Federal statistics saturate political debate about how we’re doing as a nation. Consider the function of the simple trend line—what trend is going down? What is going up? Imagine politics without trend lines, without knowing in what direction trade balances, terrorist threats, obesity, automobile wrecks, poverty, life expectancy, and inflation are headed. Remove trends from the daily news and not much is left. Although big data is making headway, for the present, trend lines draw heavily from federal statistics (surveys and administrative records). If this is true of trend lines, it is even more so when inquiring into causation. How did things get as they are? What has to happen to move the needle? Build a wall? Drop more bombs? Pass a law?

With this in mind, I turn to a feature very present in America’s trend lines—statistical races—which are produced when population counting applies a race classification, which has long been true of American counting. In the eighteenth century, natural scientists divided the world’s population into the five races of mankind—Europeans, Africans, Asians, Pacific Islanders, and Native Americans. In time, these categories came to structure the US census. Late in the twentieth century, Hispanics joined the census count as an ethnicity, with the understanding that they could be of any race. In the millennium census of 2000, the government finally, belatedly, took note that a man and woman of different races can have babies. “More than one race” became a census category.

The next census, 2020, has in view two more possible changes. First, there are proposals to merge ethnicity and race into one ethno-racial classification. This change would allow someone to identify as Hispanic and African-American Hispanic—Hispanic and another race or ethnicity is not available in any prior census. Second, and remarkably, there are proposals to introduce Middle Eastern/North African (MENA) as a separate statistical race. First there were the five races of mankind, then they became seven as Hispanic and Multi-race edged their way into the classification, and now the eighth, MENA, is proposed. There are consequences, notably several millions of Americans who were “White” in 2010 might now appear in the census as a new racial minority. This is a big statistical step; it hastens the legal and statistical recognition of an America in which whites are no longer in the majority, underscoring what many have identified as nativism at the heart of the campaign slogan: “Make America great again.”

For several decades now, race has been at the core of the strong and increasingly nasty politics of multiculturalism and identity assertion, which pits “real Americans” against those who are “taking our country away from us.” The pressure for the MENA change comes from a conversation steeped in identity politics, but the government response to this pressure moves us deeply into statistical politics.

Statistical races from the US census structure the trend lines noted above, which in turn drive political demands for policy and legal interventions—affirmative action, disparate impact, remedial justice—and fierce opposition to those policies and laws. Political argument may invoke race based on lived identity, but what finally and consequentially matters politically is the design and policy use of statistical races.

We now live in a world where familiar assertions like “the nation needs accurate statistics to produce workable public policy” carry little weight. What does carry weight is this question: “Whose country is it?” Note the text from a draft bill (H.R. 482) now making its way through the congressional maze: “Notwithstanding any other provision of law, no Federal funds may be used to design, build, maintain, utilize, or provide access to a Federal database of geospatial information on community racial disparities or disparities in access to affordable housing.” Not only does the proposed legislation roll back the sharing of open data about housing, but it seeks to stymie the use of those data and analytical methodologies that might tell us about inequality in our country based, in part, on racial statistical data.

I do not know if this text will become law. I do know that the sentiment it represents is a powerful current in today’s politics, where the impulse to use statistics to advance public policy is diminished, and replaced by an impulse to use categories to divide. This is what distinguishes prior changes to the census categories from the recent proposals and, sadly, why the nation’s statistics are politically vulnerable.