed thought.

…And why keep silent when the media promulgate clear falsehoods as scientific truths-especially when, as Snyderman & Rothman1988 demonstrated, the media portray expert opinion on intelligence as the opposite of what it really is?

Early in my career I reported that bright boys who had attended a school for dyslexics did not enter the usual high-level jobs (medicine, law, science, and college teaching) but had nevertheless succeeded at a high level by entering prestigious or remunerative occupations that required above-average intelligence but relatively little reading or writing, specifically, top management and sales positions. A colleague accused me in that seminar of saying that “blacks can’t make it because they are dumb.”

The best informed, who are often called upon for expert comment, cannot endorse clear falsehoods without jeopardizing their own standing within the discipline, but they sometimes dispute minor issues in a manner that the uninformed mistake for wholesale repudiation (Gottfredson, 1994a; Page, 1972).

The implication of ABC’s November 22, 1994, national newscast was surely not lost on viewers when, while exposing the supposedly unsavory history of intelligence research behind The Bell Curve, news anchor Peter Jennings followed photographs of Jensen and other supposed race scientists with footage of Nazi soldiers and what appeared to be death camp doctors and prisoners.

Critics have associated a belief in the hereditary basis of intelligence with evil intent so frequently and for so long that merely mentioning “IQ” is enough to trigger in many minds the words “pseudoscience,” “racism,” and “genocide.” Even current APA president Robert Sternberg keeps the malicious association alive by regularly ridiculing and belittling empirically-minded intelligence researchers (eg. comparing Jensen, in a book meant to honor him, to a child who would not grow up; Sternberg, 2003), referring to their work as “quasi-science” (“Science and pseudoscience,” 1999, p. 27) that has “recreated a kind of night of the living dead” (Sternberg, 1997, p. 55), and sprinkling his descriptions of it with mentions of racism, slavery, and even Soviet tyranny (eg. Sternberg, 2003; see also Sternberg, 2000, Sternberg & Wagner, 1993). But why should we assume that a belief in the heritability of many human differences is dangerous and a belief in man’s infinite malleability is not? Critics have yet to explain. Why is the former belief always yoked to Hitler, but the latter never to Stalin, who outlawed both intelligence tests and genetic thinking? Stalin killed at least as many as did Hitler in his effort to reshape the Soviet citizenry (Courtois, 1999).

Behavior geneticists distinguish between two types of environmental influence: shared and non-shared (also called between-family and within-family effects). Shared influences are those that make siblings more alike. Possible such influences would include parental income, education, child-rearing style, and the like, because they would impinge on all siblings in a household. Non-shared influences are those that affect individuals one person at a time and therefore make siblings less alike. Little is yet known about them, but they might include illness, accidents, non-genetic influences on fetal development, and the concatenation of unique experiences. To the great surprise even of behavior geneticists, shared environmental effects on intelligence (within the broad range of typical environments) wash away by late adolescence. IQ differences can be traced to both genes (40%) and shared environments (25%) in early childhood, but genetic effects increase in importance with age (to 80% in adulthood) while shared effects dissipate (Plomin, DeFries, McClearn, & McGuffin, 2001). For example, adoptive siblings end up no more alike in IQ or personality by adolescence than are random strangers, and instead become similar to the biological relatives they have never met.

Currently one of the biggest puzzles for family effects theory is that academic achievement gaps do not narrow even in settings where all the supposedly important environmental resources are present (Banchero & Little, 2002). For example, its adherents are now arguing among themselves (Lee, 2002) about the proper cultural explanation for the large black-white achievement gaps that persist in the most socioeconomically advantaged, integrated, liberal, suburban school districts in the United States, such as Shaker Heights, Ohio (Ogbu, 2003) and Berkeley, California (Noguera, 2001). Moreover, black-white test score gaps (IQ, SAT, etc.) tend to be larger at higher socioeconomic levels. This finding contradicts the predictions of family effects theory. It is consistent with g-based theory, however, because the latter predicts that black and white children of high-IQ parents will regress part way from their parents’ mean toward different population means, IQ 100 for whites and IQ 85 for blacks.

The fictions about intelligence essentially deny that it exists, which virtually no one really believes. Many people just want a more “democratic” view of it. Not surprisingly, psychology’s supply has risen to meet public demand, and the new egalitarian perspectives on human intelligence were instantly blessed by opinion makers. Chief among them are the “multiple intelligence” theories by psychologists Howard Gardner (1983, 1998) and Robert Sternberg (1997). The eager acceptance of their theories by educators, psychologists, and others has occurred despite neither of them providing credible evidence that their proposed intelligences actually exist, that is, as independent abilities of comparable generality and practical importance to g. Gardner has rejected even measuring his eight intelligences, let alone demonstrating that they predict anything (Hunt, 2001; Lubinski & Benbow, 1995). Study-by-study dissections of Sternberg’s multiple-intelligence research program reveal no such evidence (Brody, 2003a, b; Gottfredson, 2003a, c). If anything, they confirm that all three of his proposed intelligences are just different flavors of g itself, as probably are most of Gardner’s too (Carroll, 1993, p. 641).

In 1991, the U.S. Congress voted overwhelmingly to outlaw race-norming in employment after it learned that the Labor Department had already been race-norming its employment tests for a decade and that the U. S. Equal Employment Opportunity Commission (EEOC) had started threatening private employers if they did not adopt the “scientifically-justified” practice. The racial preferences that race-norming entails are hardly trivial. What the NRC report did not say was that blacks scoring at the 15th percentile in skill level on DOL’s test would have been judged equal to whites and Asians scoring at the 50th percentile, and blacks at the 50th percentile would be rated comparably skilled as whites and Asians at the 84th (Blits & Gottfredson, 1990a). Seldom being apprised of such facts, most people greatly underestimate how discrepant the pools of qualified applicants are from which racial balance is supposed to emerge. Another illustration, pertinent to the next example, is that about 75% of whites vs. only 28% of blacks exceed the minimum IQ level (~IQ 91)-a ratio of 3 to 1-usually required for minimally satisfactory performance in the skilled trades, fire and police work, and mid-level clerical jobs such as bank teller (Gottfredson, 1986, pp. 400-401). The potential pools become increasingly racially lopsided for more cognitively demanding jobs. Workers in professional jobs such as engineer, lawyer, and physician typically need an IQ of at least 114 to perform satisfactorily. About 23% of whites but only 1% of blacks exceed this minimum….Developing tests that measure cognitive skills more effectively tends only to worsen the proscribed disparate impact. Adding relevant non-cognitive predictors to the mix does little to reduce the racial imbalance (Schmitt, Rogers, Chan, Sheppard, & Jennings, 1997).

The police selection test developed in 1994 for Nassau County, NY, represents one such “technical advance.” The 10 members of a joint Nassau County-U.S. Department of Justice (DOJ) team had set out to develop a police selection test with less disparate impact (more racially balanced results). The county had not been able to satisfy the DOJ’s employment discrimination unit in several tries under its various consent decrees since 1977. (Recall the 3 to 1 ratio given above for the proportion of whites vs. blacks exceeding the ability level below which performance in police work tends to be unsatisfactory.) Seven of the team’s eight psychologists constituted a Who’s Who of APA’s large Division 14 (Industrial and Organizational Psychology), four of them having previously served as its president. Several years and millions of dollars later, this high-powered team claimed to have succeeded in developing a test that virtually eliminated disparate impact while simultaneously improving selection validity. Water could run up-hill, after all. Once again, leading psychologists found a seemingly scientific solution to an intractable political-legal dilemma. DOJ immediately began pressing other police jurisdictions nationwide to replace their more “discriminatory” tests with the new selection battery. A close look at the several-volume technical report for the Nassau test battery revealed that the team had succeeded in reducing disparate impact by, in effect, gerrymandering the test to assess only traits on which the races differed little or not at all (Gottfredson, 1996a, b). The joint Nassau-DOJ team had administered its nearly day-long, 25-part experimental battery to all 25,000 applicants, but settled on the battery’s final composition only after examining the scores it yielded for different races. The experimental battery was then apparently stripped of virtually all parts demanding cognitive ability. The only parts actually used to rank applicants were eight non-cognitive personality scales (all commercial products owned by members of the team) and being able to read above the 1st percentile of currently employed police officers (near illiteracy). Selection for cognitive competence had been reduced to little more than the toss of a coin, despite the team’s own careful job analysis having shown that “reasoning, judgment, and inferential thinking” were the most critical skills for good police work. The new police test was made to appear more valid than the county’s previous ones by, among other things, omitting key results required by legal and professional guidelines, transforming the data in ways that artificially reduced the apparent validity of the cognitive subtests relative to the non-cognitive ones, and making a series of statistical errors that more than doubled the final battery’s apparent predictive validity (.14 → .35). When exposed, the test created a scandal in Division 14 (“The Great Debate of 1997” in Hakel, 1997, p. 116), partly because other leading selection psychologists expected its use would produce less effective policing and degrade public safety (Schmidt, 1996).

Even the most objective, most carefully vetted procedures for identifying talent are instantly pronounced guilty of bias or “exclusion” when they yield disparate impact in hiring, college admissions, placement in gifted education, and the like. Indeed, the very notions of objectivity and merit are now under attack by influential intellectual elites (Farber & Sherry, 1997). When faithful and fair application of the law yields disparate impact in arrest or incarceration rates, American jurisprudence must be considered inherently racist (see arguments in Crenshaw, Gotanda, Peller, & Thomas, 1995). When earnest, socially liberal teachers fail to narrow the stubborn achievement gaps between races and classes, they must be unconsciously discriminatory and require diversity training. Because American institutions still routinely and almost everywhere fail to yield the desired racial balance, the Americans who created and supposedly control those institutions-majority Americans-must be judged deeply, unconsciously, inveterately racist and to have created a society where appearances to the contrary are just a smokescreen to hide their built-in privileges. Under the equipotentiality fiction, there can be no other legitimate explanation, and any attempt at one serves only to evade responsibility.

…Fewer but still many social scientists hold to a fourth false credo-that intelligence has little or no functional utility, at least outside schools. Moreover, they often add that the advantages and disadvantages of high or low IQ are mostly “socially constructed” to serve the interests of the privileged. This view was articulated in an influential article published soon after Jensen’s1969 article by economists Samuel Bowles and Herbert Gintis (1972/1973). They argued that higher IQ does not have any functional utility, even within schools, and that IQ tests are simply a tool created by the upper classes to maintain and justify their privileges. They dismissed talk of “objectivity” and “merit” as just smoke blown to obscure this fact. Psychologist Robert Sternberg implies much the same when he suggests that the g factor dimension of intellectual differences is an artifact of Western schooling (Sternberg et al 2000, p. 9) and that using cognitive tests such as the SAT to sort people is akin to the way slavery and religious prejudice were once used to keep disfavored groups down (Sternberg, 2003).

However, when critics argue that IQ differences have little or no functional meaning beyond that which cultures or their elites arbitrarily attach to them for selfish purposes, they simultaneously turn attention away from the very real problems that lower intelligence creates for less able persons. As Herrnstein & Murray1994 note, the critics generally have little contact with the downtrodden they would protect. These bright opinion makers may be living comfortably with their fictions and benevolent lies, but lower-IQ individuals must live daily with the consequences of their weaker learning and reasoning skills. Their distant protectors would seem to be the limousine liberals of intelligence.

I focus below on everyday tasks that higher-IQ individuals consider so simple that they do not realize how such tasks might create obstacles to the well-being of others less cognitively blessed.

Functional literacy and daily self-maintenance. Citizens of literate societies take for granted that they are routinely called upon to read instructions, fill out forms, determine best buys, decipher bus schedules, and otherwise read and write to cope with the myriad details of everyday life. But such tasks are difficult for many people. The problem is seldom that they cannot read or write the words, but usually that they are unable to carry out the mental operations the task calls for-to compare two items, grasp an abstract concept, provide comprehensible and accurate information about themselves, follow a set of instructions, and so on. This is what it means to have poor “functional literacy.” Functional literacy has been a major public policy concern, as illustrated by the U.S. Department of Education’s various efforts to gauge its level in different segments of the American population. Tests of functional literacy essentially mimic individually-administered intelligence tests, except that all their items come from everyday life, such as calculating a tip (see extended discussion in Gottfredson, 1997b). As on intelligence tests, differences in item difficulty rest on the items’ cognitive complexity (their abstractness, amount of distracting irrelevant information, and degree of inference required), not on their readability per se or the level of education test takers have completed. Literacy researchers have concluded, with some surprise, that functional literacy represents a general capacity to learn, reason, and solve problems-a veritable description of g.

