David Harris: Changing the melting pot, census categories to a 'matrix of race'

David Harris

By Franklin Crawford

Federal government forms now allow people to officially identify with up to six different racial groups -- a fundamental change that is designed for multiracial categorization. But the "mark one or more boxes" system has flaws -- most evident in the 2000 census -- that are now rising to the surface, according to Cornell professor of sociology David Harris.

In tabulating race data, multiracial self-reporting is often defaulted back into a single race category, he said in a talk at the American Association for the Advancement of Science in Washington, D.C., on Feb. 19. The most glaring example, he said, is with people who indicate they are both white and Native American: For some purposes, the census count automatically defaults to Native American, exaggerating the demographics of an entire population.

Harris, who also is director of the Institute for Social Sciences at Cornell, said the census approach to racial identification "raises difficult questions about how census data is used to assess the conditions of racial populations, as well as how it can be used to monitor and enforce civil rights laws." Instead, Harris proposed what he called a "matrix of race" that embraces the "multidimensional, socially constructed nature of racial classification."

The matrix of race identifies the most common bases for racial classification in three vertical columns: Genotype-Ancestry, Phenotype and Culture. These are intersected by three horizontal "perspectives" for racial classification: Internal, Expressed and External, creating nine "cells" of identification.

"Because race is a social construct, the allocation of individuals into racial categories is a social process," Harris explained. "As a result, a person whose expressed race is white and Asian might be treated as Asian by some people, but multiracial or white in contexts where alternative racial ideologies are operational. Similarly, people who express a strong white-black identity with friends might mark black on the census to indicate their political and cultural connection to blacks."

He observed: "Race is important, but the individual is the wrong unit of analysis. I am not attempting to impose a theory, I am arguing for the use of observations from the real world to inform how we collect and use racial data."

The 1997 revision of the Office of Management and Budget (OMB) Directive No. 15 (OMB15) provides up to six categories of race in the census and allows people to identify with more than one race (leaving more than 128 possible outcomes). It was issued, noted Harris, in response to mounting opposition from "people who objected to having to choose only one race on school registration and other governmental forms."