Economist Thomas Sowell argues that single-factor explanations of disparities between individuals and groups (e.g., genetics, discrimination, or exploitation) fail to take life’s complexity or basic probability into account. Sowell reviews studies across numerous disciplines to build a case that misunderstanding the causes and acting on half-baked definitions of discrimination have often led to policies that harm the very people those policies were designed to help.

There are a lot of disparate outcomes among people, groups, institutions, and nations. This fact has led to the question of why these differences exist and to a variety of interpretations. Some have posited that differences are due to genetics; others favor discrimination of one group by another as the best explanation. Between genes and discrimination are a slew of other theories. What can be agreed upon is that the vast differences found in life do not converge with what random chance would have produced.  

Success is comprised of multiple prerequisites and missing several or even just one can lead to disparate outcomes. Take literacy as an example. Even in the mid-twentieth century, more than 40 percent of the world’s population was illiterate. Today, however, literacy is necessary for numerous—if not the majority—of careers. Someone might be an upstanding individual, with no criminal record, of good reputation, and possessing significant experience relative to a profession, but will ultimately fail if he can’t read.

Another instance of multiple prerequisites would be Terman’s IQ testing a century ago. Lewis Terman was a professor at Stanford University. He selected over 1,000 elementary-aged students and followed their development over several decades. Students were selected because they had IQs of over 140. These were the brightest of the bright.  Terman predicted that they would all have remarkably successful careers because of their intelligence, but the results were underwhelming. Many from this group had only moderately successful careers, and many others most would classify as failures.

Terman had to concede that intelligence was not the only prerequisite to consider. Family background mattered a great deal as well. Students from middle and upper class backgrounds usually had college-educated fathers and grew up in homes containing books. An impressive IQ alone, as noted, couldn’t guarantee success. Some people go to college because that’s what they see everyone else doing, or it’s simply what is expected. There are others, however, with great potential who have simply never had someone encourage them to make the most of their intellect. This is a simple, but sometimes unfulfilled prerequisite for many individuals.

The expectation that an even playing field would emerge if only bias or genetic differences were removed is naïve, but this fallacy, far from dying out, is perpetuated across the ideological spectrum. Lopsided distribution of outcomes is a common occurrence in the real world, which flies in the face of assumptions on the left and the right. Many people have a deeply-held, but deeply-flawed, understanding of probabilities. The dissonance between expectations of how things should pan out and how they usually pan out in the real world has given rise to ideologies and misguided political movements.

Single-variable explanations are part of the problem. Whether genetics at one extreme or discrimination on the other, such explanations are rarely satisfactory. There are other factors that influence success (sometimes profoundly), but they are often overlooked by the media and academics.

Geography is another factor that can contribute heavily to disparities. Historically, coastal areas have been more prosperous than landlocked areas, and groups living in river valleys have fared far better than those in the mountains. A casual glance at ancient history will reveal that prosperous civilizations arose in river valleys with seasonal rains, like the Indus River Valley in South Asia, the Tigris and Euphrates Rivers which ran through Mesopotamia, the Yellow River Valley in China, and the Nile in Ancient Egypt.

Things change quickly, too. Ancient civilizations began in river valleys with seasonal rains, but these original focal points of civilization are no longer the most influential regions.

The pattern of uneven distribution is everywhere: in groups, institutions, and nations. The first-born in families with five siblings are disproportionately National Merit Scholar Finalists. In fact, more finalists from five-sibling families are first-born than all others (i.e., second-born through fifth-born) combined. A similar pattern was found in the study of birth order in the United States, Britain, and Germany. Disproportionately, the average of IQ of first-born children was higher than second-born children, which, in turn was higher than the average IQ among third-born, and so on. Similar studies have revealed the same trend in Norway and the Netherlands. It seems that, more than nature or nurture, there’s also a question of number in the birth order.

Scotland’s history provides another example of diverse factors contributing to success. For centuries, Scotland was poor, backwards, and uneducated. So why were the Scottish so well-represented among the intellectual giants of the eighteenth and nineteenth centuries? Philosopher David Hume, economist Adam Smith, the chemist Joseph Black, poet Sir Walter Scott, and economic and political theorist John Stuart Mill—all Scottish or of Scottish descent. One of the significant changes was Protestant churches encouraging literacy so people could read the Bible for themselves instead of relying on Catholic priests to read and interpret it for them. Another factor was the growing desire to learn English, rather than just the native Gaelic. This opened up new political, economic, and educational opportunities to the Scots, particularly among the lowlanders, who were more willing to learn English than the highlanders.

