Don’t let anyone tell you that standardized tests are not accurate measures. The truth of the matter is they offer a remarkably precise method for gauging the size of the houses near the school where the test was administered. Every empirical investigation of this question has found that socioeconomic status (SES) in all its particulars accounts for an overwhelming proportion of the variance in test scores when different schools, towns, or states are compared. Thus ignorance would be the most charitable explanation for why charts are published that rank schools (or towns or states) by these scores -- or why anyone would use those rankings to draw conclusions about classroom quality.

However, if this were the only problem with standardized tests, we probably would not have sufficient reason to work for their elimination. After all, one could factor in SES in evaluating test results to determine the “true” score. And one could track a given school’s (or district’s) results over time; assuming no major demographic changes, a statistically significant shift in scores would then seem to be meaningful. But here’s the problem: even results corrected for SES are not very useful because the tests themselves are inherently flawed. This assessment is borne out by research finding a statistical association between high scores on standardized tests and relatively shallow thinking....