An investigation into methods of predicting underachievement of above average I.Q. pupils at a boys' school
Powell, Allan Steytler
MetadataShow full item record
Educators have long been concerned at the fact that there are pupils at all three levels of education who appear to have mental abilities far in excess of their actual academic performance. Such people are called underachievers, but the exact definition of underachievement tends to fluctuate according to the purposes of the investigator or the statistical method used. Various personality and social factors have been found to be associated with underachievement, such as poor self image, poor study methods and home conditions, but the relationships have not been found to be necessarily causal. Doubt has been expressed about the accuracy of predicting academic success since underachievers may be victims of over prediction. However, the highly significant correlations of 0,50 at secondary level between verbal I.Q. scores and academic results have made researchers conclude that the former provides researchers with a valid tool to measure academic potential, even though only 50% of the variation has been explained. All indications are that pupils who underachieve at secondary level are doomed to leading unfulfilled lives because of their unrealised potential. This investigation therefore tried to establish a valid method to be used by teachers to predict underachievement at secondary level amongst boys of above intelligence at a semi-urban South African English speaking school. A total of 164 subjects were chosen, representing three consecutive groups of standard seven pupils to reach standard ten without failing. Four models of predicting underachievement were investigated: the linear regression, the stanine difference, the performance index and the step-wise regression. Variables used were the test results from Junior Aptitude tests, Junior Scholastic Proficiency Battery tests, five standard seven subject examinations, the total standard seven and ten examination results, as well as the verbal I.Q. totals of the pupils. The verbal I.Q. total was used as a predictor of academic potential in the linear regression, stanine difference and performance index models at standard seven level and the subsequent performance of the pupils was used to refine the critical level of each predictive model. The results of the three models were compared, and statistical definitions found for each which described the area they had in corrunon. The linear regression model was found to be the most satisfactory to use, with a critical cut-off line at 0,75 standard deviations below the regression line. The stepwise regression results were analysed but discarded as impractical. The secondary aims of the study were the use of an estimated I.Q. to predict underachievement, the relationship between aptitudes and underachievement, and underachievement and the changing of schools. The estimated I.Q. was found to be an adequate substitute instead of verbal I.Q. to predict underachievement. No discernable relationship was found in the three dimensional histograms relating defined aptitudes to both achievement and ability. Underachievement was found amongst pupils who change schools before the end of standard seven at the 5% level of confidence, but not amongst those who change seven. We therefore conclude that a reliable tool has been found to predict underachievement at standard seven level, and an estimated I.Q. can be used as a substitute predictor for statistical purposes. Using both, a teacher is able to identify potential underachievers and with the help of wise counselling help pupils lead more academically rewarding lives.