Teachers' Knowledge of Students: Defining a Domain and Its Relationship to Student Achievement

• In Event: 66.051 - Tracing the Production Function: Connecting Teacher Preparation, Knowledge, Beliefs, and Instruction to Student Achievement

Abstract

Teachers’ knowledge of their students ranks high on the list of competencies recommended by professional standards and scholars (NCTM, 1991; Shulman, 1986). Many observe that such knowledge enables a variety of effective classroom strategies, including adjusting the pacing of instruction (Clark & Peterson, 1986), forming appropriate instructional groups (Shavelson & Borko, 1979), and designing tasks and questions to further student understanding. Improving teachers’ knowledge of their students has also formed the basis for several significant professional development efforts, with some evidence suggesting that such approaches have worked to improve teacher knowledge and student achievement (Bell, Wilson, Higgins & McCoach, 2010; Carpenter, Fennema, Peterson, Chiang, & Loef, 1989; Black & Wiliam, 1998).

A review of this research literature shows diverse approaches to measuring teachers’ knowledge of students. Beginning in the 1960s, numerous scholars measured teacher accuracy in predicting performance on cognitive assessments (Helmke & Schrader, 1987; Hoge & Coladarci, 1989). A more recent strand (Carpenter, Fennema, Peterson & Carey, 1988; Sadler et al., 2013) focuses on teachers’ understanding of student thinking, including students’ prior knowledge, likely developmental pathways, and prevalent misconceptions. However, these two metrics have seldom been compared to one another or subject to rigorous assessments of reliability and validity.

To address this issue, we create measures of these constructs and then explore their reliability and validity. We draw on data collected over three academic years (2010-2013) from roughly 300 fourth and fifth grade teachers and their students in four urban school districts. To measure accuracy, we first asked teachers to predict the percent of students within their class who would answer specific items on a project-administered test correctly. Accuracy was estimated as the difference between teacher predictions and actual student performance. To measure understanding of student thinking, we asked teachers to predict the most likely incorrect response on each item, then credited teachers when those predictions matched the modal incorrect student response.
We find a weak to moderate degree of reliability in teachers’ scores on these two measures. Cross-year correlations range from 0.03 to 0.56, depending on the measure, grade, and school year. Internal consistency reliability estimates range from 0.39 to 0.76. At the same time, we find moderate convergent and predictive validity evidence. Attenuated correlations between these scores and teachers’ mathematics knowledge are significant and range up to 0.25. In addition, teachers with higher scores on the accuracy measure have students who post better achievement on state standardized assessments (r = 0.11, p<0.05) and a low-stakes assessment (r = 0.15, p<0.01). Future analyses will investigate the causal relationship between teacher knowledge of students and student outcomes, first by analyzing data from a year in which teachers were randomly assigned to class lists, then by using a quasi-experimental design that relies on variation in teacher knowledge observed within schools across adjacent grades and years. Our results highlight the potential for teacher’s knowledge of students to be used as a measure of a teacher’s effectiveness, but also stress the importance of continued exploration of this construct.  

Authors

• David Blazar, Harvard University

• Heather C. Hill, Harvard Graduate School of Education

• Mark Chin, Harvard University

• Daniel McGinn, Harvard University