Racist attitudes of whites towards blacks have long become socially unacceptable in America, although the reverse, racism of a minority directed at the white majority, is still tolerated or even encouraged. However, statistical racial disparities persist. African Americans, as a population, continue to suffer income, crime and incarceration rate, health, housing and family-structure deficits by comparison with the white population.
These disparities cannot easily be attributed to racist behavior by whites. The disparities have either increased or remained the same while individual racist behavior has declined. What then is the cause of these disparities? There are two possibilities: causes within individuals, what I have elsewhere called endogenous causes; or external, exogenous causes.
Endogenous Causes of Black-White Disparities
Endogenous causes were in fact the first ones to be studied, with unfortunate results. Bigots stigmatized the entire “black race” as inferior because of lower average scores on, for example, IQ tests. Blacks’ under-performance in terms of status, health, incomes, etc. was then comfortably attributed to their alleged built-in inadequacy.
The usual presumption was that IQ is fixed at birth, that it is the most important factor in life success and that it cannot be altered by later experience. None of these is true; although the fixity-of-IQ view seemed to be supported by several studies showing relatively high (statistical) heritability for IQ. But heritable is not the same as fixed: high statistical heritability for a behavioral trait does not imply that it is fixed at birth and independent of the rearing environment. Language is the most obvious counter-example. It is a learned behavior that also has high heritability. Language is 100 percent learned and 100 percent heritable—kids learn the language of their parents.
The only reason we know that language is not a sort of instinct is the “natural experiments” provided by adoption. Despite the high heritability of language, adopted infants learn the language of their adoptive, not biological, parents. It follows that high heritability does not mean genetic determinism. Statistical heritability depends on rearing environment as well as genetics.
What kind of experiment would be needed to prove that intelligence, which is also (statistically) heritable, is in fact genetically determined? What would it take to show that there are irreducible average-IQ differences between races: that no matter how rich the environment, blacks and whites would still have differing average IQ, leaving genes as the only cause? Only a very elaborate, unethical, and in practice un-doable, experiment could do it.
We lack now, and for the foreseeable future, a detailed understanding of how the human genotype produces the human brain. Nor do we know exactly how the brain works to produce the human mind and behavior. We cannot therefore map out step-by-step, in detail, how genes-build-the-brain-makes-behavior. To understand the genetic determination of IQ we would be forced to resort to experiments with whole human infants. To address the black-white issue in a completely rigorous way, we would need to begin with two fetal genotypes, one “black” and one “white.” Next, we would need an indefinite number of identical copies, clones, of each genotype. Then, each clone would be exposed to a different rearing environment. (Have we tried out the full range—all relevant environments? How would we know?) Then, perhaps 16 years later, we would give IQ tests to each child. We would get a distribution of IQs for each genotype. We could then ask: how do these distributions differ? Are their means the same or different? Which is higher? Is one distribution more variable than the other?
Suppose we find a mean difference. Does that settle the issue? Well, no, not yet. We are talking about race differences here, not differences between individual genotypes. A race is a set, a range, of genotypes. So, we need to repeat this impossible process with an indefinitely large sample of “white” and “black” genotypes (there would, of course, be debate about which genotypes should go into which group). Only after we had this two-dimensional array of genotypes vs. IQ could we compare them and come up with a valid conclusion about race difference in IQ.
An elaborate study of this sort is of course impossible. But what about adoption? Adopted children learn the language of their adoptive parents. Do adopted children acquire the IQ of their adoptive parents? There have been several studies along these lines. A landmark 1997 study (Plomin et al.) looked at 245 adoptions within a mostly white population. Their conclusion was unequivocal:
Children increasingly resemble their parents in cognitive abilities from infancy through adolescence. Results obtained from a 20-year longitudinal adoption study of 245 adopted children and their biological and adoptive parents, as well as 245 matched nonadoptive (control) parents and offspring, show that this increasing resemblance is due to genetic factors. Adopted children resemble their adoptive parents slightly in early childhood but not at all in middle childhood or adolescence.1
The convergence of IQ between biological parents and adopted children as the children age is also true of siblings, who grow more similar to their parents in many psychological measures well into adulthood: the (statistical) heritability of IQ increases substantially with age. These results are discouraging for advocates of “nurture” as the exclusive source of racial IQ differences.
