One of the many questions concerning intelligence that has intrigued me for years is the following: What are the consequences of the social structure of our society if groups of people differ in intellectual capacity?
I realize that this question is sensitive and that many people would rather think about it in silence than to discuss it openly. This issue is nonetheless far too important to ignore. From a purely scientific perspective – from my desire to understand the social world as it is – I believe that no issue should be left unexplored even if the truth may make us disquiet. From a more practical perspective, the insights in intelligence and its relationship to the social structure of society may help us make proper judgments on the effectiveness of government policies with regards to the assistance of certain social groups.
Richard J. Herrnstein & Charles Murray’s The Bell Curve: Intelligence and Class Structure in American Life (1994) explores the aforementioned question. The book has been highly controversial upon its publication, and has also been a best-seller; selling over 500,000 copies. The thesis of the book is that intelligence is a much better predictor of many social variables than social economic status (SES). Their research is mostly dependent on the National Longitudinal Study of Youth (NLSY), which is a study that began in 1979. Originally, 12,586 participants between the ages of 14 and 22 took part of the study. They were chosen “to provide adequate sample sizes for analyzing crucial groups (for example, by oversampling blacks, Latinos, and low-income whites), and also incorporating a weighting system so that analysts could determine the correct estimates for nationally representative samples of their age group” (Herrnstein & Murray, 1994, p. 119).
In this review, I will first explain three different conceptions of intelligence and which conception Herrnstein & Murray have taken as the foundation for their research. Thereafter, I will present some of the book’s research outcomes on the relationship between intelligence and a wide variety of social variables. The book itself consists of the following 4 parts from which I will present excerpts that I have found most interesting, surprising or astonishing:
I. The emergence of a cognitive elite;
II. Cognitive classes and social behavior;
III. The national context;
IV. Living together.
I will not explain the overall content chapter by chapter. Finally, I will provide some personal thoughts on the book.
Intelligence: classicists, revisionists, and radicals
Herrnstein, who was at the time an experimental psychologist at Harvard, had already released an article in Atlantic Monthly (1971) in which he contended that “because IQ is substantially heritable, because economic success in life depends in part on the talents measured by IQ tests, and because social standing depends in part on economic success, it follows that social standing is bound to be based to some extent on inherited differences” (Herrnstein & Murray, 1994, p. 10). This proposition however, was considered too controversial and Herrnstein became slandered by the politically correct who called him a racist or a scientific charlatan. The professional community hence learned that it was best not to speak out on such issues in public. The research in measurable human intellect has nonetheless carried outside the public eye, and according to Herrnstein & Murray the professional community has already reached a general consensus on many issues with regards to intelligence – including the notion that intelligence is partly genetically heritable. (Herrnstein & Murray, 1994, pp. 13-14) There are still disagreements, but these disagreements are along different lines. By the early 1990s, one could roughly divide the professionals into one of the following three groups: (a) the classicists, (b) the revisionists, and (c) the radicals.
(a) The classicists
The classicists believe in intelligence as a structure which can be quantified as ‘g’ which stands for general intelligence. They believe that modern IQ tests do a reasonably good job of measuring g, and that it is unbiased against socioeconomic, ethnic, or racial subgroups if one administers it properly. (Herrnstein & Murray, 1994, pp. 14-15)
(b) The revisionists
The revisionists believe in intelligence as information processing. They accept that there is a general mental ability (g), but they differ in their approach to measure g. The revisionists emphasize intelligence as a form of information processing. Sternberg for example asserts that IQ tests should test people’s abilities of adaptation (the ability to make the best of a situation), shaping (ability to shape the external environment so that it conforms to one’s desired state of affairs), and selecting (ability to select new environments) (Herrnstein & Murray, 1994, p. 17).
(c) The radicals
The radicals deny the existence of the structural intelligence g. Howard Gardner, a radical, for example argues that there are seven distinct intelligences: linguistic, musical, logical-mathematical, spatial, bodily-kinesthetic, intrapersonal, and interpersonal. Radicals in addition also deny that their thesis can be defended with quantitative data. (Herrnstein & Murray, 1994, p. 18)
Although Herrnstein & Murray sympathize with the revisionists and the radicals, they “will be drawing most heavily from the classical tradition” (Herrnstein & Murray, 1994, p. 19). Despite persistent disagreements among professionals, the following six conclusions stand beyond technical significant dispute:
1) There is such a thing as a general factor of cognitive ability on which human beings differ.
