Predicting Pre-service Classroom Teachers’ Civil Servant Recruitment Examination’s Educational Sciences Test Scores Using Artificial Neural Networks
Metin DemirDepartment of Elementary Education, Faculty of Education, Dumlupınar University, Kutahya Turkey
This study predicts the number of correct answers given by pre-service classroom teachers in Civil Servant Recruitment Examination’s (CSRE) educational sciences test based on their high school grade point averages, university entrance scores, and grades (mid-term and final exams) from their undergraduate educational courses. This study was therefore designed by using a general survey model. The participants were 219 graduates of the departments of classroom teacher education from the education faculties of two different state universities. Artificial neural networks (ANNs) were used to predict the numbers of correct answers from the CSRE educational sciences test. As a result of different trials, the correlation between the predicted and actual numbers of correct answers was examined, and 10 ANN models were included in the study. Statistically, significant positive correlations were found between the numbers of correct answers predicted by the ANN and the students’ actual correct answers in the CSRE. The highest loading was r = .63 (p < .01), and the lowest was r = .43 (p < .05).