The reason for this seemingly dotty result is that the errors from scientific sampling are easier to quantify (and minimise) than those arising from an imperfect head-count.

ECONOMIST: Statistics