The National Adult Literacy Survey (NALS; Kirsch, Jungeblut, Jenkins, & Kolstad, 1993) groups literacy scores into five levels. Individuals scoring in Level 1 have an 80% chance of successfully performing tasks similar in difficulty to locating an expiration date on a driver’s license and totaling a bank deposit slip. They are not routinely able to perform Level 2 tasks, such as determining the price difference between two show tickets or filling in background information on an application for a social security card. Level 3 difficulty includes writing a brief letter explaining an error in a credit card bill and using a flight schedule to plan travel. Level 4 tasks include restating an argument made in a lengthy news article and calculating the money needed to raise a child based on information in a news article. Only at Level 5 are individuals routinely able to perform mental tasks as complex as summarizing two ways that lawyers challenge prospective jurors (based on a passage discussing such practices) and, with a calculator, determining the total cost of carpet to cover a room.

Although these tasks might seem to represent only the inconsequential minutiae of everyday life, they sample the large universe of mostly untutored tasks that modern life demands of adults. Consistently failing them is not just a daily inconvenience, but a compounding problem. Likening functional literacy to money-it always helps to have more-, literacy researchers point out that rates of socioeconomic distress and pathology (unemployment, adult poverty, etc.) rise steadily at successively lower levels of functional literacy (as is the pattern for IQ too; Gottfredson, 2002a)…Such disadvantage is common, too, because 40% of the adult white population and 80% of the adult black population cannot routinely perform above Level 2. Fully 14% and 40%, respectively, cannot routinely perform even above Level 1 (Kirsch et al 1993, pp. 119121). To claim that lower-ability citizens will only be victimized by the public knowing that differences in intelligence are real, stubborn, and important is to ignore the practical hurdles they face.

Health literacy, IQ, and health self-care. The challenges in self-care for lower-IQ individuals are especially striking in health matters, where the consequences of poor performance are tallied in excess morbidity and mortality. Health psychologists have ignored the role of competence in health behavior, focusing instead on volition. Patient “non-compliance” is indeed a huge problem in medicine, but health literacy researchers, unlike health psychologists, have concluded that it is more a matter of patients not understanding what is required of them than being unwilling to implement it (reviews in Gottfredson, 2002a, in press).

…For example, 26% of outpatients in several large urban hospitals could not determine from an appointment slip when the next visit was scheduled and 42% could not understand instructions for taking medicine on an empty stomach. Among those with “inadequate” literacy, the failure rates on these two tasks were 40% and 65%, respectively. Substantial percentages of this low-literacy group were unable to report, when given prescription labels containing the necessary information, how to take the medication four times a day (24%), how many times the prescription could be refilled (42%), or how many pills of the prescription should be taken (70%). Taking medications improperly can be as harmful as not taking them at all, and the pharmacy profession has estimated that about half of all prescriptions are taken incorrectly. As in other performance domains, training and motivation do not erase the disadvantages of lower comprehension abilities. For instance, many patients who are under treatment for insulin-dependent diabetes do not understand the most elemental facts for maintaining daily control of their disease. In one study, about half of those with “inadequate” literacy did not know the signs of very low or very high blood sugar, both of which require expeditious correction, and 60% did not know the corrective actions to take. Like hypertension and many other chronic illnesses, diabetes requires continual self-monitoring and frequent judgments by patients to keep their physiological processes within safe limits during the day. Persistently high blood sugar levels can lead to blindness, heart disease, limb amputation, and much more. For persons in general, low functional literacy has been linked to number and severity of illnesses, worse self-rated health, far higher medical costs, and (prospectively) more frequent hospitalization. These relations are not eliminated by controlling for education, socioeconomic resources, access to health care, demographic characteristics, and other such variables.

Because health literacy is a rough surrogate for g, it produces results consistent with research on IQ and health. To take several examples, intelligence at time of diagnosis correlates .36 with diabetes knowledge measured one year later (Taylor, Frier, Gold, & Deary, in press). IQ measured at age 11 predicts longevity, incidence of cancer, and functional independence in old age, and these relations remain robust after controlling for deprived living conditions (Deary, Whiteman, Starr, & Whalley, in press). Another prospective epidemiological study found that the motor vehicle death rate for men of IQ 80-85 was triple and for men of IQ 85-100 it was double the rate for men of IQ 100-115 (O’Toole, 1990). Youthful IQ was the best predictor of all-cause mortality by age 40 in this large national sample of Australian Army veterans, and IQ’s predictive value remained significant after controlling for all 56 demographic, health, and other attributes measured (O’Toole & Stankov, 1992). As in education, equal resources do not produce equal outcomes in health. Like educational inequalities, health inequalities increase when health resources become equally available to all, such as happened to the British government’s dismay after it instituted free national health care. Health improves overall, but least for less educated and lower income persons. They seek more but not necessarily appropriate care when cost is no barrier; adhere less often to treatment regimens; learn and understand less about how to protect their health; seek less preventive care, even when free; and less often practice the healthy behaviors so important for preventing or slowing the progression of chronic diseases, the major killers and disablers in developed nations.

…Infusing more knowledge into the public sphere about health risks (smoking) and new diagnostic options (Pap smears) results in already-informed persons learning the most and more often acting on the new information. This may explain why an SES-mortality gradient favoring educated women developed for cervical cancer after Pap smears became available.

…After it became clear that health inequalities could not be explained by inequalities in material resources and access to health care, it became fashionable in health epidemiology to blame class and race differences in health on the psychic damage done by social inequality. We are now to believe that social inequality per se is literally a killer (Wilkinson, 1996). Physicians, like teachers, are increasingly being accused of racism and given sensitivity training when they fail to produce racial parity in outcomes (Satel, 2000). Mindful of ideologically correct thought, health literacy researchers who mention intelligence do so only to reject out of hand the notion that literacy might reflect intelligence, because any such notion would be racist and demeaning.

In the meantime, inadequate learning and reasoning abilities put many people at risk of taking medications in health-damaging ways, not grasping the merits of preventive precautions against chronic disease and accidents, and failing to properly implement potentially more effective but complex new treatment regimens for heart disease, hypertension, and other killers.

…To intentionally ignore differences in mental competence is unconscionable. It is social science malpractice against the very people whom the “untruth” is supposedly meant to protect.

Gottfredson, Linda S. 2002. “Where and Why g Matters: Not a Mystery.” Human Performance 15 (1/2): 25-46.

g is a highly general capability for processing complex information of any type. This explains its great value in predicting job performance. Complexity is the major distinction among jobs, which explains why g is more important further up the occupational hierarchy. The predictive validities of g are moderated by the criteria and other predictors considered in selection research, but the resulting gradients of g’s effects are systematic. The pattern provides personnel psychologists a road map for how to design better selection batteries.

…One of the simplest facts about mental abilities provides one of the most important clues to the nature of g. People who do well on one kind of mental test tend to do well on all others. When the scores on a large, diverse battery of mental ability tests are factor analyzed, they yield a large common factor, labeled g. Pick any test of mental aptitude or achievement-say, verbal aptitude, spatial visualization, the SAT, a standardized test of academic achievement in 8th grade, or the Block De- sign or Memory for Sentences subtests of the Stanford-Binet intelligence test- and you will find that it measures mostly g. All efforts to build meaningful mental tests that do not measure g have failed…In contrast, no general factor emerges from personality inventories, which shows that general factors are not a necessary outcome of factor analysis. (See Jensen1998, and Gottfredson, 1997, 2000a, 2002, for fuller discussion and documentation of these and following points on g.)

The important point is that the predictive validities of g behave lawfully. They vary, but they vary systematically and for reasons that are beginning to be well understood. Over 2 decades of meta-analyses have shown that they are not sensitive to small variations in job duties and circumstance, after controlling for sampling error and other statistical artifacts. Complex jobs will always put a premium on higher g. Their performance will always be notably enhanced by higher g, all else equal. Higher g will also enhance performance in simple jobs, but to a much smaller degree.

“Genome-wide association studies establish that human intelligence is highly heritable and polygenic”:

General intelligence is an important human quantitative trait that accounts for much of the variation in diverse cognitive abilities. Individual differences in intelligence are strongly associated with many important life outcomes, including educational and occupational attainments, income, health and lifespan1,2. Data from twin and family studies are consistent with a high heritability of intelligence3, but this inference has been controversial. We conducted a genome-wide analysis of 3511 unrelated adults with data on 549 692 SNPs and detailed phenotypes on cognitive traits. We estimate that 40% of the variation in crystallized-type intelligence and 51% of the variation in fluid-type intelligence between individuals is accounted for by linkage disequilibrium between genotyped common SNP markers and unknown causal variants. These estimates provide lower bounds for the narrow-sense heritability of the traits. We partitioned genetic variation on individual chromosomes and found that, on average, longer chromosomes explain more variation. Finally, using just SNP data we predicted approximately 1% of the variance of crystallized and fluid cognitive phenotypes in an independent sample (P = 0.009 and 0.028, respectively). Our results unequivocally confirm that a substantial proportion of individual differences in human intelligence is due to genetic variation, and are consistent with many genes of small effects underlying the additive genetic influences on intelligence.

“Common DNA Markers Can Account for More Than Half of the Genetic Influence on Cognitive Abilities”:

For nearly a century, twin and adoption studies have yielded substantial estimates of heritability for cognitive abilities, although it has proved difficult for genome-wide-association studies to identify the genetic variants that account for this heritability (ie. the missing-heritability problem). However, a new approach, genome-wide complex-trait analysis (GCTA), forgoes the identification of individual variants to estimate the total heritability captured by common DNA markers on genotyping arrays. In the same sample of 3,154 pairs of 12-year-old twins, we directly compared twin-study heritability estimates for cognitive abilities (language, verbal, nonverbal, and general) with GCTA estimates captured by 1.7 million DNA markers. We found that DNA markers tagged by the array accounted for .66 of the estimated heritability, reaffirming that cognitive abilities are heritable. Larger sample sizes alone will be sufficient to identify many of the genetic variants that influence cognitive abilities.

…Cognitive abilities predict educational attainment, income, health, and longevity, and thus contribute importantly to the intellectual capital of knowledge-based societies (Deary, 2012). Since the 1920s, twin and adoption studies have investigated the genetic and environmental origins of individual differences in cognitive abilities; scores of such studies have consistently yielded estimates of substantial heritability (ie. the extent to which genetic variance can account for observed, or phenotypic, variance; Deary, Johnson, & Houlihan, 2009). Meta-analyses of these studies have yielded heritability estimates of about .50 for general cognitive ability, the most well-studied cognitive trait (Plomin, DeFries, Knopik, & Neuhouser, 2013).

…One of the most far-reaching results of GWA studies is to show that there are no genes of large effect size in the population, which means that the heritability of complex traits is probably due to many genes of small effect size, and this means that associations will be difficult to detect and replicate (Plomin, 2012). For example, the first GWA studies of general cognitive ability (Davies et al 2011; Davis et al 2010) were powered to detect associations that account for as little as .01 of the variance, but they came up empty-handed because the associations with the largest effect accounted for less than .005 of the variance. One of many possible reasons for the missing-heritability problem is that the common SNPs (ie. SNPs for which the frequency of the less frequent allele is greater than .01) incorporated in commercially available DNA arrays miss the contribution of rare DNA variants (Cirulli & Goldstein, 2010). Another possibility is that heritability has been overestimated by twin and adoption studies.

