CHAPTER 1

INTRODUCTION AND OVERVIEW

There was a racial classification scheme in America's first census(1790), as there was in the next twenty-two censuses, which brings usto the present. Though the classification was altered in response to thepolitical and intellectual fashions of the day, the underlying definitionof America's racial hierarchy never escaped its origins in the eighteenth-century.Even the enormous changing of the racial landscape in the civilrights era failed to challenge a dysfunctional classification, though it didbend it to new purposes. Nor has the demographic upheaval of our presenttime led to much fresh thinking about how to measure America. It is,finally, time to escape that past. Twenty-first-century statistics should notbe governed by race thinking that is two and a half centuries out of date.They poorly serve the nation, especially how it understands and managesthe color line and the nativity line—what separates us as races and whatseparates us as native born and foreign born.

What Are Statistical Races?

On April 1, 2010, the American population numbered more than 308million. When the Census Bureau finished with its decade populationcount it hurried to inform the president and the Congress how manyof those 308 million Americans resided in each of our fifty states. Thenation requires this basic fact to reapportion congressional seats andelectoral college votes, allowing America's representative democracy towork according to its constitutional design (see chapter 2).

Immediately after this most basic population fact was announced,the Census Bureau told us how many of the 308 million Americansbelonged to one of these five races: White, African American, AmericanIndian, Asian, Native Hawaiian. The bureau reported that a fewmillion Americans belonged to not just one of these five but to twoor more. Simultaneously, the bureau reported how many Americanswere Hispanics—which, the government insists, is not a race at all butan ethnic group. Incidentally, not all Hispanics got that message, becauseabout half of them filled in a census line allowing Americans tosay they belonged to "some other race." Hispanics, however, are not arace. Hispanics are expected to be Hispanics and also to self-identify asone or more of the five major race groups listed above (this is explainedin chapter 6).

What perhaps puzzles the reader is why race statistics are so terriblyimportant that they are publicly announced simultaneously withthe population figures mandated for reapportionment. You may also bepuzzled that the census form (fig. 1) dedicates so much of its space tothe race and Hispanic question but has no space for education, health,employment, or marital status questions. Are such matters less importantthan the country's racial profile? We will examine such puzzles. It isimportant that we do so because the race and Hispanic questions usedin the census have a very long reach. A version of these questions is usedin hundreds of government surveys—federal, state, and local—and inofficial administrative record keeping that captures traits of Americansfrom the moment of birth to their death: vital statistics, military records,and education and health data. Further, because the statistics resultingfrom a voracious appetite for information in our modern nation-stateare embedded in law, regulations, and policies, there are thousands ofprivate-sector institutions—universities, hospitals, corporations, voluntaryorganizations—doing business with the government that collectmatching race statistics.

America has statistical races. What they are, how we got them, how we usethem, and whether today we want or need them are questions that shapethis book. America's statistical races are not accidents of history. Theyhave been deliberately constructed and reconstructed by the government.They are tools of government, with political purposes and policyconsequences—more so even than the biological races of the nineteenthcentury or the socially constructed races from twentieth-century anthropologyor what are termed identity races in our current times. Whetherthese biological, socially constructed, or identity races are "real" is a seriousmatter, but they are of interest in this book only as they conditionwhat the government defines as our statistical races.

What, specifically, are statistical races? Organized counting of anykind—and certainly a census is organized counting—requires countersto know what they are counting, which in turn depends on a classificationscheme. Statistical races are by-products of the categories usedin the government's racial classification. And what do the actual 2010census categories produce? Though you cannot easily tell by looking atthe census form, the categories are designed to produce two statisticalethnicities and five statistical races. The ethnicities are Hispanic andNon Hispanic, though this is not evident from a question in which theterm ethnicity does not appear. The five statistical races are White, Black,American Indian, Asian, and Native Hawaiian/Pacific Islander, whichwe will learn in chapter 2 directly derive from a color-based division ofthe world's population by eighteenth-century natural scientists—white,black, red, yellow, and brown. With these basics established, the censusform then unleashes the combinations that result from the "mark oneor more" instruction. We will see later how these many combinationshave not, as yet, been used to make public policy. They became part ofthe census to fulfill expressive demands for recognition. Then there iswhatever appears on the "some other" line, though again these countsdo not become statistical races. So, whatever you think might be goingon in these census questions, the political and policy intent is to countthe Hispanics separately from everyone else, and to then sort everyAmerican, including Hispanics, into five primary races. When I usethe term statistical races, it refers to these five groups plus the Hispanicethnicity.

