ABSTRACT
College admissions test scores are not designed, nor should they be, to be used as measures of school effectiveness. Designed to be relatively "curriculum proof," they measure verbal and quantitative reasoning abilities that develop over extended periods of time in the family, peer group, electronic media, and solitary reading and thinking environments, as well as in school. As measures of these broad reasoning abilities, they
are highly developed and reliable instruments for those who choose to take them. However, because individuals themselves select whether or not to take these tests, and because this self-selection process varies widely from state to state and across social-economic status and ethnic groups, the high school seniors who take the SAT tests (as well as the American College Testing tests) are a biased sample of all college-bound seniors, and changes in subgroup mean scores from year to year do not necessarily reflect ability changes in the entire college-bound population. Such mean score changes may reflect changing perceptions among different demographic groups of the availability of financial support and of benefits to be gained by attendance at more selective colleges, or changing admission requirements by large state university systems, as much as or more than they reflect changes in student ability. If trends in the developed ability of the group of interest are to be monitored, what is needed is a test administered to a random, not a self-selected, sample of college-bound seniors.