…The study reported here addressed both of these possibilities by comparing twin-based estimates of heritability for cognitive abilities with estimates from a new method that is population based rather than family based. The method, called genome-wide complex-trait analysis (GCTA), can be used to estimate genetic variance accounted for by all the SNPs that have been genotyped in any sample, not just samples consisting of special family members such as twins or adoptees (Lee, Wray, Goddard, & Visscher, 2011; Yang, Lee, Goddard, & Visscher, 2011; Yang, Manolio, et al 2011)…GCTA does not identify specific genes associated with traits. Instead, it uses chance similarity across hundreds of thousands of SNPs to predict phenotypic similarity pair by pair in a large sample of unrelated individuals. The essence of GCTA is to estimate genetic influence on a trait by predicting phenotypic similarity for each pair of individuals in the sample from their total SNP similarity. In contrast to the twin method, which estimates heritability by comparing phenotypic similarity of identical and fraternal twin pairs, whose genetic similarity is roughly 1.00 and .50, respectively, GCTA relies on comparisons of pairs of individuals whose genetic similarity varies .00–.02. GCTA extracts this tiny genetic signal from the noise of hundreds of thousands of SNPs using the massive information available from a matrix of thousands of individuals, each compared pair by pair with every other individual in the sample; for example, the 3,000-plus individuals in the present sample provided nearly 5 million pairwise comparisons

…GCTA has been used to estimate heritability as captured by genotyping arrays for height (Yang et al 2010), weight (Yang, Manolio, et al 2011), psychiatric and other medical disorders (Lee et al 2012; Lee et al 2011; Lubke et al 2012), and personality (Vinkhuyzen, Pedersen, et al 2012). GCTA was first applied to cognitive ability in a study of 3,500 unrelated adults, which yielded heritability estimates of .40 and .51 for crystallized and fluid intelligence, respectively (Davies et al 2011). The GCTA estimate for general cognitive ability was .47 in a meta-analysis across three studies involving nearly 10,000 adults (Chabris et al 2012) and .48 in a study of nearly 2 thousand 11-year-old children (Deary et al 2012)…GCTA has been used to estimate heritability as captured by genotyping arrays for height (), weight (Yang, Manolio, et al 2011), psychiatric and other medical disorders (; ; ), and personality (Vinkhuyzen, Pedersen, et al 2012). GCTA was first applied to cognitive ability in a study of 3,500 unrelated adults, which yielded heritability estimates of .40 and .51 for crystallized and fluid intelligence, respectively (). The GCTA estimate for general cognitive ability was .47 in a meta-analysis across three studies involving nearly 10,000 adults () and .48 in a study of nearly 2 thousand 11-year-old children ().

…GCTA has been used to estimate heritability as captured by genotyping arrays for height (Yang et al 2010), weight (Yang, Manolio, et al 2011), psychiatric and other medical disorders (Lee et al 2012; Lee et al 2011; Lubke et al 2012), and personality (Vinkhuyzen, Pedersen, et al 2012). GCTA was first applied to cognitive ability in a study of 3,500 unrelated adults, which yielded heritability estimates of .40 and .51 for crystallized and fluid intelligence, respectively (Davies et al 2011). The GCTA estimate for general cognitive ability was .47 in a meta-analysis across three studies involving nearly 10,000 adults (Chabris et al 2012) and .48 in a study of nearly 2 thousand 11-year-old children (Deary et al 2012)…This is the first study in which GCTA estimates of heritability for diverse cognitive abilities were compared directly with twin-based estimates using the same measures at the same age in the same sample. The Affymetrix 6.0 DNA array yielded GCTA estimates that accounted on average for .66 of the twin heritability estimates for language, verbal, nonverbal, and general cognitive abilities. Note that the GCTA estimates accounted for a greater proportion of the twin heritability estimates in the case of cognitive abilities than in the case of height (.44) and weight (.50).

…Why might these common SNPs tag general cognitive ability more than height and weight? Common SNPs are likely to be common because they are old, having spread through the population over many generations, but there seems no obvious reason why the evolutionary architecture for general cognitive ability should differ from height in this way. However, there is one major genetic difference between cognitive and physical traits: Assortative mating (nonrandom mating) is at least twice as great for general cognitive ability (correlation between spouses: ~.45) as for height and weight (~.20; Plomin et al 2013). The effect of assortative mating is to increase additive genetic variance because children receive correlated genetic influences from their parents, which spreads out the distribution; moreover, the effects of assortative mating accumulate generation after generation. If assortative mating is responsible for the fact that common SNPs tag general cognitive ability more than height and weight, then verbal abilities should show greater GCTA/twin heritability ratios than nonverbal abilities do because verbal abilities show more assortative mating than nonverbal abilities (correlation between spouses: ~.50 vs. .30). The results in Table 1 are consistent with this hypothesis: The GCTA/twin heritability ratio is .65 for verbal ability and .48 for nonverbal ability.

Above a certain level, intelligence doesn’t matter. There was no significant difference in maximum income earned by men with IQs in the 110-115 range and men with IQs higher than 150.

TODO: what’s going on there? weasel wording on ‘maximum’? not a big enough sample size to reach statistical-significance

“Deliberate practice: Is that all it takes to become an expert?”, Hambrick et al 2014:

…Global measures of intelligence (IQ) have also been found to correlate with performance in chess and music, consistent with the possibility that a relatively high level of intelligence is necessary for success in these domains. Frydman & Lynn 1992 found that young chess players had an average performance IQ of 129, compared to a population average of 100, and that the average was higher for the best players (top-third avg. = 131) in the sample than the weakest players (bottom-third avg. = 124). Furthermore, Grabner, Neubauer, and Stern (2006) found that, even in highly rated players, IQ positively predicted performance on representative chess tasks (eg. next best move). Bilalić et al 2007 found that IQ was not a significant predictor of chess rating in the sample of elite young chess players listed in Table 1 after statistically controlling for practice. However, the sample size for the elite group was only 23, and mean IQ was significantly higher for the elite group (M = 133) than for the rest of the sample (M = 114).

  • Frydman, M., & Lynn, R. (1992). The general intelligence and spatial abilities of gifted young Belgian chess players. British Journal of Psychology, 83, 233-235.

  • Grabner, R. H., Neubauer, A. C., & Stern, E. (2006). Superior performance and neural efficiency: The impact of intelligence and expertise. Brain Research Bulletin, 69, 422-439.

  • Bilalić, M., McLeod, P., & Gobet, F. (2007). Does chess need intelligence? A study with young chess players. Intelligence, 35, 457-470.

IQ correlates positively with music performance, as well. Luce (1965) found a correlation of .53 (p b .01) between IQ and sight-reading performance in high school band members, and Salis (1977) reported a correlation of .58 between these variables in a university sample. Gromko (2004) found positive correlations between both verbal ability and spatial ability (rs = .35-.49) and sight-reading performance in high school wind players, and Hayward & Gromko 2009 found a significant positive correlation (r = .24) between a measure of spatial ability based on three ETS tests and sight-reading performance in university wind players. Ruthsatz et al 2008 found that Raven’s scores correlated positively and significantly with musical achievement in high school band members (r = .25). This correlation was not statistically significant in a sample of more highly accomplished conservatory students and music majors, but this could have been due to a ceiling effect on Raven’s, as these participants had been heavily selected for cognitive ability.

  • Luce, J. R. (1965). Sight-reading and ear-playing abilities as related to instrumental music students. Journal of Research in Music Education, 13, 101-109.

  • Salis, D. L. (1977). The identification and assessment of cognitive variables associated with reading of advanced music at the piano (unpublished doctoral dissertation). Pittsburgh, PA: University of Pittsburgh.

  • Gromko, J. E. (2004). Predictors of music sight-reading ability in high school wind players. Journal of Research in Music Education, 52, 6-15.

  • Hayward, C. M., & Gromko, J. E. (2009). Relationships among music sight-reading and technical proficiency, spatial visualization, and aural discrimination. Journal of Research in Music Education, 57, 29-36.

  • Ruthsatz, J., Detterman, D. K., Griscom, W. S., & Cirullo, B. A. (2008). Becoming an expert in the musical domain: It takes more than just practice. Intelligence, 36, 330-338.

Ruthsatz & Urbach2012 administered a standardized IQ test (the Stanford-Binet) to eight child prodigies, six of whom were musical prodigies. Despite full-scale IQs that ranged 108–147-just above average to above the conventional cutoff for “genius”-all of the prodigies were at or above the 99th percentile for working memory (indeed, six scored at the 99.9th percentile).

  • Ruthsatz, J., & Urbach, J. B. (2012). Child prodigy: A novel cognitive profile places elevated general intelligence, exceptional working memory and attention to detail at the root of prodigiousness. Intelligence, 40, 419-426.

…General intelligence does not always predict performance. In a study of football players, Lyons, Hoffman, and Michel (2009) found that scores on the Wonderlic Personnel Test, a widely administered group intelligence test, correlated essentially zero with success in the National Football League, even in the quarterback position, which is believed to place the highest demand on information processing. Furthermore, Hambrick et al 2012 found that spatial ability positively predicted success in a complex geological problem solving task in novice geologists, but not in experts.

  • Lyons, B., Hoffman, B., & Michel, J. (2009). Not much more than g? An examination of the impact of intelligence on NFL performance. Human Performance, 22, 225-245.

  • Hambrick, D. Z., & Meinz, E. J. (2012). Working memory capacity and musical skill. In T. P. Alloway, & R. G. Alloway (Eds.), Working memory: The connected intelligence (pp. 137-155). New York: Psychology Press.

“Intelligence and semen quality are positively correlated”, Arden et al 2009

If the correlations among cognitive abilities are part of a larger matrix of positive associations among fitness-related traits, then intelligence ought to correlate with seemingly unrelated traits that affect fitness-such as semen quality. We found significant positive correlations between intelligence and 3 key indices of semen quality: log sperm concentration (r = .15, p = .002), log sperm count (r = .19, p b .001), and sperm motility (r = .14, p = .002) in a large sample of US Army Veterans. None was mediated by age, body mass index, days of sexual abstinence, service in Vietnam, or use of alcohol, tobacco, marijuana, or hard drugs.

…intelligence correlates with many important health outcomes, even longevity (Batty, Deary, & Gottfredson, 2007).

  • Batty, G. D., Deary, I. J., & Gottfredson, L. S. (2007). Premorbid (early life) IQ and later mortality risk: Systematic review. Annals of Epidemiology, 17 (4), 278−288.

…the effect size is congruent with phenotypic correlations observed for other bodily correlates of intelligence such as height (r = .14, r = .15) (Silventoinen, Posthuma, van Beijsterveldt, Bartels, & Boomsma, 2006; Sundet, Tambs, Harris, Magnus, & Torjussen, 2005).

  • Silventoinen, K., Posthuma, D., van Beijsterveldt, T., Bartels, M., & Boomsma, D. I. (2006). Genetic contributions to the association between height and intelligence: Evidence from Dutch twin data from childhood to middle age. Genes, Brain and Behavior, 5(8), 585−595.

  • Sundet, J. M., Tambs, K., Harris, J. R., Magnus, P., & Torjussen, T. M. (2005). Resolving the genetic and environmental sources of the correlation between height and intelligence: A study of nearly 2600 Norwegian male twin pairs. Twin Research and Human Genetics, 8(4), 307−311.


  1. Deary IJ, Strand S, Smith P, Fernandes C (2007) Intelligence and educational achievement. Intelligence 35: 13-21 doi:10.1016/j.intell.2006.02.001. . doi: 10.1016/j.intell.2006.02.001. IQ scores are often used as an index of general cognitive abilities. Such IQ measures exhibit substantial correlations from late childhood through adulthood (eg. IQ scores were estimated to correlate 0.73 from ages 11 through 77 in a longitudinal study [2]).

  2. Deary IJ, Whalley LJ, Lemmon H, Crawford J, Starr JM (2000) The Stability of Individual Differences in Mental Ability from Childhood to Old Age: Follow-up of the 1932 Scottish Mental Survey. Intelligence 28: 49-55 doi:10.1016/S0160-2896(99)00031-8. . doi: 10.1016/S0160-2896(99)00031-8.

“Childhood cognitive ability accounts for associations between cognitive ability and brain cortical thickness in old age”, Karama et al 2013

We analyzed data on 588 subjects from the Lothian Birth Cohort 1936 who had intelligence quotient (IQ) scores from the same cognitive test available at both 11 and 70 years of age as well as high-resolution brain magnetic resonance imaging data obtained at approximately 73 years of age. Cortical thickness was estimated at 81,924 sampling points across the cortex for each subject using an automated pipeline. Multiple regression was used to assess associations between cortical thickness and the IQ measures at 11 and 70 years. Childhood IQ accounted for more than two-third of the association between IQ at 70 years and cortical thickness measured at age 73 years. This warns against ascribing a causal interpretation to the association between cognitive ability and cortical tissue in old age based on assumptions about, and exclusive reference to, the aging process and any associated disease.

“Brain Fiber Architecture, Genetics, and Intelligence: A High Angular Resolution Diffusion Imaging (HARDI) Study”, Chiang et al 2011

We developed an analysis pipeline enabling population studies of HARDI data, and applied it to map genetic influences on fiber architecture in 90 twin subjects. We applied tensor-driven 3D fluid registration to HARDI, re-sampling the spherical fiber orientation distribution functions (ODFs) in appropriate Riemannian manifolds, after ODF regularization and sharpening. Fitting structural equation models (SEM) from quantitative genetics, we evaluated genetic influences on the Jensen-Shannon divergence (JSD), a novel measure of fiber spatial coherence, and on the generalized fiber anisotropy (GFA) a measure of fiber integrity. With random-effects regression, we mapped regions where diffusion profiles were highly correlated with subjects’ intelligence quotient (IQ). Fiber complexity was predominantly under genetic control, and higher in more highly anisotropic regions; the proportion of genetic versus environmental control varied spatially. Our methods show promise for discovering genes affecting fiber connectivity in the brain.

