Publication Date




Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PHD)


Educational and Psychological Studies (Education)

Date of Defense


First Committee Member

Walter G. Secada

Second Committee Member

Ji Shen

Third Committee Member

Batya Elbaum

Fourth Committee Member

Soyeon Ahn

Fifth Committee Member

Robert C. Schoen


Teachers’ knowledge of their individual students’ success is an essential component for fostering content-rich mathematical discussion – a central tenet of mathematics education reform. The existing research, however, focuses only on either teachers’ accuracy or teachers’ biased inaccurate expectations without considering how these concepts are related to each other. The conceptual framework presented in this study seeks to bridge the gap in the literature by considering both accurate and inaccurate teacher knowledge of their students in tandem and by using a finer grain of analysis for the concept of teachers’ inaccuracy: overestimation and underestimation. Cognitively guided instruction (CGI) is a professional development program that aims at improving teachers’ understanding of children’s mathematical thinking to improve instruction and student learning. This study examined (1) the effectiveness of the CGI training to increase teachers’ knowledge of their students’ performance, (2) student and teacher contextual factors that may affect the effectiveness of the training, and (3) the relationship between the three theoretical components of teachers’ knowledge of their students. Using an inverse-variance weighted analysis of variance, the study evaluated the effects of the treatment and of ten possible moderators of treatment on teachers accuracy vs. overprediction and accuracy vs. underprediction odds ratios. The professional development was found to have a significant effect on teachers’ knowledge by increasing their predictive accuracy and reducing their overestimations of student performance. Students’ performance, gender, race/ethnicity, English language learner status, learning disability designation, and teachers’ race/ethnicity each had a significant effect in moderating the impact of CGI training. Teachers’ accuracy was increased as overestimation of students’ performance decreased, although there were some instances where training had an effect on underestimation. The implications of these findings for research and professional development practices are discussed.


mathematics education; professional development; teacher knowledge; CGI; teacher accuracy

Available for download on Saturday, August 01, 2020