PART III
“CLASS IS A FUNCTION OF PRIVILEGE”
The system is rigged in favor of heterosexual white males. The privilege accorded them accounts for who gets ahead in America and who is kept on the bottom. That’s one sound-bite version of a core element of the orthodoxy. It began in academia in the 1960s, spilled over into American politics after the turn of the century, and by the 2016 election had become a common position among people who self-identify as progressives. Its more nuanced version is that the system is not completely rigged, but the dice are loaded in favor of whites, males, and heterosexuals—they don’t win all the time, but they win far more often than they deserve.
Another sound-bite version of “class is a function of privilege” is that class is a function of wealth: The system is rigged in favor of the rich, who pass their money to the next generation, who in turn become the next generation of the upper class. The more nuanced version is that social mobility has diminished in recent decades, symptomatic of an entrenched upper class.
Meanwhile, those who self-identify as conservatives commonly believe that class is a function of character, determination, and hard work. It draws from the traditional American credo: In America, people can become anything they want to be if they try hard enough. The more nuanced version is that people differ in their talents, but for most occupations and roles in life, innate talent is not nearly as important as character, determination, and hard work.
Part III is about a third narrative, not as dark as the orthodoxy’s nor as idealistic as the traditional one. Class is a function of the genetic lottery plus character, determination, hard work, and idiosyncratic circumstances. The sociological, economic, and psychometric evidence for it has been available since at least the 1980s and on some topics for longer. That’s why a quarter of a century ago, Richard Herrnstein and I were able to write a book of 800-plus pages with the subtitle “Intelligence and Class Structure in American Life.”1
The book’s main title was The Bell Curve. In many ways, it documents the ways in which a segment of American society is a indeed morphing into a castelike upper class. But inherited wealth is a tangential contributor. The bare bones of its argument are that the last half of the twentieth century saw two developments of epochal importance: First, technology, the economy, and the legal system became ever more complex, making the value of the intellectual ability to deal with that complexity soar. Second, the latter half of the twentieth century saw America’s system of higher education become accessible to everyone with enough cognitive talent. The most prestigious schools, formerly training grounds for children of the socioeconomic elite, began to be populated by the students in the top few percentiles of IQ no matter what their family background might be—an emerging cognitive elite. By 2012, what had been predictions about the emerging cognitive elite as we were writing in the early 1990s had become established social facts that I described in another book, Coming Apart.
The purpose of Part III is to bring the third narrative up to date, explicitly addressing not just the role of IQ but of other abilities, the role of genes in determining those other abilities, the distinctions among different kinds of environmental influences, and the interactions between genes and environment.
But What About White Male Privilege and Intersectionality?
In 1989, legal scholar Kimberlé Crenshaw provided a new vocabulary for conceptualizing why class is a function of privilege, not talent. “In race discrimination [legal] cases, discrimination tends to be viewed in terms of sex-or class-privileged Blacks; in sex discrimination cases, the focus is on race-and class-privileged women,” she wrote. “This focus on the most privileged group members marginalizes those who are multiply-burdened and obscures claims that cannot be understood as resulting from discrete sources of discrimination.”2 Separating issues of racial discrimination, gender discrimination, and socioeconomic class was theoretically and empirically wrong.
It was not long before Crenshaw’s ideas and her introduction of the word intersectionality had been expanded into a full-blown theoretical approach that posits an interaction effect across different kinds of oppression. The original focus on women and blacks expanded to apply to all people who suffered from their identities as women, blacks, poor people, gays, the elderly, the disabled, and others. Two leading scholars of intersectionality theory, Margaret Andersen and Patricia Collins, put it this way when introducing the ninth edition of their anthology, widely used as a college textbook:
Fundamentally, race, class, and gender are intersecting categories of experience that affect all aspects of human life; they simultaneously structure the experiences of all people in this society. At any moment, race, class, or gender may feel more salient or meaningful in a given person’s life, but they are overlapping and cumulative in their effects.3 [Emphasis in the original.]
Together, the dimensions of intersectionality combine to form what Andersen and Collins labeled a “matrix of domination.”
The rhetoric is compelling and fuels endless ideological arguments. But the empirical situation is less fraught. I suggest there is room for agreement on two broad statements: Racism and sexism still play a role in determining who rises to the top, but that role is not decisive. We can have a range of opinions about whether the roles of racism and sexism merit the adjectives “large” or “small,” and advocate different public policies depending on our different perspectives, without affecting the relevance of the roles of genes, environment, and their interactions that constitute the topic of Part III.
A COMPROMISE
My proposition is that racism and sexism are no longer decisively important in determining who rises to the top. To support that proposition, I am about to demonstrate that ethnic differences in two major components of class—educational attainment and income—nearly disappear (or in some cases favor ethnic minorities) for people at similar IQ levels. Let’s suppose that you think the exercise is meaningless because you reject this use of IQ scores to make racial comparisons. You can nonetheless read Part III profitably if you are willing to consider the evidence that class structure within ethnic groups is shaped by the dynamics I describe—that, to take the most important example, white class structure is shaped by these dynamics.