Plomin et al.’s study did not compare white and black kids. It is quite possible, therefore, that black kids will show less influence of parental IQ than white kids do. In which case trans-racial adoptions, black kids adopted by white parents, might well show more similarities with their adoptive parents than Plomin’s white kids did. In a delicately worded sentence, the Minnesota Transracial Adoption Study follow-up concludes: “Putative genetic racial differences do not account for a major portion of the IQ performance difference between racial groups.” Or, stated more directly: racial differences account for less than half the black-white IQ difference. Clearly the role of genetics on IQ is not negligible, as it seems to be for language learning; but neither is it the dominant factor.
The complicating issue is that statistical heritability depends on the environment. In a “rich” environment, one which allows every genotype its full expression, differences that remain will tend to reflect genetic rather than environmental effects. This may be the reason that heritability increases with age. In an impoverished environment, on the other hand, individual development will be stunted, and individual differences will likely more reflect chance. An infant raised without ever hearing speech will not learn to speak—even though, under normal conditions, language is almost 100 percent heritable.
We don’t understand how genes affect the development of the individual brain and we cannot do the kind of experiments necessary to get a definitive answer to the genetic black-white-IQ-difference question. Adoption “natural experiments” are necessarily flawed: neither adoptees nor adoptive parents are randomly selected; age at adoption cannot be controlled—some children are adopted as infants, others as pre-teens or older. The impossibility of scientific proof has left the question wide open to other influences, as we will see.
It is unfortunate that the political climate in the US has for many years been strongly opposed to even the possibility that behavioral traits are in any way pre-determined. Best-sellers like Malcolm Gladwell’s Outliers (2008) claimed that persistent hard work—the “10,000-hour rule”—will propel anyone to success, with the implication that there is no such thing as natural talent. We could all compose like Mozart if we just practiced long enough.2
The playing field of the nature-nurture debate has not been, and is not now, level. Honest writers on this topic have been attacked and their employment threatened by unfair attacks on their purely academic, non-ideological writing. Moderate, evidence-based views—IQ is a product of nature as much as nurture, “racial” groups differ in IQ—are caricatured as racist extremes. The view that IQ is genetically fixed—like an instinct—is in fact held by none of the leading researchers in this field. Yet it has been used to stigmatize distinguished scholars such as Linda Gottfredsen and Charles Murray.
Philosopher Michael Levin was called an unabashed white supremacist following the publication of his 1997 book Why Race Matters.3 His crime was to take too seriously the fact that blacks and whites as groups differ in terms of IQ and possibly other socially relevant psychological measures and that these differences should be taken into account in evaluating racial disparities. Lurid links were concocted by well-known opinion writers between those who dared to discuss the heritability of IQ and coercive eugenics laws, not to mention ‘white supremacy’ and Nazi genocide. Data and arguments were buried under a hailstorm of ad hominem attacks.
Fear of treatment like this from the Southern Poverty Law Center (SPLC) and similar groups with a political agenda and no interest in understanding the complex issues involved has, in effect, shut down research on endogenous racial differences as a cause of racial disparities. We still do not understand the causes of individual differences in IQ and other racially differentiated personality characteristics. What is the role of rearing environment and social context? Of genetical endowment? How do these individual differences interact with the social environment to bring about the disparities that so dismay many social scientists?
There is an average-IQ difference between white and black Americans. But research has not yet established this IQ difference as a cause, or even a partial cause, for real-world black-white disparities. Nor will it, given the rancor and disinformation that surrounds the question.
Exogenous Causes of Racial Disparities
With endogenous causes rendered taboo, all that is left to account for racial disparities is exogenous causes, especially racial discrimination. But everyone admits that discrimination by individuals has decreased in recent decades. The picture below shows the results of a University of Illinois survey which concluded that, “One of the most substantial changes in white racial attitudes has been the movement from very substantial opposition to the principle of racial equality to one of almost universal support.”