2) All standardized tests of academic aptitude or achievement measure this general factor to some degree, but IQ tests expressly designed for that purpose measure it most accurately.
3) IQ scores match, to a first degree, whatever it is that people mean when they use the word intelligent or smart in ordinary language.
4) IQ scores are stable, although not perfectly so, over much of a person’s life.
5) Properly administered IQ tests are not demonstrably biased against social, economic, ethnic, or racial groups.
6) Cognitive ability is substantially heritable, apparently no less than 40 percent and no more than 80 percent. (Herrnstein & Murray, 1994, pp. 22-23)
Herrnstein & Murray recognize that “[M]easures of intelligence have reliable statistical relationships with important social phenomena, but they are a limited tool for deciding what to make of any given individual” (Herrnstein & Murray, 1994, p. 21).
PART I. THE EMERGENCE OF A COGNITIVE ELITE
According to Herrnstein & Murray, the proportion of people who went to college increased rapidly (fifteen-fold) from 1900 to 1990. With the influx of more students into colleges, many colleges became more selective since the 1950s and some attracted high school graduates from the top few percentiles of cognitive ability as measured by SAT tests. For instance, the admittance rate of Harvard in 1952 was 66 percent against 33 percent in 1960. The SAT verbal score was 583 in 1952 against 678 in 1960. Also the SAT math score increased by almost 100 points in those eight years. “The average Harvard freshman in 1952 would have placed in the bottom 10 percent of the incoming class by 1960” (Herrnstein & Murray, 1994, p. 30). With the growth in college population, a more efficient recruiting process by cognitive ability, and the sorting of students by cognitive ability among colleges, American society became more cognitively stratified.
Comparing the graphs of Americans with and without college degrees in 1930 and 1990, one can see that the total proportion of college graduates has increased tremendously. The mean IQ of all college graduates in 1930 has risen somewhat, but not as significant as those graduates from the Ivy League & Seven Sisters. The mean IQ of college graduates has just risen from 111 in 1930 to 113 in 1990. By 1990, the mean IQ of graduates from these prestigious universities was 2.7 SDs higher than the mean IQ of the general population. This means that their mean IQ has shot up from 120 in 1930 to 142 in 1990. (Herrnstein & Murray, 1994, pp. 45-46) To provide the Dutch readers with a way to compare these IQ scores, I will provide here the mean IQ scores of VMBO (100), Havo (108), VWO Atheneum (116), and VWO Gymnasium (118) students in their third year (van Dijk & Tellegen, 2004, p. 15). Please note that the mean IQs also differ among the disciplines in college. Click here for my post in which I discuss the mean IQs per college major.
There is furthermore also a strong correlation between IQ and the probability that one enters and finishes college. The mean IQ of those who finish college is significantly higher than those who enter college. (Herrnstein & Murray, 1994, p. 37) Please note that this would mean that the average IQ of entering students of elite universities is, although I do not have specific data in this regards, probably significantly lower than those who graduate from these universities as well.
Cognitive partitioning also happens with respect to occupation. The higher pay associated with occupations that are more cognitively demanding attracts people who are more cognitively abled. IQ is moreover also a better predictor of someone’s job performance than for example biographical data, reference checks, education, interview, college grades, interest, and age (Herrnstein & Murray, 1994, p. 81). If an employee’s job performance is dependent on intelligence, then the economic value of the employee and his wages are also dependent on intelligence. If “[G]etting high quality for a complex job can be worth large multiples of what it is worth to get equally high quality for a simpler job” (Herrnstein & Murray, 1994, p. 83), then employers will naturally look for ways to hire the more cognitively abled. This has for example resulted in strict educational requirements for jobs like a minimum grade point average.
Herrnstein & Murray believe that these trends in cognitive partitioning by education, occupation, and economic pressure are both “heartening and frightening, more or less in equal measure” (Herrnstein & Murray, 1994, p. 50). It is frightening, because cognitive partitioning makes people live in ‘bubbles’ – encapsulated from the general population – who nonetheless yield the greatest influence on society. It is however heartening that so many people have the opportunity to enjoy a college education.