If you are now confused, you are on your way to understandingwhy I've written this book. That statistical races are real there is nodoubt. Law courts, legislatures, executive agencies, media, electioncampaigns, advocacy groups, corporate planners, university admissionoffices, hospitals, employment agencies, and others endlessly talk about"how many" African Americans, Asians, Pacific Islanders, American Indians,Whites, and Hispanics there are—and how fast their numbersare growing, how many have jobs, graduate from high school, are inprison, serve in the military, are obese or smoke, own their homes, ormarry each other.

If the statistical races are real and important, why does the censusform fail to make that clear? In fact, if you take a closer look at the questionsyou will be even more confused. You should be perplexed that onecensus-designated race—White—is simply a color. Nothing else is said.The next race is a color, Black (and Negro, which is another way to sayBlack), but also a descent group—that is, Americans whose ancestorsare from the African continent—and in some respects an "ethnicity" aswell. Today's immigrants from Ghana or Ethiopia also go into that category.Then color drops out of the picture altogether. A civil status enters.American Indians/Alaska Natives belong to a race by virtue of tribalmembership, which has a clear definition in American law; they can alsobelong to that race by declaring membership in a principal tribe, whichis not a legal status but a self-identification. Look at the census questionagain. With Whites, Blacks, and Native Americans now listed, there followsa long list of nationality groups. If we read the question stem literally,each of these is a race. The Chinese, the Koreans, the Samoans arepresented as if they are independent races. We are not, however, supposedto understand the question literally, but to understand that webecome part of the Asian race or the Native Hawaiian/Pacific Islanderrace by checking a national origin or writing one in. Oddly, however, theterm Asian only incidentally appears in the question, defining personsfrom India in a way that doesn't confuse them with American Indians,and inviting write-in responses, where the examples listed are again nationalitygroups.

With this nationality nomenclature in mind you might look backat the question on Hispanic/Latino/Spanish origin (terms used interchangeably),where you will see that it is similarly constructed. There isno box indicating Hispanic, but several boxes labeled with nationalitiesand a write-in space again guided by nationality examples. In a nationfamous for its ethnic diversity, you might now be asking, is the censustelling us that there are only two ethnic groups that matter (Hispanicand non-Hispanic) while ignoring all white European national origingroups (Swedes, Germans, Italians, Irish, Poles, and Russians, amongothers)? It seems so.

It's hard to find the underlying rationale for what appears, and whatdoesn't, in these two ethnoracial questions. We will discuss in detail theabsence of a coherent rationale. As an astute scholar has written, theCensus Bureau "has no choice but to rely on incoherent categories ifit hopes to measure race in the United States" because, he continues,"race arises out of (fundamentally irrational) social practices." A largepart of the story told in the chapters to follow explains how "incoherentcategories" result in incoherent statistical races, which derive not onlyfrom social practices but equally from policy goals.

It matters if America measures races, and then, of course, how thegovernment decides what those races are. It matters because law andpolicy are not about an abstraction called race but are about races as theyare made intelligible and acquire their numerical size in our statistical system.When we politically ask why black men are jailed at extra ordinarilyhigh rates, whether undocumented Mexican laborers are taking jobsaway from working-class whites, or whether Asians have become themodel minority in America, we start from a count of jailed blacks, thecomparative employment patterns of Mexicans and whites, and Asianeducational achievements. When our political questions are shaped byhow many of which races are doing what, and when policies addressingthose conditions follow, we should worry about whether the "howmany" and the "which races" tell us what we need to know about whatis going on in our polity, economy, and society. We should worry aboutwhether we should have statistical races at all, and if so, whether we havethe right ones. My answer, worked out in chapter 11, argues for incrementallytransforming our racial statistics in order to match them withthe governing challenges of the twenty-first century. This argument, andthe tactical advice offered to realize it, makes sense only in the contextof a historical account of statistical races.

Chapter 2 starts with basics that frame this American history. A Germandoctor in 1776 divided the human species into five races. Today,nearly two and a half centuries later, these are the same five races intowhich the U.S. Census divides the American population, making Americathe only country in the world firmly wedded to an eighteenth-centuryracial taxonomy. Embedded in this science were theories of a racial hierarchy:there were not just different races but superior and inferior races.American politics and policy held onto this assumption for nearly twocenturies.