I will also mention, however, that defending your belief that ethnic differences in IQ are meaningless is tough. The ways of defending it that first come to mind don’t work, for reasons described in the note.[4]
I use two indicators, educational attainment and earned income, to make that case. The data come from the 2018 Current Population Survey (CPS) and two cohorts of the National Longitudinal Survey of Youth (NLSY). The details for each of the following empirical claims are given in the notes.
Educational attainment by sex. Even without adjusting for anything, there’s no female disadvantage to worry about when it comes to educational attainment. Women now have higher mean years of education and a higher percentage of college degrees than men and have enjoyed that advantage for many years. These advantages persist over all IQ levels.[5]
Educational attainment by ethnicity. In terms of the raw numbers, Asians have higher educational attainment than any other ethnic group. Blacks and Latinos have substantially lower educational attainment than whites, but these discrepancies are more than eliminated after adjusting for IQ.[6] Blacks have more mean years of education and higher proportions of college degrees than whites at comparable IQ levels. After taking IQ into account, Latino and white levels of educational attainment are similar. Asians retain their advantage over whites after adjusting for IQ.[7]
Earned income by sex. A substantial female disadvantage in earned income exists, but it is almost entirely explained by marriage or children in the household. Using Current Population Survey data for 2018, earnings for women who were not married, had no children living at home, and worked full-time were 93 percent of the earnings of comparable men.[8] Married women with children in the house have considerably lower earned income even after adju
sting for IQ, but the main source of the income discrepancy is not that married women in the labor force earn less than unmarried women, but that married men earn more than unmarried men.[9]
Earned income by ethnicity. Using raw 2018 data from the CPS, Asians have higher mean earned income than whites, while Blacks and Latinos have substantially lower mean earned income than whites.[10] Once again, adjusting for IQ changes that picture dramatically. The note reports multivariate results for two large, nationally representative longitudinal surveys. In the earlier survey, adjusting for IQ wipes out the ethnic income differential among whites, blacks, and Latinos (Asians were not included in this survey). In the latter survey, whites and Latinos have effectively the same earned income while the fitted mean for blacks is 84 percent of the fitted mean for whites. The fitted mean for Asians is 57 percent higher than the fitted mean for whites.[11]
Let me be clear: I am not using these numbers to say that women, blacks, and Latinos do not still face problems because of sexism and racism. These numbers say nothing about individuals being passed over for promotions because of their sex or ethnicity, about glass ceilings, or about discriminatory or harassing interactions in the workplace. But there can be many people who legitimately think they haven’t gotten fair treatment without justifying the rhetoric that the orthodoxy uses about white male privilege. If we’re comparing men and women with similar IQs or members of different ethnicities with similar IQs, there’s only one American group that appears to be privileged for mysterious reasons. Martian sociologists investigating us with an unprejudiced eye would have no trouble identifying it: Americans of Asian ancestry.
There are a host of “Yes, but…” responses that different readers will have. The leading one is probably that IQ is in itself a function of privilege produced by affluence and good schools. And that brings us back to the topic of Part III.
Part III is about the role of genes in shaping this new class structure. Describing that role involves three steps, each of which gets a chapter of its own:
Establishing the heritability of cognitive repertoires and the relative unimportance of family background.
Demonstrating that those cognitive repertoires are important causes of success.
Examining the potential ways to mitigate the role of genes in determining success.
But first you need to be familiar with the framework for disentangling the roles of nature and nurture, to which I now turn.
10
A Framework for Thinking About Heritability and Class
We know a great deal about genes and class, far more than we know about genes and gender or genes and race, and the basics have been known for decades—that’s what I meant in the introduction when I said that the archaeological site for exploring class had been effectively closed until the genome was sequenced.
It may sound odd to put it that way, because the orthodoxy still barely acknowledges that genes play any role in human behavior, let alone shape socioeconomic classes. On almost any campus in the country, you can find sociologists who still assure their students that it’s all hereditarian pseudoscience.1 Class is driven by white privilege and the oppression of the patriarchy. But among psychologists who are familiar with the data, such views are exasperating without being an impediment to their work. There are still a few holdouts, but psychologists’ debates about heritability generally start from common understandings.2
Francis Galton was the first person to try to study heritability scientifically. His book Hereditary Genius (1869) presented evidence from British history that people with excellence in the same field—judges, parliamentarians, poets, scientists, even wrestlers and oarsmen—tended to be related by blood.3 Twentieth-century scientists took up where he left off. The intuitive thought here is that if genes are important, people who are more closely related will resemble each other more—siblings will resemble each other more than half siblings, for example.4 And so it has turned out in practice.