PART II. COGNITIVE CLASSES AND SOCIAL BEHAVIOR
Part II of the book investigates only non-Hispanic whites, and examines the relationship between their intelligence and various social behaviours. Herrnstein & Murray start the section with a cognitive division of the population into five different classes:
– (I) the ‘very bright’ with an IQ of 125 and higher comprising 5% of the population;
– (II) the ‘bright’ with an IQ between 110 and 125 comprising 20% of the population;
– (III) the ‘normal’ with an IQ between 90 and 110 comprising 50% of the population;
– (IV) the ‘dull’ with an IQ of 75 to 90 comprising 20% of the population;
– (V) the ‘very dull’ with an IQ of less than 75 comprising 5% of the population. (Herrnstein & Murray, 1994, pp. 121-122)
The distribution of IQ, defined for these cognitive classes, is represented in the following graph:
The common procedure of Herrnstein & Murray is to “run regression analyses in which the independent variables include IQ and parental SES. The result is a statement of the form: ‘Here is the relationship of IQ to social behavior X after the effects of socioeconomic background have been extracted,’ or vice versa” (Herrnstein & Murray, 1994, p. 123).
The first variable that Herrnstein & Murray examine is ‘poverty’. They maintain that
low intelligence is a stronger precursor of poverty than low socioeconomic background. Whites with IQs in the bottom 5 percent of the distribution of cognitive ability are fifteen times more likely to be poor than those with IQs in the top 5 percent. (Herrnstein & Murray, 1994, p. 127)
In the graph below, one can indeed find that IQ is a much better predictor of someone’s probability of being in poverty than SES. A person with an IQ of +2 SDs above the mean (130) has less than 2 percent chance of being in poverty. A person of -2 SDs (70) has 26 percent chance of being in poverty.
This is another graph that illustrates the strong correlation between IQ and poverty among the white population.
Herrnstein & Murray sum up the correlation between poverty and intelligence as follows: “Low intelligence means a comparatively high risk of poverty. If a white child of the next generation could be given a choice between being disadvantaged in socioeconomic status or disadvantaged in intelligence, there is no question about the right choice” (Herrnstein & Murray, 1994, p. 135).
Another interesting finding in the chapter is the trend line of poverty from WWII to the 1990s. I find it remarkable that the war on poverty which started in the 1960s has not helped reduce poverty. According to the Institute for Research on Poverty, poverty had been on a decades-long declining trend before the Johnson administration jumpstarted the programme. See the graph below.
Does this mean that the war on poverty has been highly ineffective or maybe even inducing poverty? This is something I would like to explore more in further readings.
As with poverty; the higher your IQ, the greater the probability that you finish high school or college. Again, IQ is a much better predictor of someone’s success in education than SES. See the figure below for the predictive power of IQ and a person’s chance of finishing high school (Herrnstein & Murray, 1994, pp. 148-149).
If a student would be in the top 2 percent of the socioeconomic scale, he would still have just 40 percent chance of getting a college degree if he is of average intelligence. In contrast, a student who is of average SES, but who is in the top 2 percent of IQ has more than 75 percent chance of graduating from college. The stereotype that one’s socioeconomic status heavily influences one’s educational success does therefore not seem to be true in the real world. Even students who are in the top quartile of SES, but who are of below average IQs, had only 12 percent chance of getting a college degree. This means that for white youths, being smart is more important than being privileged in getting a college degree. (Herrnstein & Murray, 1994, pp. 152-154) See the figure below:
Unemployment, Idleness, and Injury
Surprisingly, a higher IQ also correlates with a person’s physical problems which can prevent the person from holding an occupation. Herrnstein & Murray admit though that they do not know why this is the case, but the reason could be that less intelligent people are less able to perceive and estimate risks accordingly. (Herrnstein & Murray, 1994, 161)
Herrnstein & Murray have also found that people with higher intelligence are less likely to be unemployed or out of the labour force than people with lower intelligence (Herrnstein & Murray, 1994, pp. 158-163).
Herrnstein & Murray have found similar relational results for IQ and illegitimacy, welfare dependency, parental abuse, parental neglect, crime, and middle class values.