The absence of discriminatory attitudes doesn’t mean the playing field has been completely leveled, of course. A past bias, once justified, may persist long after the world has changed. Robert Moses’s racist-inspired building practices in New York leave their mark today. It’s not hard to point to lingering effects of past racism on parents, children, and neighborhoods.
These effects act in complex ways. Proving their existence, much less measuring them with precision, is almost impossible. Many effects are delayed, different individuals vary in their reactions, and allegedly discriminatory actions are often unintentional.
The problem is that these complex and little-understood effects have been bundled together into one toxic package, labeled “systemic racism” (the terms institutional and structural racism are also used).
The word racism is thrown around a lot these days, but a precise definition is hard to find. The Stanford Encyclopedia of Philosophy passes. Wikipedia comes up with: “Racism is the belief in the superiority of one race over another…”
Other definitions add reasons for racism. Particularly popular is the idea that “inferior” races have unchangeable (“genetically determined”) attributes that render them less able than the favored (usually white) race. But a definition of racism doesn’t have to make reference to the cause. Basically, racism is intentional discrimination against a race for no reason at all: You just think white people are better than black people and so whites and blacks should be treated differently. Preference by itself—you prefer to spend time with members of your own race, say—is not ipso facto racist. Only if you think that a person should be treated differently just because of her race, does it qualify as racist.
By none of these definitions can the complex, lagged effects of past discrimination be termed (systemically) racist. The effects I will discuss are not intentional—if they were, they would be individual racism, not systemic. They are not based directly on race, for the most part. If a black child fails to get into a good college because she scored poorly on the SAT, she fails because of her score not her race—even if her low score might be traced to poor rearing conditions that are a legacy of segregation or past discrimination.
Effects like this are, possibly, effects of racism in the past. They are not racism now, systemic or otherwise, by any reasonable definition. Failure to acknowledge this distinction has unjustly stigmatized white people and is a cause of needless conflict. Let me give some examples of the problems with this pernicious concept.
A 2017 review paper in the respected medical journal The Lancet, authored by several public health researchers, looks at the health implications of what they term “structural racism.” They refer to the “rich social science literature conceptualizing structural racism” emphasizing that the idea goes beyond “unfair treatment as experienced by individuals.” Yet in the next paragraph they say, “Any account of structural racism within the USA must start with the experiences of black people…” This inconsistency is never resolved: are the individual experiences of racism by African Americans relevant or not? Apparently not, as the paper goes on to discuss racial disparities as evidence of structural racism.
One example is this: “The legacy of these [ostensibly race-neutral] policies is that the annual rate of incarceration of black men is 3.8–10.5 times greater than that of white men, across all age groups…” which is obviously unfair, hence racist. Unfair—unless the rate of offending is also skewed. Men notoriously commit more crimes, especially violent crimes, than women. We do not cry “sexism” when more males than females wind up in prison. Is disparity in incarceration rates between black and white men another example of racism? Or do black men in fact commit proportionately more crimes than white?
The SPLC has weighed in on this issue, citing a relevant report from the U.S. Bureau of Justice, “Race and Hispanic Origin of Victims and Offenders, 2012-15.” They headline the report: “White supremacists’ favorite myths about black crime rates take another hit from BJS study: Vast majority of most crimes are committed by a person of the same race as the victim, Bureau of Justice Statistics reports.” It goes on to say that “White supremacists… claim that … African-Americans, are far more crime-prone and the source of most violent crime against whites.”
Either willfully or because the writer doesn’t understand the issue, the headline misses the point of the BJS report. First, the U.S. population is 65 percent white and 12 percent black. It is likely, therefore, that more crimes can be attributed to whites than to blacks. No surprise there. And it has been known for many years that most violent crimes are intra, not inter-racial. Blacks are the victims largely of black criminals, whites of white.
