ABSTRACT
This chapter focuses on gathering evidence of systemic misconduct by educators at the classroom, school, or district level. It presents the modeling strategy which, uses aggregated results of an examinee-level nonlinear regression model to make unit-level inferences regarding the reasonableness of observed proportions of test takers classified into performance level categories, conditional on students' prior scale scores. The chapter presents the modeling strategy and provides a basic framework upon which a researcher can build a functioning longitudinal model to identify classrooms, schools, or districts with unusual changes in performance level classifications relative to prior student achievement. It considers two longitudinal measurements, which will be characterized between-cohort and within-cohort modeling options. Setting an appropriate flagging criterion to identify unusual changes in performance is a critical decision that requires careful consideration of both theoretical and practical concerns. Flagging rules can be adjusted with this pursuit in mind standardized residuals could be rank-ordered to prioritize units that appear to be most unusual.