Even in nature, patterns of uneven distribution are the rule rather than the exception. 90 percent of world’s tornado incidents occur in the United States. Thunderstorms occur 20 times more often in South Florida than along California’s coast. There are scores of peaks over 20,000 feet in Asia, but there are zero mountains that tall in Africa. There are stretches of Amazonian rivers the size of a tennis court that contain eight times the biodiversity found in all of Europe’s lakes and rivers.

What is discrimination? Do we mean the same thing when we talk about someone’s discriminating tastes in wines and art as we do when we talk about how someone’s behavior towards others varies arbitrarily, based on their group, without any regard for the qualities that an individual may or may not possess?

By both understandings of discrimination, there are significant differences in outcomes for things and people. Connoisseurs capable of identifying and selecting a quality bottle of chardonnay will consistently pick some vintages over others.  Throughout the world, differences in outcome among groups can also be pronounced, for example, in college admissions, unemployment rates, and occupations.

It is clear that not all discrimination is bad. It is also clear that some forms are, in fact, pernicious. It is possible to make appropriate and accurate distinctions between individuals and groups based on available empirical evidence, just as it is possible to make distinctions based on arbitrary hatred and unsubstantiated information.

In other words, the disparities in outcomes can be due to internal differences in behavior and competence or external differences. Let us also bear in mind that the answer is not the same for all disparities. There is a need for admitting complexity. Clearer definitions of discrimination will help us.

These questions must be settled by empirical data, not our feelings, good intentions, or sense of certitude—however zealous that may be. Neither Hitler’s conclusions in Mein Kampf nor Marx’s in Das Capital were arrived at through testing hypotheses; they were presumed to be true without much evidence, and the results of policies built upon fascist and Communist ideas were catastrophic. It’s important to define our terms. The word “discrimination” is no exception.

Towards nuancing the discussion of discrimination, let’s consider several categories:

The first type of discrimination—let’s call it Discrimination 1—will refer to discrimination broadly speaking, to the capacity of distinguishing traits of people or things. The second type—let’s call it Discrimination 2—will refer to the meaning typically intended: treating others poorly, due to illogical or capricious judgment of an individual’s qualities, like race, sex, and so on. It’s this latter variety that’s given rise to anti-discrimination laws and public policy.

Ideally, our decisions would be based on Discrimination 1, in which a person is able to judge another based on knowledge of that individual, regardless of the group from which he or she comes. Unfortunately, this ideal is hard to find where flesh-and-blood human beings are involved.

There are plenty of circumstances in which knowing someone as an individual before making a judgment simply is not possible, or at least the cost of judging incorrectly would be high. For example, if you’re walking down a street alone late at night in a poorly lit area, and you see someone in the shadows up ahead, do you judge that person as an individual or cross the street to avoid walking too close to a person you know nothing about? Maybe that person in the shadows is the kindest person in the neighborhood who harbors no malicious intent. But do you take such a chance? The cost of being wrong could be dangerous or even fatal.

A decision to cross the street to avoid the person in the shadows would not fall under Discrimination 1, but neither would it fall under Discrimination 2 (i.e., decisions founded on falsehood or hatred). We need a further distinction, let’s call them Discrimination 1a and Discrimination 1b, where Discrimination 1a is the most ideal (making a decision based on knowledge of the individual) and Discrimination 1b, (in which one makes a decision based on group evidence).

There is always a cost to be counted. That cost is not merely monetary. We can all agree that Discrimination 1 is the preferred approach, but there are contexts in which that is not advisable or possible.

For another example, let’s say there’s an employer looking to make a hire in a town comprised of two groups: Groups A and B. If rates of alcoholism are 40 percent among Group A and only one percent among Group B, an employer with a knowledge of the empirical evidence would likely pick someone from Group B. He doesn’t know the particular individual from Group A. Maybe the individual from Group A is a kind and responsible person with no history of alcoholism. But the cost of being wrong is high. It could endanger the other employees, the customer—if the products are faultily designed—and the business’ profits.

This example does not aim to justify or criticize the employer’s choice, but to show that different decision-making processes are required in different situations, and they should be assessed according to clearer classifications.

The situations above were hypothetical, but they illustrate the need for more granular categories of “discrimination.” Unfortunately, sloppier categories have had harmful implications in real-life situations for flesh-and-blood human beings.