“Investigating America’s elite: Cognitive ability, education, and sex differences”, Wai2013

Are the American elite drawn from the cognitive elite?…However, whether the elite are primarily composed of individuals in the top percentiles of the ability distribution who have attended the most prestigious colleges and universities has not yet been empirically examined…To address this, five groups of America’s elite (total N = 2254) were examined: Fortune 500 CEOs, federal judges, billionaires, Senators, and members of the House of Representatives. Within each of these groups, nearly all had attended college with the majority having attended either a highly selective undergraduate institution or graduate school of some kind. High average test scores required for admission to these institutions indicated those who rise to or are selected for these positions are highly filtered for ability…Females were underrepresented among all groups, but to a lesser degree among federal judges and Democrats and to a larger degree among Republicans and CEOs. America’s elite are largely drawn from the intellectually gifted, with many in the top 1% of ability…Murray (2008) was correct that a large portion of America’s elite are drawn from the intellectually gifted. This held for every group except the House of Representatives, which had a lower percentage having attended an Elite School. If the definition of elite is broadened to include either attendance at an Elite School or Graduate School then the majority met these criteria and are likely in the top percentiles of ability. This would include 56.6% of the billionaires, 67.0% of the CEOs, 68.1% of the House, 83.0% of the Senate, and all of the judges. All the federal judges and Senators and nearly all the other groups attended college…This study used average SAT or ACT scores of a college or university (America’s Best Colleges, 2013) as an approximation for ability level (Frey & Detterman, 2004; Koenig et al 2008), which may not hold for each individual case. It would have been optimal to have access to individual test scores, but unfortunately this data was not publicly available. However, using average SAT and ACT scores as an approximation for ability level may give an underestimate because extremely smart people may not have chosen to attend a top school for multiple reasons (eg. financial, scholarship, staying close to home). Alternatively, using this method may also give an overestimate because there are many legacies and athletic admits to elite institutions who do not usually meet the typical test score criteria (Espenshade & Radford, 2009). …This study demonstrates that in America, Democrats were more likely than Republicans to have a higher percentage of Senate and House members who attended an Elite School which places these individuals in the top 1% in ability (see Fig. 1 panel B and Appendix A). Therefore, among the elected elite, Democrats had a higher ability and education level, on average, than Republicans…This also shows that Bill Gates and Mark Zuckerberg (included in the Technology sector), who are often used as prominent examples in the media as to why going to college is not necessary for success (eg. Lin, 2010: “Top 10 college dropouts”; Williams, 2012: “Saying no to college”), are actually exceptions to the rule. Within the billionaire sample, 37 (8.7%) were clearly marked as a college drop out by the Forbes staff who compiled the data. The majority of the billionaires (88%) went to college and graduated. …Even within a group in the top 0.0000001% of wealth and a group of CEOs who were compensated quite highly (well within the top 1% of wealth), there were differences in the education and ability level between those who earned more money compared to those who earned less. The analyses in Table 3a and b demonstrate that even within billionaires and CEOs, higher education and ability level is related to higher net worth and compensation. Prior research demonstrated that even within a group in the top 1% in ability, higher ability is associated with higher income (Wai et al 2005). The analyses in Table 3c demonstrated that even within the top 1% of ability, higher ability is associated with higher net worth and compensation. Therefore, this study adds to, expands, and strengthens the literature linking education, ability, and wealth (Murray, 1998; Nyborg & Jensen, 2001; Zax & Rees, 2002), and provides further evidence that does not support an ability threshold hypothesis (Kuncel & Hezlett, 2010; Park et al 2007; ) - or the idea that more ability does not matter beyond a certain point in predicting real world outcomes.

  • Kuncel, N. R., & Hezlett, S. A. (2010). “Fact and fiction in cognitive ability testing for admissions and hiring decisions”. Current Directions in Psychological Science, 19, 339-345

  • Murray, C. (1998). Income inequality and IQ. Washington, D.C.: AEI Press.

  • Nyborg, H., & Jensen, A. R. (2001). “Occupation and income related to psychometric g”. Intelligence, 29, 45-55

  • Park, G., Lubinski, D., & Benbow, C. P. (2007). “Contrasting intellectual patterns predict creativity in the arts and sciences”. Psychological Science, 18, 948-952

  • Wai, J., Lubinski, D., & Benbow, C. P. (2005). “Creativity and occupational accomplishments among intellectually precocious youths: An age 13 to age 33 longitudinal study”. Journal of Educational Psychology, 97, 484-492

  • Zax, J. S., & Rees, D. L. (2002). “IQ, academic performance, environment, and earnings”. The Review of Economics and Statistics, 84, 600-614.

This might seem obvious (“elite schools produce the elites”) but is worth verifying. It’s also interesting that being elected doesn’t substantially affect the educational credentials & inferred IQ (even Representatives are still 20x more likely to be from an elite school), since given their behavior/policies/statements one might assume elected politicians are mediocrities or otherwise not especially intelligent. I can give a personal example: I grew up in New York, where one of the Senators has been for as long as I can remember, Charles “Chuck” Schumer, who never struck me as very subtle or intelligent, an impression furthered when he made his ill-fated public call years ago for the - still operating - Silk Road to be shut down; so I was somewhat shocked to learn via Steve Sailer that he claimed a perfect 1600 on the pre-centering SAT (and then went to Harvard where he was Phi Beta Kappa) which implies that he was more intelligent than myself or most of LessWrong, and combined with his flawless election record, further suggests that I have badly misunderstood him and he is actually a brilliant political mastermind.

“Polygenic Risk for Schizophrenia Is Associated with Cognitive Change Between Childhood and Old Age”, McIntosh et al 2013:

Background: Genome-wide association studies (GWAS) have shown a polygenic component to the risk of schizophrenia. The disorder is associated with impairments in general cognitive ability that also have a substantial genetic contribution. No study has determined whether cognitive impairments can be attributed to schizophrenia’s polygenic architecture using data from GWAS.

Methods: Members of the Lothian Birth Cohort 1936 (LBC1936, n = 14,937) were assessed using the Moray House Test at age 11 and with the Moray House Test and a further cognitive battery at age 70. To create polygenic risk scores for schizophrenia, we obtained data from the latest GWAS of the Psychiatric GWAS Consortium on Schizophrenia. Schizophrenia polygenic risk profile scores were calculated using information from the Psychiatric GWAS Consortium on Schizophrenia GWAS.

Results: In LBC1936, polygenic risk for schizophrenia was negatively associated with IQ at age 70 but not at age 11. Greater polygenic risk for schizophrenia was associated with more relative decline in IQ between these ages. These findings were maintained when the results of LBC1936 were combined with that of the independent Lothian Birth Cohort 1921 (n 14,517) in a meta-analysis.

Conclusions: Increased polygenic risk of schizophrenia is associated with lower cognitive ability at age 70 and greater relative decline in general cognitive ability between the ages of 11 and 70. Common genetic variants may underlie both cognitive aging and risk of schizophrenia.

fat/obesity; “Childhood Intelligence and Adult Obesity”, Kanazawa2013

Design and Methods: A population (n=17,419) of British babies has been followed since birth in 1958 in a prospectively longitudinal study. Childhood general intelligence is measured at 7, 11, and 16, and adult BMI and obesity are measured at 51. Results: Childhood general intelligence has a direct effect on adult BMI, obesity, and weight gain, net of education, earnings, mother’s BMI, father’s BMI, childhood social class, and sex. More intelligent children grow up to eat more healthy foods and exercise more frequently as adults. Conclusion: Childhood intelligence has a direct effect on adult obesity unmediated by education or earnings. General intelligence decreases BMI only in adulthood when individuals have complete control over what they eat.

…Obesity is just one of a large number of health problems that afflict less intelligent individuals, increase their mortality, and decrease their life expectancy (7-9)…Thus, relative to their less intelligent counterparts, more intelligent children are more likely to grow up to espouse the evolutionarily novel values of left-wing liberalism or atheism (13); to be nocturnal (15); to consume the evolutionarily novel substances of alcohol, tobacco, and psychoactive drugs (16); to prefer evolutionarily novel instrumental music such as classical and light music (17); and regardless of their genetic and hormonal predisposition, to engage in evolutionarily novel homosexual behavior (18)…The available evidence suggests that more intelligent individuals are more likely to exercise more frequently than less intelligent individuals (21,22)…Consistent with the prediction of the Hypothesis, Teasdale et al (23) report that, among a sample of 26,274 young Danish men, intelligence and body-mass index (BMI) are significantly negatively correlated even net of education.

  • Batty GD, Deary IJ, Gottfredson LS. Premorbid (early life) IQ and later mortality risk: systematic review. Ann Epidemiol2007;17:278-288.

  • Gottfredson LS, Deary IJ. “Intelligence predicts health and longevity, but why?” Curr Direct Psychol Sci 2004;13:1-4

  • Kanazawa S. Mind the gap…in intelligence: reexamining the relationship between inequality and health. Br J Health Psychol2006;11:623-642

  • Kanazawa S. Why liberals and atheists are more intelligent. Soc Psychol Q 2010; 73:33-57

  • Kanazawa S, Perina K. Why night owls are more intelligent. Pers Individ Differences 2009;47:685-690

  • Kanazawa S, Hellberg JEEU. Intelligence and substance use. Rev Gen Psychol2010;14:382-396

  • Kanazawa S, Perina K. Why more intelligent individuals like classical music. J Behavior Decis Making 2012;25:264-275.

  • Kanazawa S. Intelligence and homosexuality. J Biosoc Sci 2012;44:595-623.

  • Hall PA, Elias LJ, Fong GT, Harrison AH, Borowsky R, Sarty GE. A social neuro-

  • science perspective on physical activity. J Sport Exercise Psychol2008;30:432-449.

  • Kingma EM, Tak LM, Huisman M, Rosmalen JGM. Intelligence is negatively associated with the number of functional somatic symptoms. J Epidemiol Commun Health2009;63:900-906.

  • Teasdale TW, Sørensen TIA, Stunkard AJ. Intelligence and educational level in relation to body mass index of adult males. Hum Biol1992;64:99-106.

“Rare Copy Number Deletions Predict Individual Variation in Intelligence”, Yeo et al 2011

Phenotypic variation in human intellectual functioning shows substantial heritability, as demonstrated by a long history of behavior genetic studies. Many recent molecular genetic studies have attempted to uncover specific genetic variations responsible for this heritability, but identified effects capture little variance and have proven difficult to replicate. The present study, motivated an interest in “mutation load” emerging from evolutionary perspectives, examined the importance of the number of rare (or infrequent) copy number variations (CNVs), and the total number of base pairs included in such deletions, for psychometric intelligence. Genetic data was collected using the Illumina 1MDuoBeadChip Array from a sample of 202 adult individuals with alcohol dependence, and a subset of these (n = 77) had been administered the Wechsler Abbreviated Scale of Intelligence (WASI). After removing CNV outliers, the impact of rare genetic deletions on psychometric intelligence was investigated in 74 individuals. The total length of the rare deletions significantly and negatively predicted intelligence (r = −.30, p = .01). As prior studies have indicated greater heritability in individuals with relatively higher parental socioeconomic status (SES), we also examined the impact of ethnicity (Anglo/White vs. Other), as a proxy measure of SES; these groups did not differ on any genetic variable. This categorical variable significantly moderated the effect of length of deletions on intelligence, with larger effects being noted in the Anglo/White group. Overall, these results suggest that rare deletions (between 5% and 1% population frequency or less) adversely affect intellectual functioning, and that pleiotropic effects might partly account for the association of intelligence with health and mental health status. Significant limitations of this research, including issues of generalizability and CNV measurement, are discussed.

“Efficiency of functional brain networks and intellectual performance”, van den Heuvel2009

Our brain is a complex network in which information is continuously processed and transported between spatially distributed but functionally linked regions. Recent studies have shown that the functional connections of the brain network are organized in a highly efficient small-world manner, indicating a high level of local neighborhood clustering, together with the existence of more long-distance connections that ensure a high level of global communication efficiency within the overall network. Such an efficient network architecture of our functional brain raises the question of a possible association between how efficiently the regions of our brain are functionally connected and our level of intelligence. Examining the overall organization of the brain network using graph analysis, we show a strong negative association between the normalized characteristic path length lambda of the resting-state brain network and intelligence quotient (IQ). This suggests that human intellectual performance is likely to be related to how efficiently our brain integrates information between multiple brain regions. Most pronounced effects between normalized path length and IQ were found in frontal and parietal regions. Our findings indicate a strong positive association between the global efficiency of functional brain networks and intellectual performance.

