The construction and validation of a scale for predicting academic success in the eighth grade
The purpose of this study was to construct a scale for predicting academic success in the eighth grade from scores made on selected psychological tests and academic performance in the seventh grade. The six variables selected consisted of (1) two subtotals from the Academic Promise Tests, (2) three scores on tests from the Science Research Associates' Achievement Series, Multilevel Edition, Form C, Battery Green, and (3) grade point averages computed from four seventh grade basic courses. The sample group consisted of 353 male and female students who had attended a certain large junior high school at least since September of 1965 and had completed grade eight in the Spring of 1967. This group was divided on a random or chance basis into two subgroups of 200 and 153, respectively. The first subgroup was used to construct the scale, the second was used to validate the scale. School records provided the data, test scores, and school marks received during the seventh and eighth grade school years. Frequency distributions of the scores were made for each of the six variables according to a dichotomy of 'successful' and 'unsuccessful' performance. 'Successful' performance was defined as a yearly grade point average of 3.0 (C) or better. These frequency distributions were arranged to give percentages of success that varied directly with the magnitude of grades. These percentage figures constituted the 'Score Sheet.' A 'Total Predictive Score' was calculated for each individual and a frequency distribution of the 200 'Total Predictive Scores' was constructed in terms of a successful- unsuccessful dichotomy based on eighth grade grades. This distribution was the 'Predictive Scale' and the 'Percent Successful' column of this table became the means by which predictions of success in terms of odds were made. This scale was validated by using the remaining 153 students of the original 353. The scores for these students on the six variables were converted to 'Total Predictive Scores,' by using the 'Score Sheet' and these were placed in the appropriate interval. If 50 percent or more of the individuals in any interval were successful, it was predicted that all those students of the validation group whose scores fell in these intervals would be successful. If less than 50 percent of the students in any interval were successful, it was predicted that all those students in the validation group whose scores fell in these intervals would be unsuccessful. The grade point average a student actually made in the eighth grade was then compared to the predictions made. Accuracy of prediction was determined for each class interval and the scale as a whole. Predictions for the scale as a whole were correct in 80 percent of the cases. This method provides a relatively simple and yet accurate and useful way to construct a scale for predicting academic success.