Quantifying Intrinsic Motivation Effects on NCR Private School Outcomes

Abstract

Intrinsic motivation (IM) is a recognised determinant of cognitive engagement, yet its quantifiable causal influence on student academic outcomes in National Capital Region (NCR) private schools has not been adequately examined through empirically rigorous models. This research seeks to establish a multi-faceted, data-driven approach to measure the impact of IM on academic performance metrics and to discover moderating factors, such as teaching approaches, socioeconomic factors, and exposure to digital learning. This study fills an important gap in the literature by bringing together psychometric, behavioural and machine learning approaches into a unified model. The study adopted a mixed methods approach using Structural Equation Modeling (SEM), Hierarchical Linear Modeling (HLM) and machine learning regression. A stratified random sample of 450 students from 15 private schools across three tiers in NCR was used. Motivation was defined as intrinsic motivation measured by an adapted Academic Self-Regulation Questionnaire and related behaviours such as assignment completion and voluntary participation. Academic performance was assessed through standardised test scores and continuous assessment scores. Qualitative perspectives were gathered via semi-structured interviews with teachers. Findings show that IM has a statistically significant direct impact on overall GPA (beta = 0.22, p < 0.001) and a much larger total effect (beta = 0.76, p < 0.001) via intermediary factors of academic engagement, self-efficacy, and teaching support. Random Forest had the best fit (R² = 0.782, RMSE = 5.64) for predicting academic performance, with IM as the most influential predictor. School tier moderated the IM-GPA relationship, with elite-tier students displaying higher levels of motivation (M = 78.3, standard-tier: M = 59.8). These findings advance the theoretical understanding of motivation-performance dynamics in high-density urban private school contexts and provide a scalable predictive framework applicable to policy intervention design.