The next section covers the nineteenth century, showing how assumptionsof racial superiority and inferiority tightly bound together statisticalraces, social science, and public policy.

Policy, Statistics, and Science Join Forces

The starting point—as is true of many features of American government—takesus to constitutional language (chapter 3). The U.S. Constitutionrequired a census of the white, the black, and the red races. The foundersfaced an extra ordinary challenge—how to join the original thirteencolonies into a republic of "united states." They met this challenge witha political compromise that brought slaveholding states into the Union.Without this statistical compromise there would not have been a UnitedStates as we know it today. In the early censuses slaves were countedas three-fifths of a person, a ratio demanded by slaveholder interests asthe price of joining the Union. Holding their noses, the northern statesagreed. A deep policy disagreement at the moment of founding the nationwas resolved in the deliberate creation of a statistical race. In thiscase, the policy need shaped the statistical practice.

Later in American history the reverse frequently occurred. Specificpolicies—affirmative action, for example—took the shape they did becausethe statistical races were already at hand. One of my major arguments,especially starting with chapter 8, is that we should learn a lesson fromthe founding period: start with agreement on public purposes and thendesign suitable statistics to meet policy challenges. Without clarityon why the nation should measure race, clarity on what to measure isimpossible.

The political understanding in the nineteenth century that countingthe population by race could do nationally significant policy work lednaturally to a close partnership between race science and census statistics,setting the stage for what 150 years later we call evidence-based policy.It's a fascinating if also depressing story, resting as it does on the nearuniversal assumption that there is a biologically determined racial hierarchy:whites at the top, blacks at the bottom, with the yellow, red, andbrown races arrayed between. Chapter 4 tells the race science story, givingemphasis to features that mark American history to the present day.Among the more important was the shift from simply counting races, aswas needed to make the three-fifths policy work, to investigating characteristicsconsidered unique to different races. The policy goal was to determinewho was fit for citizenship responsibilities: whites, certainly; theAmerican Indian, probably not; the African, clearly not. The statisticalraces helped fix the color line in American politics, essentially drawingpolicy boundaries that gradually governed all aspects of life: schooling,housing, employment, marriage, travel, and—of course—politicalparticipation.

Drawing internal policy boundaries became more pressing when theCivil War ended slavery and presented the specter of four million freeblacks in the society. Starting in the 1880s, the Jim Crow racial ordersustained the color line initially put in place for slaveholders. Yet morecomplicated boundaries were drawn with the arrival of Chinese labor tothe mines and fields of the western states in the mid-nineteenth century,and later in the century when massive flows of immigrants from southernand central Europe arrived as labor for the eastern states' factories.Chapter 5 covers how America continually readjusted its color line whenthe economy's need for workers resulted in immigration-driven populationgrowth but the polity required a monopoly of power in the handsof the "right" whites—that is, European Protestants.

In the stressful half century starting with the Civil War, the social sciencesentered the scene. Their methods and theories joined with the greatresource of national census statistics lent the authority of a new scienceto the policy choices of the day, a social science that gradually displacedthe biologically based race science popular earlier in the nineteenth century.In ways familiar to us today, the social sciences embraced the statisticalraces as a key to informing policy makers across a broad range ofissues including, early in the twentieth century, stopping the flow of the"wrong" European immigrants—Catholic and Jewish—without interferingwith immigration from the ancestral Protestant Europe. Statisticalraces at the time reflected assumptions of a color line but also of a nativityline, that drawn between the native born and the foreign born.

When You Have a Hammer,Everything Looks Like a Nail

The nation's statistical races were four as the twentieth century arrived:European White, African Black, American Indian Red, and Asian Yellow.They were put to policy work in restricting immigration in the 1920sand for racial segregation more generally until the civil rights challengedramatically arrived in the 1960s. The simple version of what then happened,recounted in chapter 6, is how a policy instrument used to politically,economically, and socially exclude since the nation's founding madea 180-degree turn and was used to include the racial groups historicallysent to the back of the bus—both literally and figuratively. It is a story ofhow proactive policies of racial justice were shaped with racial statisticsnever intended for the policy uses to which they were put. But the fitmade sense. The policy tool—statistical races—could be shaped to matchthe policy goal of racial justice. It also made sense given the greater effectivenessof the Census Bureau at measuring population characteristics,and the availability of the social sciences to advise policy design atlevels of detail previously unimagined.