No abstract

No abstract

Student-reported past grades were comparedwith school-reported grades as to their usefulness in predicting success i n matheviatics courses. Student-reported grades were as valid as srhool-reported grades i n predicting criteria of algebra success but werv slightly less valid in predicting criteria of geometry success. I t is concluded that for the purposes described here, student-reported grades may be used. EDUCATORS have long been aware that students' past grades improve the predictive efficiency of special-purpose aptitude measures. However, school personnel often lack time to look up grades and apply appropriate weights to grades and test scores.A way in which schools can utilize students' past grades without making undue demands on guidance workers' time was devised by the American College Testing Program Examination. At the time of testing, students are asked to report four grades on their answer sheets. Scores on the tests proper are then combined with these student-reported grades during the computerized processing of the answer sheets. In this way, appropriate weights are assigned to the grades and test scores.Special-purpose aptitude tests have more recently employed noncomputerized, hand-and machine-scoring procedures by which grades and test scores are conveniently weighted and combined to yield composite scores. Thus, by hav-

EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT 1971, 31, 521-523. THE procedure of cross-validation is a necessary step in establishing the predictive power of a prognostic instrument in which weights are applied to the separate parts making up the total score. The technique involves the determination of the &dquo;best&dquo; weights with an initial sample and the subsequent verification of these weights with a second sample from the same population.A total score on the Orleans-Hanna Algebra Prognosis Test and on the Orleans-Hanna Geometry Prognosis Test, is a composite of (a) four student-reported past course-grades, (b) student-predicted course-grade in algebra or geometry, and (c) number right of the Prognosis Test work-sample items. For both of the Orleans-Hanna tests, an a priori weight of two was assigned to each of the five course-grade variables and a weight of one was assigned to the test score representing the number of work-sample items answered correctly. Validation studies (Orleans and Hanna, 1968 a, b) were undertaken during the norming of the tests to determine the multiple regression weights for predicting each of the four criteria of success: mid-year and final course grades, and mid-year and endof-year achievement test scores. The multiple correlations were then compared to the zero-order correlations obtained with the a priori 1

No abstract