Measuring Up: What Educational Testing Really Tells Us

by Daniel Koretz

A pure matrix-sampling design does not provide useful scores for individual students because students take different, and therefore not comparable, subtests. Hence the controversy.

.c2