Construction and validation of a scale for predicting graduation from a college of optometry
Taulbee, George C., Sr.
MetadataShow full item record
The purpose of this study was to construct and validate a scale for the prediction of successful completion of the educational requirements for the practice of Optometry. The subjects for this study were males who had been admitted as full-time students to the College of Optometry, University of Houston, Houston, Texas, during the academic years, 1952 through 1956. Other criteria established for the inclusion of a subject in the total sample were the following: he must have taken all or part of the Optometry Test Battery prior to or during his first year of residence in the College, and he must have been awarded grades for at least one semester of study in the College. The general sample was divided into two groups. Group V, the validation sample, was comprised of students entering the College of Optometry during any of the academic years, 1952, 1954, or 1956. The subjects for Group CV, the cross-validation sample, were selected from students admitted to the College during the academic years of 1955 or 1955. Students in Groups V and CV were classified as "Failures" or "Successes," the criterion-measures. The "Success" category (Groups V-S and CV-S) was comprised of those subjects who successfully completed the three-year Optometry curriculum. The "Failure" category (Groups V-F and CV-F) consisted of those students who failed to graduate from the College and who, in addition, had failed to earn, during the last semester attended, the qualitypoint average necessary to qualify them as students in good academic standing for the subsequent semester. The Optometry Test Battery, administered to all applicants at the time of admission to the College, but in no case used as a criterion for admission, consisted of nine tests yielding a total of twenty-nine scores. To select those measures which would most likely prove to be efficient predictors, an estimate was made of the relative efficiency of each in discriminating between the "Success" and "Failure" of subjects in Group V. These estimates were made by using the t test of the differences between the medians of each group on the 29 variables. Of the 29 measures, 11 had P-values equal to or less than .05. These were 1. Cooperative English Test, Form PM, Usage, Vocabulary, and Total Score (three measures) 2. Cooperative General Achievement Tests: Test II: Natural Sciences (one measure) 5. Cooperative General Achievement Tests: Test III: Matnematics (one measure) 4. Survey of Mechanical Insight (one measure) 5. American Council on Education Psychological Examination for College Freshmen, 1947 Edition, Q- Seo re and Total Score (two measures) 6. The Use of Library and Study Materials, Part I, Part II, and Total (three measures The number of predictor-measures was reduced from eleven to eight, however, by the elimination of the three scores on The Use of Library and Study Materials, for this test had not been administered to a sufficient number of the subjects in Group CV to warrant its use in this study. Frequency distributions of the scores made by the validation group on each of the eight predictor-measures were prepared showing, by class interval, the number of successful and unsuccessful students. A "Score Sheet" was then constructed to show the percentages of successful and unsuccessful students falling in each class interval of each measure. The test scores of the 59 subjects in Group V were converted into their respective "Weighted Scores" by use of the "Score Sheet" and the sum of the "Weighted Scores" for each student constituted his "Total Predictive Score." A frequency distribution of the 59 "Total Predictive Scores" was then constructed showing the numbers and percentages of successful and unsuccessful students in each of the class intervals. These data comprised the "Predictive Scale." The "Total Predictive Score" for each of the 54 subjects in the cross-validational group (Group CV) was computed by summing his "Weighted Scores" obtained by means of his scores on the predictor-measures and the "Score Sheet." A frequency distribution of the "Total Predictive Scores" for this group was then constructed and predictions made, by class interval, relating to the percentage of CV subjects expected to be successful in Optometry. The accuracy of these predictions by class interval ranged from 50 to 100%. The accuracy of prediction for the total cross- validational group was 96%. Following the aforementioned procedures, predictions of percentages of success for the cross-validational group were made using each of the predictor-measures independently. The percentages of accuracy of these predictions for the total group ranged from 85% to 89%, somewhat less than the accuracy of prediction achieved by means of the "Total Fredictive Score" based upon all eight criterion-measures.