Introduction and Importance of the Study

• In Event: Feasibility of Teacher-Centric Analyses Using NAEP Data

Abstract

This presentation will explain a new initiative of NCES to expand uses of data from the NAEP teacher survey by providing representative teacher-level sampling weights. In this presentation, an overview will be provided on the new initiative of NCES. The presenter will describe the current reporting practices (student-centric reporting) of NAEP and its limitations, benefits of teacher-centric analyses of students’ NAEP achievement, challenges to implement teacher-centric analyses, and tasks that need to be undertaken by NCES. The presenter will use publicly available NAEP data to explain the purpose, tasks, and challenges involved in this project. The study builds on prior work conducted for NCES (Rust, 2013; Rust, 2014; Karr, 2014).
NAEP, a congressionally mandated survey assessment, offers authentic information about how students are performing nationally and within states, and as such it contributes vital information regarding the relationship between teacher characteristics and student outcomes. In each NAEP operational assessment, thousands of teachers of sampled fourth and eighth grade students are asked to complete a survey with items describing their background, teaching practices, and qualifications. Data from this teacher survey have traditionally been disaggregated to the level of students and analyzed using student-level weights for student-centric reporting. NAEP reports on teacher data in this way because it is not designed to survey a representative sample of teachers. Nevertheless, if all or most of the teachers are sampled from each sampled school, the teacher sample is necessarily representative by virtue of having a representative sample of schools.
Researchers and policy makers interested in teacher-centric topics such as the relationship between teacher characteristics and student performance are likely to be interested in the possibility of conducting teacher-weighted analyses using NAEP data. In addition to this general appeal, users of NCES data in particular will be interested in teacher-centric analyses of NAEP data. First, inclusion of teacher sampling weights will increase the comparability between NAEP and other NCES surveys with representative samples of teachers such as SASS. In this way, NAEP would combine the analysis of teacher characteristics and outcomes (which is typical of SASS) with the analysis of student characteristics and outcomes (which is not possible using SASS). Second, teacher weights open up the possibility of conducting representative two-level analyses using student-level outcomes and teacher-level predictors. Third, inclusion of teacher sampling weights permits analyses of teacher-level outcomes, something not previously considered possible using NAEP data.
In summary, the session attendees will benefit from knowledge of how NAEP can expand its reporting dimension to include more information about teachers.

Author

• Taslima Rahman, University of Maryland - College Park