PART III. THE NATIONAL CONTEXT
Part III of the book starts with the highly controversial chapters on cognitive differences between ethnic groups. Herrnstein & Murray have looked into the NLSY data to find the IQ differences between Whites and African American Blacks; and according to their findings the average IQ of Blacks is 1.2 SDs below the average IQ of Whites. Herrnstein & Murray strongly emphasize the importance that we should not treat individuals differently despite the lower mean in IQ score:
a difference of 1.2 standard deviation means considerable overlap in the cognitive ability distribution for blacks and whites… For any equal number of blacks and whites, a large proportion have IQs that can be matched up. This is the distribution to keep in mind whenever thinking about individuals. (Herrnstein & Murray, 1994, pp. 278-279)
The finding that the mean IQ of Blacks is lower than those of Whites has been observed in numerous other studies. The following graph shows an overview of 156 studies concerning the B/W IQ gap. The mean of the B/W difference of these studies stands at 1.08 SDs (16 IQ points). (Herrnstein & Murray, 1994, pp. 276-278)
How would the Blacks’ social standing be in comparison with Non-Latino Whites and Latinos after controlling for IQ? The answer to this question has been strikingly surprising to me. I will give an overview of Herrnstein & Murray’s findings with respect to the probability of holding a bachelor’s degree, probability of being in a high-IQ occupation, wages, probability of being in poverty, probability of being employed for a month or more, probability of women bearing their children out of wedlock, probability that a poor mother has been on welfare, and the probability of ever having been interviewed in a correctional facility.
Probability of holding a bachelor’s degree
The conventional view is that Blacks and Latinos lag far behind in education compared to Whites. This view is indeed supported by the NLSY data if one would not control the research sample for IQ; Whites of average age (29) have a 27 percent chance of holding a bachelor’s degree compared to 11 percent for Blacks and 10 percent for Latinos. However, the picture changes after controlling for IQ. Looking only at people of average age and average IQ for college graduates (114), we will find that the chances of holding a bachelor degree for Whites and Latinos are more or less equal (50 percent), whereas Blacks have a significantly higher chance (68 percent). (Herrnstein & Murray, 1994, p. 320)
In a separate chapter in Part IV, Herrnstein & Murray argue that the Blacks’ higher chances can be attributed to reverse discrimination policies among college admission bodies. Looking into the SAT data, they find that Blacks have an advantage over other ethnicities when it comes to college enrollment. The average Black person, admitted at America’s top 26 colleges, scores at the 10th percentile of white students. Surprisingly, Asians seem to suffer from reverse discrimination. The average Asian was at the 60th percentile of white students. The Black edge corresponds to a 180 SAT points advantage, whereas Asians faced a median penalty of 30 SAT points. (Herrnstein & Murray, 1994, pp. 451-452)
Probability of holding high-IQ occupations
A similar relationship seems to be prevalent in the probabilities of being in a high-IQ occupation. Before controlling for IQ, Whites of average age (29) have a 5 percent chance of being in a high-IQ occupation. Both Blacks and Latinos have 3 percent chance. After controlling for IQ, this picture reverses and Whites have a 10 percent chance, Blacks 26 percent, and Latinos 16 percent. (Herrnstein & Murray, 1994, pp. 321-322)
Herrnstein & Murray substantiate these findings in Part IV, where they show the B/W difference in SDs per job category. They suggest that employers are using dual standards for Black and White job applicants. The dual standards are partly or possibly even mainly influenced by affirmative action policies – pressures brought on to employers by government policies regarding the representation of minority groups. (Herrnstein & Murray, 1994, pp. 488-489) The findings have led Herrnstein & Murray to suggest that:
aggressive affirmative action does produce large racial discrepancies in job performance in a given workplace. It is time that this important area be explored systematically. In coming to grips with policy, a few hard truths have to be accepted. First, there are no good ways to implement current job discrimination law without incurring costs in economic efficiency and fairness to both employers and employees. Second, after controlling for IQ, it is hard to demonstrate that the United States still suffers from a major problem of racial discrimination in occupations and pay. (Herrnstein & Murray, 1994, pp. 479-480)
Annual wages for a year-round worker
Again, Whites seem to earn significantly more than Blacks and Latinos before controlling for IQ in 1989. For a person of average age (29), Whites earn on average $27,372, Blacks $20,994, and Latinos $23,409. This means that Blacks earn on average 23 percent less than Whites, and Latinos 14 percent. However, after controlling for IQ, the wage gap between these three ethnic groups more or less disappears: Whites $25,546, Blacks $25,001, and Latinos $25,159. What would happen if we do not control for IQ, but instead control for education? “[I]n the 1992 national data collected by the Bureau of the Census median earnings of year-round, full-time workers in 1992 were $41,005 for white male graduates with a bachelor’s degree and only $31,001 for black males with the same degree” (Herrnstein & Murray, 1994, p. 324). What if we would control for age, education, and socioeconomic background, but not for IQ? “The result is that black wages are still only 84 percent of white wages, again suggesting continuing racial discrimination” (Herrnstein & Murray, 1994, p. 324). Finally, what would happen to the wage gap if we would add gender to age and IQ in the analysis? “Controlling for age, IQ, and gender (ignoring education and parental SES), the average wage for year-round black workers in the NLSY sample was 101 percent of the average white wage” (Herrnstein & Murray, 1994, p. 326).