“Genetics of Brain Fiber Architecture and Intellectual Performance”, Chiang et al 2009

The study is the first to analyze genetic and environmental factors that affect brain fiber architecture and its genetic linkage with cognitive function. We assessed white matter integrity voxel-wise using diffusion tensor imaging at high magnetic field (4 Tesla), in 92 identical and fraternal twins. White matter integrity, quantified using fractional anisotropy (FA), was used to fit structural equation models (SEM) at each point in the brain, generating three-dimensional maps of heritability. We visualized the anatomical profile of correlations between white matter integrity and full-scale, verbal, and performance intelligence quotients (FIQ, VIQ, and PIQ). White matter integrity (FA) was under strong genetic control and was highly heritable in bilateral frontal (a2 = 0.55, p = 0.04, left; a2 = 0.74, p = 0.006, right), bilateral parietal (a2 = 0.85, p < 0.001, left; a2 = 0.84, p < 0.001, right), and left occipital (a2 = 0.76, p = 0.003) lobes, and was correlated with FIQ and PIQ in the cingulum, optic radiations, superior fronto-occipital fasciculus, internal capsule, callosal isthmus, and the corona radiata (p = 0.04 for FIQ and p = 0.01 for PIQ, corrected for multiple comparisons). In a cross-trait mapping approach, common genetic factors mediated the correlation between IQ and white matter integrity, suggesting a common physiological mechanism for both, and common genetic determination. These genetic brain maps reveal heritable aspects of white matter integrity and should expedite the discovery of single-nucleotide polymorphisms affecting fiber connectivity and cognition.

“Smarter people are (a bit) more symmetrical: A meta-analysis of the relationship between intelligence and fluctuating asymmetry”, Banks et al 2010

Individual differences in general mental ability (g) have important implications across multiple disciplines. Research suggests that the variance in g may be due to a general fitness factor. If this is the case, a relationship should exist between g and other reliable indicators of fitness. Some empirical results indicate a relationship between g and fluctuating asymmetry. However, there have been inconsistencies in the results, some of which may be due to random sampling error, and some of which may be due to moderating variables, publication bias, and methodological issues. To help clarify the literature, a meta-analysis was conducted on the relationship between g and fluctuating asymmetry. Based on 14 samples across 1871 people, estimates of the population correlation ranged from −.12 to −.20. There was a difference in the magnitude of the correlation between published studies and unpublished studies with published studies showing larger magnitude negative correlations and unpublished studies yielding results closer to zero. The implications for our understanding of g and its relationship with fluctuating asymmetry are discussed.

“The Inheritance of Inequality”, Bowles & Gintis2002

Correlations of IQ between parents and offspring range 0.42–0.72, where the higher figure refers to measures of average parental and average offspring IQ (Bouchard and McGue, 1981; Plomin et al 2000).

We have located 65 estimates of the normalized regression coefficient of a test score in an earnings equation in 24 different studies of U.S. data over a period of three decades. Our meta-analysis of these studies is presented in Bowles, Gintis & Osborne2002. The mean of these estimates is 0.15, indicating that a standard deviation change in the cognitive score, holding constant the remaining variables (including schooling), changes the natural logarithm of earnings by about one-seventh of a standard deviation. By contrast, the mean value of the normalized regression coefficient of years of schooling in the same equation predicting the natural log of earnings in these studies is 0.22, suggesting a somewhat larger independent effect of schooling. We checked to see if these results were dependent on the weight of overrepresented authors, the type of cognitive test used, at what age the test was taken and other differences among the studies and found no significant effects. An estimate of the causal impact of childhood IQ on years of schooling (also normalized) is 0.53 (Winship and Korenman, 1999). A rough estimate of the direct and indirect effect of IQ on earnings, call it b, is then b ϭ 0.15 ϩ (0.53)(0.22) ϭ 0.266.

  • Bowles, Gintis & Osborne2002 “The Determinants of Individual Earnings: Skills, Preferences, and Schooling.” Journal of Economic Literature. December, 39:4, pp. 1137-176

  • Winship and Korenman, 1999 “Economic Success and the Evolution of Schooling with Mental Ability”, in Earning and Learning: How Schools Matter. Susan Mayer and Paul Peterson, eds. Washington, D.C.: Brookings Institution, pp. 49-78

A meta-analysis of 63 studies showed a significant negative association between intelligence and religiosity. The association was stronger for college students and the general population than for participants younger than college age; it was also stronger for religious beliefs than religious behavior. For college students and the general population, means of weighted and unweighted correlations between intelligence and the strength of religious beliefs ranged from −.20 to −.25 (mean r = −.24). Three possible interpretations were discussed. First, intelligent people are less likely to conform and, thus, are more likely to resist religious dogma. Second, intelligent people tend to adopt an analytic (as opposed to intuitive) thinking style, which has been shown to undermine religious beliefs. Third, several functions of religiosity, including compensatory control, self-regulation, self-enhancement, and secure attachment, are also conferred by intelligence. Intelligent people may therefore have less need for religious beliefs and practices.

“The Relation Between Intelligence and Religiosity: A Meta-Analysis and Some Proposed Explanations” Zuckerman et al 2013:

A meta-analysis of 63 studies showed a significant negative association between intelligence and religiosity. The association was stronger for college students and the general population than for participants younger than college age; it was also stronger for religious beliefs than religious behavior. For college students and the general population, means of weighted and unweighted correlations between intelligence and the strength of religious beliefs ranged from −.20 to −.25 (mean r = −.24). Three possible interpretations were discussed. First, intelligent people are less likely to conform and, thus, are more likely to resist religious dogma. Second, intelligent people tend to adopt an analytic (as opposed to intuitive) thinking style, which has been shown to undermine religious beliefs. Third, several functions of religiosity, including compensatory control, self-regulation, self-enhancement, and secure attachment, are also conferred by intelligence. Intelligent people may therefore have less need for religious beliefs and practices.

…To our knowledge, the first studies on intelligence and religiosity appeared in 1928, in the University of Iowa Studies series, Studies in Character (Howells, 1928; Sinclair, 1928). These studies examined sensory, motor, and cognitive correlates of religiosity. Intelligence tests were included in the battery of administered tasks. Both Howells (1928) and Sinclair (1928) found that higher levels of intelligence were related to lower levels of religiosity. Accumulation of additional research during the subsequent three decades prompted Argyle (1958) to review the available evidence. He concluded that “intelligent students are much less likely to accept orthodox beliefs, and rather less likely to have pro-religious attitudes” (p. 96). Argyle also noted that, as of 1958, all available evidence was based on children or college student samples. He speculated, however, that the same results might be observed for adults of post-college age. In the subsequent decade, the pendulum swung in the opposite direction. Kosa & Schommer1961 and Hoge (1969) drew conclusions from their data that were inconsistent with those of Argyle (1958). According to Kosa and Schommer, “social environment regulates the relationship of mental abilities and religious attitudes by channeling the intelligence into certain approved directions: a secular-oriented environment may direct it toward skepticism, a church-oriented environment may direct it toward increased religious interest” (p. 90). They found that in a Catholic college, more intelligent students knew more about religious doctrine and participated more in strictly religious organizations.

…Hoge (1969, 1974) tracked changes in religious attitudes on 13 American campuses. He compared survey data, most of which were collected between 1930 and 1948, with data that he collected himself in 1967 and 1968. On four campuses, Hoge also examined the relation between SAT scores and religious attitudes. Correlations were small and mostly negative. Hoge (1969) concluded that “no organic or psychic relationship exists between intelligence and religious attitudes and . . . the relationships found by researchers are either due to educational influences or biases in the intelligence tests” (p. 215). Hoge acknowledged that range restrictions of college students’ intelligence scores may decrease correlations between intelligence and other variables. Nevertheless, he concluded that the low negative-intelligence-religiosity correlations implied that there is no relation between intelligence and religiosity.

…As if in response to Beit-Hallahmi and Argyle’s (1997) call, the last decade has seen a number of large-scale studies that examined the relation between intelligence and religiosity (Kanazawa, 2010a; Lewis, Ritchie, & Bates, 2011; Nyborg, 2009; Sherkat, 2010). Kanazawa (2010a), Sherkat (2010), and Lewis et al 2011 all found negative relations between intelligence and religiosity in post-college adults. Nyborg (2009) found that young atheists (age 12 to 17) scored significantly higher on an intelligence test than religious youth. The last decade also saw studies on the relation between intelligence and religiosity at the group level. Using data from 137 nations, Lynn, Harvey, and Nyborg (2009) found a negative relation between mean intelligence scores (computed for each nation) and mean religiosity scores. However, IQ scores from undeveloped and/or non-Westernized countries might have limited validity because most tests were developed for Western cultures. Low levels of literacy and problems in obtaining representative samples in some countries may also undermine the validity of these findings (Hunt, 2011; Richards, 2002; Volken, 2003). In response to these critiques, Reeve (2009) repeated the analysis but set all national IQ scores lower than 90 to 90. The resulting IQ-religiosity correlation was not lower than what had been reported in prior studies (see Reeve, 2009, for a discussion of his truncating procedure). In the same vein, Pesta, McDaniel, and Bertsch (2010) found a negative relation between intelligence and religiosity scores that were computed for all 50 states in the United States. These results are less susceptible to the problems (eg. cultural differences) that plagued studies at the country level.

…Studies in this area have found that, relative to the general public, scientists are less likely to believe in God. For example, Leuba (1916) reported that 58% of randomly selected scientists in the United States expressed disbelief in, or doubt regarding the existence of God; this proportion rose to nearly 70% for the most eminent scientists. Larson & Witham1998 reported similar results, as evidenced by the title of their article-“Leading scientists still reject God.” Of course, higher intelligence is only one of a number of factors that can account for these results.

…Outside of academic journals, however, there have been at least two reviews (Beckwith, 1986; Bell, 2002). Beckwith (1986) concluded that 39 of the 43 studies that he summarized supported a negative relation between intelligence and religiosity, and Bell (2002) simply repeated this tally. However, some of the studies reviewed by Beckwith were only indirectly relevant (eg. comparisons between more and less prestigious universities), and some relevant studies were excluded.

…The first row of Table 2 presents basic statistics describing the relation between intelligence and religiosity for all 63 studies. Results are presented for random-effects analyses (unweighted mean correlations) and fixed-effects analyses (weighted mean correlations). Fifty-three studies showed negative correlations while 10 studies showed positive correlations. Thirty-seven studies showed significant correlations; of these, 35 were negative and 2 were positive. The unweighted mean correlation (r) between intelligence and religiosity was −.16, the median r was −.14, and the weighted mean r was −.13. The similarity of these three indicators of central tendency indicates that the distribution was approximately symmetrical and was not skewed by several very large studies that were in the database. Random- and fixed-effects models yielded significant evidence that the higher a person’s intelligence, the lower the person scored on the religiosity measures.

…When GPA[grade-point-average]-religiosity correlations from the five studies using only GPA are combined with GPA-religiosity correlations from the four studies using GPA as well as other intelligence measures, the mean GPA-religiosity correlation was not significantly different from zero, MGPA = −.027, p = .33. It was concluded that GPA had no meaningful relation to religiosity and, accordingly, all subsequent analyses omitted the five studies that used only GPA.

…As expected, the extreme groups effect size (M = −.43) that was significantly more negative than that of the unbiased studies (p < .001 by post hoc least significant difference [LSD] test).

…The fixed-effects trim and fill method for detecting possible publication bias yielded negligible impact for the pre-college and non-college groups. For the college group, however, there was evidence of publication bias, such that nine negative effect sizes would need to be added to yield a symmetrical distribution. The imputation of these effects resulted in an adjusted mean effect of −.21, noticeably quite different from the observed weighted mean effect of −.15. Because the adjusted effect size is hypothetical, it will not be incorporated into subsequent analyses. However, this result and the range restriction in intelligence scores in this group suggest that the true intelligence-religiosity relation in the college population may be more negative than the literature indicates.

