Context: Improving teacher quality and granting every student equal access to high-quality instruction have been a shared goal of many parents, educators, and policy makers for decades. In practice, teacher qualification measures (e.g. teacher degree, years of experience etc.) are often used as proxies of teacher quality. However, Shulman (1986) introduced the concept of pedagogical content knowledge (PCK) as a pivotal determinant of teacher quality, which amalgamates subject matter knowledge and knowledge of effective teaching strategies. This suggests that PCK may offer a more reliable indicator of teacher quality compared to traditional proxy measures. Nevertheless, the causal relationship between PCK is causally related to the quality of instruction received by students, which ultimately influences students' learning outcome, remain inadequately studied. Given that instructional interactions in the classroom serve as the primary conduit for teaching and learning, understanding how instructional practices impact instructional quality could not only advance our comprehension of effective teaching practices but also devise more targeted approaches to support teachers in delivering high-quality instruction to all students. Research Question: Substantively, this research intends to produce robust evidence to addresses research questions including, whether teachers with a higher level of PCK tend to deliver instruction with higher quality, which facets of instructional delivery exhibit the most pronounced response to enhancements in PCK. Methodologically, this research outlines an empirical strategy to analyze experimental data, wherein teacher allocation to exchangeable classrooms was randomized. This analysis employs hierarchical linear modeling, and to rigorously construct assessment scores for instructional quality from multiple observation tools. Description of Data: This research will use the Measures of Effective Teaching (MET) Longitudinal Database as the analytic sample. Overall, the MET project collected data from approximately 2,700 teacher volunteers and their 160,000 students in the six urban school districts of U.S. and contains high-quality data on three major categories of measures: classroom observation in the form of video coding and professional ratings, student perception surveys of their classroom experience, and administrative data on students' demographic information and prior scores of state standardized tests. This paper will mainly use the data from the 2010-2011 Academic Year, where the MET project implemented an experimental design, and focus on the mathematics classrooms. Research Design and Intervention: Prior to year 2 of the MET project, the school principals were asked to form "exchangeable" class portfolios across grade-subject combinations to ensure similarity and comparability in student composition. In year 2, eligible teachers were randomly assigned to these exchangeable classes, resulting in identical student compositions across participated classes within each randomized block. The randomization, conducted at school level, can be viewed as a multisite randomized experiment, with each school acting as a site where the same experiment was replicated. Teachers who have different levels of PCK may also differ in other characteristics such as teaching experience, and additional training obtained by pursuing advanced degree. These teacher characteristics may confound the causal effect of PCK on instructional quality. This study investigates and presents empirical evidence on whether these teacher qualification characteristics confounds the causal relationship of PCK and instructional quality. Another potential threat to causality is noncompliance and attrition, which are not significant concerns in this study. Data Analysis: The analysis of this paper includes two key components: constructing the outcome variable for instructional quality and conducting a causal analysis of PCK's impact on instructional quality factors. To address differences in student development and school structure across grades, observations are segmented into three sub-sample, i.e., elementary schools (4th to 5th grade), middle (6th to 8th grade) and high schools (9th grade), each analyzed separately using parallel approach described as followed. To construct instructional quality factors, I first conducted an exploratory factor analysis to uncover underlying structures across two commonly used measurement frameworks, Classroom Assessment Scoring System (CLASS) and Mathematical Quality of Instruction (MQI) along with Tripod survey, followed by confirmatory factor analysis using structural equation modeling (SEM) to identify the best-fit model. The exploratory analysis indicated weak associations among items across frameworks but stronger correlations within each framework, suggesting potential for dimensionality reduction on top of the current dimensions. Furthermore, the confirmatory factor analysis suggested a classroom management dimension measured by three CLASS items, an instructional interaction dimension measured by the rest of the 9 items of CLASS, a mathematical coursework planning dimension measured by 3 items of MQI, and student feedback dimensions of items from Tripod. Based on the experimental design and data structure, I specified a three-level hierarchical linear model (HLM) that classrooms are nested within randomized blocks, and randomized blocks are nested within schools. The SEM-constructed instructional factors and original additive scores were analyzed as the outcomes, with predictors centered within randomized blocks, where classrooms of the same grade are made exchangeable in students' demographics and prior academic background. As aforementioned, observable teacher qualification characteristics were considered potential confounders of the causal relationship between PCK and instructional quality. Using the same model set-up, results showed that although years of experience and racial identity significantly associated with teachers' average PCK, none of these observational qualification characteristics predict instructional quality under the randomized design, thus ruling out their confounding effects in this study. Findings: In elementary school, mathematical teachers' PCK primarily affects the mathematical quality of instruction. Specifically, classes taught by teachers with higher PCK feature richer content related to Mathematics and more time allocation for working with students in Math problems than teachers with lower PCK. In middle school, apart from the mathematical quality of instruction, instructional quality ratings related to teacher-student interaction, such as on student participation in meaning making and reasoning, and teachers' respect for students' perspectives, significantly increase with teachers' PCK levels. In high school, teachers generally possess higher levels of PCK on average, the impact of PCK is primarily reflected on students' perception of their classroom experience. Students reported a more positive classroom experience with teachers who have higher PCK compared to those with lower PCK.