Math A Test and Standards-based Tests

Following my work on the NY Regents Math A Panel last summer, which examined the causes of the high failure rate on the June 2003 Math A test, I undertook a study of the psychometric methodology, based on Item Response Theory, that New York State Education Department used on the Math A test. The Math A Panel had access to a vast array of test data and, as I sifted through it, patterns emerged of serious shortcomings in the methodology for maintaining a constant performance standard. These problems will probably arise in many other state standards-based tests, calling into question the validity of the year-to-year comparisons of student performance in mathematics, mandated by the No Child Left Behind Act.

Click here for a short version of the findings
Click here for a long version of the findings