Census Director Kenneth Prewitt officially launched the 2000 Census effort in January by riding a dog sled into Unalakleet, Alaska, where he was feted by residents of this tiny fishing village on the Bering Sea with a potluck dinner of moose liver and muktuk (whale skin) and entertained by enumeration-minded cheerleaders who chanted, "Census, census, count us first!" The next morning Prewitt obliged, knocking on the door of an aged Inuit couple and filling out the very first census form. It was a bright beginning for America's demographic Iditarod, but the singular clarity and good cheer of that moment may seem quite quaint by the time the government reports to America the final census outcome.
Clouding the horizon are two large questions, both centering on race:
First, the Census Bureau right now is on track to produce two separate and different counts of the American people. One will be the actual head count it makes of census day, April 1, the day in the life of the American people that the decennial census is supposed to record. The other population count will be adjusted, based upon the results of a statistical sample to be done over the summer, in order to correct for the usual undercount in the field, which especially tends to miss Asians, blacks, Hispanics, and American Indians. Because blacks and Hispanics--who in sheer numbers are the ones most undercounted--tend to be Democrats, Democrats almost universally see the wisdom in sampling and adjusting the totals, while Republicans, with very few exceptions (Senator John McCain was one of them), do not. But even if the Census Bureau, as now seems likely, produces both adjusted and unadjusted numbers, it remains to be seen whether the Supreme Court will permit adjusted numbers to be used for what, in the political realm, is their most important purpose--redistricting.
The second question, which has emerged as a sort of Rubik's Cube of race, is how to count, for a variety of governmental purposes, those people who will take advantage of a new freedom on this census to identify themselves as members not just of a single race, but of as many races as they see fit. The Census Bureau is expecting only about one in 50 people to check more than one box. But that does not change the reality that the 2000 Census is ushering in a new, more complicated racial order, one that may more faithfully reflect a changing America even as it may undermine the whole practice of counting by race.
For some, of course, that may be the whole idea. But for most in the civil rights community, the multiplication of possible racial classifications raises the specter of a new system so scattered and unwieldy that it no longer can be usefully applied to measuring America's racial gaps and remediating its disorders. As the census forms were being delivered to homes across America, the federal Office of Management and Budget (OMB) responded to these concerns. In March it issued a directive instructing federal agencies how they should crunch the new race numbers coming out of the 2000 Census for civil rights purposes: Collapse the many possible race combinations back to the standard few.
Where these two quite separate controversies--sampling and tabulating race--will join in mayhem is in the redistricting process. Barring definitive federal action, every state will have to separately choose whether to use adjusted or unadjusted population figures in drawing new congressional and other district lines. Then, for purposes where race matters, the states will have to decide whether to use the full range of categories the census will provide, or the collapsed categories that, under the OMB directive, the Justice Department will use to evaluate the race-fairness of their handiwork. Amidst all this confusion will be one serene certainty--no matter what choices they make, they will be sued.
Heightening the hysteria is the fact that the stakes here are enormous. At present, Republicans hold a bare six-seat advantage in the House of Representatives. And not since the 1950 redistricting have the two parties been so evenly matched in their share of power in the states.
In 1990 Republicans had total control of redistricting in only two multidistrict states--Utah and New Hampshire--for a grand total of five congressional districts. Today, of the 435 congressional districts, Republicans totally control the process in states with 113 districts, and Democrats have total control over 144 districts--numbers that could swing wildly depending on the outcome of legislative elections across America this fall. Republicans are also in better shape in most of the states that are gaining or losing seats through reapportionment. The 1999 Census estimates indicate that eight states will gain seats (two each in Arizona and Texas; one each in California, Colorado, Florida, Georgia, Montana, and Nevada) and that eight states will lose seats (two each in New York and Pennsylvania; one each in Connecticut, Illinois, Mississippi, Ohio, Oklahoma, and Wisconsin). Projections by Election Data Services suggest the same outcome except that Georgia would gain two seats and Montana none.
The biggest prize, California, should remain in Democratic hands, though there is still an outside chance a proposition will appear on the fall ballot seeking to remove redistricting from legislative control. In the meantime, Republicans think they can more than make up for any gerrymandered losses they may suffer in California if they are able to take control of the process in Texas and undo the Democrats' 1990 line-drawing legerdemain that helped keep 17 of the state's 30 congressional seats in Democratic hands to this day. Republicans now have a one-vote margin in Texas Senate seats and are within four seats of gaining control of the House.
Republican redistricters recognize that political control of the process is exponentially more important to the outcome than whether adjusted or unadjusted population figures are used. But with the balance of power so tick-tight, as Dan Rather might put it, even small advantages are being hyped to the heavens.
Thus, in a May 1997 memo, Republican National Chairman Jim Nicholson warned state party chairs, "The Clinton administration is implementing a radical new way of taking the next census that effectively will add nearly four-and-one-half million Democrats to the nation's population."