Let’s return to our running example, height. If you divide a perfectly random assortment of people into two groups and correlate their heights, the correlation coefficient will be around zero. If the assortment of people consists instead of pairs of half siblings, the correlation will be around +.25. For full siblings it will be about +.50. For identical twins it will approach +1.00.5 The rising correlation reflects the rising percentage of genes that the pairs share. It forms the basis for a powerful research methodology for calculating the heritability of a trait versus the contribution of the environment: Compare the results for identical twins and fraternal twins—more formally, monozygotic (MZ) twins, formed from a single egg that splits, and dizygotic (DZ) twins, created by two fertilized eggs. But to explain why the method is so powerful, first I need to unpack the meaning of heritability.
Heritability
Definition
People from time immemorial have noticed the resemblances of parents and children. Languages around the world have adages reflecting them—in English, for example, “the apple doesn’t fall far from the tree” and “a chip off the old block.”
The scientific definition of heritability is unrecognizably different. Expressed in words, heritability is a ratio calculated as the variance attributable to genes divided by total variance in the phenotype. Mathematically, the kind of heritability that I will be discussing, narrow heritability, is denoted as h2.
A MINI-INTERLUDE
I don’t need to get into as many technicalities about heritability as were required for the discussions of sex and population differences, but you need to be aware of the distinction between broad heritability and narrow heritability. Broad heritability, denoted as H2, refers to the combination of both additive and nonadditive sources of variation. Narrow heritability, denoted as h2, is limited to additive variation.
To illustrate what “additive” means, consider a genetic site in a flower in which one allele codes for red and the other codes for white. The flower is red if the two alleles in the SNP both code for red, white if they both code for white, and pink if one codes for red and the other codes for white. The color of the flower is the result of adding the effects of the pair of alleles in the person’s genotype.
If an interaction between the two alleles is involved, the effects of the alleles don’t add up in the same simple way. For example, suppose that one site codes for red or white and another site codes for whether color pigment will be produced. To get a pink or red flower not only depends on the site coding for color but also requires that the site coding for pigment production is a “yes.” That’s one type of nonadditive heritability, called epistasis. The other main type of interaction is dominance, involving alleles at the same site. Hemophilia is an example. It occurs only if both alleles code for hemophilia. The allele coding for hemophilia is recessive while the “normal” allele is dominant.
Additive variation is the most prevalent form, as I will document. Standard practice has been to fit a model based on additive variation and test to see how well it fits. If the fit is poor, the possibility of nonadditive variance needs to be explored.
Heritability is subject to misunderstandings. A common one is to confuse heritability, which refers specifically to the role of genes, with inherited, which can refer to things that are passed down through generations whether by genes, parenting, family traditions, or wills.
Another common misunderstanding is to think that the heritability of a trait refers to individuals. Mathematically, heritability refers to a whole population. Suppose that genes explain 70 percent of a population’s variance in height. You can use this information to conclude that “genes probably have a lot to do with how tall Joe is,” but it does not mean that “genes explain 70 percent of how tall Joe is.”
Heritability is not a fixed number for a given trait. It can vary by age, for example. We will encounter an example of this when we get to the heritability of IQ: Counterintuitively, it increases as people get older.
Heritability also varies by population. For
example, suppose you want to know the heritability of performance on the SAT and you compare two sets of students. One sample is from an ordinary New York City public high school and the other is from Stuyvesant, a famous high school for the intellectually gifted. For practical purposes, Stuyvesant scores will be concentrated in a narrow range—probably 1500 to 1600. The scores for the sample from an ordinary high school will vary from 400 to 1600. The denominator for the heritability ratio calculated from students at Stuyvesant will be smaller than the denominator from the sample from the ordinary high school. Other things equal, the heritability of SAT scores in the Stuyvesant sample will be higher than the heritability for the sample from the ordinary high school.[6]
Heritability can also vary over populations, or over the same population over time, for an important reason that is too seldom recognized: As society does a better job of enabling all of its citizens to realize their talents, the heritability of those talents will rise. It is a statistical necessity: The phenotype is the result of genes and environment. In a perfect world where everyone had completely full opportunity to realize their talents, heritability of those talents would converge on 100 percent because the environment relevant to those talents would no longer vary. If it doesn’t vary, it can’t explain anything.
Heritability rises even in an imperfect world. Consider educational attainment as measured by years of education. For the first half of the twentieth century, Norway was a country in which the amount of schooling you got depended strongly on where you lived (many remote places did not have secondary schools) and your family’s social class. In 1960, the average years of education for Norwegian adults was 5.9. After World War II, access to elementary and secondary school became nearly universal. By 2000, the average Norwegian adult had 11.9 years of education. Norwegian allele frequencies for the SNPs that are associated with years of education cannot have changed appreciably from 1960 to 2000. The absolute genetic contribution was effectively constant. But the heritability of educational attainment for Norwegian male twins born before 1940 was 40 percent. For their counterparts born after 1940, it was approximately 70 percent.7
Human Diversity Page 25