THE SIGI PREDICTION SYSTEM: PREDICTING COLLEGE GRADES WITH AND WITHOUT TESTS

This paper gives an example of how to predict college grades in the absence of tests. It describes components of the prediction system which is one section of the computer‐based System of Interactive Guidance and Information (SIGI). Students who use the SIGI prediction system are provided with detailed criterion information and then asked to rate themselves on “achievement factors” and give estimates of their performance. These ratings and estimates, as well as self‐reported high school record and biographical data, are used as predictors in a regression analysis. Predictive validities for final grades in four courses at Illinois State University were examined. Since students at ISU are required to take ACT, separate analyses were run using SIGI variables, ACT scores, and SIGI plus ACT. Consistent findings were found across the four courses: (1) multiple R's, using two or three SIGI variables, are quite high, falling in the range of .45 to .66, (2) SIGI variables afford a higher level of prediction than do ACT scores, and (3) the addition of ACT scores to SIGI variables does not appreciably improve the prediction levels.