Underlying this rhetorical paroxysm is the following calculation: In 1990, according to the Census Bureau's estimates, 0.7 percent of whites were not counted, compared to 2.3 percent of Asians, 4.4 percent of blacks, 5 percent of Hispanics, and 12 percent of American Indians living on reservations. Most of the undercounted black and Hispanic people live in Democratic districts. If statistical methods were used to erase the undercount this time, those black and Hispanic numbers would swell the population in those districts and the overflow would wash into adjoining districts, making those other districts more Democratic.
Nicholson sounded the alarm based on a study by Republican redistricting consultant Clark Bensen warning that adjusted census numbers could hurt Republican chances in 24 congressional, 113 state senate, and 297 state house districts. "That's greatly overblown," says Jeff Wice, a leading Democratic redistricting guru. But Bensen says it will matter a lot even if he is only a tiny bit right. "The House is so close that any seat could flip control."
And yet, there are occasions when sampling could help Republicans. For example, a larger black or Hispanic count could trigger the creation of a new majority-minority district. That means taking a district that would have already elected a Democrat and concentrating enough minority voters there so it is virtually certain to elect a Democrat the color the district is designed for. Meanwhile, adjoining districts are bleached of minority Democrats and made riper for Republicans.
This is precisely what happened across the South in the last redistricting: Blacks and Republicans worked together, with a crucial assist from the Justice Department, to create new black districts. The result: In 1992, 12 new black members of Congress were elected from the South, even as in 1992 and 1994, for that reason among others, southern white Democrats suffered a net loss of 23 seats in Congress.
The prospects for new majority-black districts have about run dry, but much the same political dynamic that made allies of blacks and Republicans in the last redistricting will apply increasingly to Hispanics and Republicans. Arturo Vargas, executive director of the National Association of Latino Elected and Appointed Officials (NALEO), was deeply involved in the 1990 redistricting in California. "The Democrats were our adversaries," recalls Vargas, who was then with the Mexican American Legal Defense and Educational Fund. "The Republicans were falling over themselves to give us data and computers and advice. We said, 'No, thank you.'"
Both sides of the debate about sampling contend that the Supreme Court has already effectively decided in their favor. In a January 1999 decision, the Court ruled that sampling could not be used to arrive at the count that is used for reapportionment (which in common parlance refers to determining how many seats in Congress each of the states gets) but could be used for other purposes. Supporters of sampling read that to mean that an adjusted count could be used for drawing district lines within each state. But opponents of sampling contend that in Supreme Court-speak, "reapportionment" includes redistricting.
In any case, Republicans in a number of states have moved to prohibit the use of sampled numbers in their redistricting, and, sooner or later, the Supreme Court will settle the question. The question may also figure in the presidential campaign because a new president will take office before the adjusted figures are ready for release. Vice President Al Gore has indicated he would release the adjusted numbers. But Texas Governor George W. Bush, while expressing his opposition to sampling, has not been crystal clear about whether, as president, he would release adjusted numbers.
It is certainly an issue that Democrats would like to try to use against Bush. But even some staunch advocates of an adjusted census recognize it is difficult to put the sizzle in statistical sampling. "It's a very difficult issue to sell," says Larry Gonzalez, who directs NALEO's Washington office. "The only way you sell it is us versus them: They don't want us to be counted; they don't want us to have power. I'm not saying that a lot of that isn't true."
By contrast, the controversy over how to count those who choose more than one race on the census is at once more gut-grabbing and yet, so far, further removed from the national political debate.
In late 1997, capping four years of marathon meetings and deliberations, OMB brokered a compromise that changed the way Americans could identify by race. OMB rejected the demands of some for a separate multiracial category on the next census--an option that was anathema to many civil rights groups, which feared that a loss in the population count would diminish their strength and jeopardize their protection in law. Instead, individuals would be permitted to identify with more than one race.
The change was made in response to a simple and compelling human story. Children of the soaring numbers of interracial marriages should not have to choose either their father's or mother's race as their sole identity, to the exclusion of the other.
Martha Farnsworth Riche, who was census director at the time, says the change was a case study of the power of a letter-writing and lobbying campaign by a small but committed group of middle-class families. They enlisted some powerful allies, like then-Speaker Newt Gingrich, who became passionate in his support for a multiracial category as a step toward the elimination of all racial counting. And, coincidentally, at a key moment, along came Tiger Woods. Hot off his star-making Masters triumph in 1997, Woods announced on The Oprah Winfrey Show that he was not just black but "Cablinasian"--a mix of Caucasian, black, American Indian, and Asian. Riche says she knew at that moment change was inevitable.
But in permitting persons to check more than one race, OMB was also opening a Pandora's box.
In the 1990 Census, Americans could choose to be white; black; American Indian, Eskimo, and Aleut; Asian and Pacific Islander; or other. There was also, as there will still be, a separate ethnicity question in which people could identify as Hispanic, itself a bureaucratic designation invented for census purposes but one that has since emerged as a de facto racial category and, increasingly, a genuine marker of collective identity.