The real issue is “crime-proneness,” which depends both on the number of black and white perpetrators and on the sizes of the black and white populations. Population size is not mentioned in the SPLC article. There are fewer blacks than whites in the U.S., therefore we can expect fewer black than white perpetrators. But how much fewer? Well, if we include population figures (which also appear in the BJS report), we see that there are 5.3 times as many whites as blacks in the US. So, the real question is, are there 5.3 times as many white as black criminals? Well, no. From Table 1 in the BJS report, 43.8 percent of perpetrators are white and 22.7 percent are black, so the ratio of white to black criminals is just 1.93. In other words, blacks are 5.3/1.93 = 2.79 times as likely to be perpetrators as whites. A disparity in criminality may have something to do with the incarceration disparity, although it may not be the whole story. But the differential incarceration rate by itself proves nothing.
Again, these data need to be unpacked to understand what is really going on. Young males are more likely to act violently than older ones. The black population tends to be younger than the white. Does controlling for age reduce the black-white disparity? No doubt other relevant variables should be examined. The point is that the incarceration disparity may have a non-racial cause. Absent a lot more research, it should not be blamed immediately on racism. Which is not to take away from cogent criticisms of the excessive US incarceration rate in general. The point is that incarceration-rate disparities are not necessarily evidence of racism. This is an example of what Quillette contributor Coleman Hughes calls the disparity fallacy, which “holds that unequal outcomes between two groups must be caused primarily by discrimination…” Indeed, I will argue the opposite: that a charge of “racial discrimination” can be justified only when other possible causes of disparities have been eliminated.
The BJS report deals with all violent crime, a measure that is imprecise for several reasons. The definition of “violent” varies from one person to another, there may be differences in reporting between black and white communities—blacks may be more (or less) reluctant to report crimes than whites, for example. Murder statistics are necessarily more certain than reports of non-lethal crimes. Here, the data are unequivocal. The Centers for Disease Control (CDC) in a 2017 report concludes: “the homicide rate among African-Americans is nearly quadruple that of the national average” and some eight times the rate for whites.
So, there is some non-racial basis for the apparently “racist” fact that blacks are incarcerated at a higher proportion than whites. Blacks are also more likely to commit violent crimes, and much more likely to commit murder, than whites. Of course, the details of the causality—in particular, the possible links between black criminality and racism suffered in the past—are not at all understood. But what should be understood is that this disparity is not necessarily evidence for contemporary racism. Much the same argument applies to racial profiling: racial disparities in police “stop-and-search,” for example.
Incarceration rate is a clear case where a racial disparity can be traced to a non-racial cause. The Lancet article goes on to list other examples of systemic/structural racism:
One key example, with ongoing intergenerational effects, is the historic Social Security Act of 1935, which created an important system of employment-based old-age insurance and unemployment compensation. The Act also, however, deliberately excluded agricultural workers and domestic servants—occupations largely held by black men and women.
The authors attribute the act to an attempt “to secure the votes of Democrats in the South.” Redistricting to restrict black voter participation aka “racial gerrymandering,” is often cited as another example of systemic racism. But in both these cases the aim is to secure votes not exclude black people because they are black. If the African-American population changed its preferences from Democrat to Republican, no doubt gerrymandering practices would adapt accordingly. The point is that these practices are not aimed at black people because they are black, but because of the way they vote. Gerrymandering is sleazy, but it is not racist—systemic or otherwise.
Another example that seems clearly racist is hiring by private companies. The Lancet article quotes “one study that used identical résumés, which differed only in the name of the applicant, hiring managers called back those with traditionally white names (eg, Brad or Emily) 50% more often than those with traditionally black names (eg, Jamal or Lakisha).” Is this racism? We don’t know the hirer’s previous experience. Perhaps she is racist; but in any case, this is individual not systemic racism.
This kind of muddle infuses the Lancet article and many others like it. But the fact remains that there are certainly lingering effects of past racism. How should they be understood? Labeling them as systemic racism is not helpful, because it suggests a single cause for what are in fact multiple interacting causes, most of them non-racial.
