New research raises issue with SAT predictability
- New research from a team that has been questioning the reliability of the SAT's predictive power for years finds that the test is inaccurate at some colleges across the country, affecting hundreds of thousands of students.
- Inside Higher Ed reports study authors analyzed data from 475,000 students at 176 colleges that was first used by the College Board to show the test does have predictive validity across students, but the authors found it doesn't have the same predictive validity across institutions.
- Because grading policies vary across institutions, the study finds the SAT math portion is an inaccurate predictor by gender at 16% of colleges and at 19% of colleges when comparing Latino students to whites; on the critical reading section, predictions didn't work when looking at black and white students' scores at 20% of colleges.
Lead author Herman Aguinis, of the Kelley School of Business at Indiana University at Bloomington, told Inside Higher Ed his team's research underscores a need for the College Board to release more information about its test so colleges can decide how important to make it in their admissions decisions.
Already, though, the College Board says it offers the free Admitted Class Evaluation Service that helps institutions think about the best combination of measures that will predict student performance on their campuses.
The National Center for Fair and Open Testing counts hundreds of colleges among those who do not require the SAT or ACT for undergraduate admissions. These tests have been criticized by civil rights activists as requiring an amount of cultural capital to prepare for and pass that leaves low-income and minority students at a disadvantage.
With the College Board on the verge of releasing its new SAT, administrators will have decide how prominent a place to give it in the admissions decision.
- Inside Higher Ed Faulty predictions?
Follow Tara García Mathewson on Twitter