For the first time, the 2000 Census will count "Native Hawaiians and Other Pacific Islanders" as a separate category. There was a desire on the part of some native Hawaiians to leave the larger "Asian and" category and join the smaller indigenous category, "American Indian or Alaska Native" (its 2000 wording), but they were spurned there.
Add it up, and there will now be six discrete race categories--and 15 potential combinations of two races, 20 combinations of three races, 15 combinations of four races, six combinations of five races and, finally, one ultimate combo for those Americans who choose to be all that they can be. Altogether, that yields 63 different permutations, each of which can also be Hispanic or not, for a grand total of 126. (Interestingly, because the question of Hispanic identity is a separate, yes-or-no question, the one combination one cannot be is Hispanic and non-Hispanic.)
If 126 combinations may be exhilarating and, perhaps, more exactly correct than what came before, they are nightmarish for those charged with making sure that employers or schools do not discriminate on the basis of race or those who endeavor to make affirmative action programs work, for those who keep vital records and literally depend on the census for the common denominator they use in computing all their statistics, and, of course, for those who must draw legislative lines mindful of minority voting rights.
"The goal was to allow people to self-express, and that's fine," says Todd Cox, an attorney with the NAACP Legal Defense and Educational Fund (LDEF) in Washington. "The question now is, how do we take that information and make it useful for civil rights enforcement?"
With that in mind, the new OMB directive indicates how businesses and institutions that are required to report race data to the federal government should "aggregate" the data, and how federal agencies should "allocate" the multiple race responses for purposes of civil rights monitoring and enforcement. Instead of 63 categories of race, the aggregation rules would require reporting only the five single race categories (ignoring "other"), plus what they expect to be the four most popular two-way combinations (American Indian and white, Asian and white, black and white, and American Indian and black), and then any other combination within the relevant jurisdiction that totals more than 1 percent. For civil rights enforcement, federal agencies would allocate mixed-race individuals back to the minority race or, if there are two or more minority races, back to the race that makes sense in the particular enforcement context. (A whole other methodology is being developed for "bridging" new race data with old race data in order to be able to conduct trend or time-series analysis.)
The new rules were announced to a very small gathering of reporters in an OMB board room at 2:30 on a Friday afternoon, the time and day official Washington makes known those things it wants to make least known and expose to the least criticism. The multiracial activists were disappointed, if not surprised. The one-drop rule they despised was now, in important ways, back with a vengeance.
Susan Graham, the founder of Project RACE, which lobbied for the creation of a single multiracial category, listed her children as white on the 1990 Census because she was told, erroneously, by a census official in Atlanta, where she lived at the time, that the race of the mother was determinative. Graham, who now lives in Tallahassee, is white and her husband is black. But this year, if they identify their children as white and black, for many government purposes her children will be counted simply as black.
Curiously, even as the OMB decision was certainly a victory for LDEF and its allies, their reaction has been notably tentative. "We don't know what it will mean in the real world context," says Deepa Iyer, a staff attorney with the National Asian Pacific American Legal Consortium. Push come to shove, OMB's methodology may be rejected by others, challenged in court, or countermanded by a new administration less sympathetic to the civil rights community.
"We are entering into unchartered territory," says Roderick Harrison, the former head of the Census Bureau's race statistics branch, who is now at the Joint Center for Political and Economic Studies in Washington. He thinks the existence of two different race counts may undermine the confidence of a public "that already figures statistics are things people lie with. They may think, 'If it can be these two numbers, what else could it be?'"
The counting issue is difficult because it precisely locates the tenderest spot of American and, more particularly, liberal ambivalence on race. Up close, the individual's desire to break out of the straitjacket of racial categories is compelling. The whole notion of requiring people to officially identify themselves by race, or even races, is at some level noxious. And yet, whatever its origins, counting by race is the only way to understand and calibrate how race is in fact experienced, and to fashion methods of amelioration.
"We agreed upon the convenient fiction that we would classify people as if they were a single race because we saw a larger goal that would be addressed," says Joshua Goldstein, an associate professor affiliated with Princeton University's Office of Population Research. Goldstein's own analysis of census data suggests that the bureau may be underestimating how many people will choose more than one race. He suggests it may be two or three times the 2 percent of the population that the Census Bureau projects. And, of course, in some places, the numbers will be much higher. At the 1998 dress rehearsal census in Sacramento, California, 5.4 percent of the population--and in some parts of town 20 percent of the population--chose more than one race.
Goldstein says that the old racial classification system is coming undone not just in the face of rising intermarriage rates, but also in the context of a declining consensus about civil rights. "It's a combination of the changing political climate and the demographic change that is making it difficult to maintain this single-race classification system."
The National Research Council estimates that by 2050, a fifth of all Americans--36 percent of Asians and 45 percent of Hispanics--would be of multiple ancestry. And the more shaded the choices people have for identifying themselves racially, the more likely they are to try out different identities over the course of a lifetime. It may be that not very many people will check more than one box this census and that those who do will make the work of government, at least in the short run, very messy. But in the long run, this April 1 may eventually be remembered as a signal day in the melting away of race as a solid construct into a more fluid reality. ¤