Panel:
Achievement Tests: On the Validity of Comparisons across Cohort, Grade, Subject, and Place
(Education)
*Names in bold indicate Presenter
This session presents new results on the use of achievement tests to compare performance in different cohorts, grades, subjects, and geographic populations of test takers. The first paper describes the common scale underlying the new Stanford Education Data Archive, and its use and limitations in cross-state and cross-district comparisons within the U.S. The second paper links achievement tests for 82 countries onto a common scale, and documents cross-national human capital variation, identifying several new facts about the global relationship between education, human capital, and income. The third paper links achievement tests across 50 cohorts of American test takers, showing credibly that gaps in achievement based on income, and trends in achievement, show little improvement. The fourth paper compares student growth measured by achievement tests across different tested grades and subjects, and shows that differences among tests are related to misalignment of the taught curriculum with tested content. All of these papers advance our knowledge of how to compare achievement testing results across diverse test instruments and populations - and provide cautionary tales for policymakers trying to make decisions or set incentives on the basis of tests that are comparable only with great care.