One solution has been to invoke systems theory as a way to model a complex situation. Systems theory cannot be accused of excessive precision. A system is “an organized entity made up of interrelated and interdependent parts.” The basic idea is that no part can be considered in isolation because all are connected. An action on one part will have effects on many others. Systems thinking has been successfully applied to collective phenomena such as flocking in birds, schooling in fish and the building of termite mounds, explaining coordinated behavior by leaderless groups in terms of simple rules followed by each individual group member. In this case the emergent, coordinated behavior can be measured precisely, and hypothetical rules can be rigorously tested by computer simulation.
But systems theory has been invoked much more widely, in areas from political economy to social work. A sociology review paper begins with this: “An example of an emergent property is wetness. Neither hydrogen nor oxygen alone has or can produce wetness; wetness occurs only in a chemical system that includes both hydrogen and oxygen in the correct proportions.” The problem is that the sociological equivalents of the elements hydrogen and oxygen can neither be accurately identified nor independently manipulated.
A paper that seems to have influenced contemporary thinking in sociology is Thomas Schelling’s 1971 “Dynamic Models of Segregation,” which has as one conclusion: “In some cases small incentives, almost imperceptible differentials, can lead to strikingly polarized results.” This is an early version of the “tipping point.” This is an idea that appeals to students of racism because it allows the possibility that even small vestiges of racism may be amplified by “race discrimination system” interactions to produce what Barbara Reskin has called über racism. It’s certainly possible: chaos theory has given us the butterfly effect, after all. But large effects from small causes, or even from large but mostly unrelated, non-racial, causes, are the exception rather than rule. Über racism requires much more convincing proof than the kind of qualitative, even impressionistic, accounts offered by researchers in this area.
There are two other problems with this approach. The first afflicts all of sociology. There is much discussion of “causal models.” But since experiment is generally impossible, all that is available are correlations. If these correlations are sufficiently compelling, perhaps causality can be inferred, but in fact they rarely are. The second problem is that in the racism area there are almost no quantitative correlations at all. What is offered are not so much models as visual metaphors. The picture above, labelled “The race discrimination system and emergent discrimination,” is a case in point. No doubt other researchers have other diagrams. This diagram paints a plausible but completely untestable, hence unscientific, “proof” of how historical racism and its contemporary after-effects might work together to produce disparate racial outcomes. Yet it is used to justify the concept of systemic racism.
Systemic racism is a poor concept. First, it is almost impossible to prove, because racism is discrimination without any reason other than race. To prove discrimination, all other possible reasons—reasons like differential ability, interests, criminality, etc., as in the examples I gave earlier—must be eliminated. Does the tech industry discriminate against women? Does the nursing profession discriminate against men? To show racism, which is differential treatment for no reason other than race, alternative explanations for disparities must be eliminated. But in practice not only are they not eliminated, efforts to explore these other causes are actively suppressed.
So, the second, and perhaps most important, problem with the charge of systemic discrimination is that it deflects attention from the proximal causes, endogenous as well as exogenous, of the racial disparities that led to its invention. Disparities—racial, ethnic, or gender-based—are not proof of anything. Disparities raise questions about their cause. Absent further information, a racial disparity does not favor one answer over others. To say, as some academic critics have, that “When I See Racial Disparities, I See Racism” is simply wrong. If only things were that simple!
The beauty of “systemic racism” is its air of permanence. It is here forever, and its victims must be compensated in perpetuity. It has become the elusive and inexpugnable cause of all the ills of people of color. And it provides an endless supply of ammunition for those whose careers depend on the persistence of racism. It has become a cause of racial division rather than part of the cure. It should be abandoned.
1 Robert Plomin, David W. Fulker, Robin Corley and John C. DeFries (1997) Nature, Nurture, and Cognitive Development from 1 to 16 Years: A Parent-Offspring Adoption Study. Psychological Science, Vol. 8, No. 6 (Nov., 1997), pp. 442-447
2 No, actually: Mosing, M. et al. (2014) Practice does not make perfect: no causal effect of music practice on music ability. Psychological Science, 25(9) 1795-1803.
3 The initial print run for the book was small and soon sold out. The author was savagely criticized, so the publisher apparently declined to print any more. The book was re-published in 2005 by a company that has itself been attacked as racist, rendering the project tainted.
Join the newsletter to receive the latest updates in your inbox.