Do We Have to Count One by One?

摘要

During my tenure as an employee of the U.S. Census Bureau, I wrote computer programs to address uncertainties in the proposed statistical adjustment of the 1990 census. At one point, a friend's father, who was a physics professor, asked me to describe my work. I explained that the unadjusted decennial census implicitly assumed that the attempted headcount was equally accurate everywhere, whereas the proposed adjustment assumed that the attempted headcount was equally accurate within broad population subgroups called poststrata. He promptly responded, "Oh, so it's like a first-order correction."