The role of teacher support in students' academic performance in low- and high-stakes assessments.

Teachers' impact on learning outcomes is well supported, but more robust evidence is needed on the complex indirect pathways through which teacher support impacts performance in low-versus high-stakes examinations. This study sought to understand the divergent mechanisms through which teachers' support for students affects their performance on low- and high-stakes assessments. Bayesian structural equation modeling with the Markov Chain Monte Carlo algorithm and data from Ghana were used to test mediated, unmediated, and moderated-mediated relationships. The results show that a typical student's high-stakes exam performance increases by approximately 13 % for each one-unit increase in teacher support. A chain mediation effect also exists through student homework behavior and low-stakes everyday performance. Student gender is also a moderator. These results could aid in identifying the malleable leverage points associated with low- and high-stakes assessments. Such empirical clarity would help education administrators to develop appropriate professional development programs that enhance teachers' support roles, enabling them to respond better to learning disparities and related challenges. • Teacher support significantly predicts low and high-stakes academic performance. • Teacher support predicts end-of-year academic performance directly and daily schoolwork performance indirectly. • Homework behavior mediates the relationship between teacher support and student academic performance. • Teacher support contributes to improvement in girls' homework behavior more than boys. [ABSTRACT FROM AUTHOR]