Consider black employment and background checks. The U.S. Equal Employment Opportunity Commission has sued employers that do background checks for job applicants on grounds of racial discrimination. They’ve made such cases even when background checks are required of any applicant, regardless of his or her race. What is more, they’ve sued even though the empirical evidence clearly shows that employers who use background checks are more likely to hire black males.

The policy of illegalizing, or disincentivizing criminal background checks for all potential employees raised the cost of hiring young black males. Employers were forced to use Discrimination 1b, which relied on an understanding of group averages, and, on average, there is a higher likelihood of a young black male having a criminal record. Maintaining the background check policy would have allowed employers to use the more ideal Discrimination 1a, because the decision would have been based on more detailed knowledge of an individual applicant rather than group averages.

More background checks means more employment for young blacks—not less. But the empirical evidence doesn’t seem to affect those committed to a particular social vision, however contrary to the facts that vision may be.

There are also times when decisions are made not based on a group itself, but on its chemistry—or lack thereof—with other groups. In the 1800s, for example, immigrants sometimes brought their prejudices with them from Europe to the United States. Employers learned what would happen when Irish Protestants and Irish Catholics worked together—violence and reduced productivity.

Similar discriminations have been made when groups get along all too well. An employer may be happy to hire a competent man or woman, but for purposes of work efficiency, chooses to hire one or the other.  Nursing is a predominantly female profession, and male nurses can have difficulty finding work, because of the preference for females. Lumberjacks are usually men, and a man is more likely to be selected than a woman for the profession—even a woman who is just as competent as her fellow male applicant. The point here is that equal qualification is not the only consideration.

Without thoughtful distinctions between various kinds of discrimination, there’s greater chance of slandering someone’s attempt at making a sound, empirically-based judgment. There’s also a greater chance that policies based on inaccuracies will do more harm than good.

Misdiagnosing situations according to our idealized notions of what we’d like the world to be can lead to policies that harm more than they help. Policies are less likely to achieve their goals if those who design them don’t understand the given situation—no matter how well-intentioned the policies are.

Even in the most crime-ridden communities, the majority are law-abiding citizens. It is these law-abiding citizens who pay the price for the crimes of others in their neighborhood. The process of engaging in Discrimination 1 behavior, which evaluates an individual's merits, is ideal, but has its costs (e.g., time, and the risk of misjudging). Thus, a cruder Discrimination 1b approach is usually adopted, which relies on empirically-based generalizations about a given area or population.

The law-abiding citizens in high-crime areas pay a high price. Businesses are very reluctant to deliver goods and services to these areas for fear of bodily harm or even death to their drivers. Taxi drivers refuse to give rides to those areas. Businesses and chains avoid putting shops in those areas, which deprives good citizens of opportunities for employment. Costs are literally higher in monetary terms as well. The costs of everyday goods are more expensive because the costs of conducting business in such areas is higher. Rates of shoplifting, vandalism, and burglary are higher, which raises business insurance premiums.

It is true that “the poor pay more.” But a study by the same title concluded that the reason the poor pay more is exploitation by businesses. The government, the media, and academics picked up on this messaging. The authors of this study, however, made the mistake tragically common to those analyzing statistics: they assumed that by identifying a problem, that they had special knowledge regarding the causes of that problem. That information can’t be readily extrapolated from the statistics themselves.

But the people who list prices at a store are not the cause of higher prices. Those who are not familiar with economics are quick to denounce the higher prices as “price gouging” and “greed,” another case of Discrimination 2 that requires government involvement to rectify. The assumption that higher pricing means higher rates of profit is mistaken. In low-income, inner-city neighborhoods, profits are actually far below average, even though the goods are typically priced higher.

If rates of profit were higher, there would obviously be more businesses coming to low-income, inner-city areas. But the risks and costs of doing business in the inner-city are often prohibitive. The motivation for doing business elsewhere or making items more expensive is rarely based on malice or avarice, or discrimination against the poor or against minorities, but rather reducing those risks and costs. While it’s probably not much of a consolation to the majority of citizens in those communities who pay this high price, it’s critical to clarify the reasons for these disparities, so that policies will be built on economic facts rather than economic fallacies. A theme throughout this book is the painful disconnect between how people wish the world to be, and how it actually is. Good intentions are insufficient to bring about good results. By perpetually misattributing the causes of disparities to genetics, discrimination, or exploitation, existing problems will be worsened rather than ameliorated.