…As an exploratory analysis, we examined the relation between percentage of males in each study and effect size of the intelligence-religiosity relation. In the 34 studies in which it could be determined, percentage of males was positively correlated with unweighted effect sizes, r(32) = .50, p < .01. This correlation indicates that the negative intelligence-religiosity relation was less negative in studies with more males. This relation held in terms of magnitude for the pre-college and college groups, r(6) = .48, ns, and r(12) = .51, p = .06, but was weaker at the non-college level, r(10) = .19, ns. When analyzed as a fixed-effects regression, the relation between percentage of males and effect size was also markedly positive, p < .001. A more direct test of the possibility that the intelligence-religiosity relation is less negative for males is a within-study comparison between males and females. Kanazawa1 conducted this test for two studies (Kanazawa, 2010a; combined N = 21,437). If anything, the results pointed in the opposite direction. The intelligence-religiosity correlations for females and males, respectively, were −.11 and −.12 in Study 1, and −.14 and −.16 in Study 2. Although the difference between females and males was not significant, even when combined meta-analytically across studies (Z = 1.39, p = .16), the direction of this difference is inconsistent with the between-studies finding of the meta-analysis.

…As previously noted, some investigators suggested that education mediates the relation between intelligence and religiosity (Hoge, 1974; Reeve & Basalik, 2011). Interestingly, Kanazawa (S. Kanazawa, personal communication, January 2012) espouses an opposing view, namely that intelligence accounts for any negative relation between education and religiosity. Table 8 presents results that address the two competing hypotheses. The analyses are based on seven studies from three sources. Results from the student sample studied by Blanchard-Fields, Hertzog, Stein, and Pak (2001; first row in Table 8) can be excluded because of range restriction for intelligence and education (indeed, all correlations for that study were weak). The results of the remaining six studies indicate that education does not mediate the intelligence-religiosity relation. To begin with, intelligence was more negatively related to religiosity than was education (unweighted mean correlations were −.18 and −.06, respectively). We tested the significance of this difference separately for each study, using a procedure for comparing nonindependent correlations (Meng, Rosenthal, & Rubin, 1992); the combined difference across the six studies was highly significant, Z = 9.32, p < .001. Furthermore, controlling for education did not have much of an effect on the intelligence-religiosity relation-unweighted means of the six zero-order and partial correlations were −.18 and −.17, respectively. In contrast, controlling for intelligence led to a somewhat greater change in the education-religiosity relation; the unweighted means for the six zero-order and partial correlations were −.06 and .00, respectively. This finding is consistent with S. Kanazawa’s (personal communication, January 2010) view that intelligence accounts for the education-religiosity relation. However, given that the analysis is based on only six studies, our conclusions are tentative.

…Table 9 presents the findings. In all four comparisons, Terman’s sample scored significantly lower on religiosity than the general public (the average of these effects was used in the meta-analysis as one of the extreme groups’ studies). Admittedly, the years of data collection and ages of the two groups do not match perfectly. However, the results are so strong that it is difficult to imagine that more exact matching would make a difference. These results are even more striking if the Termites’ religious upbringing is considered. Terman & Oden1959 reported that close to 60% of Termites reported that they received “very strict” or “considerable” religious training; approximately 33% reported receiving little training, and about 6% reported no religious training. This suggests that the Termites underwent changes in their religiosity after their childhood. …In Terman’s sample (n = 410), 1.2% checked the religious option, compared with .4% in the Hunter group (n = 139), Z < 1. These results suggest that on an absolute level, religion was relatively unimportant to middle-aged adults who were identified as gifted in childhood in both samples. In addition, we speculate that if the Hunter sample is similar to the Terman sample with respect to religiosity, it too may be less religious than the general population. In the Terman and the Hunter samples, a high intelligence level at an early age preceded lower religiosity many years later. However, our analyses of these results neither controlled for possible relevant factors at an early age (eg. socioeconomic status) nor examined possible mediators (eg. occupation) of this relation.

…Intelligence can be reliably measured at a very early age while religiosity cannot (eg. Jensen1998; Larsen, Hartmann, & Nyborg, 2008). In their classic study, for example, H. E. Jones & Bayley1941 showed that the mean of intelligence scores assessed at ages 17 and 18 (a) correlated .86 with the mean scores assessed at ages 5, 6, and 7; and (b) correlated .96 with the mean of intelligence scores assessed at ages 11, 12, and 13. Because intelligence can be measured at an early age, it can be used to predict outcomes observed years later. For example, Deary, Strand, Smith, and Fernandes (2007) reported a .69 correlation between intelligence measured at age 11 and educational achievement at age 16. Unlike intelligence, religiosity assessed at an early age is a weak predictor of religiosity assessed years later. For example, Willits & Crider1989 found only small to moderate correlations between religiosity at age 16 and that at 27 (.28 for church attendance and .36 for beliefs). O’Connor, Hoge, and Alexander (2002) found no relationship between measures of church involvement at ages 16 and 38.

…First, although the prevalence of religiosity varies widely among countries and cultures, more than 50% of the world population consider themselves religious. Using survey data collected by P. Zuckerman (2007) from 137 countries, Lynn et al 2009 and Reeve (2009) observed a prevalence of 89.9% believers in the world and 89.5% believers in the United States. However, a recent Win-Gallup International (2012) poll of 59,927 persons in 57 countries found that only 59% of the respondents (60% in the United States) consider themselves religious, a decline of 9% (13% in the United States) from a similar 2005 poll. Atheism might be considered a case of nonconformity in societies where the majority is religious. This is not so, however, if one grows up in largely atheist societies, such as those that exist in Scandinavia (P. Zuckerman, 2008).

…There is also empirical evidence suggesting that religiosity may be an in-group phenomenon, reinforcing prosocial tendencies within the group (see a review by Norenzayan & Gervais, 2012), but also predisposing believers to reject out-groups members (see meta-analysis by D. L. Hall, Matz, & Wood, 2010). To become an atheist, therefore, it may be necessary to resist the in-group dogma of religious beliefs. Not surprisingly, there is evidence of anti-atheist distrust and prejudice (Gervais, Shariff, & Norenzayan, 2011; Gervais & Norenzayan, 2012b; for a review, see Norenzayan & Gervais, 2012).

…Intelligence also confers a sense of personal control. We identified eight studies that reported correlations between intelligence and belief in personal control (Grover & Hertzog, 1991; Lachman, 1983; Lachman, Baltes, Nesselroade, & Willis, 1982; Martel, McKelvie, & Standing, 1987; Miller & Lachman, 2000; Prenda & Lachman, 2001; Tolor & Reznikoff, 1967; P. Wood & Englert, 2009). All eight correlations were positive, with a mean correlation (weighted by df of each study) of .29. In addition, higher intelligence is associated with greater self-efficacy-the belief in one’s own ability to achieve valued goals (Bandura, 1997). This construct is similar to personal control beliefs but has been examined separately in the literature. In a meta-analysis of 26 studies, the mean correlation between intelligence and self-efficacy was .20 (Judge, Jackson, Shaw, Scott, & Rich, 2007).

…Choosing the large delayed reward serves as an indicator of self-control. Shamosh & Gray2008 meta-analyzed the relation between intelligence and delay discounting (the latter construct is identical to delay of gratification except that high delay discounting indicates poor self-control). Their analysis, based on 26 studies, yielded a mean r of −.23. This suggests that intelligent people are more likely to delay gratification (ie. less likely to engage in delay discounting).

…On the other hand and in line with Kanazawa’s (2010a) model, genetic influences have been implicated not only in intelligence (cf., Nisbett et al 2012b), but also in religiosity (D’Onofrio, Eaves, Murrelle, Maes, & Spilka, 1999; Koenig, McGue, & Iacono, 2008). Furthermore, the model was used to predict other correlates of intelligence (eg. political liberalism and, for men, monogamy), and those predictions received empirical support. In conclusion, Kanazawa’s (2010a) interpretation remains an intriguing possibility.

…This function was not included in our discussion of functional equivalence because, to the best of our knowledge, there is no evidence pertaining to the relation between intelligence and death anxiety. Although this logic suggests that the negative relation between intelligence and religiosity might decline at the end of life, the relevant evidence we have indicates otherwise. The highly intelligent members of Terman’s sample retained lower religiosity scores (relative to the general population) even at 75 to 91 years of age (Table 9). Additional research is needed to resolve this issue.

“Intelligence (IQ) as a Predictor of Life Success”, Firkowska-Mankiewicz2002

“Leading scientists still reject God”, Larson & Witham1998

Research on this topic began with the eminent US psychologist James H. Leuba and his landmark survey of 1914. He found that 58% of 1,000 randomly selected US scientists expressed disbelief or doubt in the existence of God, and that this figure rose to near 70% among the 400 “greater” scientists within his sample^1. Leuba repeated his survey in somewhat different form 20 years later, and found that these percentages had increased to 67 and 85, respectively^2.

In 1996, we repeated Leuba’s 1914 survey and reported our results in Nature^3. We found little change from 1914 for American scientists generally, with 60.7% expressing disbelief or doubt. This year, we closely imitated the second phase of Leuba’s1914 survey to gauge belief among “greater” scientists, and find the rate of belief lower than ever - a mere 7% of respondents.

…Our chosen group of “greater” scientists were members of the National Academy of Sciences (NAS). Our survey found near universal rejection of the transcendent by NAS natural scientists. Disbelief in God and immortality among NAS biological scientists was 65.2% and 69.0%, respectively, and among NAS physical scientists it was 79.0% and 76.3%. Most of the rest were agnostics on both issues, with few believers. We found the highest percentage of belief among NAS mathematicians (14.3% in God, 15.0% in immortality). Biological scientists had the lowest rate of belief (5.5% in God, 7.1% in immortality), with physicists and astronomers slightly higher (7.5% in God, 7.5% in immortality). Overall comparison figures for the 1914, 1933 and 1998 surveys appear in Table 1.

The significance of variations in intelligence has also been examined among individuals with alcohol dependence, as lower intelligence as assessed in childhood or in early adulthood predicts greater comorbidity [14], a greater propensity for hangovers [15], greater mortality from alcohol-related health problems [16], and poor treatment outcomes [17].

“Early-Life Intelligence Predicts Midlife Biological Age”, Schaefer et al 2015:

Objectives. Early-life intelligence has been shown to predict multiple causes of death in populations around the world. This finding suggests that intelligence might influence mortality through its effects on a general process of physiological deterioration (ie. individual variation in “biological age”). We examined whether intelligence could predict measures of aging at midlife before the onset of most age-related disease.

Methods. We tested whether intelligence assessed in early childhood, middle childhood, and midlife predicted midlife biological age in members of the Dunedin Study, a population- representative birth cohort.

Results. Lower intelligence predicted more advanced biological age at midlife as captured by perceived facial age, a 10-biomarker algorithm based on data from the National Health and Nutrition Examination Survey (NHANES), and Framingham heart age (r = 0.1-0.2). Correlations between intelligence and telomere length were less consistent. The associations between intelligence and biological age were not explained by differences in childhood health or parental socioeconomic status, and intelligence remained a significant predictor of biological age even when intelligence was assessed before Study members began their formal schooling.

Discussion. These results suggest that accelerated aging may serve as one of the factors linking low early-life intelligence to increased rates of morbidity and mortality.

“Intelligence Predicts Health and Longevity, but Why?”, Gottfredson & Deary2004:

Large epidemiological studies of almost an entire population in Scotland have found that intelligence (as measured by an IQ-type test) in childhood predicts substantial differences in adult morbidity and mortality, including deaths from cancers and cardiovascular diseases. These relations remain significant after controlling for socioeconomic variables. One possible, partial explanation of these results is that intelligence enhances individuals’ care of their own health because it represents learning, reasoning, and problem-solving skills useful in preventing chronic disease and accidental injury and in adhering to complex treatment regimens.

O’Toole & Stankov1992 used IQ at induction into the military, along with 56 other psychological, behavioral, health, and demographic variables, to predict noncombat deaths by age 40 among 2,309 Australian veterans. When all other variables were statistically controlled, each additional IQ point predicted a 1% decrease in risk of death. Also, IQ was the best predictor of the major cause of death, motor vehicle accidents. Vehicular death rates doubled and then tripled at successively lower IQ ranges (100-115, 85-100, 80-85; O’Toole, 1990).

