AI-enabled Personalized Economics Tutoring Platform for MSc Education Impact

Thesis Abstract

This study investigates the impact of an AI-enabled personalized economics tutoring platform on MSc economics education, addressing persistent gaps in learner engagement, differential attainment, and the attainment of core econometrics and macroeconomics competencies in graduate programs. The central aim is to evaluate how adaptive feedback, individualized learning paths, and intelligent tutoring features influence student performance, motivation, and perceived self-efficacy among master's students. Specific objectives are (1) to assess changes in course achievement in advanced economics modules after platform integration; (2) to examine variations in engagement, time-on-task, and persistence across learner profiles; (3) to identify how personalized interventions affect mastery of econometric methods and applied microeconomic theory; (4) to explore instructors’ perceptions of workload, instructional quality, and scalability; and (5) to determine the platform’s cost-effectiveness relative to traditional tutoring approaches. The study adopts a quasi-experimental mixed-methods design over two academic semesters in a large research university. The population comprises MSc economics students enrolled in core and elective modules, with a target sample of 320 students across four cohorts. A purposive sample of 24 instructors participates for qualitative insights. The intervention group (n=160) engages with the AI tutoring platform integrated into the course management system, while the control group (n=160) follows standard instruction without AI-enabled support. Data collection instruments include (i) platform analytics capturing login frequency, sequence of learning activities, time-on-task, hint usage, and mastery checks; (ii) standardized assessments for econometrics, macroeconomics, and microeconomic theory; (iii) validated motivation and self-efficacy scales; (iv) instructor surveys focusing on perceived instructional quality and workload; and (v) semi-structured interviews with students (n=40) and instructors (n=12). Validity and reliability of instruments are established through pilot testing, Cronbach's alpha assessment, and content validity checks by economics education experts. Data analysis employs a multi-step approach. Quantitative data are analyzed using ANCOVA to compare post-test scores between intervention and control groups while controlling for pre-test performance, with regression analyses to identify predictors of learning gains. Multilevel modeling accounts for nesting of students within modules. Time-series analysis of platform usage patterns reveals engagement trajectories. Mediation analyses test whether engagement and perceived self-efficacy mediate the relationship between platform exposure and learning outcomes. Qualitative data from interviews are analyzed via thematic analysis to extract themes on user experience, perceived authenticity of feedback, and implementation challenges, with coding validated through intercoder reliability checks. A cost-effectiveness analysis compares the incremental cost per learning outcome achieved versus traditional tutoring. Expected findings include statistically significant improvements in post-test performance for the intervention group, particularly in econometric reasoning, data interpretation, and application of theory to case-based problems. Enhanced engagement metrics are anticipated, including higher average weekly active days and reduced dropout risk in challenging modules. The study anticipates that improvements are moderated by prior attainment and self-regulatory skills, with the greatest gains observed among mid-level performers who actively utilize personalized feedback and mastery checks. Qualitative insights are expected to reveal positive perceptions of adaptive feedback, more efficient use of instructor time, and perceived scalability, alongside challenges related to system integration, data privacy, and the need for ongoing content updates. The study contributes to knowledge by providing empirical evidence on the effectiveness and cost-efficiency of AI-driven personalized tutoring in graduate economics education, extending theories of intelligent tutoring systems and self-regulated learning to advanced econometrics and economic theory instruction. It extends the literature on educational technology in higher education by detailing implementation considerations, teacher roles, and equity of access within MSc programs. The main conclusion is that AI-enabled personalized tutoring can significantly enhance mastery of complex economics concepts and improve student engagement when integrated with rigorous assessment and continuous professional development for instructors. Recommendations include scaling the platform across additional modules, investing in robust data governance frameworks, ensuring transparent explainable AI feedback, training faculty in data-informed instructional design, and conducting longitudinal follow-ups to assess lasting impact on research productivity and career-readiness.