When English teacher Dawna Vanderpool returned to school last fall, she did what many teachers do: She pored over her eighth graders’ test results from the previous year, searching for clues about how much they had learned and what aspects of her teaching had or had not been effective. Instead of relying on a single benchmark to measure achievement, however—for instance, whether a student scored as “proficient” on a statewide language arts test—Vanderpool measured each student’s progress by comparing his or her actual test score to the score the student had been expected to receive, a prediction calculated on the basis of performance on annual tests in previous years.