…To date, Scotland is the only country to have conducted IQ testing on almost a whole year-of-birth cohort. This took place in the remarkable Scottish Mental Survey of 1932 (SMS1932). Using these procedures, the researchers traced 2,230 (79.9%) of those children who took the MHT in Aberdeen: 1,084 were dead, 1,101 were alive, and 45 had moved away from Scotland. In addition, 562 were untraced… IQ at age 11 had a significant association with survival to about age 76. On average, individuals who were at a 1standard-deviation (15-point) disadvantage in IQ relative to other participants were only 79% as likely to live to age 76. The effect of IQ was stronger for women (71%) than for men (83%), partly because men who died in active service during World War II had relatively high mean IQ scores. Further analyses of the Aberdeen subjects found that a drop of 1 standard deviation in IQ was associated with a 27% increase in cancer deaths among men and a 40% increase in cancer deaths among women (Deary, Whalley, & Starr, 2003). The effect was especially pronounced for stomach and lung cancers, which are specifically associated with low socioeconomic status (SES) in childhood.

…Higher intelligence might lower mortality from all causes and from specific causes partly by affecting known risk factors for disease, such as smoking. In the combined SMS1932-Midspan database, there was no significant childhood IQ difference between participants who had ever smoked and those who had never smoked (Taylor, Hart, et al 2003). However, at the time of the Midspan studies, participants who were current smokers had significantly lower childhood IQs than ex-smokers. For each standard deviation increase in IQ, there was a 33% increased rate of quitting smoking. Adjusting for social class reduced this rate only mildly, to 25%. Thus, childhood IQ was not associated with starting smoking (mostly in the 1930s, when the public were not aware of health risks), but was associated with giving up smoking as health risks became evident.

…However, health inequalities tend to increase when health resources become more available to everyone (Gottfredson, in press). That is, increased availability of health resources improves health overall, but the improvements are smaller for people who are poorly educated and have low incomes than for people with more education and better incomes. Compared with people in high-SES groups, people with low SES seek more but not necessarily appropriate care when cost is no barrier; adhere less often to treatment regimens; learn and understand less about how to protect their health; seek less preventive care, even when it is free; and less often practice the healthy behaviors so important for preventing or slowing the progression of chronic diseases, the major killers and disablers in developed nations today.

Yet social class correlates with virtually every indicator of health, health behavior, and health knowledge. The link between SES and health transcends the particulars of material advantage, decade, nation, health system, social change, or disease, regardless of its treatability. Health scientists view the pervasiveness and finely graded nature of this relationship between SES and health as a paradox, leading them to speculate that SES creates health inequality via some yet-to-be-identified, highly generalizable “fundamental cause” (Gottfredson, in press). The socioeconomic measures that best predict health inequality also correlate most with intelligence (education best, then occupation, then income). This means that instead of IQ being a proxy for SES in health matters, SES measures might be operating primarily as rough proxies for social-class differences in mental rather than material resources.

…Health workers can diagnose and treat incubating problems, such as high blood pressure or diabetes, but only when people seek preventive screening and follow treatment regimens. Many do not. In fact, perhaps a third of all prescription medications are taken in a manner that jeopardizes the patient’s health. Non-adherence to prescribed treatment regimens doubles the risk of death among heart patients (Gallagher, Viscoli, & Horwitz, 1993). For better or worse, people are substantially their own primary health care providers.

For instance, one study (Williams et al 1995) found that, overall, 26% of the outpatients at two urban hospitals were unable to determine from an appointment slip when their next appointment was scheduled, and 42% did not understand directions for taking medicine on an empty stomach. The percentages specifically among outpatients with “inadequate” literacy were worse: 40% and 65%, respectively. In comparison, the percentages were 5% and 24% among outpatients with “adequate” literacy. In another study (Williams, Baker, Parker, & Nurss, 1998), many insulin-dependent diabetics did not understand fundamental facts for maintaining daily control of their disease: Among those classified as having inadequate literacy, about half did not know the signs of very low or very high blood sugar, and 60% did not know the corrective actions they needed to take if their blood sugar was too low or too high. Among diabetics, intelligence at time of diagnosis correlates significantly (.36) with diabetes knowledge measured 1 year later (Taylor, Frier, et al 2003).

Duarte, Crawford, Stern, Haidt, Jussim, and Tetlock:

[T]he observed relationship between intelligence and conservatism largely depends on how conservatism is operationalized. Social conservatism correlates with lower cognitive ability test scores, but economic conservatism correlates with higher scores (Iyer, Koleva, Graham, Ditto, & Haidt, 2012; Kemmelmeier 2008). Similarly, Feldman & Johnston 2014 find in multiple nationally representative samples that social conservatism negatively predicted educational attainment, whereas economic conservatism positively predicted educational attainment. Together, these results likely explain why both Heaven et al 2011 and Hodson & Busseri2012 found a negative correlation between IQ and conservatism–because “conservatism” was operationalized as Right-Wing Authoritarianism, which is more strongly related to social than economic conservatism (van Hiel et al 2004). In fact, Carl (2014) found that Republicans have higher mean verbal intelligence (up to 5.48 IQ points equivalent, when covariates are excluded), and this effect is driven by economic conservatism (which, as a European, he called economic liberalism, because of its emphasis on free markets). Carl suggests that libertarian Republicans overpower the negative correlation between social conservatism and verbal intelligence, to yield the aggregate mean advantage for Republicans. Moreover, the largest political effect in Kemmelmeier’s (2008) study was the positive correlation between anti-regulation views and SAT-V scores, where β = .117, p < .001 (by comparison, the regression coefficient for conservatism was β = −.088, p < .01, and for being African American, β = −.169, p < .001).


Many key determinants of well-being correlate highly with the results of IQ tests, and other measures of intelligence. Many specific life outcomes have been shown to correlate highly with intelligence (Herrnstein and Murray 1994; Kirsch et al 1993). While no causal link has been demonstrated between higher levels of cognition and happiness (Gow et al 2005), numerous studies have highlighted that increased cognition improves the likelihood of specific markers of wellbeing, while lower levels of general intelligence predisposes an individual to various forms of social disadvantage (Herrnstein and Murray1994; ).

Several prominent intelligence studies have demonstrated that higher general intelligence correlates with such life outcomes as increased income (Rowe et al 1998; Zagorsky 2007; Gottfredson 2003), improved quality of health and reduced mortality (Batty et al 2007; Whalley and Deary 2001; Gottfredson and Deary 2004) and overall increased life chances (Murray 2002; Herrnstein and Murray 1994; Kirsch et al 1993; Gottfredson2011). Intelligence appears to have a prominent effect over a broad range of social and economic life outcomes. Life is difficult for individuals with borderline intellectual disability; an IQ below 75. This group is at a high risk of failing elementary education, being unable to master simple daily tasks, being classified as unemployable, and are at an increased risk of being socially isolated (Edgerton1993; Koegel and Edgerton1984; Herrnstein and Murray1994; ). Individuals with borderline intellectual disability are at a great risk of living in poverty (30 %), having illegitimate children (32 %), being a chronic welfare dependent (31 %) and have very poor employment opportunities (Gottfredson1997; Herrnstein and ; ). While individuals within this group are capable of leading satisfying lives, they will most likely require significant social support in order to do so. Those in this group who do live independently have a tendency to live volatile and unpredictable lives due to the lack of stabilizing resources that come with increased intellectual competence (; ).

…Over half of these individuals fail to reach the minimum recruitment standards for the US military (Hunter and Schmidt 1996). Individuals with minor to moderate lower, normal intelligence are still at large risk of living in poverty (16 %), being a chronic welfare dependent (17 %) and are much more likely to drop out of school (35 %) compared to individuals with average intelligence (Herrnstein and Murray 1994; Kirsch et al 1993; Gottfredson2011). The odds of incarceration remain steady for all lower, normal intelligence groups (7 %) but reduce by more than half for average intelligence levels (3 %), indicating a particular susceptibility to incarceration at lower intelligence levels (Gottfredson1997; Herrnstein and Murray1994).

  • Batty, G.D., I.J. Deary, and L.S. Gottfredson. 2007. Pre-morbid (early life) IQ and later mortality risk: Systematic review. Annals of Epidemiology 17(4): 278-288.

  • Edgerton, R.B. 1993. The cloak of competence, revised and updated edition. Berkeley: University of California Press

  • Gottfredson, L.S. 1997. Why g matters: The complexity of everyday life. Intelligence 24: 79-132.

  • Gottfredson, L.S. 2003. g, jobs, and life. In The scientific study of general intelligence: Tribute to Arthur R. Jensen, ed. H. Nyborg. New York: Pergamon.

  • Gottfredson, L.S. 2011. Intelligence and social inequality: Why the biological link? In Handbook of individual differences, ed. T. Chamorro-Premuzic, A. Furhnam, and S. von strumm. New York: Wiley.

  • Gottfredson, L.S., and I.J. Deary. 2004. “Intelligence predicts health and longevity, but why?” Current Directions in Psychological Science 13: 1-4.

  • Gow, A.J., M.C. Whiteman, A. Pattie, L. Whalley, J. Starr, and I.J. Deary. 2005. Lifetime intellectual function and satisfaction with life in old age: Longitudinal cohort study. BMJ 331: 141-142.

  • Herrnstein, R.J., and C.A. Murray. 1994. The bell curve: Intelligence and class structure in American life. Salt Lake: Free Press.

  • Hunter, J.E., and F.L. Schmidt. 1996. Intelligence and job performance: Economic and social implications. Psychology, Public Policy, and Law 2: 447-472.

  • Kirsch, I.S., A. Jungeblut, L. Jenkins, and A. Kolstad. 1993. Adult literacy in America: A first look at the results of the national adult literacy survey. Princeton: Educational Testing Service.

  • Koegel, P., and R.B. Edgerton. 1984. “Black ‘six-hour retarded children’ as young adults”. Monographs of the American Association on Mental Deficiency 6: 145-171.

  • Murray, C. 2002. IQ and income inequality in a sample of sibling pairs from advantaged family backgrounds. The American Economic Review 92: 339-343.

  • Rowe, D.C., W.J. Vesterdal, and J.L. Rodgers. 1998. Herrnstein’s syllogism: genetic and shared environmental influences on IQ, education, and income. Intelligence 26: 405-423

  • Whalley, L.J., and I.J. Deary. 2001. Longitudinal cohort study of childhood IQ and survival up to age 76. BMJ 322: 819.

  • Zagorsky, J.L. 2007. Do you have to be smart to be rich? The impact of IQ on wealth, income and financial distress. Intelligence 35: 489-501.

“Clever Enough to Tell the Truth”, Ruffle & Tobol2017:

We conduct a field experiment on 427 Israeli soldiers who each rolled a six-sided die in private and reported the outcome. For every point reported, the soldier received an additional half-hour early release from the army base on Thursday afternoon. We find that the higher a soldier’s military entrance score, the more honest he is on average. We replicate this finding on a sample of 156 civilians paid in cash for their die reports. Furthermore, the civilian experiments reveal that two measures of cognitive ability predict honesty, whereas self-report honesty questions and a consistency check among them are of no value. We provide a rationale for the relationship between cognitive ability and honesty and discuss the generalizability of this result.

“Intelligence and socioeconomic success: A meta-analytic review of longitudinal research”, Strenze2007

The relationship between intelligence and socioeconomic success has been the source of numerous controversies. The present paper conducted a meta-analysis of the longitudinal studies that have investigated intelligence as a predictor of success (as measured by education, occupation, and income). In order to better evaluate the predictive power of intelligence, the paper also includes metaanalyses of parental socioeconomic status (SES) and academic performance (school grades) as predictors of success. The results demonstrate that intelligence is a powerful predictor of success but, on the whole, not an overwhelmingly better predictor than parental SES or grades. Moderator analyses showed that the relationship between intelligence and success is dependent on the age of the sample but there is little evidence of any historical trend in the relationship.

“Does intelligence foster generalized trust? An empirical test using the UK birth cohort studies”, Sturgis et al 2010

Social, or ‘generalized’ trust is often characterised as the ‘attitudinal dimension’ of social capital. It has been posited as key to a host of normatively desirable outcomes at the societal and individual levels. Yet the origins of individual variation in trust remain something of a mystery and continue to be a source of dissensus amongst researchers across and within academic disciplines. In this paper we use data from two British birth cohort studies to test the hypothesis that a propensity to express generalized trust varies systematically as a function of individual intelligence. Intelligence, we argue, fosters greater trust in one’s fellow citizens because more intelligent individuals are more accurate in their assessments of the trustworthiness of others. This means that, over the life-course, their trust is less often betrayed and they are able to accrue the benefits of norms of reciprocity. Our results show that standard measures of intelligence administered when cohort members were aged 10 and 11 can explain variability in expressed trust in early middle age, net of a broad range of theoretically related covariates.

