Teacher Effectiveness Index as an Aid to Determine Performing Teachers for Promoting Excellence in Education

• Activity Number: 42
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section
• Author(s): Avi Singh*+ and Eric C. Hedberg and Tom B. Hoffer and Arend M. Kuyper
• Companies: NORC at the University of Chicago and NORC at the University of Chicago and NORC at the University of Chicago and Northwestern University
• Keywords: Latent Variable Models ; Teacher-Induced and School-Induced Test Score Biases ; Unobserved Covariates ; Value Added Models

Abstract

Direct measures such as average class test or teacher evaluation score by students are not adequate due to potential measurement biases induced by teacher or school. As a supplement,a bias-free indirect measure of teacher effectiveness would be useful. Value-added models attempt to provide such a measure for a subject and grade by treating teacher effect specific to a student as random. The effect varies in general with the subject,student characteristics,and school policies. It is averaged to obtain an indirect measure of teacher's overall effect. There are several issues such as lack of a standardized student population resulting in confounding of the parameter definition with student, teacher, and school covariates, conceptual problem in interpreting student's gain-in-score as a causal effect due to nonrandom assignment,and bias in parameter estimates. We propose an alternative measure based on a new construct of latent teacher effectiveness index (which may represent several behavioral characteristics) introduced explicitly in the model as an unobserved random covariate with pre-specified values of mean 0 and variance 1. It can overcome several concerns of the existing measure.