ERIC Number: ED563440
Record Type: Non-Journal
Publication Date: 2011-Apr-10
Abstractor: As Provided
Reference Count: 5
A Case for Transforming the Criterion of a Predictive Validity Study
Patterson, Brian F.; Kobrin, Jennifer L.
College Board, Slides presented at the Annual Meeting of the American Educational Research Association (AERA) (New Orleans, LA, Apr 10th, 2011)
This study presents a case for applying a transformation (Box and Cox, 1964) of the criterion used in predictive validity studies. The goals of the transformation were to better meet the assumptions of the linear regression model and to reduce the residual variance of fitted (i.e., predicted) values. Using data for the 2008 cohort of first-time, first-year students attending 129 different colleges and universities, this study demonstrated how the transformation may be incorporated into the process of validating an admission test for predicting students' first-year grade point average (FYGPA). In this setting, we regularly found substantial negative skewness in both the criterion and its residuals and the use of a Box-Cox transformation of the criterion yielded residuals that much better approximated the assumption of normally distributed residuals. The result is that with relatively little additional effort, researchers may perform predictive validity studies for college admission measures with one of the most meaningful criteria (i.e., FYGPA) and they can better meet the assumptions of the linear regression model.
Descriptors: Predictive Validity, Evaluation Criteria, Regression (Statistics), College Freshmen, Cohort Analysis, Admission Criteria, College Admission, Grade Point Average, Evaluation Problems, Least Squares Statistics, College Entrance Examinations
College Board. 250 Vesey Street, New York, NY 10281. Tel: 212-713-8000; e-mail: firstname.lastname@example.org; Web site: http://research.collegeboard.org
Publication Type: Non-Print Media; Reference Materials - General
Education Level: Higher Education; Postsecondary Education
Authoring Institution: College Board
Identifiers - Assessments and Surveys: SAT (College Admission Test)