There are however also other variables like the probability of being unemployed for a month or more, the probability of having married by age 30, probability that women bear their children out of wedlock, and probability of having been on welfare etc. for which IQ explains away ethnic differences just moderately or not at all. There must be other factors involved, and it is up to future researchers to pick up on this investigation. I will show some of these findings here.
Finally, I would like to present Herrnstein & Murray’s findings on dysgenesis. According to Herrnstein & Murray, the United States have been in a downward trend with regards to cognitive ability. This is instigated by lower birth rates of women with greater cognitive ability who, in general, spend more of their reproductive years in school than women of lower cognitive ability. The opportunity costs of bearing children on the privileged are much greater than the opportunity costs on the poor for whom a career does not seem to be a realistic option. In addition, Herrnstein & Murray suggest that welfare for poor people can also lower their opportunity costs and raise their incentives to have babies. (Herrnstein & Murray, 1994, pp. 344-345) To illustrate findings that women with higher education and greater cognitive ability have fewer children, I will add the following graphs.
The Bell Curve still stands strong, twenty years after its first publication. This book is excellent for anyone who wants to have a well-reasoned reading into intelligence and how it influences the social structure of American society. Many findings in the book will appear to the reader as common sense though – such as the statistics that college attracts the more intellectually abled among people or that certain jobs require more cognitive abilities and therefore attract people with higher IQs, and that the cognitive partitioning in both education and occupation must both necessarily lead to cognitive stratification in society. Most people also already know that intelligence is partly heritable and that there are generic traits among different ethnic or social groups. In this sense, many of the findings are already known to the wider public – although not with the same scientific rational and evidence that Herrnstein & Murray provides.
The book however, also harbors some controversies. The main controversies are centred around chapter 13 and 14 which discuss ‘Ethnic Differences in Cognitive Ability’ and ‘Ethnic Inequalities in Relation to IQ’. Even though this section makes up just a small part of the book, most criticizers have only focused on this section. Despite Herrnstein & Murray’s scrupulous efforts to present the book as objectively and scientifically as possible, some reviewers like Bob Herbert (1994) had called it “a scabrous piece of racial pornography masquerading as serious scholarship” (Herbert, 1994, October 26) and that the “book is just a genteel way of calling somebody a nigger” (Herbert, 1994, October 26). I entirely disagree with Herbert. My impression is that many reviews and opinions expressed on the book are entirely unsubstantiated – devoid of any appeal to scientific reasoning.
I hold two suspicions with regards to the many reviews I have read. The first one is that despite its large sales numbers, very few people have actually read the book from cover to cover due to its size (850 pages if one includes the appendices and notes) and its frequent usage of statistical jargon which make it not easily comprehendible for the layman.
My second suspicion is that many people have projected their personal anxieties about racism unto Herrnstein & Murray’s proposition that ethnic groups have different IQ scores. These anxieties have prevented them from reaching objective and rational conclusions about the data presented here in the book. For these reasons, I have felt that I could not count on others’ reviews and that I should read the book myself as to form an objective opinion. It is important to understand that the book discusses means of social groups. Nowhere, does it encourage us to form prejudices against individuals. So no, contrary to some beliefs, I do not believe that it is an angry book, and no, it is not racist, nor does it have any political agenda.
