Predictors of Licensure Examination for Teachers: A Comparative Analysis of Multiple Linear Regression and Artificial Neural Network

Abstract

The growing demand for high-quality education underscores the vital role of Teacher Education Institutions (TEIs) in shaping future educators who can meet the challenges of a dynamic society. Among the key indicators of a TEI's effectiveness is the success of its graduates in the Licensure Examination for Teachers (LET), a critical benchmark for professional readiness and teaching competence in the Philippines. As such, it is imperative to identify the predictors of board examination performance to help TEIs refine their academic programs, enhance their graduates' preparedness, and improve success rates in LET. Hence, this study aimed to identify the best predictors of LET among graduates of Bachelor of Secondary Education (BSE) program at Isabela State UniversityEchague Campus. To achieve this goal, the study uses advanced statistical models, namely, Multiple Linear Regression (MLR) and Artificial Neural Network (ANN), to analyze data on the academic performance of Batch 2019 graduates who immediately took the LET after graduation. The study found a significant positive correlation between academic performance and LET performance among BSE graduates. The study also compared the predictive abilities of MLR and ANN models using Mean Absolute Percentage Error. Results revealed that the ANN model had a lower forecast error compared to the MLR model for the General Education component of LET in the BSE program. However, the MLR model had a lower forecast error for the Professional Education component of LET in the same program. Moreover, the predictive abilities of the two models vary across the six major courses of the LET. While the data used courses in the old curriculum, it is recommended that the same study be conducted using data from the graduates of the new curriculum.