ERIC Number: ED113369
Record Type: Non-Journal
Publication Date: 1975-Jun
Reference Count: N/A
Study of Academic Growth Using Simplex Models. Final Report.
Werts, Charles E.; Linn, Robert L.
Forming a sequence covering the various aspects of the simplex model, four articles are presented here under the following titles: "A Simplex Model for Analyzing Academic Growth", "Analyzing Ratings With Correlated Intrajudge Measurement Errors", "The Correlation of States With Gain", and "The Reliability of College Grades from Longitudinal Data". The most important finding of this study is that a simplex model which allows for measurement error, fits a variety of longitudinal academic data quite well. This allows for attenuation corrections when only one measure is available at each time. More importantly, the results suggest that the commonly used split-half or parallel form procedures for estimating reliability may typically yield overestimates or reliability due to "method" variance, i.e., nonindependent measurement errors resulting from the use of closely similar item types. The simplex model appears less subject to this problem because both item format and content change over time. It has been demonstrated that accurate corrections for attenuation are essential to a study of the determinants of academic growth. (Author/RC)
Descriptors: Academic Achievement, Achievement Gains, Analysis of Covariance, College Students, Computer Programs, Correlation, Error of Measurement, Goodness of Fit, Grades (Scholastic), Longitudinal Studies, Mathematical Models, Matrices, Measurement Techniques, Predictor Variables, Scores, Statistical Analysis, Test Reliability, Testing Problems, True Scores
Publication Type: Reports - Research
Education Level: N/A
Sponsor: National Inst. of Education (DHEW), Washington, DC.
Authoring Institution: Educational Testing Service, Princeton, NJ.