The construction of scales for predicting academic success in Grade 7

The need for a more effective method of predicting academic success within the junior high school has become an increasingly greater problem, since this is the level of education in which the alarming dropout rate has its inception. Although the numerous factors influencing academic success receive more attention as changes occur, the time and information required to ferret out many of the crucial factors involved in an individual student's decision are not always available. One means of simplifying the problem and at the same time providing usable information was developed by Ford (1945) as he adapted a method first used by Glueck and Glueck (1940) in their studies and predictions of recidivism of delinquent behavior among juveniles. In his adaption. Ford considered seven factors which proved to have a predictive validity ranging from 78 percent to 100 percent for various aspects of academic attainment at the high school level. The same method was employed in the present study for the purpose of constructing a scale for predicting academic success in Grade 7 of the junior high school and to determine the accuracy with which such predictions could be made. Success at Grade 7 level was defined in terms of quality point average. All students achieving an average of 2.50 and above were termed successful whereas those whose average fell below 2.50 were termed unsuccessful. The quality point average was computed on the basis of a letter grade of 'A' receiving four quality points for each particular subject area, 'B' receiving three points, 'C' two points, 'D' one point, and 'F' zero quality points. The sample group was comprised of 371 students currently attending a Junior high school of a large metropolitan area and who met the criteria of (1) completing both sixth and seventh grade in the same area of a much larger school district, and (2) having scores available on the Stanford Achievement Test, Arithmetic Diagnostic Test, Otis Quick- Scoring Mental Ability Test, and a Grade 6 quality point average. Ten variables, nine scores from the three tests and the students Grade 6 quality point average, were used to determine each student's total predictive score. The method used was as follows: 1. Frequency distributions were made for each of the ten variables, locating each student in an interval according to his score on each test and in the appropriate successful-unsuccessful column determined by the aforementioned definition of success. 2. The percent of successful students was calculated for each Interval, then adjacent intervals were combined when necessary to produce a percent scale which varied with the magnitude of the actual scores. 3. The resulting percents for each interval for the ten variables comprised the score sheet (Table 1). 4. The sum of the points earned by a student on each variable on the score sheet became his total predictive score. 5. The resulting total predictive scores for the 371 students of the sample group were then tallied into a frequency distribution with the tabulations made according to the interval in which the scores fell and the column determined by the definition of success. The scale was then validated by using 100 randomly selected students who met the same criteria as the sample group but completed Grade 7 one year later than the sample group. The method for validation was as follows: 1. Students were divided according to the definition of success into successful and unsuccessful. 2. Total predictive scores were computed from the score sheet for each student on each of the ten variables. 3. The ten scores for each student were summed and the total predictive score located in the appropriate interval of the 'Scale for Predicting a Quality Point Average of 2.50 or Above' (Table II). In making the predictions, a gross prediction of success was made for all students whose total predictive scores placed them in or above the 450-524 interval, since the odds in these intervals, of the students in the original sample group, were greater than 50 in 100 that a quality point average of 2.50 or above would be earned. A gross prediction of unsuccessful was made for all students whose total predictive scores placed them in intervals below the 450-524 interval. Compared on the basis of these gross predictions, the accuracy of prediction within the definition of success was 82 percent, whereas the accuracy of prediction within the definition of unsuccessful was 91 percent. The accuracy of prediction for each separate interval ranged from 57 percent to 100 percent, with the greatest error represented in the middle of the scale as anticipated. The accuracy of prediction for the total validation group was 86 percent.