The Teacher’s College Record, an education journal out of Columbia College for Teachers, just released its May volume (Vol. 106, Issue 6, pp. 1124-1144) online. In it they highlight the study by Jay Greene, Marcus Winters, and Greg Forster on the believability of high-stakes tests. In other words, do these tests correlate with school-level tests, or not?
Although the new report is only available through subscription to the journal, or on a fee-per-article basis, the foundational study by the same authors was published by the Manhattan Institute in February of 2003 (Civic Report 33).
According to Winters, the MI report uses the same numbers, and the conclusions are the same: “high and low stakes tests produce very similar score level results,” and “the stakes of the test do not distort information about the general level at which the students are performing.”
The significant thing about this study is that it used data from states that give both kinds of tests, allowing for reasonable high stakes-low/no stakes comparisons.