To conclude my review, I would like to present a joint statement of 52 scientific professionals in support of Herrnstein & Murray’s findings that appeared in The Wall Street Journal shortly after the publication of The Bell Curve. It suggests that what Herrnstein & Murray have written has already been generally accepted by the scientific community at least twenty years ago.
The statement contains the following assertions:
The meaning and measurement of intelligence
1. Intelligence is a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test taking smarts.
2. Intelligence, so defined, can be measured, and intelligence tests measure it well.
3. While there are different types of intelligence tests, they all measure the same intelligence.
4. The spread of people along the IQ continuum, from low to high, can be represented well by the bell curve (in statistical jargon, the “normal curve”).
5. Intelligence tests are not culturally biased against American blacks or other native-born, English-speaking peoples in the U.S. Rather, IQ scores predict equally accurately for all such Americans, regardless of race and social class.
6. The brain processes underlying intelligence are still little understood.
7. Members of all racial-ethnic groups can be found at every IQ level.
8. The bell curve for whites is centered roughly around IQ 100; the bell curve for American blacks roughly around 85; and those of different subgroups of Hisanics roughly midway between those for whites and blacks. The evidence is less definitive for exactly where above IQ 100 the bell curves for Jews and Asians are centered.
9. IQ is strongly related, probably more so than any other single measurable human trait, to many important educational, occupational, economic, and social outcomes.
10. 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.
11. The practical advantages of having a higher IQ increase as life settings become more complex (novel, ambiguous, changing, unpredictable, or multifaceted).
12. Differences in intelligence certainly are not the only factor affecting performance in education, training, and highly complex jobs (no one claims they are), but intelligence is often the most important.
13. Certain personality traits, special talents, aptitudes, physical capabilities, experience, and the like are important (sometimes essential) for successful performance in many jobs, but they have narrower (or unknown) applicability or “transferability” across tasks and settings compared with general intelligence.
Source and stability of within-group differences
14. Individuals differ in intelligence due to differences in both their environments and genetic heritage. Heritability estimates range from 0.4 to 0.8 (on a scale from 0 to 1), most thereby indicating that genetics plays a bigger role than does environment in creating IQ differences among individuals.
15. Members of the same family also tend to differ substantially in intelligence (by an average of about 12 IQ points) for both genetic and environmental reasons.
16. That IQ may be highly heritable does not mean that it is not affected by the environment.
17. Although the environment is important in creating IQ differences, we do not know yet how to manipulate it to raise low IQs permanently.
18. Genetically caused differences are not necessarily irremediable (consider diabetes, poor vision, and phenylketonuria), or are environmentally caused ones necessarily remediable (consider injuries, poisons, severe neglect, and some diseases).
Source and stability of between-group differences
19. There is no persuasive evidence that the IQ bell curves for different racial-ethnic groups are converging.
20. Racial ethnic differences in IQ bell curves are essentially the same when youngsters leave high school as when they enter first grade.
21. The reasons that blacks differ among themselves in intelligence appear to be basically the same as those for why whites (or Asians or Hispanics) differ among themselves.
22. There is no definitive answer to why IQ bell curves differ across racial-ethnic groups.
23. Racial-ethnic differences are somewhat smaller but still substantial for individuals from the same socioeconomic backgrounds.
24. Almost all Americans who identify themselves as black have white ancestors – the white admixture is about 20%, on average – and many self-designated whites.
Implications for social policy
25. The research findings neither dictate nor preclude any particular social policy, because they can never determine our goals. They can, however, help us estimate the likely success and side-effects of pursuing those goals via different means. (Gottfredson, 1997, pp. 13-15)
Dijk van, H., & Tellegen, P. (2004). Nederlandse Intelligentietest voor Onderwijsniveau: Handleiding en Verantwoording. Retrieved from www.testresearch.nl/nio/nioverkort.doc
Gottfredson, L.S. (1997). Mainstream Science on Intelligence: An Editorial With 52 Signatories, History, and Bibliography. Intelligence, 24, 1, pp. 13-23.
Herbert, B. (1994, October 26). In America; Throwing a Curve. Retrieved from www.nytimes.com/1994/10/26/opinion/in-america-throwing-a-curve.html
Herrnstein, R.J., & Murray, C. (1994). The Bell Curve: Intelligence and Class Structure in American Life. New York: The Free Press.