The learning crisis in low- and middle-income countries (LMICs) is significantly marked by a high degree of heterogeneity in student learning outcomes. Such variations, extending across student abilities, backgrounds, and learning speeds, are believed to be partly a result of teachers failing to devote sufficient attention to students who are low-performing. This lack of attention to low-achieving students can widen learning gaps, exacerbating the already critical learning crisis. With this in mind, my dissertation takes a novel approach to investigate the accuracy of teachers' beliefs about their students' academic abilities, which could significantly influence how they cater to different learners. My focus is on identifying and implementing an intervention, designed to enhance this accuracy with the ultimate aim to shift teacher behavior at scale in an LMICs, paving the way for improved learning outcomes across the board. Drawing on existing literature, I first develop a theoretical framework to understand the factors that influence teachers' belief formation about their students' academic skills in LMICs. This ecological framework examines the interplay of elements that shape teacher beliefs, ranging from data availability and quality to institutional factors. Building on this framework, I design a low-cost intervention providing diagnostic feedback to teachers on their students' foundational skills. Using a large-scale randomized experiment in Bangladesh's secondary schools, I estimate the impact of the intervention on student learning outcomes. The results reveal a series of interconnected changes triggered by providing teachers with diagnostic feedback. First, the intervention helps teachers become more accurate in their estimation of student abilities, a crucial factor in tailoring instruction to individual students' needs. Following this improved accuracy in perception, I find evidence of a shift in teaching practices, with teachers more likely to interact with low-achieving students. This targeted attention to struggling students appears to cultivate positive changes in students' socio-emotional skills, specifically in areas such as self-management, growth mindsets, and self-efficacy. These improved student beliefs, in turn, seem to stimulate an increase in student effort. Finally, this series of effects culminate in improved student achievement: an increase of 0.21[sigma] in standardized math test scores, with effects notably concentrated among students in the bottom two terciles of the achievement distribution, where the increase reaches up to 0.25[sigma] .These findings suggest that providing targeted diagnostic feedback on foundational skills can effectively enhance learning outcomes, especially for low-achieving students, by nudging teachers to re-optimize their instructional strategies independently. From a public policy perspective, the results of my dissertation highlight the importance of providing teachers with pertinent and actionable data in a comprehensible format to enable them to better tailor their instructional techniques. Based on the empirical evidence generated, I argue that by focusing on low-cost scalable interventions, such as diagnostic feedback, policymakers can foster improved learning outcomes in LMICs, particularly for the most vulnerable students. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]