“Associations between IQ and cigarette smoking among Swedish male twins”, Wennerstad et al 2010

It has been suggested that certain health behaviors, such as smoking, may operate as mediators of the well-established inverse association between IQ and mortality risk. Previous research may be afflicted by unadjusted confounding by socioeconomic or psychosocial factors. Twin designs offer a unique possibility to take genetic and shared environmental factors into account. The aim of the present national twin study was to determine the interrelations between IQ at age 18, childhood and attained social factors and smoking status in young adulthood and mid-life. We studied the association between IQ at age 18 and smoking in later life in a population of 11 589 male Swedish twins. IQ was measured at military conscription, and data on smoking and zygosity was obtained from the Swedish Twin Register. Information on social factors was extracted from censuses. Data on smoking was self-reported by the twins at the age of 22-47 years. Logistic regression models estimated with generalised estimating equations were used to explore possible associations between IQ and smoking among the twins as individuals as well as between-and within twin-pairs.

A strong inverse association between IQ and smoking status emerged in unmatched analyses over the entire range of IQ distribution. In within-pair and between-pair analyses it transpired that shared environmental factors explained most of the inverse IQ-smoking relationship. In addition, these analyses indicated that non-shared and genetic factors contributed only slightly (and non-statistically-significantly) to the IQ-smoking association. Analysis of twin pairs discordant for IQ and smoking status displayed no evidence that non-shared factors contribute substantially to the association. The question of which shared environmental factors might explain the IQ-smoking association is an intriguing one for future research.

“The association between county-level IQ and county-level crime rates”, Beaver & Wright2011:

An impressive body of research has revealed that individual-level IQ scores are negatively associated with criminal and delinquent involvement. Recently, this line of research has been extended to show that state-level IQ scores are associated with state-level crime rates. The current study uses this literature as a springboard to examine the potential association between county-level IQ and county-level crime rates. Analysis of data drawn from the National Longitudinal Study of Adolescent Health revealed statistically significant and negative associations between county-level IQ and the property crime rate, the burglary rate, the larceny rate, the motor vehicle theft rate, the violent crime rate, the robbery rate, and the aggravated assault rate. Additional analyses revealed that these associations were not confounded by a measure of concentrated disadvantage that captures the effects of race, poverty, and other social disadvantages of the county. We discuss the implications of the results and note the limitations of the study.

tattoos and piercings:

pg406, Strenze, “Intelligence and Success”, ch25 of Handbook of Intelligence Evolutionary Theory, Historical Perspective, and Current Concepts, ed Goldstein et al 2015:

Table 25.1 Relationship between intelligence and measures of success (Results from meta-analyses)

Measure of success





Academic performance in primary education




Poropat (2009)

Educational attainment




Strenze (2007)

Job performance (supervisory rating)




Hunter & Hunter1984

Occupational attainment




Strenze (2007)

Job performance (work sample)




Roth et al 2005

Skill acquisition in work training




Colquitt et al 2000

Degree attainment speed in graduate school




Kuncel et al 2004

Group leadership success (group productivity)



Judge et al 2004

Promotions at work




Schmitt et al 1984

Interview success (interviewer rating of applicant)




Berry et al 2007

Reading performance among problem children




Nelson et al 2003

Becoming a leader in group



Judge et al 2004

Academic performance in secondary education




Poropat (2009)

Academic performance in tertiary education




Poropat (2009)





Strenze (2007)

Having anorexia nervosa




Lopez et al 2010

Research productivity in graduate school




Kuncel et al 2004

Participation in group activities



Mann (1959)

Group leadership success (group member rating)



Judge et al 2004




Kim (2005)

Popularity among group members



Mann (1959)





DeNeve & Cooper (1998)

Procrastination (needless delay of action)




Steel (2007)

Changing jobs




Griffeth et al 2000

Physical attractiveness




Feingold (1992)

Recidivism (repeated criminal behavior)




Gendreau et al 1996

Number of children



Lynn (1996)

Traffic accident involvement




Arthur et al 1991

Conformity to persuasion



Rhodes & Wood1992

Communication anxiety




Bourhis & Allen1992

Having schizophrenia



Woodberry et al 2008

r correlation between intelligence and the measure of success, k number of studies included in the meta-analysis, N number of individuals included in the meta-analysis

“Intelligence in young adulthood and cause-specific mortality in the Danish Conscription Database – A cohort study of 728,160 men”, Christensen et al 2016:

An inverse association has been reported between early life intelligence and all-cause mortality. The aim of this study was to investigate whether this well-established association differed according to the underlying cause of death and across different birth cohorts. The associations between young adult intelligence and mortality from natural and external causes were investigated in the Danish Conscription Database (DCD), which is a cohort of more than 700,000 men born 1939–1959 and followed in Danish registers from young adulthood until late mid-life. Young adult intelligence was inversely related to all-cause mortality with a 28% higher risk of dying during the study period per 1 standard deviation (SD) decrease in intelligence test score (HR = 1.28 95% CI = 1.27–1.29). The strength of the observed inverse associations did not vary much across main groups of natural and external causes with the exception of the associations for mortality from respiratory diseases (HR = 1.61 95% CI = 1.55–1.67) and homicide (HR = 1.65 95% CI = 1.46–1.87) which were more pronounced compared to the rest. Moreover, for skin cancer mortality, each SD increase in intelligence test score was associated with a small increase in mortality risk (HR = 1.03 95% CI = 1.01–1.15). Furthermore, the association between intelligence and mortality was stronger for those born 1950–1959 compared to those born 1939–1949 for almost all natural and external causes of death.

“Association of Fluid Intelligence and Psychiatric Disorders in a Population-Representative Sample of US Adolescents”, Keyes et al 2017:

Objective: To investigate the association of fluid intelligence with past-year and lifetime psychiatric disorders, disorder age at onset, and disorder severity in a nationally representative sample of US adolescents.

Design, Setting, & Participants: National sample of adolescents ascertained from schools and households from the National Comorbidity Survey Replication-Adolescent Supplement, collected 2001 through 2004. Face-to-face household interviews with adolescents and questionnaires from parents were obtained. The data were analyzed from February to December 2016. DSM-IV mental disorders were assessed with the World Health Organization Composite International Diagnostic Interview, and included a broad range of fear, distress, behavior, substance use, and other disorders. Disorder severity was measured with the Sheehan Disability Scale.

Main Outcomes & Measures: Fluid IQ measured with the Kaufman Brief Intelligence Test, normed within the sample by 6-month age groups.

Results: The sample included 10 073 adolescents (mean [SD] age, 15.2 [1.50] years; 49.0% female) with valid data on fluid intelligence. Lower mean (SE) IQ was observed among adolescents with past-year bipolar disorder (94.2 [1.69]; P = .004), attention-deficit/hyperactivity disorder (96.3 [0.91]; P = .002), oppositional defiant disorder (97.3 [0.66]; P = .007), conduct disorder (97.1 [0.82]; P = .02), substance use disorders (alcohol abuse, 96.5 [0.67]; P < .001; drug abuse, 97.6 [0.64]; P = .02), and specific phobia (97.1 [0.39]; P = .001) after adjustment for a wide range of potential confounders. Intelligence was not associated with post-traumatic stress disorder, eating disorders, and anxiety disorders other than specific phobia, and was positively associated with past-year major depression (mean [SE], 100 [0.5]; P = .01). Associations of fluid intelligence with lifetime disorders that had remitted were attenuated compared with past-year disorders, with the exception of separation anxiety disorder. Multiple past-year disorders had a larger proportion of adolescents less than 1 SD below the mean IQ range than those without a disorder. Across disorders, higher disorder severity was associated with lower fluid intelligence. For example, among adolescents with specific phobia, those with severe disorder had a mean (SE) of 4.4 (0.72) points lower IQ than those without severe disorder (P < .001), and those with alcohol abuse had a mean (SE) of 5.6 (1.2) points lower IQ than those without severe disorder (P < .001).

Conclusions & Relevance: Numerous psychiatric disorders were associated with reductions in fluid intelligence; associations were generally small in magnitude. Stronger associations of current than past disorders with intelligence suggest that active symptoms of psychiatric disorders interfere with cognitive functioning. Early identification and treatment of children with mental disorders in school settings is critical to promote academic achievement and long-term success.



  • /doc/iq/ses/1997-gottfredson.pdf




  • Lead:

  • :

These examples show that, contrary to Shalizi’s claims, all cognitive abilities are inter-correlated. We can be confident about this because the best evidence for it comes not from the proponents of g but from numerous competent researchers who were hell-bent on disproving the generality of the positive manifold, only to be refuted by their own work .

Depending type of job and how performance is measured GMA explains between 30% and 70% of the variation in people’s work performance (ie. correlations of between .56 and .84), which is larger than any other known predictor.4 [“When performance is measured objectively using carefully constructed work sample tests (samples of actual job tasks), the correlation (validity) with intelligence measures is about .84—84% as large as the maximum possible value of 1.00, which represents perfect prediction. When performance is measured using ratings of job performance by supervisors, the correlation with intelligence measures is .66 for medium complexity jobs (over 60% of all jobs). For more complex jobs, this value is larger (eg. .74 for professional and managerial jobs), and for simpler jobs this value is not as high (eg. .56 for semi-skilled jobs). Another performance measure that is important is the amount learned in job training programs (Hunter et al 2006). Regardless of job level, intelligence measures predict amount learned in training with validity of about .74 (Schmidt, Shaffer, and Oh, 2008).” From: Schmidt, Frank L, and John E Hunter. “Select on intelligence.” Handbook of principles of organizational behavior(2000): 3-14.]

Evidence from several meta-studies shows that when performance is measured using work-sample tests, the correlation between GMA and performance is 0.84. When supervisor ratings are used, the correlation is lower, at 0.74 for high-complexity jobs.5 [Schmidt, Frank L, and John E Hunter. “Select on intelligence.” Handbook of principles of organizational behavior(2000): 3-14.]

GMA also predicts how high up you get in the job hierarchy - i.e. your occupational level.6 US Employment Service data shows a strong correlation (0.72) between GMA and occupational level and US military data shows that mean GMA scores are higher at higher occupational levels. Also, there is a wider variety of GMA scores at lower occupational levels than at higher ones. It seems that there are high-scoring people in low-level occupations, but low-scoring people are unlikely to get promoted to higher levels.7[Schmidt, Frank L, and John Hunter. “General mental ability in the world of work: occupational attainment and job performance.” Journal of personality and social psychology 86.1 (2004): 162.]

But to fully show the link we need to track people with known GMA over time to see if high GMA individuals end up being more successful. This has been done.8 [Schmidt, Frank L, and John Hunter. “General mental ability in the world of work: occupational attainment and job performance.” Journal of personality and social psychology 86.1 (2004): 162.] In a longitudinal study of 3,887 young adults, GMA predicted movement in the job hierarchy 5 years later. Another study found that if people were in a job that was less complex than their GMA would predict, they moved up to a more complex job and vice versa. The predictivity of GMA even holds when controlling for socioeconomic status by comparing biological siblings. “When the siblings were in their late 20s (in 1993), a person with average GMA was earning on average almost $18,000 less per year than his brighter sibling who had an IQ of 120 or higher and was earning more than $9,000 more than his duller sibling who had an IQ of less than 80.”9 [Schmidt, Frank L, and John Hunter. “General mental ability in the world of work: occupational attainment and job performance.” Journal of personality and social psychology 86.1 (2004): 162.]

The link has also been confirmed by two natural experiments. [Schmidt, Frank L, and John E Hunter. “Select on intelligence.” Handbook of principles of organizational behavior(2000): 3-14.]

With high GMA, people are more able to go beyond existing job knowledge and make judgements in unfamiliar situations.12 [Schmidt, Frank L, and John E Hunter. “Select on intelligence.” Handbook of principles of organizational behavior(2000): 3-14.]

Although GMA predicts performance in all jobs the more complex the job is13, the stronger the relationship between GMA and performance.14[Hunter, John E. “Cognitive ability, cognitive aptitudes, job knowledge, and job performance.” Journal of vocational behavior 29.3 (1986): 340-362.] And the more complex the job, the more variation there is between top performers and bottom performers.15 [Hunter, John E, Frank L Schmidt, and Michael K Judiesch. “Individual differences in output variability as a function of job complexity.” Journal of Applied Psychology 75.1 (1990): 28.] So if you have one of the highest levels of GMA in a highly complex job, you’ll have a high output compared to the average performer.

“Childhood intelligence in relation to major causes of death in 68 year follow-up: prospective population study”, Deary et al 2017:

“Intelligence and persisting with medication for two years: Analysis in a randomized controlled trial”, Deary et al 2009:

Similar Links

[Similar links by topic]