Research accumulated has suggested that narrowing instructional quality gaps can improve educational equity and the well-being of children in social and economic backgrounds. Considering that the disparity of instructional quality may affect educational inequality across different regions in China, this study explored how teaching quality varied in 30 lessons primary English classrooms in an economically disadvantaged province in China. This study adopted a mixed-method strategy with quantitative classroom observation data to select four lessons contrastive in teaching quality for subsequent qualitative analysis to explore classroom processes in-depth. Using two internationally validated classroom observation instruments, ICALT and TEACH, added a further dimension to examine how characteristics of instruments might influence perceived instructional quality. Results revealed that while both high-inference instruments were theoretically comparable in distinguishing teaching quality, only ICALT predicted learner engagement. While quantitative instruments could not provide detailed accounts of classroom processes, qualitative accounts of the four lessons could uncover the deep relationships between teacher-student interactions and differences in instructional quality. These findings suggest that conceptually similar instruments may vary in predictive power and that systematic qualitative analysis is indispensable in complementing high-inference instruments to provide an objective teacher evaluation.