Sets of randomly generated numbers are used to produce simulated data sets to illustrate that high heritability of a property within each of two populations may be consistent with a vanishingly small heritability in the combined population. (A response by Jensen appears in the next issue of Journal of Educational Statistics, p.385). (CTM)

This paper focuses on the use of the variable race in educational research. Researchers are clearly considered to have the right to choose their variables. But, the use of race in a nonscholarly fashion is held to be professionally inadequate and often detrimental to black Americans. For years, researchers using race to make comparison s between…